Bardet–Biedl syndrome 3 protein promotes ciliary exit of the signaling protein phospholipase D via the BBSome

Certain ciliary signaling proteins couple with the BBSome, a conserved complex of Bardet–Biedl syndrome (BBS) proteins, to load onto retrograde intraflagellar transport (IFT) trains for their removal out of cilia in Chlamydomonas reinhardtii. Here, we show that loss of the Arf-like 6 (ARL6) GTPase BBS3 causes the signaling protein phospholipase D (PLD) to accumulate in cilia. Upon targeting to the basal body, BBSomes enter and cycle through cilia via IFT, while BBS3 in a GTP-bound state separates from BBSomes, associates with the membrane, and translocates from the basal body to cilia by diffusion. Upon arriving at the ciliary tip, GTP-bound BBS3 binds and recruits BBSomes to the ciliary membrane for interacting with PLD, thus making the PLD-laden BBSomes available to load onto retrograde IFT trains for ciliary exit. Therefore, BBS3 promotes PLD exit from cilia via the BBSome, providing a regulatory mechanism for ciliary signaling protein removal out of cilia.

The BBSome performs IFT-dependent ciliary cycling in four continuous steps: coupling with IFT at the ciliary base, entry and anterograde traffic (from the base to the tip), remodeling and turnaround at the ciliary tip, and retrograde traffic (from the tip to the base) and ciliary exit (Eguether et al., 2014;Iomini et al., 2001;Keady et al., 2012;Lechtreck et al., 2013;Liew et al., 2014;Pedersen et al., 2006;Pedersen et al., 2005;Wei et al., 2012). During this process, BBSome cargoes, such as somatostatin receptor 3 (Ssr3) and melanin-concentrating hormone receptor 1 (Mchr1), couple with the BBSome at the ciliary base for ciliary entry (Berbari et al., 2008;Jin et al., 2010). In contrast, others like dopamine receptor 1 (D1), GPR161, and PLD load onto the BBSome at the ciliary tip for ciliary exit (Domire et al., 2011;Liew et al., 2014;Liu and Lechtreck, 2018;Ye et al., 2018). It was shown that the BBSome is recruited from the cell body to the ciliary base as the major effector of the Arf-like 6 (ARL6) GTPase BBS3 (Jin et al., 2010;Xue et al., 2020). Such a scenario could allow cells to control the amount of BBSomes available at the basal body and in turn regulate the presence and amount of signaling protein cargoes in the ciliary membrane (Jin et al., 2010;Xue et al., 2020). When in its GTP-bound state, BBS3 enters cilia (Liew et al., 2014;Xue et al., 2020) and undergoes GTPase cycling (Liew et al., 2014). Upon finishing a GTPase cycle with the aid of the Rab-like 4 (RABL4) GTPase IFT27 as a BBS3-specific guanine nucleotide exchange factor (GEF), GTP-loaded BBS3 mimics its role in the cell body to bind the cargo-laden BBSomes at the ciliary tip and loads cargoes onto retrograde IFT trains for ciliary exit, thus regulating the cargo content in the ciliary membrane (Liew et al., 2014). Currently, the precise molecular activity by which BBS3 maintains the ciliary dynamics of signaling protein cargoes through the BBSome remains to be determined, while both in vitro biochemical analysis and structural studies have implicated that membrane association of BBS3 in its active GTP-bound state is a prerequisite for signaling protein cargoes to couple with BBSomes (Jin et al., 2010;Liew et al., 2014;Loktev et al., 2008;Mourão et al., 2014;Nachury et al., 2007).
Our current limited understanding of how BBS3 and the BBSome interact for regulating signaling protein content in the ciliary membrane was derived mainly from biochemical assays performed on whole-cell extracts but not ciliary extracts of mammalian cells (Jin et al., 2010;Liew et al., 2014). Chlamydomonas reinhardtii has a clear advantage over other ciliated model organisms as its cilia can be easily isolated for biochemical analysis. Hence, we explored how BBS3 and the BBSome cross-talk in C. reinhardtii for targeting to the basal body, for entering cilia from the basal body, and for maintaining their dynamics in cilia and how they coordinate to mediate ciliary exit of PLD in C. reinhardtii (Liu and Lechtreck, 2018). By performing functional, in vivo biochemical, and single-particle in vivo imaging assays, we showed that the BBSome depends on BBS3 for targeting to the basal body but not vice versa. Upon targeting to the basal body, GTP-bound BBS3 separates from the BBSome and they enter cilia by diffusion (BBS3) and via IFT (the BBSome). Upon reaching the ciliary tip, BBS3 mediates the sorting of PLD out of cilia through promoting its coupling with the BBSome, thus filling a gap in our understanding of how BBS3 regulates ciliary removal of signaling proteins through the BBSome in C. reinhardtii.

Results
BBS3 is not required for the BBSome to enter cilia from the basal body It was previously reported that the BBSome relies on BBS3 for entering cilia in human cells (Jin et al., 2010). Our previous study showed that Chlamydomonas BBS3 is actually required for recruiting the BBSome to the basal body, thus making the BBSome available for loading onto anterograde IFT trains for ciliary entry (Xue et al., 2020). This finding raises the question of whether the BBSome, upon targeting to the basal body, requires BBS3 for entering cilia. To answer this question, we examined the so-called BBS3 CLiP mutant (LMJ.RY0402.149010) that we named clip1. The clip1 strain contains a 1923 bp paromomycin gene insertion in 3' distal region of the fifth intron of the BBS3 gene (Figure 1-figure supplement 1A-C). This insertion does not cause changes in BBS3 cDNA sequence (Figure 1-figure supplement 1D, E), and the endogenous BBS3 and the BBSome subunits BBS1 and BBS5 were maintained at wild-type levels ( Figure 1A), demonstrating that the clip1 strain is not a real BBS3-null mutant. Strikingly, BBS3 was not present in cilia of the clip1 strain, while BBS1 and BBS5 were maintained at wild-type levels in cilia, a result showing that the BBSome does not require BBS3 for entering cilia ( Figure 1B). To verify that BBS3 is not able to enter cilia in clip1 cells, we expressed BBS3 attached at its C-terminus to a yellow fluorescent protein (YFP) (BBS3::YFP) in wild-type CC-5325 and clip1 cells (resulting strains BBS3::YFP 5325 and BBS3::YFP clip1 ) ( Figure 1C). The transgenic BBS3::YFP, similar to the endogenous BBS3, was present in cilia of BBS3::YFP 5325 cells but not of BBS3::YFP clip1 cells ( Figure 1D, E). As expected, BBS1 and BBS5 were maintained at wild-type levels in whole-cell samples and cilia of both strains ( Figure 1C, D). In addition, BBS5 and BBS3::YFP were concentrated and colocalized at the basal bodies in both BBS3:: YFP 5325 and BBS3::YFP clip1 cells ( Figure 1E), and BBS3::YFP immunoprecipitated BBS1 and BBS5 in the cell body extracts of BBS3::YFP clip1 cells ( Figure 1F), suggesting that BBS3 retains its ability to bind and recruit the BBSome to the basal body in clip1 cells (Xue et al., 2020 be investigated why BBS3 concentrates at the basal bodies but is prevented from entering cilia in clip1 cells, while our data clearly show that Chlamydomonas BBS3 is not required for entry of the BBSome into cilia from the basal bodies ( Figure 1G). This notion was confirmed when BBS3 knockdown and rescue studies were performed ( Figure 7D, E).

BBS3 enters cilia without relying on the BBSome
It was previously reported that BBS3 relies on the BBSome for entering cilia in human cells (Jin et al., 2010). To investigate whether this applies in C. reinhardtii, we examined the BBS1-null bbs1-1 mutant (Lechtreck et al., 2009), in which protein levels of BBS3 and the BBSome subunits BBS4, BBS5, and BBS7 were at wild-type levels (  absent from the basal bodies in bbs1-1 cells ( Figure 2C), indicating that knockout of BBS1 disrupts BBSome assembly and no BBSome is available to be recruited to the basal body for ciliary entry in bbs1-1 cells. Of note, BBS3 was concentrated at the basal bodies and distributed along the length of cilia in bbs1-1 cells ( Figure 2D). Together with the result that BBS3 is present in cilia of bbs1-1 cells at wild-type level ( Figure 2B), our data revealed that Chlamydomonas BBS3 does not require the BBSome for targeting to the basal body nor for entering cilia. Our previous study revealed that the BBSome loads onto anterograde IFT trains at the basal body and enters cilia via IFT in C. reinhardtii (Xue et al., 2020). In mammalian cells, BBS3 binds the BBSome through a direct interaction between BBS3 and BBS1 (Mourão et al., 2014) and was proposed to enter cilia via IFT through binding the BBSome (Jin et al., 2010;Liew et al., 2014). Since Chlamydomonas BBS3 can enter cilia without relying on the BBSome, we proposed that BBS3 does not enter cilia via IFT in C. reinhardtii. Supportive of this notion, when expressed at similar levels in CC-125 and bbs1-1 cells (resulting strains BBS3::YFP and BBS3::YFP 1-1 ) ( Figure 2E), BBS3::YFP entered cilia of both strains and was maintained in cilia of both strains at similar amounts ( Figure 2F). As reflected by total internal reflection fluorescence (TIRF) imaging, BBS3::YFP statically distributed along the length of cilia but did not undergo IFT in either strain ( Figure 2G; and Figure 2-videos 1 and 2). Thus, BBS3 can enter cilia independent of the BBSome and does not traffic bidirectionally in cilia via IFT but likely diffuses inside cilia ( Figure 2H).

Membrane association is a prerequisite for BBS3 to enter cilia
Other studies have demonstrated that GTP-loaded small GTPase of the Arf family associates with the membrane via its N-terminal amphipathic helix (Amor et al., 1994;Liu et al., 2010;Zhang et al., 2011), and the N-terminal residues 1-15 are essential for BBS3 to associate with the membrane in vitro (Jin et al., 2010;Mourão et al., 2014). When incubated with the synthetic liposomes, bacterially expressed BBS3::YFP associated with liposomes only in the presence of GTPgS, which locks BBS3::YFP in a GTP-bound state ( Figure 3A, B; Liew et al., 2014). In contrast, BBS34N::YFP, which lacks the N-terminal residues 1-15 of BBS3, did not associate with liposomes even in the presence of GTPgS, revealing that the N-terminal amphipathic helix is required for BBS3 to associate with the membrane, and BBS3 associates with the membrane in a GTP-dependent manner ( Figure 3A, B). To examine whether the N-terminal residues 1-15 are essential for BBS3 to enter cilia in vivo, we expressed BBS34N::YFP at similar levels in CC-125 and bbs1-1 cells (resulting strains BBS34N::YFP and BBS34N::YFP 1-1 ) ( Figure 3C). In contrast to BBS3::YFP that enters cilia of both CC-125 and bbs1-1 cells ( Figure 2F, G), BBS34N::YFP was absent from cilia of both BBS34N::YFP and BBS34N::YFP 1-1 strains ( Figure 3D, E), indicating that depletion of the N-terminal residues 1-15 prevents BBS3 from entering cilia. Of note, BBS34N::YFP was concentrated at the basal bodies of both BBS34N::YFP and BBS34N::YFP 1-1 strains, revealing that depletion of the N-terminal residues 1-15 retains the ability of BBS3 to traffic from the cell body to the basal body in a BBSomeindependent manner ( Figure 3E). In BBS34N::YFP 1-1 cells, the BBSome was not assembled due to the lack of BBS1 and thus was not available for targeting to the basal bodies for ciliary entry ( Figure 3D, E). In contrast, BBS34N::YFP colocalized with BBS5 at the basal bodies in BBS34N::YFP cells ( Figure 3E) and immunoprecipitated the BBSome subunits BBS1 and BBS5 in the cell body extracts of BBS34N::YFP cells ( Figure 3F), revealing that BBS34N::YFP retains the ability to bind and recruit the BBSome to the basal body (Xue et al., 2020). Therefore, loss of the N-terminal residues 1-15 of BBS3 does not prevent BBS3 from binding and targeting the BBSome to the basal body, while membrane association is necessary for BBS3 to enter cilia from the basal body in C. reinhardtii ( Figure 3G).
BBS3 associates with the ciliary membrane in a GTP-dependent manner BBS3 of several species is unique in that they contain an alanine (73A for C. reinhardtii) rather than a glutamine (Q) residue at the position critical for Ras family GTPases to hydrolyze GTP (     . Similar stationary pattern was also recorded for the C-terminal green fluorescent protein (GFP)tagged BBS3 and the A73L mutant in rescuing strains BBS3 Res-WT and BBS3 Res-A73L , in which BBS3:: GFP and BBS3 A73L ::GFP are expressed in similar amounts in the BBS3-knockdown strain BBS3 miRNA ( Figure 4D; and Figure 4-videos 3 and 4; Xue et al., 2020). Upon entering cilia, BBS3 A73L ::YFP was detected in the membrane fraction but not in the matrix fraction isolated from cilia of the BBS3 A73L ::YFP strain, while the majority of BBS3::YFP was present in the matrix faction of cilia of the BBS3::YFP strain ( Figure 4E; and Figure 4-figure supplement 2). To confirm this observation, we performed ciliary fraction analysis on CC-125 cells in the presence of GTPgS, GDP, or neither, which are proposed to lock BBS3 at a GTP-bound, GDP-bound, or wild-type state, respectively (Liew et al., 2014). Our data showed that BBS3 is mainly present in the matrix fraction in the absence of nucleotide or completely present in the matrix fraction when GDP was added, while the majority of BBS3 was detected in the membrane faction in the presence of GTPgS ( Figure 4F). Therefore, GTP binding allows BBS3 to attach to the ciliary membrane, and GDP binding detaches BBS3 from the ciliary membrane in cilia, consistent with the observation that GTP-but not GDPbound BBS3 associates with the membrane of liposomes in vitro ( Figure 3B; Mourão et al., 2014).
BBS3 interacts with the BBSome in cilia in a GTP-dependent manner Rodent BBS3, in its GTP-bound state, binds and recruits the BBSome to the membrane of liposomes in vitro, implying that, in cilia, GTP loading confers BBS3 to bring the BBSome to the ciliary membrane (Liew et al., 2014). To investigate whether and how BBS3 interacts with the BBSome in cilia of C. reinhardtii cells, we performed sucrose gradient density centrifugation on the ciliary extracts of BBS3::YFP and BBS3 A73L ::YFP transgenic strains. Of note, a minority of BBS3::YFP co-sedimented with the BBSome subunits BBS1 and BBS5 ( Figure 5A). In contrast, the majority of BBS3 A73L ::YFP co-sedimented with the two BBSome proteins although partial BBS3 A73L ::YFP remained to be existing as a free form independent of the two BBSome proteins in cilia ( Figure 5A). These results showed that GTP-loaded BBS3 is able to bind the BBSome in cilia ( Figure 5A). This notion was confirmed by observing that the BBS3 A73L variant but not BBS3::YFP recovered BBS1 and BBS5 in the ciliary extracts ( Figure 5B). To further verify this result, we added GTPgS or GDP, which is proposed to lock GTPases in a GTP-or GDP-bound state, respectively, to the ciliary extracts of BBS3:: YFP cells (Liew et al., 2014). Sucrose gradient density centrifugation assay identified BBS3::YFP sediments separately from BBS1 and BBS5 in the presence of GDP but became partially co-sedimented with the two BBSome proteins when GTPgS was present ( Figure 5C). Together with the observation that BBS3::YFP immunoprecipitated BBS1 and BBS5 in the ciliary extracts of the BBS3:: YFP transgenic strain only when GTPgS was present ( Figure 5D), we concluded that BBS3 in a GTPbut not GDP-bound configuration binds the BBSome in cilia. Of note, partial BBS3, even when locked in a GTP-bound configuration, remained not to interact with the BBSome, revealing that GTP loading cannot confer all BBS3 molecules to interact with the BBSome in cilia ( Figure 5A, C).

The BBSome cycles through cilia normally in the absence of BBS3 in cilia
Once at the ciliary tip, the BBSome/IFT train is proposed to remodel before undergoing turnaround to exit cilia (Wei et al., 2012). During this process, rodent BBS3 was reported to mediate BBSome exit out of cilia by promoting its loading onto retrograde IFT trains at the ciliary tip (Liew et al., 2014). To investigate whether this applies in C. reinhardtii, we examined clip1 cells and found that loss of BBS3 in cilia did not cause ciliary hyperaccumulation of the BBSome subunits BBS1, BBS4, BBS5, and BBS7, excluding BBS3 from being required for the BBSome to exit cilia in C. reinhardtii ( Figure 6A and Figure 1B; Liew et al., 2014). Our previous study has shown that loss of IFT25 blocks loading of the BBSome onto retrograde IFT trains (Dong et al., 2017b). Reflecting this observation, BBS1 and BBS5 hyperaccumulated at the ciliary tip in the IFT25-knockdown cells ( Figure 6B). BBS1 and BBS5 co-sedimented with each other and peaked at the same fraction as IFT-B1 (checked with IFT46 and IFT70) in the ciliary extracts of wild-type CC-5325 cells ( Figure 6C; Xue et al., 2020). In contrast, these two BBSome proteins rather than IFT-B1 were found to shift to lower fractions in the ciliary extracts of IFT25 miRNA cells, suggesting that they are not able to assemble to form intact BBSomes after remodeling at the ciliary tip ( Figure 6C). As compared to the IFT25-knockdown cells, clip1 cells did not accumulate BBS1 and BBS5 at the ciliary tip and the two BBSome subunit proteins co-sedimented in the same fractions as IFT-B1 in the ciliary extracts of clip1 cells, demonstrating that ciliary dynamics of the BBSome is normally maintained when BBS3 is absent from cilia ( Figure 6B, C). These results further argued against BBS3 being required for the BBSome to load onto retrograde IFT trains for ciliary exit (Liew et al., 2014). Next, we expressed BBS5::YFP in similar amounts in CC-5325 and clip1 cells (resulting strains BBS5::YFP 5325 and BBS5::YFP clip1 ) ( Figure 6D). BBS5::YFP entered cilia of both strains and was maintained at similar amounts ( Figure 6E). As determined by TIRF imaging, BBS5::YFP underwent IFT with normal speeds and velocities in cilia of clip1 cells, suggesting that IFT of the BBSome is not impaired when BBS3 is absent from cilia ( Figure 6F, G; Figure 6-source data 1; and Figure 6-videos 1 and 2). Therefore, we concluded that BBS3 is not required for maintaining BBSome dynamics in cilia, nor for promoting BBSome coupling/uncoupling with IFT trains at the ciliary tip in C. reinhardtii ( Figure 6H).
BBS3 is essential for PLD to associate with the BBSome for ciliary exit In C. reinhardtii, the membrane-associated signaling protein PLD enters cilia mostly by diffusion but is removed from cilia mainly via IFT by associating with the BBSome. Therefore, the bbs mutants accumulates PLD in cilia as the BBSome is not available for entering cilia to bridge PLD to retrograde IFT trains for ciliary exit in those mutants (Lechtreck et al., 2013;Liu and Lechtreck, 2018). Of note, the clip1 strain contained PLD at wild-type level in whole-cell sample ( Figure 7A) and accumulated PLD in cilia, mostly at the ciliary tip ( Figure 7B, C), revealing that BBS3 is essential for promoting PLD to exit cilia. In clip1 cells, BBS3 was maintained at wild-type level in whole-cell sample ( Figure 7A) but was absent from cilia ( Figure 7B), and its ciliary absence did not affect BBSome  Figure 6 continued on next page abundance in cilia and uncoupling/recoupling of the BBSome with IFT at the ciliary tip ( Figure 6A-F); we thus concluded that Chlamydomonas BBS3 is essential for PLD to associate with the BBSome at the ciliary tip for ciliary exit via IFT. To further verify this notion, we examined the BBS3-knockdown strain, BBS3 miRNA , in which the BBSome (checked with BBS1 and BBS5) was maintained at wild-type level in whole-cell sample but was strongly reduced in cilia ( Figure 7D, E; Xue et al., 2020). BBS3 knockdown did not alter cellular PLD level but accumulated PLD in cilia, mostly at the ciliary tip ( Figure 7D-F). When BBS3::GFP was expressed in BBS3 miRNA cells (resulting strain BBS3 Res-WT ), it entered cilia and restored BBS1, BBS5, and PLD to wild-type levels ( Figure 7D-F; Xue et al., 2020). When BBS34N::YFP was expressed in BBS3 miRNA cells (resulting strain BBS3 Res-4 N ), BBS1 and BBS5 were rescued to wild-type levels in cilia, while PLD remained to be accumulated mostly at the ciliary tip ( Figure 7D-F). Since BBS3 was absent from cilia of both BBS3 miRNA and BBS3 Res-4N cells and BBS1 and BBS5 were restored to wild-type levels in cilia of BBS3 Res-4N cells, our data thus showed that ciliary absence of BBS3 causes PLD to accumulate mostly at the ciliary tip even when intact BBSomes are present in cilia ( Figure 7E-G), confirming that BBS3 is essential for PLD to associate with the BBSome at the ciliary tip for ciliary exit. Remarkably, the BBS3 A73L ::GFPexpressing BBS3 miRNA strain BBS3 Res-A73L also restored the ciliary PLD content to wild-type level, excluding GTPase cycling of BBS3 in cilia from mediating PLD association with the BBSome for ciliary exit ( Figure 7D-F). Together with the observation that PLD rarely co-sedimented with the BBSome (checked with BBS1 and BBS5) in the ciliary extracts of clip1 cells; a minority of PLD co-sedimented with the BBSome in the ciliary extracts of BBS3 Res-4N cells; and the majority of PLD became co-sedimented with the BBSome in the ciliary extracts of BBS3 Res-A73L cells ( Figure 7H), our data suggest that GTP-bound BBS3 efficiently enables PLD to associate with the BBSome in cilia at the ciliary tip to undergo retrograde IFT ( Figure 7I). In addition, BBS34N::YFP did not enter cilia itself but restored BBS1 and BBS5 to wild-type levels in cilia of the BBS3-knockdown BBS3 miRNA cells ( Figure 7D-F), providing compelling evidence to confirm that ciliary entry of the BBSome from the ciliary base does not depend on BBS3 (Figure 1).

Discussion
Using C. reinhardtii as a model organism, we elucidated the interplay between BBS3 and the BBSome for ciliary entry and investigated their dynamics in cilia and the role of BBS3 in promoting ciliary exit of the signaling protein PLD via the BBSome. Our data show that BBS3 plays a crucial role in mediating ciliary exit of PLD through promoting its association with the BBSome at the ciliary tip, thus closing a gap in our understanding of the role of BBS3 in regulating ciliary exit of signaling protein cargoes.

BBS3 and the BBSome interplay non-reciprocally for targeting to the basal body
The BBSome is the major effector of BBS3 in human cells (Jin et al., 2010). BBS3 binds the BBSome through a direct interaction with the BBSome subunit BBS1 in human, mouse, and Chlamydomonas cells (Mourão et al., 2014;Zhang et al., 2012). In the cell body of C. reinhardtii, this binding does not rely on the BBS3 nucleotide state; however, only GTP-bound BBS3 recruits the BBSome to the Figure 6 continued BBS5::YFP inside cilia of BBS5::YFP 5325 and BBS5::YFP clip1 cells are similar. The speeds of BBS5::YFP were 2.01 ± 0.12 (anterograde) and 3.31 ± 0.27 mm/ sec (retrograde), and the frequencies were 0.69 ± 0.10 (anterograde) and 0.39 ± 0.10 particles/sec (retrograde) for BBS5::YFP 5325 cells. In cilia of BBS5:: YFP clip1 cells, the speeds of BBS5::YFP were 2.04 ± 0.11 (anterograde) and 3.34 ± 0.19 mm/sec (retrograde), and the frequencies were 0.70 ± 0.09 (anterograde) and 0.41 ± 0.08 particles/sec (retrograde). Error bar indicates SD. n: number of cilia analyzed; n.s.: non-significance. (H) Schematic presentation showing that the BBSome and IFT trains remodel at the ciliary tip in a BBS3-independent manner. The online version of this article includes the following video and source data for figure 6: Source data 1. Source data for the speed and frequency experiment shown in Figure 6G. basal body, suggesting that basal body recruitment of the BBSome is regulated by the nucleotide state of BBS3 (Xue et al., 2020). Such a scenario would allow cells to control the amount of BBSomes available for pick-up by anterograde IFT trains at the basal body and in turn could regulate the presence and amount of signaling proteins in cilia, which rely on the BBSome for ciliary export and import via IFT (Wingfield et al., 2018;Xue et al., 2020). Indeed, basal body-associated pools have been also described for other IFT cargoes (Dai et al., 2018). Of note, the BBSome relies on GTP-bound BBS3 to target to the basal body, while GTP-bound BBS3 concentrates at the basal body even in the absence of the BBSome, suggesting that BBS3 does not rely on the BBSome for (I) Schematic presentation of how BBS3 binds and recruits the BBSome to the ciliary membrane for interacting with PLD and how PLD-associated BBSome loads onto IFT trains for ciliary exit. For panels A, B, D, and E, a-tubulin and Actubulin were used to adjust the loading for WCE and CE, respectively. For panels C and F, the inset shows ciliary tips. Scale bar 10 mm.
trafficking from the cell body to the basal body in C. reinhardtii. Thus, unlike the observation that GTP-bound BBS3 and the BBSome target to the ciliary base of human cells in an interdependent manner (Jin et al., 2010), the BBSome relies on GTP-bound BBS3 to target to the basal body but not vice versa in C. reinhardtii.

BBS3 and the BBSome translocate from the basal body to cilia independently
Upon targeting to the basal body, the BBSome loads onto anterograde IFT trains through associating with IFT-B1 to enter cilia (Xue et al., 2020). An analysis of both the clip1 mutant and the BBS3knockdown strain BBS3 miRNA shows that this happens even when BBS3 is not able to enter cilia. Furthermore, BBS3 translocates from the basal body to cilia even in the absence of the BBSome. These results reveal that BBS3 and the BBSome do not rely on each other for translocating from the basal body to cilia in C. reinhardtii. Loss of the N-terminal amphipathic helix essential for BBS3 to associate with the membrane prevents BBS3 from translocating from the basal body to cilia but has no effect on the ability of BBS3 to traffic from the cell body to the basal body. These results reveal that membrane association is required for BBS3 to translocate from the basal body to cilia but not for BBS3 to traffic from the cell body to the basal body. Thus, our data provide in vivo evidence, for the first time, to show that BBS3 enters cilia from the basal body potentially by lateral transport on the membrane, consistent with the notion previously proposed for rodent cells (Jin et al., 2010;Liew et al., 2014;Mourão et al., 2014;Zhang et al., 2011). Our previous study has shown that RABL5/IFT22 binds and stabilizes BBS3 independent of their nucleotide states in the cell body (Xue et al., 2020). Since the GTP-bound configuration is related to BBS3's ability to associate with the membrane (Jin et al., 2010;Klink et al., 2020;Liew et al., 2014;Mourão et al., 2014;Singh et al., 2020;Zhang et al., 2011), IFT22 binding to BBS3 in the cell body could prevent BBS3 from associating with the cell membrane. Before ciliary entry, IFT22 is released from BBS3 at the basal body, which will enable the GTP-bound BBS3 to be in a state to attach to the membrane for lateral transport into cilia (Figure 8). Therefore, the IFT22-dependent recruitment of BBS3 to the basal body and then separation from BBS3 could also function to regulate BBS3 activity at the basal body, ensuring a spatial restriction of its association with the membrane to the basal body and ciliary compartment (Xue et al., 2020).

Where and how does BBS3 bind the BBSome in cilia?
Unlike its counterparts in Caenorhabditis elegans and mammalian olfactory neurons that undergo IFT in cilia (Fan et al., 2004;Williams et al., 2014), Chlamydomonas BBS3, upon entering cilia, diffuses between the ciliary base and tip ( Figure 4C, D). This excludes BBS3 from coupling with BBSomes during its transportation between the ciliary base and tip, a notion also supported by the observation that partial BBS3 molecules remain to be separated from the BBSome in cilia even in its GTP-locked configuration (BBS3 A73L ) or in the presence of GTPgS ( Figure 5A, C). However, GTPbound BBS3 recovers BBSomes in cilia by immunoprecipitation, and a majority of GTP-locked BBS3 co-sediments with BBSomes in cilia ( Figure 5A-D), revealing that GTP-bound BBS3 binds BBSomes most likely at the ciliary tip. These observations demonstrate that GTP loading does not confer BBS3 to bind BBSomes during its transportation between the ciliary base and tip. While at the ciliary tip, BBS3, however, gains the ability to bind BBSomes in a GTP-dependent manner, in agreement with the report that GTP loading confers BBS3 to bind BBSomes in cilia of human cells (Jin et al., 2010). Why GTP-locked BBS3 selects to bind or not to bind the BBSome in different ciliary compartments remains unknown at present, and the underlying molecular mechanism deserves to be further investigated (Figure 8).
BBS3 likely promotes the association of PLD with the BBSome at the ciliary tip BBS3 is assumed to convert to and exist as a GDP-bound configuration upon entering cilia in rodent cells (Liew et al., 2014). It was also proposed that BBS3 undergoes a GTPase cycle at the ciliary tip, and RABL4/IFT27 activates GDP-bound BBS3 as a BBS3-specific GEF during this process (Liew et al., 2014). In rodent cells, GTP-bound BBS3 promotes the cargo-laden BBSome to reload onto retrograde IFT trains at the ciliary tip for ciliary exit via IFT (Liew et al., 2014). In C. reinhardtii, Figure 8. Hypothetical model of how BBS3 enters cilia to promote ciliary exit of PLD via the BBSome in C. reinhardtii. IFT22 and BBS3 both in a GTPbound configuration recruit the BBSome to the basal body, where the BBSome separates from BBS3 and loads onto the anterograde IFT trains by coupling with IFT-B1 and enters cilia via IFT (Xue et al., 2020). GTP-bound IFT22 is released from BBS3 and binds IFT-B1 through a direct interaction between IFT22 and the IFT74/81 sub-complex for ciliary entry (Bhogaraju et al., 2013;Lucker et al., 2010;Taschner et al., 2014). GTP-bound BBS3 Figure 8 continued on next page an analysis of the mutant strains indicates that the BBSome migrates from the basal body to cilia and cycles through cilia normally in the absence of BBS3, excluding BBS3 from mediating loading of the BBSome onto retrograde IFT trains for ciliary exit at the ciliary tip. Instead, GTP-bound BBS3 likely couples with the ciliary membrane and binds BBSomes simultaneously at the ciliary tip. This could recruit BBSomes to the ciliary membrane for interacting with PLD ( Figure 8). Therefore, BBS3 is proposed to participate in mediating the association of ciliary signaling cargoes, for example, PLD, with BBSomes at the ciliary tip (Jin et al., 2010;Klink et al., 2020;Klink et al., 2017;Liew et al., 2014;Mourão et al., 2014). Interestingly, a BBS3 A73L mutant can rescue the accumulated PLD to wild-type level in cilia, suggesting that, at least in cilia of Chlamydomonas cells, BBS3 that is locked in a GTPbound configuration is able to promote ciliary signaling molecules to associate with BBSomes for ciliary exit via IFT. This eventually excludes BBS3 GTPase cycling at the ciliary tip from being involved in promoting PLD to exit cilia in C. reinhardtii (Liew et al., 2014). Continued on next page then associates with the cell membrane to diffuse into cilia. PLD enters cilia by diffusion on membrane (Liu and Lechtreck, 2018). At the ciliary tip, the BBSome is released from anterograde IFT trains through BBSome/IFT train remodeling (Xue et al., 2020). GTP-bound BBS3 binds and recruits the BBSome to the ciliary membrane for interacting with PLD (Liu and Lechtreck, 2018). PLD-laden BBSomes then load onto retrograde IFT trains by coupling with IFT-B1 for ciliary exit (Xue et al., 2020). The kinesin-II anterograde motor and the cytoplasmic dynein-1b retrograde motor are also shown (Cole et al., 1998;Pazour et al., 1998). After turnaround at the ciliary tip, kinesin-II diffuses back to the ciliary base (Chien et al., 2017;Hendel et al., 2018).

Antibodies and immunoblotting
Polyclonal antibodies raised in rabbits have been described previously (Dong et al., 2017b;Xue et al., 2020;Zhu et al., 2017). Rabbit-originated antibodies against BBS4, BBS7, and PLD were generated by Beijing Protein Innovation. Monoclonal antibodies against GFP (YFP) (mAbs 7.1 and 13.1, Roche), a-tubulin (mAb B512, Sigma), and Ac-tubulin (mAb 6-11B-1, Sigma-Aldrich) were commercially purchased (key resources table). Preparation of whole cell, cell body, and ciliary extracts, SDS-PAGE electrophoresis, and immunoblotting were done as detailed previously (Xue et al., 2020). If not otherwise specified, 20 mg of total protein from each sample was loaded for SDS-PAGE. In immunoblotting assays, dilution used for primary and secondary antibodies is listed in the key resources table.

Isolation of cilia and cell bodies
Methods for the isolation of cilia and cell bodies have been described previously (Xue et al., 2020

Preparation of ciliary fractions
After cilia were isolated, they were dissolved in HMDEK buffer (50 mM HEPES pH 7.2, 5 mM MgCl 2 , 1 mM DTT, 0.5 mM EDTA, and 25 mM KCl) plus protein inhibitors (PI) (1 mM PMSF, 50 mg/ml soybean trypsin inhibitor, 1 mg/ml pepstatin A, 2 mg/ml aprotinin, and 1 mg/ml leupeptin) and fresh-frozen in liquid nitrogen. The ciliary matrix fraction was obtained by freezing and thawing cilia followed by centrifugation (27,000 Âg, 4˚C, 15 min), and the pellets were dissolved in an HMEDK buffer containing 0.5% NP-40 and stayed on ice for 15 min. The supernatant and pellet were collected after centrifugation (14,000 Âg at 4˚C for 10 min) as membrane and axonemal fractions, respectively, according to a previously described method .

Preparation of liposome and binding assays
Liposomes were prepared exactly as described previously (Jin et al., 2010). BBS3 association with liposome was conducted in an HMEK buffer. Liposomes (2 mg) were incubated with 100 mg of bacterially expressed C-terminal 6ÂHis tagged BBS3::YFP or BBS34N::YFP in the presence of GTPgS (100 mM), GDP (100 mM), or neither in a 100 mL reaction at 30˚C for 1 hr. The reactions were centrifuged at 385,000 Âg for 30 min at 24˚C, and equal portions of the resulting pellets were resolved by SDS-PAGE and immunoblotted with a-BBS3.

Immunoprecipitation
Cell bodies and cilia isolated from Chlamydomonas strains expressing YFP-tagged BBS3 or its variants were resuspended in HMEK+PI supplemented with 50 mM NaCl and lysed by adding nonidet P-40 (NP-40) to 1%. The supernatants were collected by centrifugation (14,000 Âg, 4˚C, 10 min) and incubated with agitation with 5% BSA-pretreated camel anti-GFP antibody-conjugated agarose beads (V-nanoab Biotechnology) for 2 hr at 4˚C. The beads were then washed with HMEK containing 150 mM NaCl, 50 mM NaCl, and finally 0 mM NaCl. The beads were then added with Laemmli SDS sample buffer and boiled for 5 min before centrifuging at 2,500 Âg for 5 min. The supernatants were then analyzed by immunoblotting as described above.

Immunofluorescence
Immunofluorescence staining was performed according to our published method (Fan et al., 2010). The primary antibodies against PLD, YFP, BBS3, BBS1, and BBS5 and the secondary antibodies -Alexa-Fluor594-conjugated goat anti-rabbit IgG and Alexa-Fluor488-conjugated goat anti-mouse IgG (Molecular Probes, Eugene, OR) -are listed in the key resources table with their suggested dilutions for immunofluorescence staining. Images were captured with an IX83 inverted fluorescent microscope (Olympus) equipped with a back illuminated scientific CMOS camera (Prime 95B, Photometrics) at 100Â amplification and processed with CellSens Dimension (version 2.1, Olympus).

Sucrose density gradient centrifugation
Sucrose density gradient centrifugation of ciliary extracts was performed according to our published method (Wang et al., 2009). In brief, linear 12 mL of 10-25% sucrose density gradients in 1Â HMDEK buffer plus protease inhibitors and 1% NP-40 were generated using the Jule Gradient Former (Jule Inc, Milford, CT) and used within 1 hr. The cilia were opened with liquid nitrogen for three rounds of frozen-and-thaw cycles and centrifuged at 12,000 rpm, 4˚C, for 10 min. Seven hundred microliters of ciliary extracts were then loaded on the top of the gradients and separated at 38,000 rpm, 4˚C, for 14 hr in a SW41Ti rotor (Beckman Coulter). The gradients were fractioned into 24-26 0.5 ml aliquots using a Pharmacia LKB Pump P-1 coupled with a FRAC-100 fraction collector. The standards used to calculate S-values were BSA (4.4S), aldolase (7.35S), catalase (11.3S), and thyroglobulin (19.4S). Twenty microliters of each fraction was analyzed by immunoblotting as described above. If necessary, the assay was carried out in the presence of GTPgS (20 mM) or GDP (20 mM), respectively.

IFT video imaging and speed measurements
The motility of GFP-and YFP-tagged proteins in cilia was imaged at~15 fps using TIRF microscopy on an inverted microscope (IX83, Olympus) equipped with a through-the-objective TIRF system, a 100Â/1.49 NA TIRF oil immersion objective (Olympus), and a back-illuminated scientific CMOS camera (Prime 95B, Photometrics) as detailed previously (Xue et al., 2020). To quantify IFT speeds and frequencies, kymograms were generated and measured with CellSens Dimension (version 2.1, Olympus).

Protein expression and purification
The cDNAs encoding BBS3, BBS3 A73L , and BBS3 T31R were amplified from pGEX-6P-cBBS3, pGEX-6P-cBBS3 A73L , and pGEX-6P-cBBS3 T31R using the primer pair cBBS3-FOR and cBBS3-REV as described earlier. The cDNAs of BBS3 and its mutants were inserted into the BamHI and Hind III sites of pET-28a (Novagen) to result in pET-28a-cBBS3, pET-28a-cBBS3 A73L , and pET-28a-cBBS3 T31R , respectively. To express YFP, the cDNA of YFP was cut from pBKS-gBBS3::YFP-Ble with EcoRI and XhoI and inserted into pET-28a (Novagen) to result in pET-28a-YFP. To express the C-terminal YFP-tagged BBS3 and BBS3DN, the cDNAs of BBS3 and BBS3DN were amplified from pGEX-6P-cBBS3 using the primer pair cBBS3-FOR and cBBS3-REV (for BBS3) and cBBS3DN-FOR (5'-CCGGATCCATGCTCTTGTCGCTG-3') and cBBS3-REV (for BBS3DN) and inserted into BamHI and EcoRI sites of pET-28a-YFP to result in pET-28a-BBS3::YFP and pET-28a-BBS3DN::YFP. After these plasmids were transformed into bacteria, the bacterially expressed recombinant proteins were purified with Ni-NTA beads and cleaved with thrombin (Solarbio) to get rid of the N-terminal 6Â His tag according to our previous report (Xue et al., 2020). If necessary, 10 mg of proteins from elutes was resolved on 12% SDS-PAGE gels and visualized with Coomassie Blue staining.

Small GTPase assay
Small GTPase assay was performed according to our previous report (Xue et al., 2020). Intrinsic GTP hydrolysis of the bacterially expressed BBS3, BBS3 A73L , and BBS3 T31R was measured by optical assay for the release of inorganic phosphate with reagents from the QuantiChrom ATPase/GTPase assay kit (Bioassay Systems) (Pan et al., 2006).