The microvillar protocadherin CDHR5 associates with EBP50 to promote brush border assembly

Abstract

Transporting epithelial cells of the gut and kidney interact with their luminal environment through a densely packed collection of apical microvilli known as a brush border (BB). Proper brush border assembly depends on the intermicrovillar adhesion complex (IMAC), a protocadherin-based adhesion complex found at the distal tips of microvilli that mediates adhesion between neighboring protrusions to promote their organized packing. Loss of the IMAC adhesion molecule Cadherin-related family member 5 (CDHR5) results in significant brush border defects, though the functional properties of this protocadherin have not been thoroughly explored. Here, we show that the cytoplasmic tail of CDHR5 contributes to its correct apical targeting and functional properties in an isoform-specific manner. Library screening identified the Ezrin-associated scaffolds EBP50 and E3KARP as cytoplasmic binding partners for CDHR5. Consistent with this, loss of EBP50 disrupted proper brush border assembly with cells exhibiting markedly reduced apical IMAC levels. Together, our results shed light on the apical targeting determinants of CDHR5 and further define the interactome of the IMAC involved in brush border assembly.

INTRODUCTION

During their terminal differentiation, transporting epithelial cells of the gut and kidney assemble a dense array of actin-based microvilli on their apical surface that are specialized to mediate solute transport (Crawley et al., 2014a). Collectively, these microvilli are referred to as a brush border (BB) and act to amplify the apical surface area of the cell that is in contact with the external environment. As part of this optimization, these epithelial cells organize their microvilli into hexagonal arrays: a geometric pattern that allows for the maximum number of microvilli to occupy the apical surface of each cell to ensure peak solute transport. Indeed, each transporting epithelial cell assembles an organized collection of ∼1000 apical microvilli in their mature state. Perturbations to BB structure can be life threatening unless treated, as seen with infections of the gut by the attaching/effacing microbe enterohemorrhagic Escherichia coli (In et al., 2016) and Antibrush Border Antibody Disease in which autoantibodies damage the BB of the kidney (Arcoverde Fechine Brito et al., 2021).

Previous studies focused on understanding the underlying mechanism of BB development identified the intermicrovillar adhesion complex (IMAC), a protocadherin-based complex found at the distal tips of BB microvilli that creates physical linkages that connect neighboring microvilli together to control their organization (Figure 1A; Crawley et al., 2014b). These “intermicrovillar adhesion links” are a trans-heterophilic adhesion complex between Cadherin-related family member 2 (CDHR2; also known as protocadherin-24) and Cadherin-related family member 5 (CDHR5; also known as mucin-like protocadherin; Figure 1A; Crawley et al., 2014b). By forming physical connections between the distal tips of neighboring microvilli, the IMAC organizes these microvilli into discrete “tee pee-like” apical clusters during the early stages of BB assembly. As these clusters grow larger with the addition of new microvilli, they eventually amalgamate to form a unified, highly organized BB. Both in vitro and in vivo studies have shown that IMAC-mediated intermicrovillar adhesion plays a critical role in proper BB assembly (Crawley et al., 2014b, 2016; Weck et al., 2016; Choi et al., 2020). CDHR2 and CDHR5 knockout (KO) mice exhibit malformed enterocyte BBs, having short microvilli that do not attain their normal ordered dense packing (Pinette et al., 2019; Modl et al., 2023). For CDHR2, this correlates with the KO mice having a lower body weight compared with heterozygous littermates, likely due to a reduced functional capacity of their intestinal tissue in nutrient absorption. Studies revealed that proper targeting and function of CDHR2 requires association with a cytoplasmic complex composed of two scaffolding molecules, Ankyrin repeat and sterile α-motif domain containing 4B (ANKS4B) and USH1C (also known as harmonin), a myosin motor protein Myosin-7b (Myo7b), and the myosin light chain Calmodulin-like protein-4 (CALML4; Figure 1A; (Crawley et al., 2014b, 2016; Choi et al., 2020). In contrast, little is currently known about the cytoplasmic binding partners for CDHR5 and their role in correct function of this cadherin.

FIGURE 1:

The splice isoforms of CDHR5 exhibit different targeting properties in LLC-PK1-CL4 kidney epithelial cells. (A) Cartoon showing the location of intermicrovillar adhesion links found at the tips of BB microvilli and the interactome of the IMAC. (B) Diagram of the two CDHR5 splice isoforms, showing the MLRs of the human long isoform. (C–F) Confocal images of 4-d polarized LLC-PK1-CL4 cells stably expressing EGFP-tagged constructs tested (green) stained for F-actin (red). Boxed regions denote area in zoomed image panels. Dashed lines indicate the positions where the x-z section were taken; x-z sections are shown below each en face image. Arrows point to signal accumulation found in subapical and cytoplasmic puncta in CDHR5-L-EGFP expressing cells. Scale bars, 10 μm. Deletion of the cytoplasmic tail of CDHR5-L, but not CDHR5-S, influences apical targeting. (G) Scatterplot quantification of the BB:cytosol ratios of EGFP signal from cells expressing the CDHR5-S-EGFP and CDHR5-L-EGFP constructs tested. Bars indicate mean and SD. R.U = relative units. ***p < 0.0001, two-tailed t test. Measurements: CDHR5-S-EGFP, n = 35 BB:cyto ratios; CDHR5-L-EGFP, n = 35 BB:cyto ratios. (H and I) Scatterplot quantification of the BB:cytosol ratios of EGFP signal from cells expressing the EGFP-tagged CDHR5 variant constructs tested. Bars indicate mean and SD. R.U = relative units. ns = not significant, ***p < 0.0001, two-tailed t test. Measurements: CDHR5-S-EGFP, n = 31 BB:cyto ratios; CDHR5-S_ΔCD-EGFP, n = 35 BB:cyto ratios; CDHR5-L-EGFP, n = 60 BB:cyto ratios; CDHR5-L_ΔCD-EGFP, n = 60 BB:cyto ratios.

CDHR5 was originally discovered in a screen for proteins involved in branching morphogenesis that occurs during development of the kidney (Goldberg et al., 1997, 2000). Subsequently, CDHR5 was also identified to be highly expressed in the gut as two distinct splice isoforms (Goldberg et al., 2000; Moulton et al., 2004) and was found to be lost in most colorectal cancer tissue samples analyzed for the protein (Losi et al., 2011). Truncation of the cytoplasmic tail of CDHR5 inhibits its ability to promote BB assembly in cultured enterocytes, suggesting that cytoplasmic factors that associate with the tail are necessary for its proper function. In this study, we explored the functional properties of CDHR5 using cultured proximal tubule kidney epithelial cells and enterocytes as cell models that generate an apical BB. Using these systems, we questioned whether the two dominant splice isoforms of CDHR5 exhibited different targeting properties, and, if so, what sequence features guided these properties. These studies led us to further screen for potential cytoplasmic binding partners for CDHR5 to understand how they may contribute to its function. Together, our results shed light on how CDHR5 is delivered to and promotes formation of an apical BB, a critical interface found between transporting epithelia of the gut and kidney and their luminal environments.

RESULTS

The cytoplasmic tail of CDHR5-L, but not CDHR5-S, contributes to proper apical targeting

CDHR5 is expressed in transporting epithelial cells predominantly as two splice isoforms, a short isoform (CDHR5-S) and a long isoform (CDHR5-L; Figure 1B; Goldberg et al., 2000, 2002). The two CDHR5 isoforms are identical in sequence with the exception that the long isoform contains an additional membrane proximal mucin-like repeat (MLR) domain in its extracellular region. In humans, the CDHR5 MLR domain is a 4X tandem repeat sequence of 31 amino acids that is rich in threonine, serine and proline (TSP) residues, similar to a conventional TSP-rich region found in the secreted and transmembrane mucins of the gut (Figure 1B; Supplemental Figure S1A). Interestingly, the MLR sequence of CDHR5-L is divergent between species; mice have a 3X tandem TSP-repeat sequence in which the repeats exhibit a relatively low repeat identity (Supplemental Figure S1B), while rats have a 4X TSP-tandem repeat sequence with a repeat identity level similar to humans (Supplemental Figure S1C). In contrast, pigs have an MLR domain that more aptly fits a 2X TSP-rich tandem repeat sequence pattern of ∼70 amino acids (Supplemental Figure S1D). Despite this diversity, every species we examined had some form of MLR domain in their CDHR5 sequence. For our analysis, we focused on characterizing the functional differences between the two dominant human splice isoforms, CDHR5-L and CDHR5-S.

To compare their targeting properties, full-length human constructs of both CDHR5-S and CDHR5-L were stably expressed as C-terminally tagged EGFP-fusion proteins in LLC-PK1-CL4 kidney epithelial cells (Figure 1, C and D). LLC-PK1-CL4 cells are a pig kidney proximal tubule cell line that differentiate rapidly (∼3–4 d) to establish a robust BB on their apical surface (Tyska and Mooseker, 2002). We quantified the distribution of CDHR5 by measuring the EGFP signal found in the BB versus the cytosol. We observed that when overexpressed, CDHR5-S targets efficiently to apical microvilli, with very little material found in the cytoplasm (Figure 1, C and G). In striking contrast, a significant amount of the CDHR5-L signal appeared trapped internally with a much lower amount reaching the apical domain (Figure 1, D and G). This CDHR5-L signal often appeared as relatively large puncta in the cytoplasm (Figure 1D; arrows). To further explore the sequence features of CDHR5 that contribute to its proper apical targeting, we first assessed whether deletion of the tail would impact targeting of either of the two splice isoforms when overexpressed. While the absence of the tail from CDHR5-S had little impact on apical targeting (Figure 1, E and H), deleting the cytoplasmic tail from CDHR5-L resulted in a near-complete loss of protein targeting to apical microvilli (Figure 1, F and I). Together, these results demonstrate that the two CDHR5 splice isoforms exhibit significantly different targeting properties, and that the MLR domain found specific to CDHR5-L impedes efficient targeting to apical microvilli which is further compounded upon loss of the entire cytoplasmic tail.

The C-terminal PDZ-binding motif of CDHR5-L promotes apical delivery

The cytoplasmic tail of human CDHR5 is relatively small in size (154 amino acids), is rich in glycine and proline residues with potential poly-proline motifs, and also ends with a canonical PDZ binding motif (PBM) (Figure 2A). Sequence alignment across a number of different species revealed that the C-terminal PBM is somewhat variable, conforming to either a class I or class II PBM consensus sequence (Figure 2A). We speculated that tagging CDHR5-L with a C-terminal EGFP would likely disrupt the proper function of its PBM, because a canonical PBM requires a free C-terminal hydroxyl group as part of its binding interaction (Doyle et al., 1996; Hung and Sheng, 2002). To determine whether tagging CDHR5-L with a C-terminal EGFP was interfering with its targeting ability, we expressed an untagged version of this isoform and visualized its targeting using immunofluorescence. We first confirmed that our antibody directed against human CDHR5 did not detect signal from endogenous porcine CDHR5 that could be potentially found in our LLC-PK1-CL4 cells (Figure 2B). Strikingly, we observed that the untagged CDHR5-L had a significantly higher level of BB targeting compared with the EGFP-tagged molecule, though the levels were still below the targeting seen with CDHR5-S (Figure 2, C, D, and F). This result was confirmed using a second independent antibody against CDHR5 (Supplemental Figure S2, A and B). Both antibodies also detected the CDHR5 signal in large cytoplasmic puncta (Figure 2, C and D; Supplemental Figure S2B; arrows), as we observed with our EGFP-tagged CDHR5-L construct. In contrast, a similar experiment done with CDHR5-S did not detect differences between the tagged and untagged versions, further confirming that these two CDHR5 splice isoforms exhibit different sequence-dependent targeting properties (Supplemental Figure S2, C–E). To confirm the importance of the PBM of CDHR5-L, we mutated the C-terminal hydrophobic position to a charged residue (I845R) which is predicted to ablate the motif and expressed the mutant without an EGFP tag. Similar to the EGFP-tagged wild-type protein, mutation of the PBM significantly blocked apical targeting of CDHR5-L (Figure 2, E and F). In sum, these results highlight that the C-terminal PBM of CDHR5-L is essential for efficient targeting of this splice isoform.

FIGURE 2:

The cytoplasmic tail of CDHR5-L requires an intact C-terminal PBM to promote apical targeting of the cadherin. (A) Sequence alignment of the cytoplasmic tail of CDHR5 from Human (Homo sapiens), Rhesus Macaque (Macaca mulatta), Mouse (Mus musculus), Goat (Capra hircus), and Pig (Sus scrofa). Amino acids shaded in black are identical. Amino acids shaded in gray are similar. Amino acids colored in red denote PDZ binding motifs (PBM) X = any residue, Φ = hydrophobic residue. (B–E) Confocal images of 4-d polarized LLC-PK1-CL4 cells stably transduced with an empty-vector control, CDHR5-L with no tag (PBM intact) and CDHR5-L-EGFP (PBM blocked) and CDHR5-L_I845R with no tag. Monolayers have been stained for F-actin (red) and an anti-CDHR5 antibody (green; antibody HPA009081) raised against the human CDHR5 sequence. Boxed regions denote area in zoomed image panels. Dashed lines indicate the positions where the x-z section were taken; x-z sections are shown below each en face image. Arrowheads denotes localization to BB microvilli, while arrows point to signal accumulation found in subapical and cytoplasmic puncta. Scale bars, 10 μm. CDHR5-L with an intact PBM exhibits enhanced apical targeting compared with CDHR5-L-EGFP or the PBM mutant CDHR5-L_I845R. (F) Scatterplot quantification of the BB:cytosol ratios of CDHR5 signal from cells expressing the CDHR5-L with no tag (PBM intact), CDHR5-L-EGFP (PBM blocked), CDHR5-L_I845R with no tag, and CDHR5-S-EGFP (PBM blocked). Bars indicate mean and SD. R.U = relative units. ns = not significant, *p < 0.01, ***p < 0.0001, two-tailed t test. Measurements: CDHR5-L-EGFP (PBM blocked), n = 80 BB:cyto ratios; CDHR5-L with no tag (PBM intact), n = 43 BB:cyto ratios; CDHR5-L_I845R with no tag, n = 65 BB:cyto ratios; CDHR5-S-EGFP, n = 71 BB:cyto ratios.

The cytoplasmic tail of CDHR5-S is necessary to promote robust microvillar elongation

During the process of working with the two CDHR5 splice isoforms, we observed that overexpression of CDHR5-S resulted in quite a dramatic change in the appearance of the BB microvilli. To explore this further, we measured the length of microvilli from cells overexpressing CDHR5-S and CDHR5-L. For comparison, we also measured the length of BB microvilli of neighboring cells that were not overexpressing these cadherins in our stable cell lines. We observed a significant increase in microvillar length from cells overexpressing CDHR5-S, suggesting that this cadherin may help support microvillar elongation upon reaching the apical surface (Figure 3, A and C). Deletion of the cytoplasmic tail of CDHR5-S partially reduced this effect (Figure 3, B and C). In many cases for cells overexpressing high amounts of CDHR5-S, the apical domain often appeared to herniate outwards with long microvilli splaying from the herniated membrane (Supplemental Figure S3A). This could be easily seen using scanning electron microscopy to observe the general surface features of the cell monolayer from our CDHR5-S stable cell line versus nontransfected control cells (Supplemental Figure S3, B and C). Overexpressing an untagged version of CDHR5-L also promoted the partial lengthening of microvilli, while EGFP-tagged or PBM-deleted variants of CDHR5-L failed to have a similar effect (Supplemental Figure S3, D–G). In combination with the recent discovery that CDHR5 KO mice exhibit shortened microvilli (Modl et al., 2023), our results further implicate CDHR5 in playing a key role in microvillar elongation when delivered to the apical domain.

FIGURE 3:

Overexpression of CDHR5-S promotes elongation of microvilli. (A and B) Confocal images of 4-d polarized LLC-PK1-CL4 cells stably expressing either CDHR5-S-EGFP or CDHR5-S_ΔCD-EGFP (green) and stained for F-actin (red). Boxed regions denote area in zoomed image panels. Dashed lines indicate the position where the x-z sections were taken; Individual channels for the x-z sections are shown below the en face images. Arrows in the x-z section merge channels point to long microvilli, while arrowheads denote nonexpressing cells that have normal, shorter microvilli. Scale bars, 10 μm. Overexpression of CDHR5-S promotes the elongation of BB microvilli, an activity partially dependent upon its cytoplasmic tail. (C) Quantification of microvillar length from cells overexpressing CDHR5-S-EGFP and CDHR5-S_ΔCD-EGFP compared with non-expressing cells. Bars indicate mean ± SD. Measurements: CDHR5-S-EGFP expressing, n = 91 microvilli, CDHR5-S_ΔCD-EGFP expressing, n = 103 microvilli. CDHR5-S nonexpressing, n = 58 microvilli. ***p < 0.0001, two-tailed t test.

CDHR5 interacts biochemically with EBP50/E3KARP

We performed a yeast-two-hybrid screen of the isolated CDHR5 cytoplasmic tail against a human kidney cDNA library in order to identify potential cytoplasmic binding partners for the cadherin. Three prey proteins were recovered as multiple clones from this screen (Table 1), including a fragment of the Ezrin-associated PDZ-based scaffold NHE3 kinase A regulatory protein (E3KARP; also known as NHERF-2 and SLC9A3R2) (Figure 4A). A similar yeast-two-hybrid screen using the cytoplasmic tail of CDHR2 recovered a unique set of binding partners, including the previously identified PDZ-based scaffold USH1C (Crawley et al., 2014b; Li et al., 2016; Yu et al., 2017; Table 2). These independent screens suggest that CDHR5 and CDHR2 likely do not associate with an overlapping cytoplasmic complex, even though both cadherins are found highly enriched at microvillar tips and each end with a C-terminal PBM. Previous work has shown that E3KARP exhibits a restricted tissue distribution, being found mostly in the epithelia of the lung and small amounts in the kidney (Ingraffea et al., 2002). Our discovery of E3KARP as a candidate binding partner prompted us to also explore whether the closely related homolog Ezrin-radixin-moesin (ERM)-binding phosphoprotein 50 (EBP50; also known as NHERF-1 and SLC9A3R1) may also interact with CDHR5. EBP50 is highly expressed in transporting the epithelia of both the kidney and small intestine, and, like E3KARP, is known to act as a scaffold to link Ezrin to a variety of transmembrane proteins found in the BB (Ingraffea et al., 2002; Sauvanet et al., 2015). As a cursory assessment, we first performed a pairwise yeast-two-hybrid test directly between the CDHR5 cytoplasmic tail and EBP50 and observed a positive interaction (Figure 4A). We further confirmed that EBP50 was expressed in the intestine by staining mouse duodenal tissue sections and CACO-2_BBE enterocytes (Figure 4B; Supplemental Figure S4A). In both cases, EBP50 was found highly enriched in BB microvilli. In contrast to the prominent expression of EBP50, very low amounts of E3KARP were detected in duodenal intestinal tissue and in CACO-2_BBE cells (Figure 4C; Supplemental Figure S4B).

TABLE 1:

Yeast-Two-Hybrid results for the screen against the cytoplasmic tail of CDHR5. The following are putative CDHR5 interaction partners that were recovered in multiple clones from our yeast-two-hybrid screen. Shown are their names, domain diagrams, and cellular functions if known.

CDHR5 Interactor	Domain Diagram	Function
SLC7A13		Scaffold
SLC7A13		Amino acid transporter
MAPKAPK5		Protein kinase

FIGURE 4:

The cytoplasmic tail of CDHR5 interacts with the Ezrin-associated scaffolds E3KARP and EBP50. (A) Screening of a yeast-two-hybrid kidney cDNA library identified E3KARP as a binding partner for the cytoplasmic tail of CDHR5. Shown are yeast cotransformed with either the CDHR5 cytoplasmic tail bait and the recovered E3KARP prey construct or empty bait vector and the recovered E3KARP prey construct. Pairwise yeast-two-hybrid interaction tests also show that the cytoplasmic tail of CDHR5 can interact with EBP50. A separate screen of the same kidney cDNA library against the cytoplasmic tail of CDHR2 confirmed the known interaction of CDHR2 with USH1C, another PDZ-based scaffold. Yeast were plated on a medium formulated for plasmid selection alone or a medium formulated for plasmid and interaction selection. (B and C) Confocal images of mouse duodenal tissue stained for EBP50 or E3KARP (green) and Villin (red). Arrows point to the BB. Boxed regions denote area in zoomed image inset panels. Scale bars, 10 μm. EBP50 exhibits robust targeting to BB microvilli. (D) Domain diagram showing the sequence identity between EBP50 and E3KARP. (E) Diagram of EBP50 constructs used to map interaction with CDHR5. (F and G) Mapping the interaction between EBP50 and the cytoplasmic tail of CDHR5. Beads coated with bacterially expressed GST-CDHR5 CD served as bait, while HEK293T cell lysates expressing myc-tagged EBP50 constructs served as pulldown material containing the prey. CD = cytoplasmic domain, FL = Full-length, PDZ12 = fragment composed of PDZ1 and PDZ2, EBD = EBD. The cytoplasmic tail of CDHR5 interacts with the open, active version of EBP50 through PDZ1. (H) Interaction between EBP50 and CDHR5 using purified recombinant protein from bacteria. GST-tagged purified CDHR5 CD proteins served as bait whereas mNEON-tagged EBP50-PDZ12 served as prey. GST alone is included as a negative control.

TABLE 2:

Yeast-Two-Hybrid results for the screen against the cytoplasmic tail of CDHR2. The following are putative CDHR2 interaction partners that were recovered in multiple clones from our yeast-two-hybrid screen. Shown are their names, domain diagrams, and cellular functions if known.

CDHR2 Interactor	Domain Diagram	Function
USH1C (harmonin)		Scaffold
HAX1		Cytoskeleton remodeling
CD24		Cell surface signal transducer

E3KARP and EBP50 have an identical domain structure, having two PDZ domains followed by an Ezrin binding domain (EBD) that terminates with a canonical class I PBM (Figure 4D; Sauvanet et al., 2015). We proceeded to map the binding interactions between CDHR5 and EBP50/E3KARP using protein pulldowns. A GST-fusion protein of the cytoplasmic tail of CDHR5 was coupled to beads and used to perform pulldowns from cell lysates of HEK293T cells transfected with various fragments of EBP50/E3KARP (Figure 4E; Supplemental Figure S4C). These pulldowns revealed that the CDHR5 tail interacts with PDZ1 in EBP50 and PDZ2 in E3KARP (Figure 4F; Supplemental Figure S4D). However, we observed that a concatenated fragment of PDZ1-PDZ2 for both EBP50 and E3KARP displayed a potentiated interaction with CDHR5 (Figure 4F; Supplemental Figure S4D). Interestingly, our pulldowns also demonstrated that CDHR5 only weakly interacts with either of the full-length scaffolds (Figure 4F; Supplemental Figure S4D). We speculated that this may be due to the full-length scaffolds being found in autoinhibited “closed” conformations, in which their C-terminal PBMs bind back to one of their PDZ motifs (Sauvanet et al., 2015). It is currently proposed that EBP50/E3KARP must first interact with activated Ezrin to relieve autoinhibition before they can interact with transmembrane proteins (Finnerty et al., 2004; Terawaki et al., 2006). To test this, we deleted the C-terminal PBMs from the scaffolds to lock them in an open, active state (Figure 4E; Supplemental Figure S4C). Consistent with this idea, we observed that the cytoplasmic tail of CDHR5 exhibited a robust interaction only with the open, active versions of both scaffolds (Figure 4G; Supplemental Figure S4E). Finally, to validate that our observed interaction between EBP50 and CDHR5 is direct, we purified a PDZ1-PDZ2 fragment of EBP50 as a His-mNeon-tagged protein from bacteria to test with our GST-CDHR5 cytoplasmic tail. We also created a variant of the CDHR5 cytoplasmic tail in which we mutated the C-terminal PBM (I845R). As expected, the PDZ1-PDZ2 fragment of EBP50 could only bind to wild-type CDHR5 cytoplasmic tail and not the CDHR5 variant lacking a canonical PBM (Figure 4H). We conclude that CDHR5 uses its C-terminal PBM to interact directly with EBP50 and that the scaffold must be in the open conformation for robust binding to occur.

CDHR5 interacts with EBP50/E3KARP in cells

To explore whether CDHR5 and EBP50 interact in a cellular context, we stained for both proteins in polarized CACO-2_BBE cells and used proximity ligation assay (PLA) analysis to assess a putative interaction. We observed that both CDHR5 and EBP50 overlapped extensively in the apical BB (Figure 5A), and resulted in formation of PLA puncta not seen in control PLA conditions (Figure 5, B–D). We noted that in mature, longer microvilli CDHR5 appeared to be more restricted to the distal tips while EBP50 was found all along the microvillar axis (Figure 5E; see arrows in zoom panel). This was consistent with dual staining for CDHR5 and EBP50 in mouse intestinal tissue (Supplemental Figure S5). To further validate that EBP50/E3KARP can interact with CDHR5 in a cellular context, we took advantage of a “nanotrap” pulldown approach that allows protein–protein interactions to be rapidly interrogated in live cells. This strategy involves fusing a bait protein to a truncated version of the myosin, Myo10, which uses its robust motor activity to constitutively target to the tips of filopodia in cells. Interaction of a bait with a prey protein can then be easily assessed by visualizing whether the prey protein has been forced to the tips of filopodia by interacting with the bait (Bird et al., 2017). We fused various EBP50 and E3KARP constructs to mCherry to utilize as prey, while the cytoplasmic tail of CDHR5 was fused to the truncated EGFP-Myosin-10. We transfected EGFP-Myo10-CDHR5-cytoplasmic tail along with mCherry-EBP50/E3KARP constructs in HeLa cells and plotted the correlation between bait and prey fluorescence at the tips of filopodia (Figure 6, A–F; Supplemental Figure S6, A–E). We observed the strongest colocalization between bait and prey when using the PDZ1–PDZ2 constructs of EBP50/E3KARP (Figure 6, D and F; Supplemental Figure S6, D and E). Consistent with our conventional protein pulldown data, the observed interactions were dependent upon having EBP50/E3KARP scaffolds in the open conformation, as well as having an intact C-terminal PBM for CDHR5. In sum, our results show that the Ezrin-associated scaffolds EBP50 and E3KARP can interact with CDHR5 both biochemically and in cells.

FIGURE 5:

EBP50 and CDHR5 are both highly enriched in enterocyte BBs. (A) Confocal image of 12-d polarized CACO-2_BBE cells stained for endogenous EBP50 (red) and CDHR5 (green). Dashed line indicates the position where the x-z section was taken; x-z section channels are shown to the right of the en face image. Scale bars, 10 μm. EBP50 and CDHR5 colocalize to BB microvilli. (B–D) Confocal images of 12-d polarized CACO-2_BBE cells visualized for PLA interaction using antibodies for CDHR5 (rabbit; antibody HPA009081) and EBP50 (mouse; AFFN-SLC9A3R1-9B6). Control experiments were performed using each primary antibody alone. Scale bars, 10 μm. (E) Zoom panels of X-Z section from 12-d polarized CACO-2_BBE cells stained for endogenous EBP50 (red) and CDHR5 (green). In longer, more mature microvilli CDHR5 is restricted to the distal tips as shown by the arrows in zoom panel 1, while EBP50 is found more uniform along the axis of microvilli. Zoom panel 2 shows shorter microvilli with more uniform staining of CDHR5 along the axis of microvilli.

FIGURE 6:

EBP50 can interact with the cytoplasmic tail of CDHR5 in a cellular context. (A–E) Confocal images of nanotrap pulldown experiments between the cytoplasmic tail of CDHR5 and EBP50 constructs. See text for nanotrap pulldown experimental details. GFP-tagged Myo10- -CDHR5 CD constructs served as bait, mCherry-tagged EBP50 constructs served as prey. Boxed regions in merge panels denote area in zoomed image panels. Scale bars,10 μm. EBP50 can interact with the cytoplasmic tail of CDHR5 in a cellular context. Arrows in (D) show examples of extensive colocalization of bait and prey at filopodia tips. (F) Quantification of the correlation between bait (x-axis) and prey (y-axis) fluorescence at individual filopodia tips. Lines represent best linear fit. CD = cytoplasmic domain, FL = Full-length, PDZ12 = fragment composed of PDZ1 and PDZ2.

CDHR5 forms a complex with Ezrin via EBP50/E3KARP in a regulated manner

The primary function of EBP50 in microvilli is to connect transmembrane proteins to Ezrin, a member of the ERM family (Sauvanet et al., 2015). Ezrin is known to play an essential role in the formation of microvilli by linking the plasma membrane to the underlying actin cytoskeleton (Algrain et al., 1993; Turunen et al., 1994; Fehon et al., 2010). Ezrin is composed of an N-terminal 4.1 protein, Ezrin, Radixin and Moesin (FERM) domain joined to a C-terminal ERM association domain (C-ERMAD) through a long α-helical coiled-coil region. In the inactive (closed) form of Ezrin, the N-terminal FERM domain binds with high affinity to the C-terminal C-ERMAD. Binding of PI(4,5)P2 to the FERM domain allows phosphorylation of a specific threonine residue at position 567 (T567) in the C-ERMAD to occur, resulting in the open active conformation of Ezrin (Pelaseyed et al., 2017; Pelaseyed and Bretscher, 2018). Active Ezrin can then bind to and activate EBP50/E3KARP using its FERM domain (Reczek et al., 1997; Reczek and Bretscher, 1998), while its C-terminal C-ERMAD sequence binds directly to F-actin. After mapping the CDHR5 binding site of EBP50/E3KARP, we explored whether the interaction of active Ezrin with these scaffolds would promote their open conformations to allow a ternary complex to form between CDHR5-EBP50/E3KARP-Ezrin. To test this, we generated full-length Ezrin and an Ezrin T567D phosphomimetic mutation construct, as well as fragments of the isolated FERM domain and C-ERMAD (Figure 7A; Supplemental Figure S7A). We coexpressed these Ezrin constructs along with full-length EPB50/E3KARP in HEK293T cells, and used cell lysates derived from these transfections for protein pulldowns with beads coated with GST-CDHR5 cytoplasmic tail as bait. Both the full-length and the T567D phosphomimetic Ezrin mutant failed to initiate ternary complex formation, suggesting that the full-length inactive version of Ezrin cannot mediate this interaction and that the phosphomimetic mutant alone does not bypass the need for PI(4,5)P2 for full activation of Ezrin (Figure 7B; Supplemental Figure S7B). In contrast, the isolated FERM domain of Ezrin appeared to activate EBP50/E3KARP to allow the ternary complex to form (Figure 7B; Supplemental Figure S7B). Importantly, the cytoplasmic domain of CDHR5 could not directly bind to Ezrin, demonstrating that the complex formed was indeed a ternary interaction in which Ezrin first must activate EBP50/E3KARP, which then associates with the cytoplasmic tail of CDHR5 (Figure 7C). Together, these data show that the available FERM domain of activated Ezrin can promote the interaction of CDHR5 with the Ezrin-associated scaffolds EBP50/E3KARP (Figure 7D).

FIGURE 7:

Ternary complex formation between CDHR5, EBP50, and Ezrin. (A) Cartoon schematic of the different Ezrin constructs used in pulldown assays with EBP50 and the cytoplasmic tail of CDHR5. (B) Ternary complex formation between CDHR5, EBP50, and Ezrin. Beads coated with bacterially expressed GST-CDHR5 CD served as bait, while HEK293T cell lysates expressing myc-tagged EBP50 and various EGFP-tagged Ezrin constructs served as pulldown material containing the prey. A fragment of Ezrin can activate EBP50 to allow it to interact with the cytoplasmic tail of CDHR5. (C) Pulldown analysis testing for potential interaction between CDHR5-CD and Ezrin constructs. (D) Cartoon schematic of the ternary complex formed between CDHR5, EBP50, and Ezrin.

EBP50 is required for proper apical polarization and IMAC assembly

Given the prominent role that EBP50 plays in linking BB-resident transmembrane proteins to the actin cytoskeleton via Ezrin, we were interested to explore what the impact of loss of EBP50 would be on IMAC localization and BB assembly. We screened a panel of six independent shRNA knockdown (KD) constructs targeting EBP50 in our CACO- 2BBE cells and identified two that had a significant KD as judged by both immunostaining and immunoblot analysis (Supplemental Figure S8, A and B). Consistent with previous findings in other epithelial cell systems (Morales et al., 2004; LaLonde et al., 2010), loss of EBP50 had a severe effect on the proper development of apical microvilli in our polarized CACO-2_BBE cells (Figure 8A). Apical microvilli appeared small and rudimentary in these EBP50 KD cells, and did not cluster together indicating defective IMAC activity (Figure 8, A and B). Indeed, staining for various components of the IMAC (CDHR5, CDHR2, and USH1C) in the EBP50 KD cells revealed significant defects in their proper targeting as compared with cells transduced with a scramble control shRNA (Figure 8, C and D; Supplemental Figure S8C). We quantified this by measuring the ratio of signal of the IMAC components found in the apical domain compared with the total cell content. In each case, depletion of EBP50 resulted in loss of CDHR5, CDHR2, and USH1C out of BB microvilli (Figure 8E; Supplemental Figure S8D). Immunoblot analysis of total cell lysates from these cell lines further demonstrated that the overall levels of these IMAC components were markedly reduced in the absence of EBP50 expression (Supplemental Figure S8E). We performed the complimentary experiment of knocking down CDHR5 in CACO-2_BBE cells to understand whether EBP50 was dependent upon this IMAC component for its proper apical targeting. Similar to loss of EBP50, KD of CDHR5 resulted in significant defects in BB assembly (Supplemental Figure S9, A and B), with cells having only small microvilli that did not exhibit the typical clustering phenotype as seen in the scramble control cell line. In contrast however, loss of CDHR5 only had a minor impact on the efficiency of EBP50 targeting to the BB (Figure 9, A and B; Supplemental Figure S9C). We extended this analysis to look at the BB targeting efficiency of total and activated Ezrin (P-ERM; see figure legend for details) in CDHR5 KD cells. In both cases, the targeting efficiency of total and activated Ezrin did not appear to be impacted with loss of CDHR5 (Figure 9, A and B; Supplemental Figure S9C). Together, these studies show that EBP50 acts as an upstream factor to promote proper targeting and function of the IMAC and allows us to propose a new interactome model of the enterocyte IMAC in which CDHR5 is associated with active Ezrin through the scaffold EBP50 (Figure 9C).

FIGURE 8:

Loss of EBP50 results in severe BB assembly defects in CACO-2_BBE cells. (A) Confocal images of 12-d polarized CACO-2_BBE cells stably expressing either a scramble shRNA construct or shRNAs targeting EBP50 (KD 37 and 64), stained for EBP50 (green) and F-actin (red). Loss of EBP50 expression disrupts proper apical assembly. (B) Quantification of microvillar clustering in scramble and EBP50 KD CACO-2_BBE cell lines. Measurements: scramble control, n = 742 cells; EBP50 KD 37, n = 556 cells; EBP50 KD 64, n = 448 cells. Bars indicate mean ± SD. ***p < 0.0001, two-tailed t test. (C) Confocal images of 12-d polarized CACO-2_BBE cells stably expressing either a scramble shRNA construct or an shRNA targeting EBP50 (KD 64), stained for F-actin (red) and various IMAC components (CDHR5, CDHR2, or USH1C; green). Boxed regions denote the area in zoomed image panels. Scale bars, 10 μm. (D) Example x-z sections taken from confocal images of 12-d polarized scramble shRNA control or shRNA EBP50 KD 64 stable CACO-2_BBE cell lines. Monolayers were stained for F-actin (red) and CDHR5, CDHR2, or USH1C (green) as labeled. (E) Scatterplot quantification of apical/total signal ratios of the IMAC components in scramble and EBP50 KD 64 CACO-2_BBE cells. Each data point represents the ratio of the entire apical signal found in the x-z section compared with the total signal of the x-z section. Measurements of apical to total signal: CDHR5 signal, scramble n = 28, KD 64 n = 31; CDHR2 signal, scramble n = 27, KD 64 n = 44; USH1C signal, scramble n = 33, KD 64 n = 38. ***p < 0.0001, two-tailed t test. Loss of EBP50 causes defects in IMAC targeting.

FIGURE 9:

Loss of CDHR5 results in BB assembly defects in CACO-2_BBE cells. (A) Confocal images of 12-d polarized CACO-2_BBE cells stably expressing either a scramble shRNA construct or an shRNA targeting CDHR5, stained for CDHR5, EBP50, Ezrin, and Phospho-ERM (P-ERM; green) and F-actin (red). Boxed regions denote the area in zoomed image panels. Scale bars, 10 μm. Loss of CDHR5 expression disrupts proper apical assembly, but does not block apical targeting of Ezrin, activated Ezrin (detected by P-ERM antibody) and EBP50. (B) Scatterplot quantification of apical/total signal ratios of Ezrin, activated Ezrin (P-ERM) and EBP50 in scramble and CDHR5 KD CACO-2_BBE cells. Each data point represents the ratio of the entire apical signal found in the x-z section compared with the total signal of the x-z section. Measurements of apical to total signal: EBP50 signal, scramble n = 78, KD n = 65; Ezrin signal, scramble n = 74, KD n = 56; P-ERM signal, scramble n = 62, KD n = 69. ns = not significant, *p < 0.01, two-tailed t test. (C) Cartoon proposing the new interactome of the IMAC, in which CDHR5 is cross-linked to the actin cytoskeleton via EBP50-Ezrin.

DISCUSSION

Proper formation of a BB on the surface of transporting epithelial cells of the gut and kidney creates an impressive apical membrane reservoir that mediates the solute transport function of the tissue. Within the intestine, the BB membrane is specifically enriched in nutrient processing enzymes and transport channels, along with host defense factors (Crawley et al., 2014a). Defects in the intestinal BB can result in both malabsorption and forms of inflammatory bowel disease due to loss of proper nutrient transport and barrier function. A recent study demonstrated that CDHR5 KO mice exhibit pronounced BB defects in the gut, with enterocytes assembling a lower density of microvilli on their apical surface that are shorter in length (Modl et al., 2023). These defects correlated with an overall lower level of key BB-resident proteins that are involved in nutrient processing and host defense including Dipeptidyl peptidase IV, Intestinal alkaline phosphatase, and Na+/H+ Exchanger 3. In turn, CDHR5 deficient mice were more sensitive to developing DSS-induced colitis compared with wild-type littermates (Modl et al., 2023). These findings are consistent with the observation that patients with Crohn’s disease and ulcerative colitis have a downregulation of CDHR5 in their intestinal mucosa, with enterocytes also exhibiting shorter microvilli (VanDussen et al., 2018; Modl et al., 2023). While our study provides evidence that the two splice isoforms of CDHR5 behave differently in terms of their apical targeting requirements, it is still unclear whether they exhibit different biological roles in the gut. It was previously demonstrated that both isoforms possess an identical ability to interact with CDHR2, suggesting that direct differences in intermicrovillar adhesion are not likely the underlying biological purpose of expressing two distinct CDHR5 isoforms (Crawley et al., 2014b). Having CDHR5 expressed as two isoforms with different ectodomain lengths would generate intermicrovillar adhesion links of slightly different sizes. However, whether this contributes to plasticity of intermicrovillar adhesion during BB assembly is currently unknown (Meenderink et al., 2019; Cencer et al., 2023). Given that the CDHR5 MLR domain is subject to heavy O-glycosylation (Goldberg et al., 2000), we speculate that the long isoform of this cadherin may play an important role in helping form the dense glycocalyx barrier normally associated with the BB membrane. This may explain the specific positioning of the MLR domain just outside the exoplasmic face of the plasma membrane, where it could serve as a final barrier against microbe attachment to the plasma membrane. Furthermore, having the two CDHR5 splice isoforms exhibit different sequence-dependent targeting properties may represent a mechanism to temporally control their specific delivery to the apical domain. EBP50 expression is known to be upregulated in response to inflammatory stimuli (Leslie et al., 2013), which could lead to more apical delivery of CDHR5-L to potentially reinforce the epithelial barrier during infection. The underlying basis of how the MLR domain promotes retention of CDHR5-L in the cytoplasm, and how EBP50 interaction circumvents this retention will be important questions to address going forward. Creating a CDHR5-L specific KO mouse line, as well as generating an isoform-specific antibody against the long isoform, will help discern the distinct roles of these two CDHR5 isoforms in transporting epithelia.

One of the most striking effects we observed in our study was that overexpression of CDHR5 resulted in lengthening of BB microvilli. This lengthening effect was partially dependent upon an intact cytoplasmic tail of CDHR5, though did not appear to require its interaction with EBP50. This suggests that the formation of microvilli may be governed by a balance between activities that promote lengthening of microvilli and those that support the stabilization of the length. Similar to our observation with CDHR5, modulation of the Drosophila melanogaster cadherin Cad99C was previously observed to control microvillar length in the fly system (D’Alterio et al., 2005; Schlichting et al., 2006). Cad99C is found highly enriched in the microvilli of follicle cells that are in contact with the Drosophila oocytes, and flies depleted of Cad99C had short microvilli on their follicle cells, whereas Cad99C overexpression promoted pronounced elongation of these microvilli (D’Alterio et al., 2005). However, this lengthening effect was not dependent upon the cytoplasmic tail of Cad99C. In combination with our study here and the discovery that loss of either CDHR2 or CDHR5 results in short microvilli (Pinette et al., 2019; Modl et al., 2023), these findings highlight a potential conserved function of protrusion-associated cadherins in promoting microvillar elongation. The molecular basis of this observation may relate to the recent discovery that mucin biopolymers and long-chain polysaccharides found within the cell glycocalyx can generate entropic forces that promote the formation of de novo finger-like membrane extensions from the cell surface (Shurer et al., 2019). Shuer et al. (2019) discovered that ectopic overexpression of transmembrane mucin proteins induced the extensive formation of finger-like extensions from the cell surface, similar in nature to microvilli. This activity correlated directly to the number of mucin repeats found in the transmembrane mucin proteins, as well as their glycosylation level. Whether the levels of glycosylation of the ectodomains of Cad99C and CDHR5 correlates with their ability to promote microvillar elongation will be an interesting future question to address.

MATERIALS AND METHODS

Request a protocol through Bio-protocol.

Molecular Biology

The human cDNA constructs of CDHR2 (NM_001171976.2;UniProtKB-Q9BYE9), CDHR5-L (NM_021924.5;UniProtKB-Q9HBB8-1), CDHR5-S (NM_031264.5;UniProtKB-Q9HBB8-2), EBP50 (NM_004252.5;UniProtKB-O14745-1), E3KARP (NM_001130012.3;UniProtKB-Q15599-1), and Ezrin (NM_003379.5;UniProtKB- P15311) were used in this study. All fragment and truncation constructs including CDHR5-CD (aa 697–845), CDHR5-S_ΔCD (aa 1–501), CDHR5-L_ΔCD (aa 1–695), CDHR2-CD (aa 1178-13110), EBP50-PDZ1 (aa 1–148), EBP50-PDZ2 (aa 149–298), EBP50-PDZ12 (aa 1–298), EBP50-PDZ2 EBD (aa 149–358), EBP50-FL ΔPBM (aa 1–353), E3KARP-PDZ1 (aa 1–149), E3KARP-PDZ2 (aa 150–284), E3KARP -PDZ12 (aa 1–284), E3KARP-PDZ2 EBD (aa 150–337), E3KARP-FL ΔPBM (aa 1–329), Ezrin-WT (aa 1–586), Ezrin-NERMAD (aa 1–297), Ezrin-CERMAD (aa 461–586), were generated by PCR and TOPO cloned into the pCR8 Gateway Entry vector (Invitrogen). The Quikchange site-directed mutagenesis kit (Aligent) was used to generate CDHR5-L_I845R, CDHR5-CD I845R, and Ezrin-T567D mutants within their entry vectors. All entry vectors clones were verified by DNA sequencing. Entry clones for targeting and interaction assessment in mammalian cell culture lines were shuttled into the appropriate destination vectors including: pEGFP-N3 (Clontech), pmCherry-C1 (Clontech), and pLVX-N3-EGFP (Clontech) that had been Gateway-adapted using the Gateway Vector Conversion kit (Invitrogen). Vectors used for expression of recombinant protein for pulldown assays including His-mNEON tagged, FLAG tagged, myc tagged, and FC tagged, and EGFP-tagged proteins in this study have been previously described (Crawley et al., 2014b, 2016; Graves et al., 2020). KD studies utilized a nontargeting scramble control shRNA (Addgene; plasmid #46896), CDHR5 KD shRNA (TRC clone TRCN000054168; KD68), and EBP50 KD shRNA clones (TRCN0000043733; KD33), (TRCN0000043734; KD34), (TRCN0000043735; KD35), (TRCN0000043737; KD37), (TRCN0000440444; KD44), and (TRCN0000437164; KD64) sequ­ences that were expressed from the pLKO.1 plasmid. The pcDNA3.1-MYO10-HMM-Nanotrap system was a gift from Thomas Friedman (Addgene plasmid # 87255; http://n2t.net/addgene: 87255; RRID: Addgene_87255). Flagged-tagged Human EBP50-FL (aa 1-358), EBP50-PDZ1(aa 1-148), EBP50-PDZ2 (aa 148-298), and EBP50-PDZ12 (aa 1-298) cloned into the pCMV-2-FLAG vector were gifts from Maria-Magdalena Georgescu (Addgene plasmids #28291-28297).

Yeast-two-hybrid screening

Yeast-two-hybrid library screening was performed using the MATCHMAKER system (Takara) with a human kidney yeast-two-hybrid cDNA library. Library-adapted cDNAs were expressed from the pACT2 vector in the MATα Y187 yeast strain, while the CDHR5-CD and CDHR2-CD bait construct was expressed from the pGBKT7 vector in the MATa AH109 strain. Library screening, including the use of controls, was performed using standard methods outlined by the MATCHMAKER system.

Cell culture and stable cell line generation

CACO-2_BBE (American Type Culture Collection [ATCC] CRL-2102), HeLa (ATCC CCL-2), HEK293T (ATCC CRL-3216), HEK293FT (Invitrogen R70007) and LLC-PK1-CL4 (gift from Matthew Tyska) cells were grown in a 5% CO₂ humidified incubator at 37°C in DMEM supplemented with high glucose and 2 mM L-glutamine. CACO-2_BBE cells were cultured in DMEM with 20% fetal bovine serum (FBS) and LLC-PK1-CL4, HeLa, HEK293T, and HEK293FT cells in DMEM with 10% FBS. All cell lines were authenticated using Short Tandem Repeat (STR) profiling to detect potential cell misidentification, cross-contamination or genetic drift. Regular monthly Mycoplasma testing of all cell culture lines were performed using the LookOut Mycoplasma PCR Detection Kit (Sigma MP0035). Experiments using all cell lines were done with low passage number cultures (LLC-PK1-CL4 cells ∼15 passages, HEK293TFT/HeLa cells ∼15 passages and CACO-2_BBE ∼10 passages) to help prevent possible genetic drift effects on data. HEK293FT cells (10-cm dish at 80% confluency) were used for lentivirus production by cotransfecting 6 μg of Lentiviral expression plasmid with 4 μg psPAX2 packaging plasmid and 0.8 μg pMD2.G envelope plasmid using the polyethylenimine transfection reagent (Polysciences). Cells with transfection medium were exchanged with fresh medium 12 h after transfection. Cells were returned to the 37°C incubator with 5% CO₂ to allow for lentiviral production into the medium. Two days later, supernatant containing secreted lentiviral particles was collected, filtered with a 0.45 μm syringe filter and concentrated using Lenti-X Concentrator (Clontech) and stored at –80°C for further use. To transduce LLC-PK1-CL4 and CACO-2_BBE cells with lentivirus, cells were seeded and grown to 80% confluency in a T25 flask, after which the medium was supplemented with 8 μg/ml polybrene and ∼300 μl of concentrated lentivirus was added. Cells were then incubated overnight with this transduction medium. For plasmid transfections, LLC-PK1-CL4 cells were first seeded and grown to 80% confluency in a T25 flask. Plasmids were mixed with Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions to create a transfection mixture. This transfection mixture was added to the culture medium of cells and allowed to incubate for 3 h. Fresh medium was then swapped in and cells were allowed to recover for 24 h. To select for stable cell lines after viral transduction or plasmid transfection, T25 flasks containing the transduced/transfected cells were reseeded to 10-cm dishes and grown for 3 d in the absence of antibiotics. Next, cells were reseeded into T182 flasks with medium containing 50 μg/ml puromycin for viral transductions or 1 mg/ml G418 for plasmid transfections. Cells were then continuously grown for numerous (∼10) passages to select for stable integration of DNA.

Recombinant protein purification from E. coli

For pulldown assays using the GST fusion of the CD of CDHR5, the pGEX-4T3-CDHR5 CD construct was transformed into BL21(DE3) bacteria (Thermo Fisher Scientific), expressed, and purified using GSH resin using standard conditions. For recombinant production of EBP50, the pNCS-NEON-EBP50 PDZ1PDZ2 construct was expressed in BL21(DE3) bacteria. Bacteria were grown to a O.D = 1 at 37°C, after which the temperature was lowered to 16°C and expression induced for 8 h using ITPG (Invitrogen). Cells were then collected by centrifugation at 4000 × g for 20 min and purified using Ni–NTA–agarose (Qiagen) using standard conditions. Briefly, cells were lysed using cold lysis buffer (50 mMNaPO4, pH 8.0, 300 mM NaCl, 10 mM imidazole, 1 mg/ml lysozyme, 1 mM PMSF) along with sonication. The cell lysate was then centrifuged at 15,000 × g for 1 h. The soluble fraction of the lysate was incubated with Ni-NTA agarose resin (Qiagen) on a rocking platform for 1 h at 4°C. The lysate–Ni–NTA resin mixture was applied to a column, washed with Wash Buffer (50 mM NaPO4, pH 8.0, 300 mM NaCl, 25 mM imidazole) and the His fusion protein eluted with Elution Buffer (50 mM NaPO4, pH 8.0, 300 mM NaCl, 250 mM imidazole). Eluates were analyzed by SDS–PAGE and those fractions containing His-NEON-EBP50 PDZ1PDZ2 were pooled, dialyzed, and utilized for protein pulldown assays.

Biochemical pulldown and nanotrap live-cell pulldown assays

For biochemical pulldowns, HEK293T cells were grown in T75 flasks to 90% confluency for transfection with pulldown constructs (40 μg total DNA) using 100 μg of Polyethylenimine Linear, MW 25,000 (PEI; Polysciences). Cells were incubated with transfection media for 16 h, after which transfection media was removed and replaced with fresh media. After 48 h, cells were washed once in phosphate-buffered saline (PBS), recovered by a cell scraper, and frozen for storage at –80°C. For pulldowns, cells were thawed on ice, resuspended in ice-cold CelLytic M buffer (Sigma) containing 2 mM ATP, 1X cOmplete ULTRA protease inhibitor cocktail (Roche), and 1 mM Pefabloc SC (Roche), and lysed by several rounds of needle aspiration (27G X 11/4 needle). Cell lysates were centrifuged at 16,000 X g and the soluble material recovered and incubated with a 50 μl bed volume of either preequilibrated anti-FLAG M2 resin (Sigma) or glutathione resin (Sigma) with GST fused to the CDs of the IMAC protocadherins. Resins were incubated with cell lysates for 2–3 h rocking at 4°C, pelleted by a low-speed spin, washed four times using 0.5X RIPA buffer supplemented with 2 mM ATP, 1X cOmplete ULTRA protease inhibitor cocktail (Roche), and 1 mM Pefabloc SC (Roche), and eluted to recover bound material by boiling in 2X sodium dodecyl sulfate buffer supplemented with 20 mM reduced glutathione for GST pulldowns or eluted using wash buffer containing 200 μg/ml FLAG peptide for FLAG pulldowns. Resin-bound material was detected by either staining with Coomassie blue or western analysis with the following antibody dilutions: mouse anti-FLAG M2 (1:1000; Sigma catalogue#F3165), mouse anti-V5 (1:5000; Invitrogen catalogue#R960-25), or mouse anti-myc clone 9E10 (1:1000; Sigma catalogue#M4439). For nanotrap live-cell pulldowns, HeLa cells were seeded onto coverslips at 50% confluency for transfection with nanoscale pulldown constructs (3.5 μg pEGFP-C1-Myo10-bait fusion constructs and 3.5 μg pmCherry-C1 prey construct DNA) using 100 μg of Polyethylenimine Linear, MW 25,000 (PEI; Polysciences). Cells were incubated with transfection media for 16 h, after which transfection media was removed and replaced with fresh media. After 24 h, cells were washed once in warm PBS, and processed for microscopy (see Confocal Microcopy section).

Confocal Microscopy

CACO-2_BBE, LLC-PK1-CL4, HeLa cells seeded on coverslips were fixed in 4% paraformaldehyde (Electron Microscopy Sciences) in PBS for 15 min at RT, washed with PBS, and permeabilized with 0.1% Triton X-100 in PBS for 7 min. After fixation, cells were washed four times with PBS and blocked overnight with 5% bovine serum albumin at 4°C. Cells were stained for 1 h at room temperature using primary antibody for anti-CDHR5 (1:200; Sigma catalogue no. HPA009081; second independent CDHR5 antibody used 1:500, Sigma catalogue no. HPA009173), anti-CDHR2 (1:75; Sigma catalogue no. HPA012569), anti-USH1C (1:70; Sigma catalogue no. HPA027398), anti-E3KARP (1:200; Sigma catalogue no. HPA001672), anti-EBP50 (1:200; Sigma catalogue no. HPA027247), anti-EBP50 (1:150; Developmental Studies Hybridoma Bank catalogue no. AFFN-SLC9A3R1-9B6- used for PLA analysis only), anti-Ezrin (1:200; Cell Signaling catalogue no. 3145), anti-P-ERM (1:200; Cell Signaling catalogue no. 3726) or anti-GFP (1:200; Aves Labs catalogue no. GFP1020). Cells were then washed three times with PBS and incubated with Alexa Fluor 488 goat antirabbit or Alexa Fluor 488 goat antichicken (where appropriate) and Alexa Fluor 568 phalloidin diluted 1:200 in PBS for 1 h at RT. Cells were washed four times with PBS and coverslips were mounted using ProLong Diamond Anti-fade reagent (Invitrogen). Paraffin-embedded intestinal tissue sections from mice were prepared as previously described (Crawley et al., 2014b). Cells and tissue sections were imaged using a Leica TCS SP8 laser-scanning confocal microscope equipped with HyVolution deconvolution software.

Analysis of fluorescence colocalization for nanotrap live-cell pulldown assays

Image analysis was performed as previously described (Bird et al., 2017). Briefly, cells that had been transfected with both bait (EGFP) and prey (mCherry) constructs were fixed and stained for F-actin (phalloidin) and utilized to determine the correlation of fluorescence intensity of EGFP/mCherry signal at filopodial tips. Line scans along individual filopodia were performed in ImageJ using the bait and/or phalloidin (F-actin) fluorescence signal as a guide to visual filopodia. Filopodia-tip localized fluorescence intensities of EGFP/mCherry signal were collected and plotted against each other and were fir with liner regression curves.

BB:cytosol quantification

X–Z cross section images were used to quantify CDHR5 construct targeting to microvilli. For each construct, ∼12 cross section images were analyzed derived from ∼5 independent image Z-stacks. Z-stacks were acquired across three independent preparations of stable cell lines for each construct. BB:cytosol ratio values were generated as follows: for a given X–Z section, microvilli were first visualized using the F-actin (phalloidin) channel. The intensity of EGFP signal found in BB microvilli was measured at three to five points in cells that were expressing the CDHR5 construct. Another three to five intensity points were acquired from a region that corresponded to subapical cytosol. Intensity values from microvillar and cytosolic regions were averaged separately, and the resulting means were used to obtain BB:cytosol intensity ratios.

Proximity Ligation Assay (PLA)

PLAs were performed in 12-d polarized CACO-2_BBE cells using antibodies for CDHR5 (1:200 rabbit; antibody HPA009081) and EBP50 (1:150 mouse; AFFN-SLC9A3R1-9B6) according to the manufacturers protocol. Cells were visualized for PLA interaction (red) and DNA (blue). Control experiments utilized each antibody alone, as recommended by the manufacturers protocol.

Electron microscopy

LLC-PK1-CL4 cells were seeded into 0.4-μm collagen coated 12-mm Transwell-COL inserts (Corning) and allowed to polarize for 4 d. Samples were washed once with warm SEM buffer (100 mM sucrose and 100 mM Na-phosphate buffer, pH 7.4) and fixed overnight at 4°C with 3% glutaraldehyde in SEM buffer. Samples were washed with SEM buffer followed by incubation with 1% OsO4 in SEM buffer on ice for 1 h, and subsequently washed with SEM buffer. Samples were dehydrated in a graded ethanol series, dried, mounted on aluminum stubs, and coated with gold/platinum using a sputter coater. Imaging was performed using a JEOL JSM-7500F field emission scanning electron microscope operated in high vacuum mode with an accelerating voltage of 0.5–2 kV. All SEM reagents were purchased from Electron Microscopy Sciences.

Statistical analysis

All graphs were generated and statistical analyses performed using Prism version 6 (GraphPad). For all figures, error bars represent SD Unpaired t tests were employed to determine statistical significance between reported values. Statistical details of individual experiments can be found in figure legends (***, p < 0.0001; **, p < 0.001; *, p < 0.01).

Distribution of Material

All material from this study will be freely distributed upon request.

Abbreviations used:

BB

brush border

GST

glutathione S-transferase

IMAC

intermicrovillar adhesion complex

KD

knockdown

MLR

mucin-like repeat

PBM

PDZ binding motif

PBS

phosphate-buffered saline