Borg3 controls septin polymerization for primary cilia formation

Summary

Septin GTPases form hexa- or octameric rods that polymerize into higher order structures and integrate into the cytoskeleton, playing crucial roles in cellular functions. Among these, they are involved in the formation of primary cilia—cellular signaling hubs. It is established that septins localize to cilia and contribute to their formation and function; here, we aim to gain further insights into their oligomeric composition, assembly, and regulation in the confined ciliary compartment. Using cultured cells we demonstrate, that septins enter cilia as octamers and require polymerization for ciliary enrichment. Ciliary localization of septin filaments depends on Borg3, also known as Cdc42ep5, which we identify as an essential component of primary cilia. Knockout of Borg3 as well as dysregulation of Cdc42 impairs septin dynamics and enrichment within cilia. Borg3 localization is regulated by the cycling of the Rho-GTPase Cdc42 between its inactive- and active states confined at the ciliary base.

Graphical abstract

Highlights

• •Septin polymerization is essential for their enrichment at the cilium
• •Cdc42 GTPase cycling facilitates the recruitment of Borg3 to the cilium
• •Borg3 controls septin dynamics and polymerization at the ciliary base
• •Borg3 acts upstream of septins as a regulator of ciliogenesis

Biochemistry; Cell biology; Organizational aspects of cell biology

Introduction

Primary cilia are antenna-like projections found on the surface of nearly all eukaryotic cells. These non-motile structures play a critical role in signal transduction and environmental sensing, contributing significantly to the development of organs and the overall organism.¹ Dysfunction of primary cilia and cilia-associated proteins has been linked to diseases, including Bardet-Biedel syndrome, Meckel-Gruber syndrome, nephronophthisis, and polycystic kidney disease.^2,3

Septins (Sept) are a highly conserved family of small GTP-binding proteins ranging from 30 to 65 kDa. In humans 13 distinct septins are encoded, which are categorized into four groups based on sequence homology. Septins assemble into hexameric or octameric building blocks which can further associate into higher-order structures, such as filaments, rings, and cage-like formations. These structures frequently interact with other components of the cytoskeleton, including microtubules and actin filaments. Additionally, septins play a pivotal role in cell signaling and membrane organization.⁴

The role of septins in ciliogenesis has been previously studied and linked to diverse functions. Initially, septins were identified as forming a ring-like structure at the base of cilia,⁵ which is hypothesized to act as a diffusion barrier. However, subsequent observations revealed that septins more frequently exhibit non-ring-like localizations, either at the ciliary base or along the ciliary axoneme. This has led to the identification of additional roles for cilia-associated septins. Septins have been shown to regulate ciliary length⁶ and contribute to ciliary stability by balancing the length of the distal tip.⁷ Furthermore, they play critical scaffolding roles at the ciliary transition zone. Septin-dependent localization and interactions have been reported for several transition zone proteins like CC2D2A, B9D1, TMEM231,⁸ or DZIP1L.⁹ Recent findings have demonstrated that Sept9 regulates the recruitment of transition zone proteins on Golgi-derived vesicles by activating the exocyst complex via the ARHGEF18 and RhoA.¹⁰ Most studies agree that inhibition of septins - regardless of the method employed—leads to a significant reduction in cilia formation.^5,6,10 While the importance of septins in ciliary function continues to emerge, the mechanisms underlying the regulation of septin accumulation and assembly at cilia remain poorly understood.

Binder of Rho GTPases (Borg) proteins, also known as Cdc42 effector proteins (Cdc42eps) not only interact with septins, but also regulate their function.^11,12 Our previous studies suggest that the Rho GTPase Cdc42 and its effector proteins Borg play a critical role in regulating septin assembly during various cellular processes, such as the formation of toxin-induced protrusions¹³ and the establishment of septin barriers against P. aeruginosa cellular uptake.¹⁴ Research has shown that expression of either constitutively active or -inactive Cdc42, as well as Cdc42-binding deficient Borg2, disrupts filamentous septins structures that coalign with actin fibers in melanoma cells and cancer-associated fibroblasts.^11,15,16 This highlights a connection between Cdc42, downstream Borg proteins, septins and changes in actin organization. These processes are located in the cytosol, where the actin- and septin cytoskeleton mutually influence each other.¹⁷ This raises an intriguing question: how is septin accumulation and assembly into higher-order structures regulated in the confined compartment of the primary cilium? To address this, we investigate the role of Borg proteins in septin localization and dynamics at primary cilia. Our findings reveal that Borg3, in particular, is critical for septin accumulation at cilia and efficient ciliogenesis. We demonstrate that the Rho GTPase Cdc42 facilitates Borg3-driven septin assembly through a GTPase-dependent switch, which recruits Borg3 and septins to the ciliary compartment.

Results

Septins localize to cilia

Septin assemblies have been associated with diverse functions and localizations at cilia.^5,6 To investigate this further, we analyzed the localization of individual septins at the cilium in Madin-Darby canine kidney cells (MDCK) and mouse inner medullary collecting duct (IMCD) cells, two commonly used models for studying primary cilia formation. Immunofluorescence studies in both cell types for a common member of each septin group (i.e., Sept2, -3, -6, or -7 group, here Sept9 was stained as a Sept3 group member) revealed that septins preferentially localize along the entire length of the cilium as well as in a spot-like accumulation more restricted to the ciliary base (Figure 1A). Representative images with a larger field of view are shown in Figure S1A. Approximately 70% of cilia exhibited septin association along the axoneme, while in ∼30% of cilia, septins were restricted to the ciliary base. Only ∼5% of cilia showed no detectable septin association. These observations were consistent across both cell lines and align with findings from previous studies.^6,7,9 Since we did not stain for the ciliary base, we validated the exclusive localization of septins to the axoneme or the ciliary base by immunostainings in upright standing cilia (Figure S1B). This was achieved by a floating fixation that preserves the ciliary orientation and facilitates the identification of the ciliary base. To this end a polar accumulation of septins represents the ciliary base. Representative super resolution images for both localizations in the xy-plane, which allows for higher resolution imaging, are shown in Figure 1B (along axoneme) and Figures S1C and S1D (ciliary base, arrows). A ring-like localization at the base of cilia⁵ was the rarest septin assembly observed in our experiments. Occasionally, we detected structures that might resemble partially closed rings (representative super resolution image in Figure S1E, arrow). This suggests that the formation of septin ring-like structures may be linked to specific stages of ciliary development or decay. The presence of a septin from each group implies the existence of hetero-octameric protomers within cilia, which are increasingly recognized as the physiological building blocks of septin filaments. These filaments, in turn, are critical for the biological functions of septins.¹⁸

Figure 1.

Septin localizations in primary cilia

(A) Classification and quantification of different septin localizations at primary cilia. base: septins only found at the base of the primary cilium. Ring-like base: septins form a ring at the base of the primary cilium. Along axoneme: septins were found throughout the length of the primary cilium. No: no septins were found at the primary cilium. The illustration was created with BioRender. Cells were grown for 48 h under starving conditions. A random panel of 9 images was acquired at 40x magnification and all observed primary cilia were classified into one of four groups regarding their septin immunofluorescence in IMCD and MDCK cells. Data are given as percentage of total cilia observed in each immunostaining, mean ± SEM. n = 3.

(B) Representative SIM image (maximum intensity projection, MIP) of IMCD cells immunostained for endogenous acetylated tubulin and Sept2, -6, -7, or -9. Cells were serum starved for 48 h before fixation. Scale bars 1 μm.

(C) Representative SIM image (MIP) of MDCK cells immunostained for Sept7 and acetylated tubulin. Cells expressed shRNA targeting Sept9 or a non-targeting control shRNA and were starved for 48 h before fixation. Scale bars, 1 μm.

(D) Sept7 mean fluorescence intensity in cilia of MDCK cells treated as in C. Acetylated tubulin was used as mask for intensity measurements (AU = arbitrary units). Data are given ±SEM. > 40 cilia analyzed from n = 3, t test ∗∗∗∗p < 0.001.

(E and F) MDCK cells were transfected with non-targeting shRNAs or shRNAs targeting Sept7 or Sept2, respectively. Cells were starved for 48 h, fixed and stained for acetylated tubulin by immunofluorescence and with DAPI to visualize nuclei. GFP additionally expressed from the shRNA vector was used to identify KD cells. The percentage of cells with a cilium was quantified. The average of the controls was set to 100% and used for normalization. Data are given as mean ± SEM. ≥30 random fields of view were analyzed from n = 3, t test, ∗∗∗∗p < 0.001 and ∗p < 0.05, respectively.

(G) Representative SIM image (MIP) of IMCD cells transfected with unmodified (wt) or non-polimerizable (NCmut) Sept2-GFP and immunostained for endogenous acetylated tubulin and Sept9. Cells were serum starved for 48 h before fixation. Scale bars, 1 μm.

(H) GFP intensity ratio in IMCD cells treated as in G. Acetylated tubulin was used as mask for intensity measurements in the cilium and divided by the mean fluorescence intensity of a rectangle (1 μm × 1 μm) positioned at an area adjacent to the cilium. Data are given ±SEM. > 100 cilia analyzed from n = 10, t test ∗∗∗∗p < 0.001.

(I) Sept9 mean fluorescence intensity in cilia of IMCD cells treated as in G. Acetylated tubulin was used as mask for intensity measurements (AU = arbitrary units). Data are given ±SEM. > 100 cilia analyzed from n = 10, t test ∗∗∗∗p < 0.001.

The presence of septin octamers was further confirmed by knockdown (KD) of Sept9 (Figure S2A) which impairs the formation of functional Sept9-containing octamers. This led to reduced septin accumulation within the ciliary compartment (Figures 1C, 1D, and S2B) and formation of cilia (Figure S1C). Similarly, KD of Sept7 (Figure S2D), which disrupts formation of septin hexameric and octameric building blocks, or KD of Sept2 (Figure S1E, right), which is also required for filament elongation of hexa- and octamers, also inhibited ciliogenesis. This verifies the expected dependence of octameric building blocks for polymerization of functional septin structures in cilia. The effect was more pronounced for Sept7 (Figure 1E), likely due to Sept7 being the sole member of the Sept7 group. In contrast, KD of Sept2 (Figure 1F) may be partially compensated by the expression of other members of the Sept2-group.

Overexpression of exogenous Sept2-GFP with simultaneous siRNA KD (non-coding region) of endogenous Sept2 (Figure S2F) revealed that wild type (wt) Sept2-GFP is incorporated into ciliary septin structures. In contrast, Sept2 carrying mutations (F20D; V27D, hereafter referred to as NCmut) that abolish filament polymerization^18,19 but not hexa- or octamer formation was less enriched in the ciliary compartment (Figures 1G and 1H). A similar effect was observed for endogenous Sept9 following the expression of Sept2NCmut. Less endogenous Sept9 was detected in the ciliary compartment when septin polymerization was inhibited (Figure 1I). This indicates that proper ciliary septin localization requires septin basic building blocks (shRNA Sept7), relies on octamers (shRNA Sept9) and depends on the ability to assemble into filaments (Sept2NCmut).

While septin-related extra ciliary functions, such as vesicle transport,^10,20,21 may also influence cilia number and length, the reduced cilia association of all observed septins—caused by KD or introduction of non-polymerizable Sept2—underscore the critical role of septin polymerization and higher-order organization in the development of primary cilia.

Borg proteins colocalize with septins at cilia

The family of Borg proteins has been shown to bind to septins via the conserved BORG homology 3 domain (BD3), playing a role in septin reorganization within cells.^12,22 However, Borg proteins have never been observed or studied in the ciliary compartment. They remain largely uncharacterized, and little is known about their regulation in the ciliary context. To address this, we investigated BORG localization in cilia. Borg1, -2 and -3 were previously shown to be expressed in the kidney.²³ In subconfluent MDCK cells expressing GFP-tagged versions of Borg1-3, Borgs associated with cytoplasmic septin-positive filaments (Figure S2G), as previously described in various cell-types.^16,24,25,26 To further investigate endogenous Borg protein localization, we stained confluent IMCD cells (Figures 2A–2D, Video S1). Notably, Borg3 exhibited the strongest signal in the ciliary compartment compared to Borg1 and Borg2 (Figures 2A and 2B). Similar to septins, structured illumination microscopy (SIM) revealed that Borg proteins primarily localize along the cilium length in a spot-like pattern. While septins exhibited a more even distribution along the cilium, Borg proteins appeared to form more distinct clusters (Figures 2C and 2D). Frequently, Borg3 clusters were distributed along the cilium in an almost equidistant scattering (Video S2).

Figure 2.

Borg3 localizes to primary cilia and is involved in cilia formation

(A) Representative SIM image (MIP) of IMCD cells immunostained for endogenous Borg1, -2 or -3 and acetylated tubulin. Cells were serum starved for 48 h before fixation. Scale bars, 1 μm.

(B) Quantification of Borg immunofluorescence average intensity in cilia like in (A). Acetylated tubulin was used as mask for intensity measurements (AU = arbitrary units). Data are given ±SEM. ≥ 3 random fields of view were analyzed from n = 3, oneway ANOVA ∗∗∗∗p < 0.001.

(C) Representative SIM image (MIP) of IMCD cells immunostained for endogenous acetylated tubulin, Borg3 and Sept2, -6, -7, or -9. Cells were serum starved for 48 h before fixation. Borg3 and Sept were stained with directly labeled primary antibodies. Scale bars, 1 μm.

(D) Representative SIM image (MIP) of IMCD cells immunostained for endogenous Borg3 and acetylated tubulin. IMCD cells were fixed in a μ-slide (Ibidi) enabling analysis of standing cilia. Image shows zoom of cilium from Video S1. Colored arrows indicate all three dimensions with the viewing plane indicated by the horizontal and vertical arrow. Scale bar, 1 μm.

(E, F, and G) IMCD cells were transfected with control siRNA or with siRNAs targeting Borg1, Borg2, or Borg3 in (E), (F), or (G), respectively. siRNA-mediated KD was confirmed by western blot. α-tubulin was used as loading control.

(H) Representative confocal image of IMCD cells immunostained for endogenous Arl13b and acetylated tubulin. Cells were transfected as in (E), (F) and (G) and serum starved for 72 h before fixation. DAPI staining labeled nuclei. Scale bars, 10 μm.

(I) Quantification of cilia formation in cells treated as in (E), (F), (G), and (H). The percentage of cells with a cilium was quantified. Data are given ±SEM. > 4000 cells were analyzed from ≥ 25 random fields of view n = 4, oneway ANOVA ∗∗∗∗p < 0.001. Symbol shapes indicate individual data points from biological replicates.

Video S1. IMCD cells fixed in μ-slides (Ibidi) and stained for Borg3 (green) and acetylated tubulin (red) by immunofluorescence

Z-stack was 3D-rendered and animated by imaris. Scale bar indicates magnification. Stills of zoomed in cilium in Fig. 2 D.2

Video S2. IMCD cell stained for Borg3 (green) and acetylated tubulin (red) by immunofluorescence

Z-stack was 3D-rendered and animated by imaris. Scale bar indicates magnification.3

Borg3 influences formation of cilia

Given the specific localization of Borg proteins at cilia, we analyzed the effect of Borg depletion on cilia formation. IMCD cells were transfected with siRNAs targeting Borg1-3, and after 48 h, siRNA treatment led to a reduction in Borg expression at both the mRNA (Figure S2H) and protein level (Figures 2E–2G). Notably, Borg3 KD significantly reduced the number of cells developing a cilium (Figure 2H and 2I), whereas no such effect was observed for Borg1 or Borg2. Overall, no major changes in cell morphology were observed across all Borg family KDs.

To further investigate the role of Borg3 in cilia, we generated Borg3 knockout (KO) IMCD cells using CRISPR-Cas9.²⁷ KO analysis was performed by TIDE (https://tide.nki.nl; Brinkman et al.²⁸). The inserted mutations for two generated cell lines are outlined in Figure 3A (representative sequencing result Figure S3A). In KO#1, one allele contains a nucleotide insertion leading to a frameshift after 12 amino acids of the wt protein, while the other allele has a two-nucleotide deletion causing a frameshift after 10 amino acids.

Figure 3.

Borg3 KO inhibits formation of cilia

(A) Schematic representation of the sequencing results from the generated CRISPR-Cas9 mediated KO of Borg3 in IMCD cells. The nucleotides lost or gained in the subsequently used KO IMCD cell lines as well as the effect on the protein translation is shown. KO cell line 1 (KO#1) has two allele specific mutations while KO cell line 2 (KO#2) has a homozygous mutation. The corresponding sequencing chromatogram is shown in Figure S3 A.

(B) Quantification of cilia formation in IMCD wt and Borg3 KO cells. Cells were grown to confluency and starved for 48 h. In comparison cells transfected with GFP-Borg3 were analyzed to observe rescue effects. Cells were stained for acetylated tubulin to label cilia and with DAPI to label nuclei. Data are given as ±SEM. > 4000 cells were analyzed from 9 random field of views from n = 3, oneway ANOVA ∗∗∗∗p < 0.001. Symbol shapes indicate individual data points from biological replicates.

(C) Cilia growth over time in wild type and KO IMCD cell lines stably expressing Arl13b-tomato. Cells were seeded in high and equal densities to reach immediate confluency at imaging start. Cells were observed by Video Smicroscopy for 100 h with images taken every 30 min. Cilia were identified by fluorescence using an intensity and size dependent filter (Figure S3C). KO cells were normalized to the highest cilia count of the WT control. For every condition each biological replicate consisted of 3 technical replicates in which 4 random fields of view per replicate were analyzed. Data are given ±SEM. n = 3.

(D) Borg3 KO IMCD cell line 1 stably expressing Arl13b was treated as in (C). Additionally, KO cell line 1 stably expressing Arl13b was transfected with GFP-Borg3 1 day before the experiment and enriched by FACS as a rescue. Cells were observed and analyzed as in (C). The rescue was normalized to the highest cilia count of KO cell line 1 rescue. Data are given ±SEM. n = 3.

In KO#2, both alleles carry the same mutation—an insertion of one nucleotide—resulting in a frameshift after 13 amino acids. Consequently, both functional domains of Borg3, the CRIB domain and the BD3 domains, are disrupted. The CRIB domain mediates interaction with Cdc42, while the BD3 domain interacts with septins Sept6/7.^22,23,29 Immunofluorescence staining of the Borg3 KO#1 cell line confirmed the absence of BORG3 staining in cilia (Figure S3B). Consistent with the KD experiments, of BORG3 KO reduced the number of cells developing a cilium (Figure 3B). The phenotype was rescued by transient re-expression of GFP-Borg3 in KO cells, leading to a significant increase in the fraction of ciliated cells in both KO cell lines.

Since quantifying cilia-bearing cells after fixation and immunofluorescence staining provides only a snapshot of cilia formation—and recent studies argue that fragile parts of cilia can be lost during fixation and sample processing⁷—we tracked cilia formation in real time over up to 4 days. To do this, cells stably expressing ARL13B-tomato were seeded at high density, transferred to FCS-free medium and monitored by Video Smicroscopy. These real-time studies confirmed the immunofluorescence endpoint studies, showing that all KO cell lines exhibited a reduced and delayed cilia formation (Figures 3C and S3C). The KO#1 phenotype was rescued by transient re-expression of GFP-Borg3. After transfection, cells were enriched by FACS and GFP-Borg3 expression led to an increase in cilia formation (Figure 3D).

Septins at cilia are regulated by Cdc42 and its downstream effector Borg3

All Borg proteins share a conserved CRIB domain and function as downstream effectors of the small Rho GTPases Cdc42 and TC10.^11,13,15,23 This raises the intriguing possibility that the interplay between Cdc42 and Borg proteins play a crucial role in septin assembly at cilia. Furthermore, previous studies suggest that Cdc42 contributes to ciliogenesis, for example, by localizing and regulating the exocyst complex.^30,31

Confluent MDCK cells expressing 5HT6-tomato as a ciliary marker were stained for Sept2 and GTP-bound Cdc42 using immunofluorescence. Fixation and staining conditions were modified to preserve the upright orientation of cilia. Confocal imaging revealed that active, GTP-bound Cdc42 was confined to the ciliary base (Figure 4A), where Sept2 was also present and extended further into the cilium. Higher-resolution 3D-SIM imaging of upright cilia, combined with statistical analysis of the first 3 μm from the base (Figure S4A), confirmed that active GTP-bound Cdc42 remained restricted to the ciliary base. In contrast, septin intensity gradually increased beyond this region. Further super resolution microscopy of cells stained for GTP-bound Cdc42 and Borg3 revealed that Borg3 localizes in close proximity to GTP-bound Cdc42 at the ciliary base (Figure 4B) and is also present within the cilium. Notably, the Cdc42 guanine nucleotide exchange factors (GEFs) Tuba and intersectin 2, previously identified as positive regulators of ciliogenesis also exhibit ciliary localization.^32,33 We observed colocalization of Tuba and Centrin2, a marker of the ciliary base, following expression of both proteins (Figure S4B). Tuba was enriched at the ciliary base compared to the cytosol (Figure S4C). Additionally, both expression of Cdc42 or immunofluorescence staining of Cdc42—independent of its activity state-revealed its accumulation at the cilium (Figures S4D and S4E). However, unlike GTP-bound Cdc42, the signal of total Cdc42 was more broadly distributed and less restricted to the ciliary base. Cdc42 within the cilium did not appear to be membrane-associated as the ciliary membrane was resolved by SIM using an ARL13B staining (Figure S4E, line scan). Taken together, these experiments demonstrate that Cdc42 localizes to primary cilia, and the restricted, activity-dependent localization of GTP-bound Cdc42 suggests a specific spatial regulation of this Rho GTPase. To further investigate the role of Cdc42 in ciliogenesis, we used MDCK cells stably expressing shRNAs targeting Cdc42. KD of Cdc42 reduced the percentage of ciliated cells up to 45% compared to the control (Figures 4D and S4F). Next, we investigated the impact of Cdc42 nucleotide exchange on cilia formation. We hypothesized that dynamic cycling of Cdc42 between its active GTP-bound and the inactive GDP-bound states influences the efficacy of Borg3 recruitment and ciliogenesis. To test this, MDCK cells were transfected with GFP-tagged constitutively active (Q61E) or inactive (T17N) Cdc42 mutants, and the number of transfected cells with cilia was quantified (Figure 5A). Expression of dominant-negative Cdc42 (T17N) and constitutively active Cdc42 (Q61E), both led to a reduced number of ciliated cells. Similarly, expression of the septin binding domain of Borg3 (BD3), which lacks the regulatory Cdc42 interaction site, decreased ciliogenesis to a similar extent (50%). These findings suggest that the precise spatiotemporal regulation of Cdc42 and its downstream effector Borg3 is crucial for septin regulation during primary ciliogenesis. We speculate that an excess of constitutively active Cdc42 may mis-localize or sequester Borg3 from the cilium, while dominant-inactive Cdc42 could inhibit Cdc42 activation at the ciliary base. Additionally, an unregulated excess of the BD3 domain might sequester septins away from the cilium. Therefore, global deregulation of Cdc42 or BORG3 negatively impacts cilia formation.

Figure 4.

Formation of cilia is regulated by Cdc42

(A) Representative confocal image of MDCK cells transfected with 5HT6-tomato to visualize the primary cilium immunostained for Sept2 and active Cdc42. The images show the top view on the cilium (upper row), the side view (bottom left) and a 3D-line scan. Arrows in the images and the line scan indicate accumulation of active Cdc42 at the base of the cilium. Scale bars, 5 μm.

(B) Representative SIM image (MIP) of IMCD cells immunostained for endogenous Borg3 and active Cdc42. Arrow indicates accumulation of active Cdc42 with surrounding Borg3 at the ciliary base. Borg3 also locates along the axoneme. Scale bar, 1 μm.

(C) Representative confocal images of stable MDCK KD cell lines immunostained for acetylated tubulin. Cells that express shRNA targeting Cdc42 and/or TC10 or non-silencing control (ns control) are marked by GFP and/or RFP fluorescence. Cells were serum starved for 48 h before fixation. Nuclei were stained with DAPI. Scale bar, 5 μm.

(D) Graph shows quantification of cilia formation in TC10 and Cdc42 KD cells (representative images shown in C). Data are given ±SEM, ≥30 fields of view were analyzed from n = 3, oneway ANOVA ∗∗∗∗p < 0.001.

(E) MDCK cells were transfected with Cdc42 (T17N)-GFP, Cdc42 (Q61E)-GFP, the septin binding domain BD3 or GFP as control and were starved for 48 h. Cells were stained for acetylated tubulin and nuclei. Data are given ±SEM, >30 fields of view were analyzed from n = 4, oneway ANOVA ∗∗∗∗p < 0.001.

(F) Representative confocal images of IMCD cells immunostained for TC10 and acetylated tubulin. Cells were serum starved 48 h before fixation. Scale bar, 5 μm.

(G) 3D-intensity surface-blot shows quantification of active Cdc42 immunofluorescence intensity at the ciliary base in a 2.6 μm square (400 pixels). 15 images of active Cdc42 at the base of individual cilia were analyzed. For all cilia the z-plane sectioning the ciliary base was used. All images were centered on the base of the cilium. The fluorescence intensity of each pixel was measured for each image and summed up to generate a 3D-graph (AU = arbitrary units). The Anova pixel analysis (small graph, left) shows every pixel in red with statistically elevated intensity (∗p < 0.05). n = 15.

(H) Same quantification as in G) for TC10 intensity at the cilium. Here a TC10 antibody was used that recognizes active- and inactive TC10. n = 15.

Figure 5.

Localization of Septins and Borg3 to the primary cilium depends on Cdc42

(A) Representative SIM images (MIP) of cilia of IMCD wt and Borg3 KO cell lines immunostained for endogenous Sept2 and acetylated tubulin. Cells were serum starved 48 h before fixation. Scale bars, 1 μm.

(B) Quantification of Sept2 immunofluorescence average intensity in cilia from IMCD cells treated as in A. Acetylated tubulin was used as mask for intensity measurements (AU = arbitrary units). Data are given ±SEM. ≥ 9 random fields of view were analyzed from n = 3, oneway ANOVA ∗∗∗∗p < 0.001.

(C) Representative SIM images (MIP) of cilia of IMCD wt and Borg3 KO cell lines immunostained for endogenous Sept7 and acetylated tubulin. Cells were serum starved 48 h before fixation. Scale bars, 1 μm.

(D) Quantification of Sept7 immunofluorescence average intensity in cilia from IMCD cells treated as in C. Acetylated tubulin was used as mask for intensity measurements (AU = arbitrary units). Data are given ±SEM. ≥ 9 random fields of view were analyzed from n = 3, oneway ANOVA ∗∗∗∗p < 0.001.

(E) 3xHA-Borg3 and 3xHA-Borg3 (I23A, S24A) were expressed in IMCD cells starved for 48 h. Cells were immunostained for the HA-tag and for Arl13b. Confocal images were acquired. A mask was generated by the Arl13b channel to measure the HA average fluorescence intensity in the cilium. This value was divided by the average intensity adjacent to the cilium in the focal plane to yield a cilium/cell intensity ratio. Data are given ±SEM from n = 4, t test ∗∗p < 0.01.

(F) Representative confocal images of IMCD cell lines stably expressing shRNA targeting Cdc42, TC10 or a non-targeting shRNA control as well as the IMCD Borg3 KO#2 cell line stably expressing the non-targeting shRNA control transfected with Sept6-GFP (for shTC10 and respective control) or Sept6-tom (for shCdc42 and respective control) 24 h before imaging. SiR-tubulin was added to the cells as a ciliary marker 1 h before imaging. Cells were bleached with the corresponding laser until 50% of the previously observed fluorescence intensity inside the cilium was lost and recovery was observed for 10 min. Scale bars, 1 μm.

(G) Quantification of ciliary Sept6 intensity in IMCD wt and Borg3 KO#2 cell lines treated as in F. Data are given ±SEM. n = 10.

In addition to Cdc42, Borg proteins also interact with TC10 known as RhoQ,²³ a small Rho GTPase closely related to Cdc42. While Borg3 is exclusively regulated by Cdc42,²³ TC10, and Cdc42 have been shown to exhibit functional redundancy in several cellular processes.^34,35 TC10 is primarily localized at the cell membrane and vesicles, where it plays a key role in regulating exocytosis in adipocytes and neurons.^36,37 However, the role of TC10 in BORG regulation and its potential involvement in downstream septin regulation at the cilium has not yet been investigated.

Immunofluorescence microscopy and of GFP-TC10 expression revealed that TC10 accumulates around the base of cilia (Figures 4F and S5B). Quantitative spatial analysis of fluorescence signals for GTP-bound Cdc42 (Figure 4G) and TC10 (Figure 4H) demonstrated a significant difference in their localization patterns, highlighting distinct spatial preferences between these two Rho GTPases. Whereas GTP-bound Cdc42 predominantly localizes to a distinct spot at the base of cilia, TC10 is excluded from this region and instead surrounds the ciliary base in small clusters. We propose that TC10 positive vesicles arrange around the base of the cilium. Notably, Figure 4H does not indicate that TC10 forms a continuous ring around cilia; rather, it represents the cumulative fluorescence distribution from 15 cilia, highlighting the regions where TC10 fluorescence was most frequently observed within the optical plane of the ciliary base. These findings suggest that TC10 and Cdc42 may play distinct roles in ciliogenesis. Furthermore, the non-overlapping localization of TC10 and Borgs indicates that TC10 is unlikely to be involved in the recruitment of Borg3 to cilia. We did not observe a robust localization of TC10 within the ciliary compartment (Figure S5C), as was the case for Cdc42 (Figures S4D and S4E).

To gain further insight into the roles of these Rho GTPases in ciliogenesis upstream of Borg3, stable KDs of TC10, Cdc42, and a combined KD of both were generated in MDCK cells (Figures 4D and S5A). The TC10 single KD had no effect on the number of ciliated cells, while the double KD of TC10 and Cdc42 reduced ciliogenesis to the same extent as the Cdc42 single KD (Figures 4C and 4D), with no additional effect observed. These findings further support the idea that Cdc42, rather than TC10, serves as the upstream regulator of Borg3 localization at cilia. It is tempting to speculate that Cdc42 and Borg3 affect the formation of cilia by regulating of septin recruitment and assembly.

Since Borg3 interacts with the active, GTP-bound form of Cdc42,²³ we hypothesized that the dynamic cycling of Cdc42 between its inactive (GDP-bound) and active (GTP-bound) states is crucial for recruiting Borg3 in sufficient amounts. This recruitment, in turn, may be necessary to drive septin accumulation and subsequent polymerization at cilia.

Borg3 regulates septin dynamics at cilia

To study the role of Borg3 in septin recruitment to cilia, we analyzed septin localization in Borg3 KO cells, comparing it to wt cells²⁹ (Figures 5A–5D). Immunofluorescence staining was used to visualize Sept2 (Figures 5A and 5B) and Sept7 (Figures 5C and 5D), with acetylated tubulin staining serving as a mask to assess septin fluorescence intensity within the ciliary compartment. In Borg3 KO cells, the fluorescence intensity of both Sept2 and Sept7 was significantly reduced. These findings suggest that Borg3 plays a role in either the recruitment of septins to the ciliary compartment or in the regulation of septin assembly within cilia.

To investigate the function of Borg3 in septin assembly at cilia, we utilized a Borg3 variant carrying I23A, S24A mutations in the CRIB domain, which disrupt its interaction with Cdc42²³ (Figures 5E and S5D). Following the expression of HA-tagged wt and mutant Borg3, significantly less mutant Borg3 was detected in the ciliary compartment compared to the cytosol. Borg3’s interaction with Cdc42 appears crucial for its localization to cilia and potentially for its role in septin assembly.

To further examine the impact of Borg3 on septin dynamics at cilia, we performed fluorescence recovery after photo bleaching (FRAP) experiments (Figures 5F and 5G). Sept6-GFP was expressed in both wt and KO#2 cells and microtubules were stained with SiR-tubulin to identify cilia. Shorter cilia were chosen for bleaching to ensure complete photobleaching of the cilium while minimizing bleaching of fluorescent Sept6 in the cytosol. Smaller cilia might exhibit more dynamic Sept6 behavior due to active growth phases. After bleaching, we observed more efficient entry of Sept6-GFP into the ciliary compartment when functional Borg3 was present, suggesting that Borg3 plays a role in regulating septin dynamics at cilia. Notably, the entry of Sept6-GFP was slow taking nearly 10 min to recover 80% of the pre-bleach fluorescence intensity. Borg3 is potentially regulated or recruited by Cdc42, which must cycle between the GDP-bound inactive and GTP-bound active states for proper function. Additional experiments with cells expressing shRNAs targeting Cdc42 (Figure S5E) or TC10 (Figure S5F) further supported the idea that Cdc42 acts as the upstream regulator of septin localization to primary cilia. Cells with Cdc42 KD cells showed reduced fluorescence recovery compared to the respective control mimicking the effects of Borg3 KO. In contrast, TC10 KD had no impact on fluorescence recovery (Figures 5F and 5G) compared to the respective control, suggesting that Borg3 downstream of Cdc42 significantly influences septin dynamics at cilia.

Discussion

Here, we identify Borg3 as a novel regulator of primary cilia formation. Our findings demonstrate that disrupting Borg3 function through either KD or KO reduces primary cilia formation. Borg3 localizes both at the ciliary base and within the cilium, where it colocalizes with septin structures. Ciliary proteomes have been analyzed and septin family members have been identified,^38,39,40 whereas Borg proteins were not detected in these studies. This discrepancy may be due to the challenging nature of Borg3 detection by mass spectrometry, as tryptic digestion produces a particularly long peptide and another peptide prone to crosslinking, resulting in low sequence coverage.

Only two domains of Borg proteins have assigned functions. The first is the CRIB domain, which mediates interaction with Cdc42. However, Borg3 does not appear to function as a classical Cdc42 effector, as it may not exert an enzymatic function upon activation. Instead, active Cdc42 may be required solely for Borg3 localization.^11,15,25

The second functional domain of Borg proteins is the BD3 domain. Previous studies demonstrated that the BD3 domain interacts with the C-terminal domains of Sept6 and Sept7.^22,29 This interaction may not be limited to a single filament but could also involve adjacent filaments, potentially promoting the formation of bundles and higher order structures. Furthermore, BD3-domain binding to the C-terminal domains of Sept6 and Sept7 has been shown to induce the polymerization of septin oligomers into filaments in vitro.²⁹

We observed that both Borg3 and septin polymerization are required for efficient septin accumulation in the ciliary compartment. Furthermore, septin octamers containing Sept9 appear to serve as the primary building blocks for ciliary septin structures. The interaction of Borg3 with Sept6 and Sept7 does not explain the preference for octamer-based filaments in cilia. However, recent studies suggest that the septin octamer is the functional unit of the septin cytoskeleton.^18,41 Moreover, the frequent observation of octamer-based septin filaments on contractile F-actin structures or microtubules,^18,41 along with the presence of Borg3 at the latter,^13,26,42 suggests a potential correlation between these components. Further studies are necessary to analyze the higher order structures septins form within cilia and how Borgs influence these structures.

Supporting the idea that Borgs interact with Cdc42, we identified Cdc42 as a regulator of ciliary septin assembly and Borg3 recruitment. Locking Cdc42 in either its active or inactive state, as well as downregulating it via KD, led to reduced cilia formation and impaired recruitment of both Borg3 and septins. This suggests that the spatio-temporally regulated cycling of Cdc42 between its inactive GDP-bound and active GTP-bound states is crucial for Borg3-assisted septin recruitment and assembly. Furthermore, the repetitive switching of Cdc42 between these states may enhance the process by continuously attracting and releasing Borg3, thereby amplifying septin dynamics at cilia.

We propose a mechanism in which a GTPase-dependent switch at the base of cilia increases Borg3 concentration thereby promoting efficient septin assembly at cilia (Figure 5H). This process likely requires the presence of both a Cdc42 guanine nucleotide exchange factor (GEF) and a GTPase-activating protein (GAP) at cilia. Tuba and intersectin 2 have emerged as promising as GEF candidates,^32,33 while GAPs for Rho-GTPases have been identified in ciliary proteomes^38,40 and may serve as potential interaction partners for septins.⁴³ Through this regulation, a localized concentration of Borg3 and septins could be established, creating favorable conditions for septin assembly at cilia.

Most in vitro studies induce septin polymerization by shifting ionic strength from high to low,⁴⁴ but this is unlikely to be the physiological mechanism controlling septin dynamics in cells. Interestingly, in the presence of the BD3 domain, septin polymerization is independent of ionic strength shifts in vitro.²⁹ This raises the question of how specificity for Borg3 is achieved, given that all Borg proteins share the conserved CRIB and BD3 domains. We observed that active Cdc42 is enriched at the base of the cilium forming a confined spot surrounded by a region with TC10 localization. Notably, Borg3 is the only Borg protein regulated exclusively by Cdc42 and not by TC10 as well.²³ Thus, the spatial distribution of Cdc42 and TC10 activity at the ciliary base may play a role in selectively recruiting Borg3. It is tempting to speculate that the other Borgs are also recruited to the periphery of the cilium, while Borg3 is exclusively directed to it. This does not exclude the possibility that TC10, other Borgs and downstream septins have additional cilia-related functions—such as vesicle tethering and subsequent exocytosis^45,46—which may play a crucial role in directing membrane protein transport to the cilium.

The localization of active GTP-bound Cdc42 at the base of cilia, followed by Borg3 recruitment, may create additional synergistic effects with the adjacent membrane. An isoprene derivative, linked to the cysteine residue of the C-terminal CaaX-box of Cdc42, anchors the Rho-GTPase to the membrane.⁴⁷ Septins preferentially accumulate at curved membranes^48,49 found at the base of cellular protrusions. This provides additional enrichment mechanisms that concentrate essential components at the site of polymerization. Notably, in fungi, plasma membrane associated Cdc42 and its upstream regulators dictate septin recruitment and assembly during bud formation.⁵⁰

The cytoskeletal partners of septins, actin, and microtubules, rely on numerous binding proteins to regulate and fine-tune their polymerization. Well studied examples include formins and +end binding proteins, which regulate actin- and microtubule polymerization, respectively. It is plausible that the septin cytoskeleton employs similar regulatory mechanisms. Among the known candidates, Borg proteins are the most extensively documented in this role. Our findings highlight Borg3 as a key regulator of septin polymerization at cilia. However, further studies are needed to explore the broader involvement of Borgs in other septin-related cellular functions. Additionally, the precise molecular mechanisms by which Borg proteins facilitate septin filament formation and higher-order assembly remain unclear, warranting further investigation.

Limitations of the study

Our study investigates the impact of Borg3 proteins on septin dynamics within the confined environment of the primary cilium. Using polarized monolayers of commonly employed cell lines as a model for cilia formation, we demonstrate that Borg3 influences the assembly of higher-order septin structures within cilia. However, our findings do not establish whether these effects persist under in vivo conditions.

While our data suggest a role for Borg3 in septin organization, the specific structural changes in septin assemblies remain unexamined. Recent studies propose that Borg proteins may be involved in interfilament interactions, highlighting the need for further research to elucidate their influence on filament composition and organization in greater detail.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Carsten Schwan (carsten.schwan@pharmakol.uni-freiburg.de).

Materials availability

This study did not generate new unique reagents other materials are available from the lead contact upon request.

Data and code availability

• Data: data sheets for all quantification and raw images are available from the lead contact upon request.

• Code: this study did not generate new unique codes.

• All other items: any additional information needed for data presented in this paper is available from the lead contact upon request.

Acknowledgments

We thank Feng Zhang, Joseph Gleeson, Channing Der, Kirk Mykytyn, Manos Mavrakis, Pietro De Camilli, and Tamara Caspary for Addgene plasmids. We thank Ian Macara for 3xHA-Borg3 plasmid. Work was supported by Deutsche Forschungsgemeinschaft (DFG) with a research grant (SCHW 1708/2-1) to C.S. The Elyra microscope system was funded through the DFG major research instrumentation grant INST 39/1170-1 FUGG. We thank Otilia Wunderlich and Peter Gebhardt for technical assistance.

Author contributions

J.N.S., F.L., A.V.M., and C.S. designed and performed the experiments. J.N.S. performed most of the experiments. J.N.S., F.L., A.V.M., and C.S. analyzed the data. J.N.S. and C.S. wrote the manuscript. C.S. conceived and supervised the project and acquired funding for the project.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
CDC42EP1	Sigma-Aldrich	RRID: AB_1078486; Cat# HPA006379
CDC42EP2	Sigma-Aldrich	RRID: AB_10671745; Cat# HPA038562
CDC42EP3	Sigma-Aldrich	RRID: AB_2732670; Cat# HPA061792
CDC42EP5	ThermoFisher	RRID: AB_2854305; Cat# PA5-106637
CDC42EP5	Sigma-Aldrich	RRID: AB_10961041; Cat# HPA043449
SEPT2	Sigma-Aldrich	RRID: AB_1856684; Cat# HPA018481
SEPT7	Sigma-Aldrich	RRID: AB_10601573; Cat# HPA029524
SEPT6	Santa Cruz	RRID: AB_2184999; Cat# sc-20180
SEPT9	Sigma-Aldrich	RRID: AB_2678054; Cat# HPA042564
H3	Cell Signaling	RRID: AB_3289584; Cat# 4499S
GAPDH	Millipore	RRID: AB_2107445; Cat# MAB374
alpha tubulin	Invitrogen	RRID: AB_10984311; Cat# PA5-19489
acetylated α Tubulin	Santa Cruz	RRID: AB_628409; Cat# sc-23950
CDC42	Santa Cruz	RRID: AB_627233; Cat# sc-8401
active CDC42	New East Biosciences	RRID: AB_1961759; Cat# 26905
TC10	Sigma-Aldrich	RRID: AB_477591; Cat# T8950
Oligonucleotides
mCdc42ep2_fw qPCR (TCCCCATCTATTTGAAACGTGG)	N/A	N/A
mCdc42ep2_rv qPCR (CCGCTGTTCCTGGAAGGAG)	N/A	N/A
mCdc42ep3_fw qPCR (CCAAGACCCCAATTTACCTGAAA)	N/A	N/A
mCdc42ep3_rv qPCR (CCCTCTTTGCCGATGTGTATAGT)	N/A	N/A
mCdc42ep5_fw qPCR (GGGATGCCCACCCTAGAGT)	N/A	N/A
mCdc42ep5_rv qPCR (TGGAGGTCAGCATTTGAGCAG)	N/A	N/A
Borg3 p3 fw (KO#1) CRISPR/Cas9 (CACCGAGCCGCTTCTTGGGTTGTGC)	N/A	N/A
Borg3 p3 rv (KO#1) CRISPR/Cas9 (AAACGCACAACCCAAGAAGCGGCTC)	N/A	N/A
Borg3 p5 fw (KO#2) CRISPR/Cas9 (CACCGCCCTGCACAACCCAAGAAG)	N/A	N/A
Borg3 p5 rv (KO#2) CRISPR/Cas9 (AAACCTTCTTGGGTTGTGCAGGGC)	N/A	N/A
Borg3 seq. fw sequencing (GATCAGGTACAGTTATGGGCG)	N/A	N/A
Borg3 seq. rv sequencing (GTGGAGTGCTGGGAGGGAG)	N/A	N/A
Borg3 seq. 2 fw sequencing (CAGGAGGGACCTTGAGAACCT)	N/A	N/A
Borg3 seq. 2 rv sequencing (CGACTCAGGAATGAGGTGTCC)	N/A	N/A
AllStars Neg. Control siRNA KD	Quiagen	1027292
siCdc42EP2#1 KD (GACCUUCCCUUCCAGUUUA)	Horizon	D-044823-01
siCdc42EP2#2 KD (GAUUAUGGAUCACGACCUA)	Horizon	D-044823-02
siCdc42EP2#3 KD (UGGCGGAGAUGACAUGUUU)	Horizon	D-044823-03
siCdc42EP2#4 KD (CGUGCAGAUUCCUACAUA)	Horizon	D-044823-04
siCdc42EP3#1 KD (GGAGCAAAGUAGUCUAUUA)	Horizon	D-046421-01
siCdc42EP3#3 KD (GAUCUUGGGCCUUCACUUU)	Horizon	D-046421-03
siCdc42EP5#1 KD (GGAGCACUCUCGAUCUCAG)	Horizon	D-063228-01
siCdc42EP5#3 KD (CGACGUCACGGGUCUGUAG)	Horizon	D-063228-03
siSEPT2 KD (AGGCAGGGAUUUACGUUUA)	Horizon	J-051534-11-0010
shSEPT2 KD (GGAGAACATCGTGCCCGTC)	N/A	N/A
shSEPT7 KD (GGCAGTATCCTTGGGGTGT)	N/A	N/A
shSEPT9 KD (GTCCATCACGCACGATATT)	N/A	N/A
shScramble KD (GATCTGATCGACACTGTAA)	N/A	N/A
shCdc42 #1 KD (CAGGAGACATGTTTTACCA)	Horizon	V3LHS_641566
shCdc42 #2 KD (TCTGTCATAATCCTCTTGC)	Horizon	V2LHS_261933
shTC10 KD (CACGTAATCAAACAACAGT)	Horizon	V3SVMM08_11985849
pGIPZ non-silencing lentiviral shRNA control KD	Horizon	RHS4348 GE
Recombinant DNA
GFP-Borg1	Nölke et al. 201613	N/A
GFP-Borg2	Nölke et al. 201613	N/A
GFP-Borg3	Nölke et al. 201613	N/A
HA-Borg3	gift from Ian Macara	N/A
HA-Borg3 (I23A, S24A) site directed mutagenesis from HA-Borg3	HA-Borg3	N/A
GFP-BD3	Nölke et al. 201613	N/A
SEPT2-GFP	Nölke et al. 201613	N/A
SEPT2NCmut-GFP	Addgene	RRID: Addgene_180320
SEPT6-GFP	Nölke et al. 201613	N/A
SEPT6-tom	Nölke et al. 201613	N/A
GFP-TC10	Addgene	RRID: Addgene_23232
GFP-Cdc42	gift from Gudula Schmidt	N/A
GFP-Cdc42 (Q61E)	gift from Gudula Schmidt	N/A
GFP-Cdc42 (T17N)	gift from Gudula Schmidt	N/A
pSpCas9(BB)-2A-GFP	Addgene	RRID: Addgene_48138
dsRed-cent2	Addgene	RRID: Addgene_29523
GFP-Tuba subcloned from pcDNA3-HA-Tuba		RRID: Addgene_22214
5ht6-tomato subcloned from pEGFPN3-5ht6		RRID: Addgene_35624
Arl13b-tom subcloned L13-Arl13bGFP		RRID: Addgene_40879
Software and algorithms
Metamorph	Molecular Devices	N/A
Imaris	Oxford Instruments	N/A
Visiview	Visitron Systems	N/A
GraphPad Prism	Dotmatics	N/A
Zen black	Zeiss	N/A

Experimental model and study participant details

MDCK and IMCD cells were provided by the Renal Division at the University Freiburg Medical Center. MDCK cells were cultured in Modified Eagle’s Medium (MEM) supplemented with 10% FCS, 1% penicillin/streptomycin and 1% nonessential amino acids. Platinum-E cells (gift from Prof. Toshio Kitamura) and IMCD cells were cultured in DMEM/F12 supplemented with 10% FCS, 1% penicillin/streptomycin. 1% nonessential amino acids and 1% sodium pyruvate (Biochrom) were additionally added for Platinum-E cells. Cell lines were tested for mycoplasma contamination monthly.

Method details

Cell culture and transient transfections

To enhance ciliogenesis cells were placed in their respective growth medium without FCS for 48h. For immunostainings, cells were plated on HCl-washed coverslips or on μ-Slide 8 wells (Ibidi). For live-cell imaging, cells were plated on glass-bottom dishes (Greiner).

Cells were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s protocol. When necessary, transfected cells were enriched by fluorescence-activated cell sorting with a BD Melody 48 h after transfection and directly reseeded on glass-bottom well-plates. After attachment of cells the medium was exchanged. Observation of ciliogenesis for 100h was done under starvation conditions.

shRNAs

The shRNAs for Sept2, for Sept7 and for Sept9 were cloned into the pSUPER.neo+GFP vector (Oligoengine).

The Cdc42 shRNA were inserted into the pGIPZ lentiviral vector with GFP (Clone-Id: V3LHS_641566 GE Dharmacon and Clone-Id: V2LHS_261933 GE Dharmacon). TC10 shRNA was inserted into pSMART lentiviral vector with Turbo RFP. Following non-silencing control was used: pGIPZ non-silencing lentiviral shRNA control (# RHS4348 GE Dharmacon). Sequences are shown in the key resources table.

Western Blot

Lysates were generated from confluent cells in RIPA lysis buffer. Supernatants were analyzed using SDS-PAGE (sodium dodecyl sulfate – polyacrylamide gel electrophoresis) followed by semi-dry transfer to PVDF membranes and subsequent immunoblotting using the antibodies in the key resources table.

Lentiviral packaging and transduction

Packaging of the lentiviral vectors pGIPZ-GFP shRNA Cdc42 #2, pGIPZ-GFP non-silencing control, pSMART-TurboRFP shRNA TC10 was performed using transient transfection into Platinum-E cells (a gift from Prof. Toshio Kitamura, University Tokyo,⁵¹) together with five packaging plasmids: pTLA1-Pak, pTLA1-Enz, pTLA1-Env, pTLA1-Rev and pTLA1-TOFF (Thermo-Fisher). 1.5 × 10⁶ Platinum-E cells were plated onto 10 cm dishes and grown for one day. Culture supernatants of the lentivirus producing cells were collected. The medium from sub-confluent MDCK cells was removed and replaced by a 1:1 mixture of lentivirus containing medium and growth medium, supplemented with 8 μg/mL polybrene. After 48 h selection of transduced cells was initialized by 3 μg/mL Puromycin. The infection efficiency of the lentivirus was determined by green or red fluorescent protein (GFP and RFP) expression. Double KD cells were first transduced with pGIPZ-GFP shRNA Cdc42#2. The expression of each construct was detected using a fluorescence microscope.

CRISPR/Cas9 mediated Knockout of Borg3

BORG3 KO cell lines were generated according to Ran et al.²⁷ In brief, targeting sequences (key resources table) for the guide RNAs were cloned into pSpCas9(BB)-2A-GFP (PX458)-plasmid (from addgene, #48138). 500 ng of the plasmid were introduced to 90% confluent IMCD cells. Cells were sorted and seeded as single cells. New cell lines carrying the desired mutation were identified by sequencing. Nucleotide sequences are shown in the key resources table.

Antibodies and fluorescent proteins

The used antibodies are listed in the key resources table.

dsRed-cent2 (Plasmid #29523)⁵² and GFP-TC10 (Plasmid #23232),⁵³ pEGFPN3-5ht6 (Plasmid #35624),⁵⁴ Sept2NCmut-msfGFP (Plasmid #180320),¹⁸ L13-Arl13bGFP (Plasmid #40879),⁵⁵ were obtained from addgene for direct use or further subcloning. 3xHA-Borg3 was provided by Ian Macara.²³ Other plasmids were previously used in Nölke et al. and Ostevold et al.^13,24

Immunostaining

Cells were washed with PBS, fixed for 15 min with 4% formaldehyde in PBS, permeabilized (10min) with 0.15% Triton X-100 in PBS, and blocked by 1% BSA or normal goat serum in PBS for 30min. Cells were incubated with the primary antibody for 60-90 min at RT. Cells were washed with PBS and incubated with the suitable secondary antibody for 1h. Cells on coverslips were washed in PBS, water, dried, and embedded with Prolong Diamond (Thermo Fisher Scientific). For nuclear staining DAPI (Sigma) was added to the normal staining or embedding medium supplemented with DAPI (Thermo Fisher Scientific) was used. For parallel immunostainings of Borg3 and different septins (all relevant antibodies produced in rabbit), the FlexAble kit for primary antibody labeling was used according to manufacturer’s protocol. Critical stainings were verified by a staining according to Hua and Ferland⁵⁶ with PFA in Cytoskeleton buffer.

For floating fixation that preserves the upright orientation of cilia, whenever reliable identification of the ciliary base was necessary, cells were seeded in μ-slides (Ibidi), and all washing, fixation, and staining solutions were applied with continuous addition and simultaneous aspiration. This ensured a larger-than 5-fold volume exchange at each step, preventing the cells from drying and maintaining the cilia in a floating state. At the end of the procedure, the cells remained in PBS.

Imaging

Cells were analyzed with an Axio Observer microscope (Carl Zeiss), driven by Visiview (Visitron) imaging software with plan-apochromat objectives, a Yokogawa CSU-X1 spinning disk confocal head with emission filter wheel, 405-, 488-, 561-,640 nm laser lines, and a Coolsnap HQ II digital CCD camera or prime bsi sCMOS camera (Teledyne Photometrics). Alternatively, cells were imaged with a LSM800 confocal microscope (Carl Zeiss), multiplex airyscan/GaAsP-detectors and 405-, 488-, 561-,640 nm laser lines or an Elyra 7 structured illumination super resolution microscope with dual PCO edge sCMOS cameras. For live-cell imaging, cells were incubated in a chamber with humidified atmosphere (5.5% CO₂) at 37°C on the microscopes mentioned above.

For real-time quantification of cilia formation, a Lionheart FX automated microscope (Agilent BioTek) with environmental control was used.

Image processing and analysis

SIM images were processed with ZEN Black software. The theoretical optical transfer function given by the manufacturer was used. ‘Baseline shift’ was deactivated for all reconstructed images as a standard, and thus the threshold was set manually into the first peak appearing in the histogram for representative images. No gray values of the histogram were influenced or cut off for further quantitative analysis. Maximum-resolution enhancement by minimal artifact structure (‘hammer finish artifacts’) was targeted by comparing background patterns with structure signals for representative images.

All SIM images in figures are maximum intensity projections derived from the apical z-planes of the cell. Approximately 1/3 (∼25 planes) of the z stack (z-plane distance ∼100nm) was used.

All microscopy images were further processed with Imaris or Metamorph software.

To quantify formation of cilia in fixed cells, cilia and nuclei of all cells per field of view were counted. Nuclei were stained with DAPI and cilia were stained with acetylated tubulin as well as Arl13b. The number of cilia in relation to the total number of nuclei was evaluated. For transfected cells, only the total number of transfected cells with cilia were counted.

Strong overexpression of Borg3 can negatively affect cilia formation, cause aberrant septin filament formation, and alter cell morphology. To minimize these effects, the transfected DNA amount was reduced (400 ng/mL). Live cells were sorted to select low-expressing populations through rigorous forward- and side-scatter gating, with the highest 20% of expressing cells excluded to prevent intensity outliers. In fixed-cell experiments, high-intensity outliers were also excluded based on morphology and intensity measurements. For intensity quantifications of immunofluorescence stainings a 2D- or 3D mask was generated by a ciliary marker (i.e., ARL13B or acetylated tubulin). The mask was applied to the channel of interest to exclusively measure fluorescence intensities at the cilium. The mean fluorescence intensity for each primary cilium was used for subsequent analyses.

For quantitative mapping of fluorescence intensities at the base of cilia, a 20x20 pixel square was centered at the base of the cilium. Intensities of all 400 pixel were measured and summed up for each pixel coordinate of the 15 cilia analyzed. All pixel values were compared by one way ANOVA with Tukeyś posttest to identify statistically significant elevations in fluorescence intensity in the averaged image. The generated 3D-blot for each pixel gives an overview where fluorescence signal accumulated.

In order to identify formed cilia in real time long-term Video microscopy a fluorescence intensity- and size-dependent filter was used to discriminate cell debris from newly formed cilia. Only intensities following the first peak in the histogram were evaluated and subsequently only counted if they covered an area of 4–90 pixels in size.

Fluorescence recovery after photo bleaching

IMCD cell lines containing shRNAs or a non-targeting control were transfected with Sept6-GFP or Sept6-tom as to not interfere with the fluorescence needed to identify shRNA expressing cells. Primary cilia were identified by SiR-tubulin and an area of 2 μm by 2 μm centered on the primary cilium was bleached with the corresponding laser until 50% of the fluorescence intensity was lost. Recovery was observed for 10 min inside the primary cilium using the SiR-tubulin fluorescence as positioning and size marker. During imaging an automated focus was used. Bleaching was performed using the corresponding laser for the transfected protein (560 nm for Sept6-tom and 488 nm for Sept6-GFP) with 10–15% laser intensity as needed from the fluorescence intensity observed before bleaching. Bleaching was stopped automatically once 50% of initial intensity was bleached.

Quantification and statistical analysis

Student’s t test was applied when two groups with normal distribution had to be compared. Whenever more than 2 groups were compared a Oneway ANOVA with Tukey posttest was used. Statistical evaluation was performed with GraphPad Prism. p values < 0.05 were considered statistically significant and marked with an asterisk (∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.005; ∗∗∗∗p < 0.001).

Published: May 12, 2025