Hypoxia-Inducible Factor-2alpha Affects the MEK/ERK Signaling Pathway via Primary Cilia in Connection with the Intraflagellar Transport Protein 88 Homolog

Abstract

The ability of cells to communicate with their surrounding is a prerequisite for essential processes such as proliferation, apoptosis, migration, and differentiation. To this purpose, primary cilia serve as antennae-like structures on the surface of most mammalian cell types. Cilia allow signaling via hedgehog, Wnt or TGF-beta pathways. Their length, in part controlled by the activity of intraflagellar transport (IFT), is a parameter for adequate function of primary cilia. Here we show, in murine neuronal cells, that intraflagellar transport protein 88 homolog (IFT88) directly interacts with the hypoxia-inducible factor-2α (HIF-2α), hitherto known as an oxygen-regulated transcription factor. Furthermore, HIF-2α accumulates in the ciliary axoneme and promotes ciliary elongation under hypoxia. Loss of HIF-2α affected ciliary signaling in neuronal cells by decreasing transcription of Mek1/2 and Erk1/2. Targets of the MEK/ERK signaling pathway, such as Fos and Jun, were significantly decreased. Our results suggest that HIF-2α influences ciliary signaling by interacting with IFT88 under hypoxic conditions. This implies an unexpected and far more extensive function of HIF-2α than described before.

INTRODUCTION

The ability of a cell to transmit signals from the environment to the cell interior describes a prerequisite for the realization of cellular processes such as proliferation, apoptosis, migration and differentiation.^1,2 Primary cilia receive signals on the cell surface and transmit them to the interior implementing signal transduction. Almost all mammalian cell types, including neural tissue,^3,4 have microtubule-based cilia as antenna-like structures on their surface.⁵ Primary cilia are fundamental for cellular signaling and responsible for the initiation of signaling pathways, e.g., by hedgehog,^6–9 Wnt,^10,11 platelet-derived growth factor (PDGF)¹² and transforming growth factor-beta (TGF-β).¹³ Disruption of signal transmission in these pathways will lead to a multitude of serious human diseases and developmental disorders, called ciliopathies. Well-known ciliopathies are the Joubert syndrome (JBTS), the Senior–Løken syndrome (SLS), nephronophthisis (NPHP), the Meckel–Gruber syndrome (MKS), or the Bardet-Biedl syndrome (BBS).¹⁴ Ciliopathies often present with malformations of the central nervous system, such as forebrain disorders, hydrocephalus or exencephaly.^15,16 Interestingly, the formation and function of primary cilia has already been linked to the oxygen supply of the cell.^17,18 Sufficient oxygen supply is required for proper cell function and survival. When oxygen supply falls short, cells must quickly react and provide relevant signals for appropriate counter-regulations to avoid hypoxic damage. Key factors in the hypoxic response are hypoxia-inducible factors (HIFs). HIFs act as transcription factors, are heterodimers of an α- and a β-subunit and are constitutively expressed. However, α-subunits are degraded by the proteasome under normoxic but stabilized under hypoxic conditions.¹⁹ The activation of the HIF pathway plays an essential role in the developing nervous system.^20,21 Especially the HIF-2 isoform promotes neurogenesis and has a protective effect on neuronal stem cells.^22,23 Through its influence on the physiological formation of new nerve cells and the production of apoptosis inhibitors, HIF-2 is a relevant factor in the regeneration of cerebral pathologies.²⁴ So far it is only known for chondrocytes, that HIF-2α accumulates in the axoneme of primary cilia by pharmacological HIF stabilization via DMOG.²⁵

We now show localization of HIF-2α in the axoneme of murine neuronal primary cilia under hypoxic conditions. The regenerative abilities of HIF-2α suggest that it can improve the outcome of diseases such as stroke by means of signals mediated by primary cilia. Therefore, we propose that HIF-2, which otherwise acts as a transcription factor, plays a functional role in the signal transmission of primary cilia.

RESULTS

HIF-2α accumulates under hypoxia in the axoneme of primary cilia in neuronal cells

Proteins, which may to cross the ciliary transition zone (TZ), mostly have a distinctive function in primary cilia. They are essential for ciliary biosynthesis or proper cilia signaling or are transported to the ciliary tip to be shed in extracellular vesicles. We detected HIF-2α by immunofluorescence staining at the ciliary base and in the ciliary axoneme of neuronal cells (Fig. 1). Interestingly, HIF-2α increasingly accumulated in the axoneme under hypoxic conditions (Fig. 1), whereas under normoxia we detected significantly more at the ciliary base. In about 75% of primary cilia at 21% O₂, we found HIF-2α at the proximal ending of the axoneme and in 10%, the HIF-2α staining correlated with the marked axoneme. In contrast, at 1% O₂ 30% of cilia showed an accumulation of HIF-2α at the ciliary base and approximately 45% inside the axoneme.

FIG 1.

Accumulation of HIF-2α in different compartments of primary cilia. (A) Counting of HIF-2α positive cilia in percentage terms revealed that at 21% O₂ HIF-2α protein mostly accumulated at the ciliary base, whereas under hypoxic conditions (1% O₂), significantly more HIF-2α was detected in the ciliary axoneme. (B) Immunofluorescence on cells migrated out of murine neurospheres. Acetylated α-tubulin in the ciliary axoneme is stained in green, HIF-2α is stained in red and γ-tubulin in stained in blue. The scale bar represents a length of 500 nm. Data are presented as mean ± SD; **P < 0.01, ***P < 0.001; ****P < 0.0001; n = 3 (>20 cilia per sample).

Ciliary elongation under hypoxia is restricted without HIF-2α due to limited IFT88

Malformation of cilia could be first evidence of impaired ciliary signal transduction. To probe for possible abnormality in the ciliary structure, ciliary length was measured in wild type (Hif-2a^+fl/+fl) and Hif-2α-knockout (Hif-2a^−/−) cells (Fig. 2A and B). Under normoxic conditions, ciliary length of Hif-2a^+fl/+fl and Hif-2a^−/− neuronal cells did not differ. Normoxic cilia had a length of about 0.9 ± 0.3 µm. Under hypoxia, the Hif-2a^+fl/+fl cilia significantly elongate to an average length of 1.25 ± 0.4 µm. The Hif-2a^−/− cilia, however, showed no significant change in length under hypoxic conditions.

FIG 2.

Loss of HIF-2α impairs ciliary length and reduces IFT88 content in the ciliary axoneme. (A, C) Immunofluorescence on murine neurosphere derived cells. (A) The ciliary axoneme is stained in green for acetylated α-tubulin and in red for Arl13b. The ciliary base is stained in blue for γ-tubulin. (B) Normalized ciliary length. Cilia are significantly longer in Hif-2a^+fl/+fl cells under hypoxia (1% O₂). At least 50 cilia per mouse (n = 3) were used for quantification. (C) The ciliary axoneme is stained in green for acetylated α-tubulin and IFT88 is stained in red. The scale bar represents 500 nm. (D) Normalized ciliary amount of IFT88. Significantly more IFT88 accumulated in the ciliary axoneme in Hif-2a^+fl/+fl cells under hypoxia (1% O₂). At least 20 cilia per mouse (Hif-2a^+fl/+fl n = 3; Hif-2a^−/− n = 5) were used for quantification. (E) Western blot and quantification of IFT88. IFT88 whole cell protein was significantly decreased in Hif-2a^−/− cells under hypoxic conditions (n = 3). IFT88 gene expression was not changed under any condition (Hif-2α^+fl/+fl n = 5; Hif-2α^−/− n = 7). (F) Co-immunoprecipitation of HIF-2α and IFT88. Whole-cell extracts were prepared and immunoprecipitated (IP) using an antibody against HIF-2α and immunoblotted (IB) using an antibody against IFT88 (n = 3–4). Data are presented as mean ± SD; *P < 0.05, ****P < 0.0001.

In order to identify a possible cause for the lack of elongation of the Hif-2a^−/− cilia under hypoxia, in a next step we investigated the biosynthesis of cilia. The protein IFT88 is essential for proper cilia synthesis and acts as a transport protein for anterograde transport. Here we observed similar effects as with the ciliary length. At 21% O₂, the relative amount of IFT88 in the cilium was not different between the genotypes (Fig. 2C to D). Under hypoxic conditions (1% O₂), however, the amount of IFT88 significantly increased in the ciliary axoneme when compared to the amount of IFT88 in Hif-2a^−/− cells. The total protein amount of IFT88 in the cell was also significantly decreased in Hif-2a^−/− cells at 1% O₂ compared to wild-type cells (Fig. 2E). Of note,IFT88 gene expression was not changed (Fig. 2E). To demonstrate a direct interaction between HIF-2α and IFT88, co-immunoprecipitation with whole-cell extracts was performed. Proteins were immunoprecipitated (IP) using an antibody against HIF-2α and immunoblotted (IB) using an antibody against IFT88. Under hypoxia, we were able to detect interaction between HIF-2α and IFT88 (Fig. 2F). Under normoxia, only a weak signal was detected.

HIF-2α regulates Mek1/2 and their downstream genes

In order to determine the functional effects of altered ciliary length and reduced IFT88 abundance in primary cilia, we considered different signaling pathways associated with ciliary transduction. For the MEK/ERK signaling pathway, we found genes upstream of Mek unchanged under either normoxic or hypoxic conditions (Fig. 3). However, Mek and Erk themselves and all investigated downstream genes were increased at 1% O₂ in Hif-2a^+fl/+fl cells (Fig. 4). In contrast, under hypoxia Hif-2a^−/− cells showed significantly reduced Mek and Erk expression compared to the wild-type cells. Likewise, the targets of MEK/ERK signaling Jun and Fos lacked hypoxic induction in the absence of HIF-2α (Fig. 5). To prove the critical role of primary cilia for MEK/ERK signaling pathway, we mechanically removed cilia by shear force. Without cilia, the target genes Jun and Fos of the MEK/ERK signaling pathway were not induced under hypoxia. This effect was comparable to Hif-2a^−/− cells with cilia (Fig. 5). In case of Mmp13, hypoxic induction remained in the absence of cilia.

FIG 3.

Quantitative PCR analysis of genes involved in MEK/ERK signaling upstream of Mek. Expression of genes upstream of Mek was not changed in Hif-2a^−/− cells. Data are presented as mean ± SD; Hif-2a^+fl/+fl n = 5; Hif-2a^−/− n = 7.

FIG 4.

Quantitative PCR analysis of Mek1/2 and Erk1/2 and Western blot of phospho-MEK1/2. (A) Expression of Mek1/2 and Erk1/2 was significantly decreased under hypoxia (1% O₂) in Hif-2a^−/− cells compared to Hif-2a^+fl/+fl cells (Hif-2a^+fl/+fl n = 5; Hif-2a^−/− n = 7). (B) Western blot and quantification of phospho-MEK1/2. Phospho-MEK1/2 is significantly decreased in Hif-2a^−/− cells under hypoxic conditions (n = 3). Data are presented as mean ± SD; *P < 0.05, **P < 0.01.

FIG 5.

Quantitative PCR and Western blot analysis of factors involved in MEK/ERK signaling downstream of Mek. (A) Expression of MEK/ERK target genes (Mmp13, Jun, Fos) was significantly decreased under hypoxia (1% O₂) in Hif-2a^−/− cells compared to Hif-2a^+fl/+fl cells. This effect on the expression of Jun and Fos under hypoxia was not observed without cilia (Hif-2a^+fl/+fl n = 5; Hif-2a^−/− n = 7; Hif-2a^+fl/+fl (no cilia) n = 3). (B) This observation could also be made in Western blot analysis of Jun and Fos (n = 3). Data are presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

DISCUSSION

Primary cilia are regulators for a variety of signaling pathways, ensuring cell development, survival and regeneration. Especially for proliferation and maintenance of neural progenitor cells, primary cilia take on a critical role.^26,27 In order to fulfill this variety of important functions, many proteins are located in the ciliary axoneme. The import and export of these proteins into and from cilia is of decisive importance and is realized—among others—by IFT proteins.^28,29 Surprisingly, the protein HIF-2α, primarily known as a nuclear transcription factor, was detected in the ciliary axoneme due to pharmacological stabilization of HIF in chondrocytes.²⁵ So far, it is completely unclear what function HIF-2α takes on in primary cilia. However, a specific function of HIF-2α in the cilium is likely, because the transition zone, the so-called ciliary gate, and the regulated transport of the cilium ensure that no proteins are transported into the cilium that are not needed in this cell compartment.^30,31 Surprisingly, we were able to detect HIF-2α in the primary cilia of murine neuronal cells and observed transport of HIF-2α into cilia under hypoxic conditions (Fig. 1). In addition to the increased transport of HIF-2α into cilia, elongation of the ciliary axoneme was observed under hypoxic conditions (Fig. 2A and B). Interestingly, Hif-2a^−/− cells did not show this elongation. In this case, different to the previously describes elongation of primary cilia due to treatment with IL-1b, the increased cilia length is HIF-2α dependent. The length of primary cilia is always a critical measure for their proper functionality and their signaling pathways. Many human diseases, called ciliopathies, are referred to abnormal short or long cilia including brain malformations.³² For renal and hepatic cells, it has already been shown that hypoxia-induced mechanisms significantly influence the length of primary cilia and lead to the prolongation of cilia.^33,34 An elongation of neuronal primary cilia under hypoxic conditions may lead to an improved signaling reacting on hazardous low oxygen levels by increasing the surface for exogenous transmitters. This reaction could potentially prevent neuronal cells from cell death by inducing signaling pathways responsible for cell survival, e.g., the MEK/ERK signaling pathway.

Our data indicate that ciliary length in neuronal cells is dependent on HIF-2α. In contrast, chondrocytes showed primary cilia elongation independent of HIF-2α expression.²⁵ Wann et al., however, only investigated the transcriptional activity of HIF but did not consider other possible functions directly at the cilium. We believe this is necessary if one recalls the transport regulation of the transition zone, which implies a direct function in the cilia to HIF-2α. A decisive factor for the biosynthesis and elongation of cilia is the protein IFT88. IFT88 is a member of the IFT-B complex and a loss of IFT88 results in the absence of cilia.^35–37 We found significantly more IFT88 in cilia under hypoxia than under normoxic conditions (Fig. 2C and D). Gene expression of Ift88 did not change (Fig. 2E). In the case of Hif-2a^−/− cells, IFT88 did not increasingly accumulate in the ciliary axoneme. Thus we show, that HIF-2α is necessary to increase the amount of IFT88 in the ciliary axoneme under hypoxic conditions. A likely explanation for the prolongation of the cilia under hypoxia could be that HIF-2α and IFT88 directly interact and perform a common function in primary cilia. This could be the reason why we see an HIF-2α dependent cilia elongation although this mechanism is described as HIF-2α independent before.²⁵ In this study only the transcriptional activity of HIF-2 was considered and thereupon inhibited in the experiments, but if HIF-2α directly binds another protein, here IFT88, at the primary cilium, the elongation could definitely relate to HIF-2α. Co-immunoprecipitation of HIF-2α and IFT88 confirmed interaction of both proteins (Fig. 2F). Most likely, there are two binding sites in the structure of HIF-2α to realize this interaction. One is ODD domain which is responsible for binding of pVHL and the PHDs and the other is the CAD domain which enables the interaction with FIH and CBP/p300. Both binding sites are known to be regulated by oxygen-dependent hydroxylation that has been structurally characterized.^38–41

As we see an induced interaction of HIF-2α and IFT88 under hypoxia, one can assume that hydroxylation of HIF-2α blocks the interaction with IFT88. Further studies focusing on the structural implementation have to be done to shed more light on this mechanism of interaction.

Ciliary HIF-2α enhances MEK/ERK signaling

Functional testing of different ciliary signaling pathways showed no change in the target gene expression of Shh, PDGF, Wnt, Hippo, mTor or TGF-β, but with continuous investigation we found the MEK/ERK signaling pathway influenced by the loss of Hif-2α. Under hypoxic conditions, the expression of Mek1/2 and Erk1/2 was significantly increased, but genes upstream from Mek and Erk showed no change (Fig. 3 and Fig. 4). Target gene expression of the MEK/ERK signaling pathway also significantly increased (Fig. 5). Interestingly, these effects were no longer observed in Hif-2a^−/− cells, suggesting that HIF-2α actively influences the MEK/ERK signaling pathway. Activated MEK1/2 protein accumulated to the same pattern as shown by Western blot analysis (Fig. 4B). The MEK/ERK signaling pathway regulates processes of cell proliferation and survival.^42–44 These become particularly important in the event of oxygen deficiency, such as in stroke, to protect cells and tissues and to regenerate damaged tissue. Connections between IFT88 and the MEK/ERK signaling pathway have already been observed in the past, but never specifically connected to HIF. In fibroblasts, mutation in the IFT88 gene leads to reduced MEK/ERK signaling.¹² From our data, we suggest that the interaction of IFT88 and HIF-2α realizes this process (Fig. 2F). In support of this notion, cells in which the cilia were removed did not show induction of the MEK/ERK target genes Jun and Fos even under hypoxic conditions (Fig. 5). This implies that primary cilia mediate the enhanced MEK/ERK target gene expression via HIF-2. Possibly, the activation of this signaling cascade by IL-1β could be a consequence of ciliary elongation and accumulation of HIF-2α in the ciliary axoneme, as shown in chondrocytes after IL-1β treatment. More likely, however, this activation is due to the function of HIF-2α itself and the effect of IL-1β to stabilize HIF-α protein.^25,45 The fact that expression of Mmp13 without cilia was not reduced (Fig. 5) may result from indirect transcriptional control by HIF-2⁴⁶ and is therefore unlikely to be mediated by cilia.

In trophoblasts, regulation of matrix metalloproteinase (MMP) expression is attributed to the endocrine gland-derived vascular endothelial growth factor (EG-VEGF), which is formed via the HIF signaling cascade.^47,48 EG-VEGF is thought to increase the expression of MMPs via the ERK signal cascade with the receptor for VEGF located at the primary cilium. However, since the expression of EG-VEGF is limited to specific tissue types and does not occur in the brain, this mechanism is most likely not present in our neuronal cells. Moreover, it is unlikely that VEGF derived from other sources and homologous to EG-VEGF is involved in the expression of Mmp13, because cells without cilia and thus VEGF receptor should not increase Mmp13 expression under hypoxia.

In summary, these studies demonstrate an unexpected function of HIF-2α, which highlights the importance of HIF-2α in primary cilia signaling. The accumulation of HIF-2α in the ciliary axoneme and the interaction of HIF-2α with IFT88, a protein important for cilia function, shed new light on the function of HIF-2α, hitherto known as a transcription factor. The influence of HIF-2α on the MEK/ERK signaling leads to the hypothesis that HIF-2α accumulated under hypoxia affects cilia signaling activity due to its transport into the cilium (Fig. 6). This new insight into HIF-signaling and connection to primary cilia may have major implications in a variety of hypoxic pathologies including brain malformation or stroke. By increasing MEK/ERK signaling via primary cilia, HIF-2 seems to promote cell survival, which appears extremely useful in situations of acute oxygen deficiency.

FIG 6.

Hypothesis for the role of HIF-2α in primary cilia. HIF-2α is transported into primary cilia by interaction with IFT88 under hypoxic conditions. This stimulates ciliary elongation and promotes MEK/ERK signaling downstream of Mek. HIF-2α-knockout (Hif-2a^−/−) does not lead to longer cilia under hypoxia due to less IFT88 in the ciliary axoneme. Therefore, the effect on the MEK/ERK signaling pathway seen in the Hif-2a^+fl/+fl cells is lost in Hif-2a^−/− cells, as well as in cells without cilia. Created with BioRender.com.

MATERIALS AND METHODS

Animals

We used male and female C57BL/6 J mice with a neuronal-specific Hif-2α knockout realized with the Cre-loxP system (Hif-2^−/−, B6.Cg-Tg(Nes-cre)1Kln/J, purchased from the Jackson Laboratory, Bar Harbor, ME, USA). Littermates negative for CRE recombinase but Hif-2α^+fl/+fl served as control animals. All animals showed a physiological habitus and normal breeding behavior. Complete pelleted feed and drinking water were administered ad libitum. The keeping and breeding of the animals took place in compliance with the German law for animal welfare and was approved by the State Agency for Nature, Environment and Consumer Protection North Rhine-Westphalia (file reference, 84–02. 04. 2016. A173).

Primary cell culture

We generated neurospheres isolated from mice aged p1–p3 as described before.²³ Neurospheres display a suitable model to investigate cell-cell communication because of their layered structure of the most common neuronal cell types: neurons, astrocytes, oligodendrocytes, progenitor cells and stem cells. Neurospheres were cultivated in neurosphere medium (DMEM/F-12 (1:1, Thermo Fisher) containing 0.2 mg/mL L-glutamine (Merck KGaA), 2% v/v B27 supplement (Thermo Fisher), 100 U/mL penicillin and 100 μg/mL streptomycin (both Merck KGaA)) at a density of 10⁵ cells/mL with addition of 20 ng/mL epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF, PreproTech GmbH) at 37 °C in 21% O₂/5% CO₂.

Experimental procedure

Adhesion and differentiation of neurospheres were induced by special differentiation medium for 72 h (neurosphere medium containing 1% v/v FCS (Merck KGaA)) in culture dishes coated with 10 μg/mL poly-L-ornithine and 10 μg/mL 445 laminin-1 (Merck KgaA). Ciliogenesis was induced by serum starvation for 24 h simultaneously to hypoxic incubation at 1% O₂. Cilia were removed by shear force using the protocol of Mitchell⁴⁹ with an enhanced shaking time of 5 min at 400 rpm under hypoxic conditions.

Polymerase chain reaction

Total RNA was isolated from neurospheres with the NucleoSpin RNA kit (MACHEREY-NAGEL GmbH & Co. KG). cDNA was synthesized using M-MLV reverse Transcriptase (Promega GmbH) and qPCR analysis was performed with Biozym Blue S'Green qPCR-Kit (Biozym Scientific GmbH) on Bio-Rad’s CFX96™ real-time system (Bio-Rad Laboratories GmbH). Used primer pairs are shown in Table 1.

TABLE 1.

Mouse primer sequences

Gene	5432p=‘5′ Primer	3′ Primer
A-Raf	GACCATGCACAATTTTGTACGG	GGCGGTTGGTACTCATGTCAA
C-Raf	TGGACTCAAAGATGCGGTGTT	AAAACCCGGATAGTATTGCTTGT
Erk1	TCCGCCATGAGAATGTTATAGGC	GGTGGTGTTGATAAGCAGATTGG
Erk2	GGTTGTTCCCAAATGCTGACT	CAACTTCAATCCTCTTGTGAGGG
Fos	CGGGTTTCAACGCCGACTA	TTGGCACTAGAGACGGACAGA
Hras	CGTGAGATTCGGCAGCATAAA	GACAGCACACATTTGCAGCTC
Ift88	TGAGGACGACCTTTACTCTGG	CTGCCATGACTGGTTCTCACT
Jun	CCTTCTACGACGATGCCCTC	GGTTCAAGGTCATGCTCTGTTT
Kras	GCAAGAGCGCCTTGACGATA	GTCCCTCATTGCACTGTACTC
Mek1	GAGTGCAACTCCCCGTACATC	TTCTCCCGAAGATAGGTCAGG
Mek2	GTTACCGGCACTCACTATCAAC	GCTCACCGACCTTAGCCTTC
Mmp13	CTTCTTCTTGTTGAGCTGGACTC	CTGTGGAGGTCACTGTAGACT
Nras	CCTCTACAGGGAGCAAATTAAGC	CCAGTGTGTAAAAGGCATCCTC
Rpl13a	CTGTGAAGGCATCAACATTTCTG	GACCACCATCCGCTTTTTCTT
Shc4	GCAGGACTCACGCTGTATGTA	GGCTCCCTTCGTCCAAAGAC

Primer sequences of specific PCR products used for RNA quantifications of murine neurospheres.

Genes of interest were normalized to 60S ribosomal protein L13a (Rpl13a) as indicated. Expression was calculated with the 2^−ΔCT method.

Antibodies

Cells were immune-stained with primary antibodies targeting acetylated α-tubulin (#sc-23950; Santa Cruz Biotechnology), HIF-2α (#NB100-122; Novus Biologicals), IFT88 (#13967-1-AP; Proteintech Group, Inc.), and γ-tubulin (#15176-1-AP; Proteintech Group, Inc.). As fluorochromes appropriate anti-mouse, anti-rabbit and anti-goat Alexa 405, Alexa 488, and Alexa 568 antibodies (Invitrogen) were used.

Immunofluorescence

For immunofluorescence, cells were fixed with 4% paraformaldehyde after experimental procedure. Fixed cells were rinsed with phosphate-buffered saline (PBS), followed by 10 min incubation with PBS/0.5% Triton X-100 for permeabilization. After washing with PBS, the samples were incubated in PBST (PBS/0.1% Triton X-100) containing 10% donkey serum (blocking solution) for at least 60 min at room temperature. Diluted primary antibodies in PBST were incubated overnight at 4 °C. Following three washing steps, incubation with fluorescent secondary antibodies (diluted in PBST) was performed at room temperature for 1 h. Before embedding with Mowiol (optionally containing DAPI), samples were washed several times in PBS.

For triple immunofluorescence staining of primary antibodies from the same host species, fab fragment antibody had to be used to circumvent this issue. Cells were fixed, washed and permeabilized as described above. Primary antibody (anti γ-tubulin) was incubated overnight in PBST at 4 °C. Cells were washed and incubated overnight with goat fab anti-rabbit antibody, following washing with PBST and incubation with secondary antibody (anti-goat Alexa 405) diluted in PBST. Cells were blocked for 1 h in normal rabbit serum and stained with the other primary antibodies as described above.

Image acquisition

Image acquisition was carried out using a Zeiss Axiovert 200M microscope (Carl Zeiss AG), a monochrome charge-coupled device camera (Axiocam 305, Carl Zeiss AG), and the Zen 3.0 Software (Carl Zeiss AG). Three single-plane images per cilium were obtained in an 8-bit grayscale modus respectively covering the specific spectrum of the used fluorochrome.

Co-immunoprecipitation and western blot

Co-immunoprecipitation (Co-IP) was performed with 500 µg protein with a total volume of 500 µL per sample. Lysates were incubated with primary antibody or appropriate control IgG overnight at 4 °C on a rotating device. Protein A/G PLUS-Agarose (#sc-2003; Santa Cruz Biotechnology, Inc.) was added and incubated for another 2 h at 4 °C rotating. Beads were pelleted by centrifugation at 11,000 rpm for 2 min at 4 °C. Supernatant was discarded and pellet washed seven times with protein lysis buffer, each time repeating the centrifugation step above. After the final wash, the supernatant was discarded and pellet resuspended and boiled for 10 min in 25 µL 4× SDS buffer. The following Western blot analysis was performed as previously described⁵⁰ with 20 μg protein for the input control. The following antibodies were used: HIF-2α (#NB100-122; Novus Biologicals), IFT88 (#13967-1-AP; Proteintech Group, Inc.), α-tubulin (#sc-8035; Santa Cruz Biotechnology), phospho-MEK1/2 (Ser217/221) (#9121; Cell Signaling Technology), JUN (#24909-1-AP; Proteintech Group, Inc.), Fos (#NB110-75039; Novus Biologicals). Proteins were detected with secondary antibodies conjugated to peroxidase (Merck KgaA) and visualization of protein bands was done on a Fusion FX (Vilber).

Quantification and presentation

Ciliary length measurements and protein staining were quantified using ImageJ software (National Institutes of Health). Measurement of protein intensity based on immunofluorescence staining was performed as described before.^30,51–53 Triplets of 8-bit single-plane grayscale images were merged via ImageJ. The ciliary length has been considered while quantifying the ciliary amount of IFT88 in different genotypes. Therefore, the area marked by acetylated α-tubulin was used as a reference and the average pixel intensity of IFT88 was quantified. Unspecific staining was excluded from the measurements by subtracting the mean value of the average pixel intensity of three neighboring regions free from specific staining. Representative images were processed via Adobe Photoshop CS2 after quantification by means of background subtraction and contrast settings.

Statistical analysis

Data are presented as mean ± SD. Analysis of variances (ANOVA) were used for all data in which more than two datasets are compared. ^*P < 0.05 was defined as statistically significant, **P < 0.01 was defined as statistically very significant, ^***P < 0.001 was defined as statistically high significant, and ****P < 0.0001 was defined as statistically extremely significant. Sample sizes are biological replicates and indicated in the figure legends. All statistical data analysis and graphical illustrations were carried out with GraphPad Prism 8 (GraphPad Software). Datasets that support the statistical analyses are openly available.⁵⁴

Glossary

ABBREVIATIONS

BBS

Bardet-Biedl syndrome

Co-IP

Co-immunoprecipitation

EG-VEGF

Endocrine gland-derived vascular endothelial growth factor

HIF-2α

Hypoxia-inducible factor 2α

HIFs

Hypoxia-inducible factors

IB

Immunoblotted

IFT

Intraflagellar transport

IFT88

Intraflagellar transport protein 88 homolog

IP

Immunoprecipitated

JBTS

Joubert syndrome

MKS

Meckel–Gruber syndrome

MMP

Matrix metalloproteinase

NPHP

Nephronophthisis

Rpl13a

60S ribosomal protein L13a

SD

Standard deviation

SLS

Senior–Løken syndrome

TGF-β

Transforming growth factor-beta

TZ

Transition zone

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are openly available in “figshare” at https://doi.org/10.6084/m9.figshare.21975563.v1], reference number ⁵³.