Arp2/3-dependent regulation of ciliogenesis governs adaptive distal tubular epithelial cell states in kidney disease

Abstract

Proteinuric kidney disease substantially affects renal tubules through incompletely understood mechanisms. We identify elongation of primary cilia in distal renal tubules in the context of glomerular nephropathy. In renal biopsies and mouse models, tubular injury correlates with ciliary elongation, tubule dilation, and disruption of the cortical actin cytoskeleton. In vitro studies implicate biophysical cues of the glomerular filtrate and subsequent dysregulation of the actin cytoskeleton as contributing factors, confirmed by conditional deletion of N-WASP and Arp2/3 in vivo and in vitro. Electron and fluorescence microscopy revealed enlarged ciliary pockets, basal body mislocalization, and intracellular cilia formation in Arp3 knockout conditions. Transcriptome analysis identifies the essential role of cilia in maintaining adaptive tubular cell states, while persistent activation leads to disease progression through extracellular matrix remodeling, exemplified by Tenascin-C. Our findings establish cilia as central mediators of tubular adaptation to injury and identify the Arp2/3-dependent actin cytoskeleton as a critical regulator, providing essential insights into the pathogenesis of chronic kidney disease.

Proteinuria causes ciliary elongation and actin cytoskeleton remodeling, revealing a key mechanism in kidney disease progression.

INTRODUCTION

Chronic kidney disease (CKD) represents a major global health burden due to rising prevalence and mortality worldwide (1, 2). Current estimates indicate that CKD affects 10 to 14% of the global population, resulting in a considerable socioeconomic impact on health care systems (3). A substantial proportion of chronic renal failure is attributed to glomerular diseases, which are usually characterized by disruption of the renal filtration barrier clinically presenting with proteinuria. Despite advances in our understanding of the individual factors contributing to kidney injury, the intricate interplay between glomerular dysfunction (e.g., proteinuria) and the renal tubular compartment remains incompletely understood (4). CKD is commonly characterized by histological features such as interstitial fibrosis, pseudo-cystic degeneration of renal parenchyma, and tubular atrophy. Another recently emerging feature of tubular injury is the observation of elongated cilia (5–12). However, underlying mechanisms that lead to ciliary lengthening and the significance for acquired kidney disease have not yet been elucidated.

Primary cilia are microtubule-based organelles that extend from the apical surface of renal tubular epithelial cells into the lumen (13). These sensory organelles consist of a central axoneme (ciliary shaft) derived from a modified centriole, commonly referred to as the basal body. Ciliogenesis is governed by a complex network of dynamic molecular machinery linked to the microtubule cytoskeleton (14). In general, ciliogenesis follows two distinct mechanisms—either the extracellular (EC) or the intracellular (IC) pathway (15). The EC pathway of ciliogenesis is predominantly observed in epithelial cells and is characterized by the docking of the basal body to the apical cell membrane as an initiating event. Regulation of ciliary length depends on a steady-state balance between ciliary assembly and disassembly, which is mediated by the intraflagellar transport (IFT) machinery (14). Unexpectedly, recent in vitro studies have identified the actin cytoskeleton, in particular, branched actin networks nucleated by the Arp2/3 complex, as an important regulator of ciliary length (16–18). Actin nucleation by the Arp2/3 complex is tightly regulated by nucleation promoting factors (i.e., N-WASP and SCAR/WAVE proteins) and coregulators [i.e., cortactin (CTTN)], which integrate a multitude of cellular signaling cascades (19, 20). Nevertheless, the mechanisms by which the Arp2/3 complex contributes to cilia length control remain a subject of considerable debate (17, 18, 21).

Primary cilia serve as central signaling hubs and mechanosensors, orchestrating a variety of signaling pathways essential for kidney development and homeostasis (13, 22). The relevance of cilia in kidney function is demonstrated by numerous hereditary ciliopathies (23), with autosomal dominant polycystic kidney disease (ADPKD) being the most prevalent (24). In ADPKD, mutations in PKD genes lead to altered calcium signaling and cilia lengthening and translate into tubular dilation and cystic degeneration of the renal parenchyma (5, 25). Beyond hereditary disease conditions, growing evidence suggests that ciliary dysfunction might also be associated with proteinuric glomerular diseases, diabetic nephropathy, and other entities of renal injury, thereby underscoring the broad pathophysiological relevance of this organelle (5–12).

Recent advances in methodologies such as single-cell RNA sequencing (scRNA-seq) and spatial biology technologies have profoundly transformed our understanding of epithelial cell plasticity in kidney disease (26, 27). These observations have been integrated into a concept, where proximal and distal renal tubular cells respond to injury within a continuous spectrum of adaptive cell states (28–30). These molecular phenotypes are either associated with repair and regeneration or, in the case of maladaptive and degenerative cell states (i.e., dedifferentiation), they promote disease progression. The correlation between these transcriptionally defined cell states and specific morphological features, in particular changes in cilia length, as well as their role in glomerular disease, has not been described.

Therefore, our study aimed to delineate the relevance of altered ciliary morphology in distal tubules in proteinuric glomerular diseases and the underlying regulatory mechanisms. Here, we elucidate the function of branched actin networks in regulating ciliary length, both in vitro and in vivo. We demonstrate a mechanistic link between actin dynamics, ciliary lengthening, and transcriptional reprogramming, unveiling a previously unrecognized aspect of epithelial cell plasticity. Collectively, our results suggest that modulation of ciliary signaling may serve as a central mechanism through which tubular epithelial cells adapt to proteinuric kidney injury.

RESULTS

Proteinuric glomerular disease is associated with cilia elongation in distal renal tubules

To investigate the involvement of cilia in cellular response to distal tubular cell injury in proteinuric glomerular diseases, we analyzed cilia in damaged tubules of patients diagnosed with focal segmental glomerulosclerosis (FSGS) and immunoglobulin A nephropathy (IgAN) (Fig. 1, A and B, and fig. S1). Here, we observed a notable elongation of cilia in the aforementioned pathologies [mean control, 3.27 μm; FSGS, 4.10 μm; and IgAN, 4.14 μm; also characterized by intratubular precipitates/proteinaceous casts preferentially in distal tubular segments (31)]. Elongated cilia were predominantly detected in tubules with only mild signs of damage (e.g., mild dilation or intraluminal cast formation). In contrast, tubules with cystic dilation and marked epithelial flattening exhibited a notable shortening and loss of cilia. These observations already indicate a continuum in the ciliary response, potentially correlating to the level of tubular damage. To further evaluate the specificity of these findings, we used different models of genetically induced proteinuric glomerular disease (fig. S2). These models encompass causative mutations in the podocyte slit diaphragm gene Nphs2 (32), the CoQ10 metabolism gene Pdss2 (33), podocyte-specific deletion of the integrin adhesion complex component Parva (34), and deletion of the Alport-Syndrome gene Col4a3 (35). While primary proteinuric models (Nphs2, Col4a3, Pdss2, and Parva) exhibited a pronounced elongation of distal (and proximal) tubule cilia, models without severe proteinuria and very low tubular damage at the analyzed stage of disease (NOH—T cell–mediated nephritis model) (36) showed no significant alterations in ciliary length (Fig. 1, C and D, and fig. S2). Detailed analysis revealed a correlation between the dilation of tubules and the length of cilia in these models (Fig. 1E). Additional histological analysis of the medullary renal compartment substantiated a strong correlation between dilation of distal tubules, cilia elongation, and the presence of intratubular proteinaceous casts (predominantly localized in distal tubular segments: Fig. 1, F to I, and fig. S2). Together, our observations indicate that upon mild tubular damage (in conditions of glomerular proteinuria), adaptive responses of distal tubular epithelial cells correspond to the elongation of primary cilia.

Fig. 1. Proteinuric glomerular disease is associated with cilia elongation in distal renal tubules.

(A and B) Immunofluorescence (IF) analysis of distal tubules in human kidney biopsies with diagnoses of FSGS, IgAN, or healthy controls (Ctrl.). Cilia were stained with acetyl–α-tubulin, the thick ascending limb (TAL) of the nephron with NKCC2, all tubule compartments with wheat germ agglutinin (WGA) lectin and nuclei with Hoechst. Maximum intensity projections of z-stacks were generated to show the entire cilium. White arrows indicate normal, yellow arrows indicate elongated, and white arrowheads indicate shortened cilia. (B) Mean ciliary length per patient (FSGS, N = 7; IgAN, N = 6; control N = 7, dots indicate individual patients analyzed; error bars indicate means and S.E.M.; *P < 0.05) and single cilia measurements (violin plot; Ctrl., N = 1367; FSGS, N = 2052; IgAN, N = 1540). (C) IF analysis of murine models of glomerular disease. White arrows indicate normal, and yellow arrows indicate elongated cilia. (D) The mean cilia length in distal tubules (TAL and collecting duct tubules) was quantified (dots indicate individual animals analyzed—Nphs2 WT, Nphs2 mutant, Parva WT, Parva KO, Pdss2 mutant, NOH, and NOH control N = 5 each; Pdss2 WT N = 4; Col4A3 WT and KO N = 6 each; error bars indicate means and S.E.M.; n.s., not significant; **P < 0.01, ***P < 0.001, and ****P < 0.0001). (E) Correlation analysis between lumen diameter and mean ciliary length per tubule in disease models with elongated cilia. (F and G) Quantification of dilated tubules and proteinaceous casts in distal (medullary) tubules per area in the indicated disease models (dots indicate individual animals analyzed). (H) Correlation analysis between proteinaceous cast formation and dilation of distal tubules. Red dots indicate animals with elongated cilia, and blue dots with normal cilia length. (I) Representative IF of a distal tubule showing elongated cilia (white arrow) and intraluminal proteinaceous cast formation [white asterisk; visualized via autofluorescence (AF)].

Combined biological and physical properties of the proteinuric glomerular filtrate affect ciliary elongation via the actin-cytoskeleton

Our observations showed a significant correlation between proteinuria and cilia elongation in glomerular kidney disease (Fig. 1 and fig. S2). Therefore, we speculated that this phenotype might be caused by changes in the glomerular filtrate in conditions of proteinuria. Relevant factors include the presence of serum proteins (albumin, growth factors, EC matrix proteins, and others) (37) and the formation of proteinaceous casts, which lead to changes in intraluminal physical properties, resulting in decreased flow rates. To investigate these mechanisms in vitro, murine inner medullary collecting duct cells (mIMCD3) and human renal proximal tubule epithelial cells (RPTECs) were used as a model for ciliated kidney cells. The use of these models demonstrated that the addition of human serum (mIMCD3) and high concentrations of growth factor–reduced basement membrane extracts (mIMCD3 and RPTEC) promoted the formation of elongated cilia (Fig. 2, A and B, and fig. S3). The sole addition of albumin resulted in no alteration (RPTEC) or shortened cilia (mIMCD3), which might be explained by exerted cytotoxicity on tubular cells (38). As cilia elongation was induced by high concentrations of basement membrane extracts, we hypothesized that these effects may be attributed to increased medium viscosity. Increased medium viscosity translated into an increase in ciliary length in the mIMCD3 and RPTEC model. Additional cosupplementation with basement membrane extracts resulted in a further increase in cilia length, indicating a potential synergistic effect of biological (composition) and physical (viscosity) properties (Fig. 2, A and B). Varying levels of physiological fluid flow rates in our model did not directly affect ciliary length (Fig. 2, A and B, and fig. S3). These findings indicate that fluid viscosity may play a more relevant role in influencing cilia length. The observed link between ciliary elongation and tubule dilation suggests the involvement of cellular processes related to morphological adaptation, such as the actin cytoskeleton. Here, we observed an impaired apical localization of the cortical branched actin cytoskeleton polymerization machinery (N-WASP, WAVE2, CTTN, and ARP3) under conditions that lead to increased cilia length in vitro and in vivo (Fig. 2, C and D).

Fig. 2. Combined biological and physical properties of the proteinuric glomerular filtrate affect ciliary elongation via the actin-cytoskeleton.

(A and B) Analysis of cilia length in ciliated mIMCD3 cells exposed to albumin (N = 3), human serum (N = 4), basement membrane extracellular matrix (ECM) [100 μg/ml (low) or 1 mg/ml (high); N = 3 each], increased medium viscosity of 150 or 1500 cP (methylcellulose; N = 4 each), 1500 cP and high ECM (N = 3), laminar flow of 0.2 or 2 dyn/cm² (N = 3 each) or negative control medium (N = 11). Scatter plots show mean cilia length per replicate (statistical comparisons to Ctrl. samples is shown; dots indicate independent experiments per condition; error bars indicate means and S.E.M.; n.s., not significant; ***P < 0.001, ****P < 0.0001). Violin plots show individual cilia measurements per condition (statistical analysis of selected comparisons is shown; n.s., not significant; ****P < 0.0001). (B) Representative images of indicated treatment conditions. Cells were stained for acetyl–α-tubulin, filamentous actin (F-actin) with phalloidin, and nuclei with Hoechst. White arrows indicate short and red arrows elongated cilia. (C) N-WASP, WAVE2, and CTTN regulate Arp2/3 complex–dependent polymerization of branched actin filaments at the cell cortex, suppressing cilia elongation. (D) Representative images of mIMCD3 cells cultured in control or 1500-cP high-viscosity medium for 24 hours. Cells were stained for CTTN and F-actin to visualize the cortical (branched) actin cytoskeleton (white asterisks indicate the leading edge, and white arrows indicate filopodia). (E to G) CTTN, N-WASP, WAVE2, and ARP3 stainings in the collecting duct (CD) of proteinuric Parva KO animals demonstrate a reduction in apical localization (arrowheads indicate the apical cell cortex). CDs were stained with Dolichos biflorus agglutinin (DBA) lectin, cilia with acetyl–α-tubulin, and nuclei with Hoechst (maximum intensity projections are shown).

Specific deletion of N-WASP (Wasl) in distal tubules results in elongation of primary cilia

Control of ciliary length is governed by a complex network of molecular mechanisms, including the actin cytoskeleton (21). Previous reports have elegantly demonstrated the correlation between reduced activity of the Arp2/3 complex (including ARP3 and the actin nucleation promoting factor N-WASP) and increased ciliary length (16, 39). However, to the best of our knowledge, this mechanism has only been established in vitro using knockdown and/or inhibitor approaches. To overcome these limitations, we generated a distal tubule-specific N-WASP knockout (KO) mouse model (Wasl^fl/fl*Cdh16-Cre) (Fig. 3, A and B). Unexpectedly, Wasl KO mice develop only minor renal phenotypes up to 5 months of age (Fig. 3, C to G). Nevertheless, the analysis of thick ascending limb (TAL) tubules and collecting duct (CD) tubules demonstrated an approximate 35% elongation of cilia in all compartments (Fig. 3, H to J). Given this significant but modest elongation of cilia and the mild phenotype of Wasl KO animals, we reasoned that ongoing compensation by other Arp2/3 complex nucleation promoting factors like SCAR/WAVE proteins (i.e., WAVE2) might be responsible.

Fig. 3. Specific deletion of N-WASP (Wasl) in distal tubules results in elongation of primary cilia.

(A) A conditional Wasl (N-WASP) KO for the distal tubule compartment was generated by combining the Wasl^fl/fl allele with the Cdh16-Cre transgene. (B) IF confirms the loss of N-WASP expression in distal renal tubules of Wasl^fl/fl*Cdh16-Cre mice. Cre recombinase–positive tubular cells were labeled by enhanced green fluorescent protein (EGFP), applying the mT/mG reporter allele [tubules were stained with WGA and nuclei (blue) with Hoechst; orange arrow indicates interstitial cells without Cre expression]. (C) Quantification of blood urea nitrogen (BUN) in Wasl WT and Wasl^fl/fl*Cdh16-Cre (KO) animals at 5 months of age (N = 5 animals per genotype; dots indicate individual animals analyzed; error bars indicate means and S.E.M.; n.s., not significant). (D to G) Representative images of PAS or IF-stained kidneys from Wasl WT and Wasl KO animals at 5 months of age. Black boxes indicate selected regions for magnified images. KO animals show occasional dilatation of CD tubules (red asterisks) and reduced cell density (red arrowheads). The staining of CD tubules by DBA lectin shows increased proportion of DBA-negative cells (orange arrows) in CDs of KO animals (AF, autofluorescence). Quantifications show no formation of tubular cysts (cystic index), interstitial fibrosis, and tubular atrophy (IF-TA score) but a reduced fraction of DBA-positive CD cells in KO kidneys (N = 5 animals per genotype; n.s., not significant; ***P < 0.001). (H) IF staining of WT and Wasl^fl/fl*Cdh16-Cre KO mice for cilia (acetyl–α-tubulin) and distal tubule compartments using NKCC2 (TAL) and DBA lectin (CD). Maximum intensity projections are shown. Orange arrows indicate representative cilia. (I and J) Quantification of mean cilia length in indicated tubule compartments (N = 5 animals per genotype at 5 months of age; dots indicate individual animals analyzed; error bars indicate means and S.E.M.; **P < 0.01, ***P < 0.001).

Specific deletion of ARP3 (Actr3) in distal tubules results in dilation of tubules accompanied by progressive kidney disease

Therefore, we generated a distal tubule–specific ARP3 KO mouse model (Actr3^fl/fl*Cdh16-Cre), entirely abolishing actin nucleation by the Arp2/3 complex (Fig. 4, A and B), also given the ubiquitous expression of components of the Arp2/3 complex within the entire tubular system of the kidney (fig. S4). Physiological analysis revealed a marked decline in renal functional parameters, as evidenced by elevated serum urea levels and macroscopic signs of kidney sclerosis in Actr3 KO mice at 3 and 4 weeks of age (Fig. 4, C and D, and fig. S4). A comprehensive histological analysis using periodic acid–Schiff (PAS) and multiplex immunofluorescence (IF) staining demonstrated the presence of tubular dilation, tubular atrophy, and interstitial fibrosis [characterized by interstitial stroma expansion, collagen-1 deposition, and an increase in SMA (smooth muscle actin) positive myofibroblasts] (Fig. 4, E to H, and fig. S5). Moreover, a notable inflammatory response (characterized by infiltration of F4/80⁺ macrophage, CD3⁺CD4⁺ T cells, CD3⁺CD8⁺ T cells, and CD3⁺FOXP3⁺ regulatory T cells) was evident in Actr3 KO kidneys (Fig. 4, E, I to K, and fig. S6). These changes, including tubular dilation and the formation of tubular cysts, were prominently present at 3 weeks of age. However, only mild pathological changes were observed at 2 weeks of age. Together, these observations provide substantial evidence that ARP3 (Actr3) and branched actin filaments play an indispensable role in renal distal tubular function.

Fig. 4. Specific deletion of ARP3 (Actr3) in distal tubules results in dilation of tubules accompanied by progressive kidney disease.

(A) Conditional Actr3 (ARP3) KO mice were generated by combining the Actr3^fl/fl allele with the Cdh16-Cre transgene. (B) IF staining confirms the loss of ARP3 expression in distal renal tubules of Actr3^fl/fl*Cdh16-Cre mice. Cre-positive tubular cells were labeled by EGFP, applying the mT/mG reporter allele [tubules were stained with WGA and nuclei (blue) with Hoechst]. The white arrow indicates the presence of ARP3-positive interstitial cells in Actr3^fl/fl*Cdh16-Cre mice. (C) Kidneys of Actr3 KO mice exhibited a pale and smaller appearance compared to WT. (D) Quantification of BUN in Actr3 WT and KO animals at 2 and 3 weeks (w) of age (N = 5 animals per time point and genotype; dots indicate individual animals analyzed; error bars indicate means and S.E.M.; **P < 0.01). (E) Representative images of PAS or multiplex IF (mIF)–stained kidneys from Actr3 WT and KO animals. Black boxes indicate selected regions for magnified images. KO animals show progressive dilatation of medullary tubules (red asterisks) and interstitial fibrosis (red arrowheads). SMA and COL1A1 indicate a marked expansion of mesenchymal cells and deposition of ECM in the interstitium of KO animals. 7-plex mIF analysis shows infiltrating immune cells in Actr3 KO kidneys (F4/80⁺ macrophages (white arrows); CD3⁺ (arrowheads), CD4⁺, CD8⁺ and FOXP3⁺, T cell populations; LAMC1, ECM and basement membranes). (F to K) Quantifications of kidney staining [N = 5 animals per time point and genotype [(F) to (I)] or N = 4 for mIF [(J) and (K)]; dots indicate individual animals analyzed; error bars indicate means and S.E.M.; n.s., not significant; *P < 0.05, **P < 0.01]. Graphs show cystic index (F), IF-TA score (G), and inflammation (immune cell infiltrates) (I), SMA within the medulla (H), F4/80⁺ macrophages per area (J), and CD3⁺ T cells per area (K).

Loss of ARP3 results in elongation of primary cilia in kidney distal tubule cells in vitro and in vivo

To analyze the impact of the Arp2/3 complex on ciliary length control of distal tubule cells in vivo, the Actr3^fl/fl*Cdh16-Cre mouse model was used at 2 weeks of age. At this early stage, KO mice exhibited only initial dilation of tubules (Fig. 4F). The analysis of the TAL tubules, distal convoluted tubules (DCT), and CD tubules demonstrated an approximate twofold elongation of cilia in Actr3 KO mice, surpassing the effect observed in Wasl KO mice (Fig. 5, A to E). Of note, Actr3 KO animals showed no indications of glomerular disease or disruption of the glomerular filtration barrier up to 3 weeks of age (fig. S7). To further validate this ciliary phenotype in vitro, two independent Actr3 KO (KO-1 and KO-2) and two wild-type (WT) control (Ctrl.-1 and Ctrl.-2) mIMCD3 cell lines were created using CRISPR-Cas9 genome editing (Fig. 5F). The subsequent loss of cortical branched actin networks in Actr3 KO cell lines was reflected by the loss of lamellipodia and cortical CTTN staining (Fig. 5, G and H). Analysis of cilia using these cell lines demonstrated an increased percentage of ciliated cells and an approximate twofold elongation of cilia in Actr3 KO mIMCD3 cells compared to WT cells (Fig. 5, I to L), thereby corroborating our in vivo observations.

Fig. 5. Loss of ARP3 results in elongation of primary cilia in kidney distal tubule cells in vitro and in vivo.

(A) IF staining of WT and Actr3^fl/fl*Cdh16-Cre KO mice for cilia (acetyl–α-tubulin) in distal tubule compartments using NKCC2 (TAL), NCC (DCT), DBA lectin (CD), and Hoechst (nuclei). Maximum intensity projections are shown. Orange arrows indicate representative cilia. (B to D) Quantification of mean cilia length (N = 4 animals per genotype at 2 weeks of age; dots indicate individual animals; error bars indicate means and S.E.M.; *P < 0.05, ***P < 0.001). (E) Violin plot of single cilia measurements (Actr3 WT N = 1722, Actr3 KO N = 1525, Wasl WT N = 1243, Wasl KO N = 1445; ****P < 0.0001). (F) Validation of mIMCD3 WT control (Ctrl.-1 and -2) and Actr3 KO (KO-1 and -2) cell lines. Western blot for ARP3 (ACTR3) and ARP2 (ACTR2) confirmed loss of ARP3 and indicated the degradation of the Arp2/3 complex. (G and H) Staining for CTTN and F-actin (phalloidin) demonstrates the loss of CTTN-positive leading edges and lamellipodia in Actr3 KO cells. White boxes indicate regions for magnified images. White asterisks indicate CTTN at the cell cortex in WT, and white arrows loss in KO cells. The percentage of CTTN-positive cell cortex circumference was quantified (N = 25 cells per genotype; dots indicate individual cells analyzed; error bars indicate means and S.E.M.; ****P < 0.0001). (I) Representative cells stained for cilia (acetyl–α-tubulin), basal bodies (γ-tubulin), and nuclei (Hoechst) demonstrate the elongation of cilia in Actr3 KO cells. White arrows indicate basal bodies. (J and K) Quantification of ciliated cells and cilia length in control and Actr3 KO cells (N = 3 independent experiments; dots indicate results per genotype and experiment; error bars indicate means and S.E.M.; **P < 0.01, ***P < 0.001). (L) Violin plot of single cilia measurements per genotype (Ctrl.-1&2 N = 374 and KO-1&2 N = 559 cilia).

Loss of Ift88 and inhibition of dynein function reduces cilia formation and length in Actr3 KO cells

Several mechanisms have been proposed to explain ciliary elongation in Arp2/3-deficient cells. These include an increased pool of soluble actin- and tubulin-monomers and delayed cilia disassembly through increased IFT (as evidenced by the accumulation and formation of IFT protein–positive ciliary bulges) (40, 41). However, assessing these hypotheses in Actr3 KO cells and mice did not further substantiate these proposed explanatory mechanisms (Fig. 6, A to H, and fig. S8). Analysis of soluble and insoluble pools of α-tubulin, acetyl–α-tubulin, and β-actin in Actr3 KO and WT cells did not reveal any detectable changes (Fig. 6A and fig. S8, A to D). Moreover, inhibition of actin polymerization by cytochalasin D increased cilia length in WT cells in a dose-dependent manner but not in respective KO cells (Fig. 6B and fig. S8E). Only high doses of cytochalasin D led to a further increase in cilia length in KO cells. These findings are more likely to be explained by partial compensation of the loss of branched actin filaments by linear filaments than by changes in free actin and tubulin levels. Furthermore, the inhibition of actin-myosin filament formation by the ROCK inhibitor Y-27632 increased cilia length in WT but not in KO cells (Fig. 6C). The analysis of cilia assembly [fetal calf serum (FCS) starvation] and disassembly (addition of FCS) did not demonstrate any significant differences between Actr3 WT and KO cells, despite the increase in cilia length in KO cells, which occurred independently of the respective culture conditions (Fig. 6, D to E). In addition, no difference in the number of IFT protein–positive ciliary bulges was observed in Actr3 KO animals. Moreover, even reduced IFT81 and IFT88 levels were noted when normalized to the length of the ciliary shaft (Fig. 6, F to H, and fig. 8, F and G). However, inhibition of cilia formation through CRISPR-Cas9–mediated deletion of Ift88 or the inhibition of dynein function via application of ciliobrevin D resulted in a pronounced and significant reduction in cilia formation as well as length in respective Actr3 KO cells (Fig. 6, I and J, and fig. S8H).

Fig. 6. Loss of Ift88 and inhibition of dynein function reduces cilia formation and length in Actr3 KO cells.

(A) Western blot quantification of the soluble and insoluble α-tubulin fractions in Actr3 cells [N = 2 independent experiments; Actr3 Ctrl.-1&2 (total N = 4) and KO-1&2 (total N = 4) cell lines were pooled for statistical analysis, dots indicate measurements per experiment and genotype; error bars indicate means and S.E.M.; n.s., not significant]. (B to E) Quantification of mean cilia length in 100 nM or 1 μM cytochalasin D, 10 μM Y-27632, FCS, or FCS readdition for 90 min treated Ctrl.-1 and Actr3 KO-1 cells (N = 4 or 3 independent experiments; dots indicate mean per genotype and experiment; error bars indicate means and S.E.M.; n.s., not significant, *P < 0.05, **P < 0.01). (F to H) Quantification of IFT81-positive ciliary bulges and percentage of IFT81-positive ciliary area in CD tubules of WT and Actr3 KO animals. Representative images of IFT81, acetyl–α-tubulin, DBA lectin, and Hoechst (nucleus) staining (maximum intensity projections are shown). Orange arrows indicate IFT81-positive regions of the ciliary shaft (N = 4 animals per genotype at 2 weeks of age; dots indicate individual animals analyzed; error bars indicate means and S.E.M.; n.s., not significant, **P < 0.01). (I to K) Analysis of cilia in Actr3 KO-1 Ift88 KO (Actr3 Ift88 double KO – “dKO”), Actr3 KO-1 IFT88 WT (KO-1), and 0.5 μM ciliobrevin D (KO-1 CB)–mIMCD3 cells. Representative images show acetyl–α-tubulin, F-actin (Phalloidin), and Hoechst (nucleus, Nucl.) in the indicated cell lines. White boxes indicate regions for magnified images. Graphs show quantification of cilia length and ciliated cells in indicated cell lines (N = 3 independent experiments; dots indicate mean per genotype and experiment; error bars indicate means and S.E.M.; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001).

Fig. 8. ARP3-dependent regulation of cilia reprograms the transcriptome of kidney tubule cells.

(A) Transcriptome analysis of kidney medullas from Actr3 KO and WT mice at 2 weeks of age, nonstarved Ctrl.-1/-2 and Actr3 KO-1/-2 and starved Ctrl.-1, KO-1, Actr3 KO-1 Ift88 KO (dKO) and ciliobrevin D–treated KO-1 (CB) cells. Datasets of Pkd1, Pkd2, Ift140, and Dync2h1 KO mIMCD3 cells were reanalyzed (experiments were color-coded). (B) Differential gene expression analysis (DGEA) of nonstarved (green) and starved (blue) Actr3 KO versus Ctrl. cell lines [red dots, adjusted P value <0.05 with absolute log₂ fold change (FC) >1]. (C-1) Intersection analysis with reference datasets (Pkd1, Pkd2, Ift140, and Dync2h1). Bar graphs show significantly regulated genes in KO versus Ctrl. datasets (adjusted P value <0.05) and in the indicated number of reference datasets. (C-2) Box plots indicate absolute log₂ FCs of genes for the comparison of Actr3 KO versus Ctrl. cells or mean absolute log₂ FCs across all datasets. (D) DGEA of starved dKO or CB-treated KO-1 versus control KO-1 cells. (E) Normalized log₂ FC of indicated experiments to respective controls. (F) Venn diagram of significantly and matching regulated genes (adjusted P value <0.05 per dataset, inverse regulation by dKO and CB). (G) GO-term analysis of genes with matching regulation in all datasets. (H) Heatmap of significant regulated cilia genes. (I) DGEA of Actr3 KO and WT mice. (J) Intersection analysis of genes (adjusted P value <0.05) in nonstarved (green) or starved KO cells (blue), and Actr3 KO mice (brown). Graphs show matching and significantly regulated genes per intersection. TFs and co-TFs filtered by FC and inverse regulation in dKO and CB cells. (K) Heatmap of identified TFs (n.s., not significant). (L and M) IF analysis of LHX1 and CITED2 in DBA lectin or NKCC2 positive tubules (arrowheads indicate increased LHX1 and arrows loss of CITED2).

IC and EC localization of the primary cilium is regulated by ARP3 and the actin cytoskeleton

To gain further insights into the ciliary phenotype resulting from the loss of ARP3 (Actr3), ultrastructural analysis using scanning electron microscopy (SEM) was performed. SEM demonstrated a pronounced elongation of the ciliary shaft region in respective Actr3 KO cells (Fig. 7A and fig. S9A). However, no evidence of ciliary bulges or additional alterations of the shaft were observed. It is generally accepted that the mIMCD3 cell model serves as a prime example of the so-called EC pathway of ciliary assembly, which forms almost no ciliary pockets (42). Unexpectedly, we observed prominently formed invaginations of the cell membrane at the ciliary base at high frequencies in Actr3-deleted mIMCD3 cells, resembling ciliary pockets. Furthermore, correlative light and electron microscopy (CLEM) confirmed not only the presence of partially IC (PIC) cilia shafts but also the existence of entirely intracellularly localized cilia (Fig. 7B). We next sought to validate these findings by analyzing IC/EC ciliary localization using an established fluorescence reporter system (sspH-Smo-GFP) (43). This system consists of a pHluorin–green fluorescent protein (GFP)–tagged Smoothened (Smo) protein labeling the entire shaft of the primary cilium, whereas the EC portion is then selectively detected with phycoerythrin (PE)–tagged anti-GFP antibodies (Fig. 7C). Using this reporter system on WT and Actr3 KO mIMCD3 cells confirmed an almost complete EC localization of cilia in WT cells (mean EC, 96%; PIC, 3%; and IC, 1%) (Fig. 7, C and D). On the contrary, respective Actr3 KO cells showed entirely IC and PIC cilia (Actr3 KO-1: mean EC, 49%; PIC, 29%; and IC, 22%; Actr3 KO-2: mean EC, 74%; PIC, 11%; and IC, 15%) (Fig. 7, C and D, and fig. S9B). Notably, the EC shaft length, as well as the total shaft length, were highly increased in Actr3 KO compared to WT cells, demonstrating a significant contribution of the EC shaft to the observed cilia phenotype in Actr3 KO cells (Fig. 7, E and F, and fig. S9C). Acute inhibition of actin polymerization by treatment with cytochalasin D resulted in a similar phenotype as observed in Actr3 KO cells, indicating loss of branched actin filaments as an underlying mechanism (Fig. 7, C to F, and fig. S9, B and C). Therefore, we speculated that cilia elongation and the IC location of the axoneme might represent either interdependent or independent ciliary phenotypes in Actr3 KO cells. However, the simulation of potential models for interdependent regulation of these phenotypes did not substantiate this hypothesis (Fig. 7G). Furthermore, no discrepancy in cilia length was detected between cells with total IC localization and cells with (partial-) EC localization, suggesting that subcellular localization does not influence cilia length control (Fig. 7H). On the contrary, a significant number of IC cilia remained detectable in Actr3 KO Ift88 KO [double KO (dKO)] cells, suggesting that cilia length does not influence subcellular localization in Actr3 KO cells (Fig. 7, C and D). In summary, our results not only demonstrate the involvement of the Arp2/3 machinery in ciliary length control but also indicate that diminished cortical actin networks independently influence ciliary assembly modes, as reflected by the presence of IC cilia in respective Actr3 KO mIMCD3 cells.

Fig. 7. IC and EC localization of the primary cilium is regulated by ARP3 and the actin cytoskeleton.

(A) SEM of Ctrl.-1 and Actr3 KO-1 mIMCD3 cells demonstrate ciliary shaft elongation (arrowheads) and the emergence of ciliary pockets (arrow) in KO cells. (B) CLEM reveals the presence of partially and completely IC localized cilia in Actr3 KO cells [cilia (acetyl–α-tubulin), basal bodies (γ-tubulin), and nuclei (Hoechst)]. (C) Analysis of IC (magenta) and EC (green) ciliary localization using the sspH-Smo-GFP reporter system. Representative images illustrate the partial and complete IC localization of cilia in Ctrl., Actr3 KO, dKO, and 1 μM cytochalasin D (Cyto. D)–treated cells. (D to F) Quantification of cilia localization, fraction of EC to total cilia shaft length, and EC shaft length in Actr3 or WT cells treated with DMSO (WT ctrl.) or 1 μM cytochalasin D. Stacked bar graphs indicate percentages of completely EC, PIC, or completely IC localized cilia (N = 3 independent experiments; dots indicate mean values per genotype and experiment; Actr3 Ctrl.-1&-2 were pooled (total N = 4); error bars indicate means and S.E.M.; **P < 0.01, ****P < 0.0001). (G) Analysis of interdependencies between cilia length and IC localization of the axoneme in Actr3 KO-1&-2 cells. Scatter dot plot demonstrates the relationship between total cilia length and RICAL of cilia with PIC localization. Graphs show linear interpolation analysis of cilia measurements and simulated datasets of models with positive (A, green), negative (B, orange), or no correlation (C, blue) between these parameters (N = 125 KO-1&-2 cilia analyzed; error bars indicate 95% confidence intervals). (H) Volcano plot of cilia with total IC or (partial) EC localization (non-IC) in Actr3 KO-1&-2 cells (N = 113 IC and 499 non-IC cilia; n.s., not significant).

ARP3-dependent regulation of cilia reprograms the transcriptome of kidney tubule cells

One of the essential functions of the primary (sensory) cilium is the regulation of transcriptional programs and cellular phenotypes (44–46). Therefore, a complementary transcriptome analysis was performed, including kidney medullas of Actr3 mice at postnatal day 14 (P14), Actr3 mIMCD3 cells (starvation versus nonstarvation), Actr3 KO Ift88 KO (dKO), and ciliobrevin D–treated Actr3 KO cells (Fig. 8A). In addition, a reference dataset for ciliary-regulated genes was curated by analysis of previously published transcriptome data derived from Pkd1 KO, Pkd2 KO, Ift140 KO, and Dync2h1 KO mIMCD3 cell lines (47, 48). Here, differential gene expression analysis of WT and Actr3 KO cell lines revealed 7685 significantly regulated genes in nonstarved cells (adjusted P value <0.05; 1491 genes with absolute log₂ fold change >1) and 9489 significantly regulated genes in starved cells (adjusted P value <0.05; 1513 genes with absolute a log₂ fold change >1) (Fig. 8B). An intersection analysis of significantly regulated genes with reference datasets for cilia-dependent regulated genes demonstrated coregulation for the majority (in at least one reference dataset; Fig. 8C). To further delineate the direct contribution of ciliary signaling for transcriptional alterations in Actr3 KO cells, an analysis of Actr3 KO Ift88 KO (dKO) and ciliobrevin D–treated Actr3 KO cells was performed (Fig. 8D). Here, differential gene expression analysis comparing Actr3 KO with dKO cells indicated 8865 significantly regulated genes (adjusted P value <0.05; 1152 genes with absolute log₂ fold change >1), and comparison of Actr3 KO with ciliobrevin D–treated Actr3 KO cells revealed 8826 significantly regulated genes (adjusted P value <0.05; 1162 genes with absolute log₂ fold change >1). Most of these genes showed an inverse regulation of the Actr3 KO phenotype (2325 inversely regulated genes in one and 2315 in both dKO and ciliobrevin D–treated Actr3 KO cell datasets) (Fig. 8E and fig. S10A). Gene set enrichment analysis (GSEA) for gene ontology (GO) terms of strongly regulated genes with matching (overlapping and rescued) regulation in all transcriptome datasets showed an overrepresentation of genes associated with organelle function (e.g., mitochondria) but also with the cell membrane and even the EC compartment (Fig. 8, F and G). Investigating the regulation of ciliome genes showed significant regulation of 30 ciliome genes in Actr3 KO cells (Fig. 8H). However, these genes showed an inverse regulation in dKO or ciliobrevin D–treated Actr3 KO cells. These results indicate that the regulation of ciliary gene expression is due to secondary transcriptional regulation at the level of already altered cilia length rather than directly resulting from ARP3 loss. Next, we aimed to transfer these findings to the in vivo situation. Differential gene expression analysis of the kidney medulla of WT and Actr3 KO mice at P14 demonstrated significant regulation of 7789 genes (adjusted P value <0.05; 2976 genes with absolute log₂ fold change >1) (Fig. 8I). Using these data, intersection analysis of significantly regulated genes between in vivo and in vitro datasets demonstrated consistent regulations for 3428 genes with reduced expression and 3250 genes with increased expression levels as a result of Actr3 deletion (Fig. 8J). Given these large alterations at the transcriptome level, we reasoned that the underlying transcriptional programs might also be affected. To investigate this hypothesis, we filtered for highly regulated transcription factors (TFs) and co-TFs with consistent regulation across all datasets (Fig. 8, J and K). This analysis led to the identification of several highly regulated TFs (Lhx1, Nfatc4, Sox8, Aebp1, Tbx1, Cited2, and Fhl2) with known roles in nephrogenesis. These transcriptional alterations were further validated by IF analysis of LHX1 and CITED2 expression in distal tubules of Actr3 KO mice (Fig. 8, L to M).

Cilia elongation in distal renal tubules is associated with an adaptive epithelial cell state

The observation of regulated TFs essential for the development and differentiation of tubule cells indicated that elongated cilia might be involved in the induction of reactive tubular programs in disease (Fig. 8). To test this hypothesis, a GSEA of reactive distal tubule cell states was conducted using gene sets that have recently been defined in human renal disease (29). This analysis indicated a coenrichment of adaptive, cycling, and degenerative cell states in the medulla of Actr3 KO animals (Fig. 9A). Actr3 KO cells showed a significant decrease in genes associated with a degenerative cell state, whereas enrichment for genes associated with an adaptive cell state, such as Itgb6, was observed (fig. S10B) (29, 49). Moreover, cilia depletion in Actr3 KO cells by additional KO of Ift88 (dKO cells) or treatment with ciliobrevin D led to a significant enrichment of gene sets associated with a degenerative cell state. Analysis of the adaptive cell state marker ITGB6 in distal tubules of WT control and proteinuric Nphs2^R321Q/A285V mice demonstrated high expression in proteinuric animals and significant enrichment in tubules with elongated cilia (Fig. 9, B and C). Despite the activation of adaptive programs in injured renal tubule cells, persistent stress (including proteinuric glomerular disease) might eventually progress to chronic renal injury (CKD). One pathological hallmark of this process is the thickening of the tubular basement membrane (TBM) and accumulation of ECM proteins, which are atypical for TBMs (Fig. 9D). Analysis of Actr3 KO mice demonstrated marked deposition of fibrotic ECM (e.g., collagen) and thickening of TBMs (Figs. 4 and 9D, and figs. S5 and S11). Further GSEA for matrisome genes showed a profound enrichment of matrisome gene sets in Actr3 KO mice and cells (Fig. 9E). Again, many of the regulated matrisome genes showed an inverse regulation upon cilia depletion in Actr3 KO cells (63 inversely regulated genes in one and 71 in both dKO and ciliobrevin D–treated Actr3 KO cells) (Fig. 9F). To investigate the functional consequences of these findings for ECM dynamics, WT and Actr3 KO cells were cultured on surfaces precoated with basement membrane extracts, which allowed for both deposition and active remodeling of preexisting ECM proteins (Fig. 9G and fig. S12). This analysis showed a reduction of TBM components (e.g., COL4A1, LAMA1, and NTN4) that might be (partially) mediated by increased degradation (e.g., via MMP14 and MMP19) but also by an enrichment of fibrosis-associated matrisome components (e.g., FN1 and TNC) (50). Further analysis of Tenascin-C (Tnc) demonstrated an increased expression in Actr3 KO cells and a decreased expression in dKO and ciliobrevin D–treated Actr3 KO cells (Fig. 9, G to J, and fig. S13, A and B). These findings were confirmed by a markedly elevated expression of TNC in CD tubules and interstitial cells of proteinuric Nphs2^R321Q/A285V mice (Fig. 9, K and L). Moreover, the expression of TNC is increased in the tubulointerstitial compartment of patients with glomerular diseases, including FSGS, IgAN, diabetic nephropathy, and lupus nephritis (fig. S13, C to F). Functional analysis demonstrated that TNC resulted in reduced adhesive properties in comparison to basement membrane proteins. This translated into an increased susceptibility to laminar flow–induced cell detachment and an elevated cell proliferation rate (Fig. 9, M to Q, and fig. S13, G to J). In conclusion, cilia-dependent signaling in renal tubule cells appears to be associated with a favorable, adaptive epithelial cell state. Concomitantly, however, maladaptive response patterns can also be initiated, as illustrated by the remodeling of the TBM via the regulation of matrisome proteins, such as Tenascin-C.

Fig. 9. Cilia elongation in distal renal tubules is associated with an adaptive epithelial cell state.

(A) GSEA for adaptive, cycling, or degenerative tubule cell states. Bubble plots indicate normalized enrichment scores (NES) (Actr3 KO versus WT mice (brown), nonstarved (green) and starved (blue) Actr3 KO versus Ctrl. cells, dKO (Actr3 KO-1 Ift88 KO), and ciliobrevin D (CB)–treated KO-1 versus KO-1). (B and C) Analysis of the adaptive marker ITGB6 (cilia stained by acetyl–α-tubulin; tubules with elongated (arrows) and nonelongated cilia (arrowheads), casts (asterisks); N = 3 WT and N = 5 Nphs2^R231Q/A286V animals; dots indicate mean per animal; dashed lines indicate the same animal; ***P < 0.001). (D) Silver staining of distal tubule basement membranes (TBM) in CKD caused by FSGS and in Actr3 KO mice (arrows indicate thickened TBMs). (E) GSEA of matrisome gene sets (NES for the indicated conditions; n.s., not significant). (F) Intersection analysis of matrisome genes with matching regulation in indicated datasets (adjusted P value <0.05; coregulation in Actr3 KO and inverse regulation by dKO and CB). (G to J) Analysis of ECM dynamics in Actr3 cells cultured on Matrigel substrates (mean fluorescence intensity, MFI; N = 30 cells per condition; n.s., not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). (I) Tnc expression in indicated datasets. (J) Representative Tenascin-C (TNC) and phalloidin (F-Actin)–stained cells. (K and L) Analysis of DBA-positive tubules show TNC negative, low, or high expressing cells (arrows indicate cilia and asterisks interstitial cells; N = 3 WT and N = 5 Nphs2^R231Q/A286V animals; dots indicate mean per animal; dashed lines indicate the same animal; *P < 0.05, ***P < 0.001, ****P < 0.0001). (M to Q) Analysis of cell size and adherent cells cultured on basement membrane (BM) Matrigel or TNC-coated surfaces under no-flow and laminar-flow conditions (N = 3 independent experiments; dots indicate mean per experiment; error bars indicate means and S.E.M.; n.s., not significant, **P < 0.01).

DISCUSSION

While the role of primary cilia as central signaling hubs in hereditary renal tubular disorders (e.g., ADPKD) is well established (13, 22), their involvement in secondary adaptive processes has remained largely unexplored. On the basis of recent advances, our understanding of kidney disease has fundamentally changed, as tubular injury is now classified as a continuous spectrum of adaptive response patterns ranging from regenerating to maladaptive/dedifferentiating patterns of tubular epithelial cell states (26). However, whether and how such adaptive programs correspond to morphological alterations (including cilia) and their role in proteinuric glomerular disease remains to be elucidated.

In this study, we focused on renal pathologies characterized by overt glomerular proteinuria (including FSGS and IgAN) and detected significant cilia elongation in distal tubules showing early signs of injury (Fig. 1). This elongation occurred predominantly in tubules with mild damage (often accompanied with formation of intratubular precipitates/casts), whereas severely injured tubules exhibited shortening and loss of cilia. Similar observations of a transient elongation in acute injury followed by normalization of ciliary length during recovery have also been seen in models of ischemic renal injury and in patient biopsies (7, 51). Furthermore, analysis of murine disease models revealed a correlation between the dilation of renal tubules, the presence of intratubular proteinaceous casts, and an increase in ciliary length that may be mediated by mechanophysical cues (Figs. 1 and 2). While a potential receptor mediating ciliary elongation is not identified so far, our findings implicate aberrant distribution of cortical branched actin (Fig. 2). Together, these observations suggest a dynamic ciliary response that corresponds to the degree of tubular injury and potential alterations in mechanosignaling.

Numerous studies have addressed molecular mechanisms controlling ciliary length (21, 46). While the cilium is primarily a microtubule-based organelle, elegant work has demonstrated that coordinated interplay between microtubule and actin cytoskeleton dynamics is essential (52). On the basis of the initial discovery in a functional genomic screen, it is now widely accepted that the branched actin cytoskeleton generated by the Arp2/3 complex plays a pivotal role in controlling ciliary length and ciliogenesis (16). However, our current knowledge largely relies on in vitro studies using either genetic titration (small interfering RNA/short hairpin RNA) or the use of inhibitory molecules, which bear the potential for exerting off-target effects (53, 54), incomplete inhibition of the Arp2/3 complex, and in vitro artifacts. By using a combination of in vitro and in vivo KO models, we demonstrate that a deficient N-WASP–Arp2/3 machinery results in the elongation of cilia (Figs. 3, 4, and 5). A plethora of explanatory models for ciliary length control in response to modulation of the actin cytoskeleton (and additional nonactin cytoskeleton related modulators) have been described (17, 18, 21). Some studies indicated that branched F-actin networks might function as physical barriers at the ciliary base, thereby limiting trafficking of ciliary cargo (16, 55). Moreover, the contractile actomyosin network has been shown to closely interact with the branched actin cytoskeleton, and deletion studies (e.g., of Myh10) indicated a pivotal role for this cellular machinery for efficient ciliogenesis (potentially via basal body positioning toward the apical membrane) (56). Here, we investigated exemplary pathways and finally observed that IFT dynamics and dynein function appear as a prerequisite for ciliary lengthening in Arp2/3-deficient cells (Fig. 6). Through detailed ultrastructural analysis, we observed frequent formation of membrane invaginations resembling ciliary pockets, as well as partial and completely internalized cilia (Fig. 7). This phenotype not only indicates a significant displacement of the basal body from the plasma membrane but can also be interpreted as a switch in the mode of ciliogenesis to the IC pathway (15). The mechanism underlying this phenotype appears to be linked to branched actin filament formation, as pharmacological disruption of actin polymerization produced similar effects. A previous study observed IC ciliary axoneme formation caused by knockdown of the cortical actin-binding protein Ezrin in multiciliated Xenopus cells (57). Here, the intertwined regulation between planar cell polarity (PCP) signaling and the ELMO-DOCK1-Rac1-Ezrin axis regulating cortical actin remodeling and basal body migration/docking have evolved as an explanatory model. Moreover, impaired signaling of polarity proteins was linked to cilia elongation via the aPKC-Par6-Cdc42-N-WASP-Arp2/3 axis (39). Our observations of disturbed apical cortical actin networks in conditions of proteinuria (Fig. 2) indicate a potential involvement of aberrant PCP in impairing actin branching. Collectively, these results demonstrate the pivotal role of the Arp2/3 complex not only for ciliary length control but also for ciliary assembly modes.

Despite previous studies suggesting that increased ciliary length as well as the ciliary pocket translate into altered transcriptional signatures, e.g., via Hedgehog signaling, the direct effect of Actr3-dependent ciliary elongation is not known (39, 58). Through comprehensive transcriptome analysis of Actr3 KO models, we identified extensive changes in gene expression that largely overlapped with ciliary-regulated transcriptional programs (Figs. 8 and 9). Notably, we found that ciliary elongation is associated with transcriptional programs projected to a regenerative adaptive epithelial cell state. This was exemplified by increased ITGB6 expression, which has recently been demonstrated to correspond to acute-epithelial response states in conditions of acute kidney injury (29, 59, 60). Moreover, we observed a correlation between ITGB6 expression and concomitant cilia lengthening in a bona fide proteinuric mouse model carrying causative mutations for human FSGS (Fig. 9). These findings further substantiated our hypothesis that cilia lengthening is an important characteristic of the cellular response to distal tubular injury caused by FSGS or IgAN (Fig. 1). Beyond adaptive gene signatures, we also identified prominent alterations of matrisome proteins, including up-regulation of Tenascin-C (TNC) (Fig. 9). TNC is an ECM glycoprotein well known to play a critical role in renal fibrosis (61, 62), and more recent work described the potential use of TNC as a biomarker indicating acute tubular injury (63). Intriguingly, coexpression of ITGB6 and TNC may establish an autocrine signaling pathway in injured tubule cells, promoting epithelial dedifferentiation (61, 64). Our observations suggest reduced adhesive and increased proliferative properties of TNC to tubular epithelial cells. Analogous functional observations have been made in renal fibroblasts, the main source of TNC in the kidney interstitium (63). Thus, TNC may stimulate regenerative cell proliferation, but in the long term, it may also promote disease progression due to epithelial cell dedifferentiation and loss.

By using conditional deletion approaches, others and we have previously demonstrated a cell and tissue type–specific functionality for the N-WASP and Arp2/3 machinery (65–69). Our current observations of ciliary elongation and concomitant progressive renal injury in respective Wasl and Actr3 KO mice and cells not only further expand this growing notion of a context-dependent role for the central Arp2/3 complex but also suggest a complex dual role for Actr3-dependent ciliary elongation in the context of tubular injury (Fig. 10). Initially functioning as an adaptive response, prolonged ciliary alterations may contribute to progressive tubular dysfunction through dysregulation of the ECM composition within the TBM. This mechanism offers a potential explanation for the transition from adaptive to maladaptive responses in the progression of CKD.

Fig. 10. Graphic summary of the proposed role of cilia elongation in distal renal tubules.

(A) Proteinuric glomerular diseases cause cilia elongation and thickening of the TBM in the distal tubules of the nephron. (B) Proteinuric glomerular filtrate influences the N-WASP/WAVE – Arp2/3 complex–dependent cortical branched actin cytoskeleton, inducing cilia elongation and a shift in the mode of ciliogenesis. Subsequent ciliary signaling alters transcriptional programs to induce an adaptive epithelial cell state in response to tubular injury. This includes altered expression of TFs such as LHX1, matrisome genes such as TNC and markers of an adaptive cell state such as ITGB6. However, prolonged activation of these programs induces features of a degenerative cell state, including pathological remodeling of the TBM and activation of dedifferentiating signaling cascades, e.g. via TNC-ITGB6.

MATERIALS AND METHODS

Animals

Conditional Wasl and Actr3 KO mice were generated by crossing of Wasl^fl/fl or Actr3^fl/fl mice with a distal tubule–specific Cdh16-Cre [B6.Cg-Tg(Cdh16-cre)91Igr/J] line. The loxP Actr3 and Wasl alleles have been previously described (65, 67, 69). The Gt(ROSA)26Sortm4(ACTB-tdTomato,-EGFP)Luo/J (mT/mG) reporter strain was obtained from JAX mice. Mice were maintained on a C57BL/6 mixed background. Serum urea (blood urea nitrogen) levels were quantified using an enzymatic kit (LT-UR 0010, Labor & Technik, Eberhard Lehmann GmbH, Germany). All animal experiments were conducted in accordance with the German legislation pertaining to the welfare of animals and the National Institutes of Health guide for the use of laboratory animals. All studies using Actr3 or Wasl mice were approved by the Regierungspräsidium Freiburg (G-17/02), Germany.

Animal models for glomerular disease have been previously described. These include the following mouse models: Parva^fl/fl*Nphs2-Cre (12 weeks old) (34), Nphs2^R231Q/A286V (15 to 20 weeks old) (32), Pdss2^V177M/V177M (20 to 26 weeks old), Col4a3 KO (8 weeks old), and NOH [nephrin, chicken ovalbumin (OVA), and hen egg lysozyme with OT-1 cell transfer and LM-OVA infection for 1 week] (36). NOH mice were previously described (36). All studies using NOH mice were approved by the Regierungspräsidium Freiburg (G21/021), Germany.

Parva^fl/fl*Nphs2-Cre mice were previously described (34). Parva^fl/fl mice were provided by E. Montanez, University of Barcelona, Spain. Urinary albumin to creatinine ratio was determined by albumin enzyme-linked immunosorbent assay (ELISA, ab108792, Abcam) and enzymatic creatinine measurement (creatinine, PAP LT-SYS LT-CR 0106, Labor & Technik, Eberhard Lehmann). All studies using Parva mice were approved by the Regierungspräsidium Freiburg (G-17/127). Germany.

Nphs2^R231Q/A286V mice were previously described (32). Urine samples were analyzed using albumin ELISA (mouse albumin ELISA kit, Bethyl Labs) and a creatinine assay (Cayman Chemical). All studies using Nphs2 mice were approved by the State Office of North Rhine-Westphalia, Department of Nature, Environment and Consumer Protection (LANUV NRW, Germany; animal approval AZ 81-02.04.2018.A325 and AZ 84-02_04_2014_A372).

Pdss2^V117M mice were obtained from the Brinkkötter Laboratory, where they were generated via CRISPR as previously described, resembling kd/kd mice (33, 70). Briefly, C57BL/6N zygotes from superovulated females were electroporated at 0.5 dpc with Cas9 ribonucleoprotein complexes [Cas9 protein, guide RNA (gRNA) (TTGATATGAGGAGCACGACC), and ssODN repair template] to introduce the V117M mutation and a silent NlaIII restriction site, cultured to the two-cell stage, transferred to pseudopregnant recipients, and mutation confirmed by Sanger sequencing. Urinary albumin to creatinine ratios were determined by ELISA (mouse albumin ELISA Kit, Bethyl Laboratories Inc., Montgomery, TX, USA) and Colorimetric Assay (Creatinine Colorimetric Assay Kit, Cayman Chemical, Ann Arbor, MI, USA) according to the manufacturer’s instructions. All studies using Pdss2 mice were approved by the Landesamt für Natur, Umwelt und Verbraucherschutz Nordrhein-Westfalen (LANUV) (81-02.04.2019.A049), Germany.

Col4A3 KO (Alport) mouse were obtained from J. Miner, Washington University in St. Louis and maintained on a 129/Sv background (35). Three female and three male Col4A3 KO and WT mice were analyzed. Urinary albumin measurements are based on ELISA, and creatinine measurements are based on photometric measurement (Siemens Atellica solution CH module, Siemens Healthineers). Urine samples were analyzed at Analysezentrum, Universitätsklinikum Heidelberg. All studies using Col4A3 mice were approved by the Regierungspräsidium Karlsruhe (G-49/23 and G-50/23), Germany.

Human kidney tissue

The analysis of human kidney biopsy samples with established diagnoses of FSGS or IgAN and the usage of normal kidney tissue from cancer nephrectomies for healthy tissue controls was conducted on the basis of informed consent and approved by the ethics committee of the University Medical Center Freiburg (EK 21/1288; EK 18/512).

Cell culture

Murine CD cells [mIMCD-3, CRL-2123, American Type Culture Collection (ATCC)] and a previously validated monoclonal mIMCD-3 cell line (“WT clone-5”) that was used for CRISPR-Cas9 genome editing were generously provided by M. Köttgen (Department of Medicine IV, Medical Centre, University of Freiburg, Freiburg, Germany) (71). Human RPTECs (RPTEC/TERT1, CRL-4031, ATCC) were used as model of proximal tubule cells. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (10565018, Thermo Fisher Scientific) supplemented with 10% FCS (S0615, Sigma-Aldrich), 5 mM Hepes (15630-056, Thermo Fisher Scientific), and penicillin-streptomycin (15140122, Thermo Fisher Scientific) in accordance with standard procedures at 37°C, 95% air and 5% CO₂. Human embryonic kidney (HEK) 293T cells (CRL-3216, ATCC) were cultured in DMEM (31966021, Thermo Fisher Scientific) supplemented with 10% FCS. The absence of mycoplasma contamination was confirmed through the use of a polymerase chain reaction (PCR)–based detection kit (Mycoplasma PCR Detection Kit, Hiss Diagnostics GmbH, Germany) and by Hoechst 33258 staining for all cell lines, including those that had undergone genetic modification.

For the purpose of analyzing cilia, cells were seeded in ibiTreat eight-well chamber slides (80806, Ibidi) and allowed to grow for 48 hours until they reached confluence. Subsequently, mIMCD3 cells were cultured in starvation medium devoid of FCS for an additional 24 hours. The following inhibitors and reagents were used: 0.1 or 1 μM cytochalasin D (C8273, Merck), 10 μM Y-27632 (S1049, Selleck Chemicals), 0.5 μM ciliobrevin D (250401, Merck), bovine albumin (10 μg/μl; 11926.03, SERVA), basement membrane ECM extract [100 μg/ml (low) or 1 mg/ml (high) growth factor–reduced Matrigel, 356231, Corning], methylcellulose (150 or 1500 cP in culture medium, M0387, Merck), and freshly collected human serum (10% in culture medium without FCS). For the analysis of cells under laminar flow, the cells were cultured in channel slides (80196, Ibidi) in accordance with the aforementioned methodology for cilia generation on eight-well chamber slides. A laminar flow system (10902, Ibidi) was used to generate defined flow rates of 0.2 or 2 dyn/cm² for 24 hours. For experiments comparing cells on 2 μg/cm² TNC (CC065, Sigma-Aldrich) or 2 μg/cm² Matrigel-coated channel slides, a laminar flow of 3 dyn/cm² was applied. Therefore, cells were precultured on these coatings for 12 hours, and laminar flow was applied for an additional 12 hours.

Antibodies

Antibodies used in this study are described in detail in data S1.

CRISPR-Cas9 genome editing and plasmid expression

For generation of WT control, Actr3 KO as well as Actr3 Ift88 dKO cells, CRISPR-Cas9 genome editing was applied using a previously validated monoclonal mIMCD-3 cell line (WT clone-5) (71). Generation of cell lines was performed by lentiviral transduction with a doxycycline-inducible CRISPR-Cas9 (TLCV2-LoxP) system followed by creation of monoclonal cell lines by limiting dilution. The TLCV2-LoxP plasmid was a gift from A. Karpf (Addgene, plasmid #127098; http://n2t.net/addgene:127098; RRID:Addgene_127098). The following gRNAs were used for the respective cell lines: WT control 1 (Ctrl.-1) nontargeting–gRNA-1: 5′- GCGGGCAGAACGACCCTGAC -3′; WT control 2 (Ctrl.-2) nontargeting–gRNA-2: 5′- GAAGACGTGCTGGCGTCACC -3′; Actr3 KO-1 (KO-1) Actr3-gRNAs: 5′- TAATAGTGGCCAATTCGCCA(TGG) -3′ and 5′- TAACCTTGTAGAGAGGACGC(CGG) -3′; Actr3 KO-2 (KO-2) Actr3-gRNAs: 5′- ACAGATAAATTTACGAACCG(TGG) -3′ and 5′- GTAGGAGAGCGGACGCTGAC(GGG) -3′; Actr3-Ift88 dKO using Actr3 KO-1 as background cell line, Ift88 gRNA: AAACCTAGATGCCTCCTACG(TGG) and nontargeting–gRNA-1: 5′- GCGGGCAGAACGACCCTGAC -3′ for generation of Actr3 KO Ift88 WT (KO-1 Ctrl.) cells (protospacer adjacent motif sequences are indicated by brackets). The pLVX-sspH-Smo-GFP reporter plasmid was a gift by D. Toomre (Yale School of Medicine, New Haven, USA) (43). pLVX-sspH-Smo-GFP reporter cell lines were generated by lentiviral transduction. Lentiviral particles were produced by transfection of HEK293T cells with respective expression and packaging (pMD2.G and pCMV-dR8.91) plasmids. pCMVR8.91 and pMD2.G (Addgene, plasmid # 12259; http://n2t.net/addgene:12259; RRID:Addgene 12259) were a gift from D. Trono.

Western blot analysis

Cell lysis in radioimmunoprecipitation assay buffer, protein quantification, denaturation in Laemmli buffer, and SDS–polyacrylamide gel electrophoresis Western blot analysis were conducted as previously described (34). The isolation of soluble and polymerized tubulin fractions was conducted as previously described (40). Isolated fractions were subjected to the same processing as that applied to standard cell lysates.

Histology and histological assessment

The processing of kidney samples for FFPE (formalin-fixed paraffin-embedded) tissue, 2 μm microtome sectioning, PAS (periodic acid, 100524, Merck; Schiff-Reagens, 3952016, Sigma-Aldrich; Hematoxylin Gill III, 3801540, Leica Biosystems), silver-methenamine, and Acid Fuchsin Orange G (AFOG) staining were conducted in accordance with standardized procedures for medical diagnostics at the Institute of Pathology, Faculty of Medicine, University of Freiburg. Murine kidneys were fixed by perfusion via the A. renalis with 4% paraformaldehyde (PFA) diluted in phosphate-buffered saline (PBS, 15714-S, Electron Microscopy Sciences), followed by immersion in fixation buffer for 24 hours and FFPE embedding following standard procedures. Assessment of structural kidney phenotypes on whole slide images was conducted by applying semiquantitative scores for glomerular sclerosis, interstitial fibrosis and tubular atrophy, and interstitial immune cell infiltrates (inflammation), as previously described (34). The cystic index was calculated as the ratio of the cyst area to the total kidney area. The cystic kidney area was estimated from whole slide images of kidney sections using QuPath 0.4.3 (72), applying a pixel classifier for thresholding of dilated tubule lumina. Fibrosis was estimated from AFOG staining by thresholding of fibrotic regions (blue staining) using FIJI ImageJ. The number of medullary tubules showing tubular dilation or proteinaceous casts was quantified from PAS-stained kidney sections. A representative region of 1 mm² from the medullary compartment was analyzed for each animal and parameter.

IF staining on tissue sections

For immunostaining, human and murine 2-μm tissue sections were deparaffinized, and heat-induced antigen retrieval (HIAR) was performed using a pressure cooker with citrate buffer (pH6) or tris-EDTA buffer (pH 9). For immunohistochemistry (IHC) staining, an additional peroxidase blocking step was performed using 3% H₂O₂ after HIAR. Tissue sections were subsequently incubated in blocking buffer [5% bovine serum albumin (BSA), 11926.03, Serva] in PBS 7.2 (PBS) for 1 hour. Primary antibodies were diluted in the aforementioned blocking buffer and incubated at 4°C for 24 hours. Samples were washed in PBS, and secondary antibodies, either fluorophore-tagged (Alexa Fluor dyes, Thermo Fisher Scientific) or horseradish peroxidase–tagged (Dako), were applied for a period of 45 min. Cell nuclei were stained using Hoechst 33342 (for fluorescence) or hematoxylin (for IHC). The 7-plex IF staining was performed using a TSA-based detection system (OPAL, NEL861001KT, Akoya Biosciences), following the instructions provided by the supplier. Stained kidney sections were imaged and analyzed using QuPath 0.4.3. Therefore, the kidney cortex and medulla were annotated as regions of interest for subsequent analysis. The positive areas for COL1A1 and SMA were segmented by thresholding. To quantify immune cell populations, cells were segmented using the QuPath-integrated cell segmentation tool. Cell classifiers for stained immune cell markers were created and used to classify segmented cells into distinct immune cell populations.

Cell IF

For IF analysis, cells were fixed in 4% PFA in PBS with 1 mM CaCl₂/MgCl₂ for 20 min. Following fixation, cells were washed in PBS, permeabilized with 0.1% Triton X-100 (T8787, Merck), and blocked with 5% BSA in PBS. Primary antibodies were diluted in blocking buffer and incubated on the cells at 4°C for 24 hours. Subsequently, cells were washed in PBS, and secondary fluorophore antibodies (Alexa Fluor dyes, Thermo Fisher Scientific) were applied for 45 min. The cell nuclei and actin filaments (F-Actin) were costained using Hoechst 33342 (H3569, Thermo Fisher Scientific) and phalloidin (Thermo Fisher Scientific), respectively. Triton X-100 permeabilization was not used in the analysis of pLVX-sspH-Smo-GFP–expressing RPE1 cells. Instead, fluorescence (PE)–labeled anti-GFP antibodies were used for the staining of the EC portion of SMO-GFP.

Microscopy

An inverted Zeiss Axio Observer microscope (Carl Zeiss AG) was used for fluorescence imaging. The microscope was equipped with the following components: Colibri 7 illumination system, Axiocam 702 mono camera, ApoTome.2 device for focal imaging, different filter sets (49 4′,6-diamidino-2-phenylindole, 38 GFP, 43 HE dsRed, and 50 Cy5), objective lenses (10×, 20×, 40×, 100×) and the ZEN 3 software. An automated scanning stage was used for the digitalization of entire kidney sections. Maximum intensity projections of z-stack images were generated using the ZEN software (Carl Zeiss AG). A Ventana DP 200 slide scanner (Roche Diagnostics, Germany) was used for the digitalization of histological stainings. The 7-plex IF stainings were imaged using an Akoya Fusion microscope (Akoya Biosciences Inc., USA).

Electron microscopy

Processing of kidney samples for transmission electron microscopy and analysis of podocyte foot process width was performed as previously described (34). For SEM analysis, mIMCD3 cells were cultured on glass-bottom dishes with an imprinted location grid (81148, Ibidi), and ciliogenesis was induced by applying the aforementioned protocol. Cells were subsequently immersion fixed in a solution of 4% PFA and 1% GA (glutaraldehyde, 4157.2, Carl Roth) in PBS. For CLEM, cells were fixed and IF stained in accordance with the aforementioned protocol for IF analysis of cilia, including permeabilization with Triton X-100. Following the acquisition of localized 100× z-stack images, the cells were immersion fixed in 4% PFA and 1% GA in PBS. Fixed samples were dehydrated through a series of ascending concentrations of ethanol for 15 min each (70, 80, 90, and 100%). After incubation in 50:50 ethanol/hexamethyldisilazane (CAS-999-97-3; Sigma-Aldrich) and 100% hexamethyldisilazane, the solvent was allowed to evaporate. All samples were coated with platinum using an electron microscopy ACE600 Sputter-Coater (Leica). Samples were imaged using a Quattro SEM (Thermo Fisher Scientific). The imprinted grid on the glass-bottom dish was used for localization of overlapping regions for CLEM. The TrakEM2 tool implemented in FIJI ImageJ v1.52 was used for nonrigid registration of SEM and IF images. Electron microscopy was conducted at the EM core facility of the Department of Nephrology, Faculty of Medicine, University of Freiburg [EMcore (RI_00555)].

Cilia assessment

The assessment of cilia in mIMCD3 cells and kidney tubules was conducted by acetyl–α-tubulin staining. Z-stack images of randomly selected regions were captured and converted to maximum intensity projections. Dolichos biflorus agglutinin– and NKCC2 [Na-K-2Cl Cotransporter 2 (SLC12A1)]-positive tubules were selected for analysis of cilia in human and murine kidney samples. Moreover, NCC-positive tubules were examined in Actr3 KO and WT animals. Given the considerable intersubject variability in the ratio of damaged to normal nephrons observed in human kidney biopsies of patients with FSGS and IgAN, tubules exhibiting evidence of tubular damage (e.g., presence of tubular casts, tubular dilation, epithelial flattening, and epithelial cell loss) were analyzed. Cilia length and number were quantified from maximum intensity projections using QuPath v0.4.3. A minimum of 100 cilia (with the exception of Fig. 5K) or ciliation of at least 100 cells per sample and replicate were measured. The diameter of the tubule lumen was measured in murine disease models, and the corresponding mean cilia length per tubule was calculated. To analyze the localization of IFT88 and IFT81 to cilia, the entire surface area of the cilium was manually masked and used as an analysis region. The area positive for IFT88 and IFT81 was determined by thresholding within this analysis mask (27 to 65 cilia per animal were analyzed for IFT88- and IFT81-positive area). The IC and EC localization of the ciliary shaft was analyzed using a pLVX-sspH-Smo-GFP reporter plasmid and anti-GFP IF staining, as previously described (43). The IC and EC length of the cilia shaft was measured for a minimum of 100 cilia per replicate. The ratio of the EC to IC length was calculated for each analyzed cilium. Unless otherwise indicated, the mean value per replicate and sample was calculated and used for statistical analysis.

Simulation of cilia datasets

To test potential models for interdependent or independent regulation of cilia elongation and IC location of the axoneme in Actr3 KO-1&-2 cells, representative datasets for selected models were simulated. For PIC cilia, total cilia length (TCL) and the relative distance of the basal body to the plasma membrane, measured experimentally as RICAL (IC shaft length/total cilia shaft length), were identified as relevant variables for the observed phenotypes in KO cells. Three types of interdependent regulation of TCL and IC translocation were hypothesized: First, a positive correlation between TCL and RICAL (simulation A), second, a negative correlation between TCL and RICAL (simulation B), or no correlation between TCL and RICAL (simulation C). Simulation A refers to explanatory models where the increased EC cilia length leads to a secondary translocation of the basal body (e.g., mechanical pushback) or where the IC axoneme extends until it reaches the plasma membrane/EC space. Simulation B refers to explanatory models where the IC localization of the basal body/axoneme inhibits cilia extension until transport to the cell cortex is complete. Simulation C refers to models in which cilia length control and the IC position of the basal body/axoneme are independently regulated features in Actr3 KO cells. The GraphPad Prism 8 software was used to generate simulated datasets, assuming a linear correlation between TCL and RICAL (described by Y = YIntercept + X*Slope, where TCL is the X and RICAL the Y variable). To simulate experimental results, the TCLs of 125 quantified PIC cilia in KO-1&-2 cells were used to calculate corresponding values for the RICAL in simulations A, B, and C. An absolute Gaussian error was applied to the dataset using the SD of the RICAL values from KO-1&-2 cilia and an outlier rate of 5%. RICAL values were constrained to an interval of (0.05, 0.95) to match the experimental resolution of measurements in PIC cilia (RICAL = 1 represents fully IC cilia and RICAL = 0 fully EC cilia. The parameters for generation of simulated data by a linear model were adjusted so that the mean of simulated RICALs approximately matched the mean of real RICAL values of KO-1&-2 cilia. Last, linear models and 95% confidence intervals were calculated for simulated and real datasets by linear interpolation.

CTTN and ECM analysis

To analyze CTTN localization in methylcellulose-treated or Actr3 WT and KO cells, the cells were cultured in ibiTreat eight-well chamber slides for 24 hours. The cells were cultivated under sparse conditions in cell culture medium supplanted with 10% FCS. Subsequently, cells were IF stained, and single cells were imaged. CTTN was used as a marker for cortical branched actin networks. For the purpose of quantification, the perimeter of the cells and the proportion of the perimeter that was positive for CTTN were measured. For analysis of ECM remodeling and deposition, the cells were seeded in eight-well chamber slides (80807, Ibidi) that had been precoated with basement membrane ECM extract (100 μg/ml; Matrigel, 356231, Corning). The processing of the subcellular ECM (deposition, substrate degradation, and remodeling) was analyzed using QuPath v0.4.3. ECM processing of single cells was imaged with the focus plane set on the basal ECM layer using the Apotome function of the microscope and a 100× objective. Single cells were segmented on the basis of phalloidin (F-actin) staining, and the mean fluorescence intensity of ECM markers was measured within the segmented cell mask. For experiments comparing TNC (CC065, Sigma-Aldrich) with Matrigel, eight-well chamber slides (80807, Ibidi) were coated with TNC (2 μg/cm²) or Matrigel, respectively.

RNA sequencing

Cells were seeded on six-well cell culture dishes (657960, Greiner Bio-One) and allowed to grow in accordance with the aforementioned protocol for cilia induction on chamber slides. Total RNA was isolated using a kit (T2010S, New England Biolabs Inc.) in accordance with the supplier’s recommendations, including on-column deoxyribonuclease I treatment. The sample quality was assessed using an Agilent Fragment Analyzer, with RQN values ranging from 9.8 to 10. Samples processing, including poly(A) mRNA selection, library preparation, and finally, Illumina NovaSeq 6000 2 × 150–base pair paired-end sequencing, was performed by the GENEWIZ Germany GmbH. Nonstarved mIMCD3 Actr3 Ctrl.-1, Ctrl.-2, KO-1, and KO-2 cells were analyzed in two independent replicates, resulting in a total of eight samples. Starved mIMCD3 Actr3 Ctrl.-1, KO-1 Ctrl., Actr3-KO Ift88-KO (dKO), and 0.5 μM ciliobrevin D–treated KO-1 Ctrl. cells were analyzed in three independent replicates, resulting in a total of 12 samples. Isolation of total RNA from the dissected medulla of two Actr3 KO and two Actr3 WT animals at P14 was performed using the Qiagen RNAeasy midi Kit (74104, Qiagen). Illumina sequencing of mice mRNA was conducted by Eurofins Genomics Germany GmbH. The Galaxy Europe platform (73) was used for analysis of transcriptome data (read mapping with HISAT2 and differential expression analysis by DEseq2) as previously described (34, 74). An analysis of previously published transcriptome datasets of mIMCD3 Pdk1 KO (GSE179947), Pkd2 KO (GSE179947), Ift140 KO (GSE109701), and Dync2h1 KO (GSE109701) cells was conducted using the GREIN platform (47, 48, 75). Gene set intersection analysis was conducted using the Multinet tool (https://multinet.app/, accessed 05 July 2024). The GenePattern resource was used for GSEA for GO-Term, matrisome DB, and literature-derived gene sets (29, 76, 77). The TcoF-DB v2 and Animal TFDB 4.0 databases were used for the annotation of TFs and transcription cofactors (78, 79). See data S2 to S5 for analysis of RNA sequencing data.

scRNA-seq analysis

Previously published scRNA-seq data of normal human kidney samples were analyzed using the CELL×GENE resource [datasets: scRNA-seq of the Adult Human Kidney (version 1.0), single-nucleus RNA-seq of the Adult Human Kidney (Version 1.0), accessed version: 11 April 2023] (80). A relative expression analysis of Arp2/3 complex genes and marker genes for kidney cell types was performed using the platform’s built-in gene expression tool. Expression analysis was conducted for predefined kidney cell types.

Quantification and statistical analysis

The GraphPad Prism 8 software was used for the generation of graphs and statistical analysis. Graphs show means ± S.E.M. and scatter dots indicating individual data points used for statistical analysis. Violin plots indicate data distribution with lines at the median and quartiles. The following statistical tests were applied on the basis of data distribution and the experimental design: one-way analysis of variance (ANOVA) test with Tukey’s multiple comparisons test (Figs. 1B, 2A, 6, B, C, J, and K; 7, E and F; and 9, C and L; and figs. S3C and S9C), unpaired two-tailed t test with Welch’s correction Figs. 1D; 3, I and J; 4, D, F, and H; 5, B, C, and D; and 9G; and figs. S1C; S4, D and E; and B, C, and D), Pearson correlation (Fig. 1, E and H), Spearman correlation (fig. S2D), two-tailed Mann-Whitney test (Fig. 4, G and I, and fig. S4, F and G), Brown-Forsythe ANOVA test with Dunnett’s T3 multiple comparisons test (Figs. 4, J and K, and 9H; and figs. S6, B, C, and D), Brown-Forsythe ANOVA test with Games-Howell’s multiple comparisons test (Fig. 5E), Kruskal-Wallis test Dunn’s multiple comparisons test (Figs. 5H and 8B), one-way ANOVA test with Sidak’s multiple comparisons test (Fig. 5, J and K), unpaired two-tailed t test (Figs. 3, C, F, and G; 6, A, G, and H; 7H; and 9, O, P, and Q; and figs. S7D, S8, B, C, F, and G; S11D, and S13, H and J), and two-way ANOVA test with Tukey’s multiple comparisons test (Fig. 7D). Statistical analysis of human disease datasets obtained from the Nephroseq v5 database (RRID:SCR_019050) was reported as calculated by Nephroseq (fig. S13, C to F, and data S6) (81). Tools used for statistical analysis of RNA-seq datasets are described in the respective methods sections. Statistical significance is indicated as n.s., nonsignificant; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Number of independent experiments and number of analyzed features are stated in the figure legends or methods section.

Supplementary Materials

The PDF file includes:

Figs. S1 to S13

Uncropped Western blots

Legends for data S1 to S6

Other Supplementary Material for this manuscript includes the following:

Data S1 to S6