Loss of OCRL increases ciliary PI(4,5)P₂ in Lowe oculocerebrorenal syndrome

Oculocerebrorenal syndrome of Lowe (MIM: 309000) is a rare X-linked recessive disorder that presents in young males with bilateral cataracts and glaucoma, as well as renal failure, muscular hypotonia, and mental retardation. The defective gene, OCRL (also known as OCRL1), encodes an inositol polyphosphate 5-phosphatase (Attree et al., 1992; Zhang et al., 1995), which acts on phosphatidylinositol-4,5-bisphosphate [PI(4,5)P₂] (Jefferson and Majerus, 1995; Zhang et al., 1995; Zhang and Majerus, 1998). Over 210 mutations in OCRL have been found in patients affected with Lowe syndrome, most of which are located in the phosphatase domain or the C-terminal RhoGAP domain. Decreased 5-phosphatase activity is found in fibroblasts from Lowe patients as well as a 2- to 3-fold elevated ratio of PI(4,5)P₂:PI4P.

PI(4,5)P₂ levels can affect a wide range of cellular processes, including membrane curvature, signaling scaffold, ion channel functions, and actin cytoskeleton (Gamper and Rohacs, 2012; Zhang et al., 2012). Internal cellular membranes also have distinct phosphoinositide composition that can define the subcellular space (Di Paolo and De Camilli, 2006; Pirruccello and De Camilli, 2012). The unique lipid composition is controlled by kinases and phosphatases that localize to these subcellular compartments, such as OCRL. Mutations in OCRL are known to cause cellular defects in endocytosis (Nández et al., 2014; Vicinanza et al., 2011), endosomal trafficking (Billcliff et al., 2016; Cauvin et al., 2016; Noakes et al., 2011; Swan et al., 2010; van Rahden et al., 2012), actin cytoskeletal rearrangements (Coon et al., 2009; Faucherre et al., 2005; Grieve et al., 2011), autophagy (De Leo et al., 2016), cytokinesis (Dambournet et al., 2011) and primary cilia signaling (see review by Mehta et al., 2014). We and others have shown that the gene product, OCRL, localizes to the primary cilia (Coon et al., 2012; Luo et al., 2012; Rbaibi et al., 2012).

Until recently, the primary cilium has been recognized as a distinct organelle with a unique lipid composition (Rohatgi and Snell, 2010). Consisting of a basal body and an axoneme, the primary cilium has a tightly regulated barrier for lipids and protein (Hu et al., 2010; Jensen et al., 2015). Several groups have shown, that in Joubert syndrome, mutation of another inositol 5-phosphatase, INPP5E, results in PI(4,5)P₂ accumulation in the cilia (Chaáez et al., 2015; Garcia-Gonzalo et al., 2015; Xu et al., 2016). The loss of INPP5E in Joubert syndrome has been found to affect G-protein-coupled receptor trafficking and downstream sonic hedgehog signaling.

Here, we present evidence that phosphoinositide levels are dysregulated in primary cilia of Lowe syndrome patients. We found that loss of OCRL resulted in abnormal distribution of PI(4,5)P₂ in the proximal regions of cilia. Re-expression of OCRL restored the balance between PI(4,5)P₂ and PI4P. We examined the ciliary phosphoinositides in cells from human and mouse models of Lowe syndrome, showing abnormal PI(4,5)P₂ distribution in the primary cilium and differential effect of sonic hedgehog signaling in response to agonistic stimulation in mouse fibroblasts.

Based on the previously described distribution of OCRL in the cilia, we hypothesized that cells derived from patients with Lowe syndrome may also exhibit dysregulation of ciliary phosphoinositides. Using a monoclonal antibody against PI(4,5)P₂, we first examined the distribution of PI(4,5)P₂ within the ciliary membrane of fibroblasts from two unrelated Lowe syndrome patients and one control fibroblast line (Luo et al., 2014). Primary cilia formation was induced by serum starvation for 48 h, and ciliary markers of acetylated α-tubulin and PI(4,5)P₂ were assessed in fibroblasts. While the normal human fibroblasts (NHF558) did not show any PI(4,5)P₂ along the cilium, Lowe 3265 fibroblasts exhibited accumulation of PI(4,5)P₂ along the length of the cilia, as well as increased staining near the base of the cilia (Fig. 1A). Similarly, Lowe 1676 cells showed PI(4,5)P₂ staining along both the base and the membrane portion of the cilia. Average PI(4,5)P₂ ciliary intensity was markedly higher in Lowe patients fibroblasts than controls (Fig. 1B). Only 4% of PI(4,5)P₂ intensity was observed in the cilia of NHF558 cells, compared with 30% in Lowe 3265 and 40% in Lowe 1676 cells.

To confirm the difference of PI(4,5)P₂ in Lowe patients and normal human fibroblasts, all three fibroblast cell lines were transfected with phosphoinositide-specific pleckstrin homology (PH) domain proteins: four-phosphate-adaptor protein 1 (FAPP1-PH), which is a marker for PI4P; and phospholipase C-δ1 (PLCδ-PH), which is a marker for PI(4,5)P₂ (Hardie et al., 2015; Lenoir et al., 2015). Consistent with the ciliary staining pattern by the monoclonal antibody for the endogenous forms of PI(4,5)P₂, we observed a build-up of PI(4,5)P₂ and a decrease of PI4P in the cilia of two Lowe fibroblast samples (Fig. 1C). PLCδ-PH levels were significantly higher in cilia of both Lowe fibroblast types, whereas FAPP1-PH levels showed a 25% presence in cilia of NHF558 cells and 12% in Lowe fibroblasts (Fig. S1), supporting the pattern observed for endogenous PI(4,5)P₂ build-up in the Lowe syndrome fibroblasts.

To further examine the role of OCRL in ciliary phosphoinositide regulation, we turned to an Ocrl-deficient mouse model (Bothwell et al., 2011) (phenotype summarized in Fig. S2). In this mouse model (Ocrl^−/−, Inpp5b^−/−, INPP5B^+/+, also known as IOB^−/− mice), human type II inositol polyphosphate 5-phosphatase (INPP5B), a paralog of OCRL, which was implicated to play a compensatory role in vivo (Jänne et al., 1998; Bothwell et al., 2010; Luo et al., 2013; Matzaris et al., 1998; Speed et al., 1995), was introduced to rescue the embryonic lethality in Ocrl^−/− Inpp5b^−/− double knockout mice (Jänne et al., 1998). Bothwell et al. (2011) reported that the IOB^−/− mice developed renal tubular disease consistent with that of Lowe syndrome patients compared with the littermate control (Ocrl^+/Y, Inpp5b^−/−, INPP5B^+/+; IOB^+/Y mice).

We hypothesized that in the mice that have deficiency in both OCRL and INPP5B, there would be an increase in PI(4,5)P₂ in the cilia and a corresponding decrease in PI4P. IOB^−/− mouse embryonic fibroblasts (MEFs) were isolated, cultured and serum starved to induce primary cilia formation. Phosphoinositide distribution was examined by immunofluorescence using monoclonal antibodies against PI4P and PI(4,5)P₂ (De Leo et al., 2016; Garcia-Gonzalo et al., 2015). We observed an increased level of PI(4,5)P₂ in the proximal regions of the cilia and very little ciliary PI4P (Fig. 1D,E). As a critical control, IOB^+/Y MEFs have a functional Ocrl gene present; in these cells, we observed positive staining of PI4P in the cilia and a corresponding absence of PI(4,5)P₂ in both the transition zone as well as along the ciliary axoneme. Furthermore, Ocrl^−/− MEFs, which have been previously shown to express wild-type INPP5B (Bothwell et al., 2011), had PI(4,5)P₂ in both the transition zone as well as along the length of the cilium, whereas wild-type MEFs had PI4P localized to the ciliary membrane along the axoneme. Taken together, the accumulation of PI(4,5)P₂ in OCRL-deficient cells, both human and mouse, indicates a critical role for OCRL in phosphoinositide regulation.

Based on the localization of OCRL to the proximal regions of the cilium (Fig. S3), we then sought to determine the subcellular distribution using specific markers within the cilia. The previous pattern of cilia membrane staining suggests a transition zone pattern for PI(4,5)P₂ accumulation in Lowe fibroblasts. To test this hypothesis, TCTN1 was used for co-immunostaining of the transition zone.

TCTN1 is a transmembrane protein localized at the transition zone that forms complexes with other ciliary proteins (Garcia-Gonzalo et al., 2011; Yee et al., 2015). Distribution of TCTN1 was detected at the ciliary base of RPE cells, with immunoreactivity tightly organized to the transition zone. Upon primary cilia induction in fibroblasts, Arl13b and γ-tubulin were used to stain for the axoneme and the base of the cilia. In both Lowe 3265 and, to a lesser extent, Lowe 1676 fibroblasts, an extension of PI(4,5)P₂ into the proximal regions of the cilia was seen, as confirmed with TCTN1 staining (Fig. 2A).

To further corroborate the distribution of PI(4,5)P₂ in Lowe patient cells, MKS5 immunostaining was performed. Merkel syndrome type 5 (MKS5) has been recently shown to localize within the transition zone of cilia, and to regulate PI(4,5)P₂ distribution in Caenorhabditis elegans (Jensen et al., 2015). Therefore, we hypothesized that the PI(4,5)P₂ distribution defect may be secondary to abnormal localization of MKS5. However, MKS5 remained localized to the transition zone along with the accumulation of PI(4,5)P₂ in Lowe fibroblasts (Fig. 2B), suggesting that either MKS5 and OCRL do not associate along the same regulatory pathway of PI(4,5)P_2, or there is a difference in MKS5 physiology between human and C. elegans cilia. Taken together, these data support that the loss of OCRL results in a build-up of PI(4,5)P₂ at the transition zone and proximal regions of the ciliary membrane.

To determine whether the elevated PI(4,5)P₂ levels is caused by OCRL, we then assessed whether the expression of wild-type OCRL can rescue the ciliary phosphoinositide accumulation defects in patient fibroblasts. Reduction of OCRL in Lowe patient cells was previously measured (Luo et al., 2012). Transfection of FLAG-tagged OCRL, either wild type, an enzymatic dead mutant of OCRL (D499A), or an intact phosphatase but defective mutant (F276S) was performed in NHF558, Lowe 3265 and Lowe 1676 cells (Chabaa et al., 2006; Song et al., 2017). Primary cilia were induced by serum starvation, followed by immunostaining for Arl13b and PI(4,5)P₂ antibodies. Consistent with the hypothesis that OCRL expression directly underlies the increased levels of PI(4,5)P₂, introduction of wild-type but not the D499A or F276S mutant form of OCRL was able to deplete ciliary PI(4,5)P₂ (Fig. 3A). PI(4,5)P₂-positive ciliary immunostaining decreased from 45% to 8% in Lowe 3265 cells, and from 44% to 8% in Lowe 1676 cells (Fig. 3B). Moreover, a significant decrease in ciliary PI(4,5)P₂ level was observed in both Ocrl-null and IOB^−/− MEFs (Fig. 3C,D).

Overall, elevated PI(4,5)P₂ levels in the absence of OCRL suggest that OCRL is required to control the normal level of PI(4,5)P₂. Several lines of evidence support the role of OCRL as a regulator of PI(4,5)P₂ in cilia. First, OCRL has higher affinity for PI(4,5)P₂ than PI(3,4,5)P₃, with a lower dissociation constant (KD) compared with INPP5E (Kisseleva et al., 2000). Second, in vivo assays for PI(4,5)P₂ function [e.g. actin cytoskeletal rearrangement (Faucherre et al., 2005; Coon et al., 2009), cytokinesis (Dambournet et al., 2011), endocytosis (Erdmann et al., 2007; Vicinanza et al., 2011)] indicated that OCRL affects PI(4,5)P₂ levels directly. Third, loss of OCRL results in increased global levels of PI(4,5)P₂ in Lowe syndrome patients (Ungewickell and Majerus, 1999). Fourth, expression of OCRL in the Lowe syndrome mouse model shows a rescue of the phenotype for PI(4,5)P₂ accumulation in the cilia. Taken together, these findings support the idea that specific phosphoinositides such as PI(4,5)P₂ have natural low levels of localization to the ciliary compartment in normal cells, with OCRL playing a critical role in their regulation. In this regard, deficiency in OCRL causes defects in PI(4,5)P₂ regulation and accumulation within the ciliary compartment, as seen in our study.

In addition to OCRL, recent work by several groups has shown that INPP5E is also present in the cilia and is a key phosphatase for PI(4,5)P₂, which is upregulated in the INPP5E MEFs, C. elegans, Drosophila and in human knockdown cells (Chávez et al., 2015; Garcia-Gonzalo et al., 2015). Studies show that accumulation of the INPP5E phosphoinositide substrates PI(4,5)P₂ and PI(3,4,5)P₃ was accentuated at the transition zone in hedgehog (Hh)-stimulated Inpp5e^−/− cells. Interestingly, these were further associated with reduced ciliary smoothened (Smo) levels which could be rescued by expression of wild-type, but not 5-phosphatase-dead INPP5E (Dyson et al., 2017), suggesting a ciliary gatekeeper role for INPP5E, and possibly OCRL, in the transition zone of primary cilia.

In this regard, we treated Lowe syndrome and NHF558 fibroblasts with an agonist for hedgehog signaling (SAG) overnight, followed by serum starvation for 24 h, and stained for GPR161 and Arl13b. There was a significant reduction of GPR161 in both mutant and normal cells after treatment with SAG (Fig. 4A,B). Furthermore, NIH3T3 fibroblasts and IOB^−/− fibroblasts were transfected with Smo–mCherry–GCamp3 (Delling et al., 2013). After treatment with SAG overnight and 24 h serum starvation, only the NIH3T3 fibroblasts exhibited a significant increase in Smo levels, without restoration to the IOB^−/− fibroblasts (Fig. 4C,D). This discrepancy between normal and mutant cells suggests that OCRL deficiency impairs normal response to SAG in promoting Smo levels to the primary cilium. Previous reports have shown that loss of INPP5E curbs the response to Hh transcriptional activity (Garcia-Gonzalo et al., 2015). Our work adds to the complexity of this picture by suggesting a role of OCRL in regulation of proteins near the ciliary base, such as Smo, as opposed to INPP5E which plays a regulatory role to ciliary GPR161.

In summary, our results suggest that Lowe syndrome cells exhibit defects in primary cilium function or regulation. These, in turn, may contribute to abnormal aqueous humor filtration and/or renal tubular filtration. Many studies have shown a close involvement of primary cilia in different settings. Of most interest is their potential role as flow sensors and mechanotransducers to regulate tissue homeostasis in different organs, for example, in eye, brain and kidneys. In this regard, we suggest that OCRL plays a crucial role in regulating PI(4,5)P₂ levels in the primary cilium and that deficiency causes an alteration of downstream pathways, which ultimately affects fluid transport in the affected organ systems.

Anti-Arl13b mouse antibody (75-287) was purchased from Antibodies Inc (David, CA). Anti-γ-tubulin rabbit antibody (620901) was purchased from Biolegend (San Diego, CA). Anti-acetylated α-tubulin rabbit antibody (5335) was purchased from Cell Signaling Technology (Danvers, MA). Anti-TCTN1 (15004-1-AP), anti-Arl13b (17711-1-AP), anti-MKS5 (55160-1-AP), rabbit polyclonal antibodies were purchased from ProteinTech. Anti-PI4P (Z-P004) and PI(4,5)P₂ (Z-P045) antibodies were obtained from Echelon (Salt Lake City, UT). Anti-Smoothened (ab72130), was purchased from Abcam (Cambridge, MA). All primary antibodies were used at 1:500 dilution, except for PI(4,5)P₂ and Smoothened (1:200). Secondary antibodies Alexa Fluor 488 (A11008) and Alexa Fluor 594 (A11012) anti-rabbit IgG, and Alexa Fluor 488 anti-mouse IgG+IgM (1857664) and Alexa Fluor 546 anti-mouse IgG (A11030) were procured from Life Technologies/Invitrogen (Carlsbad, CA). All secondary antibodies were applied at 1:1000. SAG (11914) was purchased from Cayman Chemical (Ann Arbor, MI). Treatment of SAG was conducted overnight at 400 nM in Opti-MEM.

Three human fibroblast (NHF558, Lowe3265 and Lowe1676) and human retinal pigmented epithelial (RPE) were gathered as previously described (Luo et al., 2014). Ocrl^−/− and Inpp5b^−/− Ocrl^−/− INPP5B^+/+ mice were a gift from Robert L. Nussbaum (University of California, San Francisco). NIH3T3 mouse fibroblasts were a gift from Jeffrey Goldberg (Stanford University). All MEFs were derived from E14.5 embryos without brain, limbs and internal organs. Embryos were washed with sterile PBS, minced into small pieces with sterile razor blade, digested with trypsin in 37°C for 15 min, homogenized by pipetting, and plated in a 10 cm dish with DMEM and 10% FBS. All animal experiments followed the guidelines of the Association for Research in Vision and Ophthalmology Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care and Use Committee of Indiana University School of Medicine.

Citrine-tagged FAPP-PH and Citrine-tagged PLCδ-PH constructs were gifts from Clark Wells (Indiana University, USA) (Heller et al., 2010). Smo–mCherry–GCaMP3 was a gift from David Clapham (Harvard University, USA).

For immunofluorescence, cells were washed with PBS and fixed with 4% paraformaldehyde in PBS for 20 min at RT. Cells were washed 3 times with PBS, permeabilized in PBST (0.5% Triton X-100 in PBS) for 10 min at RT, and blocked with 10% NGS blocking buffer for 1 h at 37°C. After another wash in PBS, cells were incubated with the respective primary antibody overnight at 4°C. After 3 washes, cells were incubated with secondary antibody for 1 h at 37°C, washed 3 times, stained with DAPI, washed 4 times, and sealed in anti-fade reagent. Ciliary Arl13b or acetylated α-tubulin served as markers for primary cilia. Using ImageJ (v1.47v, NIH), each cilium in the ciliary channel and merged channels and area and total fluorescence intensity were measured. The percentages of PI4P and PI(4,5)P₂ intensity in cilia were calculated by (total ciliary intensity in merged channels – ciliary channel)/total intensity in ciliary channel×100.

Results are expressed as mean values±s.d. Statistical analysis was performed using Student t-tests or ANOVA as appropriate (SPSS). A P-value of less than 0.01 was considered statistically significant.

We thank Drs Timothy Corson, Michael Boulton, and John Spence for thoughtful comments during the preparation of this manuscript, Dr Robert Nussbaum for the gift of Lowe syndrome mouse model and Dr Clark Wells for the phosphoinositide fluorescent protein constructs.