Altered iron metabolism in cystic fibrosis macrophages: the impact of CFTR modulators and implications for Pseudomonas aeruginosa survival

Cystic fibrosis (CF) is a genetic disease caused by mutations in the CF transmembrane conductance regulator (CFTR) gene, resulting in chronic bacterial lung infections and tissue damage. CF macrophages exhibit reduced bacterial killing and increased inflammatory signaling. Iron is elevated in the CF lung and is a critical nutrient for bacteria, including the common CF pathogen Pseudomonas aeruginosa (Pa). While macrophages are a key regulatory component of extracellular iron, iron metabolism has yet to be characterized in human CF macrophages. Secreted and total protein levels were analyzed in non-CF and F508del/F508del CF monocyte derived macrophages (MDMs) with and without clinically approved CFTR modulators ivacaftor/lumacaftor. CF macrophage transferrin receptor 1 (TfR1) was reduced with ivacaftor/lumacaftor treatment. When activated with LPS, CF macrophage expressed reduced ferroportin (Fpn). After the addition of exogenous iron, total iron was elevated in conditioned media from CF MDMs and reduced in conditioned media from ivacaftor/lumacaftor treated CF MDMs. Pa biofilm formation and viability were elevated in conditioned media from CF MDMs and biofilm formation was reduced in the presence of conditioned media from ivacaftor/lumacaftor treated CF MDMs. Defects in iron metabolism observed in this study may inform host–pathogen interactions between CF macrophages and Pa.

Experimental procedure. Experiments were performed in RPMI 1640 Glutamax with 50 µg/ml gentamicin with 10% HI-FBS. MDMs underwent 48-h pretreatment with 0.2% DMSO or combination 30 nM ivacaftor and 3 µM lumacaftor (modulators, Selleckchem, cat#S7059 and S1565). Modulator concentrations were based on CF subject drug plasma concentrations and those used in previously published works 27,52,53 . For lipopolysaccharide (LPS) and iron experiments, cells were washed after pretreatment and given 1 µg/ml E. coli Biofilm formation and CFU quantification. Pa clinical isolate strain DH1137 (a generous gift from Dr. Deborah Hogan, previously described as strain NC-AMT0101-2 55 ) was cultured in LB broth shaking overnight at 37 °C. Stationary-phase cultures were diluted 1:10 in 0.2% casamino acids, 0.4% l-arginine, 0.2% d-glucose, and 0.02% MgSO 4 supplemented M63 biofilm formation media (US Biological, cat#M1015). Bacteria were then cultured to optical density (OD) at λ = 600 nm equal to 0.05 (1 × 10 8 cells/ml). In vinyl 96-well plates (Ther-moFisher Scientific, cat#2101), 75 µl DH1137 was diluted with 25 µl MDM conditioned media and incubated with humidity for 6 h at 37 °C. After incubation, 50 µl supernatant from each well was collected and OD at λ = 600 nm (total bacterial density) was measured in a microplate reader. Additionally, supernatant from each well was diluted 1:1,000 in LB broth and streaked onto agar plates for quantification of viable bacterial colony forming units (CFUs). Biofilm assay plates were washed with water, stained with 125 µl/well 0.1% crystal violet for 10 min, washed, and air dried. Photographs of biofilms were taken prior their quantification. After solubilizing biofilms with 150 µl/well 30% glacial acetic acid in water, absorbance at λ = 550 nm was used to quantitate each sample. Biofilm photographs were cropped (Fig. 4d), photographs of all subjects and conditions tested are available in Supplementary Fig. S6, metrics such as contrast and brightness were not altered for publication.  Fig. 3a-d (all male CF subjects) show data from male and female subjects. The lysis buffer employed in this study does not allow for protein quantification. Although monocytes were seeded equally at 5 × 10 5 cells/well in 12-well culture plates, platelet contamination from the isolation procedure occasionally resulted in a crowding out of the monocytes and a reduced confluence of cells in some wells. We believe this to be the cause of the lower levels of β Actin in certain lanes of the western blots presented in Fig. 2a . Briefly, zero values and values that were missing due to high background on the blot were removed. Rare instances where the same patient was measured multiple times were not included in this analysis to simplify the analysis. Ratios for proteins of interest were made to β Actin values in the same lane and log base 2 transformed to improve visualization and achieve normality. Graphs depicting comparisons between CF and non-CF MDMs as well as responses to ivacaftor/lumacaftor were generated using the R package ggplot2_3.2 56 (https ://CRAN.Rproje ct.org/packa ge=gplot s, Fig. 2c-e). Statistically significant differences between CF and non-CF MDMs not exposed to ivacaftor/lumacaftor were assessed using Wilcoxon rank sum tests, a non-parametric alternative to t tests. Statistically significant differences between CF MDMs exposed to ivacaftor/lumacaftor were assessed by linear models that included factors for CF subject and exposure to ivacaftor/lumacaftor. In Supplementary  Fig. S4a, b, heatmaps were produced with the R package gplots 57 with the heatmap.2 function (gplots_3.0.1.2).

Results
Subject characteristics. Subject characteristics are summarized in Table 1. Ten cystic fibrosis (CF) subjects homozygous for the F508del CFTR mutation participated in this study. Twelve non-CF subjects participated in this study. Five of the ten CF subjects were taking CFTR modulators at the time of phlebotomy, three subjects were taking combination ivacaftor/lumacaftor, one subject was taking combination ivacaftor/tezacaftor, and one subject was taking combination ivacaftor/tezacaftor/elexacaftor. The use of clinical modulators was considered "in vivo" modulator treatment. Because we began our ex vivo studies on CF MDMs at a time when lumacaftor/ivacaftor was the only FDA approved modulator for F508del homozygous patients, this study was limited to the use of ex vivo pretreatment with lumacaftor/ivacaftor for experiments.
CF MDMs have a hyperinflammatory response to LPS. CF macrophages are well documented to express a hyper-inflammatory phenotype in response to bacterial stimuli including LPS and Burkolderia cenocepacia 46,50 . Using LPS to induce NFκB signaling and macrophage inflammatory activation, we measured several key cytokines that play an important role in inflammatory signaling and neutrophil recruitment in the CF airway. Compared to non-CF, we found that CF macrophages secreted significantly more IL-6 (P = 0.0102) and IL-8 (P = 0.0382) in response to 24 h exposure to LPS (Fig. 1a,b), but not IL-1β ( Fig. 1c) or tumor necrosis factor alpha (TNFα) (Fig. 1d). Pretreatment for 48 h with modulators did not alter CF MDM cytokine secretion ( Fig. 1a-d, pairwise comparisons within subject). We did not observe gross phenotypic differences in monocytes and MDMs over the course of differentiation (Fig. 1e, Supplementary Fig. S1), or cytotoxicity in response to treatment conditions as measured by lactate dehydrogenase (LDH) release at 24 h ( Supplementary Fig. S2). Four of the seven CF subjects whose macrophages were used for cytokine studies were taking prescribed CFTR modulator therapy at the time of phlebotomy [lumacaftor/ivacaftor (n = 3) or tezacaftor/ivacaftor (n = 1)]. Within this subset of our cohort, we found no differences in IL-1β, IL-6, or IL-8 secretion between CF subjects who were taking prescribed modulators and CF subjects who were not ( Supplementary Fig. S3a-c). Compared to CF patients not on modulators in vivo, TNFα secretion was either trending towards or significantly (P = 0.0080) reduced regardless of pretreatment with modulators ( Supplementary Fig. S3d). While the presumable in vivo exposure of monocytes to modulators in circulation may have lowered TNFα secretion ( Supplementary Fig. S3d), ex vivo pretreatment with modulators for 48 h prior to LPS stimulation did not significantly alter cytokine secretion between the two CF subpopulations (Fig. 1a-d).

CF MDMs have a different iron-related protein expression profile compared to non-CF
MDMs. Iron metabolism is dysregulated in CFBEs, and macrophages found in CF lung explants contain elevated intracellular iron 13,16,17 . To determine if CF MDMs had differential iron-related protein levels, MDMs were assayed for soluble regulators of iron sequestration 21,38,39,58 : lactoferrin and transferrin, and the ferroportin (Fpn) regulatory hormone hepcidin. We also measured total protein expression of key iron handling proteins: the iron importer transferrin receptor 1 (TfR1), the iron exporter Fpn, the iron-storage protein ferritin (FTH1), and the anti-inflammatory heme catabolism enzyme, heme oxygenase 1 (HO-1). Comparing protein expression under control conditions between non-CF and CF, we found a trend towards elevated TfR1 (P = 0.1014) in CF MDMs (Fig. 2a,c). Compared to non-CF, CF HO-1 expression was lower (Fig. 2a, www.nature.com/scientificreports/ viously published work 46 . No changes were observed in Fpn or FTH1 (Fig. 2a,c). Using pairwise comparisons to examine the impact of modulators on protein expression within subject, we found that modulator treatment reduced TfR1 levels (Fig. 2a,d). When activated with LPS, we found that CF macrophages expressed less Fpn and trended towards less HO-1 compared to LPS-activated non-CF macrophages (Fig. 2b,c). In the presence of LPS, ex vivo modulators induced a trend (P = 0.07) towards increased Fpn compared to LPS-activated non-CF macrophages (Fig. 2b,e). Across treatment groups, soluble regulators of iron sequestration and hepcidin ( Fig. 2f-h) showed no differences between CF and non-CF MDMs. Regardless of ex vivo modulator pretreatment, we found that prescribed modulator use in vivo coincided with a trend in lower hepcidin secretion in LPS-activated CF MDMs compared to LPS-activated CF MDMs from patients who did not use in vivo modulators (Supplementary Fig. S3e). Compared to LPS-activated non-CF MDMs, we found that in vitro modulator-pretreatment in LPS-activated CF MDMs from patients on prescribed modulators had elevated lactoferrin secretion (P = 0.0618, Supplementary Fig. S3f), and significantly lower transferrin (P = 0.0076, Supplementary Fig. S3g). Clinical use transformation of the densitometric values, protein was normalized to β Actin and analyzed. Graphs were generated with R version 3.6.0 (2019-04-26, https ://www.R-proje ct.org/) (c) Baseline differences between CF and non-CF protein expression and differences between LPS-treated CF and non-CF protein expression were analyzed using Wilcoxon rank sum tests ('*' indicates P < 0.05, '**' indicates P < 0.01). Within the CF population, whether a patient was using in vivo ivacaftor/lumacaftor or ivacaftor/tezacaftor was denoted by color (yellow "iva/luma" and black "iva/teza").  Fig. S4a,b). These observations demonstrate that CF may influence expression of certain iron-regulatory proteins, and that both ex vivo modulator pretreatment and in vivo use of prescribed modulators have the potential to differentially influence soluble iron-regulatory protein expression.
Non-CF MDMs deplete more exogenous iron than CF MDMs. Lung damage leads to elevated levels of red blood cells, heme, and iron in the CF airway 16 . Mean concentrations of iron ranging from 4 to 35 µM have been measured in CF airways and sputum 15,16 . The ability of macrophages to sequester iron is a key component in preventing both oxidative tissue damage and bacterial iron metabolism. Having found that CF MDMs expressed altered TfR1 and Fpn protein levels (Fig. 2c), both of which are critical for management of iron in the microenvironment, we sought to focus on CF MDM iron handling by examining total iron content of conditioned media after the addition of clinically relevant concentrations of iron 15,16 . Given that recruited macrophages would be chronically exposed to high levels of iron in the CF airway, we hypothesized that CF MDMs challenged with physiologically relevant concentrations of iron would not sequester and/or retain exogenous iron as effectively as non-CF MDMs. As a surrogate for macrophage iron uptake or sequestration, we measured the depletion of exogenously added iron from the culture medium after 48 h. www.nature.com/scientificreports/ In control groups that did not receive exogenous iron, the concentration of total iron was undetectable by our colorimetric assay (Fig. 3a). We found that 8 µM FeCl 3 was cleared to levels below the assay detection range by both non-CF and CF MDMs (Fig. 3b). At a higher concentration of exogenous iron (30 µM), conditioned media from CF MDMs contained significantly (P = 0.0044) higher levels of iron than non-CF (Fig. 3b). Additionally, 48-h pretreatment with CFTR modulators ameliorated this effect in CF MDMs (Fig. 3b). These findings imply that CF MDMs sequester and/or withhold less iron than non-CF.

CF MDM conditioned media contributes to Pa biofilm formation and viability.
We next hypothesized that elevated iron levels in CF MDM conditioned media may influence the growth of a CF Pa clinical isolate, DH1137. To do this, we collected supernatants from non-CF and CF MDMs that were given exogenous iron, with or without modulator pretreatment (Fig. 3b). We chose to test supernatants from MDMs exposed to 30 µM FeCl 3 because we observed significant and measurable changes in the amount of iron present in CF and modulator-pretreated CF MDM supernatants at this concentration (Fig. 3b). We mixed the MDM supernatants with Pa DH1137 and performed a biofilm formation assay. At the endpoint of the biofilm formation assay, we recovered the mixed supernatants to measure Pa density and viability. We found that the optical density (OD) at λ = 600 nm (OD 600 ) of the supernatant recovered from the biofilm formation assay was not altered (Fig. 4a), indicating no change in the density of bacteria in the assay supernatant. However, the OD 600 reading does not discern the relative density of live, dead, or dying bacteria. To answer this question, we measured viable bacteria recovered from the biofilm formation assay by measuring the growth of colony forming units (CFUs) recovered from the assay supernatant. We found that CF conditioned media increased recovery of viable bacteria (Fig. 4b, mean: non-CF viable CFU × 10 6 = 2.133 ± standard deviation (SD) 2.30, CF viable CFU × 10 6 = 20.60 ± SD 16.9). Additionally, CF conditioned media increased the formation of biofilms, as measured by absorbance at λ = 550 nm (A 550 ) ( Supplementary Fig. S6, Fig. 4c,d, mean: non-CF biofilm formation A 550 = 0.150 ± SD 0.104, CF biofilm formation A 550 = 0.408 ± SD 0.124). Modulator-pretreated CF conditioned media reduced the recovery of viable bacteria (P = 0.0876, Fig. 4b) and significantly reduced biofilm formation (P = 0.0033, Fig. 4c,d).

Discussion
While iron may not be readily available in the non-CF human lung, both CF sputum and airway lavage fluid contain elevated iron 15,16,59 . Iron is an essential nutrient for the antibiotic resistance and biofilm formation of Pa, one of the most common CF pathogens 13,60,61 . As manipulation of iron availability has become a therapeutic option of interest for CF lung infections [60][61][62] , expanding our understanding of iron metabolism in primary CF macrophages could offer insight into airway host-pathogen interactions.
The goal of this study was to test the hypotheses that (1) iron-regulatory proteins are altered in CF MDMs, and (2) ex vivo ivacaftor/lumacaftor corrects expression of iron-regulatory proteins in a manner relevant to host-pathogen interactions. We analyzed samples obtained from a small cohort of CF patients who all had a common and severe CFTR mutation (Class II, F508del/F508del). We exposed primary human MDMs to physiologically relevant concentrations of CFTR modulators 27,53,63 and iron 15,16 , and examined cytokines, iron-carrier proteins, the hormone hepcidin, and cell-expressed proteins. We tested the impact of macrophage conditioned media on biofilm formation and viability of a Pa CF clinical isolate.
We found that CF MDMs produced more IL-6 and IL-8 (Fig. 1a,b) and less of the anti-inflammatory heme catabolism enzyme, HO-1 (Fig. 2c). These findings are in support of extensive works of Bruscia et al. on the www.nature.com/scientificreports/ characterization of CFTR-dependent HO-1 dysregulation in both murine and human macrophages and the contribution of this dysregulation to CF macrophage hyperinflammation 46,47 . In the context of infection and inflammation, the transcription of the Fpn-regulatory hormone hepcidin is induced by IL-6-activated JAK/ STAT signaling 38 . While CF MDMs secreted significantly more IL-6 than non-CF MDMs (Fig. 1a, P = 0.0102), there was no difference in hepcidin secretion (Fig. 2f). Though we observed significantly lower Fpn in CF MDMs with LPS activation (P = 0.0015, Fig. 2c), we did not observe changes in hepcidin secretion under these conditions (Fig. 2f). While modulator treatment ex vivo did not impact hepcidin secretion, we did observe trends in reduced hepcidin secretion in subjects taking modulators in vivo ( Supplementary Fig. S3e). Though in vivo modulator use induced trends in reduced hepcidin secretion ( Supplementary Fig. S3e), we found no increase in Fpn expression with in vivo modulator use ( Supplementary Fig. S4a,b), suggesting the observed reduction in hepcidin secretion was not responsible for appreciable changes in Fpn expression. Although we found that in vivo modulator use coincided with significantly lower transferrin (P = 0.0076) in activated, modulator-pretreated CF MDMs ( Supplementary Fig. S3g), TfR1 protein expression was not altered in association with patient modulator use under any treatment conditions ( Supplementary Fig. S4a,b). These findings demonstrate that clinical use of modulators but not ex vivo modulator treatment can impact inflammatory and iron-related protein secretion (Supplementary Fig. S3d-g). This suggests that acute and chronic modulator exposure differentially impacts monocyte/macrophage iron handling and immunophenotype. This may be addressed in future studies of patients taking the same triple combination therapy.
In addition to lower Fpn in CF MDMs with LPS activation (Fig. 2c), we found that ex vivo modulator pretreatment in LPS-activated MDMs induced a trend towards elevated Fpn without altering IL-6 or hepcidin secretion (Figs. 1a, 2c,f, Supplementary Fig. S3e). While it is known that hepatocytes are the primary producers of hepcidin 64 , these findings pose interesting questions about hepcidin-independent mechanisms of Fpn regulation as well as potential autocrine dysregulation of the hepcidin-ferroportin regulatory axis. These findings also offer insight into iron-withholding in CF MDMs activated through TLR4. CF MDMs have a proven deficit to phagocytose and kill Pa 27 , suggesting that lower Fpn in activated CF MDMs does not impart an advantage to the host during direct host-pathogen interactions. Fpn transports iron (II) across the cell membrane and releases iron (II) to be oxidized and finally bound to transferrin as iron (III) 65 . We observed no change in FTH (Fig. 2c), which plays an important role preventing iron (II)-induced oxidative damage by oxidizing and storing iron (II) as iron (III) 66 . Although we did not measure intracellular iron content in this study, we can hypothesize that low CF MDM HO-1 and Fpn in conjunction with unchanged ferritin levels may contribute to elevated intracellular iron and oxidative stress (Fig. 2c). Elevated intracellular iron has been described in CFBEs and alveolar macrophages from CF lung explants, though alveolar macrophages in CF lung explants did also express elevated ferritin and ferroportin staining 13,16 . These questions may be addressed in future studies by measuring activated CF macrophage ROS production and intracellular iron content in response to exogenous iron.
CF MDMs expressed a trend towards elevated basal TfR1 (Fig. 2c) compared to non-CF MDMs, and basal TfR1 levels were significantly (P = 0.0259) reduced with modulator pretreatment (Fig. 2d). These data are intriguing and seemingly contradictory to our finding of elevated extracellular iron in CF MDM conditioned media (Fig. 3b). There are several avenues to consider when interpreting these data. TfR1 protein accumulation, in the absence of iron uptake, may stem from defects in protein internalization. This was found to be the case in a study of PBMCs and lymphocytes from human patients with a mutation in TfR1 gene, TFRC, in which internalization of TfR1 was disrupted, resulting in protein accumulation on the cell surface 67 . TfR1 mRNA stability is regulated by the zinc-finger protein tristetraprolin (TTP), which is in turn regulated by the mammalian target of rapamycin complex 1 (mTORC1) 68 . mTORC1 expression is elevated in CFBEs 69 and TTP deficiency has been linked to hypersecretion of IL-8 from IB3-1 CF lung epithelial cells 70 , providing an additional potential avenue for TfR1 upregulation in CF MDMs.
No differences were found in IL-1β secretion from non-CF and CF LPS-activated macrophages (Fig. 1c). These data suggest similar NLRP3 inflammasome activation, although further studies would need to be done to measure pro-IL-1β and caspase-1 to further characterize inflammasome activity in LPS-activated CF macrophages. Findings in the literature suggest that differences in CF inflammasome activity may be observed in a context-dependent manner: CF airway epithelial cells and CF monocytes have similar IL-1β and caspase-1 activity in response to LPS 71 , while CF MDMs challenged with Burkholderia cenocepecia secrete significantly more IL-1β than non-CF 50 , and CFTR knockout mice challenged intranasally with Aspergillus fumigatus conidia or Pa express elevated IL-1β in lung homogenates 72 .
Except for TNFα, we found that in vivo ivacaftor/lumacaftor and tezacaftor/ivacaftor use largely did not impact LPS induced cytokine secretion ( Supplementary Fig. S3a-d). Ex vivo pretreatment with ivacaftor/ lumacaftor reduced expression of TfR1 and induced a trend towards elevated Fpn expression (Fig. 2d,e). These findings are in support of a recent publication in which MDMs isolated from CF patients taking clinical ivacaftor had improved B. cenocepecia phagocytosis and CFTR expression compared to MDMs isolated from CF patients who were taking ivacaftor/lumacaftor 50 . In that study, the genotype of donors on ivacaftor was comprised of ivacaftor-responsive class III/IV/V CFTR mutations 50 . Indeed, recent findings suggest that the combination of ivacaftor/lumacaftor induces negative effects on phagocytosis in CF MDMs 27,50 . Although we did not observe a significant detriment to biofilm clearance resulting from ex vivo ivacaftor/lumacaftor treatment, our findings that ex vivo treatment with ivacaftor/lumacaftor did not recover cytokine or soluble protein levels in CF MDMs (Figs. 1a-d, 2f-h) further demonstrate that ivacaftor/lumacaftor may not rescue all aspects of CF macrophage function in patients with more severe mutations. Additionally, the ex vivo tissue culture environment lacks many components such as cytokines, chemokines, elevated iron content, and other cell types that may influence how modulator therapy impacts innate immune cell function in vivo. Our future studies involve investigating macrophage function with an emphasis on iron homeostasis in patients who have consistently taken highly www.nature.com/scientificreports/ effective modulator therapy for over one year. These studies will allow us to assess the impact of highly effective modulator therapy in vivo on macrophage iron homeostasis. In addition to reducing basal TfR1 and elevating Fpn in activated CF MDMs, we observed that ex vivo pretreatment with modulators robustly impacted CF MDM iron sequestration and biofilm formation of DH1137 (Figs. 3b, 4c). Elevated extracellular iron and the associated increase in Pa biofilm formation and viability ( Fig. 4b-d) suggest that dysfunctional iron metabolism in CF MDMs (1) impacts host-pathogen interactions and (2) may be sensitive to CFTR modulator therapy. These findings are particularly relevant in the context of the CF lung, in which iron content is positively correlated with Pa abundance and in which Pa biofilms have been found 10,15 . When considering these data, it is important to understand that these studies do not directly test how iron effects Pa, but rather how iron-rich CF conditioned media effects Pa. To help determine if there is a causative rather than an associative relationship between iron levels in CF media and biofilm formation, direct manipulation of iron levels in conditioned media would be necessary.
These data raise several questions that require future studies. The expression of Fpn and TfR1 over a more comprehensive time course and CFTR modulator dose response regime would allow for greater granularity of expression profiles and potentially answer how modulators correct iron sequestration. Peripheral blood monocytes are recruited to the lung during infection and thus, represent a relevant model in which to study CF mononuclear phagocyte function. However, it is important to note that the lung environment shapes macrophage phenotype 73 . Due to the unique niche they occupy, CF alveolar macrophages may express an iron-regulatory phenotype that is different from that of CF MDMs. The ontogeny and lifecycle of alveolar macrophages are distinct from peripheral blood MDMs in that they are long-lived, self-renewing, tissue-resident cells derived from fetal progenitors 74 . They play a critical role in regulating the immune responses of the lung, inducing tolerance to innocuous inhaled antigens 75 or triggering synchronous inflammatory signaling in response to pathogenic stimuli such as LPS 76 . Future studies on CF and non-CF alveolar macrophages would allow for greater clarity on potential mechanisms of dysregulated iron-handling in the airways themselves. Additionally, several studies have demonstrated the phenotypic impact of iron levels on inflammation, polarization, differentiation, and cellular metabolism 48,[77][78][79] . We must also consider that elevated iron in the CF lung may impact the differentiation and phenotype of recruited monocytes as well as tissue resident macrophages, and thus modulate their ironregulatory protein repertoire. Future studies in which CF macrophages are cultured in high-iron media from the time of monocyte isolation to macrophage differentiation may help answer these questions.
We have shown that CF MDMs are hyperinflammatory in phenotype and have altered expression of several iron related proteins that are altered with modulator treatment. CF MDMs have an inability to reduce levels of exogenous extracellular iron, and their iron-rich conditioned media directly facilitates bacterial biofilm formation and viability. We have shown that CFTR modulators correct aspects of dysfunctional iron metabolism in CF MDMs, although future studies are needed to further delineate the mechanisms of these results.