Detection of CFTR function and modulation in primary human nasal cell spheroids

Abstract

Background

Expansion of CFTR modulators to patients with rare/undescribed mutations will be facilitated by patient-derived models quantifying CFTR function and restoration. We aimed to generate a personalized model system of CFTR function and modulation using non-surgically obtained nasal epithelial cells (NECs).

Methods

NECs obtained by curettage from healthy volunteers and CF patients were expanded and grown in 3-dimensional culture as spheroids, characterized, and stimulated with cAMP-inducing agents to activate CFTR. Spheroid swelling was quantified as a proxy for CFTR function.

Results

NEC spheroids recapitulated characteristics of pseudostratified respiratory epithelia. When stimulated with forskolin/IBMX, spheroids swelled in the presence of functional CFTR, and shrank in its absence. Spheroid swelling quantified mutant CFTR restoration in F508del homozygous cells using clinically available CFTR modulators.

Conclusions

NEC spheroids hold promise for understanding rare CFTR mutations and personalized modulator testing to drive evaluation for CF patients with common, rare or undescribed mutations.

1. Introduction

Cystic Fibrosis (CF) is an autosomal recessive disorder affecting >70,000 people worldwide [3,4]. CF is caused by mutations in the gene encoding the Cystic Fibrosis Transmembrane Conductance Regulator protein (CFTR), a traffic ATPase that functions as a chloride and bicarbonate channel [3,5]. CFTR also regulates several epithelial ion transporters, including other chloride channels and the epithelial sodium channel (ENaC) [6]. Over 2000 variants in the CFTR gene have been identified, with multiple characterized defects [7]. CFTR dysfunction results in multisystem disease, with most morbidity and mortality stemming from pulmonary disease due to thick airway mucus, airway obstruction, chronic infection, and inflammation. This process is caused by dysregulation of airway surface liquid (ASL) homeostasis and defective mucus production governed by CFTR [8].

Novel, small-molecules termed modulators have recently been developed that directly improve CFTR function for limited mutations. Ivacaftor (VX770) is FDA-approved for patients with mutations causing defective CFTR gating or conductance (e.g.: G551D or R117H). Lumacaftor (VX809) is FDA-approved for use in combination with ivacaftor for patients with two copies of the most common CFTR mutation, F508del [9–12]. Additional modulators are in development, potentially expanding both the qualifying patient population and the treatment options for individual patients [13–15].

Current modulators were developed using human lower airway epithelial cells (LAEC) [16]. LAECs from explanted lung tissue are grown at air-liquid interface (ALI) to mature monolayers, allowing examination of ion transport, ASL homeostasis, and mucociliary clearance as proxies of CFTR function [17,18]. This model has driven development of CFTR modulators for common CFTR mutations, but has limitations in patients with rare CFTR mutations. Patient-derived models would allow improved mutation-to-disease correlation and individualized modulator testing [19]. This could be particularly useful for patients with rare or poorly characterized mutations, for whom clinical trials are impractical. Moreover, as novel modulators emerge, such models may help determine the appropriate combination of drugs to optimize clinical outcomes.

Intestinal tissue has been grown as organoids to model CFTR function [20,21]. Acquisition of GI tissue remains invasive, requiring endoscopy or suction biopsy for rectal specimens, with minor safety risks [10,22]. Limited existing work in 3-dimensional culture of nasal tissue is not clearly adaptable to monitoring CFTR function [23–25]. We hypothesized that primary human nasal epithelial cells (NECs) obtained non-surgically could be grown in 3-dimensional culture to yield a patient-derived, swelling-based model of CFTR function. Such a model could provide powerful individualized data from an easily obtained respiratory sample.

2. Methods

This protocol was approved by the Cincinnati Children’s Hospital Medical Center Institutional Review Board. All subjects/families provided written assent/consent. Cell cultures from healthy volunteers and patients with a variety of CFTR mutations were studied.

Additional detail regarding ALI culture, electrophysiology studies, immunofluorescence, and protein isolation/detection is provided in the supplemental materials.

2.1. NEC procurement, processing, and expansion

NECs were procured by curettage of the inferior turbinate using a rhinoprobe (Arlington Scientific, Inc.), pooling cells from both nostrils into a 15 mL conical filled with Media A (Table E1) and storing on ice for <24 h. Curettage was chosen based on site expertise.

Curettes were rinsed with media and the collected cells were centrifuged at 360 ×g for 5 min. Cell pellets were re-suspended in Accutase (Innovative Cell Technologies, Inc.) and centrifuged at 360 ×g for 5 min. The resulting pellet was re-suspended in Media A and placed into VitroCol (Advanced BioMatrix) coated petri dishes pre-seeded with irradiated mouse embryonic fibroblast feeder cells (Globalstem). Cells were maintained in Media A for five days, then Media B until confluent, changing media daily. Once confluent, media was removed and cells were exposed to 0.1% trypsin (Sigma-Aldrich) for 5 min to remove them from the dish. This mixture was centrifuged at 360 ×g for 5 min and the supernatant was removed. Cells were then passaged to a new dish using the protocol above or to spheroid or monolayer culture. All experiments utilized cells of passage 1 or 2.

2.2. Spheroid culture

Growth factor-reduced matrigel (Corning, Inc.) was thawed on ice. Cells were suspended at 500,000 cells/mL in 100% matrigel, vigorously but carefully pipetting to generate a single-cell suspension while avoiding the introduction of air bubbles. This mixture was seeded in 100 µL aliquots into 16 mm four-well plates (NUNC) or 35 mm glass-bottom dishes (MatTek Corporation), creating a spherical “drop” of matrigel using a 200 µL pipette tip with the distal 3–4 mm trimmed at an angle. The plates were incubated at 37 °C and 5% CO₂ for 30 min, until the matrigel set. Media C was then added to the well to cover the matrigel drop. Cells were maintained in Media C until mature (presence of a lumen and a slightly thickened spheroid wall, suggesting a pseudostratified epithelium; typically 7–10 days), changing media daily.

2.3. Spheroid stimulation and measurement

Spheroid plates/dishes were placed in an incubated chamber (37 °C, 5% CO₂) on an Olympus IX51 inverted microscope. Spheroid images at time 0 (n = 10–20 spheroids per condition for all experiments described) were taken at 20× magnification using Slidebook 5.5 (3i, Intelligent Imaging Innovations) software. Spheroids were stimulated by adding forskolin/ IBMX diluted in PBS directly to the media (final concentration of 10 µM/100 µM), and spheroid swelling was monitored with time-lapse imaging for 1 h. At completion, post-stimulation images of all spheroids were captured at time 60 min.

Images were analyzed by manually delineating the luminal area of each spheroid using MetaMorph 7.7 software; examples are provided in Fig. E1. Staff performing the analysis were blinded to mutation, condition, and pre- or post-stimulation timing of each image; total analysis time was approximately 20 min per experimental condition. Spheroid area data was imported into Microsoft Excel 2010 and percent change from time 0 to 60 min was calculated for each individual spheroid.

2.4. Statistical analyses

All measured spheroids were included in the data analysis. Paired t-tests or ANOVA were used to compare continuous data, including change in spheroid area, I_sc, and densitometry using Microsoft Excel 2010 software. An alpha (p) value < 0.05 was used to determine statistical significance. For repeated-measures analysis, variance components were estimated by restricted maximum likelihood using a linear mixed effects model available through the ‘lme’ package in R Studio (R Foundation for Statistical Computing, Vienna, Austria). Mean estimates (±SEM) are presented for comparison of continuous data.

3. Results

3.1. NEC spheroids recapitulate characteristics of respiratory epithelia

We enrolled 25 subjects (age 1–40 years) from the following groups: non-CF controls (n = 6), F508del CFTR homozygotes (n = 9), and CF patients with at least one non-F508del CFTR mutation (n = 10). This included patients with known partial function mutations [F508del/R117H(9T) (2 subjects), R117H(7T)/3849+10kbc>T, 3849+10kbc>T/3849+10kbc>T], known surface localization and variable gating defects (F508del/G178R, S549R T>G/S549R T>G), and poorly described mutations and/or clinical features consistent with partial function (F508del/I618T, F508del/I507del, G542X/Unknown, 3120+1G>A/3120+1G>A). Nasal curettage was well tolerated, with only transient discomfort or epistaxis reported. Clinical characteristics of CF donors are presented in Tables E2 and E3. Non-CF controls did not undergo CFTR genotyping, but reported no family history of CF and had no pulmonary symptoms on clinician interview.

Spheroids formed within 3–5 days of seeding, varying from 50 to 400+ µm (Fig. 1). The success rate of forming NEC spheroids was approximately 80%, with fungal contamination as the primary reason for failure. After 7–10 days, most spheroids demonstrated cilia on the luminal surface (Video E1). The density of ciliated cells per spheroid ranged widely within- and between-donors, but did not clearly predict functional response (i.e., spheroids with no cilia often responded robustly, while spheroids with extensive ciliation could have limited response, and vice-versa). Spheroid density averaged 50–100 spheroids per matrigel aliquot. Spheroids demonstrated variable lumen size at baseline, being objectively larger in wtCFTR compared to mutant CFTR spheroids. Markers of mature respiratory epithelia were detected in NEC spheroids, including e-cadherin, CFTR, luminal F actin, cilia (alpha tubulin), and the mucin MUC5AC (Fig. 1). Wt and F508del CFTR expression was confirmed by Western blot of human NECs grown at ALI (Figs. 2 and 3).

Fig. 1.

Respiratory epithelial markers of NEC spheroids. Spheroids form within 4–7 days of plating, with variability in spheroid morphology and size that is mutation-, time-, and treatment-dependent (Row A; scale = 50 µm). Spheroids demonstrate characteristics of a polarized pseudostratified epithelium including adherens junctions (E-cadherin, Panel B; scale = 100 µm) and cilia, which are localized to the spheroid lumen (Panel C; scale = 30 µm). In addition to ciliated cells, spheroids include mucus secretory cells, noted via staining for Muc5ac (Panel D; scale = 50 µm).

Fig. 2.

Functional characteristics of NEC spheroids derived from healthy volunteers. cAMP-induced swelling/shrinking of wtCFTR spheroids from a single subject in control conditions and pretreated with Inh172 is shown in Panel A. Summary data of spheroid swelling assays from all healthy volunteers (11 experiments from 6 unique subjects) is shown in Panel B; bars represent mean values (+16.0% for cAMP, ±3.1 SEM; −13.4% for cAMP/Inh172, ±3.8 SEM). Short-circuit current for NEC ALI cultures from a single subject is shown in Panel C (n = 3 per condition); CFTR-dependent short-circuit current generation is consistent with wtCFTR. Representative Western blot demonstrating expected C-band wtCFTR in NECs is shown in Panel D; each lane represents a single 6.5 mm transwell insert from the same subject. CFBE41o-controls for Western blot are stably transduced with wtCFTR or F508del CFTR (CMV promoter) as indicated. ****p < 0.0001.

Fig. 3.

Functional characteristics of F508del CFTR homozygous NEC spheroids. cAMP- and VX770-induced swelling/shrinking of F508del CFTR spheroids from a single subject in control conditions, pretreated with VX809, and pre-incubated at 27 °C is seen in Panel A. Summary data of spheroid swelling assays from all CF patients homozygous for F508del CFTR (9 unique subjects) is shown in Panel B; bars represent mean values (−5.8% for control, ±3.9 SEM; +9.8% for VX809, ±3.0 SEM). Short-circuit current for NEC ALI cultures from a single subject is shown in Panel C (black bars: untreated; grey bars: VX809 pretreatment); small CFTR-dependent short-circuit current generation is consistent with F508del CFTR, and statistically significant improvement with VX809 is demonstrated (n = 4 inserts per condition). Panel D shows a representative Western blot demonstrating minimal C-band CFTR in untreated F508del NECs grown at air:liquid interface, with a significant increase in C-band fraction following VX809 treatment; each lane represents a single 6.5 mm transwell insert from the same subject. Optical density is presented in Panels E and C:(C + B) band fraction (relative to HSP-90) is presented in Panel F (black bars: untreated; grey bars: VX809 pretreatment). T84 intestinal cells homozygous for wtCFTR⁺ were used as a C-band loading control; T84 was loaded at 10% of the total protein content as the primary cells. *p < 0.05; **p < 0.01; ****p < 0.0001.

3.2. NEC spheroids reliably demonstrate CFTR function and modulation

Across all donors/conditions, stimulation of cAMP production in the presence of functional CFTR led to luminal ion/fluid secretion causing spheroid swelling, while cAMP stimulation of spheroids with dysfunctional CFTR produced minimal swelling, or shrinking.

NEC spheroids from healthy volunteers (wtCFTR – Fig. 2) did not change in dimension over time without stimulation (60 min), and swelled when stimulated with forskolin/IBMX with an average of +16.0% (± 3.1; range 3.5 to 38.5%) from baseline. Conversely, wtCFTR spheroids pretreated with the CFTR inhibitor Inh172 (10 µM) shrank when stimulated (mean −13.4%, ±3.8; range −32.9 to 0.6%. n = 11 experiments, 6 donors; p < 0.0001 vs control). Normal CFTR function in NECs was confirmed with voltage-clamp testing of parallel ALI monolayers in Ussing chambers, with CFTR-dependent short-circuit current of 25.4 µA/cm² (± 4.3; n = 5 donors).

NEC spheroids from F508del homozygotes (Fig. 3) generally shrank following acute stimulation with forskolin/IBMX and VX770 (1 µM) (mean −7.0%, ±3.9; range −27.6 to 3.5%). Conversely, F508del/F508del spheroids pretreated with VX809 (3 µM, 24 h) swelled following stimulation, with a mean increase of +8.7% from baseline (±3.0; range −5.7 to 24.5%. n = 9 donors; p = 0.005 vs control). These responses are shown in Videos E2 and E3, respectively. Within-subject response to VX809 was not clearly related to the extent of cilia formation within the spheroid. Aggregate data is compared against wtCFTR in Fig. 4. F508del/F508del spheroids also swelled following stimulation when pre-incubated at 27 °C for 24 h, a maneuver known to increase cell surface F508del CFTR [mean + 13.2% (±8.7) versus 37 °C of −10.9% (±5.2); range −7.2 to 28.8%. n = 4 donors; p = 0.03] [26]. When possible, abnormal CFTR function and modulation with VX809 in NECs was confirmed with voltage-clamp testing of parallel ALI NEC cultures in Ussing chambers, with CFTR-dependent short-circuit current of 1.58 µA/cm² (±0.68), increasing to 2.85 µA/cm² after pre-incubation with VX809 (±1.45; n = 2 donors). Increase in mature CFTR (C-band) fraction with VX809 treatment was confirmed by Western blot of NEC lysates (Fig. 3).

Fig. 4.

Summary of spheroid responses across three donor groups. Summary functional data includes NEC spheroid swelling experiments from each unique subject with wtCFTR (n = 6) stimulated with cAMP (white bar) and cAMP + CFTR inh172 (dotted white bar); patients homozygous for F508del CFTR (n = 9) stimulated with cAMP + VX770 (light grey bar), and cAMP + VX809 + VX770 (dotted light grey bar); and patients with other CFTR mutations (n = 10) stimulated with cAMP alone (dark grey bar). INSET: expanded view of “other CFTR” group; numbers for individual data points correlate to clinical data presented in Table E3. **p < 0.01; NS: non-significant.

NEC spheroids from subjects with other CFTR mutations swelled variably between control wtCFTR and F508del homozygous CFTR spheroids when stimulated with forskolin/IBMX (+2.4%; ±1.6; range −4.2 to 10.7%. n = 14 experiments, 10 donors). Response to CFTR modulators was variable and mutation-dependent. When possible, partial CFTR function was confirmed with voltage-clamp testing of parallel ALI NEC cultures in Ussing chambers, with CFTR-dependent short-circuit current of 11.6 µA/cm² (±5.0; n = 4 donors).

Summary data of stimulated spheroid swelling across the three groups is shown in Fig. 4. Comparison of same-subject, same-sample NEC spheroid and ALI I_sc data at baseline and after pre-treatment with VX809 is shown in Fig. 5, showing strong correlation of both metrics across model systems. Only subjects with successful spheroid and ALI cultures are included; characteristics of NEC ALI culture and I_sc testing are shown in Fig. E2.

Fig. 5.

Correlation of NEC spheroid and ALI culture analysis. Each unique subject that underwent both NEC spheroid and short-circuit current analysis is represented by a single point; error bars indicate SEM (absence of an error bar indicates an SEM smaller than the point symbol). Each group is delineated by the point symbol: wtCFTR (square), other CFTR (circle; individual mutations by adjacent number: [1] F508del/I507del; [2] c.2249C>T/c.1408A>G; [3] R117H/3849+10kbC>T; [4] G542X/Unidentified), F508del homozygotes (triangles). Panel A compares stimulation (cAMP for wt and other CFTR mutation groups; cAMP + VX770 for F508del CFTR) under control conditions in both models. Panel B compares the stimulated change induced in each model by pre-incubation with VX809, plotting the change in percent swelling in spheroids (%Swelling VX809 − %Swelling Control) versus the percent change in I_sc (100 * [VX809 − Control] / Control). Both comparisons are statistically correlated by linear regression as indicated in the figure; red line = regression line, dotted line = 95% confidence interval.

Four subjects (three healthy volunteers, one CF subject) underwent serial NEC acquisition and repeated testing. Within each subject, responses to cAMP stimulation were consistent, and each individual reliably demonstrated a specific CFTR-dependent response, shown in Fig. E3. Using a linear mixed effects model, the majority of variance in this subset of samples was due to pharmacologic treatment (Inh172 vs control, nested by subject) with a SD of 17.7. Conversely, intra-subject and inter-subject variability was low (SD 0.9 and 0.006, respectively). Residual variance was SD 17.5.

4. Discussion

We present an ex vivo model system derived from nasal curettage samples that reliably differentiates CFTR function as determined by donor genotype and characteristics. This model demonstrates F508del CFTR modulation in response to ivacaftor and lumacaftor. With an easily obtained source of respiratory epithelial cells and straightforward methodologies, this model may serve to better understand relationships between rare CFTR genotypes and CFTR function, and to detect modulator biologic activity in patient-derived cells that could ultimately inform individual therapies.

NEC spheroids rapidly formed a “lumen in” configuration, with markers of mature respiratory epithelia including cilia, mucus secretion, and separate apical/basolateral membranes supporting luminal fluid secretion. It is unclear whether spheroids arise from mature or progenitor cells, but their rapid development in standard media conditions suggests recapitulation from the native, differentiation epithelium. Though cells were manipulated through conditional reprogramming culture (CRC), it has been previously reported that epithelial cells expanded with CRC techniques are similar to their native cell origins [18]. The majority of spheroids described were of passage 1, with two individuals at passage 2 with no clear difference among groups. Ongoing work is in process to determine the durability of spheroid response over sequential passages.

Clear patterns of NEC spheroid responses emerged across all subjects. For wtCFTR spheroids, these included swelling in response to cAMP, and shrinking with cAMP combined with CFTR inhibition (Fig. 2). For the F508del homozygous group, spheroids from all donors had a positive change in cAMP-induced swelling response following treatment with VX809 or incubation at 27 °C (Fig. 3). The uncorrected F508del homozygous donors’ responses were variable between slight swelling and shrinking (Fig. 3, Panel B). This is not surprising, as detectable F508del CFTR function in LAECs has previously been reported [16]. Future studies will be needed to clarify the relationship between baseline and modulated CFTR function across F508del and other CFTR mutations.

Our third patient group included CF patients with variable predicted CFTR function based on genotype. Under control conditions, spheroids from this group swelled modestly, falling between wt and F508del CFTR groups.

ALI samples with good epithelial electrophysiology characteristics (i.e. ENaC currents, resistance) show tight correlation with same-subject sphere swelling response (Fig. 5). This cross-model correlation ties spheroid swelling to a traditional measure of CFTR function, and serves as evidence of subject specificity. This correlation holds both for baseline CFTR function (Fig. 5A) and the modulator effect in those exposed to VX809 (Fig. 5B), further supporting the use of nasal spheroids as a model of CFTR modulation. NEC ALI culture was unsuccessful in several donors, limiting our ability to validate individual spheroid responses vs I_sc measured under voltage clamp conditions. Reasons for monolayer failure included infection, poor resistance, and apparent squamous transformation, as demonstrated in Fig. E2. This highlights a challenge of studying NECs under ALI, and a potential benefit of NEC spheroid analysis.

We speculate that shrinking in stimulated spheroids with inhibited or dysfunctional CFTR is related to unopposed ENaC activity as a dehydrating force, similar to ASL dehydration in vivo. Such an effect may be absent in intestinal organoids as ENaC is not routinely expressed at a functional level in these tissues. Alternatively, basolateral sodium and/or potassium transport, especially cAMP-dependent transporters, may be involved in this phenomenon. Further study is necessary to answer these questions.

Extended survival of NEC spheroids was limited to several weeks, which is offset by multiple passages under CRC, or by repeated curettage to obtain primary specimens. In addition, the close segregation between full, variable and absent CFTR function suggests a relatively small dynamic range. Importantly, this dynamic range appeared to be centered in the range of detectable correction of F508del CFTR. Work in intestinal organoids has demonstrated a non-linear relationship of forskolin stimulus and swelling response; an understanding of similar characteristics in NEC spheroids may expand the dynamic range of the assay. Through the current studies, we are unable to exclude the impact of alternate chloride channels on spheroid swelling. Nonetheless, this swelling response is CFTR-dependent, as evidenced by with in subject response to CFTR inhibition or correction, and by correlation with short-circuit current measurements.

Additional limitations of NEC spheroids include variability in spheroid responses. In repeatability testing (Fig. E3), there was more variance associated with pharmacologic treatment than serial acquisition, as expected. Low inter-subject variability is perhaps unsurprising, as 3/4 subjects were healthy controls and the remaining subject had high CFTR function. Measurement variability is skewed by inclusion of all measured spheroids in the analysis; this reduces bias, but includes nonresponsive spheroids, increasing variability. To mitigate this, an increased number of spheroids were measured per condition to provide statistical power to detect group differences. Automation of measurement could provide the ability to rapidly measure a large number of spheroids, reducing selection bias and improving statistical power. Such automation is an ongoing goal of this work, including fluorescent labeling of cells to improve contrast and automated area measurements to increase throughput.

Critically, correlations between ex vivo spheroid responses to drug must be studied against in vivo subject biomarkers to determine if this model predicts clinical response. This comparison is complex, as existing CF disease biomarkers may be normal in those with mild disease, limiting the ability to detect benefits to therapy. Moreover, in less robust modulator treatments (e.g.: ivacaftor/lumacaftor in F508del homozygotes) the biomarker (FEV₁) impact is low, but the disease stabilization effects (reduction in pulmonary exacerbations) are high. This complexity of defining patient response makes in- and ex-vivo correlations difficult; the limited number (six) of subjects treated with modulators in this data set is presented in Table E4. Three of six subjects had clinical improvement in both FEV₁ and BMI (≥3% change); all three had a corresponding ≥10% improvement in spheroid swelling. This implies a relationship between spheroid response to drug and in vivo clinical response; larger numbers of subjects will be necessary to understand the predictive value of this novel assay to patient outcomes.

In summary, our data indicates that NEC spheroids join gastrointestinal organoids and bronchial epithelial cells as preclinical model systems to test CFTR modulators in a patient-specific fashion. These models represent exciting advances in CF therapeutic personalization, and each demonstrates unique benefit and potential use. Bronchial epithelial cultures remain the gold standard for ex vivo CFTR study, but require invasive acquisition that limits personalized study. Intestinal organoids allow for such personalization and have high throughput, but acquisition can be uncomfortable, and focuses on a tissue that is not the primary source of mortality. NEC spheroids, conversely, are personalized, easily acquired, and represent the airway tissues, but currently suffer from limited throughput. For all models, correlation of model system predictions to measurable patient benefits will be necessary to guide therapeutic trials and future care decisions. CFTR modulators are likely to be prescribed for life and are expensive, highlighting the need for careful development of preclinical model systems in parallel with meaningful clinical outcome measures. This patient-derived model system holds promise to drive future pharmacotherapies and personalized therapeutic decisions regarding CFTR modulation.

Abbreviations

ALI

air-liquid interface

ASL

airway surface liquid

BMI

body mass index

cAMP

3′, 5′-cyclic adenosine monophosphate

CF

Cystic Fibrosis

CFTR

Cystic Fibrosis Transmembrane Conductance Regulator

CRC

conditional reprogramming culture

ENaC

epithelial sodium channel

IBMX

3-isobutyl-1-methylxanthine

LAEC

lower airway epithelial cell

NEC

nasal epithelial cell

PBS

phosphate-buffered saline

ppFEV₁

percent predicted forced expiratory volume in 1 s