Decision letter: A CRISPR screen in intestinal epithelial cells identifies novel factors for polarity and apical transport

Abstract

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Epithelial polarization and polarized cargo transport are highly coordinated and interdependent processes. In our search for novel regulators of epithelial polarization and protein secretion, we used a genome-wide CRISPR/Cas9 screen and combined it with an assay based on fluorescence-activated cell sorting (FACS) to measure the secretion of the apical brush-border hydrolase dipeptidyl peptidase 4 (DPP4). In this way, we performed the first CRISPR screen to date in human polarized epithelial cells. Using high-resolution microscopy, we detected polarization defects and mislocalization of DPP4 to late endosomes/lysosomes after knockout of TM9SF4, anoctamin 8, and ARHGAP33, confirming the identification of novel factors for epithelial polarization and apical cargo secretion. Thus, we provide a powerful tool suitable for studying polarization and cargo secretion in epithelial cells. In addition, we provide a dataset that serves as a resource for the study of novel mechanisms for epithelial polarization and polarized transport and facilitates the investigation of novel congenital diseases associated with these processes. Editor's evaluation In this work, Klee et al. carried out a genome-wide CRISPR/Cas9-based screen in human intestinal cell line CaCo2 to uncover factors regulating apical localization of a brush border enzyme. Their findings identified dozens of genes and characterized novel players in apical membrane transport including TM9SF4, anocatmin 8, and ARGAP33. This work provides a useful resource for the study of apical polarity and may aid in the understanding of digestive diseases. https://doi.org/10.7554/eLife.80135.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Epithelia are highly specialized tissues that line inner and outer surfaces of various organs of metazoans, performing absorption, secretion, and barrier functions. During polarization, epithelial cells assume their characteristic shape by building specialized apical- and basolateral plasma membrane (PM) domains (Rodriguez-Boulan and Macara, 2014; Apodaca et al., 2012), which are separated by junctional complexes and characterized by a specific composition of lipids and proteins (Martin-Belmonte et al., 2007). The asymmetric distribution of polarity complexes and the mutual exclusion of proteins from one domain by proteins from the other domain are critical for the maintenance of apico- basolateral domains at the cell cortex (Rodriguez-Boulan and Macara, 2014; Román-Fernández and Bryant, 2016). Additionally, tightly orchestrated transport mechanisms and machineries, as Rab-GTPases, motor proteins, soluble NSF attachment receptor (SNARE)-proteins, and specific adapter proteins, ensure the establishment and maintenance of specialized membrane domains (Gaisano et al., 1996; Low et al., 1996; Weimbs et al., 1997; Li et al., 2002). Defects in polarization and polarized traffic often cause diseases, such as congenital diarrhea and enteropathies (Thiagarajah et al., 2018; Berni Canani et al., 2010; Apodaca et al., 2012). Microvillus inclusion disease (MVID) is an autosomal-recessive enteropathy (Cutz et al., 1989), characterized by intractable diarrhea in neonates (Cutz et al., 1989; Ruemmele et al., 2006). Enterocytes of MVID patients show loss of brush-border microvilli, formation of so-called microvillus inclusions and subapical accumulation of so-called 'secretory granules' (Cutz et al., 1989; Phillips et al., 2000). Our studies identified mutations in MYO5B, STX3, and STXBP2 to be causative for MVID (Müller et al., 2008; Ruemmele et al., 2010; Wiegerinck et al., 2014; Vogel et al., 2017b); they revealed that a molecular transport machinery involving myosin Vb (myo5b), the small Rab-GTPases Rab11a and Rab8a, the t-SNARE syntaxin3 (stx3), and the v-SNAREs slp4a and vamp7 is essential for apical cargo delivery (Vogel et al., 2015b; Vogel et al., 2017b). This cascade is required for the delivery of apical transmembrane transporters that are important for proper physiological function of enterocytes, such as sodium-hydrogen exchanger 3 (NHE3), glucose transporter 5 (GLUT5), and cystic fibrosis transmembrane conductance receptor (CFTR), but not for dipeptidyl-peptidase-4 (DPP4), sucrase-isomaltase (SI), and amino-peptidase-N (APN). This suggests the presence of additional trafficking routes and transport mechanisms for these apical cargos. Because the molecular signals for sorting and transport of apical cargo are thought to vary widely, several mechanisms have been proposed to underlie epithelial protein secretion (Levic and Bagnat, 2021). A common, characteristic feature of apical cargo is the presence of post-translational modifications, such as N- and O-linked glycosylations that are recognized by specific lectins, as well as GPI-anchors that allow sorting into cholesterol-rich lipid microdomains (Weisz and Rodriguez-Boulan, 2009; Zurzolo and Simons, 2016). Additionally, recent studies have proposed that protein oligomerization coincides with sorting into specialized membrane domains in the trans-Golgi network (TGN), which depends on the pH regulation of the TGN lumen (Levic and Bagnat, 2021; Levic et al., 2020). To uncover protein functions for a wide range of cellular processes, genome-wide clustered regularly interspaced short palindromic repeats (CRISPR)-mediated screens have advanced to a state-of-the-art strategy (Shalem et al., 2014; Shalem et al., 2015; Kampmann, 2018). In addition to their application to understanding the regulation of tumor biology, viral infection, or miRNA processing, CRISPR-mediated screening approaches have recently proven highly effective in discovering novel factors for intracellular protein trafficking and secretion (He et al., 2021; Zhu et al., 2021; Hutter et al., 2020; Stewart et al., 2017; Popa et al., 2020; Bassaganyas et al., 2019). Additionally, the CRISPR-Cas9 technology has been successfully used in Madin–Darby canine kidney (MDCK) cells with the generation of a collection of Rab-GTPase knockouts, which has provided great value for phenotypic analyses of Rab-KOs in epithelial cells (Homma et al., 2019). In this study, we employed the CRISPR-screening technology as an unbiased experimental strategy to uncover novel regulators of epithelial cell polarization and trafficking by investigating factors required for the apical delivery of DPP4. The brush-border hydrolase DPP4 is a type II transmembrane protein. It is heavily modified with N- and O-linked glycans in its extracellular domain (Misumi et al., 1992; Baricault et al., 1995; Fan et al., 1997), which have been suggested to be critical apical sorting determinants of DPP4 (Alfalah et al., 2002). Even though several studies have suggested diverse trafficking routes for DPP4, the mechanisms and protein machineries underlying these processes remain enigmatic so far (Casanova et al., 1991; Baricault et al., 1993; Low et al., 1992; Sobajima et al., 2014). Here, we conducted the first CRISPR screen in human intestinal epithelial cells to date. We present an experimental strategy for applying the CRISPR screening system in polarized epithelial cells to study novel protein functions. We have developed an easy-to-use and adaptable, FACS-based assay to measure the efficiency of protein secretion in polarized epithelial cells after genome editing. In combination with a detailed characterization of selected proteins by immunofluorescence and cryo-based electron microscopy, we have identified novel factors required for proper apico-basolateral polarization and secretion of apical cargo. Therefore, our dataset serves as a foundation for future studies aimed at deciphering novel mechanisms underlying epithelial polarization and polarized cargo transport. In addition, it provides a powerful resource for the investigation and validation of new congenital disease genes to be identified. Results Development of a genome-wide CRISPR screen to identify factors required for plasma membrane localization of the apical cargo DPP4 We established an unbiased CRISPR-Cas9-loss-of-function screen to define factors involved in surface targeting of the apical model cargo DPP4 in the enterocyte like colon carcinoma cell line, CaCo2 (Figure 1). DPP4 is a type 2 transmembrane protein that can be detected with antibodies binding to the extracellular C-terminus of the protein (Figure 1A). We made use of this feature to read out the efficiency of endogenous DPP4 surface delivery by fluorescence-activated cell sorting (FACS) in CaCo2 cells after epithelial polarization. Here, we used a period of 18–21 days, during which surface DPP4 signal is significantly increased in the course of cell surface expansion and specialized polarized trafficking processes (Figure 1B). In this context, we aimed to define factors required for apical membrane differentiation and cargo trafficking, thereby leading to a strong reduction of DPP4 after surface polarization. First, we generated Cas9-expressing CaCo2 cells and then transduced two biological replicates at a low multiplicity of infection (MOI) (0.2) using the human lentiviral GeCKOv2 CRISPR-library, selecting for successful viral integration with antibiotic treatment with puromycin. We then seeded the infected CaCo2 cultures at high density and allowed the confluent monolayers to further polarize and differentiate for 18 days. Next, polarized cells were detached, stained for endogenous DPP4, and subjected to FACS, separating those cells with only 10% of surface signal left, due to CRISPR targeting (Figure 1C and D). To determine the abundance of gRNAs in sorted versus unsorted cell populations, genomic DNA was isolated and read counts were determined by next-generation sequencing. Subsequent analysis using GenePattern and Galaxy analysis tools enabled the identification of 89 gRNAs significantly enriched in the sorted cell population (p<0.05) and represented genes whose downregulation had resulted in reduced DPP4 surface release (Figure 1D and E, Supplementary file 1A and B). Figure 1 with 1 supplement see all Download asset Open asset A CRISPR-mediated loss-of-function screen in polarized enterocytes. (A) Dipeptidylpeptidase 4 (DPP4) localizes to the apical brush-border of polarized enterocytes and can be detected with a specific antibody at its extracellular stalk domain. Top view (XZ) and lateral view (YZ) of a polarized CaCo2 monolayer. Scale = 5 µm. (B) During polarization, apical DPP4 is increased due to polarized traffic and surface expansion, which can be measured by flow cytometry (right panel, CaCo2 unpolarized versus polarized). HEK293T cells, not expressing DPP4, serve as quality control for staining specificity. (C) CaCo2-Cas9 cells are transduced with the lentiGuide-Puro library and selected with puromycin. After selection, CaCo2 cells are seeded to confluent monolayers and cultured for apico-basolateral polarization. Subsequently, cells are detached, stained, and subjected to fluorescence-activated cell sorting (FACS). Sorted and unsorted control cells are processed for gDNA extraction and genomically integrated CRISPR constructs are amplified by PCR. Finally, PCR products of sorted and unsorted cell populations are analyzed by next-generation sequencing and sgRNAs are ranked by their enrichment in the sorted vs. unsorted cell polpulation. (D) Sorting was performed for 10% of the cells, with lowest surface-signal intensity, thereby enriching for the cell population that had lost 90% of surface DPP4 signal, due to efficient CRISPR targeting. (E) 89 single-guide RNAs were significantly enriched in the sorted cell population. (F) Factors enriched in the sorted cell population could functionally be associated with secretory traffic, cytoskeletal architecture, or transcription, in a manual gene -ontology analysis. To exclude the possibility of aberrant effects caused by vacuolar apical compartment (VAC) formation in our screening workflow, we repeated the FACS screening assay with CaCo2 WT cells treated with colchicine (Gilbert and Rodriguez-Boulan, 1991). This treatment would induce VAC formation, but we found no change in apical DPP4 signaling (Figure 1—figure supplement 1A and B) and therefore concluded that VAC formation should not interfere with the FACS-based screening approach. A genome-wide CRISPR screen in polarized enterocytes identifies factors associated with secretory traffic Next, we wanted to get a comprehensive overview on the gene classes represented in our list of enriched gRNAs. However, automated KEGG pathway and gene enrichment analyses of our results were insufficient. Hence, we manually analyzed the 89 identified genes for common gene ontology (GO) terms and grouped them accordingly. We listed three GO terms from each category (biological process, molecular function, cellular compartment) for each hit, including the most common GO terms captured by the QuickGO -search database, focusing on including GO terms that indicate a role in the secretory pathway (Supplementary file 1C). Our analysis highlighted several genes, with functions related to the organization of the secretory pathway (Figure 1E and F, Figure 2A and B), including general organization and maintenance of organelles such as the endoplasmic reticulum (ER), the Golgi apparatus, or protein transport at early steps of the secretory pathway (e.g., KDELR2, RTN2, GOLGA8O). Further, identified hits were related to protein modification and transport at cis- and trans-Golgi compartments (GALNT2, SYS1), lipid-biosynthesis (MTMR2, PIP5K1C), and vesicle fusion and endocytic recycling (SNAP29, DSCR3). Two genes identified were associated with ER-plasma membrane (ER-PM) contact sites (TMEM110, ANO8). Furthermore, we found several factors required for various aspects of cytoskeletal organization such as actin-filament organization/polymerization (e.g., MARCKSL1, ARPC4-TTLL3), cell adhesion (e.g., ITGA5, FREM3, MPZ) but also microtubule organizing center (MTOC)/centriole- and cilium assembly and association (e.g., CCDC61, CCDC42B, C2CD3). Finally, we found numerous factors with functions related to DNA-templated transcription and cell differentiation (e.g., ETV7, NKX2-2, ERF), as well as mRNA processing/RNA splicing (e.g., SNRPE, SFSWAP), translation (e.g., RPL30, RPL2), and DNA repair/DNA replication (e.g., SFR1, ATAD5, REV1) (Figure 1E and F, Figure 2A). Figure 2 Download asset Open asset Gene ontology (GO) analysis of hits obtained in a CRISPR-mediated loss-of-function screen in polarized CaCo2 cells. (A) Schematic representation of significantly enriched genes obtained from a CRISPR screening approach, grouped and organized according to their associated GO terms. (B) GO association analysis of the seven factors that were chosen for CRISPR screen readout validation and further characterization. CC, cellular compartment; MF, molecular function; BP, biological process. Overall, in a CRISPR-mediated loss-of-function screen, we identified a variety of factors that affect surface transmission of an apical model cargo protein, DPP4, at different cellular levels. This underscores the value of our dataset and approach to identify novel factors for secretory membrane trafficking in polarized epithelial cells. Novel factors for surface localization of the apical cargo DPP4 After setting up a CRISPR-mediated screening platform in polarized CaCo2 cells, we validated our screening approach by further characterizing potentially novel factors for apical cargo traffic and membrane polarization. Since we had identified several genes with functions related to secretory trafficking (Figure 2A), we chose seven promising candidates for further analyses (Figure 2B). These factors function on various levels of the endomembrane system: the anoctamin family member anoctamin 8 (ANO8) and the stromal interaction molecule (STIM) enhancing tethering protein STIMATE (TMEM110) are involved in the formation and maintenance of ER-PM contact sites, and in turn, in apical PM-establishment in bile-canaliculi (Jha et al., 2019; Quintana et al., 2015; Chun Chung et al., 2020). The nonaspanin-family member TM9SF4 has been suggested to be required for transmembrane domain sorting in early steps of the secretory pathway but also in the generation of specialized membrane domains in the early cis-Golgi compartment (Perrin et al., 2015; Vernay et al., 2018; Yamaji et al., 2019). Polypeptide N-acetylgalactosaminyltransferase 2 (GALNT2) regulates O-linked glycosylation of transmembrane proteins in the Golgi and was chosen as a candidate for screen validation, with a potentially global effect on secretory traffic (Wandall et al., 1997; Moremen et al., 2012). Sorting nexin 26 (SNX26/ARHGAP33) was included since it has been described as a GTPase-activating protein for Cdc42, a major player in apical domain differentiation (Kim et al., 2013). Finally, we chose the lipid kinase subunit phosphatidylinositol 4-phosphate 5-kinase type-1 gamma (PIP5K1C) and the lipid phosphatase myotubularin-related protein 2 (MTMR2) for screen validation and further analysis since they are known regulators of apical phosphatidylinositolphosphate (PIP) pools (Xu et al., 2019; Román-Fernández et al., 2018). We generated knockout (KO) cell lines of those candidates using the CRISPR-technology and those gRNAs that had proven to efficiently target in our CRISPR screen (Figure 3A). We then analyzed KO cell lines for surface localization of DPP4 by flow cytometry using the previously described polarization assay from our CRISPR screen (Figure 3B). These measurements showed that targeting of the selected candidates indeed leads to reduced surface localization of DPP4, but to varying degrees (Figure 3C). The strongest effect on DPP4 surface localization was caused by interference with PIP5K1C (~75% reduction), followed by TM9SF4, TMEM110, and GALNT2 (~50% reduction). Interestingly, ANO8-, MTMR2-, and ARHGAP33-KOs showed the mildest phenotype (~30% reduction) (Figure 3B and C). Figure 3 Download asset Open asset Validation of selected candidates identified in a CRISPR-mediated loss-of-function screen. (A) Generation of clonal knockout (KO) cell lines of seven candidates chosen for primary CRISPR screen validation and further analysis. Any remaining protein levels in the KO clones after CRISPR targeting were determined by Western blotting compared with wild-type (WT) cells. Beta-actin was used as loading control. Molecular size markers are depicted in kDa. (B) The effect of the respective KOs on DPP4 surface transport, in KO cell lines. Cell lines were polarized for 18 days and then subjected to flow cytometry to determine the KO effect on DPP4 surface targeting. (C) Geometric means of DPP4-area (DPP4 intensity on the cellular surface) were determined of respective cell lines. All KO cell lines show varying extents of DPP4 surface reduction, with PIP5K1C-KO displaying the strongest and ANO8-KO the mildest effect. By growing KO cell lines of selected candidate genes and reanalyzing them for the effects of CRISPR targeting on PM localization of DPP4, we validated our primary CRISPR loss-of-function screen and thereby identified new players for surface localization of the apical cargo protein DPP4. For the subsequent analyses of epithelial phenotypes, we generated KO cell lines with a second set of gRNAs targeting the selected seven genes (Figure 1—figure supplement 1C) and included these cell lines in the analyses as indicated. 3D cyst models and high-resolution microscopy reveal novel factors for proper epithelial polarization Because apical transport and the correct establishment of epithelial polarity are closely linked, we investigated the relevance of the newly identified factors for polarization. Therefore, we performed 3D cyst assays using WT and the corresponding KO cell lines (Figure 4). Cysts were analyzed by immunofluorescence microscopy (IF) to determine the targeting of DPP4 to apical membrane domains. We found that all KO cell lines had severe defects in forming a single, central lumen, but rather established multiple or no lumina (Figure 4A and B). Although DPP4 was localized in the apical PM domains in all KO cell lines, TM9SF4-, MTMR2-, and ANO8-KO cell lines additionally displayed aberrant intracellular accumulation of DPP4 (Figure 4A and C). Figure 4 Download asset Open asset 3D cyst cultures demonstrate disrupted epithelial polarity. (A) 3D cyst assay were performed with WT and KO cultures. Immunofluorescence micrographs of 3D cysts generated from WT and KO cell lines. All knockdown cell lines form multiple lumina or no lumina. DPP4 localizes to actin-rich structures in al KO cell lines and additionally, to intracellular, actin-negative compartments in TM9SF4-, MTMR2- and ANO8-KO clones (white arrowheads). Scale = 10 µm. (B) Single central lumen formation was quantified. The percentage of cysts with a single central lumen is substantially decreased in the respective KO cells lines (dot box plot, Mann–Whitney U test. ***p< 0.005, n ≥ 100 cells per condition). (C) Delocalized DPP4 in cysts was quantified (dot box plot, Mann–Whitney U test. ***p<0.005, n.s., not significant). To characterize the involvement of the selected candidates in apico-basolateral polarization and apical transport in greater detail, we complemented fluorescence microscopy with cryo-based electron microscopy and investigated the ultrastructural phenotype and the subcellular distribution of selected marker molecules in the respective cell lines. To this end, TM9SF4-, ANO8-, ARHGAP33-, TMEM110-, MTMR2-, PIP5K1C-, and GALNT2-KO cell lines were grown on permeable filter membranes for 18–21 days to obtain fully polarized, differentiated 2D cell monolayers. Samples were then subjected to rapid cryo-fixation (high-pressure freezing and freeze-substitution) for transmission electron microscopy (TEM) or to conventional aldehyde fixation for scanning EM (SEM) and immunogold-TEM. In contrast to CaCo2 WT cells, all KO cell lines had conspicuous clusters of ectopic microvilli (Figure 5A–E, Figure 1—figure supplement 1D and E). They appeared either as well-organized ectopic brush-border and adjacent ectopic terminal web, lining distinct intracellular lumina (i.e., typical microvillus inclusions) (Figure 5A) or as less complex inclusions filled with intertwined masses of long, curved microvilli (Figure 5B, Figure 1—figure supplement 1E). Similar configurations occurred basolaterally as paracellular spots lined by densely packed microvilli (Figure 5C and E), frequently associated with ectopic tight- and adherens-junctions (Figure 5E). In addition, numerous long, curved microvilli were found facing the occasionally widened basolateral intercellular space (Figure 5D, Figure 1—figure supplement 1D). All these patterns (summarized in Supplementary file 2 and Supplementary file 3) together confirm the highly disturbed polarity of all KO cell lines studied here. Figure 5 Download asset Open asset Electron micrographs with various configurations of abundant ectopic microvilli reflect polarity defects in 18-day-old 2D cultures of selected CaCo2 knockout (KO) cell lines. (A) Distinct intracellular lumen, lined by ectopic brush-border and adjacent terminal web, thus, a typical microvillus inclusion (arrow) inside a PIP5K1C-KO cell. Arrowheads mark lysosomes. (B) Spherical intracellular mass of tangled microvilli associated with ectopic terminal web (arrow) inside a GALNT2-KO cell. Double arrowheads mark enlarged late endocytic organelles. (C) Paracellular, basolateral spot of densely packed microvilli plus associated ectopic terminal web (arrow) in polarized TM9SF4-KO-cell culture. (D) Numerous, slightly bent microvilli facing widened intercellular space (arrows) in polarized ANO8-KO cell culture. (E) Paracellular microvillar spot with adjacent ectopic terminal web (arrow) and associated cell junctions (double arrows) in polarized TMEM110-KO cell culture. Arrowheads mark lysosomes, double arrowhead marks poorly structured late endocytic organelles. (A–E) Scale = 1 µm. Notably, remarkable ultrastructural alterations also involved late endocytic and catabolic organelles (Figure 6). In WT CaCo2 cells, the different types and/or maturation stages of multivesicular bodies (MVBs) and lysosomes, as well as some autophagic organelles, appeared normal in size, abundance, and morphology (Figure 6A and B) and resembled patterns previously mapped for other cryo-fixed mammalian cell lines (e.g., MEF, HeLa: Vogel et al., 2015a; Yordanov et al., 2019; Hess and Huber, 2021). However, in most KO lines, the late endosomal/lysosomal endomembrane system was characterized – if not even dominated− by partly giant, spherical organelles (in addition to normal MVBs) − at the expense of normal lysosomes (Figure 5B and E, Figure 6C–G, double arrowheads). These large compartments had either weakly stained, homogeneous granular contents with a few intraluminal elements (Figure 5B and E, Figure 6C–G), or different amounts of partially degraded material. According to ultrastructural criteria, we tentatively interpreted these poorly structured, faint compartments as types of peculiar endolysosomes or (autophago)lysosomes (Bright et al., 2016; Fujita et al., 1990; Remis et al., 2014). Their size and frequency varied throughout the different KO lines. Especially in TM9SF4- and ANO8-KO, they reached dimensions of up to 2 × 6 µm in diameter (e.g., Figure 6C), in other KO cell lines only diameters of approximately 500 nm. In CaCo2 WT cells, we infrequently found this type of organelle as well, but here they had rather normal dimensions (Figure 6A and B, double arrowheads; Supplementary files 2 and 3). Figure 6 Download asset Open asset Ultrastructure of late endocytic and lysosomal organelles in CaCo2 wildtype (WT) cells versus selected knockout (KO) cell lines of cryo-fixed 18-day-old polarized filter cultures under steady-state conditions. (A, B) WT CaCo2 cells with normal endosomes and lysosomes: multivesicular bodies (MVBs) with varying contents (regarding intraluminal vesicle [ILV] number, size, and staining patterns), different types of (autophago)lysosomes (arrowheads), all filled with clearly stained, finely granular material plus membrane remnants (i.e., multilamellar bodies), and/or opaque, amorphous residues (i.e., dense-core lysosomes). Double arrowheads mark examples of inconspicuous spherical organelles with weakly stained, quite homogeneous granular contents harboring only sporadic ILVs and/or other structured components, interpreted as kind of endolysosome (Bright et al., 2016). (C) Large, poorly structured putative endolysosome (double arrowhead), virtually the predominant type of late endocytic and/or catabolic organelles in ANO8 KO at steady state. (D) Moderately sized and enlarged putative endolysosomes (double arrowheads) in PIP5K1C KO cells. (E) Enlarged putative endolysosome (arrowhead), the predominant type of late endocytic/catabolic organelles in TM9SF4 KO cells. (F) Moderately enlarged putative endolysosomes (double arrowheads) occurring together with normal lysosomes in ARHGAP33 KO cells. (G) Enlarged putative endolysosomes (double arrowheads) occurring together with slightly smaller lysosomes in MTMR2 KO cells. Arrows mark lateral microvilli. (A–G) Scale = 500 nm. Regarding the general architecture of the brush-border, SEM and actin fluorescence microscopy revealed more or less severe irregularities in in all KO cell lines studied here (Figures 7—9). They included patchy distribution or complete absence of apical microvilli, together with the occurrence of extremely long microvilli (Figure 7B–H). We then combined actin labeling with immunofluorescence microscopy using antibodies against the apical components DPP4 and stx3 (Figure 8A–H, Figure 9A–H). At first glance, we detected DPP4 in most CaCo2-KO monolayers predominantly at the apical plasma membrane. However, more detailed analysis of confocal stacks revealed that DPP4 was also mislocalized to intracellular sites in ARHGAP33-, TM9SF4-, PIP5K1C-, MTMR2-, and ANO8-KO cell lines (Figure 8B, F, and I), although to varying degrees. While DPP4 was mislocalized to subapical compartments in ARHGAP33-, TM9SF4-, and ANO8-KO cell lines, PIP5K1C- and MTMR2-KOs displayed DPP4 localization to large, actin-rich, basolateral structures, reminiscent of microvillus inclusions, observed by EM. Consistent with these observations, stx3 was detected at the apical brush-border microvilli in all KO cell lines (Figure 9B, F, and I). This was accompanied by additional ectopic localization of stx3 in TM9SF4-, ARHGAP33-, MTMR2-, and PIP5K1C-KOs, with MTMR2- and PIP5K1C-KOs exhibiting stx3-positive, basolateral inclusion-like compartments (Figure 9D and F), whereas stx3-positive structures were seen in apical regions in TM9SF4- and ARHGAP33-KO cells (Figure 9B and E). We further analyzed the localization of the apical membrane proteins aminopeptiase N (APN) and sucrase-isomaltase (SI) (Figure 8—figure supplements 1 and 2). Interestingly, we found APN delocalized to actin-rich intracellular compartments only in ARHGAP33-KO cells (Figure 8—figure supplement 1E). However, the apical localization of SI remained unaltered upon deletion of the selected genes (Figure 8—figure supplement 2A–H). Figure 7 Download asset Open asset Scanning electron microscopy (SEM) surface views on apical microvilli of 18-day-old polarized CaCo2 wildtype (WT) cells versus knockout (KO) cells. (A) CaCo2 WT cells with dense, quite uniform