Characterization of TRPV4-mediated signaling pathways in an optimized human choroid plexus epithelial cell line

Abstract

The objectives of these studies were twofold: 1) to characterize the human choroid plexus papilloma (HIBCPP) cell line as a model of the blood-cerebrospinal fluid barrier (BCSFB) via morphology, tightness, and polarization of transporters in choroid plexus epithelia (CPe), and 2) to utilize Ussing-style electrophysiology to elucidate signaling pathways associated with the activation of the transient receptor potential vanilloid 4 (TRPV4) channel involved in cerebrospinal fluid (CSF) secretion. RT-PCR was implemented to determine gene expression of cell fate markers, junctional complex proteins, and transporters of interest. Scanning electron microscopy and confocal three-dimensional renderings of cultures grown on permeable supports were utilized to delineate the morphology of the brush border, junctional complexes, and polarization of key transporters. Electrophysiology was used to understand and explore TRPV4-mediated signaling in the HIBCPP cell line, considering both short-circuit current (I_sc) and conductance responses. HIBCPP cells grown under optimized culture conditions exhibited minimal multilayering, developed an intermediate resistance monolayer, retained differentiation properties, and expressed, and correctly localized, junctional proteins and native transporters. We found that activation of TRPV4 resulted in a robust, multiphasic change in electrogenic ion flux and increase in conductance accompanied by substantial fluid secretion. This response appears to be modulated by a number of different effectors, implicating phospholipase C (PLC), protein kinase C (PKC), and phosphoinositide 3-kinase (PI3K) in TRPV4-mediated ion flux. The HIBCPP cell line is a representative model of the human BCSFB, which can be utilized for studies of transporter function, intracellular signaling, and regulation of CSF production.

INTRODUCTION

The choroid plexus (CP), situated within each brain ventricle, is a specialized, branched structure composed of a single-layered epithelium surrounding a fenestrated capillary network and rich connective tissue. The CP epithelium (CPe) establishes the blood-cerebrospinal fluid barrier (BCSFB) via tight junctions between adjacent epithelial cells. The BCSFB facilitates the delivery of various compounds to the central nervous system (CNS) via regulated carrier transport, while critically controlling the exchange of solutes between the blood and the CSF (1). The CP is the primary producer of cerebrospinal fluid (CSF), in turn facilitating proper neuronal development (2). CSF functions to protect the brain and spinal cord by decreasing the effective weight of the brain (3); enabling the distribution of local mediators and hormones to the brain parenchyma (4); and mediating the clearance of waste products and toxic material from the CNS via the debated “glymphatic” system (5).

Approximately, 500 mL of CSF is secreted per day in an adult human (6). Volume and composition of the CSF varies on a diurnal basis, as well as in response to pathophysiological states such as trauma and infection (7, 8). Understanding the mechanisms that control CSF production is, therefore, essential to developing an understanding of overall brain function in both health and disease. Regulation of CSF production is known to be controlled by the specific movement of electrolytes and water, between the blood and the intraventricular space (9). This process is coordinated by a host of signaling mechanisms integrated between different transporters and channels located on either the apical (CSF-facing) or basolateral (blood-facing) membrane of the CPe. Though many of the transporters and their localization have been identified in the native CPe, how these proteins are specifically implicated in the regulation of CSF production is less clear. Moreover, how these proteins are linked in concerted signaling pathways to maintain this tightly controlled system remains an almost entirely open question and, thus, necessitates further investigation.

The localization of several of these specific transporters, notably the Na-K-ATPase pump, is uniquely polarized to the apical membrane in the CPe compared with other epithelial tissues. Therefore, it is critical to develop an in vitro model that accurately represents the structure of the CPe that can be utilized to investigate the mechanistic interactions between these transporters. Previously, primary cultures of CP epithelial cell lines have exhibited low transepithelial electrical resistance (TEER) that is inconsistent with their barrier function in vivo and renders measuring permeability changes difficult (6, 10, 11). More recently established continuous (immortalized) cell lines have been able to overcome this problem by forming a relatively high resistance monolayer when grown on permeable supports (11–14). The barrier function of the CPe is of particular importance when considering culture models due to the growing body of evidence that infectious diseases, for example COVID-19 infection (15), can impact the barrier function of the CPe and result in CNS morbidity.

The transient receptor vanilloid 4 (TRPV4) is an osmo-, temperature-, and pressure-sensitive, nonselective cation channel that has previously been implicated in the regulation of CSF production by triggering a flux of solutes across the CPe (16). Specifically, activation of TRPV4 with a specific agonist, GSK1016790A, stimulated an increase in both transepithelial ion flux and conductance (a measure of transepithelial permeability), as measured by Ussing-style electrophysiology (16). The link between TRPV4 and CSF production was reinforced through in vivo studies considering a specific pathology, hydrocephalus. Hydrocephalus is a neurological condition characterized by the pathological accumulation of CSF in the brain ventricles. If left untreated, the pressure manifested by the fluid buildup can cause the ventricles to expand, damaging the surrounding neuronal and non-neuronal tissue. Significantly, inhibiting TRPV4 with either of two structurally distinct specific TRPV4 antagonists, RN1734 or HC 067047, alleviated ventriculomegaly in a genetic model of hydrocephalus (17).

Initial in vitro studies investigating the ion flux in the CPe stimulated by TRPV4 activation were performed in a continuous porcine choroid plexus-Riems (PCP-R) line (16). Although this line developed a high resistance monolayer and expressed numerous transporters present in the native tissue, the model was limited by its inability to correctly polarize transporters specifically implicated in CSF production, including the Na-K-ATPase pump and TRPV4 (18). Thus, to further investigate the role of TRPV4 in regulating CSF production, including identifying crucial proteins involved in this process, an alternative continuous culture model that accurately represents the polarization of native CPe tissue is required. The human choroid plexus papilloma (HIBCPP) cell line is a continuous model originally subcultivated by Ishiwata et al. (14) and then developed and characterized as intermediate resistance by Schwerk et al. (12) that has the potential to fill this void (12, 14).

The current studies characterize the HIBCPP cell line to determine its utility as an in vitro continuous culture model of the CPe for future investigations into this epithelium, the BCSFB, and CSF production in health and disease. We optimized a culture protocol that enables the HIBCPP cells to grow with minimal multilayering and develop into an intermediate resistance (300–900 Ω·cm²) epithelial barrier, facilitating permeability studies. The HIBCPP cell line establishes gene expression of several CP markers, confirming correct CP lineage. These cells also demonstrate gene expression and correct localization of many junctional complex proteins, as well as a brush border with the potential of cilia formation through heterogeneous expression of acetylated tubulin and ARl13B. In addition, at the mRNA level, the HIBCPP model expresses numerous transporters present in the native CPe. More detailed analysis shows correct polarization of key transport proteins to either the apical (Na-K-ATPase subunit α1/β2 and TRPV4) or basolateral [anion exchange protein 2 (AE2)] plasma membrane. Like in the PCP-R cell line, activation of TRPV4 by GSK1016790A, a TRPV4 agonist, stimulates a multicomponent change in net transepithelial ion flux and an increase in conductance resulting in fluid production across the monolayer. This response can be inhibited by pre- or postincubation with a TRPV4-specific antagonist, RN 1734, confirming this response is specific to TRPV4. Despite a substantial, sustained, and reversible increase in conductance, treatment with the TRPV4 agonist, GSK1016790A, does not result in junctional complex disassembly in the HIBCPP cell line. Utilizing electrophysiology, several proteins were identified as modulators of TRPV4-mediated signaling in the HIBCPP cell line: specifically, phospholipase C (PLC), protein kinase C (PKC), and phosphoinositide 3-kinase (PI3K). These proteins all influence TRPV4 activity and thus, may be involved in TRPV4-mediated regulation of CSF production.

METHODS

Cell Culture

During culturing in plastic flasks, HIBCPP cultures were grown in DMEM media, high glucose with bicarbonate and phenol red (Gibco No. 11965126), supplemented with 10% fetal bovine serum (Atlanta Biologicals), 1% antibiotic (Penicillin/Streptomycin 100×, Sigma), and 5 ng/L insulin (Gibco, human recombinant), hereafter referred to as “DMEM full media.” Cultures were maintained in 25-cm² flasks, full media replaced every second day and cultures were passaged at 90% confluency. Cultures were dissociated for passaging utilizing 0.5% Trypsin-EDTA in Hank’s balanced salt solution (HBSS) and reseeded at a density of 25% in 25-cm² flasks. Cultures were not transferred into a flask with a larger surface area as this was found to hinder cell growth with the number of cells typically available in a single-frozen vial. If one wishes to grow them in 75-cm² flasks, we recommend a seeding density of 33% confluency and to increase the media volume from the standard 10 mL to 20 mL. To maintain the line, cells from the first passage after each thawing were refrozen in liquid nitrogen at a 1:1 seeding density (confluent 25-cm² flask per vial, ∼3 × 10⁶ cells). Once thawed, cells were limited to seven passages for experimentation. For all experiments in this manuscript, cells were removed from liquid nitrogen at passages 26–28 (P20 original vial from Schroten/Schwerk laboratory), and passages 29–35 were used for the studies. Cells intended for electrophysiology and immunostaining investigations were dissociated with 0.5% Trypsin-EDTA in HBSS and seeded onto polycarbonate permeable supports (EMD Millipore PIHP03050: 0.4 µm pore size) at 100% confluent density and cultured for 16 days. DMEM full media was replaced in the plates every second day for the first 10 days. From day 10 onward, the basal compartment of the permeable supports were maintained with DMEM full media, whereas the apical compartment was replaced with DMEM serum-free media (DMEM + 1% antibiotic) and 5 ng/L (insulin) to mimic more physiological conditions. These media were then replaced daily for the final 6 days to avoid significant changes in pH.

Fluid Production

Cells were seeded onto permeable supports and grown in optimized conditions provided earlier. Media were replaced in the top and bottom chambers by use of an accurate pipetman. Compounds were added to the apical compartment (<10 μL for each) and cultures were placed back in the incubator for 10 min. During this time, collection tubes were weighed for tare values using a Sartorius analytical balance with a sensitivity of 0.1 mg. Following the incubation, the apical media were collected in the tared tubes. The tubes were then weighed, and the fluid produced was calculated using the density of the media (0.9874 g/mL).

RT-PCR

RNA was extracted from cells grown as a confluent monolayer in 25-cm² flasks (passage 32; P27 + 5). Total RNA was isolated using Monarch Total RNA Miniprep Kit (T2010S). RNA was transcribed to cDNA for PCR reactions using the NEB LunaScript RT SuperMix Kit (E3010L). Some primers were sourced from the literature; in these cases, primers were considered already validated and the PCR cycle was replicated from the original manuscript. New primers were designed using Primer3Plus according to exon sequences obtained from Ensembl, ensuring the primer pairs crossed an exon-exon boundary. Forward and reverse primers were ordered from Integrated DNA Technologies. Gradient PCR was performed to determine the optimal annealing temperatures for each primer pair. Primer pairs were considered validated after their sequenced product (Eton Biosciences) generated >90% sequence identity via NCBI BLAST to the target gene with no alternative hits. A final positive and negative RT-PCR reaction was performed utilizing all validated primer pairs at their optimal annealing temperatures and selected cycle. Primer pairs and their corresponding PCR cycle are detailed in Table 1. Electrophoresis was performed with a 2% agarose gel stained with SYBRSafe Gel Stain (Invitrogen, S33102), flanked by 100 bp ladders (NEB), and visualized under UV light with a ChemiDoc XRS imager (Bio-Rad, Hercules, CA).

Table 1.

Primer table for RT-PCR

Homo sapiens Gene	Protein	Forward Primer 5′–3′	Reverse Primer 5′–3′	Product Size, bp	Optimal Annealing Temperature, °C
GADPH	GAPDH	ACAGCCTCAAGATCATCAGCAA	CCCTGTTGCTGTAGCCAAATTC	546	60
E2f-5	E2f-5	CTGGAGGTACCCATTCCAGA	TGTTGCTCAGGCAGATTTTG	236	64.5
TTR	Transthyretin	ATGGCTTCTCATCGTCTGCTC	TCATTCCTTGGGATTGGTGAC	444	65
FOXJ1	Forkhead Box J1	CCTCCCTACTCGTATGCCAC	CGAGGCACTTTGATGAAGCAC	188	60
OTX2	Orthodenticle Homeobox 2	CATGCAGAGGTCCTATCCCAT	AAGCTGGGGACTGATTGAGAT	200	60
TJP1	Zonula occludens	GAGGACCAGCTGAAGGACAG	GGCCACTTCTTGGATCATGT	243	65
OCLN	Occludin	AGGAACACATTTATGATGAGCAG	GAAGTCATCCACAGGCGAA	242	57
CLDN1	Claudin 1	GAAGATGAGGATGGCTGTCA	AAATTCGTACCTGGCATTGA	142	57
CLDN2	Claudin 2	TCTCTTGGCCTCCAACTTGT	GCATCACCCAGTGTGACATC	192	65
TRPV4	TRPV4	AAGTTCAAGGACTGGGCCTATG	ACGATCACCAGGACAGAGTAGA	473	60
ATP1A1	Na-K ATPase α1	CAGCCCAGAAATCCCAAAACAG	AGCTGTCTTCCACATCGTTGAT	202	60
ATP1B2	Na-K ATPase β2	CCAGCATGTTCAGAAGCTCAAC	AGTGCTGTAACCATAGTGGGTG	208	60
SLC12A2	NKCC1	TGGTATTCTGGCTGGAGCAAAT	ATTTTGGGAGCACTCACTAGGG	378	60
SLC4A10	NBCe2	GGATTTGGACAGTGGCTCCTTA	CTCCCAGAAGGTCCCAGTAGTA	531	65
LRRC8B	VRAC	AAAAGTCATCGACCGTCAGTGA	ATACATGCAGATGAGGCCGTAG	428	65
SLC4A4	NCBE	CCACAGATGCTAGTTCCCTTGT	ATAGGTCCAGATGGTACCCACA	741	65
SLC4A2	AE2	AAGCTGGTGAAGATCTTCCAGG	TTGAATTTGCGCAGGAAGAAGG	242	60
AQP1	AQP1	AGATCAGCATCTTCCGTG	AGTTGTGTGTGATCACCG	354	55
PLCB1	PLCβ1	CAACTCACCAAGTCTCCAGTGG	AGTGCCACCATCTGACAACCTG	140	65
PLCB2	PLCβ2	AAGGTGAAGGCCTATCTGAGCCAA	CTTGGCAAACTTCCCAAAGCGAGT	189	65
PLCB3	PLCβ3	TCCAGAACAGACAGGTGCAG	ATGCTTGTCCCTCATCTTGG	245	65
PLCB4	PLCβ4	GCACAGCACACAAAGGAATGGTCA	CGCATTTCCTTGCTTTCCCTGTCA	179	65
PLCG1	PLCγ1	GACATCACCTACGGGCAGTT	GGAGGAAGCTGAGCATGAAC	226	65
PLCG2	PLCγ2	AGTACATGCAGATGAATCACGC	ACCTGAATCCTGATTTGACTGC	435	65
PLCD1	PLCδ1	TATCCCTGGAGAACCACTGC	CTCGTCTTCGTCTGACACCA	233	65
PLCD3	PLCδ3	CCAGAACCACTCTCAGCATCCA	GCCA TTGTTGAGCACGTAGTCAG	161	65
PLCD4	PLCδ4	AGACACGTCCCAGTCTGGAACC	CTGCTTCCTCTTCCTCATATTC	806	65
PLCE	PLCε	GGGGCCACGGTCATCCAC	GGGCCTTCATACCGTCCATCCTC	530	65
PLCH1	PLCη1	CAAAATGGGCTTCCAAGAAA	CAGATACAGGCAGCGACAAA	164	57
PLCH2	PLCη2	GCTGAAGAGGACGTGGAGTC	GCCACGGACTTGGTGTACTT	245	65
PRKCA	PKCα	CCAGGAGCAAGCACAAGTT	ATGACGCATTGCTTGTGAAC	142	65
PRKCB	PKCβ	GTTGTGATCCAAGATGATGACG	CTGTAAGAAGAACAGACCGATGG	246	65
PRKCG	PKCγ	CACGAAGTCAAGAGCCACA	TAGCTATGCAGGCGGAACTT	233	65
PRKCD	PKCδ	TTTATCGCCACCTTCTTTGG	AGGTGGGGCTCATGTAGTTG	247	65
PRKCE	PKCε	CAGGATGATGACGTGGACTG	CTGCAGCATAGAACCGTGAA	208	65
PRKCH	PKCη	GAACAGAGGTTCGGGATCAA	ATATTTCCGGGTTGGAGACC	239	65
PRKCQ	PKCθ	CTCTGTGGACTGGTGGTCCT	TTGATCTCCCGAAACAAAGG	240	65
PI3KCA	PI3KCα	GGACCCGATGCGGTTAGAG	GCATGCCGATAGCAAAACCAAT	491	65
PI3KCB	PI3KCβ	GACCATTCTTAGGGCCCCGT	GCTACGATGAGCTTTCCCCC	627	65
PI3KCG	PI3KCγ	CAACACTTCCTCTGCATCTGG	TGCCCTATGCGACCTGATAA	157	60
PI3KCD	PI3KCδ	TGCCAAACCACCTCCCATTCCT	CATCTCGTTGCCGTGGAAAAGC	160	60
PI3KCR1	PI3KR1	GGTGAAGCTCGTGTGTGGA	CGGGAGAGTCAAAGCTTGTTG	735	65
PI3KCR2	PI3KR2	CTGACACCACACCACCAACT	CTCAGGGACCAGTCTGTACG	734	65
PI3KCR3	PI3KR3	GTTCCGAGACCGACCGGAG	TGTCATTTACCTCCTCCCTTGAA	785	57

The PCR cycle ran as 95°C for 2 min, 95°C for 15 s (to denature), temperature specified in table for 45 s (to anneal), 72°C for 30 s (to elongate) for all primer pairs for 40 cycles. GAPDH was used as a positive control in all RT-PCR experiments. Primer pairs for Forkhead Box J1 (FOXJ1), occludin, claudin-1 and claudin-2 were taken from Ref. 19. The primer pair for aquaporin 1 (AQP1) was taken from Ref. 20. The primer pair for transthyretin (TTR) was taken from Ref. 21. PI3K, phosphoinositide 3-kinase; PKC, protein kinase C; PLC, phospholipase C.

Electrophysiology

Ussing-style electrophysiological techniques were utilized to monitor transepithelial electrical resistance (TEER; inverse of conductance) and concurrent changes in short-circuit current (I_sc), a measure of net electrogenic transepithelial ion flux. As aforementioned, cells were cultured on a permeable support for 16 days before electrophysiology experimentation. On day 16, the permeable supports were excised and mounted in Ussing Chambers (World Precision Instruments, “WPI”), and connected to a DVC-1000 Voltage/Current Clamp (WPI) as previously described (15). Briefly, each fluid chamber was water jacketed to maintain temperature at 37°C, and the inner chambers continuously bathed the cells in DMEM base media without supplementation. Media were oxygenated by a 5% CO₂/O₂ gas lift. The spontaneous transepithelial potential difference is continuously measured and clamped to zero, resulting in a I_sc output as a measurement of net electrogenic ion flux. Transepithelial electrical resistance (TEER) is recorded continuously through the experiment at 200-s intervals via 2 mV pulses. Cultures that demonstrated basal TEERs of <400 Ω·cm² were discarded. The control and experimental cultures for each experiment were grown and analyzed in parallel as technical replicates.

Immunofluorescence

Cells were grown on permeable supports for 16 days after seeding. Cultures were then incubated with drug, vehicle, or left untreated for 10 min. Cells were washed twice with room temperature 1× phosphate-buffered saline (PBS) and fixed in ice-cold 4% paraformaldehyde in 1× PBS for 10 min. After fixation, cultures were washed thrice in 1× PBS (5 min each), then filters were cut out, divided into six sections, and moved to 12-well plates for staining. Membranes were incubated in permeabilization buffer (0.2% Triton X-100, 3% goat serum, 1% BSA in 1× PBS) for 10 min and then blocked in blocking buffer (3% goat serum, 1% BSA in 1× PBS) for 30 min. For antibodies directed against extracellular epitopes, the 10-min permeabilization step was omitted. Membranes were incubated in primary antibody diluted in blocking buffer overnight at 4°C with gentle agitation. The following day, membranes were washed thrice with 1× PBS (5 min each), and incubated with secondary antibody diluted in blocking buffer for 1 h at room temperature protected from light with gentle agitation. Membranes were washed twice with 1× PBS (5 min each), followed by incubation with Hoechst nuclear stain diluted to 0.5 µg/mL in 1× PBS for 5 min at room temperature protected from light. Membranes were washed for a final time with 1× PBS (for 5 min) and mounted on slides, covered with Aqua Poly-Mount, and cover-slipped. Slides were imaged on a Nikon Confocal C2+ Inverted Microscope. Negative controls were IgG controls for the detection of nonspecific secondary antibody binding (available upon request). Pinhole aperture = 100 μm, scanning speed 2 μs/px. Images were acquired using a ×40 water immersion objective (Nikon) and resolution parameters were 512 × 512 px. Excitation parameters via Nikon Elements software were: 405 nm blue pseudocolor: excitation = 405 nm, emission = 480 nm, laser power 1.0, gain 5.0; 488 nm green pseudocolor: excitation = 480 nm, emission = 520 nm, laser power 2.0, gain 15.0; 568 nm red pseudocolor: excitation = 564 nm, emission = 620 nm, laser 2.0, gain 15.0. Z-stacks were acquired using 0.5 micron step sizes. Z-stacks were either presented as single plane −xy images accompanied by −xz and −yz sidebars or were synthesized as three-dimensional (3-D) renderings utilizing ImageJ or Nikon Elements software for figures in the manuscript. Scale bars are 50 μm for all images unless otherwise noted. Two to three biological replicates were performed for all conditions, and a representative image was selected for presentation. Antibody validation data are found in Table 2.

Table 2.

Antibody information for immunocytochemistry

Target	Catalog	Host Species	Dilution	Validation Method
ZO-1, Synthetic peptide	abcam; ab216880	rabbit	1:50	KO tissue; KO lysate
Occludin, Human	Invitrogen; 33-1500	mouse	1:50	IL-1B cell treatment
Claudin-1, Synthetic peptide	abcam; ab211737	rabbit	1:100	KO lysate
Claudin-2, Human	Invitrogen; 710221	rabbit	1:50	Western blot and IF in positive control cell lines
Arl13B, mouse	Proteintech, 17711-1-AP	rabbit	1:200	KO lysate
AAT, human	Sigma, MABT868	mouse	1:1,000	Positive control tissue (in house)
NKA B2, rat	Alomone, ANP-012	rabbit	1:100	Control Antigen
NKA a1, rat	Alomone, ANP-001	rabbit	1:100	Control Antigen
AE2, human	Santa Cruz, sc-376632	mouse	1:100	positive control tissue human/mouse
TRPV4, rat	Alomone, ACC-124	rabbit	1:250	Control Antigen, KO mouse tissue (in house)

a1, alpha 1; AAT, acetylated α tubulin; AE2, anion exchanger 2; B2, beta 2; KO, knock out; NKA, N-K-ATPase; TRPV4, transient receptor potential vanilloid; ZO1, zona occludins 1.

Scanning Electron Microscopy

Cells were grown on permeable supports for 16 days, with the last 6 days in serum-free media in the apical compartment. Cells were fixed in 2.5% glutaraldehyde, 0.1 M sodium cacodylate, 2 mM CaCl₂ (Electron Microscopy Sciences, No. 16537-15) for 2 h at room temperature. Samples were washed with room temperature 1× PBS and then incubated in 1% OsO₄ in water for 1 h at room temperature, followed by extensive washing with water. The samples were transitioned to 100% ethanol, cut out from the permeable support, and critical point dried from CO₂. Samples were then mounted on aluminum stubs with adhesive and sputter coated with gold particles. Samples were imaged using a JEOL JSM-7800F field emission scanning electron microscope. No postacquisition editing of the images took place. The parameters for emission as well as a scale bar are embedded in the image.

Statistics

All results are displayed as means ± SE for the number of experiments indicated on the graphs. For Ussing-style electrophysiology, a multiple t test grouped analysis was performed on all data sets in GraphPad Prism to compare the significance between each condition group and the matched agonist (GSK1016790A) control group; results distinguished by color. Data presented as bar charts were analyzed using Student’s t tests in GraphPad Prism, with optional nonparametric tests. The threshold for statistical significance was set to P < 0.05 for all results.

RESULTS

Characterization of the HIBCPP Cell Line

Presence of choroid plexus markers in the HIBCPP cell line.

To confirm the correct lineage of the HIBCPP cell line in our optimized culture conditions, RT-PCR was utilized to determine gene expression of several CPe markers, including E2f-5, transthyretin (TTR), Forkhead Box J1 (FOXJ1); Orthodenticle Homeobox 2 (OTX2). CPe markers can be used to distinguish between mature CPe cells and other cell types. Specifically, E2f-5 is a transcription factor that regulates the development of the neuroepithelium into the CPe (22). Transthyretin is a thyroid hormone carrier protein that has been reported as a unique marker of the CPe (23–26). FOXJ1 is a transcription factor that is expressed in the CP, recognized as a master regulator of multiciliogenesis (27). Finally, OTX2 is a transcription factor that plays a critical role in the development and maintenance of CPe cells (28). mRNA expression of all aforementioned genes: E2f-5, Ttr, Foxj1, and Otx2, was established in the HIBCPP cell line (Fig. 1A), confirming this model retains CPe fate in our current culture conditions.

Figure 1.

Characterization of a choroid plexus epithelia (CPe) barrier epithelial phenotype of the human choroid plexus papilloma (HIBCCP) cell line. A: gel illustrates RT-PCR products confirming gene expression of several choroid plexus (CP) lineage markers. B: gel shows RT-PCR products confirming gene expression of several tight junction proteins. Gels stained with SYBRSafe (Invitrogen) and flanked with a 100 bp ladder (New England Biolabs). GAPDH utilized as a positive control. C: immunolocalization demonstrates HIBCPP cells grown on permeable supports express Zona Occludins-1 (ZO-1), Occludin, and Claudin-1 in the junctions. C′ shows magnifications of C as indicated by the white boxes. Images shown in C–E are maximum intensity projections of the z-stack. Scale bars = 50 μm. D and E: comparison of serum feeding conditions on tight junction formation evaluated by immunocytochemistry and transepithelial electrical resistance (TEER) measurements. F: schematic created in BioRender detailing the optimized workflow for the culturing of the HIBCPP cell line. The cells utilized in these experiments have been grown under the optimized culture conditions provided in the body of the manuscript. bp, base pairs; D, day; FOXJ1, Forkhead Box J1; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; OTX2, Orthodenticle Homeobox 2; TTR, transthyretin.

Presence of appropriate junctional complex proteins in the HIBCPP cell line.

Junctional complexes, including tight junctions, are critical to the development of the CPe as a tight, barrier epithelium that controls exchange of solutes between the blood and the CSF (29). RT-PCR was performed to determine gene expression of several junctional complex proteins in the HIBCPP model known to be present in the native CPe tissue. Zona Occludins-1 (ZO-1), Claudin 1, Claudin 2, and Occludin, were all found to be expressed in the cell line (Fig. 1B), confirming previous work (12). Immunofluorescence was utilized to investigate the expression of specific junctional complex proteins at the protein level. ZO-1 and Occludin localized almost exclusively within the cell junctions of the HICBPP cell line, as would be found in the native epithelium (Fig. 1C). Claudin-1 was found to be mostly expressed in the cell junctions (Fig. 1C), similar to the localization of ZO-1 and occludin, indicative of its role as a barrier-forming Claudin (29). In contrast, Claudin-2 demonstrated mostly cytoplasmic localization (Fig. 1C), suggesting Claudin-2 does not play a defining role in junctional permeability in the HIBCPP cell line (30–32). This is consistent with the literature to suggest that Claudin-2 localization varies across development and contributes to a decrease in barrier permeability when expressed in the junctional complexes (31, 32). Together, these results demonstrate the expression and localization of junctional complexes in the HIBCPP cells.

HIBCPP cultures develop an intermediate resistance monolayer under optimized culture conditions.

Several experimental conditions were applied to optimize growth, as well as the barrier phenotype of the HIBCPP cell line. Freezing the cells in liquid nitrogen at a 1:1 seeding density (3 × 10⁶ cells/vial) allowed for successful proliferation when thawed and re-cultured. Moreover, a single-frozen vial of the HIBCPP cells grew best in no greater than 25-cm² flasks as they require close contact for effective growth after seeding. We found the optimal duration to culture the cells on 30 mm permeable supports to be 12–18 days. During this time, the HIBCPP model established an intermediate resistance monolayer, demonstrating TEERs of over 400 Ω·cm² (Fig. 1, D and E). To encourage the formation of junctional proteins, tight barrier function, and ciliation, we evaluated the efficacy of a 6-day serum-starvation on the apical side of the cultures to mimic more physiological conditions. This was found to increase the TEER values of the cells and, most markedly, aided in the process of junction formation (Fig. 1, D and E). As a result of these investigations, an optimized culture protocol was delineated, visualized in Fig. 1F, and adopted for the subsequent experiments. These results support the HIBCPP cell line as an appropriate model of the BCSFB that maintains its intermediate resistance phenotype during culture.

Formation of an apical brush border in the HIBCPP cell line.

The apical borders of native CPe cells are enriched with microvilli and demonstrate bundles of primary cilia (33). Scanning electron microscopy was utilized to evaluate the morphology of the brush border in the HIBCPP cell line. These cultures were found to develop a mostly regular, cobble-stone pattern, with junctions present between cells; significantly, the cells are covered with a lawn of microvilli on their apical border when grown under our culture conditions (Fig. 2A). Although several longer villus structures were observed along the apical surface, no overt cilia bundles were noted. To further explore the cells’ capacity for ciliation, cilia markers Arl13B and acetylated α tubulin (AAT) were utilized on the HIBCPP cultures and evaluated via confocal microscopy. Limited Arl13B⁺ protrusions were observed (Fig. 2B), suggesting that the cells maintain their ability to develop cilia. Of note, some of them were also AAT⁺. It is likely that these cultures are only forming immature cilla-like protrusions and thus, additional growth conditions may be required to fully realize a CPe-like cilia phenotype. Overall, these results suggest that the HIBCPP cell line expresses many components of the cilia complex of proteins. Further development of the culture conditions necessary to enhance ciliation is beyond the scope of the current studies.

Figure 2.

Morphology of the human choroid plexus papilloma (HIBCPP) cells. A: schematic created in BioRender representing the morphology of the apical border of choroid plexus epithelium (CPe) cells according to the literature (33); CPe cells form a multiciliated brush border on the apical membrane. Scanning electron micrograph of the apical surface of HIBCPP cells. Scale bar = 1 μm. B: confocal three-dimensional (3-D) rendering for immunostaining of HIBCPP cells with cilia markers ADP-ribosylation factor-like protein 13B (Arl13B) and acetylated α tubulin (AAT). Scale bar = 50 μm. A, apical; BL, basolateral; CSF, cerebrospinal fluid.

Presence and polarization of known native CPe transporters in the HIBCPP cell line.

Multiple polarized transporters have been identified in the native CPe and implicated in the regulation of CSF production (6). RT-PCR was performed to determine gene expression of specific transporters in the HIBCPP cell line. Specifically, these cells express TRPV4, the Na-K-ATPase α1 and β2 subunit, the Na-K-Cl cotransporter 1 (NKCC1), the sodium bicarbonate cotransporter (NBCe2), and the voltage-gated anion channel (VRAC) (Fig. 3A), all apically localized transporters in the native tissue (9, 34). In addition, the cultures express the Na/HCO₃ cotransporter (NBCE) and AE2 (Fig. 3B), both transporters that localize to the basolateral membrane in the CPe (9). The HIBCPP cells also express aquaporin 1 (AQP1), (Fig. 3C), a bidirectional water channel, predominantly expressed on the apical membrane, though also present in limited numbers on the basolateral membrane in the CPe (9). These channels and their localization in vivo are illustrated in Fig. 2G. To determine the polarization of some of the key transporters in the HIBCPP line, the Na-K-ATPase pump (α1 active subunit) and AE2 were selected for evaluation with confocal microscopy. The Na-K-ATPase ɑ1 subunit was found to be correctly localized to the apical membrane in the HIBCPP cells, as illustrated by the maximal projection and associated 3-D reconstructed image presented in Fig. 3D. The Na-K-ATPase β2 subunit was similarly found to polarize to the apical membrane in this cell line via immunofluorescence (data not shown). In addition, the HIBCPP cells correctly localize the AE2 transporter to the basolateral membrane (Fig. 3E) as shown by the maximal projection. Figure 3F shows these two channels stained on the same permeable support as a 3-D rendering, demonstrating the lack of colocalization, and the clear apical label of the Na-K-ATPase α1 and the lateral/basolateral localization of AE2. These results suggest that the HIBCPP model does correctly polarize transporters to the appropriate plasma membrane as they are expressed in the native CPe tissue.

Figure 3.

Human choroid plexus papilloma (HIBCPP) cell line expresses and localizes native transporters correctly. A–C: gels displaying RT-PCR products confirming gene expression of native choroid plexus epithelium (CPe) transporters in the HIBCPP cell line. Gels stained with SYBRSafe (Invitrogen) and flanked with a 100 bp ladder (New England Biolabs). GAPDH utilized as a positive control. A: gel illustrates the expression of transporters typically localized to the apical membrane in the native tissue: transient receptor potential vanilloid 4 (TRPV4), Na-K-ATPase α1 subunit, Na-K-ATPase β2 subunit, Na⁺-K⁺-Cl⁻ cotransporter (NKCC1), Na -dependent Cl⁻/HCO³⁻ exchanger (NCBe)2, and volume-regulated anion channel (VRAC). B: gel illustrates the expression of transporters typically localized to the basolateral membrane in the native tissue: Na⁺HCO³⁻ cotransporter (NBCE) and anion exchange protein 2 (AE2). C: gel illustrates the expression of aquaporin 1 (AQP1), a water channel in the CPe. D: immunolocalization of the α1 (active) subunit of the Na-K-ATPase pump to the apical membrane of the HIBCPP cells. Three-dimensional (3-D) reconstruction below. E: immunolocalization of the AE2 transporter to the basolateral membrane of the HIBCPP cells. D and E: maximum intensity projections from the z-stack. F: combined 3-D view stained for N-K-ATPase α1 and AE2. G: schematic created in BioRender of native transporters expressed in the choroid plexus (CP) and their localization to either the apical or basolateral membrane. Scale bars = 50 μm. A, apical side; BL, basolateral side; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; NKɑ1; β2, Na-K-ATPase ɑ1; β2; NKCC1, Na⁺-K⁺-Cl⁻ cotransporter.

Activation of TRPV4 in HIBCPP Cell Line

TRPV4 activation stimulates a change in transepithelial ion flux and conductance in the HIBCPP cell line.

TRPV4 has been implicated in the regulation of CSF production through triggering ion flux and thus obligatory water movement (16, 17). Ussing-style electrophysiology was performed in the HIBCPP cell line during stimulation with a TRPV4-specific agonist, GSK1016790A. This technique monitors concurrent changes in I_sc, a measure of net electrogenic transepithelial ion flux, as well as conductance (the inverse of the TEER), as a measure of barrier permeability. Activation of TRPV4 in the HIBCPP cell line by GSK1016790A stimulates a change in both transepithelial ion flux and an increase in conductance (Fig. 4). Specifically, the TRPV4 response exemplifies a positive deflection, followed by an immediate negative deflection in I_sc, before returning to baseline (Fig. 4, A–C). By convention this represents cation absorption and/or anion secretion, followed by cation secretion and/or anion absorption. To determine the optimal concentration of the TRPV4 agonist to administer during subsequent investigations, several concentrations of GSK1016790A, ranging from 5 nM to 20 nM, were added to the HIBCPP cell line (Fig. 4A). Interestingly, the magnitude of the second phase of the response was remarkably dose dependent both in magnitude and in rapidity of response. These data suggest a multicomponent ion flux response that may be differentially regulated. 15 nM was selected as the optimal dose, stimulating a mid-range change in transepithelial ion flux and a large increase in conductance, without being toxic to the cells (i.e., without generating an irreversible increase in conductance).

Figure 4.

Transient receptor potential vanilloid 4 (TRPV4)-mediated ion flux, conductance changes, and fluid secretion in human choroid plexus papilloma (HIBCPP) cells. A–C: electrophysiology graphs are displayed as short-circuit current (I_sc) (left, a measure of net electrogenic ion flux) and conductance (right, a measure of barrier permeability). A: electrophysiological responses to varying doses of GSK1016790A, a TRPV4 agonist, added apically to all cultures at time t = 0 min. Optimal dose determined as 15 nM. B: GSK1016790A added apically, basally, or bilaterally to each set of cultures at time t = 0 min to determine sidedness of transporter. GSK1016790A stimulated a change in net transepithelial ion flux and conductance when added either bilaterally, or to the apical chamber, suggesting TRPV4 is functioning from the apical membrane in the choroid plexus epithelium (CPe) cells. C: pre- and posttreatment with TRPV4 antagonist, RN 1734 to determine specificity of the TRPV4-stimulated response. In the pink trace, RN 1734 added bilaterally at time t = −10 min. In the teal trace, RN 1734 added bilaterally at time t = 5 min. GSK1016790A added apically to all cultures at time t = 0 min. Pre- and posttreatment with RN 1734 inhibited the TRPV4 response, confirming specificity and reversibility of the TRPV4 channel activation. For Ussing-style electrophysiology, all traces represent means ± SE for the stated “n” of technical replicates. *⁁,#,P < 0.05 considered significant between different conditions and GSK control measured by multiple t tests grouped analysis and indicated by color. D: amount of fluid produced by HIBCPP cells upon stimulation with a TRPV4 agonist, GSK1016790A, expressed as microliters per cm²/10 min. GSK, GSK1016790A; ZO1, zona occludin 1.

To establish the sidedness of the TRPV4 transporter, the TRPV4-specific agonist was added bilaterally, or to either the apical or basolateral membrane of the HIBCPP cells. Addition of GSK1016790A either bilaterally or to the apical membrane stimulates a comparable and significant change in net transepithelial ion flux and conductance (Fig. 4B). In comparison, addition of GSK1016790A to the basolateral membrane stimulates a minor change in net transepithelial ion flux and conductance (Fig. 4B). These results suggest that TRPV4 is localized to, and can be activated from, the apical membrane in the HIBCPP cell line.

To confirm that the change in I_sc and conductance is specific to TRPV4, a pre- and posttreatment was performed with RN 1734, a TRPV4-specific antagonist. Pretreatment with RN 1734 completely inhibited the GSK1016790A stimulated change in net transepithelial ion flux and conductance (Fig. 4C). Similarly, posttreatment with RN 1734 prevented any further change in net transepithelial ion flux and, significantly, reversed the conductance stimulated by the addition of GSK1016790A (Fig. 4C).

TRPV4 activation stimulates a change in fluid secretion in the HIBCPP cell line.

To determine if stimulation of electrolyte flux is accompanied by a measurable increase in fluid secretion, the cells were grown on permeable supports as per the electrophysiology experiments. Both apical and serosal bathing media volumes were carefully measured before the experiment and adjusted to eliminate any hydrostatic pressure. Vehicle (DMSO), RN 1734, and GSK1016790A were all added to the serosal bathing media for 10 min. Subsequently, the volume of the apical chamber was measured and compared with the prestimulation volume. The vehicle-treated samples had a mean net increase in fluid secretion of 1.34 μL/cm²/10 min, a value not statistically different from zero. There was a significant 18.7-fold increase in fluid secretion to 25.11 μL/cm²/10 min in response to GSK1016790A. The TRPV4 agonist-mediated increase in fluid secretion was completely blocked by a preincubation with the TRPV4 antagonist (Fig. 4D).

TRPV4 localization in the HIBCPP.

The apical localization of TRPV4 was confirmed via immunocytochemistry, visualized in Fig. 5A, which is consistent with its apical polarization in the native CP tissue (16, 35). Interestingly, the localization of TRPV4 is not affected by agonist treatment (Fig. 5A). Again, this supports the use of the HIBCPP cell line as an in vitro culture model that retains appropriate localization of major transporters, including TRPV4.

Figure 5.

Effect of transient receptor potential vanilloid 4 (TRPV4) activation on channel localization and junctional integrity. A: maximum intensity projections and three-dimensional (3-D) renderings of the −xz plane for permeable supports stained for TRPV4 under three conditions: untreated, DMSO vehicle control, and GSK1016790A treated for 10 min. Immunolocalization confirms TRPV4 is expressed on the apical membrane, and targets to apical membrane under stimulation with agonist. No DMSO effect. B: maximum intensity projections of permeable supports stained for junctional protein zona-occludens 1 (ZO1) under three conditions: untreated, DMSO vehicle control, and GSK1016790A treated for 10 min. TRPV4 activation does not cause loss of ZO1 staining in the junctional complexes. Scale bars = 50 μm. A, apical; BL, basolateral; GSK, GSK1016790A.

TRPV4 conductance changes occur independently of changes in junctional protein localization.

In a 2015 study of sheep CP epithelial cells, Narita et al. (36) reported that stimulation of TRPV4 with the agonist GSK1016790A caused an increase in barrier permeability due to the breakdown of junctional proteins (36). Our laboratory found this not to be the case for the PCP-R cell line, a comparable large animal culture model (16). To replicate these investigations in the human cell line, HIBCPP cells were incubated for 10 min with apically delivered vehicle (DMSO) or 15 nM GSK1016790A, fixed and then stained for the junctional complex protein, ZO-1. Significantly, the immunofluorescence for ZO-1 was comparable between conditions, with no increase in cytosolic expression observed, indicating that the TRPV4-mediated increase in barrier permeability at low agonist concentration occurs independently of junctional complex breakdown in our model (Fig. 5B).

TRPV4-Stimulated Transepithelial Ion Flux Is Altered by Changes in Intracellular Ca²⁺

Effect of Ca²⁺ modulators on baseline and TRPV4-mediated transepithelial ion flux.

TRPV4 is a well characterized, cation permeable channel, with known selectivity for Ca²⁺ (35, 37). Since Ca²⁺ levels inside the cell are tightly regulated, it was necessary to consider the effect of Ca²⁺ modulators on baseline, as well as TRPV4-mediated transepithelial ion flux in the HIBCPP cell line. To do so, the HIBCPP cells were preincubated with two Ca²⁺ modulators: ionomycin or thapsigargin, before adding GSK1016790A, a TRPV4 agonist, to all cultures during electrophysiology. Ionomycin is an ionophore that functions to increase intracellular Ca²⁺ levels across the plasma membrane (although also depletes intracellular Ca²⁺ stores) (38); whereas thapsigargin inhibits the SERCA pump in the endoplasmic reticulum (ER), increasing cytoplasmic Ca²⁺ by inhibiting ER-mediated Ca²⁺ uptake (39).

Preincubation with ionomycin partially inhibited the TRPV4-mediated change in net transepithelial ion flux and increase in conductance (Fig. 6A). Similarly, preincubation with thapsigargin partially inhibited the TRPV4-mediated change in both ion flux and conductance, though to a greater magnitude than pretreatment with ionomycin (Fig. 6B). These results suggest that TRPV4 signaling is sensitive to intracellular Ca²⁺ levels, particularly Ca²⁺ released from the ER, visualized in Fig. 6C.

Figure 6.

Role of intracellular calcium in transient receptor potential vanilloid 4 (TRPV4)-mediated changes to ion flux and barrier permeability. Ussing-style electrophysiology graphs are displayed as short-circuit current (I_sc) (left) and conductance (right) for the same conditions. A: ionomycin was added bilaterally to the human choroid plexus papilloma (HIBCPP) cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with ionomycin partially inhibited the TRPV4-mediated change in electrogenic ion flux and increase in conductance. B: thapsigargin was added bilaterally to the HIBCPP cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with thapsigargin partially inhibited the TRPV4-mediated change in electrogenic ion flux and increase in conductance, though to a greater magnitude than the equivalent ionomycin pretreatment. Traces represent means ± SE for the stated “n” of technical replicates. *P < 0.05 considered significant between condition and GSK control measured by multiple t tests grouped analysis and indicated by color. C: schematic created in BioRender representing the role of intracellular calcium on the TRPV4 response. Ca²⁺, calcium; ER, endoplasmic reticulum; GSK, GSK1016790A; Iono, Ionomycin; Na⁺, sodium; Thap, Thapsigargin.

Investigating TRPV4-Mediated Signaling in the HIBCPP Cell Line

PLC activation inhibits TRPV4 activity in the CPe.

In HEK-293 cells, TRPV4 activation by either heat or osmotic stimuli was found to be dependent on the presence of phosphatidylinositol 4,5-bisphosphate (PIP₂). Furthermore, activation of the PLC pathway was shown to inhibit TRPV4 activity (40, 41). Conversely, in the distal nephron, G_q/11-dependent activation of PLC and, likely, PKC, was linked to potentiating TRPV4 activity (42). Canonical PLC signaling involves the conversion of PIP₂ to diacylglycerol (DAG) and inositol trisphosphate (IP₃). IP₃ can then stimulate the release of Ca²⁺ from stores in the ER. Downstream effectors of this pathway include PKC which, depending on isoform, can be activated by DAG and/or intracellular Ca²⁺ (41). To explore the potential interaction between PLC and TRPV4 in the CPe, HIBCPP cells were pretreated with either a PLC inhibitor or a PLC activator for 10 min. The goal of this study was to determine whether altering PLC activity would influence TRPV4-stimulated electrogenic ion flux and/or conductance. Importantly, RT-PCR was performed to confirm gene expression of PLC in the HIBCPP cell line; all isoforms of PLC were expressed in the line (Hulme and Blazer-Yost, unpublished data).

Bilateral pretreatment with edelfosine, a PLC inhibitor, marginally inhibited the TRPV4-mediated change in net transepithelial ion flux, but not barrier conductance (Fig. 7A). In contrast, bilateral pretreatment with m-3M3FBS, a specific PLC activator, completely inhibited the TRPV4-mediated change in ion flux and increase in conductance (Fig. 7C). Significantly, preincubation with either thapsigargin or the PLC activator, both of which functioning to increase cytosolic Ca²⁺ levels by promoting Ca²⁺ release from the ER, demonstrated inhibition of the TRPV4 response (Fig. 6B and Fig. 7, C and D). Taken together, these results indicate a link between PLC and TRPV4-mediated ion flux in the HIBCPP cell line. One potential pathway is outlined graphically in Fig. 7E, reinforcing the importance of intracellular Ca²⁺ in TRPV4-mediated signaling in the HIBCPP cells.

Figure 7.

Role of phospholipase C (PLC) in transient receptor potential vanilloid 4 (TRPV4)-mediated changes to ion flux and barrier permeability. Ussing-style electrophysiology graphs are displayed as short-circuit current (I_sc) (left) and conductance (right) for the same conditions. A: edelfosine, a PLC inhibitor, was added bilaterally to the human choroid plexus papilloma (HIBCPP) cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with edelfosine partially inhibited the TRPV4-mediated change in electrogenic ion flux but not conductance. B: edelfosine sidedness. Edelfosine was added to either the apical or basolateral chamber at t = −10 (pink trace and teal trace, respectively); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with edelfosine from the apical side (pink trace) results in an immediate, transient change in I_sc but no change to the TRPV4-mediated response. Pretreatment with edelfosine from the basolateral side (teal trace) has no effect on baseline I_sc or the TRPV4 response until the final phase. C: m-3M3FBS, a PLC activator, was added bilaterally to the HIBCPP cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with m-3M3FBS completely inhibited the TRPV4-mediated change in net transepithelial ion flux and increase in conductance. D: m-3M3FBS sidedness. m-3M3FBS was added to either the apical or basolateral chamber at t = −10 (pink trace and teal trace, respectively); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with m-3M3FBS from the apical (pink trace) and basolateral side (teal trace) partially inhibits the TRPV4 response. Preincubation of the HIBCPP cells with m-3M3FBS from the apical side inhibits the TRPV4 response to a greater magnitude. Traces represent means ± SE for the stated “n” of technical replicates. *P < 0.05 considered significant between condition and GSK control measured by multiple t tests grouped analysis and indicated by color. E: schematic created in BioRender representing the role of PLC and thus intracellular Ca²⁺ in TRPV4-mediated signaling. Ca²⁺, calcium; DAG, diacylglycerol; Ed, edelfosine; ER, endoplasmic reticulum; GPCR, G-protein coupled receptor; GSK, GSK1016790A; IP₃, inositol trisphosphate; m-3, m-3M3FBS; Na⁺, sodium; PIP₂, phosphatidylinositol 4,5-bisphosphate.

To establish if there is a polarization to the PLC effect in the HIBCPP cell line, these cells were pretreated with both a PLC inhibitor and a PLC activator, added to either the apical or basolateral side of the Ussing Chamber setup. Preincubation with edelfosine, a PLC inhibitor, either from the apical or basolateral side, had little effect on TRPV4-mediated changes in the initial phase of electrogenic ion flux. The effect on I_sc was more pronounced in the downward direction causing the edelfosine-treated tissue to return to a level below the initial baseline (Fig. 7B). Interestingly, however, the initial (minor) effect on baseline I_sc observed during bilateral pretreatment with edelfosine was only stimulated when the edelfosine was added to the apical chamber, a treatment that also resulted in a slight increase in conductance (Fig. 7B), suggesting that the link between PLC and an electrogenic ion flux mechanism is mediated from the apical membrane. Preincubation with the PLC activator, either from the apical or basolateral side, partially inhibited the TRPV4-mediated change in net transepithelial ion flux and increase in conductance (Fig. 7D). However, a greater inhibition of the TRPV4 response was observed when the PLC activator was added to the apical chamber (Fig. 7D). These results suggest that PLC is present on both membranes in the HIBCPP cell line, though, it seems PLC localized to the apical membrane plays a slightly greater role in the initial upward phase of the I_sc. Conversely, the basolateral inhibition of PLC activity had a more pronounced effect on the second phase, again resulting in a sustained I_sc that was below the starting baseline. However, this effect could be alternatively explained as PLC, although anchored to the plasma membrane, is not a transmembrane protein. For example, the basolateral membrane may simply be less permeable to effectors than the apical membrane to limit entry from the blood in vivo and maintain the blood CSF barrier.

PKC activity modulates TRPV4 responses in the CPe.

Fan et al. (43) reported that activation of PKC can potentiate TRPV4 activity in vitro in HEK293 cells (43). PKC directly phosphorylates TRPV4 at serine 824, regulating TRPV4-mediated Ca²⁺ influx (44). In the blood-brain-barrier (BBB), TRPV4 and PKC-δ were found to be functionally linked, involved in a positive feedback loop that regulated BBB permeability (43), suggesting PKC can function both upstream and downstream of TRPV4. In fact, in the aortae of hypertensive mice, TRPV4 is thought to activate cytosolic phospholipase A2 activity through regulating a Ca²⁺/PKC signaling axis (45). Therefore, we investigated the potential role of PKC in the TRPV4 response in the CP by considering the effect of PKC activity on TRPV4-mediated ion flux via electrophysiology. Importantly, a large number of the PKC isoforms are expressed at the mRNA level in the HIBCPP cell line (unpublished data).

Preincubation of the HIBCPP cell line for 10 min with a nonselective PKC inhibitor, tamoxifen, inhibited the TRPV4-mediated change in electrogenic ion flux and increase in conductance (Fig. 8A). Since tamoxifen is known to affect alternative targets, including histamine, muscarinic, and dopamine D2 receptors (46), an additional experiment with a more specific PKC inhibitor, Go 6976, was performed. Like tamoxifen, preincubation of the human CP cells with Go 6976 for 10 min before the addition of GSK1016790A inhibited both the TRPV4-mediated change in net transepithelial ion flux and conductance (Fig. 8B), confirming this effect was a direct consequence of PKC inhibition. Phorbol 12,13-dibutyrate (PDBu) was used as an indirect activator of PKC activity. Phorbol esters are analogs of DAG that is produced by PLC activation. Interestingly, addition of PDBu resulted in a significant increase in transepithelial ion flux that was similar in time and phase to the TRPV4 response (Fig. 8C). When the TRPV4 agonist was added to the cells that had been preincubated with PDBu, the HIBCPP cell line demonstrated a complex response, including an apparent inhibition of the initial increase in I_sc but a potentiation and then overshoot during the return to baseline (Fig. 8C). These results suggested that DAG-mediated PKC-activity is important for TRPV4 signaling in the HIBCPP cell line, a pathway outlined in Fig. 8D. In agreement, pretreatment with a TRPV4-specific antagonist, RN 1734, inhibited the change in I_sc stimulated by the addition of PDBu (Fig. 8C), confirming that TRPV4 activity is necessary for the DAG-mediated change in transepithelial ion transport in the HIBCPP cell line.

Figure 8.

Role of protein kinase C (PKC) in transient receptor potential vanilloid 4 (TRPV4)-mediated changes to ion flux and barrier permeability. Ussing-style electrophysiology graphs are displayed as short-circuit current (I_sc) (left) and conductance (right) for the same conditions. A: tamoxifen, a nonspecific PKC inhibitor, was added bilaterally to the human choroid plexus papilloma (HIBCPP) cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with tamoxifen inhibited the TRPV4-mediated change in electrogenic ion flux and conductance. B: Go 6976, a more specific PKC inhibitor, was added bilaterally to the HIBCPP cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with Go 6976 similarly inhibited the TRPV4-mediated change in electrogenic ion flux and conductance. C: phorbol 12,13-dibutyrate (PDBu), a PKC activator and diacylglycerol (DAG) mimetic, was added bilaterally to the HIBCPP cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Addition of PDBu immediately stimulated a peak in baseline I_sc. Pretreatment with PDBu partially inhibited the TRPV4-mediated change in net transepithelial ion flux but not conductance. Pretreatment with RN 1734, a TRPV4 antagonist, added at t = −10 min completely inhibited the change in transepithelial transport stimulated by the addition of PDBu added at t = 0 min (teal trace). Traces represent means ± SE for the stated “n” of technical replicates. *P < 0.05, **P < 0.01; ***P < 0.001 considered significant between condition and GSK control measured by multiple t tests grouped analysis and indicated by color. D: schematic representing the role of PKC in TRPV4-mediated signaling. Schematic created in BioRender. Ca²⁺, calcium; ER, endoplasmic reticulum; Go, Go 6976; GSK, GSK1016790A; Na⁺, sodium; PKC, protein kinase C; RN, RN 1734; Tam, tamoxifen.

PI3K inhibition ablates TRPV4 activity in the CP.

TRPV4 can specifically activate the PI3K/Akt signaling pathway in other cell types (47). PI3K has also been implicated in the development of primary hydrocephalus in mouse and human infants. In particular, PI3K-catalytic α (PI3KCA, or p110α) and PTEN (phosphatase and tensin homolog) signaling via the PI3K-Akt-mTOR (mammalian target of rapamycin) pathway correlate with impaired neurogenesis, disruption of the subventricular zone, and ultimately, ventricular distension (48). The major PI3K subunits, except catalytic-γ, were found to be expressed at the mRNA level in the HIBCPP cell line (data not shown). Thus, utilizing Ussing-style electrophysiology, we evaluated the effect of PI3K modulators on the TRPV4 response in the HIBCPP cells, pretreating the line for 10 min with either a pan PI3K inhibitor, LY-294002 hydrochloride, or a pan PI3K activator, 740-Y-P.

Preincubation with the PI3K inhibitor partially inhibited the TRPV4-mediated change in net transepithelial ion flux and increase in conductance (Fig. 9A), implicating PI3K as a modulator of TRPV4 signaling in HIBCPP cell line. Conversely, pretreatment with the PI3K activator had no effect on baseline ion flux or the TRPV4 response (Fig. 9B). This indicates that PI3K does not contribute to baseline electrogenic ion flux necessary for TRPV4 activation, but rather PI3K may be a necessary regulatory kinase required for downstream TRPV4 signaling. In summary, these results suggest that PI3K activation is necessary for TRPV4 activity and thus PI3K is involved in TRPV4-mediated signaling in the human CPe, outlined in Fig. 8D; however, TRPV4 activity cannot be potentiated through further activation of PI3K, suggesting TRPV4 is maximally activated by the TRPV4 agonist.

Figure 9.

Role of phosphatidylinositol 3 kinase (PI3K) in transient receptor potential vanilloid 4 (TRPV4)-mediated changes to ion flux and barrier permeability. Ussing-style electrophysiology graphs are displayed as short-circuit current (I_sc) (left) and conductance (right) for the same conditions. A: LY-294002 hydrochloride, a PI3K inhibitor, was added bilaterally to the human choroid plexus papilloma (HIBCPP) cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with LY-294002 hydrochloride partially inhibited the TRPV4-mediated change in electrogenic ion flux and conductance. B: 740-Y-P, a PI3K activator, was added bilaterally to the HIBCPP cells at time t = −10 min (pink trace); GSK1016790A was added apically at time t = 0 min to all cultures. Pretreatment with 740-Y-P had little effect on the TRPV4-mediated change in electrogenic ion flux or conductance. C: LY-294002 hydrochloride, added at t = −10 min, inhibited the change in transepithelial ion transport stimulated by the addition of phorbol 12,13-dibutyrate (PDBu), a protein kinase C (PKC) activator, added at t = 0 (pink trace). Traces represent means ± SE for the stated “n” of technical replicates. *P < 0.05 considered significant between condition and GSK control measured by multiple t tests grouped analysis and indicated by color. D: schematic created in BioRender representing the involvement of PI3K in TRPV4-mediated signaling. 740, 740-Y-P; Ca²⁺, calcium; EGFR, epithelial growth factor receptor; LY, LY-294002 Hydrochloride; GSK, GSK1016790A; Na⁺, sodium.

To establish a link between PI3K and PKC in the TRPV4-mediated signaling pathway in the CPe, the HIBCPP model was pretreated for 10 min with LY-294002 hydrochloride before PDBu, a DAG mimetic, was added to the cells. Interestingly, inhibition of PI3K activity partially inhibited and altered the phase of the response in terms of electrogenic ion flux stimulated by the addition of PDBu (Fig. 9C). Inhibition of PI3K also inhibited the moderate PDBu-mediated change in conductance (Fig. 9C). These results implicate a link between PI3K and DAG-mediated PKC activity in the HIBCPP cell line, considered in Fig. 9D.

DISCUSSION

The CPe is a unique barrier structure, poised as a master regulator of CSF composition and volume, as well as the BCSFB and, thus, immune cell permeation/activity, and communication with the parenchyma and circumventricular organs (1). The CPe controls secretion of fluid and signaling molecules into the ventricular cavities. Yet, unlike other epithelial tissues, knowledge regarding the regulation of CPe transporters remains elusive. Dysregulation of the CSF is implicated in a variety of neurological conditions, thereby underscoring the importance of studying transporter function in a human-derived CPe model.

Previously, the HIBCPP cell line was optimized for studies of bacterial invasion (12), where the cells were grown on smaller 12-mm diameter permeable supports. Unfortunately, these smaller supports are suboptimal for Ussing-style electrophysiology experiments, a time-honored, technique that facilitates investigation into transporter function in epithelial cells (49). Moreover, these cells demonstrated slow proliferation and enhanced sensitivity to culture conditions compared with other continuous CPe cell lines. Here, we have successfully devised a culture protocol that optimizes growth conditions for the HIBCPP line in 30-mm diameter permeable supports. Importantly, the cells develop into epithelial sheets that are largely monolayered, cobblestone-like in appearance, demonstrate intermediate resistance TEER values that confirms the formation of a barrier epithelium, and form an immature brush border that is densely populated with microvilli. The somewhat heterogeneous morphology of the monolayer could be explained by cells in different stages of differentiation. We do not feel there are mixed cell types present because the initial studies by Ishiwata et al. (14) demonstrated the cells were keratin positive. The HIBCPP cell line also demonstrated gene expression of CP fate markers, appropriate junctional complex proteins, and numerous transporters identified in the native CP tissue.

The polarization of various transporters in the CPe is largely conserved across species. As a result, previously reported transporter localization can be utilized to determine the expected orientation in cultured CPe cells. The HIBCPP model, unlike some other CPe models, maintains its capacity to correctly polarize transporters to either the apical or basolateral plasma membrane, including the Na-K-ATPase pump, TRPV4, and the AE2 transporter. The results displayed in Figs. 1, 2, 3, 4, and 5 exemplify the HIBCPP cell line as an accurate and representative model of the human BCSFB that can be utilized in future investigations considering barrier function, as well as CSF production.

Next, the HIBCPP cell line was used to investigate the contribution of the TRPV4 channel to transepithelial electrogenic ion flux and barrier permeability in the CPe. TRPV4 has previously been implicated as a regulatory hub protein, capable of modulating fluid homeostasis in the CPe and other tissues (16, 17, 36). In Ussing Chamber Electrophysiology, I_sc is a measure of net electrogenic ion flux across confluent epithelial monolayers. By convention, a positive deflection indicates anion secretion and/or cation absorption and a negative deflection represents the reverse. Barrier conductance is the inverse of the TEER, thus an increase in conductance denotes an increase in the permeability of the monolayer to solutes.

At baseline, cultures exhibited a modest level of net positive baseline ion transport, indicating that either active anion secretion or cation absorption, or both in tandem, function at homeostatic levels in the HIBCPP cell line. GSK1016790A, a TRPV4 agonist, was added to the HIBCPP cultures to investigate the effect of TRPV4 activity on I_sc and conductance. When the TRPV4 agonist was added either bilaterally, or to the apical bathing media, a large multiphasic change in net electrogenic ion flux occurred, coupled with a substantial and persistent increase in conductance. This response was confirmed to be specific to the TRPV4 transporter as either pre- or posttreatment with a TRPV4-specific antagonist (RN 1734) inhibited both the change in electrogenic ion flux and conductance. These results confirm that TRPV4 is involved in transepithelial ion flux in the CPe, and indicate that this channel is activated from, and thus localized to, the apical membrane in the HIBCPP cell line. The latter was confirmed via immunohistochemistry.

Positive and negative deflections in the I_sc after activation of TRPV4 appear to be more than just “on” and “off” phases of the response. Instead, these changes may be due to differential regulation of two or more electrogenic fluxes that are slightly phase shifted. This interpretation is based on the concentration dependence of the response (Fig. 4A) wherein the positive deflection is independent of TRPV4 agonist concentration, within the small range of agonist that we tested, whereas the negative deflection is concentration dependent, with the highest agonist concentration demonstrating the most rapid negative deflection. This explanation is also supported by the sustained increase in conductance when the I_sc has returned to at, or below, the initial baseline level, indicating a substantial remaining transepithelial flux and/or changes in paracellular permeability.

Since previous studies have shown that antagonism of TRPV4 deceases the amount of CSF in the ventricles of hydrocephalic rats, one can postulate that activation of TRPV4 results in secondary activation of channels in response to the TRPV4-mediated increase in Ca²⁺ and Na⁺ (17). Although the exact nature of the resulting secondary channels or transporters is unknown, it is likely that the secretory response is due to a net secretion of anions during the upward phase and a net cation secretory response during the downward phase. Although the GSK1016790A-stimulated, extended increase in barrier permeability is independent of junctional protein breakdown, we cannot exclude more physiologically relevant changes in junctional permeability. The contributions of junctional permeability and transmembrane flux changes are not mutually exclusive.

The changes in ion flux and conductance are accompanied by a substantial fluid secretion. When stimulated with the TRPV4 agonist, the HIBCPP cells secrete fluid at a rate of over 150 µL/cm² per hour which is consistent with the in vivo tissue that secretes a total of ∼500 mL/day in an adult human.

To elucidate intracellular modulators involved in the TRPV4-mediated signaling pathways that may contribute to the regulation of CSF production in the CPe, the HIBCPP cell line was pretreated with select effectors of pathways that have been shown to be modulated by TRPV4 in other tissues. In the first experiments, we investigated the canonical effects of TRPV4 activity, namely an influx of Ca²⁺. To determine if the transepithelial ion flux mediated by TRPV4 activity was simply a consequence of increased intracellular Ca²⁺, this parameter was changed by other effectors. Ionomycin is a Ca²⁺ ionophore that allows the influx of Ca²⁺ from the external milieu, whereas thapsigargin functions to inhibit the uptake of Ca²⁺ into the ER, thereby increasing intracellular Ca²⁺ levels. Interestingly, we found that TRPV4 activity is sensitive to intracellular Ca²⁺ levels in the HIBCPP cells. Preincubation with either ionomycin or thapsigargin inhibited the TRPV4-mediated positive deflection in electrogenic ion flux (Fig. 6), suggesting that increasing cytosolic Ca²⁺ directly inhibits TRPV4 activity in the HIBCPP cell line, especially Ca²⁺ influx from the ER. Although difficult to determine in the absence of a substantial positive deflection, the increased intracellular Ca²⁺ may potentiate the magnitude of the downward deflection seen during a TRPV4 response. In other tissues, the relationship between TRPV4 and Ca²⁺ is reciprocal; TRPV4 is known to regulate Ca²⁺ homeostasis (50), but also changes in intracellular Ca²⁺ are believed to modulate TRPV4 activity, either inhibiting or potentiating channel activity depending on the concentration of Ca²⁺ (51–53). It should be stressed that an increase in intracellular Ca²⁺ by either agent did not, by itself, significantly change electrogenic ion flux compared with control, indicating that the TRPV4 response requires more than simply a change in Ca²⁺ influx. The magnitude of these calcium gradient modulations remains to be explored and could be addressed by use of radiometric calcium dyes.

PLC has been associated with TRPV4 activity in other cell types (42,43, 54). Addition of edelfosine functions to reduce intracellular Ca²⁺ levels by reducing canonical PLC-DAG-IP₃ signaling and thus, Ca²⁺ influx from the ER. It may be specifically this alteration in intracellular Ca²⁺ levels that is functionally linked to a mechanism that regulates transepithelial ion flux in the HIBCPP cell line. Bilateral inhibition of PLC activity partially inhibited the TRPV4-mediated change in electrogenic ion flux (Fig. 7A). Interestingly, addition of the PLC inhibitor to each side of the polarized epithelium (individually) significantly potentiated the downward deflection of the response. Concurrent with its effect on intracellular Ca²⁺, activation of PLC causes the activation of classical PKC family members that are stimulated in response to increases in Ca²⁺ and DAG (55). It is possible that inhibition of PLC activity is inhibiting the TRPV4 response through inhibiting the activation of PKC, via DAG, a mechanism that requires further consideration.

On the other hand, bilateral activation of PLC functions to completely inhibit TRPV4-mediated electrogenic ion flux in both the upward and downward directions (Fig. 7C). Activation of PLC will stimulate an influx of Ca²⁺ from the ER into the cytosol, altering Ca²⁺ gradients inside the cell. It may be that TRPV4 activity in the CP is sensitive to a threshold effect involving intracellular Ca²⁺ levels. In theory, further activation of PLC may switch off TRPV4 activity by increasing the cytosolic Ca²⁺ levels above a threshold that is required to maintain a concentration gradient driving TRPV4 to transport Ca²⁺ into the cell. This would agree with the results recorded during pretreatment with thapsigargin in the HIBCPP cells (Fig. 6B), which similarly functions to increase cytoplasmic Ca²⁺ from the ER. The inhibitory effect of PLC inhibition on TRPV4 activity was observed when m-3M3FBS was added to either the apical or basolateral chamber in the Ussing setup, though a greater effect was noted from the apical side on the upward direction while the downward deflection was more sensitive to the basally added activator (Fig. 7D). These results suggest that PLC is present on both the apical and basolateral membrane of the human cells; however, apically localized PLC may have a greater functional link with TRPV4 simply due to its closer proximity to the channel on the apical membrane or alternatively, the apical membrane is less permeable to the PLC-directed effectors. The hypothesized signaling pathway between PLC and TRPV4 in the CP is detailed in Fig. 7D.

The link between PKC and TRPV4 has been established in many other tissues (43, 44, 56). Inhibition of PKC via the relatively nonspecific inhibitor, tamoxifen, and the more selective inhibitor, Go 6976, substantially inhibited both phases of the TRPV4 response in the HIBCPP cell line (Fig. 8, A and B), confirming the involvement of PKC in TRPV4-mediated transepithelial transport. Indirect activation of PKC through the addition of PDBu alone stimulated a response in electrogenic ion flux, similar to that observed when TRPV4 is activated directly by GSK1016790A and is inhibited by a TRPV4 antagonist. This is of particular interest because phorbol esters are considered mimetics of DAG, which can be generated by the activation of PLC. The results suggest that PKC may be functioning downstream of TRPV4 in the human cell line, directly, or via additional downstream effectors, mediating electrogenic ion transport. Independent activation of PKC therefore cannot potentiate the TRPV4 response if PKC is already maximally activated by PDBu, hence the partial inhibition of the positive deflection of the TRPV4-mediated change in net transepithelial ion flux observed when the HIBCPP cells are pretreated with the DAG mimetic (Fig. 8C). Interestingly, the second phase of the TRPV4 response appears to be potentiated by the PDBu. The hypothesized pathway linking TRPV4 and PKC in the CP is outlined in Fig. 8D.

This PDBu experiment is interesting from two perspectives. Initially, one would predict that if PLC activation (and consequent DAG production) inhibited TRPV4 activity, the phorbol esters (PDBu—a DAG mimetic) would not be expected to stimulate a baseline response that is similar to the TRPV4-induced ion flux, and which is inhibited by a TRPV4 antagonist. Although our experiments did not address this paradox, there are several logical explanations. The first is the strength of the effect on TRPV4 by the interplay between changes in intracellular Ca²⁺ and DAG, both of which are stimulated in response to PLC activation. It may be that increases in intracellular Ca²⁺ concentrations above a set threshold, as aforementioned, inhibits TRPV4 activity, regardless of the DAG-mediated effects on PKC-TRPV4 activity. Another aspect is the relative potency of exogenous phorbol ester stimulation relative to the DAG that is produced intracellularly and, likely, in modest amounts. There is also the complication that numerous isoforms of PKC are present in the CPe cells (Hulme and Blazer-Yost, unpublished data) and that these might be differentially regulated. The other noteworthy component revealed during the experiments with PDBu was the differential effects on different phases of the TRPV4-mediated electrogenic ion flux. It is necessary to investigate the complex interactions between PLC-DAG-PKC-TRPV4 in the HIBCPP cell line in future studies.

Finally, TRPV4 and PI3K have been functionally linked in other tissues (47, 48); thus, PI3K was also explored in terms of its involvement in TRPV4 signaling in the CPe. Pretreatment with a PI3K inhibitor LY-294002 hydrochloride, partially inhibited the TRPV4-mediated change in net transepithelial ion flux and conductance (Fig. 9A), implicating PI3K in TRPV4-mediated signaling in the HIBCPP cell line. In contrast, pretreatment with a PI3K activator, 740-Y-P, had little effect on the TRPV4 response (Fig. 9B). These results suggest a minor involvement of PI3K in TRPV4 activity in the CPe, though the TRPV4 response cannot be further potentiated through this signaling axis. The proposed pathway linking PI3K and TRPV4 in the CP is described in Fig. 9D. PI3K has also been linked to downstream PKC activity (57). Critically, a functional linkage between PI3K and PKC has been confirmed in the HIBCPP cell line as pretreatment with the PI3K inhibitor was found to modulate the DAG-mediated change in both electrogenic ion flux and conductance (Fig. 9C). This experiment suggests that PI3K either modulates TRPV4 activity in the CP through modulating TRPV4 directly, or via PKC, which in turn regulates electrogenic ion flux.

In summary, our studies have demonstrated a role for TRPV4 in transepithelial ion flux and changes in barrier conductance in the CPe and, further, have linked this TRPV4 response to a variety of additional effectors, including PLC, PKC, and PI3K (Fig. 10). We have also used the cell line to show a substantial secretion of fluid in response to TRPV4 stimulation. It is important to note that Ussing-style electrophysiology studies are limited to the measurement of net transepithelial electrogenic flux and will not detect the modulation of electroneutral transporters, such as the Na⁺-K⁺-Cl⁻ cotransporter (NKCC1), also known to be present in the CPe. In conclusion, the optimized protocol for the growth of the HIBCPP cell line provides a model of the BCSFB that has already offered insight into some of the modulators associated with TRPV4 activation in the CPe, visualized in Fig. 10. Additional studies with this new in vitro tool will provide further insights that may be useful for understanding the complexities of CSF production.

Figure 10.

Schematic representing the pathways explored and their potential involvement in transient receptor potential vanilloid 4 (TRPV4)-mediated signaling in the human choroid plexus papilloma (HIBCPP) cell line. Schematic created in BioRender. DAG, diacylglycerol; ER, endoplasmic reticulum; IP₃, inositol trisphosphate; Na⁺, sodium; PI3K, phosphoinositide 3-kinase; PIP₂, phosphatidylinositol 4,5-bisphosphate; PKC, protein kinase C; PLC, phospholipase C.

GRANTS

This study was supported by the US Department of Defense Investigator-Initiated Research Award W81XWH-16-PRMRP-IIRA (to B. Blazer-Yost).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

L.H., A.H., and B.B.-Y. conceived and designed research; L.H., A.H., C.-Y.T., and B.P. performed experiments; L.H. and A.H. analyzed data; L.H., A.H., and B.B.-Y. interpreted results of experiments; L.H. and A.H. prepared figures; L.H. and A.H. drafted manuscript; L.H., A.H., C.S., and B.B.-Y. edited and revised manuscript; L.H., A.H., C.S., H.S., H.I., C.-Y.T., B.P., and B.B.-Y. approved final version of manuscript.