Rab8 attenuates Wnt signaling and is required for mesenchymal differentiation into adipocytes

Differentiation of mesenchymal stem cells into adipocyte requires coordination of external stimuli and depends upon the functionality of the primary cilium. The Rab8 small GTPases are regulators of intracellular transport of membrane-bound structural and signaling cargo. However, the physiological contribution of the intrinsic trafficking network controlled by Rab8 to mesenchymal tissue differentiation has not been fully defined in vivo and in primary tissue cultures. Here, we show that mouse embryonic fibroblasts (MEFs) lacking Rab8 have severely impaired adipocyte differentiation in vivo and ex vivo. Immunofluorescent localization and biochemical analyses of Rab8a-deficient, Rab8b-deficient, and Rab8a and Rab8b double-deficient MEFs revealed that Rab8 controls the Lrp6 vesicular compartment, clearance of basal signalosome, traffic of frizzled two receptor, and thereby a proper attenuation of Wnt signaling in differentiating MEFs. Upon induction of adipogenesis program, Rab8a- and Rab8b-deficient MEFs exhibited severely defective lipid-droplet formation and abnormal cilia morphology, despite overall intact cilia growth and ciliary cargo transport. Our results suggest that intracellular Rab8 traffic regulates induction of adipogenesis via proper positioning of Wnt receptors for signaling control in mesenchymal cells.

Differentiation of mesenchymal stem cells into adipocyte requires coordination of external stimuli and depends upon the functionality of the primary cilium. The Rab8 small GTPases are regulators of intracellular transport of membrane-bound structural and signaling cargo. However, the physiological contribution of the intrinsic trafficking network controlled by Rab8 to mesenchymal tissue differentiation has not been fully defined in vivo and in primary tissue cultures. Here, we show that mouse embryonic fibroblasts (MEFs) lacking Rab8 have severely impaired adipocyte differentiation in vivo and ex vivo. Immunofluorescent localization and biochemical analyses of Rab8a-deficient, Rab8b-deficient, and Rab8a and Rab8b double-deficient MEFs revealed that Rab8 controls the Lrp6 vesicular compartment, clearance of basal signalosome, traffic of frizzled two receptor, and thereby a proper attenuation of Wnt signaling in differentiating MEFs. Upon induction of adipogenesis program, Rab8a-and Rab8b-deficient MEFs exhibited severely defective lipid-droplet formation and abnormal cilia morphology, despite overall intact cilia growth and ciliary cargo transport. Our results suggest that intracellular Rab8 traffic regulates induction of adipogenesis via proper positioning of Wnt receptors for signaling control in mesenchymal cells.
Adipocytes are differentiated mesenchymal cells that store excess food energy as fat in organelles known as lipid droplets. Metabolic disorders such as obesity disrupt lipid homeostasis and adipogenesis by inducing hypertrophy of existing adipocytes or by increasing proliferation to form new adipocytes to cope with excess fat (1,2). Adipogenesis starts with the specification of mesenchymal stem cells into committed preadipocytes that go on to terminally differentiate into adipocytes (3)(4)(5)(6). Aberrant induction of adipogenesis may be a factor in developing obesity. It is thus imperative to fully understand intracellular and intercellular pathways that regulate adipogenesis.
A hallmark of adipocyte maturation is the formation, trafficking, and fusion of lipid droplets, all processes linked to the Rab family of small GTPases (7). Approximately 30 Rab proteins have been shown to be required for lipid droplet formation and fusion (7,8). The two Rab8 mammalian isoforms, Rab8a and Rab8b, are regulators of anterograde membrane trafficking (9,10), and they function in regulating cell shape, migration, apical and basolateral trafficking, and docking of secretory vesicles to the plasma membrane (11)(12)(13)(14)(15). Rab8 activity is regulated by the guanine nucleotide-exchange factor, Rabin8, and GTPase-activating proteins (14,16). During cilia formation, Rab8 is localized to the primary cilium and may direct vesicle docking and fusion to the cilium base (13,17). As a polarized trafficking regulator, the movement of proteins up the primary cilium may also depend on Rab8 (11,13,14,(17)(18)(19)(20). Rab8 also interacts with key ciliary regulators such as the BBSome (17) and intraflagellar transport (IFT) particles to traffic cargoes within the cilium and may regulate the length of the cilium (13,21). Rab8's role in adipogenesis has not been fully defined in vivo or in primary MEFs.
Whether Rab8-mediated membrane transport affects mesenchymal cell differentiation remains incompletely defined. We show here that MEFs deficient in Rab8a, Rab8b, or both proteins exhibit a severe impairment in activating the adipogenesis program. Mechanistically, Rab8 deficiency alters the intracellular Lrp6 vesicular compartment, Fzd2 membrane trafficking, and Wnt signaling.

Rab8a-deficient MEFs show an increased cellular response to Wnt stimulation
We first investigated the role of Rab8 on induction and activation of Wnt/β-catenin signaling components in Rab8a −/− MEFs. Using the TopFlash luciferase reporter as a readout for Wnt/β-catenin transcriptional activity, Rab8a −/− MEFs displayed increased sensitivity to Wnt3a stimulation (Fig. 1A, data represent six independent experiments). WT MEFs exhibited a 4.9-fold Wnt3a-induced TopFlash activity, whereas Rab8a −/− MEFs exhibited a 30-fold induction of reporter activity.
Wnt engagement with surface receptors on ligand-receiving cells elicits rapid formation of LRP6 signalosomes (40), which are biochemically characterized by a surge of cellular pLRP6 levels due to phosphorylation of LRP6's intracellular domain by CK1γ and GSK3β (54,55). We next tested if Rab8a affects the temporal induction of pLRP6 in serum-starved cells after Wnt3a stimulation. In untreated (serum-starved) Rab8a −/− MEFs, the abundance of pLRP6 (normalized relative to the total LRP6) was approximately 14-fold of that in WT MEFs (Fig. 1, B and C). Upon Wnt3a treatment, pLRP6 levels progressively increased over 4 h after stimulation in both WT and Rab8a −/− MEFs (Fig. 1, B and C, data represent four independent experiments).
These data suggested that loss of Rab8a in MEFs appeared to elevate Lrp6 signalosome activity that was accompanied by an enhanced intracellular signaling response. To examine at what molecular level Rab8a deficiency enhanced the signaling cascade, GSK3-mediated signaling inhibition was first assessed by using CHIR99021, the GSK3 inhibitor (56). Rab8a −/− MEFs exhibited 9-fold increase in TopFlash activity in response to CHIR99021 compared with WT MEFs (Fig. 1E). We then overexpressed a constitutively active β-catenin (β-cat-ΔN) construct lacking the GSK3β phosphorylation sites (57) and found an equivalent induction of TopFlash activity in WT and Rab8a −/− MEFs (Fig. 1F). These data suggested that the enhanced Wnt signaling in Rab8a −/− MEFs occurred upstream of β-catenin, possibly at the cell surface and endocytic compartment.

Rab8 deficiency affects LRP6 signalosome
The two mammalian Rab8 isoforms, Rab8a and Rab8b, share 83% amino acid sequence homology and both participate in apical cargo transport (58). To examine whether Rab8 regulates Lrp6 vesicular compartment and MEF's signaling capacity in response to Wnt ligands, we first established stable Rab8b knockdown (KD) MEFs and an MEF line deficient in both Rab8a and Rab8b ( Fig. 2A). Rab8b KD efficacy was verified by Western blot even after multiple passages (Fig. 2B). We first examined Lrp6 intracellular localization in WT and Rab8-deficient MEFs in the presence or absence of exogenous Wnt ligands, by transiently expressing a GFP-Lrp6. In serumstarved WT MEFs treated by the vehicle, vesicular Lrp6 was rare, with approximately two Lrp6 + puncta per cell on average. Upon Wnt3a addition for 15 min, vesicular Lrp6 increased by approximately 9-fold (Fig. 2, C and D). In contrast, vehicletreated Rab8a −/− , Rab8b KD , and Rab8a −/− ;Rab8b KD MEFs had significantly increased vesicular Lrp6 compartments compared with WTs, and the total numbers of Lrp6+ puncta did not increase further upon Wnt3a addition (Fig. 2, C and D).
The above results suggested a changed basal Lrp6 compartmentalization in Rab8-deficient MEFs. To biochemically examine Lrp6 intracellular distribution, we performed sucrose gradient sedimentation on lysates of MEFs upon 4 h of Wnt3a or vehicle treatments, essentially following a previous study (40). In vehicle-treated WT MEFs, Lrp6 was primarily found in fractions 8 to 10, and pLRP6 was barely detected in any of the fractions even after a longer exposure (Fig. 2E, data represent three independent experiments). In Wnt3astimulated WT MEFs, Lrp6 was detected in fractions 6 to 11, while pLrp6 became detectable in fractions 8 to 11, where Gsk3, endocytic markers, and caveolin were found to be cosedimented (Fig. S1A).
Examination of Rab8b KD MEFs reached a similar but less robust effect on pLrp6, as Rab8a −/− MEFs (Fig. 2G). Owing to the detection of basal pLrp6 in these MEFs, we examined localization of Lrp6 puncta in starved cells and found that in addition to intracellular puncta, there were significantly increased numbers of peripheral Lrp6 + puncta in Rab8deficient MEFs (Fig. 2, H and I, data represent six independent experiments). These results collectively suggested that there was an altered Lrp6 vesicular compartment that potentially contributed to a sensitized MEF response to Wnt ligands.
Before adipogenic induction, Rab8a −/− MEFs contained small intracellular lipid droplets that were of a similar size to those of WT MEFs (Fig. 3, C and D). To test if Rab8a deficiency might impact MEFs' response to adipogenic stimuli, we treated Rab8a −/− MEFs with the same induction cocktail. Fluorescence staining did not detect the formation of large lipid droplets in Rab8a −/− MEFs after treatment with the differentiation protocol (Fig. 3, E and F). The average diameter of WT droplets increased by over 20-fold upon differentiation, while the droplets in Rab8a −/− MEFs increased by less than 2fold ( Fig. 3, C-F, data represent five independent experiments).
Western blots showed that induced WT cultures had a decreased abundance of total β-catenin and Tcf1 compared with uninduced WT MEFs, indicating reduced canonical Wnt signaling upon induction (Fig. 3, G and H, data represent three independent experiments). In contrast, induced Rab8a −/− MEFs exhibited elevated abundances of both total β-catenin and Tcf1 (Fig. 3, G and H). These data demonstrated that the impaired Rab8 −/− MEF differentiation was accompanied by aberrantly elevated Wnt signaling, an observation consistent with the above biochemical analysis ( Figs. 1 and 2).
Examination of the subcutaneous adipose tissues in newborn (P0) Rab8a −/− mice revealed a clearly reduced subcutaneous adipose layer compared with WT littermate pups (Fig. 4, A and C). Although there was the presence of subcutaneous cells marked by aP2 (red, also known as Fabp4) in Rab8a −/− mice, these cells did not appear to contain the characteristic lipid droplets that were densely packed as the aP2 subcutaneous adipocytes in WT mice (Fig. 4, B and D). Similar observations for aP2-labeled cells were made in anatomically matched adipocyte tissues adjacent to skeletal Figure 2. Loss of Rab8a led to an elevated pLrp6 and an altered vesicular Lrp6 compartment. A, western blot showed Rab8b knockdown efficiency in Rab8b KD and Rab8a −/− ;Rab8b KD MEFs. Note Rab8a and Rab8b are similar in size but could be distinguished as two isoform-specific bands. B, western blot using lysates of MEFs after 16 passages showed continued Rab8b KD. β-Actin was used as a loading control. C, serum-starved WT, Rab8a −/− , Rab8b KD , and Rab8a −/− ;Rab8b KD MEFs were transiently transfected with 0.5-μg pCS2-LRP6 GFP for 16 h. Cells were then stimulated by vehicle (Dulbecco's modified Eagle's medium) or Wnt3a for 15 min. Indirect immunofluorescence for GFP was performed to locate LRP6. Note that LRP6 puncta were rare in vehicletreated WT MEFs but became prominent in Wnt3a-stimulated cells. D, the numbers of LRP6 puncta were manually counted from individual cells for each condition. Experiments were repeated five times. *p < 0.05; **p < 0.01; ***p < 0.001 when compared with vehicle-treated WT. E-G, sucrose density cell fractionation assays were performed on serum-starved WT, Rab8a −/− , or Rab8b KD MEFs stimulated with Wnt3a or vehicle. Western blots for total Lpr6 or pLrp6 were performed. Longer exposure of pLrp6 blot for WT in panel E showed minimal but detectable signal in unstimulated condition. These data represent at least three independent experiments. H and I, serum-starved WT, Rab8a −/− , Rab8b KD , and Rab8a −/− ;Rab8b KD MEFs were transiently transfected with pCS2-LRP6 GFP for 24 h. Cells were then stained for LRP6-GFP and phalloidin. The numbers of peripheral LRP6 puncta (arrows), based on phalloidin staining, were manually counted for individual cells of designated genotypes. *p < 0.05; **p < 0.01; ***p < 0.001, when compared with WT. MEFs, mouse embryonic fibroblasts.
Rab8 controls MEF differentiation muscles (Fig. 4, E and F). Together, these results suggested that Rab8a is necessary for adipose tissue development in vivo and ex vivo.

Loss of Rab8a and Rab8b diminished lipid droplets
Rab8b KD MEFs exhibited an overall similar Lrp6 profile as Rab8a −/− MEFs. Thus, we next examined whether Rab8b KD MEFs had a similar adipogenic defect. Similar to Rab8a −/− MEFs, the differentiation of Rab8b KD MEFs was also impaired: there was a prolonged delay, taking approximately 13 days, to start the formation of significantly smaller lipid droplets than WT cultures (Fig. 5, A-C, F and G).
To further test if there remained some redundancy between Rab8a and Rab8b, we tested adipogenic capacity of Rab8a −/ − ;Rab8b KD MEFs. Rab8a −/− ;Rab8b KD MEFs, upon induction by the same protocol, failed to attach to glass coverslips around day 5 and 6 of the differentiation protocol. To ameliorate this attachment issue, we coated glass coverslips with 0.1% gelatin. After 2 weeks of induction, Rab8a −/− ;Rab8b KD MEFs only formed scattered, small lipid droplets, with a significantly decreased total BODIPY-stained lipid area compared with Rab8a −/− or Rab8b KD MEFs (Fig. 5G). However, the diameter or the size of lipid droplets in Rab8a −/− ;Rab8b KD cultures did not differ from those in Rab8a −/− or Rab8b KD MEFs (Fig. 5F). Owing to defects in primary cilia, MEFs lacking Kif3a failed to undergo adipogenesis (36), so we used Kif3a −/− MEFs as a reference to assess the degree of adipogenic impairment of Rab8a −/− ;Rab8b KD MEFs. Quantification of the lipid-droplet size and total area indicated a similar extent of impairment in Rab8a −/− ;Rab8b KD MEFs as Kif3a −/− MEFs (Fig. 5, F and G), which also displayed a delayed formation and fewer lipid droplets (Fig. 5E). These data suggested a defective activation of the adipogenic program in Rab8-deficient MEFs, which was further supported by significantly reduced transcripts of adipocyte differentiation, including Fabp4 and Glut4 in these Rab8a −/− ;Rab8b KD MEFs compared with WT MEFs (Fig. 5H). PPAR-γ expression was only slightly reduced in Rab8a −/− ; Rab8b KD MEFs.

Intact cilia induction and maintenance in the absence of Rab8
Adipogenesis requires the transient induction of the primary cilium in the early stages of differentiation (3), a process requiring Kif3a (16,36). Our above data suggested that loss of Rab8 shared some common characteristics with the loss of Kif3a (Fig. 5, E-G), in disrupting MEF differentiation into adipocytes. As Rab8 has been implicated in ciliary cargo transport and cilia development (11-13, 17, 18), and the primary cilium inhibits Wnt/β-catenin signaling activity (2,25,27), we then examined if Rab8-deficient MEFs might exhibit defects in primary cilia development and functionality, none of which has been examined during induced MEF adipogenesis.
Rab8a −/− ;Rab8b KD MEFs exhibited the shortest primary cilia before serum starvation, followed by Rab8a −/− or Rab8b KD MEFs, whereas WT MEFs possessed the longest primary cilia (Fig. 6C). After serum starvation, the percentage of ciliated cells and the cilium length increased in WT and the three Rab8-deficient MEF cell lines (Fig. 6C) MEFs remained nonciliated at all time points examined (Fig. 6,  A and B).
Although the absence of either single or double Rab8 did not appear to prevent the formation of the primary cilium, a fraction of Rab8a −/− and Rab8a −/− ;Rab8b KD MEFs contained multiple basal bodies and aberrant cilia (Fig. 6D, inset). Interestingly, in some Rab8a −/− ;Rab8b KD MEFs, multiple cilia grew from a single basal body, while some other Rab8a −/− ; Rab8b KD cells contained two sets of basal body and cilia (Fig. 6D, white arrowheads). These changes were rarely observed in WT or Rab8b KD MEFs. Thus, although the overall growth of a primary cilium may not require Rab8, the loss of both Rab8a and Rab8b did appear to affect basal body duplication and the morphological outcome of primary ciliogenesis.

Rab8-dependent Fzd2 traffic to primary cilia
Although Rab8 deficiency did not prevent the formation of the primary cilium in our analysis, literature suggested that there might be abnormal transport of certain ciliary cargos in the absence of Rab8 (17). We first examined Smo, a 7-pass transmembrane protein trafficked to the primary cilium to modulate Hh signaling activity, and it is moved along the cilium by IFT (65,66). Smo was properly localized to the primary cilia in WT, as well as Rab8a −/− , Rab8b KD , and Rab8a −/− ; Rab8b KD MEFs (Fig. 7A). Likewise, BBS1, a subunit of the Bardet-Biedl Syndrome protein complex shown to interact with Rab8, was also localized to primary cilia in WT and Rab8deficient MEFs (Fig. 7B). Thus, ciliary transport of Smo and BBS1 was not affected by Rab8 deficiency in MEFs.

Rab8 controls MEF differentiation
Planar cell polarity effectors are trafficked to the base of primary cilia through a mechanism dependent upon Fuzzy recruitment of Dishevelled to Rab8 vesicles (27,29,67). The planar cell polarity activity directly intersects with Wnt signaling (68). As we have shown that Rab8-deficient MEFs had enhanced Wnt signalosome activities, we immunofluorescently stained Fzd2, a Wnt receptor, in ciliuminduced WT and Rab8-deficient MEFs. In 55% WT MEFs, Fzd2 was localized to large membrane "patches" at the base of the primary cilium before serum starvation (0 h, Fig. 7, C and E). After 24 h of starvation, patches of Fzd2, at the base of primary cilia, was detected in 88.7% of WT cells and numerous small Fzd2-stained puncta (Fig. 7, C and E). Approximately 50% of Rab8a −/− MEFs also displayed Fzd2 membrane patches localized to primary cilia and that was increased to 80% after 24 h of serum starvation. In addition, Rab8a −/− MEFs contained more fluorescent Fzd2 puncta even before serum starvation, and these puncta increased after 24 h of serum starvation (arrowheads, Fig. 7, C and E).
The above data indicated abnormal Fzd2 trafficking in the absence of Rab8 during adipogenic induction. We examined Fzd2 protein in various MEFs by Western blots before and after adipogenic induction. Full-length Fzd2 appeared in all cell lines before and after adipogenic induction (solid arrowhead, Fig. 7F). Interestingly, we observed the induction of a truncated Fzd2 that was prominent in Rab8a −/− and Rab8a −/− ; Rab8b KD MEFs, and to a lesser extent in Rab8b KD (empty Rab8 controls MEF differentiation arrowhead, Fig. 7F), suggesting that lack of Rab8 led to a possible post-translationally modified Fzd2. Rab8a −/− and Rab8a −/− ;Rab8b KD MEFs also showed a concomitant induction of Tcf1 that was not seen in WT or Rab8b KD cells (Fig. 7, F and G, data represent three independent experiments). Together, these data collectively suggested that Rab8 plays a critical role in regulating the membrane positioning of Wnt signaling components and proper MEF differentiation in response to adipogenic signals.

Discussion
We used genetic and biochemical approaches to demonstrate the impact of Rab8 deficiency on the morphogenesis, differentiation, and signaling of mouse embryonic fibroblasts. Our data suggested a Rab8-dependent membrane traffic of signaling receptors and intracellular positioning of signalosomes in Wnt-receiving MEFs. As attenuation of Wnt signaling is essential for adipogenesis (22,69), the aberrant activation of this pathway in Rab8-deficient MEFs is likely responsible for the severe impairment of MEF differentiation into adipocytes.
Rab8a-deficient MEFs were hypersensitive to Wnt ligand stimulation. In serum-starved Rab8a −/− MEFs, there were higher basal pLrp6, the core unit of the Lrp6 signalosome, at the plasma membrane and intracellularly. As pLrp6 assembles downstream signaling components, the increased positioning of Lrp6 vesicular machinery at the plasma membrane and in the cytoplasm could be poised to induce a robust signaling output upon ligand stimulation. This may explain the enhanced ligand sensitivity and signaling in Rab8a −/− MEFs. Active Rab8driven anterograde membrane trafficking may help diminish Lrp6 intracellular aggregation. Thus, we propose that in Wntreceiving MEFs, normal Rab8 membrane trafficking may allow the clearance of the Lrp6 signalosome when there is minimal extracellular ligand, thereby preventing unwanted activation of Wnt signaling. Such Rab8 trafficking appears to be particularly important for Wnt signaling attenuation during adipogenic induction and for MEFs to differentiate into adipocytes. This is

Rab8 controls MEF differentiation
conceptually important as this Rab8-dependent mechanism may offer additional insight into how Wnt-β-catenin signaling is controlled in differentiating cells.
Our work adds to the existing literature that documented the regulation of Wnt-β-catenin signaling by Rab11, Rab14, Rab25, and Rab27 (70)(71)(72)(73). In terms of regulating adipogenesis of MEFs, Rab8a and Rab8b exhibit similar contributions. A redundancy between the two factors was revealed in Rab8a −/− ; Rab8b KD MEFs, which showed a dramatic reduction of the formation of lipid droplets even after a prolonged differentiation phase. However, Rab8a and Rab8b do display differences with regard to cilia development and Fzd2 transport. In these scenarios, lack of Rab8a induced a stronger phenotype than the lack of Rab8b. In zebrafish, Rab8b contributes to Lrp6 endocytosis and Wnt signaling (74). Exogenous Wnt stimulation is known to induce Lrp6 internalization to lipid raft domains to sequester Axin and prevent β-catenin degradation (75,76). Internalized Lrp6 vesicles have been colocalized with Rab7 endosomes and mature into multivesicular bodies to sequester GSK3-β to sustain Wnt-β-catenin signaling (77). Our sedimentation analysis showed an impact of Rab8 deficiency on endocytic compartmentalization, suggesting that Rab8 vesicular traffic is essential to maintain homeostasis in the network of endocytosis and exocytosis. This homeostasis must be critical for the activation and inhibition of Wnt-β-catenin signaling. MEFs. C and D, representative immunofluorescent images of Fzd2 (red) in MEFs before or 24 h after cilia induction. A diagram is used to summarize the findings in WT cells: Fzd2 appeared as a cilium-associated patch before induction; after induction, numerous Fzd2 puncta or vesicles appeared (red dots). White arrowheads point to increased Fzd2 puncta that were not associated with a cilium in Rab8a −/− and Rab8a −/− ;Rab8b KD MEFs before induction. Open white arrowheads point to cilia that were not associated with a patch of Fzd2. E, quantification of the percentage of cells with Fzd2 localized to the base of the primary cilium at 0 (blue bar) or 24 h (red bar) after cilia induction. F, western blots for Fzd2, Tcf1, and β-catenin were performed on total lysates of WT or Rab8-deficient MEFs under uninduced or induced conditions. Full-length Fzd2 was marked by a solid black arrowhead; truncated Fzd2 marked by an empty arrowhead; cleaved fragment marked by an asterisk. G, fold changes in Tcf1 protein abundance between uninduced and induced MEFs were quantified from three independent experiments. p values were determined by t test. MEF, mouse embryonic fibroblast.
Mesenchymal stem cells are pluripotent cells capable of differentiating into various cell fates including the fat, bone, and cartilage. Wnt-β-catenin signaling inhibits adipogenesis, and several Wnt ligands such as Wnt3, Wnt6, Wnt10a, and Wnt10b have been identified as adipogenic inhibitors (1,3,22,23). It should also be noted that Wnt3a induced de novo lipiddroplet formation in L cells and hepatocytes (78). In undifferentiated mesenchymal stem cells, Wnt-β-catenin signaling was shown to inhibit the expression of proadipogenic factors such as C/EBPs and PPAR-γ (79). When these cells are induced to differentiate, C/EBP-α, C/EBP-β, and PPAR-γ expression levels increase and inhibit nuclear translocation of β-catenin; the inhibition of nuclear localization of β-catenin allows cells to initiate an adipogenic transcriptional program (79). Suppression of PPAR-γ signaling by noncanonical Wnt signaling and tumor necrosis factor-alpha/Interleukin-1 signaling steers cells toward an osteogenic fate by suppressing adipogenesis, thus highlighting how tight regulation of these cell-fate determinants maintains tissue homeostasis (1,3,4,6,22,36). These studies are largely consistent with our observation that an enhanced Wnt signaling in Rab8-deficient MEFs impaired adipogenic potential. We also found that Rab8 abundance is increased in mature adipocytes (8) and we speculate it may function to attenuate Wnt signaling, to allow adipocyte differentiation and maturation.
We also observe an interesting localization of Fzd2 to a disk of membrane at the base of the primary cilium. In Drosophila, Fzd2 has been implicated in a nuclear import pathway during synaptic development in neuromuscular joints. In that model, Fzd2 is endocytosed and translocated to the perinuclear area where it is cleaved (80)(81)(82). The C-terminus-containing fragment of Fzd2 is then translocated to the nucleus via importin-β11, importin-α2, and GRIP (a PDZ protein) (81,82). It was also proposed that nuclear Fzd2 may act as a transcription factor to regulate synaptic development (82). We observed that in MEFs, there was a redistribution of Fzd2 upon adipogenic induction, and this redistribution appeared to correlate with a prominent processing (possibly a cleavage event) of this receptor into a smaller fragment. A distinct processing product was detected prominently in Rab8a −/− and Rab8a −/− ;Rab8b KD MEFs upon induction, suggesting that loss of Rab8a affected Fzd2 processing. We currently do not know the significance of this change, but the aberrant Fzd2 processing in the two above-mentioned MEF lines appeared to correlate with Tcf1 induction and, thereby, Wnt pathway activity. Collectively, Rab8 appears to be critical in the maintenance, distribution, and compartmentalization of key Wnt pathway molecules (i.e., Fzd and LRP6). Future studies will elucidate the precise molecular mechanisms of such regulation.
Transient formation and disassembly of primary cilium during adipogenesis (3,5,24) may be correlated with Wnt signaling and β-catenin degradation (2,25,27). Ciliary proteins are known interacting partners of Rab8. For example, the BBSome directly interacts with Rab8 at the basal body during primary cilium formation and elongation (17). Polycystin-1, a ciliary targeting signal that regulates cell polarization, also complexes with Rab8, and its trafficking is dependent on Rab8 and Arf6 (35). IFT particles are scaffolded with Rab8 through a protein called Elipsa during trafficking events within the cilium itself (21). Rab8b interacts with otoferlin, a crucial ciliary protein in cochlea; loss of otoferlin leads to hearing impairment in patients. The otoferlin and Rab8b complex is speculated to regulate tethering and fusion of endosomes (32). While depletion of either Rab8a or Rab8b was shown to be sufficient to inhibit ciliogenesis in some studies (13,14,18,35), we found that Rab8a-and Rab8b-deficient MEFs formed intact cilia (58). Rab8-deficient MEFs only showed some ciliary morphology defects, namely fewer ciliated cells, slightly shorter length, and multiciliated cells. In addition, the formation of the primary cilia was not grossly attenuated in cells depleted for both Rab8a and Rab8b. Nevertheless, Rab8a −/ − ;Rab8b KD MEFs exhibited the strongest defect in adipogenesis, to an extent similar to Kif3a −/− cells that had both impaired ciliogenesis and adipogenic potential (16). We postulate that the assembly of signaling components by Rab8 at the plasma membrane, and not the formation of a primary cilium, may thus be a critical determinant of adipogenesis. The molecular control of adipocyte maturation by Rab proteins may be exploited in the future to better understand how fat tissue is dysregulated in disease and to guide the design of effective therapeutic strategies.

Mice, MEF cell isolation, and culture
Animal studies were approved by the Institutional Animal Care and Use Committee of Rutgers University. Mice carrying the Rab8a null allele have been described previously (15,41). Rab8a+/− mice were set up for plug mating. On day E12.5, the pregnant female was sacrificed, and the embryos were dissected out. After removing the placenta, yolk sac, and uterus, the head and internal organs (the heart, liver, spleen, gut) were removed. Tails were saved for genotyping. The remaining tissue was placed into 1.5-ml Eppendorf tubes with trypsin-EDTA. With sterile scissors, the tissue was diced as finely as possible and incubated in a 37 C humidified chamber for 10 min. After suspension, cells were pelleted and plated in Dulbecco's modified Eagle's medium (DMEM) containing sodium pyruvate, 10% fetal bovine serum (FBS), 1.0 mg/ml Pen-Strep, and 0.05 mg/ml gentamicin. Cells were passaged three times to isolate fibroblasts before starting experiments.

Cloning and lentiviral KD of Rab8b
Rab8b-specific lentiviral shRNA construct targeting against the 3'UTR of mouse Rab8b was constructed by inserting the annealed complementary oligonucleotides (5'-CCGGGCCAAGAACTAACAGAACTTTCCATGGAAAG TTCTGTTAGTTCTTGGCTTTTTG-3' and 5'-AATTCAA AAAGCCAAGAACTAACAGAACTTTCCATGGAAAGTT CTGTTAGTTCTTGGC-3') into pLK0.1 lentiviral vector (Addgene) between AgeI and EcoRI sites. For viral packaging, this lentiviral vector along with pVSV-G was transfected into GP2-293 cells (Clontech) using Lipofectamine 2000 (Invitrogen). After 48 h, the supernatant was collected and subjected to ultracentrifugation (15,000g, 2 h) for viral concentration. The viral pellet was resuspended with 200 μl of DMEM and aliquoted for later usage. For Rab8b KD, MEFs were infected with diluted lentivirus stock (1:50,000) for 5 h in DMEM in the presence of polybrene (8 μg/ml) and then incubated with complete DMEM containing 10% FBS for another 24 h. Puromycin (3 μg/ml) was added into the culture medium for selection. KD efficiency and the maintenance of KD were confirmed by Western blotting.

Sucrose gradient sedimentation
This method is according to the previous study (40). Briefly, after starvation for 16 to 24 h, WT Rab8a −/− and Rab8b kd MEF cells were treated with Wnt3a (100 ng/ml) for 4 h before harvest. Cells were harvested in Hank's Balanced Salt Buffer on ice, pelleted, and lysed for 20 min in an extraction buffer which contains 30-mM Tris (pH 7.4), 140-mM sodium chloride, 1% Triton X-100, 25-mM sodium fluoride, 3-mM sodium orthovanadate, 2-mM PMSF, and protease inhibitor cocktail tablet (Roche). The lysates were centrifuged, and the supernatant was layered on top of a 15 to 40% sucrose gradient 30-mM Tris (pH 7.4), 140-mM sodium chloride, 0.02% Triton X-100, 25-mM sodium fluoride, 3-mM sodium orthovanadate, and protease inhibitors. Ultracentrifugation was performed in a Beckman SW55Ti rotor at 45,200 rpm for 4 h at 4 C. After centrifugation, fractions were collected from the bottom of the tube by a peristaltic pump and analyzed by SDS-PAGE and immunoblot.

Immunofluorescence
For BODIPY 493/503 staining, the powder was dissolved in 100% ethanol to a final concentration of 1 mg/ml. In the dark, the differentiated cells plated on coverslips were fixed in 4% PFA for 10 min, washed in 1x high-quality PBS twice, and incubated in 0.1% Triton X-100 for 10 min at RT. Cells were then incubated in BODIPY 493/503 (1:1000 of stock in 1x PBS) for 30 min at RT and counterstained with TO-PRO-3 (1:500 in 1x PBS) for 15 min at RT. Coverslips were dried Rab8 controls MEF differentiation and mounted onto slides using the ProLong Anti-fade reagent (Thermo Fisher Scientific) and sealed with clear nail polish. Imaging was performed using a Zeiss LSM 510 Confocal Microscope.
For ciliogenesis, cells were plated in 6-well plates containing coverslips. All cells were grown and maintained in overconfluency for 48 h and then incubated in a serum-free medium for 0, 2, 24, and 48 h, respectively. On the day of staining, cells were fixed in methanol at −20 C for 5 min. Permeabilization occurred in 0.1% Triton X-100 in PBS for 10 min at RT. Cells were blocked in 10% goat serum in PBS for 30 min at RT and then incubated with primary antibodies: acetylated α-tubulin (1:800; T7451, Sigma-Aldrich), γ-tubulin (1:800; 84355, Abcam) in 10% goat serum in PBS, at 4 C overnight. On the second day after washing with PBS, cells were incubated with the fluorescently conjugated secondary antibody (1:1000; Thermo Fisher Scientific) in the dark for 1 h at RT. After washing in PBS, cells were counterstained with TO-PRO-3 (1:500; T3605, Thermo Fisher Scientific) in PBS for 15 min at RT. Coverslips were dried and mounted onto slides with the ProLong Anti-fade reagent (P36930, Thermo Fisher Scientific) and sealed with clear nail polish. Imaging was performed using the Zeiss LSM 510 confocal microscope.
Immunostaining was performed as described previously (41,43,45). Cells were fixed in 4% paraformaldehyde, blocked in 2% bovine serum albumin, 2% goat serum, and 0.1% Triton X-100 blocking buffer for 1 h, and incubated with primary antibodies, acetylated α-tubulin (1:500; T7451, Sigma-Aldrich) and γ-tubulin (1:500; #ab84355, Abcam), in the blocking buffer overnight at 4 C. On the following day after washing with PBS, sections were incubated with a fluorescently conjugated secondary antibody (1:1000; Thermo Fisher Scientific) in a blocking buffer for 1 h at RT. Sections were counterstained for nuclei with TO-PRO-3 (1:500; T3605, Thermo Fisher Scientific) in PBS for 15 min at RT and then washed and mounted with the ProLong Gold Anti-fade mounting media (P36930, Thermo Fisher Scientific). Images were taken using a Zeiss LSM 510 confocal microscope.
For LPR6 immunofluorescence detection, cells were transiently transfected with pCS2+LRP6 GFP (83). At 100% confluency, cells were serum-starved for 24 h, followed by treatment with Wnt3a (100 ng/ml, diluted in DMEM) or vehicle only. After 15 min of treatment, cells were fixed in 4% PFA and indirect immunofluorescent straining was subsequently performed as described previously for GFP (1:100; 8334, Santa Cruz Biotechnology) in 4 C overnight. Images were taken using a Zeiss LSM 510 confocal microscope. The numbers of LRP puncta, or peripheral puncta (based on phalloidin staining), were manually counted from individual cells of 6 to 8 different fields. Data represent a minimum of five independent experiments.

Quantification and statistics
All results represent three or more independent experiments unless stated otherwise. Western blots, dimension of lipid droplets, and the cilia length were measured by ImageJ (NIH, version 1.49). For each cell line, ciliated cells were counted manually against total cells from six lowmagnification confocal images. The lipid-droplet number and size were quantified using particle analysis in ImageJ. Data represented the means ± SEM from independent experiments. Statistical analysis was performed by Student's t test, one-way ANOVA, or two-way ANOVA. Significance was accepted at p < 0.05. Calculations and graphs were generated using GraphPad Prism (7.04).

Data availability
All data are contained within the article.
Supporting information-This article contains supporting information.
Conflict of interest-The authors declare that they have no conflicts of interest with the contents of this article.