cystic fibrosis research paper introduction

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 15 November 2021

CyFi-MAP: an interactive pathway-based resource for cystic fibrosis

• Catarina Pereira 1 , 2 ,
• Alexander Mazein 3 , 4 ,
• Carlos M. Farinha 1 ,
• Michael A. Gray 5 ,
• Karl Kunzelmann 6 ,
• Marek Ostaszewski 3 ,
• Irina Balaur 3 , 4 ,
• Margarida D. Amaral 1 &
• Andre O. Falcao 1 , 2  

Scientific Reports volume  11 , Article number:  22223 ( 2021 ) Cite this article

3192 Accesses

4 Citations

1 Altmetric

Metrics details

• Cellular signalling networks
• Chloride channels
• Mechanisms of disease
• Molecular medicine
• Systems biology

Cystic fibrosis (CF) is a life-threatening autosomal recessive disease caused by more than 2100 mutations in the CF transmembrane conductance regulator (CFTR) gene, generating variability in disease severity among individuals with CF sharing the same CFTR genotype. Systems biology can assist in the collection and visualization of CF data to extract additional biological significance and find novel therapeutic targets. Here, we present the CyFi-MAP—a disease map repository of CFTR molecular mechanisms and pathways involved in CF. Specifically, we represented the wild-type (wt-CFTR) and the F508del associated processes (F508del-CFTR) in separate submaps, with pathways related to protein biosynthesis, endoplasmic reticulum retention, export, activation/inactivation of channel function, and recycling/degradation after endocytosis. CyFi-MAP is an open-access resource with specific, curated and continuously updated information on CFTR-related pathways available online at https://cysticfibrosismap.github.io/ . This tool was developed as a reference CF pathway data repository to be continuously updated and used worldwide in CF research.

Similar content being viewed by others

Integrative genomic meta-analysis reveals novel molecular insights into cystic fibrosis and ΔF508-CFTR rescue

Representation and quantification of module activity from omics data with rROMA

Electronic health record phenotypes associated with genetically regulated expression of CFTR and application to cystic fibrosis

Introduction.

Omics technologies revolutionized the way researchers generate data 1 . The integration of disease-specific data allows not only to capture knowledge at its multiple levels of organization but also to unbiasedly identify molecular features, such as phenotypes associated with complex and heterogeneous diseases—which would remain otherwise unnoticed 2 , 3 . The visual representation of molecular mechanisms has emerged as a powerful tool for a better understanding and analysis of specific disease-causing features. This led to the development of several pathway-based resources, such as KEGG (Kyoto Encyclopedia of Genes and Genomes) 4 , Wikipathways 5 , Reactome 6 and MetaCore™ 7 , 8 , 9 , 10 .

The need for dedicated knowledge maps as tools for the representation of mechanisms involved in specific diseases in a holistic form gave origin to the concept of a disease map. A disease map consists of a representation of disease mechanisms illustrating the major signalling, metabolic and regulatory pathways known to be involved in a specific disorder in an interchangeable format. The major feature of the disease map is to allow comparisons among different maps using a high-quality representation in a standardised format 11 . Furthermore, it is computer-readable, user-friendly and can be transformed into mathematical models for predictive analysis and new hypothesis generation 11 . This resource can store and display multiple layers of information—from subcellular, cellular, tissue, organ to whole organism systems—besides providing customizable levels of detail that can be used in the illustration of the biological mechanisms 12 . In the past, to further understand their mechanisms and potentially find therapeutic targets, several disease maps for complex disorders have been developed such as Parkinson’s 13 , Alzheimer’s 14 , asthma 15 , several types/forms of cancer 16 and rheumatoid arthritis 17 .

Cystic Fibrosis (CF) is a life-shorting rare genetic disorder affecting 90,000 to 100,000 individuals worldwide, that results from over 2,100 variants in the CF transmembrane conductance regulator ( CFTR ) gene 18 , 19 , 20 . The CFTR protein functions as a chloride/bicarbonate channel activated through cAMP-induced phosphorylation at the apical plasma membrane (PM) of epithelial cells. This protein works in a dynamic network, interacting with multiple components and regulating a significant number of other channels 21 , 22 , 23 .This channel loss of function causes the organs in which it is expressed to be impacted. Specifically, in the lung CFTR malfunction causes severe airway dehydration and thickening of the lung mucus, leading to impaired mucociliary clearance (MCC) and, subsequently, clogging of the airways. This causes progressive loss of lung function, which is the major cause of morbidity and mortality 24 , 25 . Furthermore, it is important to note that CFTR loss of function does not occur only in CF, but also in chronic obstructive pulmonary disease (COPD)—the third main cause of death worldwide 26 —revealing the crucial role this protein plays in the airway epithelium 27 .

The most common CF-causing mutation on the CFTR gene is the c, which occurs in approximately 70% of people with CF worldwide 28 . F508del- CFTR is a misfolded unstable protein, which is mostly retained in the endoplasmic reticulum (ER) and targeted for ER-associated degradation (ERAD). However, if F508del- CFTR bypasses this pathway and is rescued to the PM, the protein still presents defective channel gating and low PM stability, being rapidly endocytosed and degraded 29 , 30 , 31 . The biogenesis and intracellular trafficking of F508del- CFTR protein has been extensively studied to better understand the pathways that contribute to and drive CF progression. Notwithstanding, and owing to the inherent complexity of the mechanisms and pathways, the CFTR dynamic network is still only partially understood 20 , 32 , 33 , 34 , 35 .

Several databases and software tools have been used in efforts to represent the known CF mechanisms through the representation of pathways covering different processes—such as MetaCore™ from Clarivate Analytics (i.e., CFTR folding and maturation), Reactome (i.e., RHO GTPases regulate CFTR trafficking), KEGG (i.e., ABC transporters—Homo sapiens) and Wikipathways (i.e., ABC-family proteins mediated transport) 6 , 8 , 36 . Notwithstanding, these pathway-based databases present CFTR data incorporated with non-related processes and proteins such as other ABC-family proteins, which although allow comparison between some members of this family, may interfere with the reading and interpretation of CFTR specific data. Furthermore, most are not detailed and/or updated with recent disease features, besides not being freely accessible, nor containing CFTR variants information or allowing comparison between a wt and mutant protein interactions, essential to understanding the progression of this disease. yFi-MAP is developed and publicly available as a complementary resource to the existing tools. Moreover, CyFi-MAP content benefits from being discussed and approved by domain experts (available on the GitHub page, section Team). The incorporation of CFTR molecular mechanisms in a single resource represented in a standardized and inter-exchangeable manner represents one of the main advantages of CyFi-MAP. This tool allows the user to: (i) follow CFTR pathways; (ii) acquire its functional context and subcellular localization through the visualization of the cell compartments represented in the map; (iii) interpret differences between wt-CFTR and F508del- CFTR where unique interactions with CFTR variant are highlighted.

With this goal in mind, in this work, we aimed to build a repository of the available CFTR -related knowledge as a disease map named CyFi-MAP. CyFi-MAP is distinctive from other CFTR databases by being the first CF disease map and aiming to archive molecular mechanisms and biological pathways reported to be relevant to CFTR in a standard and consistent way. This repository is continuously updated with careful manual annotations and is expected to expand in both molecular mechanisms and in CFTR mutations representation and to be continuously updated based on the published literature. Through the map, it is possible to visualize CFTR pathways in protein cycle processes described in a sequential network, resulting in a more global biological interpretation and a faster understanding of CF pathophysiology.

Concept and features

CyFi-MAP offers a resource that accurately (i.e., adequately confirmed in the literature) and graphically illustrates CFTR molecular pathways in an easy-to-read manner by the scientific community. Given that CF is caused exclusively by mutations in the CFTR gene that alter multiple cellular functions, the CyFi-MAP information was organized according to the CFTR life cycle, from its biogenesis to degradation, where two sub-maps were developed: (i) one representative of wt- CFTR and (ii) one of F508del- CFTR .

The main differences between these submaps are given by the representation of some of the key processes side-by-side (available on the website in the Map section), in order to facilitate submap comparison at molecular mechanism level. Additionally, a scheme is also available with CFTR traffic pathways inside the cell where the major physical alterations (e.g., traffic impairment and mucus clogging) in wild-type vs the mutant in airway epithelial cells are depicted (Fig. S1 ). An overview of the key CFTR processes/modules included in CyFi-MAP is given to inform and guide the user to the map content (Fig.  1 ).

Modules available in the CyFi-MAP. wt- CFTR (left) and F508del- CFTR (right) modules include the key processes available included in the CyFi-MAP and provide a way to focus on a specific part of CFTR life cycle. Rescued F508del- CFTR (rF508del-CFTR) pathways are highlighted in yellow, indicating the processes elicited after chemical or temperature rescue of the mutant protein (see text for details).

Pathway inclusion strategy

We included information on CFTR interactors that was confirmed in a minimum of two published references. We focused on those that studied airway epithelial cells and used methodologies that allowed the detection of physical interaction between components (such as immunoprecipitation or nuclear magnetic resonance spectroscopy (NMR)). We also captured information on confidence and accuracy to each interaction in the map. Specifically, this step resulted in 296 research papers providing physical evidence between proteins and more than 1000 papers reviewed from PubMed (for more details on map curation see “ Methods ” section).

CyFi-MAP presents features that allow retrieving information visually such as (1) different types of interactions between the entities (i.e., activation, trafficking and inhibition—more details in Fig. S2 in Supplementary material), (2) the glycosylation (i.e., form B or C) and folding status (N-glycosylation) of CFTR , (3) the proteins that bind uniquely to the F508del- CFTR protein, (4) identification of eachlife cycle steps included—additional information in Supplementary material, and (5) cell organelles specific interactions represented through images, with the entities (i.e., proteins, complexes, ions, and others) adequately located in the biological compartments, differentiating between organelle lumen and the cytosol.

CyFi-MAP navigation

Map availability.

The source of these schemes are available at the map online repository ( https://cysticfibrosismap.github.io/ ) and the CyFi-MAP can be accessed online and explored interactively via the Molecular Interaction NetwoRks VisuAlization (MINERVA) 37 .

Online and interactive navigation

In the form of an interactive diagram via the MINERVA platform (Fig.  2 ), CyFi-MAP provides the capacity to easily follow CFTR interactions starting with its folding until degradation, with every intermediate step described (see supplementary material). The MINERVA platform allows easy navigation and exploration of the CF related molecular pathways available in CyFi-MAP. The user can also zoom in and find various details about proteins of interest (such as location of their interactions with other biological elements inside a cell, information on the protein name/alternative names and identification names in several databases such as Ensembl, human gene nomenclature, UniProt etc.; the annotation information is linked to the biological resources via direct urls). The user can also filter and extract information regarding the type of interaction through the edge colour (see Fig. S2 in Supplementary Material).

CyFi-MAP Github and MINERVA platform. In the image ( A ) is possible to see the GitHub page with the objective to contextualize this work and provide access to the interactive MINERVA platform. In image ( B ) and ( C ) are represented both submaps, wt- CFTR and F508del- CFTR respectively, where it is possible to navigate and explore the pathways and interactions included in CyFi-MAP, providing generic information about the proteins and components.

Content of the CyFi-MAP

The progression of CF disease is driven by the deregulation of multiple cell processes due to the loss of CFTR function. Hence, CyFi-MAP has greatly focused on the CFTR proteostasis network, as it encompasses alterations occurring on all those processes. Along with the map it is possible to observe several numbers which represent each step on the CFTR life cycle (a brief description of these is given in supplementary material) although the order indicated is not deterministic, serving only as an indication of all the processes in which CFTR is involved. The content included in the CyFi-MAP can be divided into four mains aspects: (i) CFTR synthesis and production, (ii) maintenance at the PM in the functional state, (iii) traffic and (iv) degradation.

CFTR synthesis and production

The folding of CFTR is highly regulated in the ER before the protein is allowed to proceed along the secretory pathway to the PM. This information is represented in the Folding module at both submaps hence, it is possible to visualize it side-by-side in the GitHub link.

Beginning with thewt- CFTR , folding starts with the nascent CFTR CFTR polypeptide chain being translocated to the ER membrane (step [1]), after which the N-glycosylation occurs (step [2]), a posttranslational modification during protein synthesis in the ER that is critical for PM expression and function (Fig.  3 A) 38 , 39 , 40 .In CyFi-MAP, the different glycosylation states of wt- CFTR are represented and possible to follow. N-glycosylation starts with the addition of oligosaccharide residues (glucose3-mannose9-N-acetylglucosamine2) to CFTR (Fig.  3 A, step [2]). At this step, several chaperones and co-chaperones bind to wt- CFTR to assist with folding. The folding process includes at least four ER quality control (ERQC) checkpoints that are involved in assessing CFTR correct folding 41 .As the process consists of the subsequent trimming of glucose, they are initially identified with three glucose (G3), the first checkpoint, before binding to calnexin ( CANX ) with only one (G1), the second checkpoint (Fig.  3 A, steps [3–6]) 42 .

Representation of folding at ER. Wt- CFTR ( A ) and F508del- CFTR ( B ) are subjected to the sequential ERQC checkpoints, indicated in black. On the wt- CFTR is possible to see the four checkpoints of the CFTR protein (in blue). In the case of F508del- CFTR is possible to see a bifurcation in the pathway, where it can be targeted to degradation or rescued, where it appears red at the cytosol. It is also indicated the beginning of the sumoylation degradation pathway. The proteins are represented in beige with the orange lines indicating the movement of the protein and the black lines indicating stimulation. The white boxes containing several proteins indicate a complex. Unique interactors to F508del- CFTR are represented in yellow.

F508del- CFTR is mostly targeted to degradation in the first checkpoint (Fig.  3 B, step [2]), hence the other ERQC checkpoints are only represented at the wt- CFTR submap There, is possible to follow the second checkpoint, where the protein enters the CANX cycle for additional rounds of refolding, the third checkpoint, where specific signals when exposed lead to ER retention, or the fourth checkpoint, with the recognition of an export motif to leave ER (Fig.  3 A, step [7] and step [8]) 43 . During these checkpoints, wt- CFTR can be recognized as misfolded and move to the degradation or achieves an incomplete glycosylated state known as form B, which allows it to proceed to Golgi. A more detailed description of these checkpoints is in the degradation chapter.

Although insertion in the membrane of ER and ERQC checkpoints that lead to degradation are identified as different steps, there is evidence that co-translational folding and degradation occur.

After CFTR folding has been successfully achieved, the protein is ready to proceed along the secretory pathway to the Golgi apparatus, where its oligosaccharide structure is further modified by multiple glycosylation events generating its mature form, known as C form, which will be transported to the PM 44 . F508del- CFTR , because is highly degraded at ER, from the moment is depicted outside this organelle acquires a dark red colour, representing a rescue protein rF508del- CFTR .

Maintenance at the plasma membrane in the functional state

After delivery to the PM, wt- CFTR is regulated at multiple levels namely: (1) PM stabilization, at specific PM sites; (2) activation/channel shut down, where phosphorylation/dephosphorylation cycles activate/inactivate the channels; (3) ion channels and transporters regulation, concerning the regulation of and by other PM proteins; and (4) endocytosis, in either CCVsor caveolae vesicles.

In contrast, rF508del- CFTR is characterized by (1) PM stabilization; (2) channel shut-down; and (3) endocytosis, with consequent degradation. The reduced number of modules at PM is representative of the instability, loss of function and accelerated endocytosis that characterize this mutated protein.

In the PM stabilization module, Postsynaptic density 95, disks large, zonula occludens-1 (PDZ) domain-containing proteins are the main characters, responsible for anchoring CFTR to the PM 45 . Mechanisms such as cytoskeletal activation are represented, which enable PM anchoring and tethering of wt- CFTR to the PM (Fig.  4 A, step [11]). TherF508del- CFTR stabilization module includes a lower number of interactions with PDZ proteins and acquisition of new interactions when compared to the wt- CFTR module (Fig.  4 B, step [5]).

PM stabilization in wt- CFTR ( A ) and F508del- CFTR ( B ) submaps. Wt- CFTR represented in blue is delivered to the apical PM in the C form, as indicated at the protein, and several proteins responsible for the anchoring to the PM bind. The lines show different types of interactions, where orange lines indicate traffic, being possible to see in chloride (Cl − ) transport across the membrane in orange. The black lines indicate binding/stimulation and the red one’s inhibition. F-actin and Arp2/3 are represented as complexes, with several proteins with a grey line separating them from the rest. In blue are represented 1-phosphatidyl1D-myo-inositol-4-phosphate (PIP) which by the action of Phosphoinositide Kinase, FYVE-Type Zinc Finger Containing ( PIKFYVE ) is converted to 1-phosphatidyl-1D-myo-inositol-4,5-bisphosphate (PIP2) with a blue line depicting synthesis. Although rF508del- CFTR reaches the PM, the binding to new proteins highlighted in yellow diminishes its stabilization and anchoring therefore accelerating its endocytosis and consequent degradation.

PDZK1 ( CAP70 ) is illustrated at PM in the wt- CFTR submap to be able to potentiate the CFTR chloride channel activity by cluster two CFTR molecules (wt- CFTR submap, step [12]) 46 (Fig.  4 A, step [12]).

Gating and channel shut down depict proteins involved in CFTR activation/inactivation (Fig.  5 ). CFTR is regulated through cAMP and phosphorylated by proteins such as protein kinase A ( PRKACA ) and Protein Kinase C Epsilon ( PRKCE ) 47 , 48 (Fig.  5 step [13]).This includes proteins such as β2-adrenergic receptor ( ADRB2 ) 49 , A2B receptor ( ADORAB2 ) 50 and adenylyl cyclase I ( ADCY1 ) 51 which participate in cAMP/PKA signalling therefore activating wt- CFTR and the transport of chloride.

PM pathways in wt-CFTR submap. ( A ) Gating where is possible to see adenosine 3,5-cyclic monophosphate (cAMP) role in Protein Kinase CAMP-Activated Catalytic Subunit Alpha (PRKACA) phosphorylation that leads to CFTR activation and transport of chloride across the membrane. Other proteins are involved such as Adrenoceptor Beta 2 (ADRB2), Protein kinase CK2 (formerly known as casein kinase II), Adenosine A2b Receptor (ADORA2B) and Protein Kinase C Epsilon (PRKCE). ( B ) Shut-down with the action of Lysophosphatidic Acid Receptor 2 (LPAR2), serine/threonine protein kinase complex (AMPK), Phospholipase C beta ½ (PLC beta ½) and Protein phosphatase 2 (PP2A) where the chloride transport is inhibited. ( C ) Transporters and Ion channel regulation, in which the proteins that regulate and are regulated by CFTR at the PM are represented with the interactions detected.

The Channel shut-down present in both submaps includes dephosphorylation of CFTR and the proteins involved in its triggering—such as protein phosphatases, receptors, phospholipases, and others 52 , 53 .

The transporters and Ion channel regulation module is only present at the wt- CFTR submap, and besides directly binding to wt- CFTR also allows to visualize PDZ proteins role as intermediates between them, maintaining the proteins in close proximity and leading to changes in their respective functions (Fig.  5 B/C, steps [16], [17] and [18]). Proteins such as Solute Carrier Family 26 Member 3 ( SLC26A3 , also known as DRA ), Solute Carrier Family 26 Member 6 ( SLC26A6 , also known as PAT1 ), Anoctamin 1 ( ANO1 , also known as TMEM16A ) and epithelial sodium channel ( ENaC ), were included in this module 54 , 55 , 56 .

In the secretory pathway, traffic is essential for all processes since folding/processing and function to degradation. CFTR traffic processes start by Coat Protein complex II (COPII) vesicles, responsible for its transport between ER and Golgi, from where wt- and rF508del- CFTR reach the PM (wt- CFTR submap, step [9]) 43 . rF508del- CFTR traffic between ER and Golgi is depicted through the COPII vesicles module with the same mechanisms as wt- CFTR (rF508del- CFTR submap, step [3]) 33 , 34 , 57 . Although this information is only supported by high-throughput research articles and with only one paper for each interaction with the mutated protein, it was one of the exceptional cases that were selected to CyFi-MAP given that is explained by the functional context provided in wt- CFTR . A list with the exceptions isin the additional information section of the supplementary material.

From this point, the endocytosis module depicts a wt- CFTR association with endocytic adaptors undergoing CCV-mediated and caveolae endocytosis (Fig.  6 A, steps [19, 20] respectively), whereas in rF508del- CFTR only the last one is depicted (Fig.  6 B, step [9]).

wt-CFTR ( A ) and rF508del-CFTR ( B ) endocytosis mechanisms of internalization from the PM. In the image A two types of internalization of wt-CFTR are present, through clathrin coated vesicles [19] and through caveolae vesicles [20]. As is represented, in the first, several proteins are involved, since the clathrin triskelion complex to cytoskeletal F-actin-MYO6 complex and other proteins assisting the process. In rF508del-CFTR, only the complex Caveolin 1 (CAV1)/ Caveolin 2 (CAV2) was found in this protein endocytosis [9], with assistance of Flotillin 2 (FLOT2).

CFTR is endocytosed and arrives at the sorting endosome from which wt- CFTR moves back to the PM, either directly—Recycling Module (Fig.  7 A, step [22], [23])—or through the Golgi—Golgi Module (Fig.  7 A, step [24])—or to degradation—Degradation Module (Fig.  7 A, step [26]). In these processes, several proteins of the Rab family are represented as they are essential for CFTR traffic 58 . In the case of rF508del- CFTR , at the sorting endosome, it is sent to degradation (Fig.  7 B, step [10]).

Sorting endosome in wt-CFTR ( A ) and F508del-CFTR ( B ) submaps. In the image ( A ), is possible to see wt-CFTR arriving at the sorting endosome [21] and the possible pathways to follow, either recycling [22], [23] and [24], or degradation [25] in orange lines. Bellow in image ( B ), rF508del-CFTR in red arrives at the endosome [10] in a complex with ubiquitin (represented with Ub) where several proteins target it to degradation directly, representing the low/absence of recycling of this protein. The proteins highlighted in yellow are unique to F508del-CFTR.

Degradation

During folding and processing, several quality control proteins target misfolded CFTR to degradation, therefore, in each of the organelles—ER and Golgi—there is a module called Degradation.

The four ERQC checkpoints can lead towt- CFTR degradationat the proteasome. In the case of F508del- CFTR , only the first ERQC checkpoint, called chaperone trap, is represented with the binding of four chaperones that strongly attach to the misfolded protein (i.e., Heat Shock Protein Family A ( Hsp70 ) Member 4 ( HSPA4 ), Heat Shock Protein Family A ( Hsp70 ) Member 8 ( HSPA8 ), DnaJ Heat Shock Protein Family ( Hsp40 ) Member A1 ( DNAJA1 ) and DnaJ Heat Shock Protein Family ( Hsp40 ) Member B1 ( DNAJB1 )) (F508del- CFTR submap, step [2A]) 42 , 59 . Furthermore, the F508del- CFTR presents a degradation pathway in contrast to wt- CFTR : the sumoylation (Fig.  3 B) 60 . At Golgi, CFTR degradation depicts its targeting to degradation via the lysosome. In the PM, the rF508del- CFTR is targeted to degradation by ubiquitinationas a consequence of the destabilization of the protein (F508del- CFTR submap, steps [7], [8] and [9]). There, proteins involved in its degradation at ER attach to itin the PM (i.e., HSPA8 , STIP1 Homology And U-Box Containing Protein 1 ( STUB1/CHIP ), Stress-Induced Phosphoprotein 1 ( STIP1 ) and others) 61 . At the sorting endosome, in both cells, the protein is targeted to degradation by lysosome/proteasome although with different interactors assisting (i.e., Tumor Susceptibility 101 ( TSG101 ), Hepatocyte Growth Factor-Regulated Tyrosine Kinase Substrate ( HGS ), Charged Multivesicular Body Protein 4B ( CHMP4B ), and others) bind uniquely to rF508del- CFTR (Fig.  7 B, step [10]) 62 .

CF is the most common life-threatening autosomal recessive disease in the Caucasian population 63 . Caused by an absent/dysfunctional CFTR channel that leads to an impairing balance of ions across the membrane, CF is characterized by affecting several organs, especially the lung 20 . CFTR seems to be involved not only in the transport of ions but also in the regulation of other channels, working in a dynamic network that modulates its activity 30 . There have now been more than 30 years of scientific discoveries with new milestones achieved every year in our understanding of intracellular interactions after CFTR loss of function that control the progression of this disease. The knowledge obtained over this research has enabled diagnosis and discovery of therapies that increased life expectancy 20 . Yet no final curative treatment has yet been developed for CF disease 64 , 65 , 66 .

The increasing data available on public databases lead to the improvement of tools to filter and extract relevant knowledge required for the discovery of therapeutic targets. With this need, disease maps were developed as a multilayer-readable network that allows representing increasingly complex and extensive information in an easily updatable manner 11 . The visual representation of CFTR key processes in a cell can act as a powerful tool to understand and share knowledge. In this work, we built the CyFi-MAP, a manually curated disease map of CFTR -related available information, as a resource that permits a deeper understanding and interpretation of the disease mechanisms. CyFi-MAP development is motivated by the absence of resources differentiating between wt- CFTR and its variants and has the objective of concentrating on a single free access resource the CF major hallmarks, representing the data scattered across different platforms/research papers in form of pathways and interactions. This tool was designed to be useful for CF scientists as a reference source to analyse previous knowledge and assist in the whole-organism level perspective as well.

CyFi-MAP included data: wt-CFTR versus F508del-CFTR submap

CFTR life cycle can be impacted at several steps, with the most common mutation, F508del- CFTR , subjected to ER retention and degradation when not rescued through low temperature or chemical compounds. In fact, rF508del- CFTR is characterized by barely reaching the PM under physiological conditions, presenting reduced chloride/bicarbonate transport after being rescued, as well as enhanced endocytosis and degradation, with consequent dysregulation of other PM proteins 53 , 67 . Due to F508del- CFTR incapability to achieve a competent intermediate, the protein becomes trapped in the first steps of ERQC and is mostly targeted for degradation as soon as the polypeptide is synthesized—indicated in CyFi-MAP by the absence of the others ERQC processes 42 , 43 , 68 , 69 . Furthermore, the location and function of CFTR at PM are affected in the case of rF508del- CFTR , visible by the lack of interactions and by the absence of key processes when compared with wt- CFTR in CyFi-MAP. rF508del- CFTR in CyFi-MAP lacks proteins involved in the activation of the channel and regulation of other channels and transporters at the PM. This is a consequence of its instability where unique proteins that interact with rF508del- CFTR , such as Calpain 1 ( CAPN1 ) and Calpain 2 ( CAPN2 ), play a role in its destabilization by impairing its binding to PDZ anchor proteins. Besides that, ubiquitination and subsequent targeting to endocytosis and degradation involve additional proteins such as Ring Finger And FYVE Like Domain Containing E3 Ubiquitin Protein Ligase ( RFFL ), TSG101 , HGS , CHMP4B and others that prevent the recycling to the PM 62 , 70 , 71 .

Some proteins (e.g. SLC9A3R1 , PDZK1 and others) involved in several roles along the CFTR life cycle (including stabilization, anchoring and function at PM) are present on the map more than once. PDZ proteins are essential elements also that act as intermediates that connect other channels with CFTR . Additionally, PDZK1 is found binding to two CFTR proteins, maintaining CFTR proteins functioning in close association.

Besides the conventional trafficking, unconventional secretion pathways have been described for membrane proteins such as CFTR and usually involve bypassing the Golgi, a route that was identified through blocking the conventional Golgi-mediated exocytic pathway 72 , 73 . Pathways such as this, were not included in this version of CyFi-MAP as they are unlikely to represent the cell in its physiological state. Notwithstanding, it can be helpful to provide a broader view of the possible interactions and they may appear in a future version of CyFi-MAP with less stringent criteria for inclusion.

CyFi-MAP expansion and future work

Given the fact that new data are generated continuously, and CF aspects are yet to be included, the CyFi-Map is constantly developed with support from the community and funding agencies.CyFi-MAP benefits from major features of the MINERVA platform via its online distribution: comments and suggestions from users with regard to changes in the map content (addition, removal, update) can be analysed directly by curators and addressed potentially in the map after further refinements. In this way, users can promote active discussions and knowledge exchange to build an increasingly accurate and continuously updated CF disease map. Notwithstanding, of specific interest as a future direction, is to include in CyFi-MAP the specific steps in CFTR processes targeted by compounds, depicting this way the specific target/mechanism where each of them acts. Additionally, we anticipate including a diagram focusing on the process description layer is anticipated of the CF molecular processes (e.g., the N-glycosylation of CFTR in ER) in order to provide a deeper understanding of such interactions. Furthermore, the creation of submaps representing other CFTR mutations would be relevant to study the molecular mechanisms affected.

Altogether, CyFi-MAP represents the first stable milestone into a robust and reliable CF knowledge base integrating information on key pathways involved in molecular pathophysiological CF mechanisms, based on curated literature and expert-domain-approval. CyFi-MAP offers an integrative and system-level view of CFTR knowledge. CyFi-MAP may support the interpretation of CF progression and may facilitate the development of novel therapeutic targets and strategies. In fact, a better understanding of CFTR mechanisms can not only assist in the design of improved therapies for CF but also identify factors that work in other lung diseases, such as COPD or disseminated bronchiectasis. The next steps can also involve the integration of the knowledge acquired using CyFi-MAP as a basis for mathematical models to generate new data through network inference, modelling and creation of new hypotheses to be tested.

CyFi-MAP construction

The development of the CyFi-MAP follows the disease map development protocol, using primarily Kondratova et al. and Mazein et al. 11 , 74 . Specifically, three main steps entail the construction of CyFi-CFTR (Fig.  8 ):

The first step consisted of searching relevant CFTR -related information, selecting a total of 297 research papers and more than 1000 reviewed articles. A complete list of publications consulted for the CyFi-MAP development is available in https://cysticfibrosismap.github.io/ . The CF disease hallmarks were obtained from peer-reviewed research papers, domain experts’ suggestions and advice, previously documented and validated pathways, and curated up-to-date databases (including Reactome 6 /KEGG 4 /MetaCore from Clarivate Analytics. Please see Content curation subsection for details). This task also involved the analysis of the collected pool of data, followed by the curation of the most relevant CFTR -related knowledge (Fig.  9 ).

The second step comprised the effective diagram building, assuring the correct level of detail and the most appropriate and aesthetically output, to guarantee that the resulting map is as readable and user-friendly as possible. The biological mechanisms representation follows the Systems Biology Graphical Notation (SBGN) notation and was built in the yEd Graph Editor using the SBGN Palette ( https://yed.yworks.com/support/manual/layout_sbgn.html ). The yEd Graph Editor is a freely available graph editor providing functionality to manage large-scale graphs including: (i) features that considerably facilitate the diagram drawing process such as friendly user interface, drawing guides, zooming on the diagram and easy application of specific aesthetics (e.g. same colour for nodes/ edges, curved connectors) to an individual or multiple elements; (ii) algorithms for automatic layout (details on using yEd to automatically layout SBGN-related diagrams are given in e.g. 75 ). The yEd Editor also incorporates the SBGN Palette that permits the direct representation of the SBGN-specific elements into the yEd inner GraphML format. After the CyFi-MAP was developed in yEd, we converted it into the SBGN standard format by using the ySBGN converter, (a bi-directional converter between the SBGN and yEd GraphML formats, available at https://github.com/sbgn/ySBGN ). Further, the CyFi-Map SBGN diagram was loaded to the MINERVA online platform. The organelle images (developed manually and expert-revised) aim to facilitate visualisation of the mechanisms at the top level; thus, special attention was given to the localization of the interactions in each organelle.

The third step in the construction of the CyFi-MAP was the map exploration via the MINERVA platform 37 . In a first approach, the construction focused on the creation of small organelle-specific maps, illustrating CFTR -relevant processes on those locations. The maps included CFTR interactions covering its intracellular and intraorganellar traffic. Later, these were improved upon by the addition of other, more widespread CFTR processes and pathways, which allowed a more effective integration of the existing data. The resulting cell-wide map is expected to be continuously evolving with user input and consistent expert curation. The map is available through the web platform MINERVA that provides interactive and exploratory features.

Workflow of CyFi-MAP construction. Starting by the research and curation of data contained in general and CFTR-specific databases and peer-reviewed research papers, CyFi-MAP was constructed based on domain experts’ suggestions and users’ comments. The organization of the curated data required the selection of a general format to be used throughout the platform, which determined the map assembly and content visualization. After the map was built, domain experts were once again consulted, in order to review and provide feedback on the accuracy of the representation of the disease mechanisms, as well as the usability of the platform.

CyFi-MAP curation process. The curation process developed presents 5 levels. 1st level filter the data to studies with proteins that interact directly with CFTR, meaning that only experimental techniques which confirmed a direct interaction were considered such as e.g. Immunoprecipitation, Surface Plasmon Resonance (SRP), and others. The 2nd level is related to the type of cell culture used in these experiments, focusing on human airway epithelial cells, although other cell types were included when described on review publications. In both levels, if the studie do not agree with the criteria is rejected. The 3rd level consist on finding the location of the interaction inside the cell (e.g. ER, Golgi, cytosol, PM) followed by the type of interaction (binding or inhibition). The 4th level confers confidence to the interaction, consisting on the search for publications that support the information. The 5th level confirms information related to the protein after being selected (e.g. does it belong to a complex? Which pathway does it belong to?). The protein is manually added to the yEd Graph Editor used to built CyFi-MAP with the name accordingly to the HGNC nomenclature.

The current version of CyFi-MAP has been manually curated by CF domain experts. To ensure a continuous updating of this resource, both regular expert verification of new information, as well as regular user input are deemed essential to achieve an accurate representation of current CF data. Constant feedback from cell biologists, biochemists, physiologists and bioinformaticians contributed to a comprehensive representation of the various layers of information.

Inside CyFi-MAP, each process comprises pathways that include proteins (as individual entities or as complexes) and different types of chemical species (ions and lipids) interacting among themselves. Nodes represent entities (i.e., proteins, ions or complexes) and edge colours correspond to processes (i.e., activation, inhibition, synthesis, or in some cases movement of entities inside the cell).

CyFi-MAP currently comprises 618 nodes and 420 edges, with 426 nodes and 307 edges in wt- CFTR and 216 nodes and 117 edges in F508del- CFTR . In total, entities presented at both submaps are classified into 193proteins, 25complexes, 5ions and 5 simple molecules in wt- CFTR and 98 proteins, 12 complexes, 1 ion, and 2 simple molecules in F508del- CFTR .

Content curation

The data used in CyFi-MAP was obtained by manual human search, curation and validation with domain experts in three main sources:

Pathway databases

The curation process started by reviewing previous attempts to summarize CF information in these signalling networks. Pathways from MetaCore (Clarivate Analytics), Reactome and KEGG were reviewed in order to analyse the pathway availability for the CF disease 6 , 76 . Major CF-related pathways were retrieved from these databases and confirmed in the literature for their accuracy.

The main hallmarks of CF were extensively searched in PubMed. As CFTR is the protein that plays a central role in the map, direct interactions with it were very carefully selected, following strict criteria. Considering that the lung is the most affected organ, the focus was on human airway epithelial cells studies. The massive number of CFTR reviewed articles and studies available were analysed and selected as particularly relevant; results obtained from essays with other relevant cell types (such as intestinal epithelial cells) were also included when validated by review papers, meaning they are accepted by the scientific research community. Priority in the selection process was given to the molecular mechanisms involving protein folding and traffic, as these are the main processes that are impaired in F508del- CFTR . Most studies regarding this mutant’s behaviour at the PM resulted from experiments on rF508del- CFTR , either chemically or temperature-dependent. The inclusion of information from proteomic studies was dependent on the functional context provided by already documented interactions. Although each direct interaction with CFTR on CyFi-MAP was confirmed in a minimum of two papers, some exceptions may apply such as information retrieved from recent articles (2018on) and protein interactions that were part of well-characterized pathways involved in CF referred in more than one peer-review research paper. An example of the last is the interaction between STX3 with CFTR , as only one research paper was found with a physical interaction although is mentioned in peer-review articles 77 and hence included as one interaction accepted by the scientific community.

Among the web resources used for data gathering, the most significant were GeneCards 78 , Stringdb 79 , Biogrid 80 , UniProt 81 and HGNC (HUGO Gene Nomenclature Comittee) 82 which were used to confirm the correct names of proteins/genes, their known function and their interactors. For each protein, the name was checked in HGNC. UniProt and Genecards were used to search for alternative names for the same protein so as to find the correct HGNC designation 78 , 81 , 82 . Often, although a protein complex is known to interact/participate in a CFTR process, the specific proteins that constitute that complex are not described in the original literature report. Accordingly, proteins reported in the literature to interact with CFTR as part of larger complexes were searched for in databases to find the protein components of the complex.

During this step, name disambiguation must be considered in order to find all data related to that protein and also to not repeat proteins. For instance, by looking for syntaxin 5, names such as Syn5 and STX5 are also available for the same protein. The same happens for Golgi Associated PDZ And Coiled-Coil Motif Containing protein, known as GOPC although other names such as CAL and FIG are referred to on research papers and used to retrieve as much information as possible.

Diagram building

CyFi-MAP was built using yEd graph editor ( https://www.yworks.com/ ) using the SBGN Palette, and the data was represented based on the SBGN standard 12 . This notation provides a knowledge representation language used in the illustration of molecular pathways and protein interactions as the standard notation for disease maps 11 . Presenting three languages that provide different types of knowledge illustration allow to adapt on the level of detail intended to be highlighted on the map, including the following layers: Activity Flow to depict interactions with process direction, Process Description detailed specific mechanisms, and Entity Relationships which describe the mechanisms without a sequential process 12 .

CyFi-MAP was implemented following the SBGN Activity Flow, in order to provide a compact, sequential and easy-read format or involving signalling pathways. This language is useful to represent the flow of information in biological sequences/pathways in a way that the information can still be captured for underlying mechanisms of unknown influence 12 .

Each subcellular organelle (ER, Golgi, endosome, etc.) were drawn manually and added to the map as a background image for graphical representation of the different subcellular compartments. Additionally, each interaction provides information on itself. Depending on the selected edge different types of interactions can be found on CyFi-MAP, namely (see Fig. S1 on supplementary material for more details): (1) activation, representing a normal binding, (2) synthesis, when an altered product is released, (3) trafficking, representing movement inside the map, and (4) inhibition, when the interaction inhibits a function.

Map exploration via the MINERVA platform

CyFi-MAP diagrams are available in the platform MINERVA accessible through GitHub ( https://cysticfibrosismap.github.io/ ). The project description and key processes (shown side-by-side, represented through images to allow the comparison between wt- CFTR and F508del- CFTR submaps on the CyFi-MAP) are given on the website. Starting at the cell level, it is possible to identify the main differences between the submaps (Fig.  1 ). This view is relevant in order to compare the cells in presence of the two proteins since wt- CFTR is transported across the secretory pathway to the PM and endocytosed to be either degraded or recycled back to the PM, whereas most F508del- CFTR is retained at the ER from where it is sent for degradation. This impairment leads to the so-called ‘CF pathogenesis cascade’, which does not occur for wt- CFTR . In the cell level view, it is possible to observe these features, allowing the extraction of relevant knowledge.

Additionally, wt- CFTR and F508del- CFTR submaps were divided into modules, each representing key processes of its life cycle in order to guide the navigation through the map. To compare information in both submaps, images placing the modules folding, stabilization and sorting side-by-side are available. Mutation-specific proteins are highlighted in a different colour to emphasize differences between wild-type and mutated phenotypes.

The interactive web platform MINERVA allows accessing an interactive CyFi-MAP to navigate and explore its molecular networks. This tool provides automated content annotation, direct feedback to content curators and SBGN-compliant format 83 . Navigation in CyFi-MAP is similar to navigation in Google Maps being possible to through MINERVA search elements that are highlighted by markers and also retrieve additional information on each element on the panel on the left side, presenting several identifications names using HGNC and UniProt as sources.

The zoom feature allows a high-level view of the intracellular organelles and a close view inside each providing easier access to the complex and extensive information it contains. Every CFTR interaction represented in the CyFi-MAP is validated by PubMed references. The user can curate the data by commenting, given they provide the respective reference as well.

All suggestions will be analysed by curators and CF domain experts to maintain CyFi-MAP quality and accuracy.

The user can contribute to CyFi-MAP by adding comments with questions, corrections or additions to the map. These will be visible to other users and developers. To add a comment to CyFi-MAP during navigation, right-click on the specific location and choose to add a comment. It is possible to choose a specific identity to link the comment, such as the protein or reaction, or to remain ‘general’, which will link the comment to the location the user chooses. The remaining fields allow the user to fill in the name and email in order to facilitate communication with the developers and to clarify any questions that may emerge. Last, there will be a box where comments can be added (Fig. S2 ). Any supporting information provided will be helpful to incorporate the changes into the map. After sending the comment, it will not be possible to correct it and it will be visible on the map publicly. Details on adding user’s comments in the underlaying MINERVA platform are given at https://minerva.pages.uni.lu/doc/user_manual/v15.0/index/#add-comment .

The CyFi-MAP allow exploring the map with and without seeing the comments provided by the users by clicking on the checkbox Comments above in the map toolbar. These comments will allow the map users to benefit from the domain knowledge and expertise of researchers and to collect valuable information for the research community. All suggestions will be analysed by curators and CF domain experts in agreement with the pre-established curation process to maintain CyFi-MAP quality and accuracy.

Data availability

The CyFi-MAP is available at https://cysticfibrosismap.github.io/ .

Micheel, C. M., Nass, S. J., Omenn, G. S. & Policy, H. S. Evolution of translational omics lessons learned and the path forward. Evolution https://doi.org/10.17226/13297 (2012).

Article   Google Scholar  

Saqi, M. et al. Medicine: The future of medical genomics, healthcare, and wellness. In Methods in Molecular Biology (2016). https://doi.org/10.1007/978-1-4939-3283-2_3

Auffray, C., Chen, Z. & Hood, L. Systems medicine: The future of medical genomics and healthcare. Genome Med. https://doi.org/10.1186/gm2 (2009).

Article   PubMed   PubMed Central   Google Scholar  

Ogata, H. et al. KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Res. https://doi.org/10.1093/nar/27.1.29 (1999).

Pico, A. R. et al. WikiPathways: Pathway editing for the people. PLoS Biol. https://doi.org/10.1371/journal.pbio.0060184 (2008).

Fabregat, A. et al. The reactome pathway knowledgebase. Nucleic Acids Res. https://doi.org/10.1093/nar/gkx1132 (2018).

Article   PubMed   Google Scholar  

Kutmon, M. et al. PathVisio 3: An extendable pathway analysis toolbox. PLoS Comput. Biol. https://doi.org/10.1371/journal.pcbi.1004085 (2015).

Kanehisa, M., Sato, Y., Furumichi, M., Morishima, K. & Tanabe, M. New approach for understanding genome variations in KEGG. Nucleic Acids Res. https://doi.org/10.1093/nar/gky962 (2019).

Cirillo, E., Parnell, L. D. & Evelo, C. T. A review of pathway-based analysis tools that visualize genetic variants. Front. Genet. https://doi.org/10.3389/fgene.2017.00174 (2017).

Friend, S. H. & Norman, T. C. Metcalfe’s law and the biology information commons. Nat. Biotechnol. https://doi.org/10.1038/nbt.2555 (2013).

Mazein, A. et al. Systems medicine disease maps: Community-driven comprehensive representation of disease mechanisms. npj Syst. Biol. Appl. https://doi.org/10.1038/s41540-018-0059-y (2018).

Novère, N. L. et al. The systems biology graphical notation. Nat. Biotechnol. 27 , 735–742 (2009).

Fujita, K. A. et al. Integrating pathways of Parkinson’s disease in a molecular interaction map. Mol. Neurobiol. https://doi.org/10.1007/s12035-013-8489-4 (2014).

Mizuno, S. et al. AlzPathway: A comprehensive map of signaling pathways of Alzheimer’s disease. BMC Syst. Biol. https://doi.org/10.1186/1752-0509-6-52 (2012).

Mazein, A. et al. AsthmaMap: An expert-driven computational representation of disease mechanisms. Clin. Exp. Allergy https://doi.org/10.1111/cea.13211 (2018).

Bonnet, E. et al. NaviCell Web Service for network-based data visualization. Nucleic Acids Res. https://doi.org/10.1093/nar/gkv450 (2015).

Singh, V. et al. RA-map: Building a state-of-the-art interactive knowledge base for rheumatoid arthritis. Database J. Biol. Databases Curation https://doi.org/10.1093/database/baaa017 (2020).

Cystic Fibrosis Mutation Database. version October 2020; last update 25 April 2011 . http://www.genet.sickkids.on.ca

Riordan, J. R. et al. Identification of the cystic fibrosis gene: Cloning and characterization of complementary DNA. Science https://doi.org/10.1126/science.2475911 (1989).

Lopes-Pacheco, M. CFTR modulators: The changing face of cystic fibrosis in the era of precision medicine. Front. Pharmacol. https://doi.org/10.3389/fphar.2019.01662 (2020).

Bear, C. E. et al. Purification and functional reconstitution of the cystic fibrosis transmembrane conductance regulator (CFTR). Cell https://doi.org/10.1016/0092-8674(92)90155-6 (1992).

Jia, Y., Mathews, C. J. & Hanrahan, J. W. Phosphorylation by protein kinase C is required for acute activation of cystic fibrosis transmembrane conductance regulator by protein kinase A. J. Biol. Chem. https://doi.org/10.1074/jbc.272.8.4978 (1997).

Schultz, B. D., Singh, A. K., Devor, D. C. & Bridges, R. J. Pharmacology of CFTR chloride channel activity. Physiol. Rev. https://doi.org/10.1152/physrev.1999.79.1.S109 (1999).

Haq, I. J., Gray, M. A., Garnett, J. P., Ward, C. & Brodlie, M. Airway surface liquid homeostasis in cystic fibrosis: Pathophysiology and therapeutic targets. Thorax https://doi.org/10.1136/thoraxjnl-2015-207588 (2016).

Verkman, A. S., Song, Y. & Thiagarajah, J. R. Role of airway surface liquid and submucosal glands in cystic fibrosis lung disease. Am. J. Phys. Cell Physiol. https://doi.org/10.1152/ajpcell.00417.2002 (2003).

World Health Organization (WHO). The top 10 causes of death. https://www.who.int/news-room/fact-sheets/detail/the-top-10-causes-of-death

Gentzsch, M. et al. Endocytic trafficking routes of wild type and ΔF508 cystic fibrosis transmembrane conductance regulator. Mol. Biol. Cell https://doi.org/10.1091/mbc.E04-03-0176 (2004).

Sosnay, P. R. et al. Defining the disease liability of variants in the cystic fibrosis transmembrane conductance regulator gene. Nat. Genet. https://doi.org/10.1038/ng.2745 (2013).

Farinha, C. M. et al. Biochemical methods to assess CFTR expression and membrane localization. J. Cyst. Fibros. https://doi.org/10.1016/j.jcf.2004.05.017 (2004).

Amaral, M. D. & Balch, W. E. Hallmarks of therapeutic management of the cystic fibrosis functional landscape. J. Cyst. Fibros. https://doi.org/10.1016/j.jcf.2015.09.006 (2015).

Cheng, S. H. et al. Defective intracellular transport and processing of CFTR is the molecular basis of most cystic fibrosis. Cell https://doi.org/10.1016/0092-8674(90)90148-8 (1990).

Wang, X. et al. Hsp90 cochaperone Aha1 downregulation rescues misfolding of CFTR in cystic fibrosis. Cell 127 , 803–815 (2006).

Article   CAS   Google Scholar  

Pankow, S. et al. Δf508 CFTR interactome remodelling promotes rescue of cystic fibrosis. Nature https://doi.org/10.1038/nature15729 (2015).

Reilly, R. et al. Targeting the PI3K/Akt/mTOR signalling pathway in cystic fibrosis. Sci. Rep. https://doi.org/10.1038/s41598-017-06588-z (2017).

Pollard, H. B., Ji, X. D., Jozwik, C. & Jacobowitz, D. M. High abundance protein profiling of cystic fibrosis lung epithelial cells. Proteomics https://doi.org/10.1002/pmic.200401120 (2005).

Slenter, D. N. et al. WikiPathways: A multifaceted pathway database bridging metabolomics to other omics research. Nucleic Acids Res. https://doi.org/10.1093/nar/gkx1064 (2018).

Gawron, P. et al. MINERVA—A platform for visualization and curation of molecular interaction networks. npj Syst. Biol. Appl. https://doi.org/10.1038/npjsba.2016.20 (2016).

Sadlish, H. & Skach, W. R. Biogenesis of CFTR and other polytopic membrane proteins: New roles for the ribosome-translocon complex. J. Membr. Biol. https://doi.org/10.1007/s00232-004-0715-6 (2004).

Chen, M. & Zhang, J. T. Membrane insertion, processing, and topology of cystic fibrosis transmembrane conductance regulator (CFTR) in microsomal membranes. Mol. Membr. Biol. https://doi.org/10.3109/09687689609160572 (1996).

Kim, S. J. & Skach, W. R. Mechanisms of CFTR folding at the endoplasmic reticulum. Front. Pharmacol. 3 , 1–11 (2012).

Google Scholar  

Younger, J. M. et al. Sequential quality-control checkpoints triage misfolded cystic fibrosis transmembrane conductance regulator. Cell https://doi.org/10.1016/j.cell.2006.06.041 (2006).

Coppinger, J. A. et al. A chaperone trap contributes to the onset of cystic fibrosis. PLoS ONE https://doi.org/10.1371/journal.pone.0037682 (2012).

Wang, X. et al. COPII-dependent export of cystic fibrosis transmembrane conductance regulator from the ER uses di-acidic exit code. J. Cell Biol. https://doi.org/10.1083/jcb.200401035 (2004).

Amaral, M. D., Farinha, C. M., Matos, P. & Botelho, H. M. Investigating alternative transport of integral plasma membrane proteins from the ER to the golgi: Lessons from the cystic fibrosis transmembrane conductance regulator (CFTR). Methods Mol. Biol. https://doi.org/10.1007/978-1-4939-3804-9_7 (2016).

Guggino, W. B. The cystic fibrosis transmembrane regulator forms macromolecular complexes with PDZ domain scaffold proteins. Proc. Am. Thorac. Soc. https://doi.org/10.1513/pats.2306011 (2004).

Wang, S., Yue, H., Derin, R. B., Guggino, W. B. & Li, M. Accessory protein facilitated CFTR-CFTR interaction, a molecular mechanism to potentiate the chloride channel activity. Cell https://doi.org/10.1016/S0092-8674(00)00096-9 (2000).

Chappe, V. et al. Phosphorylation of protein kinase C sites in NBD1 and the R domain control CFTR channel activation by PKA. J. Physiol. https://doi.org/10.1113/jphysiol.2002.035790 (2003).

Picciotto, M. R., Cohn, J. A., Bertuzzi, G., Greengard, P. & Nairn, A. C. Phosphorylation of the cystic fibrosis transmembrane conductance regulator. J. Biol. Chem. https://doi.org/10.1016/s0021-9258(18)42339-3 (1992).

Naren, A. P. et al. A macromolecular complex of β2 adrenergic receptor, CFTR, and ezrin/radixin/moesin-binding phosphoprotein 50 is regulated by PKA. Proc. Natl. Acad. Sci. U.S.A. https://doi.org/10.1073/pnas.0135434100 (2003).

Watson, M. J. et al. Defective adenosine-stimulated cAMP production in cystic fibrosis airway epithelia: a novel role for CFTR in cell signaling. FASEB J. https://doi.org/10.1096/fj.11-186080 (2011).

Namkung, W., Finkbeiner, W. E. & Verkman, A. S. CFTR-adenylyl cyclase I association responsible for UTP activation of CFTR in well-differentiated primary human bronchial cell cultures. Mol. Biol. Cell https://doi.org/10.1091/mbc.E09-12-1004 (2010).

Thelin, W. R. et al. The cystic fibrosis transmembrane conductance regulator is regulated by a direct interaction with the protein phosphatase 2A. J. Biol. Chem. https://doi.org/10.1074/jbc.M507308200 (2005).

Zhang, W., Zhang, Z., Zhang, Y. & Naren, A. P. CFTR-NHERF2-LPA2 complex in the airway and gut epithelia. Int. J. Mol. Sci. https://doi.org/10.3390/ijms18091896 (2017).

Bertrand, C. A. et al. The CFTR trafficking mutation F508del inhibits the constitutive activity of SLC26A9. Am. J. Physiol. Lung Cell. Mol. Physiol. https://doi.org/10.1152/ajplung.00178.2016 (2017).

Benedetto, R. et al. Epithelial chloride transport by CFTR requires TMEM16A. Sci. Rep. https://doi.org/10.1038/s41598-017-10910-0 (2017).

Mohler, P. J. et al. Yes-associated protein 65 localizes p62(c-Yes) to the apical compartment of airway epithelia by association with EBP50. J. Cell Biol. https://doi.org/10.1083/jcb.147.4.879 (1999).

Canato, S. et al. Proteomic interaction profiling reveals KIFC1 as a factor involved in early targeting of F508del-CFTR to degradation. Cell. Mol. Life Sci. https://doi.org/10.1007/s00018-018-2896-7 (2018).

Swiatecka-Urban, A. et al. Myosin Vb is required for trafficking of the cystic fibrosis transmembrane conductance regulator in Rab11a-specific apical recycling endosomes in polarized human airway epithelial cells. J. Biol. Chem. 282 , 23725–23736 (2007).

McClure, M. L., Barnes, S., Brodsky, J. L. & Sorscher, E. J. Trafficking and function of the cystic fibrosis transmembrane conductance regulator: A complex network of posttranslational modifications. Am. J. Physiol. Lung Cell. Mol. Physiol. https://doi.org/10.1152/ajplung.00431.2015 (2016).

Ahner, A., Gong, X. & Frizzell, R. A. Cystic fibrosis transmembrane conductance regulator degradation: Cross-talk between the ubiquitylation and SUMOylation pathways. FEBS J. https://doi.org/10.1111/febs.12415 (2013).

Okiyoneda, T. et al. Peripheral protein quality control removes unfolded CFTR from the plasma membrane. Science https://doi.org/10.1126/science.1191542 (2010).

Sharma, M. et al. Misfolding diverts CFTR from recycling to degradation: Quality control at early endosomes. J. Cell Biol. 164 , 923–933 (2004).

De Boeck, K. Cystic fibrosis in the year 2020: A disease with a new face. Acta Paediatrica Int. J. Paediatr. https://doi.org/10.1111/apa.15155 (2020).

Estabrooks, S. & Brodsky, J. L. Regulation of CFTR biogenesis by the proteostatic network and pharmacological modulators. Int. J. Mol. Sci. 21 , 452 (2020).

Perrem, L. & Ratjen, F. Anti-inflammatories and mucociliary clearance therapies in the age of CFTR modulators. Pediatr. Pulmonol. 54 , S46–S55 (2019).

Veit, G. et al. From CFTR biology toward combinatorial pharmacotherapy: Expanded classification of cystic fibrosis mutations. Mol. Biol. Cell https://doi.org/10.1091/mbc.E14-04-0935 (2016).

Meng, X., Clews, J., Kargas, V., Wang, X. & Ford, R. C. The cystic fibrosis transmembrane conductance regulator (CFTR) and its stability. Cell. Mol. Life Sci. https://doi.org/10.1007/s00018-016-2386-8 (2017).

Chang, X. B. et al. Removal of multiple arginine-framed trafficking signals overcomes misprocessing of ΔF508 CFTR present in most patients with cystic fibrosis. Mol. Cell https://doi.org/10.1016/S1097-2765(00)80196-3 (1999).

Wendeler, M. W., Nufer, O. & Hauri, H.-P. Improved maturation of CFTR by an ER export signal. FASEB J. https://doi.org/10.1096/fj.07-8128com (2007).

Okiyoneda, T. et al. Chaperone-independent peripheral quality control of CFTR by RFFL E3 ligase. Dev. Cell https://doi.org/10.1016/j.devcel.2018.02.001 (2018).

Matos, A. M. et al. Inhibition of calpain 1 restores plasma membrane stability to pharmacologically rescued Phe508del-CFTR variant. J. Biol. Chem. https://doi.org/10.1074/jbc.RA119.008738 (2019).

Piao, H. et al. Sec16A is critical for both conventional and unconventional secretion of CFTR. Sci. Rep. https://doi.org/10.1038/srep39887 (2017).

Gee, H. Y., Noh, S. H., Tang, B. L., Kim, K. H. & Lee, M. G. Rescue of Δf508-CFTR trafficking via a GRASP-dependent unconventional secretion pathway. Cell https://doi.org/10.1016/j.cell.2011.07.021 (2011).

Kondratova, M., Sompairac, N., Barillot, E., Zinovyev, A. & Kuperstein, I. Signalling maps in cancer research: Construction and data analysis. Database https://doi.org/10.1093/database/bay036 (2018).

Siebenhaller, M. et al. Human-like layout algorithms for signalling hypergraphs: Outlining requirements. Brief. Bioinform. 21 , 62–72 (2018).

Kanehisa, M., Furumichi, M., Tanabe, M., Sato, Y. & Morishima, K. KEGG: New perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. https://doi.org/10.1093/nar/gkw1092 (2017).

Tang, B. L., Gee, H. Y. & Lee, M. G. The cystic fibrosis transmembrane conductance regulator’s expanding SNARE interactome. Traffic 12 , 364–371 (2011).

Stelzer, G. et al. The GeneCards suite: From gene data mining to disease genome sequence analyses. Curr. Protoc. Bioinform. https://doi.org/10.1002/cpbi.5 (2016).

Szklarczyk, D. et al. STRING v11: Protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. https://doi.org/10.1093/nar/gky1131 (2019).

Oughtred, R. et al. The BioGRID interaction database: 2019 update. Nucleic Acids Res. https://doi.org/10.1093/nar/gky1079 (2019).

Bateman, A. UniProt: A worldwide hub of protein knowledge. Nucleic Acids Res. https://doi.org/10.1093/nar/gky1049 (2019).

Braschi, B. et al. Genenames.org: The HGNC and VGNC resources in 2019. Nucleic Acids Res. https://doi.org/10.1093/nar/gky930 (2019).

Hoksza, D., Gawron, P., Ostaszewski, M., Smula, E. & Schneider, R. MINERVA API and plugins: Opening molecular network analysis and visualization to the community. Bioinformatics 35 , 4496–4498 (2019).

Download references

Acknowledgements

The authors would like to thank the domain experts that participated in stimulating discussions and valuable feedback and to the colleagues from FunGP—Functional Genomics and Proteostasis who tested CyFi-MAP and provided valuable suggestions. Work in MDA lab is supported by UIDB/04046/2020 and UIDP/04046/2020 centre grants (to BioISI) from FCT/MCTES Portugal. CP was recipient of fellowship SFRH/PD/BD/131405/2017 and through funding of LASIGE Research Unit, ref. UIDB/00408/2020 and ref. UIDP/00408/2020. AM and IB were supported in part by the Innovative Medicines Initiative Joint Undertaking under Grant Agreement No. IMI 115446 (eTRIKS), resources of which are composed of financial contributions from the European Union’s Seventh Framework Programme (FP7/2007–2013) and EFPIA companies.

Author information

Authors and affiliations.

Faculty of Sciences, BioISI—Biosystems Integrative Sciences Institute, University of Lisboa, Campo Grande, 1749-016, Lisbon, Portugal

Catarina Pereira, Carlos M. Farinha, Margarida D. Amaral & Andre O. Falcao

LASIGE, Faculty of Sciences, University of Lisboa, Campo Grande, 1749-016, Lisbon, Portugal

Catarina Pereira & Andre O. Falcao

Luxembourg Centre for Systems Biomedicine, University of Luxembourg, 6 Avenue du Swing, 4367, Belvaux, Luxembourg

Alexander Mazein, Marek Ostaszewski & Irina Balaur

CIRI UMR5308, CNRS-ENS-UCBL-INSERM, European Institute for Systems Biology and Medicine, Université de Lyon, 50 Avenue Tony Garnier, 69007, Lyon, France

Alexander Mazein & Irina Balaur

Biosciences Institute, University Medical School, Newcastle University, Framlington Place, Newcastle upon Tyne, NE2 4HH, UK

Michael A. Gray

Universität Regensburg, 9147, Regensburg, Germany

Karl Kunzelmann

You can also search for this author in PubMed   Google Scholar

Contributions

C.P.—conceptualization; data curation; formal analysis; investigation; methodology; original draft preparation; review and editing; validation. A.M.—conceptualization; methodology; review and editing; supervision. C.F.—review and editing. M.G.—review and editing. K.K.—review and editing. M.O.—MINERVA online version. I.B.—MINERVA online version. M.D.A.—conceptualization; funding acquisition; resources; review and editing supervision. A.O.F.—conceptualization; data curation; methodology; resources; review and editing; supervision; validation. All authors have read and agreed to the published version of the manuscript.

Corresponding author

Correspondence to Andre O. Falcao .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary information., rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Pereira, C., Mazein, A., Farinha, C.M. et al. CyFi-MAP: an interactive pathway-based resource for cystic fibrosis. Sci Rep 11 , 22223 (2021). https://doi.org/10.1038/s41598-021-01618-3

Download citation

Received : 28 July 2021

Accepted : 27 October 2021

Published : 15 November 2021

DOI : https://doi.org/10.1038/s41598-021-01618-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

An official website of the United States government

Here’s how you know

Official websites use .gov A .gov website belongs to an official government organization in the United States.

Secure .gov websites use HTTPS A lock ( A locked padlock ) or https:// means you’ve safely connected to the .gov website. Share sensitive information only on official, secure websites.

• Heart-Healthy Living
• High Blood Pressure
• Sickle Cell Disease
• Sleep Apnea
• Information & Resources on COVID-19
• The Heart Truth®
• Learn More Breathe Better®
• Blood Diseases & Disorders Education Program
• Publications and Resources
• Clinical Trials
• Blood Disorders and Blood Safety
• Sleep Science and Sleep Disorders
• Lung Diseases
• Health Disparities and Inequities
• Heart and Vascular Diseases
• Precision Medicine Activities
• Obesity, Nutrition, and Physical Activity
• Population and Epidemiology Studies
• Women’s Health
• Research Topics
• All Science A-Z
• Grants and Training Home
• Policies and Guidelines
• Funding Opportunities and Contacts
• Training and Career Development
• Email Alerts
• NHLBI in the Press
• Research Features
• Ask a Scientist
• Past Events
• Upcoming Events
• Mission and Strategic Vision
• Divisions, Offices and Centers
• Advisory Committees
• Budget and Legislative Information
• Jobs and Working at the NHLBI
• Contact and FAQs
• NIH Sleep Research Plan
• < Back To Research Topics

Cystic Fibrosis Research

Language switcher.

Cystic fibrosis is a life-shortening genetic disease that affects many organs of the body, especially the lungs. No cure for cystic fibrosis exists yet, but decades of NHLBI leadership in and support for research have led to more and better treatment options.

Effective treatments now allow many people with cystic fibrosis to live well into adulthood. Close to 40,000 children and adults are living with cystic fibrosis in the United States. 

The condition is caused by mutations, or changes, in the gene that codes for the cystic fibrosis transmembrane conductance regulator (CFTR) protein. CFTR helps control how chloride – a component of salt – and other ions are secreted by cells. The chloride attracts water and thins out the mucus the cells produce.

Current research supported by the NHLBI focuses on understanding how changes in the CFTR protein lead to the development of cystic fibrosis. The Institute carries out and supports studies that could lead to gene therapies and other treatments that may help the lives of people who have the condition.

NHLBI research that really made a difference

• Medicines that address underlying causes: NHLBI’s lung program supported decades of research to understand the structure and function of a lung protein that is genetically changed by cystic fibrosis. This foundational work led to the industry discovery of ivacaftor , the first drug to treat the underlying cause of cystic fibrosis, which the U.S. Food and Drug Administration (FDA) approved in 2012.
• Effective combination therapies: NHLBI-supported clinical trials led to the approval of a triple combination of CFTR modulator medicines that improves lung function in about 90% of people who have cystic fibrosis. However, further studies have shown that people of color are less likely than white people to have mutations that are eligible for treatment with current CFTR modulators. Researchers supported in part by the NHLBI are investigating these disparities. The Institute is also funding projects to develop treatments that will work for people who have less common CFTR mutations.
• Remote monitoring of lung health: As COVID-19 made in-person healthcare visits more challenging, an NHLBI-funded study developed an at-home diagnostic tool for measuring how well the lungs work. The FDA-cleared system uses a handheld breathing device and a mobile app called Breathe Easy. Patients blow into the device and communicate directly with their healthcare provider, who can monitor how well their lungs are working and adjust medicines as needed. Between April 2020 and May 2021, the Cystic Fibrosis Foundation distributed nearly 20,000 of the devices to people with cystic fibrosis in the United States. A study showed results were reliable 81% of the time for a sample of 48 patients who used the portable device with the app.

Find  funding opportunities  and  program contacts for cystic fibrosis research. 

Current research funded by the NHLBI

Our Division of Lung Diseases and its Airway Biology and Disease Branch oversee much of the research on cystic fibrosis that we fund.

• Biomarkers of declining lung function: The NHLBI Catalyze Program is supporting the development of a biomarker-based platform that predicts lung function decline in patients with cystic fibrosis and a web-based application to inform physicians when a patient may require therapeutic interventions.
• More options for disease modulators: Investigators are studying CFTR protein folding and the use of modulators for patients with rare variant mutations not typically eligible for cystic fibrosis modulator therapy.
• Monitoring methods with fewer side effects: Researchers are combining noninvasive, radiation-free imaging and proteomic biomarkers to diagnose and monitor lung disease progression in kids with cystic fibrosis.

Current research on cystic fibrosis treatments

• New treatment strategies: Although treatments improve lung function for many living with cystic fibrosis, some people have CFTR mutations that do not respond to available CFTR modulators. Even those whose disease responds to CFTR modulators still have trouble clearing bacteria out of their lungs. We are supporting a project, Novel Strategies to Clear Bacteria from the CF Lung , that aims to develop an inhaled medicine to shift the lung’s immune balance to help clear bacteria from lungs of people with cystic fibrosis.
• New treatment mechanisms: NHLBI-funded studies are testing whether a medicine to correct acid problems in the blood can also help reduce acid levels in the airways , which can then prevent or slow the development of cystic fibrosis.
• Better medicines to fight mucus: Researchers are developing new medicines to help clear and target the thick mucus found in cystic fibrosis lungs and improve how well the lungs work. This can also help prevent inflammation and infection.
• Antibiotic alternatives: Investigators are studying the thiocyanate (SCN−) analog selenocyanate (SeCN−) as an alternative therapeutic for the treatment of cystic fibrosis lung pathogens that are difficult to treat with current antibiotics.

Watch videos of a 2022 workshop that brought together experts from the NHLBI and other parts of NIH, as well as the Cystic Fibrosis Foundation, to identify future research needs for treating cystic fibrosis.

Current research on gene editing and cystic fibrosis

The NHLBI is supporting research on new genetic therapies to treat cystic fibrosis. For example, researchers are studying state-of-the-art gene delivery tools and technologies that may be better at delivering a corrected gene to lung cells. Researchers are also working on better methods to improve genetic therapies in the laboratory before moving to clinical trials.

Through the NIH Common Fund Somatic Cell Genome Editing (SCGE) Program , the NHLBI supports studies that explore new genetic therapy approaches to repair the cystic fibrosis gene, among others.

• Gene editing for lung disease: Studies using CRISPR gene editing tools will correct genes in the cells that line the airways. Using these tools could lead to new treatments for genetic and acquired lung disease.
• Gene editing for cystic fibrosis: A research program is developing combinatorial nonviral and viral CRISPR delivery for lung diseases. These studies focus on efficiently targeting gene editing tools to diseased lung cells in people who have cystic fibrosis.

To identify research barriers and challenges to using gene editing as a means to cure cystic fibrosis, the NHLBI participated in a joint workshop with the Cystic Fibrosis Foundation in 2018. The Institute also participated in a 2020 virtual workshop to discuss challenges and opportunities that could be addressed in a potential second phase of the SCGE program.

The NHLBI funds other studies of gene editing for cystic fibrosis as well.

• Molecular targets to treat cystic fibrosis: The NHLBI supports research for new molecular therapies , including gene editing. Molecular therapies have helped restore the CFTR protein function for some but not all people with cystic fibrosis. New research uses airway cells and animal models to look for more ways to prevent and treat this condition.
• How nanoparticles may improve treatment effectiveness: To improve the delivery of gene editing tools and other therapeutics, researchers are developing more effective virus-inspired nanoparticles to penetrate mucus barriers in diseases with thick mucus like cystic fibrosis. 

Find more NHLBI-funded studies on gene editing and cystic fibrosis at NIH RePORTER.

Read more about NHLBI-supported research on gene editing, which involves making changes to a specific DNA sequence to correct the mutation in the cystic fibrosis gene: Genome editing for cystic fibrosis: A Q&A with Peter Glazer, Ph.D., M.D.

Current research on understanding the causes of cystic fibrosis

• New treatment and prevention: NHLBI-supported scientists are carrying out Multi-Scale Investigations of Respiratory Mucus/Mucin Structure and Function in Health and Disease designed to build a solid foundation of knowledge about mucus, how it forms, and how it works to protect the body from infections. The resulting knowledge could lead to new ways to treat and prevent lung problems that result from the thick, sticky mucus caused by CFTR mutations.
• Origins of disease: Scientists are using a cystic fibrosis animal model to study the origins of cystic fibrosis airway disease . Researchers hope this will help speed up the development of new treatments for early lung disease.
• High-resolution imaging to better understand lung disease mechanisms: Another NHLBI-funded study uses an imaging method called optical coherence tomography (OCT) to take high-resolution images of the lungs and the nose. OCT can help researchers better understand how mucus is cleared and how cystic fibrosis affects this process.

Find more NHLBI-funded studies on the causes of cystic fibrosis at NIH RePORTER.

Learn more about how cystic fibrosis changes the airways’ cellular makeup: Molecular analysis identifies differences between healthy lung and lungs of people who have cystic fibrosis .

Cystic fibrosis research labs at the NHLBI

The Division of Intramural Research , which includes investigators from the Pulmonary Branch , is actively engaged in the study of cystic fibrosis.

Related cystic fibrosis programs

• The  Trans-Omics for Precision Medicine (TOPMed) program includes  participants who have cystic fibrosis, which may help researchers understand how genes affect people and how individuals respond to treatment.
• The NHLBI-funded LungMAP research centers are creating molecular maps of the lungs to better understand rare lung diseases in children, including cystic fibrosis.

Explore more NHLBI research on cystic fibrosis

The sections above provide you with the highlights of NHLBI-supported research on cystic fibrosis. You can explore the full list of NHLBI-funded studies on the NIH RePORTER .

To find more studies:

• Type your search words into the  Quick Search  box and press enter. 
• Check  Active Projects  if you want current research.
• Select the  Agencies  arrow, then the  NIH  arrow, then check  NHLBI .

If you want to sort the projects by budget size — from the biggest to the smallest — click on the  FY Total Cost by IC  column heading.

Select Your Interests

Customize your JAMA Network experience by selecting one or more topics from the list below.

• Academic Medicine
• Acid Base, Electrolytes, Fluids
• Allergy and Clinical Immunology
• American Indian or Alaska Natives
• Anesthesiology
• Anticoagulation
• Art and Images in Psychiatry
• Artificial Intelligence
• Assisted Reproduction
• Bleeding and Transfusion
• Caring for the Critically Ill Patient
• Challenges in Clinical Electrocardiography
• Climate and Health
• Climate Change
• Clinical Challenge
• Clinical Decision Support
• Clinical Implications of Basic Neuroscience
• Clinical Pharmacy and Pharmacology
• Complementary and Alternative Medicine
• Consensus Statements
• Coronavirus (COVID-19)
• Critical Care Medicine
• Cultural Competency
• Dental Medicine
• Dermatology
• Diabetes and Endocrinology
• Diagnostic Test Interpretation
• Drug Development
• Electronic Health Records
• Emergency Medicine
• End of Life, Hospice, Palliative Care
• Environmental Health
• Equity, Diversity, and Inclusion
• Facial Plastic Surgery
• Gastroenterology and Hepatology
• Genetics and Genomics
• Genomics and Precision Health
• Global Health
• Guide to Statistics and Methods
• Hair Disorders
• Health Care Delivery Models
• Health Care Economics, Insurance, Payment
• Health Care Quality
• Health Care Reform
• Health Care Safety
• Health Care Workforce
• Health Disparities
• Health Inequities
• Health Policy
• Health Systems Science
• History of Medicine
• Hypertension
• Images in Neurology
• Implementation Science
• Infectious Diseases
• Innovations in Health Care Delivery
• JAMA Infographic
• Law and Medicine
• Leading Change
• Less is More
• LGBTQIA Medicine
• Lifestyle Behaviors
• Medical Coding
• Medical Devices and Equipment
• Medical Education
• Medical Education and Training
• Medical Journals and Publishing
• Mobile Health and Telemedicine
• Narrative Medicine
• Neuroscience and Psychiatry
• Notable Notes
• Nutrition, Obesity, Exercise
• Obstetrics and Gynecology
• Occupational Health
• Ophthalmology
• Orthopedics
• Otolaryngology
• Pain Medicine
• Palliative Care
• Pathology and Laboratory Medicine
• Patient Care
• Patient Information
• Performance Improvement
• Performance Measures
• Perioperative Care and Consultation
• Pharmacoeconomics
• Pharmacoepidemiology
• Pharmacogenetics
• Pharmacy and Clinical Pharmacology
• Physical Medicine and Rehabilitation
• Physical Therapy
• Physician Leadership
• Population Health
• Primary Care
• Professional Well-being
• Professionalism
• Psychiatry and Behavioral Health
• Public Health
• Pulmonary Medicine
• Regulatory Agencies
• Reproductive Health
• Research, Methods, Statistics
• Resuscitation
• Rheumatology
• Risk Management
• Scientific Discovery and the Future of Medicine
• Shared Decision Making and Communication
• Sleep Medicine
• Sports Medicine
• Stem Cell Transplantation
• Substance Use and Addiction Medicine
• Surgical Innovation
• Surgical Pearls
• Teachable Moment
• Technology and Finance
• The Art of JAMA
• The Arts and Medicine
• The Rational Clinical Examination
• Tobacco and e-Cigarettes
• Translational Medicine
• Trauma and Injury
• Treatment Adherence
• Ultrasonography
• Users' Guide to the Medical Literature
• Vaccination
• Venous Thromboembolism
• Veterans Health
• Women's Health
• Workflow and Process
• Wound Care, Infection, Healing
• Download PDF
• Share X Facebook Email LinkedIn
• Permissions

What Is Cystic Fibrosis?

• 1 Case Western Reserve University School of Medicine, Cleveland, Ohio
• 2 Rainbow Babies and Children’s Hospital, Cleveland, Ohio
• Clinical Trials Update New Combination Therapy for Cystic Fibrosis Anita Slomski, MA JAMA
• News From the Food and Drug Administration Patients With Cystic Fibrosis Have New Triple-Drug Combination Rebecca Voelker, MSJ JAMA
• Review Cystic Fibrosis—A Review Thida Ong, MD; Bonnie W. Ramsey, MD JAMA

Cystic fibrosis ( CF ) is a genetic disorder caused by an abnormal gene that is inherited from both biological parents.

The cystic fibrosis transmembrane conductance regulator ( CFTR ) gene is responsible for salt transport across different tissues in the body. In CF, the protein made by the abnormal CFTR gene is absent or dysfunctional, resulting in reduced salt transport and decreased water movement, which causes thick mucus to accumulate in various parts of the body.

How Common Is Cystic Fibrosis?

Cystic fibrosis affects more than 30 000 children and adults in the US and 70 000 people worldwide. Although CF affects people of all races and ethnicities, it is most common in non-Hispanic White individuals.

Diagnosis of Cystic Fibrosis

Newborn screening programs measure a chemical in the blood called immunoreactive trypsinogen, which is elevated in CF. Most newborn screening programs also include genetic testing for the most common mutations that cause CF. Diagnosis is confirmed by finding elevated chloride levels in sweat. The sweat chloride test is also used for diagnosis of CF if the condition is clinically suspected in a child or adult.

How Does Cystic Fibrosis Affect Body Functions?

Abnormal or absent CFTR protein causes accumulation of thick mucus, which blocks the airways, leads to repeated episodes of infection and inflammation, and results in damage to the lungs. Respiratory failure is the most common cause of death in people with CF.

People with CF also have thick mucus in the pancreas, which limits release of digestive enzymes and leads to difficulty digesting food, malabsorption of nutrients, and poor weight gain. In addition, damage to the pancreas can lead to development of CF-related diabetes. People with CF may have chronic diarrhea and episodes of constipation.

Other parts of the body can also be affected. Most men with CF are infertile, and women with CF may have difficulty becoming pregnant. People with CF often have chronic sinusitis. Also, people with CF are at increased risk of dehydration in hot weather because of abnormal functioning of their sweat glands.

Treatments for Cystic Fibrosis

Most people with CF take inhaled medications daily to thin their mucus and use mechanical devices several times daily to dislodge mucus from the airways. Oral and inhaled antibiotics may be prescribed to help control infection, while intravenous antibiotics are used to treat flares of infection.

Most people with CF require a high-calorie diet with fat-soluble vitamin supplements and pancreatic enzyme replacement medications with each meal.

CFTR modulators, the first class of drugs aimed at treating the underlying cause of CF, became available in 2011. CFTR modulators have effects on abnormal CFTR proteins within cells and improve salt transport. In the US, an estimated 90% of people with CF have CFTR gene mutations that may respond to CFTR modulators. These drugs improve lung function, weight gain, and quality of life; decrease episodes of infection; and are expected to improve survival. Among children born with CF in 2019, half are predicted to live to age 48 years or older, an increase of about 10 years since 2009.

A double lung transplant may be an option for some patients with advanced lung disease due to CF.

For More Information

Cystic Fibrosis Foundation www.cff.org/intro-cf/about-cystic-fibrosis

Conflict of Interest Disclosures: Dr Konstan reported serving on advisory boards for Vertex Pharmaceuticals. No other disclosures were reported.

Source: McBennett KA, Davis PB, Konstan MW. Increasing life expectancy in cystic fibrosis: advances and challenges. Pediatr Pulmonol . Published online October 21, 2021. doi:10.1002/ppul.25733

See More About

Endres TM , Konstan MW. What Is Cystic Fibrosis? JAMA. 2022;327(2):191. doi:10.1001/jama.2021.23280

Manage citations:

© 2024

Artificial Intelligence Resource Center

Cardiology in JAMA : Read the Latest

Browse and subscribe to JAMA Network podcasts!

Others Also Liked

• Register for email alerts with links to free full-text articles
• Access PDFs of free articles
• Manage your interests
• Save searches and receive search alerts

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings
• My Bibliography
• Collections
• Citation manager

Save citation to file

Email citation, add to collections.

• Create a new collection
• Add to an existing collection

Add to My Bibliography

Your saved search, create a file for external citation management software, your rss feed.

• Search in PubMed
• Search in NLM Catalog
• Add to Search

Cystic Fibrosis

Affiliation.

• 1 *Joint and equal first authors; Medizinische Klinik V (Pneumology), LMU University of Munich, Pneumology, Medizinische Klinik Innenstadt, University of Munich; Department of Pediatrics, Justus-Liebig-University Gießen.
• PMID: 28855057
• PMCID: PMC5596161
• DOI: 10.3238/arztebl.2017.0564

Background: Universal screening of newborn babies for cystic fibrosis was launched in Germany on 1 September 2016. Here we present up-to-date information on the diagnosis, treatment, and prognosis of this disease.

Methods: This article is based on relevant publications retrieved by a selective search in PubMed, along with guidelines from Germany and abroad and systematic reviews.

Results: Cystic fibrosis is caused by a gene mutation leading to dysfunction of the cystic fibrosis transmembrane conductance regulator (CFTR) protein. It affects multiple organ systems-the lungs, pancreas, upper airways, liver, intestine, and reproductive organs-to varying degrees. Its incidence among newborn babies in Germany is between 1 in 3300 and 1 in 4800. Its diagnosis requires both clinical evidence (positive newborn screening, sibling[s] with cystic fibrosis, clinical signs) and the demonstration of CFTR dysfunction by an elevated chloride concentration in sweat, and/or two disease-causing mutations, and/or abnormal electrophysiological findings (nasal potential difference measurement, intestinal short-circuit current measurement). Patients should be cared for by specialized cystic fibrosis centers in close cooperation with their primary care physicians. The median life span of patients with this disease has risen steadily to the current value of 40 years. Aside from symptomatic treatment, the first mutation- specific treatments have recently become available.

Conclusion: Early diagnosis and optimized treatment prolong the lives of persons with cystic fibrosis and improve their quality of life. Causally directed treatment for all patients and their effects on the course of disease are now central issues for further research.

PubMed Disclaimer

Neonatal screening, modified from (5)…

Neonatal screening, modified from (5) *The cutoff value of PAP depends on the…

Complications of CF according to…

Complications of CF according to the American CF registry (e15) (with permission of…

Similar articles

• Cystic Fibrosis Diagnosis and Newborn Screening. Rosenfeld M, Sontag MK, Ren CL. Rosenfeld M, et al. Pediatr Clin North Am. 2016 Aug;63(4):599-615. doi: 10.1016/j.pcl.2016.04.004. Pediatr Clin North Am. 2016. PMID: 27469178 Review.
• The diagnosis of cystic fibrosis. De Boeck K, Vermeulen F, Dupont L. De Boeck K, et al. Presse Med. 2017 Jun;46(6 Pt 2):e97-e108. doi: 10.1016/j.lpm.2017.04.010. Epub 2017 May 31. Presse Med. 2017. PMID: 28576637 Review.
• [Update on cystic fibrosis : From neonatal screening to causal treatment]. Sommerburg O, Wielpütz MO. Sommerburg O, et al. Radiologie (Heidelb). 2022 Nov;62(11):981-994. doi: 10.1007/s00117-022-01076-6. Epub 2022 Oct 24. Radiologie (Heidelb). 2022. PMID: 36278998 German.
• Laboratory tests for the diagnosis of cystic fibrosis. Wang L, Freedman SD. Wang L, et al. Am J Clin Pathol. 2002 Jun;117 Suppl:S109-15. doi: 10.1309/XTM9-E4BU-C1Y5-JN10. Am J Clin Pathol. 2002. PMID: 14569807 Review.
• Applying Cystic Fibrosis Transmembrane Conductance Regulator Genetics and CFTR2 Data to Facilitate Diagnoses. Sosnay PR, Salinas DB, White TB, Ren CL, Farrell PM, Raraigh KS, Girodon E, Castellani C. Sosnay PR, et al. J Pediatr. 2017 Feb;181S:S27-S32.e1. doi: 10.1016/j.jpeds.2016.09.063. J Pediatr. 2017. PMID: 28129809
• Secondary use of patient data within decentralized studies using the example of rare diseases in Germany: A data scientist's exploration of process and lessons learned. Zoch M, Gierschner C, Andreeff AK, Henke E, Sedlmayr M, Müller G, Tippmann J, Hebestreit H, Choukair D, Hoffmann GF, Fritz-Kebede F, Toepfner N, Berner R, Biergans S, Verbücheln R, Schaaf J, Fleck J, Wirth FN, Schepers J, Prasser F. Zoch M, et al. Digit Health. 2024 Aug 10;10:20552076241265219. doi: 10.1177/20552076241265219. eCollection 2024 Jan-Dec. Digit Health. 2024. PMID: 39130526 Free PMC article.
• Cystic Fibrosis in an Adolescent: A "Miranda Warning" Against Blaming TB-A Case-Based Scholarly Update. Adela AY, Kebede AG, Zewdneh D, Kifle M, Dias AB. Adela AY, et al. Adolesc Health Med Ther. 2024 Feb 2;15:19-29. doi: 10.2147/AHMT.S451251. eCollection 2024. Adolesc Health Med Ther. 2024. PMID: 38328573 Free PMC article.
• [Comparison of diagnostic data after introduction of cystic fibrosis screening in newborns in Germany]. Nährlich L, Brockow I. Nährlich L, et al. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz. 2023 Nov;66(11):1243-1248. doi: 10.1007/s00103-023-03778-1. Epub 2023 Oct 13. Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz. 2023. PMID: 37831094 Free PMC article. German.
• Impaired Nitric Oxide Synthetase Activity in Primary Ciliary Dyskinesia-Data-Driven Hypothesis. Eggenkemper L, Schlegtendal A, Maier C, Lücke T, Brinkmann F, Beckmann B, Tsikas D, Koerner-Rettberg C. Eggenkemper L, et al. J Clin Med. 2023 Sep 16;12(18):6010. doi: 10.3390/jcm12186010. J Clin Med. 2023. PMID: 37762950 Free PMC article.
• Role of Olive Bioactive Compounds in Respiratory Diseases. Vijakumaran U, Goh NY, Razali RA, Abdullah NAH, Yazid MD, Sulaiman N. Vijakumaran U, et al. Antioxidants (Basel). 2023 May 23;12(6):1140. doi: 10.3390/antiox12061140. Antioxidants (Basel). 2023. PMID: 37371870 Free PMC article. Review.
• World Health Organization. The molecular genetic epidemiology of cystic fibrosis. www.who.int/genomics/publications/reports/en/index.html (last accessed on 14 February 2017)
• Sommerburg O, Hammermann J, Lindner M, et al. Five years of experience with biochemical cystic fibrosis newborn screening based on IRT/PAP in Germany. Pediatr Pulmonol. 2015;50:655–664. - PubMed
• Sens B, Stern M. Qualitätssicherung Mukoviszidose 2012. Bad Honnef: Hippocampus Verlag. 2013
• Nährlich L, Burkhart M, Wiese B. Deutsches Mukoviszidoseregister- Berichtsband 2015. www.muko.info/fileadmin/redaktion/datei_gruppen/muko_institut/Register/B... (last accessed on 10 December 2016
• Gemeinsamer Bundesausschuss. Kinder-Richtlinie: Formale und inhaltliche Überarbeitung (Neustrukturierung) - Neufassung. www.g-ba.de/informationen/beschluesse/2287/ (last accessed on 10 September 2016)

Publication types

• Search in MeSH

Related information

• Cited in Books

LinkOut - more resources

Full text sources.

• Deutsches Aerzteblatt International
• Europe PubMed Central
• PubMed Central

Other Literature Sources

• scite Smart Citations
• Genetic Alliance
• MedlinePlus Health Information

• Citation Manager

NCBI Literature Resources

MeSH PMC Bookshelf Disclaimer

The PubMed wordmark and PubMed logo are registered trademarks of the U.S. Department of Health and Human Services (HHS). Unauthorized use of these marks is strictly prohibited.

Loading metrics

Open Access

Peer-reviewed

Research Article

Metabolic specialization drives reduced pathogenicity in Pseudomonas aeruginosa isolates from cystic fibrosis patients

Roles Data curation, Formal analysis, Investigation, Visualization, Writing – original draft

Affiliation The Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Kongens Lyngby, Denmark

Roles Investigation

Affiliation Department of Clinical Microbiology 9301, Rigshospitalet, Copenhagen, Denmark

Roles Methodology, Resources, Writing – review & editing

Affiliation Department of Biochemistry, University of Cambridge, Cambridge, United Kingdom

Roles Funding acquisition, Resources, Writing – review & editing

Affiliations Department of Clinical Microbiology 9301, Rigshospitalet, Copenhagen, Denmark, Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark

Roles Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing

Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Supervision, Visualization, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliations The Novo Nordisk Foundation Center for Biosustainability, Technical University of Denmark, Kongens Lyngby, Denmark, Department of Clinical Microbiology 9301, Rigshospitalet, Copenhagen, Denmark

• Bjarke Haldrup Pedersen, 
• Filipa Bica Simões, 
• Ivan Pogrebnyakov, 
• Martin Welch, 
• Helle Krogh Johansen, 
• Søren Molin, 
• Ruggero La Rosa

• Published: August 23, 2024
• https://doi.org/10.1371/journal.pbio.3002781
• Peer Review
• Reader Comments

This is an uncorrected proof.

Metabolism provides the foundation for all cellular functions. During persistent infections, in adapted pathogenic bacteria metabolism functions radically differently compared with more naïve strains. Whether this is simply a necessary accommodation to the persistence phenotype or if metabolism plays a direct role in achieving persistence in the host is still unclear. Here, we characterize a convergent shift in metabolic function(s) linked with the persistence phenotype during Pseudomonas aeruginosa colonization in the airways of people with cystic fibrosis. We show that clinically relevant mutations in the key metabolic enzyme, pyruvate dehydrogenase, lead to a host-specialized metabolism together with a lower virulence and immune response recruitment. These changes in infection phenotype are mediated by impaired type III secretion system activity and by secretion of the antioxidant metabolite, pyruvate, respectively. Our results show how metabolic adaptations directly impinge on persistence and pathogenicity in this organism.

Citation: Pedersen BH, Simões FB, Pogrebnyakov I, Welch M, Johansen HK, Molin S, et al. (2024) Metabolic specialization drives reduced pathogenicity in Pseudomonas aeruginosa isolates from cystic fibrosis patients. PLoS Biol 22(8): e3002781. https://doi.org/10.1371/journal.pbio.3002781

Academic Editor: Alice Prince, Columbia University, UNITED STATES OF AMERICA

Received: January 17, 2024; Accepted: August 1, 2024; Published: August 23, 2024

Copyright: © 2024 Pedersen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The authors declare that all data necessary for supporting the findings of this study are enclosed in this paper ( S1 – S4 Data). All genomic data is publicly available through the SRA database and has been published previously by Marvig et al., (2015), doi: 10.1038/ng.3148 .

Funding: The work at the Novo Nordisk Foundation Center for Biosustainability is supported by the Novo Nordisk Foundation www.novonordiskfonden.dk (grant number NNF20CC0035580). This work was supported by the UK Cystic Fibrosis Trust www.cysticfibrosis.org.uk (grant number SRC017 - MW, SM, HKJ) and the Independent Research Fund Denmark/Natural Sciences www.dff.dk (grant number 9040-00106B - SM). HKL was supported by the Novo Nordisk Foundation www.novonordiskfonden.dk (Challenge grant NNF19OC0056411). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Abbreviations: ALI, air–liquid interface; COG, Clusters of Orthologous Groups; HCA, hierarchical cluster analysis; HCN, hydrogen cyanide; LPS, lipopolysaccharide; PCA, principal component analysis; PDHc, pyruvate dehydrogenase complex; PFA, paraformaldehyde; PQS, Pseudomonas quinolone signal; pwCF, people with cystic fibrosis; ROS, reactive oxygen species; SCFM2, synthetic cystic fibrosis medium 2; T3SS, type III secretion system; TEER, transepithelial electrical resistance

Introduction

Difficult to treat bacterial infections are increasing around the world [ 1 ]. While antibiotic resistance is a major cause of treatment failure, other less characterized mechanisms rooted in the complexity of the host–pathogen interactions are also substantial contributors to persistence [ 2 ]. Genetic variants with high tolerance to stresses, host and immune evasion capabilities, and low virulence are often specifically selected for in persistent infections since they provide higher within-host fitness [ 3 , 4 ]. In people with cystic fibrosis (pwCF), for example, opportunistic pathogens such as Pseudomonas aeruginosa colonize the airways and establish persistent infections that can last for more than 30 years. During the infection, the bacteria differentiate into heterogeneous populations specifically evolved for the host microenvironment [ 5 – 8 ]. Surprisingly, antibiotic resistance is not the initial driver of persistence, since bacteria retain antibiotic susceptibility for years after colonization [ 9 ]. Seemingly, these populations use mechanisms of persistence such as biofilm formation, loss of flagella and virulence factors, growth rate reduction, immune escape, and metabolic specialization to “hide” from the immune system and withstand the antibiotic treatment [ 2 ]. These mechanisms usually fall below the detection radar since, in simple laboratory conditions, without the intrinsic complexity of the host environment and its interactions with the bacteria, no assumptions on the persistence of such populations can be made. However, these mechanisms are relevant for the overall persistence of the infecting population, possibly being the main drivers of the initial host colonization before the insurgence of antibiotic resistance.

One hypothesis is that metabolic specialization strongly influences the host–pathogen interactions and leads to persistence [ 10 , 11 ]. In many infections, the nutrient composition of the host microenvironment provides an environmental cue for bacterial pathogens to activate their virulence repertoire [ 11 ]. Abnormalities in calcium homeostasis, nutrient limitation, or change in pH can, for example, trigger the type III secretion system (T3SS) cascade in several bacterial species, activating the secretion of virulence effectors promoting colonization of the host [ 12 – 15 ]. In P . aeruginosa , the T3SS has a key role during infection, since through its injectosome and secreted factors, it subverts the host cell machinery influencing both invasion, growth, and host immune response [ 12 , 16 , 17 ]. Similarly, metabolism-dependent processes such as biofilm production and mucoidy provide shielding from antibiotics and immune cells and are generally associated with worse prognoses [ 18 – 20 ]. Furthermore, crosstalk between bacteria and immune cells through exchange of bacterial and host metabolites has additionally revealed the importance of metabolism in determining the outcome of an infection [ 21 – 23 ]. Importantly, metabolism is not a static function; it dynamically changes to support cell growth and virulence specifically for each host and type of infection. Clinical strains of P . aeruginosa that infect pwCF have been shown to modify their metabolic preference to accommodate the specific nutrient composition of the airways [ 24 , 25 ]. Auxotrophy, specialized assimilation of carbon sources, secretion of high value metabolites and differential oxygen requirements of adapted clinical strains ensure appropriate functionality of the cell and support the phenotype requirements in the host [ 10 ]. However, it is still unclear if and how metabolic specialization directly contribute(s) to persistence [ 24 , 26 ]. Furthermore, the specific selective forces (for example, antibiotic treatment or the immune system) leading to metabolic specialization still remain uncharacterized. A few examples of laboratory studies in P . aeruginosa and Escherichia coli have suggested the involvement of specific metabolic mutations with changes in antibiotic susceptibility and virulence [ 27 – 31 ]. However, limited knowledge is available on their relevance in clinical isolates of P . aeruginosa during an infection. Moreover, the extent to which metabolic specialization per se provides a specific fitness advantage or if it is merely a downstream effect of accommodating other essential phenotypes still remains unknown. For example, it has been shown that overexpression of multidrug efflux pumps causing antibiotic resistance, can lead to rewired metabolism, thereby compensating for the associated fitness cost [ 32 ]. However, if such a mechanism is generalizable and whether it specifically contributes to persistence remains still unexplored. Importantly, previous metabolic characterizations of clinical strains were carried out on only a limited number of isolates and/or on bacterial cultures at one specific growth phase (exponential or stationary phase) lacking the dynamics of metabolic processes [ 24 , 33 , 34 ]. Because of technical challenges related to the complexity of dynamic metabolomic analysis and its interpretation, large-scale analyses of populations of clinical isolates that account for the inherent dynamics of metabolism have so far been lacking. Moreover, the effect of metabolic specialization on the host–pathogen interactions remains unclear, limiting the understanding of its contribution to persistence [ 2 , 35 ]. It is, therefore, crucial to systematically identify and characterize new mechanisms of metabolic specialization to build a comprehensive understanding of the contribution of metabolism to persistence. Such efforts will provide new understanding of treatment failure and unravel new pathogen vulnerabilities and therapeutic options, which are currently overshadowed by the focus on increasing antibiotic resistance.

Here, we identify and characterize molecular mechanisms of metabolic specialization occurring in clinical strains of P . aeruginosa associated with persistence in pwCF. We further show that these mechanisms have an impact on the relationship between the host and the pathogen. By analyzing, in detail, the metabolic and proteomic profiles of clinical strains from pwCF at different stages of within-host evolution, we identified distinct metabolic configurations characterized by CF-specific nutrient assimilation and secretion patterns. These changes in metabolism and proteome allocation are directly linked with mutations affecting key metabolic genes. We also characterize 1 specific mechanism of metabolic specialization involving the pyruvate dehydrogenase complex (PDHc), which is essential for the processive flux of pyruvate through central carbon metabolism. Surprisingly, recombinant strains containing single mutations in the PDHc show decreased infection capabilities and are associated with inflammation in an air–liquid interface (ALI) infection model system. These strains display a chronic infection phenotype and an increased secretion of pyruvate, which is an important scavenger of reactive oxygen species (ROS) and has known anti-inflammatory properties [ 36 – 38 ]. Importantly, we show that these mutations are widespread in clinical isolates of P . aeruginosa from different patients and infection scenarios. This suggests that metabolic specialization might be specifically selected for during the early stages of an infection to limit host-dependent inflammation. Altogether, these results provide a rationale for metabolic specialization during CF airway infections.

Strain collection for metabolomic analyses

To identify molecular mechanisms of metabolic specialization which modify the interactions between the host and the pathogen occurring during the infection of pwCF, we selected from our longitudinal collection of 474 P . aeruginosa clinical strains [ 39 ] pairs of early (*_E; first P . aeruginosa isolated from the person) and late (*_L) clinical isolates, longitudinally isolated from eight different pwCF ( Fig 1A ). Previous studies have shown that slow growth is an important phenotype associated with long term adaptive evolution to the CF airways, appearing within the first 2 to 3 years of the infection [ 9 ]. Therefore, we used growth rate as a proxy for within-host evolution. This allowed us to increase the odds of selecting strains presenting unexplored mechanisms of metabolic specialization. Accordingly, we investigated metabolic adaptation in a diverse collection of 16 clinical strains from 8 distinct pwCF ( Fig 1A ).

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

(A) Growth rate (hour -1 ) of early (circles), late (triangles), and PAO1 (square) strains colored by their clone type. The fold change of growth rate from early to late is shown as bar chart on the right y axis and as number value at the base of each bar. (B) The maximal optical density (maxOD) of all isolates, grouped by early vs. late comparison. Statistical significance was calculated by unpaired Welch t test and indicated as *** ( p < 0.001). (C) Growth rate (hour -1 ) of each strain in M9 minimal media containing 20 mM of a single carbon source glucose (Glu), lactate (Lac), or succinate (Scc). The data underlying this figure can be found in S4 Data . maxOD, maximal optical density.

https://doi.org/10.1371/journal.pbio.3002781.g001

As by selection criteria, when grown in synthetic cystic fibrosis medium 2 (SCFM2) (which has been formulated to resemble the CF mucus composition [ 40 ]), early isolates showed a high growth rate, comparable with the reference strain, PAO1. Conversely, late isolates showed a 2.2- to 5.1-fold reduction in growth rate, in addition to a lower maximal optical density (paired t test, p = 0.002) ( Fig 1A and 1B ). Moreover, some late isolates were unable to grow in minimal medium in presence of single carbon sources such as glucose, lactate, or succinate, indicative of a metabolic constraint such as auxotrophy [ 41 ] ( Fig 1C ).

The genome of each isolate had been previously sequenced, and their genotype and phylogenetic relationship determined [ 39 ]. Specifically, each pair of early and late clinical isolates are phylogenetically related, share a common ancestor, and belong to the same “clone type” (defined as genomes sharing more than 10,000 SNPs) ( Fig 1A ). Each pwCF was colonized by a distinct clone type with evolutionary histories spanning between 1.4 and 7 years ( S1 Table ). On average, the difference in mutations between early and late strains was 46 ± 5 with the exception of strains of the DK36 clone type which show an increased number of mutations (402 mutation differences) due to hypermutation ( S1 Table ) [ 39 ]. As previously reported [ 42 ], each single isolate represents the most abundant representative of the population from a sputum sample. These selection criteria, therefore, allowed us to analyze isolates from different individuals with distinct clinical and evolutionary histories, rather than focusing on the heterogeneous population present in pwCF.

Within-host evolution selects for specialized metabotypes

To evaluate the degree of metabolic specialization in the clinical isolates, dynamic exo-metabolomics was performed on cells growing in SCFM2 ( S1 Data ). This approach allowed for analysis of the assimilation and secretion patterns of specific metabolites during the growth of each isolate, providing a dynamic profile of their metabolic activity. Principal component analysis (PCA), k-means clustering, and hierarchical cluster analysis (HCA) of the extracellular metabolomes revealed the presence of 3 distinct specialized metabolic configurations in late strains, hereafter defined as “adapted metabotypes” (DK15 and DK55 metabotype 1; DK12 metabotype 2; DK03, DK13, and DK36 metabotype 3). These metabotypes separate the late isolates from each respective early isolate and from those strains not exhibiting metabolic changes, hereafter referred to as “naïve metabotypes” ( Fig 2A ). Note that whereas PCA emphasizes the largest differences in the metabolomic profiles, the HCA clearly separates the early strains from the late strains, indicating a certain degree of metabolic specialization in all late isolates ( Fig 2B ). Indeed, both the hierarchy of assimilation (metabolite half-lives, OD 50 ) which represents the order of assimilation of the available carbon sources, and the secretion of metabolites differ between early and late isolates ( Fig 2C and 2D ). Of note, late strains of DK12, DK36, and DK55 secreted high amounts of pyruvate which is a key metabolite connecting glucose metabolism to the TCA cycle. Detailed information on the specific metabolic preferences of each metabotype is presented in S1 Text . Furthermore, the net balance between assimilated and secreted metabolites (total mM) varies between early and late strains ( Fig 2E ), which positively correlates with the lower biomass of the late isolates (Pearson’s r = 0.9; p = 0.0023) (Figs 1B and 2E ). In other words, a high percentage of assimilated carbon sources are catabolized in the late isolates and secreted back into the culture medium, thus limiting their availability for biomass production. This suggests either a specific metabolic configuration for the late isolates—the objective of which is not biomass accumulation—or an apparent inefficient metabolism.

(A) PCA showing separation of strains based on their time-resolved exo-metabolomes. Strains are shown as circles (early), triangles (late), and square (PAO1), with dashed arrows indicating the most notable trajectories from early to late, colored by clone type. Metabotypes were designated based on iterative k-means clustering analysis and HCA. See Material and methods for details. (B) HCA of the exo-metabolomics data, showing a general separation between early and late strains. Branches for early isolates are indicated in blue, late isolates in red and PAO1 in black. Accuracy of the HCA was tested by bootstrapping where gray values within the branches represent the % of bootstrap values for 10,000 replicates. (C) Assimilation hierarchies of the analyzed metabolites. Each symbol represents the half-life (OD 50 ) of a specific metabolite while each line connects the same metabolite for each pair of early and late isolates. Missing connecting curves indicates that either the strain did not assimilate the metabolite, or that the OD 50 was outside the analyzed assimilation window. (D) Secretion plots showing variations in the concentration of acetate, pyruvate, and formate (mM) relative to normalized growth (OD 600 ). Shaded areas indicate the 95% confidence intervals. (E) Table showing the total amount (mM) of carbon sources assimilated for each strain. Secreted metabolites were subtracted from the total to account for their excretion in the medium. Clone types are separated by gray boxes with “E” indicating early and “L” indicating late isolate. The data underlying this figure can be found in S1 and S4 Data. HCA, hierarchical cluster analysis; PCA, principal component analysis.

https://doi.org/10.1371/journal.pbio.3002781.g002

Changes in proteome allocation supports the metabolic specialization of clinical isolates

To test the hypothesis that the observed metabolic specialization is rooted in changes in expression of proteins involved in cellular metabolism, we analyzed the proteome of each clinical strain. To this end, we performed whole cell proteomics to compare protein expression profiles between PAO1, early, and late strains (comparisons being early versus PAO1, late versus PAO1, and late versus early). This allowed us to evaluate proteome changes in the clinical isolates relative to a laboratory reference strain (early versus PAO1, late versus PAO1) and to identify in the clinical isolates possible molecular mechanism(s) underpinning the metabolic specialization we observed (late versus early). Of the 2,061 proteins identified, 740 were differentially expressed in at least 1 comparison (early versus PAO1 or late versus early) ( S2 Data ). Similar to the metabolomic analysis, when comparing the normalized expression profile of all identified proteins, PCA readily separated the proteome of the late isolates from that of the early isolates, suggesting that changes in metabolite profiles might be rooted in the proteome ( Fig 3A and 3B ). Most of the early and some of the late proteomes co-localized with the reference strain PAO1, indicating little or no changes in proteome allocation ( Fig 3A and 3B ). However, the proteome of the late isolates of lineages DK12, DK17, DK36, and DK55 separated from the respective early strains to form a cluster of adapted proteomes ( Fig 3A and 3B ). Notably, the DK03 strains formed an independent cluster, indicative of a lineage-specific proteomic signature (Figs 1A , 3A , and 3B ).

(A) PCA showing separation of strains based on their mid-exponential proteomes. Strains are shown as circles (early), triangles (late), and square (PAO1), with dashed arrows indicating most notable trajectories from early to late, colored by clone type. (B) HCA of the proteomic data containing 2,061 proteins identified, showing a general separation between early and late strains. Branches for early isolates are indicated in blue, late isolates in red, and PAO1 in black. Accuracy of the HCA was tested by bootstrapping where gray values within the branches represent the % of values for 10,000 replicates. (C) HCA of the 235 differentially expressed proteins when comparing late vs. early strains within the COG categories related to metabolism. Proteomes are represented by dashed lines for “cluster A” and continuous lines for “cluster B.” Accuracy of the HCA was tested by bootstrapping where gray values within the branches represent the % of bootstrap values for 10,000 replicates. (D) Parallel plots of the number of DE proteins within the COG categories related to metabolism in cluster A (blue, shaded) and B (red, transparent). For each category, the mean of each cluster is indicated. The difference in the total number of DE proteins between clusters when matching by category was computed by two-tailed paired t test where ** indicates p = 0.0047. The data underlying this figure can be found in S2 Data . COG, Clusters of Orthologous Groups; DE, differentially expressed; HCA, hierarchical cluster analysis; PCA, principal component analysis.

https://doi.org/10.1371/journal.pbio.3002781.g003

A similar relationship between the proteomes is obtained through HCA, specifically when considering the 235 differentially expressed proteins (late versus early comparison) belonging to the Clusters of Orthologous Groups (COG) categories involved in metabolism ( Fig 3C ). When comparing differentially expressed proteins between late versus early strains, 2 clusters of proteomes are identified. Cluster A represents proteomic changes in clone types without any apparent metabolic specialization. In contrast, cluster B represents proteomic changes in strains which underwent metabolic specialization in the adaptive evolution process from early to late (DK12, DK36, and DK55). This result supports our hypothesis that metabolic specialization is likely rooted in changes in the expression of proteins involved in cellular metabolism ( Fig 3C ). Notably, cluster B is characterized by a significantly higher number of differentially expressed proteins in the COG categories involved in metabolism relative to cluster A (two-tailed paired t test p = 0.0047) (Figs 3D and S1A ). Similarly, several categories of proteins involved in the metabolism of amino acids, fatty acids, and sugars are statistically enriched in early and late strains ( S1B Fig ) providing a molecular explanation for the altered hierarchy of assimilation of the carbon sources and the reduced growth rate of the adapted metabotypes ( Fig 2C and 2D ). Detailed information on the convergent expression at the pathway level of metabolic proteins is presented in S1 Text ( S2 Fig ).

Although late strains of DK12, DK36, and DK55 belong to different metabotypes, their proteomes move in the same direction ( Fig 3A–3C ). This indicates that largely similar proteomes can sustain distinct metabolic configurations which ultimately depend on the metabolic fluxes thorough specific pathways and on their regulation. We note that, in the case of the DK17 strains comparison (cluster B) ( Fig 3A–3C ), although our metabolomic analysis did not detect any major metabolic rewiring ( Fig 2A and 2B ), proteome reorganization could be related to other metabolic processes which were not analyzed in our study.

Changes in virulence traits during within patient evolution

To evaluate the relationship between metabolic specialization and virulence, we analyzed which differentially expressed proteins were enriched when comparing late versus early isolates. Interestingly, most of the changes (based on their KEGG and GO categories) were related to adaptation to the infection environment, redox balance, and virulence (Figs 4A and S1B ). For example, most of the lineages showed expression changes in phenazine biosynthesis (KEGG) and secondary metabolite(s) biosynthetic process (GO), both of which are deeply involved in redox-balance, cell homeostasis, metabolism, and virulence [ 43 , 44 ].

(A) Enrichment analysis for KEGG and GO terms by the specific comparisons (Early vs. PAO1; Late vs. PAO1; Late vs. Early). Icons represent individual lineages and are colored by fold-enrichment. (B) Differential expression of virulence factors in early clinical strains vs. PAO1. Only differentially expressed proteins are represented and are colored by clone type. (C) HCA of differentially expressed proteins of virulence factors in late vs. early clinical strains. The data underlying this figure can be found in S2 Data . HCA, hierarchical cluster analysis.

https://doi.org/10.1371/journal.pbio.3002781.g004

Several changes in protein expression are already apparent in early strains relative to PAO1 indicating a different response of clinical strains to the airway-like conditions accompanying growth in SCFM2 ( Fig 4A ). Specifically, early strains show increased expression of proteins involved in alginate production, hydrogen cyanide (HCN), phenazine biosynthesis, and PQS (the P seudomonas q uinolone s ignal), and decreased expression of proteins involved in flagella biogenesis, lipopolysaccharide (LPS) O-antigen metabolism, and pyoverdine production, which are all hallmarks of an acute infection phenotype ( Fig 4B ). Early stage colonization and acute infection are thought to require expression of several virulence factors promoting host tissue injury and immune response impairment [ 45 ]. By contrast, when comparing late versus early strains, the adapted metabotypes—and specifically strains belonging to cluster B (DK12, DK17, DK36, and DK55)—show convergent up-regulation of proteins related to alginate, phenazine, PQS, and the type VI secretion system (T6SS), suggesting that metabolism plays an active role in regulating virulence and the chronic infection phenotype (Figs 4C and S4 ). Reduced virulence and cytotoxicity are advantageous for long-term infection and the establishment of a persistent infection [ 46 ]. Interestingly, the DK12 late strain also shows lower expression (-3-fold) of the T3SS toxin, ExoT, which is known to play a role in preventing phagocytosis, in the induction of cytoskeletal reorganization, and in host cell apoptosis [ 16 ]. Overall, the identified pattern of proteomic changes suggests that early isolates show reduced expression of virulence factors relative to PAO1, with an even greater reduction being associated with the late isolates. This is consistent with their persistence during CF airway infections [ 46 ].

aceE and aceF mutations leads to metabolic specialization and impaired virulence

From the previously published genome sequences [ 39 ], we searched for mutations in the genome of each clinical isolate which might explain the secretion of pyruvate and the metabolic specialization of late isolates. In particular, we focused on those genes encoding the phospho enol pyruvate-pyruvate-oxaloacetate node ( S2 Fig ), since such mutations might be expected to lead to elevated pyruvate secretion. This led to the identification of mutations in genes encoding the PDHc, aceE and aceF , in late strains of DK12 and DK36, respectively. The PDHc catalyzes the conversion of pyruvate to acetyl-CoA, connecting sugar metabolism with the TCA cycle ( S2 Fig ). Interestingly, the aceE and aceF genes are candidate pathoadaptive genes, suggesting positive selection for such mutations in pwCF [ 39 ]. In our strain collection of 474 isolates from 34 pwCF [ 39 ], we identified the presence of 18 different and independent aceE / aceF mutations (4 indels and 14 SNPs, of which 3 were synonymous) in 18 separate lineages. These lineages were present in more than half of the patients, supporting the hypothesis that modulation of the PDHc might be selected for during airway infections ( S5 Fig ). Moreover, in situ expression of the aceE and aceF genes is reduced in sputum samples collected from chronically infected pwCF [ 47 ]. This suggests an undescribed role of the PDHc activity which connects central carbon metabolism and pathogenicity during infection.

The DK12_L and DK36_L strains contained a +TCCC duplication at position 813 in aceF and a T→C transition at position 551 in aceE , respectively ( Fig 5A ). To evaluate the contribution of these mutations to the bacterial phenotype, independent of the underlying historical contingency of the clinical strains, we generated recombinant PAO1 derivative strains containing the same mutations. The aceF mutation leads to a frameshift starting from Lys 273, whereas the aceE mutation leads to a Phe→Ser amino acid change at position 184. The recombinant aceE and aceF mutant strains show reduced growth rate ( Fig 5B ) and increased secretion of pyruvate (maximal pyruvate accumulation by the aceE mutant was 0.5 mM, whereas for aceF , the value was 13 mM) ( Fig 5C ). By comparison, maximal pyruvate accumulation by the DK12_L and DK36_L late strains was 8.5 mM and 4.6 mM, respectively ( Fig 2D ). The PAO1-derived aceE mutant strain clearly has a milder phenotype (a smaller reduction in growth rate and lower pyruvate secretion) compared with the aceF mutant strain, suggesting partial functionality of the PDHc. As previously reported, in P . aeruginosa , acetate is catabolized into acetyl-CoA, and can, therefore, metabolically complement the growth defects of PDH mutants [ 31 , 48 , 49 ]. Indeed, we were able to fully restore the growth phenotype of the aceE strain by supplementing the bacterial culture with acetate to replenish the pool of acetyl-CoA, whereas this only partially restored growth in the aceF mutant strain ( Fig 5B ). A similar result was obtained by expressing a wild-type copy of the aceE gene under its native promoter which complements the growth defect of the strain ( S6A Fig ). These data confirm that the frameshift in aceF has a much greater impact on PDHc activity than the SNP in aceE . Surprisingly, in laboratory conditions, phenotypes such as biofilm formation, motility, redox susceptibility, pyoverdine production, and antibiotic susceptibility showed no statistical difference between the PAO1 wild-type strain and the aceE and aceF mutant strains, except for a slight decrease in twitching motility and mildly increased tobramycin susceptibility ( S6 Fig ).

(A) Schematic of the aceE and aceF genes including location and type of mutation found in the DK36 ( aceE ) and DK12 ( aceF ) late strains. (B) Growth rate (hour -1 ) in SCFM2 (blue) and SCFM2 supplemented with 5 mM acetate (red) for PAO1, aceE and aceF mutant strains, and DK12 and DK36 late clinical isolates. Bars indicate mean ± SEM, with icons representing biological replicates. Statistical significance is assessed by two-tailed unpaired parametric Welch t test and indicated as * ( p < 0.05), ** ( p < 0.01), or **** ( p < 0.0001). (C) Pyruvate secretion (mM) for PAO1 wt (yellow), aceE (cyan), and aceF (magenta) mutant strains over 24 h. Icons represent biological replicates. (D) PCA and HCA of whole-cell proteomics for PAO1 wt and aceE and aceF mutant strains. Filled icons indicate without (−) and unfilled icons indicate with (+) 5 mM acetate. E Number of differentially expressed proteins for mutant strain vs. PAO1 wt comparisons: in orange in absence of acetate (+/− mutation) and in green in presence of acetate (+/− mutation (ace)). In purple, the number of differentially expressed proteins for either the PAO1 wt or the aceE or aceF mutant strain in presence vs. absence of acetate (+/− acetate). (F) Metabolic map of enzymes related to pyruvate and acetyl-CoA metabolism for mutant strains vs. PAO1 in absence (left) or presence (right) of acetate. Reactions are colored by their pathway if the underlining enzyme is differentially expressed (dashed if down-regulated or continuous if up-regulated). Enzymes responsible for each reaction are indicated by numbered circles. For details on individual enzymes, see S7 Fig . (G) Differentially expressed proteins involved in virulence in aceE and aceF mutant strains vs. PAO1 in absence of acetate. Jittered icons indicate specific proteins that are up-regulated (red plus) or down-regulated (blue circle). The data underlying this figure can be found in S3 and S4 Data. HCA, hierarchical cluster analysis; PCA, principal component analysis; SCFM2, synthetic cystic fibrosis medium 2.

https://doi.org/10.1371/journal.pbio.3002781.g005

To characterize the effect of the aceE and aceF mutations more broadly, we performed whole cell proteomics on the aceE and aceF recombinant strains (and on the PAO1 progenitor) in presence and absence of acetate. In total, we quantified 3,246 proteins and identified 449 as differentially expressed in at least 1 comparison of either the mutant versus wild-type or of the mutant in presence versus absence of acetate ( S3 Data ). In the absence of acetate, the proteome of the aceE and aceF mutants was clearly different compared with that of the wild-type progenitor, PAO1 ( Fig 5D ). PC1, which encompasses >50% of the variance in the data set, separated the aceF mutant proteome from that of the wild type, whereas the proteome of the aceE mutant was more similar to that of the wild type, separating along PC2 (accounting for just 20% of the total variance in the data set, Fig 5D ). As previously noted, supplementation of the growth medium with acetate altered both the aceE and aceF mutant proteomes, with both moving closer to that of the wild type ( Fig 5D ). This effect was greater in the aceF mutant, where growth in acetate decreased the number of up-regulated proteins ( cf . PAO1) by more than half (from 301 to 126) ( Fig 5E ). Unsurprisingly, the expression of several enzymes involved in pyruvate and acetyl-CoA metabolism (e.g., lactate dehydrogenase (41), acetyl-coenzyme A synthetase (6), citrate synthase (28)) return to wild-type levels following the addition of acetate (Figs 5F and S7A ). Similarly, for the clinical isolates DK12_L and DK36_L, the categories of proteins involved in amino acid/lipid metabolism and energy conversion presented the largest number of differentially expressed proteins ( S7B Fig ). Moreover, proteins involved in terpene, propionate, isoprenoid and branched-chain amino acid metabolism (all of which are directly connected to pyruvate and acetyl-CoA metabolism) were statistically enriched in the PAO1-derived aceF mutant, indicating a reorganization of both central and peripherical pathways to cope with reduced synthesis of acetyl-CoA ( S7C Fig ). The PAO1-derived aceE mutant also shows an extreme down-regulation of the oxygen-sensing transcriptional regulator Dnr, which is known to be required for denitrification but also regulates acetate metabolism and the T6SS [ 50 – 52 ]. This mirrors the down-regulation seen specifically for cluster B of the clinical isolates ( Fig 3C ), which may suggest a role for Dnr down-regulation in metabolic specialization ( S7D Fig ).

Importantly, the aceE and aceF mutants showed increased expression of proteins involved in alginate production and T6SS, as well as decreased expression of proteins involved in the T3SS, including the secreted factor ExoT (in strains aceF and DK12_L) ( Fig 5G and S3 Data ). This indicates that the aceE and aceF mutations contribute to the expression profile of virulence determinants shown by the late clinical isolates of DK12 and DK36 (Figs 4 and 5G and S3 Data ).

Pyruvate dehydrogenase mutations modulate pathogenicity in ALI culture infections

To test whether the secretion of pyruvate and the reduced expression of virulence determinants in the PAO1-derived aceF and aceE mutants lead to reduced infectivity, we performed host-bacteria infections using an ALI infection model system. This model system is composed of mucociliated differentiated airway epithelial cells which represent the airways and provides insights into the host response including epithelial damage and recruitment of the immune system. Overall, the aceF mutant displayed a broad suppression of virulence, including reduced epithelium damage and innate immune recognition during the infection ( Fig 6 ). The transepithelial electrical resistance (TEER) which quantifies the integrity and permeability of the epithelial layer, LDH release which quantifies the epithelium cellular damage, and the bacterial count which quantifies the growth and penetration of the bacteria through the epithelium to the basolateral side of the ALI transwells, all showed reduced values after 14 h of infection by the aceF mutant, compared with the PAO1 progenitor ( Fig 6A ). This is in line with the behavior of a Δ pscC mutant defective in T3SS, which also shows severely reduced virulence ( Fig 6A ). By contrast, the aceE mutant elicited epithelial damage similar to PAO1, indicating that the mutation does not influence bacterial penetration ( Fig 6A ). This is consistent with the proteomic analysis showing no significant differential expression in this strain of proteins involved in the T3SS, which is the leading cause of epithelial damage ( Fig 5G ). However, both the aceE and aceF mutants, together with the Δ pscC mutant, elicit lower interleukin 8 (IL-8) release, a cytokine that is secreted by epithelial cells and is necessary for the recruitment of the immune system at the site of infection ( Fig 6B ). These results are corroborated by confocal microscopy data, which show a similar colonization profile on the airway epithelium of the wild-type and aceE mutant, and of the aceF and Δ pscC mutant, respectively. It is worth noting that nuclei shedding, seen as nuclei (in blue) being pushed to the apical side of the epithelium (in red), was observed only for PAO1 and not for the aceE mutant ( Fig 6C ), which may suggest a slight reduction in the aggressiveness of the infection. This mechanism of reduced cellular damage and reduced inflammation seems not to depend on the growth defect associated with the aceF mutant, since the Δ pscC mutant shows comparable aggressiveness to the aceF mutant but a wild-type–like growth rate.

(A) Mean ± SEM of TEER (Ω∙cm 2 ), LDH release, and CFUs in the apical and basolateral ALI compartments after 14 h of infection in fully differentiated BCi-NS1.1 cells. (B) IL-8 release into the basolateral media indicating inflammation caused by the invading bacteria. Icons represent each biological replicates. Mock represents un-infected control cells. Statistical significance was determined by two-way ANOVA for TEER and CFU measurements and one-way ANOVA for LDH and IL-8 measurements and indicated as * ( p < 0.05), ** ( p < 0.01), *** ( p < 0.001), and **** ( p < 0.0001). (C) Confocal images of ALI transwells following infection with P . aeruginosa in green (GFP), epithelium in red (Phalloidin), and nuclei in blue (To-pro). Scale bar = 40 μm. The data underlying this figure can be found in S4 Data . ALI, air–liquid interface; CFU, colony-forming unit; TEER, transepithelial electrical resistance.

https://doi.org/10.1371/journal.pbio.3002781.g006

Importantly, in our longitudinal collection of clinical P . aeruginosa strains, 18 different lineages (specific clone types obtained from different pwCF; 89 isolates) harbored mutations in the aceE and aceF genes and/or carried additional mutations in T3SS genes [ 39 ]. In 9 lineages, PDHc mutations are present alone or anticipate additional T3SS mutations. In 7 lineages, T3SS mutations precede PDHc mutation or both systems are mutated at the same time (4 lineages) already in the first isolate of the lineage ( S8 Fig ).

Altogether, these results suggest a role for the aceE and aceF mutations in persistent infections and a link between pyruvate metabolism and virulence.

Metabolism constitutes a central process which is specifically regulated and optimized according to the nutritional resources available and the phenotypic requirements of the cell. Several studies have shown that metabolic specialization occurs in clinical isolates of P . aeruginosa [ 5 , 10 , 24 , 25 , 53 ]. However, little is known about whether this metabolic specialization is simply an accommodation for other significant fitness gains, i.e., metabolic compensation for phenotypes with high fitness cost, or whether it is specifically selected to enable increased persistence in the host.

By analyzing a range of longitudinal clinical isolates of P . aeruginosa , we found that all infection lineages underwent a substantial metabolic rewiring. This includes CF-specific patterns of carbon source (amino acids, sugars, and organic acids) assimilation and fermentation (acetate secretion). These metabolic changes, in turn, are associated with changes in proteome allocation potentially driven by mutations in metabolic genes. Adapted strains harboring mutations in genes ( aceE and aceF ) encoding the PDHc are characterized by a chronic-like phenotype (T3SS down-regulation, up-regulation of the T6SS, phenazines, the alginate biosynthetic pathway, and slow growth) suggesting that these mutations confer a selective advantage in the host [ 4 ]. Moreover, adapted strains secrete large amounts of pyruvate, which is an important mediator of inflammation in the host [ 37 ]. Reconstitution of clinical aceE and aceF mutations in a defined laboratory strain (PAO1) revealed that these mutations elicit reduced stimulation of the immune system and reduced epithelial layer penetration when compared with the isogenic wild-type progenitor strain (PAO1), which exhibits a highly virulent phenotype similar to that associated with acute infection. Importantly, this phenotype is only evident in the ALI infection model system and not in standard laboratory conditions since virulence and pathogenicity are only elicited in the presence of the host. These findings are consistent with the notion that strains with reduced virulence are selected during within-patient evolution in pwCF to reduce host tissue damage and limit the activity of the immune system [ 54 ]. Such reduced aggressiveness in the aceF mutant appears to be mediated by lower expression of the T3SS effector, ExoT, which induces apoptosis in host cells [ 16 , 17 ]. This is consistent with previous studies showing that mutations in the PDHc-encoding genes lead to repression of T3S in P . aeruginosa [ 31 , 55 ]. Mutations in the enzyme iso citrate lyase, involved in the glyoxylate shunt of the TCA cycle, have also been associated with reduced T3SS activity, suggestive of a direct correlation of metabolism to virulence [ 29 ]. This is the case for the DK15 late isolate in our collection, which shows reduced expression of iso citrate lyase together with suppression of the ExoT exotoxin. Given the tendency in our collection for aceE and aceF mutations to precede mutations in the T3SS-encoding genes, it is interesting to consider if mutations in the PDHc and/or modified PDHc activity could play a broader role in promoting a persistence phenotype during evolution in the host by reducing T3SS-dependent virulence. However, further investigation would be needed to establish such a mechanism. Nevertheless, the diversity of aceE and aceF mutations indicate that the activity of the PDHc can be fine-tuned accordingly to the specific host environment and consequently lead to different infection outcomes. Indeed, while the aceE mutant (carrying just an SNP) shows comparable aggressiveness to PAO1 but reduced recruitment of the immune system (IL-8 secretion), the aceF mutant (carrying a frameshifting indel) combines the effect of the reduced aggressiveness and that of the reduced immune recruitment.

It is worth noting that pyruvate, in addition to its crucial role as intermediate of central carbon metabolism, is a potent scavenger of ROS which use has been suggested as an anti-inflammatory treatment in several diseases [ 36 , 56 ]. Due to the sustained activation of neutrophils and other defects in the homeostatic processes of pulmonary epithelia in CF airways, an abnormal flux of ROS is present which can be alleviated by the bacterial pyruvate secretion [ 57 ]. During exacerbation events, the concentration of pyruvate increases (0.3 mM) which might counterbalance the increased inflammation [ 58 ]. Indeed, it has been shown that treatment of human lung epithelial cells with pyruvate reduces the inflammatory response by reducing the production of IL-8 [ 59 , 60 ]. It induces a potent shutdown of the response to LPS in dendritic cells [ 61 ] and reduces alveolar tissue destruction in a COPD model system [ 62 ]. Similarly, inhibition of the PDHc and redirection of the pyruvate flux has been suggested both as a potential anti-inflammation therapy for chronic metabolic diseases [ 63 ] and as a cancer treatment [ 64 ]. Moreover, fine-tuning of the flux through the mitochondrial PDHc in macrophages and dendritic cells is key for regulating their polarization and thus the balancing between pro- and anti-inflammatory responses [ 38 , 65 ]. LPS-induced polarization of macrophages is prevented by pharmacological inhibition of pyruvate import into the mitochondria [ 66 ]. Moreover, flux through the PDHc, possibly altered by the pyruvate secreted by the aceE and aceF mutants, controls the production of the antibacterial immunometabolite, itaconate, by macrophages [ 67 – 70 ]. Therefore, the reduced recognition from the epithelial layer of the aceE and aceF mutants in this study may depend on the secretion of bacterial pyruvate, which is known to mediate an important inflammatory response in both the epithelium and the immune cells [ 37 , 38 , 56 ].

The metabolic specialization and specifically the PDHc dysregulation seen in the clinical isolates may, therefore, serve as a mechanism for ensuring active pyruvate secretion and suppression of the immune system by a mechanism of cross-feeding. The evolutionary benefit of such a mechanism of host–pathogen crosstalk during persistent infections may explain why, in many cases, ROS produced in the CF airways appear to select for metabolic specialization in P . aeruginosa [ 71 ]. Moreover, the consequent phenotypic change of metabolically adapted strains limits epithelial damage to avoid any further inflammation. Indeed, changes in the T3SS, the T6SS, phenazines, alginate biosynthesis, and growth physiology are all hallmarks of a transition from an acute (high virulence) to a chronic (low virulence) phenotype which are specifically selected for during colonization of pwCF [ 4 , 54 ]. These combined mechanisms of cross-feeding and reduced virulence in turn promote persistence in the host by a change in the infection microenvironment which can benefit the entire population of infecting bacteria. Moreover, they can provide a rationale for the broad accumulation and maintenance of PDHc mutations in P . aeruginosa clinical CF isolates and the reduced expression of the PDHc in sputum samples [ 39 , 47 ]. This suggests a highly complex and near universal role of the pyruvate node in regulation of both host and pathogen (and their respective functions in vivo) through a conserved central metabolite. Such a mechanism might be engaged years before the development of antibiotic resistance, because of its high relevance for the establishment of a chronic phenotype.

The mechanism of metabolic specialization presented here is only one of many different mechanisms that P . aeruginosa might use for long-term survival in the host. However, the connection between metabolic behaviors and virulence phenotype (acute or chronic) are still poorly characterized, limiting our capacity to design treatments that counteract such processes. While our work focuses on a limited collection of clinical strains, a systematic analysis of the heterogeneous populations infecting pwCF is necessary to fully evaluate the role of metabolic specialization in persistence. Designing new methods, e.g., for high-throughput metabolomics analyses under in vivo-like conditions, such as those recreated by ALI or organoid cultures [ 71 ] which go beyond the limitation of laboratory conditions, will allow the characterization of new persistence mechanisms by clarifying the relationship between the host, the pathogen, and other microbial species of the host microbiota. Indeed, laboratory conditions and classical phenotype screening approaches are inadequate to unravel complex host–pathogen interactions. Still, it is important to note that the pyruvate secretion and the associated phenotypic changes described here might be beneficial not only for the PDHc mutated population but also for other nearby “wild-type” cells, since they alleviate the airway inflammation and limit the recruitment of the immune system. Importantly, metabolic specialization might be equally relevant in other persistent infections such as those caused by Escherichia coli , Staphylococcus aureus , and Mycobacterium tuberculosis which are similarly threatening and difficult to tackle [ 2 ]. Finally, understanding specific mechanisms linking metabolism, energy balance and virulence, and most importantly how the relationship between the host and the pathogen changes during an infection, could provide new opportunities for more efficient and/or complimentary treatments, beyond the classical antibiotic treatment, which are greatly needed.

Strains and media

The collection of clinical isolates used in this study is a subset of the collection published in Marvig and colleagues and Bartell and colleagues [ 9 , 39 ]. The local ethics committee at the Capital Region of Denmark (Region Hovedstaden) approved the use of the stored P . aeruginosa isolates (registration number H-21078844). For information about specific clinical isolates, see S1 Table .

For construction of aceE , aceF , and Δ pscC mutant strains, derivatives of pACRISPR were constructed with the Uracil Specific Excision Reagent (USER) cloning [ 72 ]. The plasmids and primers used are listed in S1 and S2 Tables, respectively. Target DNA fragments were amplified from gDNA of clinical strains DK12 Late and DK36 Late using the Phusion U polymerase kit (Thermo Fisher Scientific, United States of America). For the Δ pscC mutant strain primers were designed to create matching short overlaps between the DNA fragments that should be stitched together in the final plasmid. The resulting products were treated with FastDigest DpnI enzyme (Thermo Fisher Scientific, USA) and ligated with the USER Enzyme (New England Biolabs, USA). PCR was performed on random E . coli DH5alpha colonies using OneTaq 2X Master Mix (New England Biolabs, USA) to confirm the correct insertions and further sequenced with Sanger method (Eurofins Scientific, Luxembourg). aceE , aceF , and Δ pscC mutations in the genome of P . aeruginosa PAO1 were introduced using the previously developed CRISPR-Cas9 system and following the indicated protocol [ 73 ] with the addition of 0.2% arabinose in the growth medium during transformation. To confirm the presence of desired mutations, PCR fragment of the genome around the mutations was amplified and sequenced with Sanger sequencing (Eurofins Scientific, Luxembourg). To perform confocal microscopy analyses, strains were tagged with GFP using 4 parental mating using a mini-Tn7 delivery method [ 74 ] but for strains aceE and aceF with the additional supplementation of 5 mM acetate to ensure growth of PDHc dysregulated target strains. Transformants were identified by green fluorescence and validated by comparing growth rate to the untagged target strain.

To complement the mutation in the PDH ( aceE gene), we constructed a complementation strain aceE(rev) containing a wild-type copy of the aceE gene under the control of its native promoter and integrated it into the genome of the aceE mutant using the mini-Tn7 delivery method [ 74 ] as above. The plasmid pIP281 carrying the aceE gene and the Tn7 system was assembled with USER cloning [ 72 ]. The plasmids and primers used are listed in S1 and S2 Tables, respectively.

Bacteria were grown in Synthetic Cystic Fibrosis Media 2 (SCFM2) [ 75 ]. To reduce viscosity and allow for HPLC analysis, DNA and mucins were excluded [ 76 ]. Cultures were grown at 37°C and 250 rpm.

Dynamic exo-metabolomics analyses

Sampling of supernatants was performed in 96-well deep well plates (Cat. No. 0030502302; Eppendorf, Hamburg, Germany) with an air:liquid ratio of 1:1 using a high-throughput dilution-based growth method as previously described [ 77 ]. Immediately before sampling, OD 600 of the cultures were measured in a Synergy MX microtiter plate reader (BioTek Instruments, Winooski, Vermont, USA). Supernatants were stored at −80°C until HPLC-analysis. For organic acids and sugars (glucose, lactate, formate, acetate, and pyruvate), a Dionex Ultimate 3000 system (Thermo Scientific, Waltham, USA) with a HPx87H ion exclusion column (125–0140, Aminex, Dublin, Ireland), equipped with a guard column (125–0129, BioRad, Hercules, California, USA) and guard column holder (125–0131, BioRad, Hercules, California, USA) was used. Samples were injected with an injection volume of 20 μl and eluted, using a 5 mM H 2 SO 4 mobile phase, at an isocratic flow of 0.6 ml min −1 at 45°C for 30 min. Pyruvate was analyzed by UV detection at a wavelength of 210 nm, using a System Gold 166 UV-detector (Beckman Coulter, Brea, USA), while the rest of the metabolites were analyzed by RI detection, using a Smartline RI detector 2300 (KNAUER Wissenschaftliche Geräte, Berlin, Germany). For amino acids (aspartic acid, glutamic acid, serine, histidine, glycine, threonine, arginine, alanine, tyrosine, valine, phenylalanine, isoleucine, leucine, lysine, and proline), 20 μl of the thawed sample was diluted 1:10 by mixing with 80 μl Ultrapure MilliQ water and 100 μl of internal standard (20 μg/ml 2-aminobutyric acid and sarcosine) before injection of 56.5 μl into the instrument. Prior to injection into the column, derivatization of amino acids was performed in the HPLC-instrument by automatic mixing with the following eluents: (i) 0.5% (v/v) 3-mercaptopropionic acid in borate buffer 0.4 M at pH 10.2; (ii) 120 mM iodoacetic acid in 140 mM NaOH; (iii) OPA reagent (10 mg/ml o -phtalaldehyde and 3-mercaptopropionic acid in 0.4 M borate buffer); (iv) FMOC reagent (2.5 mg/ml 9-fluorenylmethyl chloroformate in acetonitrile); and (v) buffer A (40 mM Na 2 HPO 4 , 0.02% (w/v) NaN 3 at pH = 7). Following derivatization, the samples were separated isocratically on a Dionex Ultimater 3000 HPLC with fluorescence detector (Thermo Scientific, Waltham, USA) through a Gemini C18 column (00F-4439-E0, Phenomenex, Værløse, Denmark) equipped with a SecurityGuard Gemini C18 guard column (AJ0-7597, Phenomenex, Værløse, Denmark) with 5 mM H 2 SO 4 at a flowrate of 1 ml min −1 at 37°C for 31 min. Amino acids were detected using an UltiMate 3000 Fluorescence variable wavelength UV detector (FLD-3400RS, Waltham, Massachusetts, USA).

Proteomic analyses

To maintain a 1:1 air:liquid ratio, as in the metabolomic analysis, 25 ml SCFM2 cultures were inoculated from ON cultures in 50 ml Falcon tubes at OD 600 of 0.05 and then incubated in an orbital shaker at 37°C and 250 rpm until sampling by centrifugation at mid-exponential phase. Five biological replicates were analyzed for PAO1 and the clinical isolates while 3 biological replicates for the PAO1 derivative strains aceE and aceF both in presence and absence of 5 mM acetate. Pellets were washed twice with PBS and stored at −80°C until protein extraction. No specific preparation or enrichments were performed to identify secreted proteins. For experiment with clinical isolates, protein extraction was done in Lysis buffer A (100 mM Tris, 50 mM NaCl, 1 mM tris(2-carboxyethyl)phosphine (TCEP), 10% glycerol, pH = 7.5) with cOmplete Mini protease inhibitor (Roche) by sonication at amplitude 10 for 3 × 10 s cycles with 20 s cooling between, followed by acetone precipitation and resuspension in Lysis buffer B (6 M Guanidinium hydrochloride, 5 mM TCEP, 10 mM chloroacetamide, 100 mM Tris-HCl, pH = 8.5). For experiment with PAO1 derivative mutant strains, protein extraction was done in Lysis buffer B by bead beating with 3-mm zirconium oxide beads at 99°C for 5 min in a Tissuelyzer (Retsch, MM 400) at 25 Hz, then boiled, still at 99°C, in heat block (Eppendorf Thermomixer C) for 10 min while shaking/mixing at 2,000 rpm. In both cases, protein concentrations were determined by micro BCA Protein Assay Kit (Thermo Scientific, prod #23235) and 100 μg protein was used for trypsin digest in Trypsin/Lys-C Buffer (0.1 μg/μl trypsin, 50 mM Ammonium Bicarbonate). The reaction was stopped by addition of 10 μl 10% TFA and samples were desalted by stagetipping with C18 filter plugs (Solaμ HRP 2 mg/1 ml 96-well plate, Thermo Scientific). Peptide samples were stored in 40 μl of 0.1% formic acid at 4°C until LC-MS analysis.

For the experiment with clinical isolates, LC-MS/MS was carried out using a CapLC system (Thermo Fisher Scientific, Waltham, USA) coupled to an Orbitrap Q-exactive HF-X mass spectrometer (Thermo Fisher Scientific, Waltham, USA). The first samples were captured at a flow of 10 μl/min on a precolumn (μ-pre-column C18 PepMap 100, 5 μm, 100 Å) and then at a flow of 1.2 μl/min. Peptides were separated on a 15 cm C18 easy spray column (PepMap RSLC C18 2 μm, 100Å, 150 μm × 15 cm). The applied gradient went from 4% acetonitrile in water to 76% over a total of 60 min. MS-level scans were performed with Orbitrap resolution set to 60,000; AGC Target 3.0e6; maximum injection time 50 ms; intensity threshold 5.0e3; dynamic exclusion 25 s. Data-dependent MS2 selection was performed in Top 20 Speed mode with HCD collision energy set to 28% (AGC target 1.0e4, maximum injection time 22 ms, Isolation window 1.2 m/z). For the experiment with the aceE and aceF mutants, peptides were loaded onto a 2 cm C18 trap column (Thermo Fisher 164946), connected in-line to a 15 cm C18 reverse-phase analytical column (Thermo EasySpray ES904) using 100% Buffer A (0.1% formic acid in water) at 750 bar, using the Thermo EasyLC 1200 HPLC system, and the column oven operating at 30°C. Peptides were eluted over a 70 min gradient ranging from 10% to 60% of 80% acetonitrile, 0.1% formic acid at 250 nL/min, and the Orbitrap Exploris instrument (Thermo Fisher Scientific) was run in DIA mode with FAIMS Pro Interface (Thermo Fisher Scientific) with CV of −45 V. Full MS spectra were collected at a resolution of 120,000, with an AGC target of 300% or maximum injection time set to “auto” and a scan range of 400 to 1,000 m/z. The MS2 spectra were obtained in DIA mode in the orbitrap operating at a resolution of 60.000, with an AGC target 1,000% or maximum injection time set to “auto,” and a normalized HCD collision energy of 32. The isolation window was set to 6 m/z with a 1 m/z overlap and window placement on. Each DIA experiment covered a range of 200 m/z resulting in 3 DIA experiments (400 to 600 m/z, 600 to 800 m/z, and 800 to 1,000 m/z). Between the DIA experiments a full MS scan is performed. MS performance was verified for consistency by running complex cell lysate quality control standards, and chromatography was monitored to check for reproducibility.

Phenotypic characterizations

Growth curves were performed by inoculating 1 μl of overnight culture in 149 μl of media, using 96-well microtiter plates (Cat. No. 650001; Greiner Bio-One, Kremsmünster, Austria), covered with plate seals (Ref. 4306311, Thermo Fisher Scientific, United Kingdom) and incubated at 37°C and 250 rpm in a BioTek ELx808 Absorbance Microtiter Reader (BioTek Instruments, Winooski, Vermont, USA) for 24 to 48 h. Antibiotic MICs were determined by microdilution. ON cultures were standardized to OD 600 = 0.5 and diluted 1:2,500 to reach 5 × 10 5 CFU/ml in fresh SCFM2 media (LB for Azithromycin). Growth assays were performed at increasing antibiotic concentrations and MIC determined based on final OD. Resistance to oxidative stress was measured as the diameter of clearance zones around diffusion disks saturated with 5 μl of fresh 30% H 2 O 2 after 24 h incubation at 37°C on LB agar plates cast with 3 ml overlay agar containing 100 μl of LB ON culture standardized to OD 600 = 1. Pyoverdine production was measured as the fluorescence at 400/460 nm excitation/emission on a Synergy H1 Hybrid Multi-Mode Reader (BioTek Instruments, Winooski, Vermont, USA) of supernatants normalized against OD 600 of ON cultures in King’s B medium. Biofilm formation assay was done as previously described for NUNC peg lids [ 9 ] (NUNC cat no. 445497). Motility was measured as the diameter of the motility zone around single colonies deposited in middle layer of 0.3% (swimming), surface layer of 0.6% (swarming), or bottom layer of 1.5% (twitching) LB agar motility plates, after incubation at 37°C for 24 to 48 h. For PAO1 wild-type and aceEF mutant strains, pyruvate secretion was determined from supernatants of samples taken after 0, 4, 8, and 24 h of growth in SCFM2. Supernatants were stored at −80°C and analyzed by the HPLC method also used for exo-metabolomics.

Infection of ALI cultures

For the ALI infections, the BCi-NS1.1 cells were used [ 78 ]. Cells were cultured in Pneumacult-Ex Plus medium (STEMCELL Technologies, 05040) supplemented with Pneumacult-Ex 50x supplement (STEMCELL Technologies, 05008), 96 ng/ml hydrocortisone (STEMCELL Technologies, 07925), and 10 μm Y-27632 ROCK inhibitor (Bio-Techne #1254/10) in a 37°C, 5% CO 2 humidified incubator. Following expansion, 1.5 × 10 5 cells were seeded onto 6.5-diameter-size transwells with 0.4 μm pore polyester membrane inserts (Corning Incorporated, 3470) previously coated with human type I collagen (Gibco, A1048301). ALI was established once cells reached full confluency by removing media from the apical chamber and replacing media in the basolateral chamber with Pneumacult-ALI maintenance medium (STEMCELL Technologies, 05001). Pneumacult-ALI maintenance medium was supplemented with Pneumacult-ALI 10× supplement (STEMCELL Technologies, 05003), Pneumacult-ALI maintenance supplement (STEMCELL Technologies, 05006), 480 ng/ml hydrocortisone, and 4 μg/ml heparin (STEMCELL Technologies, 07980). ALI cultures were grown in a 37°C, 5% CO 2 humidified incubator for 30 days, with media replacement every 2 to 3 days. Epithelial polarization was monitored by measurements of the TEER using a chopstick electrode (STX2; World Precision Instruments). Following 15 days under ALI conditions, the apical surface was washed with PBS every media change to remove accumulated mucus. Biological replicates of bacterial strains were obtained from single colonies on LB agar plates grown ON as precultures in LB and then diluted to an OD 600 of 0.05 before sampling at mid-exponential phase by centrifugation, washing with PBS and resuspending in PBS at a density of 10 5 CFU/ml. Fully differentiated BCi-NS1.1 cells were inoculated with 10 3 CFU from the apical side, diluted in 10 μl PBS. Control wells were incubated with bacteria-free PBS. Cells were incubated for 14 h at 37°C, followed by addition of 200 μl PBS to the apical side and measurement of the TEER. CFUs were determined by platting 10 μl of 6-fold serial dilutions on LB-agar plates in technical triplicates both for the initial inoculum, as well as for the apical and basolateral solutions following TEER measurements. The basolateral media was also used for measurements of LDH and IL-8 release in technical triplicates, using the Invitrogen CyQUANT LDH Cytotoxicity Assay Kit (Invitrogen, C20301) and Human IL-8/CXCL8 DuoSet ELISA Kit (R&D Systems, DY208) according to the manufacturer’s instructions. BCi-NS1.1 cells on transwell inserts were rinsed once with PBS and fixed by adding 4% (wt/vol) paraformaldehyde (PFA) to both apical and basolateral chambers for 20 min at 4°C. After washing, cells were permeabilized and blocked for 1 h with a buffer containing 3% BSA, 1% Saponin, and 1% Triton X-100 in PBS. Cells were stained on the apical side with Phalloidin-AF488 (Invitrogen, 65-0863-14) and TO-PRO3 (Biolegend, 304008) diluted in a staining buffer (3% BSA and 1% Saponin in PBS) at a 1:500 dilution for 2 h at room temperature. Transwells were removed from their supports with a scalpel and mounted on glass slides with VECTASHIELD Antifade Mounting Medium (VWR, VECTH-1000). Images were acquired with a Carl Zeiss LSM 510 Confocal Laser Scanning Microscope (40× magnification, 1.3 oil) and analyzed using the ImageJ software.

Data analysis

For exometabolomics, all chromatograms were analyzed and used to construct standard curves constructed for absolute quantification of concentrations for all 20 metabolites, using software Chromeleon v7.2.9. All other analysis of exo-metabolomic data was done in JMP Pro 15.0. To compare between different strains and media batch-effects, OD 600 was normalized against final OD 600 of that strain in the experiment and concentrations were normalized against specific concentrations in SCFM2 controls for each batch. Missing values were replaced with 20% of lowest value detected for any metabolite in any sample. PCA and HCA were done on C metabolite of all quantified metabolites for 16 samples (8 time points from biological duplicates) of each strain. PCA and iterative k-means clustering was done on covariance using JMP Pro 15.0 software. HCA was performed using the “average” clustering method and “correlation” for distance using the R package pvclust 2.2. Reliability of the clusters were analyzed by 10,000 bootstraps. Metabolite half-life (OD 50 ), defined as the normalized OD 600 -value, where 50% of the starting concentration of a metabolite is present, was calculated from the sigmoidal mechanistic growth model (Equation: a(1 –bExp(–cx)) where a = asymptote, b = scale, and c = rate) [ 79 ]. Naïve and adapted metabotypes were designated based on PCA, iterative k-means clustering (range 3 to 10) followed by curation based on HCA. Specifically, iterative k-means clustering readily identified adapted metabotypes 2 and 3 based on 3 clusters and the highest Cubic Cluster Criterion statistical analysis. Cluster 1 is determined based on 5 k-means and agrees with the grouping generated by HCA followed by bootstrapping. Differences of metabolic preferences between metabotypes were analyzed by comparing the assimilation and secretion profiles between strains using OD 50 values. The net balance of carbon sources assimilated during the growth was calculated by summing the amount (mM concentration) of assimilated metabolites and subtracting the secreted ones. Metabolomics data are enclosed in S1 Data .

For the experiment with clinical isolates, proteomic raw files were analyzed using Proteome discoverer 2.4 software with the following settings: Fixed modifications: Carbamidomethyl (C) and Variable modifications: oxidation of methionine residues. First search mass tolerance 20 ppm and a MS/MS tolerance of 20 ppm. Trypsin as enzyme and allowing 1 missed cleavage. FDR was set at 0.1%. The Match between runs window was set to 0.7 min. Quantification was only based on unique peptides and normalization between samples was based on total peptide amount. For the experiment with mutant strains, raw files were analyzed using Spectronaut (version 16.2). Dynamic modifications were set as Oxidation (M) and Acetyl on protein N-termini. Cysteine carbamidomethyl was set as a static modification. All results were filtered to a 1% FDR, and protein quantitation done on the MS1 level. The data was normalized by RT dependent local regression model [ 80 ] and protein groups were inferred by IDPicker. In both cases, spectra were matched against the P . aeruginosa PAO1 reference proteome (UniProt Proteome ID UP000002438). In both experiments, any protein that was not quantified in at least 3 of 5 or 2 of 3 biological replicates for all strains were excluded from the analysis. Using JMP Pro 15.0, abundances were normalized by Log 2 -transformation and biological replicates used for missing value imputation. PCA Wide was done on correlations using JMP Pro 15.0 and HCA was performed on mean Log 2 (abundance) and Log 2 (Fold change) for the relevant comparisons. HCA was performed using the “average” clustering method and “correlation” for distance using the R package pvclust 2.2. Reliability of the clusters were analyzed by 10,000 bootstraps. Differential expression was determined by two-way ANOVA with Tukey’s multiple comparisons test, using GraphPad Prism 9.3.1, and defined as those protein-comparisons where adjusted P value ≤0.05 and Log 2 (Fold Change) ≥ |0.6|. Enrichment analysis was done using the DAVID Functional Annotation Bioinformatics Microarray Analysis tool from lists of Locus Tags of proteins that were differentially expressed, separated into lists of up-regulated and down-regulated proteins, respectively, for each of the relevant strain comparisons. The reference genome was set as Pseudomonas aeruginosa . Proteomics data are enclosed in S2 and S3 Data.

Growth rates and maxOD were calculated in JMP Pro 15.0. Blanks were first subtracted from OD 600 -measurements and values converted to cm -1 (using Greiner dimensions for pathlength). The stationary phase was excluded, and growth rates were calculated by fitting the Exponential 3P model to the exponential phase (r 2 > 0.99). Mean ± SD was calculated from biological replicates. GraphPad Prism 9.3.1 was used for statistical analysis of biofilm formation, motility, redox sensitivity, pyoverdine production, and MICs (see S6 Fig for details), as well as for ALI infection experiments. ALI data are represented as mean ± SEM. Replicates represent independent experiments performed with cells from different passages. Statistical comparisons were calculated using two-way ANOVA for TEER and CFU measurements and one-way ANOVA for LDH and IL-8 measurements. Statistical significance was considered for p value < 0.05. All figures were finalized in Adobe Illustrator Artwork 27.0.

Supporting information

(A) Number of differentially expressed proteins (top) and Log 2 (Fold change) of individual differentially expressed proteins in late vs. early clinical isolates, separated by lineage and on the x-axis by COG categories. (B) Complete enrichment analysis showing fold-enrichments separated by comparison and lineage on x-axis and by KEGG and GO terms on y-axis. The data underlying this figure can be found in S2 Data .

https://doi.org/10.1371/journal.pbio.3002781.s001

Metabolic maps showing the differences between early clinical strains vs. PAO1 (left) and late vs. early clinical strains (right) in pathways related to the catabolism of nutrients present in SCFM2 through central carbon metabolism. Arrows represent individual metabolite-conversions colored by their pathway. Reactions are colored if the underlining enzyme is differentially expressed in more than one clone type—either with dashed (down-regulated) or full (up-regulated) lines, or both. Enzymes responsible for each reaction are indicated by numbered circles. For details on individual enzymes, see S3 Fig . Transporters are shown at the bottom as arrows crossing bacterial cell membrane with their specific transported metabolites. Asp, Glu, and acetate are labeled with * to indicate that they are represented in 2 places on the map, due to their involvement in different parts of central carbon metabolism. The data underlying this figure can be found in S2 Data .

https://doi.org/10.1371/journal.pbio.3002781.s002

(A) Metabolic map in the same style as S2 Fig for late clinical strains vs. PAO1. (B) Table of genes/Locus Tags of all metabolic enzymes included in the map. (C, D) Bar charts showing Log 2 (Fold change) of all differentially expressed proteins highlighted on the metabolic maps. Icons colored by lineage and separated into 3 comparisons on x-axis (Early vs. PAO1; Late vs. Early; Late vs. PAO1). Panel c contains metabolic enzymes and panel d contains transporters. The data underlying this figure can be found in S2 Data .

https://doi.org/10.1371/journal.pbio.3002781.s003

(A) Hierarchical cluster analysis in the same style as Fig 4C , showing differential expression in late vs. early clinical strains for several more virulence categories. (B) The Log 2 (Fold change) of differentially expressed proteins in late vs. early clinical strains separated into the relevant virulence categories and further separated into the 2 main clusters from panel a. Icons are colored by lineage and lines depict the difference in mean Log 2 (Fold change) between the 2 clusters. The data underlying this figure can be found in S2 Data .

https://doi.org/10.1371/journal.pbio.3002781.s004

Schematic of all 18 unique mutations observed for aceE (top) and aceF (bottom) in our collection of clinical isolates of P . aeruginosa . Gray lines indicate mutation site and letters indicate nucleotide-sequence colored by type of mutation. The data underlying this figure can be found in Marvig and colleagues [ 39 ].

https://doi.org/10.1371/journal.pbio.3002781.s005

(A) Complementation analysis of aceE mutant. Growth rate (hour -1 ) in SCFM for PAO1, aceE mutant and complementation strain aceE(rev) . Bars indicate mean ± SEM, with icons representing biological replicates. Statistical significance was assessed by one-way ANOVA with Tukey’s multiple comparisons test and indicated as * ( p < 0.05). (B) Potential for biofilm formation (surface attachment) of PAO1 wt (blue), as well as aceE (red) and aceF (green) mutant strains and DK12 and DK36 late clinical strains after 24 and 48 h, respectively. For the clinical isolates, strain is indicated on the x-axes and incubation time is indicated as blue (24 h) and red (48 h). Attachment is measured as the ratio of surface-attached cells (OD 590 ) to total number of cells (OD 600 ) after incubation. For PAO1 wt and aceE and aceF mutants, all strains were compared by one-way ANOVA. For clinical isolates, each strain was compared to itself after 24 and 48 h of incubation, respectively, using two-tailed unpaired parametric Welch t test. Significance is indicated as “ns” ( p > 0.05), * ( p < 0.05), ** ( p < 0.01), *** ( p < 0.001). Same coloring and statistical analysis for PAO1 and mutant strains in panels B–D. (C) Motility measured as the diameter (mm) of the zone of growth in motility plates (LB agar). Swimming motility was clearly visible after 24 h of incubation. Swarming and twitching plates required 48 h of incubation. (D) Redox sensitivity measured as the diameter of the clearance zone (mm) on bacterial lawns (LB agar) of each strain from H 2 O 2 diffusion disks after 24 h. (E) Pyoverdine production measured as relative fluorescence (F/OD 600 ) of each strain after 24 h of growth in King’s B medium. For panels A–D, icons indicate biological replicates and bars represent the mean ± SEM. (F) MICs of PAO1 wt and aceE and aceF mutant strains. Each cell represents the maxOD of growth curves under the given condition, following the color gradient to the right of each heatmap. The MIC is the concentration where no growth is observed (white). Each heat map shows the MIC for all 3 strains for a given antibiotic (Ceftazidime, Meropenem, Piperacillin, Azithromycin, Chloramphenicol, Ciprofloxacin, and Tobramycin). Concentrations (μg/ml) increase 2-fold downward on the vertical axis and the specific strain is given on the horizontal axis. Azithromycin MICs were determined in LB, while all other MICs were determined in SCFM2. Piperacillin was used in combination with the β-lactamase inhibitor Tazobactam. The data underlying this figure can be found in S4 Data .

https://doi.org/10.1371/journal.pbio.3002781.s006

(A) Metabolic map in the same style as S2 Fig for aceF mutant strain vs. PAO1 reference in SCFM2 in absence (left) and presence (right) of acetate. (B) Parallel plot showing the number of differentially expressed proteins, separated by metabolic COG categories, in aceE (cyan) and aceF (magenta) mutant strains vs. PAO1. (C) Enrichment analysis showing fold enrichment on x-axis separated by KEGG and GO terms on y-axis for aceE (left) and aceF (right) mutant strains vs. PAO1. (D) The -Log2(Fold change) of the Dnr transcriptional regulator in late vs. early clinical isolates (same holds for late strains vs. PAO1) for lineages in Cluster A (left) and B (middle), as well as in the aceE mutant strain (right) when compared to PAO1 wt in SCFM2 in absence (+/− mutation) and presence (Ace) of acetate or when compared to itself in presence of acetate (+/− Acetate). The data underlying this figure can be found in S3 Data .

https://doi.org/10.1371/journal.pbio.3002781.s007

S8 Fig. Longitudinal collection of P . aeruginosa clinical isolates from CF airways.

Icons indicate isolates carrying mutations in genes encoding PDHc (red squares) and/or T3SS (purple triangles) proteins, as well as isolates with no mutations in either (gray dots). Isolates are separated by Clone Type and further separated by patient-specific lineages within Clone Types (A, B, C). The x-axis shows the length of infection (years) since the first isolate of the given lineage. The data underlying this figure can be found in Marvig and colleagues [ 39 ].

https://doi.org/10.1371/journal.pbio.3002781.s008

S1 Data. Dynamic exometabolomics data.

Values correspond to mM concentrations of the analyzed metabolites by their collection OD and strain.

https://doi.org/10.1371/journal.pbio.3002781.s009

S2 Data. Proteomic data for the clinical isolates.

Values correspond to the Log 2 (Fold-change) for the indicated comparison. Statistical significance is reported by q values.

https://doi.org/10.1371/journal.pbio.3002781.s010

S3 Data. Proteomic data for the aceE and aceF recombinant strains.

https://doi.org/10.1371/journal.pbio.3002781.s011

S4 Data. Individual numerical values underlying all figures.

https://doi.org/10.1371/journal.pbio.3002781.s012

S1 Table. List of strains and plasmid.

https://doi.org/10.1371/journal.pbio.3002781.s013

S2 Table. List of primers.

https://doi.org/10.1371/journal.pbio.3002781.s014

S1 Text. Detailed information on metabolic preferences and convergent proteomic changes.

https://doi.org/10.1371/journal.pbio.3002781.s015

Acknowledgments

The Basal Cell Immortalized Non-Smoker 1.1 (BCi-NS1.1) cell line was a kind gift from Professor Ronald G. Cristal (Weil Cornell Medical College, New York, USA).

• View Article
• PubMed/NCBI
• Google Scholar
• 79. Behrends V, Williams HD, Bundy JG. Metabolic Footprinting: Extracellular Metabolomic Analysis. In: Filloux A, Ramos J-L, editors. Pseudomonas Methods and Protocols. New York, NY: Springer New York; 2014. p. 281–292. https://doi.org/10.1007/978-1-4939-0473-0_23 pmid:24818913

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• v.335(7632); 2007 Dec 15

Clinical Review

Cystic fibrosis, jane c davies.

1 Department of Gene Therapy, Imperial College, London

2 Department of Paediatric Respiratory Medicine, Royal Brompton Hospital, London

Eric W F W Alton

3 Department of Adult Respiratory Medicine, Royal Brompton Hospital, London

Andrew Bush

Summary points.

• Cystic fibrosis is the commonest inherited disease in white populations, with an incidence of 1 in 2500 newborns; over 7000 people in the United Kingdom currently have the disease
• Until recently, the diagnosis has been largely clinical, although the widespread implementation of a screening programme for newborns is now complete in the UK
• Cystic fibrosis is a multiorgan disease best managed in a multidisciplinary setting in conjunction with a specialist centre for cystic fibrosis, with treatment tailored to the individual
• The cornerstones of management are proactive treatment of airway infection and encouragement of good nutrition and an active lifestyle
• Conventional treatment has improved greatly over the past few decades; however, current treatments at best slow the decline in lung function. Newer approaches such as gene and small molecule based treatments may have more potential to halt disease progression

Over 7000 people have cystic fibrosis in the United Kingdom. It is the commonest genetically inherited disease in white populations (1 in 2500 newborns), although it is increasingly recognised as being important in non-white populations. However, most general practitioners have only one or two patients on their list, and as management generally takes place in specialist centres, many general paediatricians will be involved in the care of only a small number of patients.

Progress in our understanding of the disease and the impact of this on management has been rapid over the past 20 years. Cystic fibrosis used to be a digestive and lung disease of young children but more recently has become a complex, multisystem disease extending into adulthood; there will soon be more adults than children with the condition. The predicted median survival for babies born in the 21st century is now more than 50 years. 1 This increased survival—together with changes in standard treatment, the increasing implementation of newborn screening, and the focus on new therapeutic strategies—leads us to consider that an update on this albeit relatively rare disease may be of general interest.

What is the cause of cystic fibrosis?

Cystic fibrosis is an autosomal recessive disease. It is caused by mutations in the CFTR (cystic fibrosis transmembrane conductance regulator) gene. 2 The commonest mutation is the deletion of phenylalanine at codon 508 (phe508del, until recently known as ΔF508). This occurs in about 70% of patients with cystic fibrosis ( www.ornl.gov/sci/techresources/Human_Genome/posters/chromosome/cftr.shtml ). Over 1600 mutations of the CFTR gene have been described. Different mutations in this gene have varying effects on CFTR function and can result in different phenotypes of the disease. Some mutations will result in milder forms of the disease, although there is not enough evidence about these rarer mutations to counsel patients about their prognosis. The CFTR protein is expressed in many cells and has several functions, not all of which have been linked with disease. The primary function of the CFTR protein is as an ion channel that regulates liquid volume on epithelial surfaces through chloride secretion and inhibition of sodium absorption.

The commonly accepted explanation for airway disease in cystic fibrosis is the “low volume” hypothesis. A reduced volume of airway surface liquid causes failure of mucociliary clearance, the lungs’ innate defence mechanism. 3 The mucociliary dysfunction means that a patient with cystic fibrosis cannot effectively clear inhaled bacteria. In addition, there is an excessive inflammatory response to pathogens. For a given bacterial load, a person with cystic fibrosis will have up to 10 times more inflammation than a person with a lower respiratory tract infection but without the disease. This may also be the case for other insults such as viruses or even for airborne particulate matter and pollutants. The reasons for the excessive inflammatory response to pathogens are not fully understood. The abnormal composition and secretion of mucus may also be important. At birth, the airway is uninfected and probably uninflamed, although some controversy exists in this area, 4 but the end result of the abnormalities described above is irreversible airway damage with bronchiectasis and respiratory failure in most patients (fig 1 ​ 1). ). Ion and water abnormalities may also cause disease in other epithelia-lined organs (see tables 1 ​ 1, , 3 ​ 3, , 4 ​ 4 ).”

Fig 1 Severe bronchiectasis in end stage cystic fibrosis shown in chest radiograph (top) and computed tomogram (bottom). For reasons that are not fully understood, the upper lobes are often most severely affected, although the patient has severe bronchiectasis throughout the whole of the right lung. Note presence of indwelling intravenous catheter (a “port-a-cath”) on the right lateral chest wall

 Age related presentations of cystic fibrosis

 Gastrointestinal problems and their management

* Treatments for which consensus is lacking.

† Treatments based on randomised controlled trials or meta-analyses.

 Management of other common complications of cystic fibrosis

*Treatments for which consensus is lacking.

What are the clinical features and when should the diagnosis be considered?

Disease manifests in many organs, but most notably the upper and lower airways, pancreas, bowel, and reproductive tracts (table 1 ​ 1 ). 5 For most patients, lung disease is the most important problem in terms of symptoms and the treatment required and the fact that it is the most likely cause of death. Table 1 ​ 1 ​ outlines the clinical presentation, which varies according to age.

Fig 2 Finger clubbing indicates advance suppurative lung disease. It is not characteristic of asthma, with which older patients have sometimes been misdiagnosed

How is cystic fibrosis diagnosed?

The optimal diagnostic test for cystic fibrosis is the measurement of sweat electrolyte levels. 6 Patients with the disease have raised concentrations of sodium and chloride (>60 mmol/l, diagnostic; 40-60 mmol/l, intermediate (but more likely to be diagnostic in infants); <40 mmol/l, normal). However, undoubted cases of cystic fibrosis with normal sweat electrolytes have been described. Newer techniques have reduced the amount of sweat needed (fig 3 ​ 3). ). The test needs to be done by someone trained and experienced. For this reason the diagnosis will usually be made in secondary and tertiary centres, although primary care professionals play a vital role in identifying the patients who need investigation. In the rare cases where the diagnosis remains in doubt, other diagnostic tests are available—for example, measurement of the nasal potential difference to assess altered salt transport is available in a few specialist centres in the UK. 7

Fig 3 Child’s arm during sweat test with the macroduct system. After pilocarpine iontophoresis to stimulate sweating, the closed capillary collecting system is applied to the skin of the forearm. Sweat can be seen entering the tubing (blue); electrolyte analysis can be reliably performed on as little as 50 μl of sweat

The UK now has a programme for screening all newborns for cystic fibrosis using the Guthrie blood spot test. 8 The initial screen is for raised concentrations of immunoreactive trypsinogen. Positive samples will be tested for common CFTR gene mutations followed by a second screen for immunoreactive trypsinogen if required. Screen positive infants will be referred for sweat testing.

Screening programmes have been in place in some parts of the world for many years, but they may be inappropriate in countries with a low prevalence of CFTR gene mutations. The advantages of early diagnosis include nutritional benefits; early access to specialised care; a reduction in the time of diagnostic uncertainty; and the ability to counsel parents for prenatal testing.

Screening programmes have some negative aspects, however. Programmes will identify some healthy heterozygote carriers as potential patients. This may have psychological implications and stress for the family until the diagnosis is excluded. Moreover, some patients, even those with classic cystic fibrosis, will be missed.

Once a diagnosis has been confirmed, other family members may be offered screening. All siblings need to be screened for the disease, which may be presymptomatic or unrecognised. Asymptomatic adult relatives, may wish to be screened for carrier status to enable them to make informed choices about prenatal screening. In our experience, screening and counselling of other family membersis most readily facilitated through primary care but requires coordination between genetic laboratories to ensure rapid and cost efficient testing.

Most patients in the UK and Europe receive care coordinated by a tertiary cystic fibrosis centre, which improves outcomes. However, patients benefit greatly from links with and access to local care, in many cases having formalised “shared care” with local clinics. Primary care teams can provide valuable help with surveillance and early treatment of infection; dietary and nutritional support; and social and psychological support for patients and families. Primary care also provides continuity during the difficult transition from paediatric to adult care; an informative patient’s perspective of the issues encountered during this period has recently been published. 9

Much of the current clinical practice has evolved over decades without being subjected to high quality randomised controlled trials. Tables 2-4 ​ 2-4 ​ ​ outline the various treatments and indicate those that are based on randomised controlled trials, meta-analyses, or systematic review and those that are treatments for which we consider consensus is lacking.

 Management of cystic fibrosis lung disease

Bcc=Burkholderia cepacia complex; Pa=Pseudomonas aeruginosa;Sa=Staphylococcus aureus; Hi=Haemophilus influenzae; Sm=Stenotrophomonas maltophilia; Ax=Alcaligenes xylosoxidans.

*Strategies for which consensus is lacking.

†Strategies based on randomised controlled trials or meta-analyses.

Lung disease

The aims of treating the lungs at different stages of disease vary; table 2 ​ 2 outlines the conventional management at each of these stages. Many of the treatment options have been discussed in systematic reviews. Respiratory treatments represent the greatest challenge to patients and families: doing physiotherapy and taking inhaled drugs such as antibiotics often takes up a lot of time—more than an hour a day during periods of good health and much longer during a respiratory exacerbation.

Extrapulmonary disease

Patients with cystic fibrosis often have gastrointestinal problems; table 3 ​ 3 outlines the nature and management of these (management should be in close collaboration with a specialist dietician). Table 4 ​ 4 lists other complications of the disease, plus their management strategies.

Psychological issues

Cystic fibrosis clearly poses a huge burden to patients and families in terms of the life shortening nature of the disease, the time consuming treatments prescribed, and the ongoing morbidity. Times of particular stress include diagnosis, adolescence (when adherence to treatment can often be poor), and end of life. Support and coping strategies from clinical psychologists with experience of the disease are often invaluable. 20

Sources and selection criteria

We searched PubMed for the terms “cystic fibrosis”, “therapy”, “treatment”, “management”, “complications”, and “diagnosis” in various combinations. We also searched all entries under “cystic fibrosis” in the Cochrane Library. From this search, we selected randomised controlled trials, high quality journal reviews and meta-analyses. We also drew from our own personal archives of references from known leaders in this field.

Ongoing research

• Research is being conducted into gene therapy that aims at introducing a normal copy of CFTR into lung epithelial cells. 21 Achieving expression after repeat administration of viral vectors has been a major problem owing to immune recognition. Because of this, the UK Cystic Fibrosis Gene Therapy Consortium, which has been formed in recent years to develop cystic fibrosis gene therapy for clinical benefit is focusing current efforts on a non-viral approach
• New drugs to improve ion transport 22 and osmotic agents 23 to increase airway surface liquid are currently in phase II clinical trials, as are anti-inflammatory agents, mucolytics, and pseudomonas vaccines
• New methods for administering current agents, such as the development of dry powder formulations (ease of administration) and of liposomal preparations (enhanced activity) of antibiotics are being developed

Additional educational resources

• Kerem E, Conway S, Elborn S, Heijerman H; Consensus Committee. Standards of care for patients with cystic fibrosis: a European consensus. J Cyst Fibros 2005;4:7-26.
• Bush A, Alton E, Davies JC, Griesenbach U, Jaffe A. Cystic fibrosis in the 21st century (progress in respiratory research) . Basel: Karger, 2005.
• Cystic Fibrosis Trust, 11 London Road, Bromley BR1 1BY ( www.cftrust.org.uk/ )—Charity whose work includes research into the disease and support to patients and their families
• Cystic Fibrosis Foundation ( www.cff.org/ )—US non-profit organisation whose work includes research into the disease and support to patients and their families
• Genetics Home Reference ( http://ghr.nlm.nih.gov/condition=cysticfibrosis )—US government website supplying general scientific information on the disease
• Association of Clinical Biochemistry. Guidelines for the performance of the sweat test for the investigation of cystic fibrosis in the UK. 2003. http://acb.org.uk/docs/sweat.pdf
• Cystic Fibrosis Mutation Database ( www.genet.sickkids.on.ca/cftr/ )—Aims to provide researchers and other professionals with up to date information about individual mutations in the CFTR gene and phenotypic data
• Information on the UK newborn screening programme ( www.ich.ucl.ac.uk/newborn/cf/index.htm )
• Breathing Room ( www.thebreathingroom.org/cg )— Illustrations of many issues affecting patients and their carers
• UK Cystic Fibrosis Gene Therapy Consortium ( www.cfgenetherapy.org.uk )—The research programme of three leading gene therapy groups in the UK

Contributors: JCD wrote the original draft of this paper and all authors contributed to subsequent drafts. Jackie Francis provided certain photographic images. JCD acts as guarantor for the article.

Competing interests: JCD and EWFWA are members of the UK Cystic Fibrosis Gene Therapy Consortium.

Provenance and peer review: Commissioned; externally peer reviewed.