© 2015. Published by The Company of Biologists Ltd | Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

RESEARCH ARTICLE SUMOylation regulates ciliary localization of olfactory signaling proteins Jeremy C. McIntyre1, Ariell M. Joiner2, Lian Zhang1, Jorge Iñiguez-Lluhı2́and Jeffrey R. Martens1,*

ABSTRACT of primary neurons where it is enriched in the primary cilia Cilia are evolutionarily conserved organelles found on many projecting from neurons throughout many regions of the brain mammalian cell types, including neuronal populations. Although (Berbari et al., 2007; Bishop et al., 2007). Although localization of neuronal cilia, including those on olfactory sensory neurons (OSNs), AC3 to cilia in both of these systems is necessary for proper are often delineated by localization of adenylyl cyclase 3 (AC3, function, little is known about the mechanisms that are responsible also known as ADCY3), the mechanisms responsible for targeting for its targeting (Guadiana et al., 2013; Kaneko-Goto et al., 2013; integral membrane proteins are largely unknown. Post-translational Wang et al., 2011; Wong et al., 2000). modification by small ubiquitin-like modifier (SUMO) proteins plays an For ciliary localization of transmembrane proteins, mechanisms important role in protein localization processes such as nuclear– relying on the use of specific signal sequences such as the RVxP cytosolic transport. Here, we identified through bioinformatic analysis motif and the Ax[S/A]xQ motif have been identified (Berbari et al., that adenylyl cyclases harbor conserved SUMOylation motifs, and 2008a; Dishinger et al., 2010; Geng et al., 2006). For example, the show that AC3 is a substrate for SUMO modification. Functionally, Ax[S/A]xQ motif and interactions with chaperone proteins overexpression of the SUMO protease SENP2 prevented ciliary regulates the ciliary localization of several G-protein-coupled localization of AC3, without affecting ciliation or cilia maintenance. receptors (GPCRs) (Berbari et al., 2008a; Berbari et al., 2008b; Furthermore, AC3-SUMO mutants did not localize to cilia. To test Sun et al., 2012). In the OSNs, ciliary trafficking of olfactory cyclic- whether SUMOylation is sufficient for cilia entry, we compared nucleotide-gated (CNG) channels requires formation of a localization of ANO2, which possesses a SUMO motif, and ANO1, heteromeric complex between CNGA2 and the CNGB1b subunit, which lacks SUMOylation sites and does not localize to cilia. which contains a RVxP motif (Jenkins et al., 2006; Michalakis Introduction of SUMOylation sites into ANO1 was not sufficient for et al., 2006). The proper context is essential, however, because ciliary entry. These data suggest that SUMOylation is necessary but insertion of this motif into CNGA2 is not sufficient for trafficking of not sufficient for ciliary trafficking of select constituents, further the CNGA2 subunit on its own (Jenkins et al., 2006). Notably, CNG establishing the link between ciliary and nuclear import. channel trafficking also depends on phosphorylation of CNGB1 by the serine/threonine protein kinase casein kinase II (CK2) KEY WORDS: Cilia, Protein Trafficking, SUMO and interaction with phosphofurin acidic cluster-sorting protein 1 (PACS-1) (Jenkins et al., 2009). Although these studies have INTRODUCTION identified mechanisms involved in ciliary CNG trafficking, these Cilia are evolutionarily conserved organelles projecting from the mechanisms do not account for the ciliary enrichment of other surface of cells and are important for motility, sensing external olfactory signaling proteins. For example, overexpression of a stimuli and maintaining cellular homeostasis. Defects in cilia dominant-negative PACS-1 protein in OSNs prevents CNGA2 formation and function underlie a growing number of human localization, but does not affect ciliary localization of AC3 (Jenkins diseases that affect numerous organ systems. In the olfactory et al., 2009). Retention of selective ciliary entry mechanisms has system, the multiple cilia that extend from the dendritic knobs of also been demonstrated in OSNs of Cep290 and Cetn2 mutant mice olfactory sensory neurons (OSNs) are essential for function given where ciliary transition zone and basal body function is disrupted that ciliary disruption leads to anosmia (Kulaga et al., 2004; (McEwen et al., 2007; Ying, 2014). Together these findings argue McEwen et al., 2007; McIntyre et al., 2012). Functional odorant that ciliary targeting mechanisms are protein specific and likely detection depends on a ciliary-localized signaling pathway involve multiple processes with some proteins requiring several involving odorant receptors coupled to adenylyl cyclase 3 (AC3, different mechanisms. To fully elucidate the mechanisms regulating also known as ADCY3) (Bakalyar and Reed, 1990; Bönigk et al., ciliary entry, comparisons to other selective entry compartments 1999; Buck and Axel, 1991; Feinstein et al., 2004; Menco et al., might provide additional insight. 1997; Schwarzenbacher et al., 2005; Stephan et al., 2009; Wong Mechanisms that are known to regulate nuclear import have et al., 2000). The ciliary localization of AC3 is also characteristic recently been shown to function in ciliary import. Similar to in the nucleus, molecular size and nucleoporins regulate the ability of soluble proteins to diffuse into the (Breslow et al., 2013; Kee 1Department of Pharmacology and Therapeutics, University of Florida, PO Box 100267, Gainesville, FL 32610, USA. 2Department of Pharmacology, University of et al., 2012; Kee and Verhey, 2013; Lin et al., 2013). In addition, Michigan, Ann Arbor, MI 48109, USA. basic sequences similar to nuclear localization signals have been identified in several ciliary proteins and these sequences regulate *Author for correspondence ([email protected]) localization into the cilium (Dishinger et al., 2010; Hurd et al., This is an Open Access article distributed under the terms of the Creative Commons Attribution 2011). Finally, nuclear import of many proteins involves a Ran License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, β distribution and reproduction in any medium provided that the original work is properly attributed. GTPase gradient and chaperone proteins such as importin- 2 (also known as TNPO1). These same mechanisms have also been

Received 21 October 2014; Accepted 24 March 2015 implicated in both ciliogenesis and trafficking of proteins in the Journal of Cell Science

1934 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673 cilium (Dishinger et al., 2010; Fan et al., 2011; Hurd et al., 2011; the stimulatory G protein (Fig. 1B). This sequence is highly Maiuri et al., 2013). conserved across vertebrate species; however, the specific motif is Given the identified similarities between nuclear and ciliary absent from the zebrafish (Danio rerio) sequence, although entry, it is reasonable that other mechanisms known to regulate putative SUMO motifs are present at other positions in the full protein localization into the nucleus also act at the cilium. In this sequence. AC3 belongs to the type III family of adenylyl cyclase regard, the post-translational modification of proteins through the proteins, which consist of ten distinct isoforms. Interestingly, in reversible covalent conjugation of small-ubiquitin like modifier addition to AC3, several other members are also reported to (SUMO) proteins plays important roles in nucleocytoplasmic localize to cilia, including AC2, AC4, AC6, AC5 and AC8 (Choi transport. In particular, the localization of RanGAP1 to the et al., 2011; Masyuk et al., 2008; Wong et al., 2000). Analysis of cytoplasmic face of nuclear pore filaments is essential for all ten family members from the mouse genome reveals at least establishment of the Ran GTPase gradient and depends on one predicted SUMO site in each sequence, suggesting that RanGap SUMOylation as well as on the interaction between the SUMOylation of transmembrane adenylate cyclase proteins is nucleoporin Nup358 (also known as RANBP2) and the SUMO important for their trafficking and/or function. moiety on RanGap (Hutten et al., 2008; Matunis et al., 1996). In The presence of putative modification sites indicates that AC3 is addition, the SUMOylation status of multiple proteins regulates a likely target of SUMO modification. To evaluate whether AC3 their nuclear or cytoplasmic localization mainly by altering their is modified by SUMOylation in live cells, we immunoprecipitated interactions with other proteins (Hutten et al., 2008; Klein and Nigg, GFP-tagged AC3 from HEK293 cells co-expressing HA–SUMO3 2009; Majumdar et al., 2011; Pichler and Melchior, 2002; Zhang and Ubc9, the E2 SUMO-conjugating enzyme. Our choice of et al., 2002). Notably, SUMOylation is not restricted to soluble SUMO3 over the two other SUMO isoforms (SUMO1 and proteins but also plays important roles in the function and SUMO2) is based on its higher accumulation on modified subcellular trafficking of integral membrane proteins. For proteins (Mukherjee et al., 2009) and its greater cytoplasmic example, the functional properties of several K+ channels are localization and thus availability for conjugation to plasma altered by SUMO modification (Benson et al., 2007; Plant et al., membrane proteins (Ayaydin and Dasso, 2004; Benson et al., 2010; Plant et al., 2011; Qi et al., 2014). Interestingly, 2007). Western blot analysis of the samples using an anti-HA SUMOylation also regulates the trafficking and surface expression antibody revealed that HA-immunoreactive bands can be detected of a subset of transmembrane proteins such as the membrane in samples expressing AC3–GFP but not in those derived from cells insertion of the GluA1 AMPAR subunit (also known as GRIA1) expressing HA–SUMO3 and GFP (Fig. 1C, left). Comparison to and the activity dependent increase in AMPAR surface expression samples probed with an anti-AC3 antibody indicated that the HA (Jaafari et al., 2013). These observations led us to consider that the immunoreactive species migrate with an apparent molecular mass conjugation of SUMO to polytopic membrane proteins could that is greater than the main AC3–GFP form but coincided with influence their ciliary localization. minor AC3 immunoreactive species (Fig. 1C, right). The minor Here, we demonstrate a new and direct role of SUMOylation in intensity of these bands is consistent with the relatively low steady the ciliary localization of the adenylyl cyclase isoform AC3. The state of SUMO modification of most proteins (Benson et al., 2007; results from this study demonstrate that AC3 is a substrate for Johnson, 2004). Notably, analysis of samples from cells expressing SUMOylation and that this modification is found on endogenous an AC3-SUMO mutant showed that the upper HA-immunoreactive AC3 from olfactory cilia. Altering the SUMOylation status of band was not present, further indicating a specificity of SUMO AC3, either through overexpression of SUMO peptidases or binding to K465. These data demonstrate that AC3 can be modified mutation of the SUMO acceptor site within the protein, inhibits by SUMOylation in a heterologous system. To test whether ciliary localization. Using the phylogenetically related integral endogenous AC3 is a SUMO substrate, olfactory mucosa was membrane proteins, annoctamin 1 (ANO1) and annoctamin 2 isolated from wild-type mice in either the presence or absence of (ANO2) which differ in their ciliary localization and possession N-ethylmaleimide (NEM), a SUMO-specific deconjugating of SUMOylation motifs, we also show that introduction of enzyme inhibitor (Benson et al., 2007). Immunoprecipitation of SUMOylation sites is not sufficient to drive entry into the AC3 was performed with a rabbit anti-AC3 antibody from the tissue cilium. These results are the first to show that SUMO modification lysate. Blotting for endogenous SUMO2 and SUMO3, we were able of transmembrane proteins regulates their enrichment in the to detect high molecular mass bands in samples that had been treated primary cilium and further highlight the similarities in the with NEM (Fig. 1D). These bands were not present in the samples mechanisms used for nuclear and ciliary entry. prepared in the absence of NEM, indicating that SUMO2 and SUMO3 were interacting with AC3. As endogenous AC3 from RESULTS the olfactory epithelium is highly glycosylated, the native band Adenylyl cyclase 3 contains a conserved consensus runs higher than the predicted molecular mass. In the olfactory SUMOylation motif and is a SUMO substrate in vitro and epithelium samples several AC3 bands that had been modified in vivo with SUMO2 or SUMO3 appear to be closer in size to the Given the established role for SUMO modifications in regulation glycosylated protein as the shift in molecular mass is not as great. of both membrane and nuclear trafficking, we sought Additionally several SUMO bands were also identified running to investigate a potential role for SUMOylation in regulating between 120 and 150 kDa. These bands might correspond to ciliary localization of AC3. Bioinformatic analysis of AC3 modifications of the native AC3 proteins, which has a molecular identified the presence of the consensus SUMOylation motif mass of ∼120 kDa. Ψ-K-X-D/E (Ψ is any large, hydrophobic residue), centered on K465 in the predicted third intracellular loop of the protein SUMO isoforms differentially associate with cilia (Fig. 1A) (Ren et al., 2009). Modeling of SUMO3 in conjunction Our results provide strong evidence that AC3 is covalently modified with the sequence of AC3 reveals that binding of SUMO3 would by SUMO both in vivo and in vitro. However, because AC3 also occur in a manner that would not inhibit interaction of AC3 with localizes to other regions of the cell, we sought to determine whether Journal of Cell Science

1935 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Fig. 1. AC3 harbors conserved SUMOylation motifs and is modified by SUMOylation in vitro and in vivo. (A) Diagram of AC3 protein structure with the 12 ransmembrane domains, and two catalytic domains highlighted. SUMOylation occurs at the consensus sequence Ψ-K-X-D/E (Ψ is any large hydrophobic residue), with SUMO being conjugated to the lysine residue. SUMOsp 2.0 software revealed the presence of predicted SUMOylation sites in AC3 proteins from multiple species. (B) Proposed model of AC3, with a single conjugated SUMO protein. This model predicts that interaction of AC3 with SUMO would not necessarily interfere with Gαs binding. (C). Immunoprecipitation experiments from HEK293 cells co-transfected with GFP, AC3–GFP (AC3:GFP), or AC3K465R– GFP (AC3K465R:GFP), and HA–SUMO3 and Ubc9. Immunoprecipitation was performed with rabbit anti-GFP and immunoblotted (IB) with mouse anti-HA antibody (left panel). The membrane was then stripped and blotted with rabbit anti-AC3 antibody (right panel). Arrowheads denote SUMO-modified AC3 bands. The arrow marks the major AC3 band. (D) Endogenous AC3 is SUMOylated. Whole olfactory epithelium was homogenized either in the absence (–) or presence (+) of N-ethylmaleimide (NEM). Homogenates were pre-cleared with Protein A beads, and immunoprecipated with rabbit anti-AC3 antibody. Samples were probed with a monoclonal antibody against SUMO2 and SUMO3 (left panel), which only revealed bands in the NEM-treated sample. Probing for AC3 (right panel) shows similar efficiencies in immunoprecipitation. Numbers to the left of blots indicate molecular mass (kDa). participants in the SUMOylation pathway associate with the cilium. the apical surface of the olfactory epithelium (Fig. 2B). To further In mammals, three SUMO isoforms competent for conjugation have localize GFP–SUMO3 in OSNs and avoid possible fixation artifacts been identified, of which SUMO3 expression is considered more we used live en face imaging, which allows for a top down cytoplasmic. To determine whether the extra-nuclear expression of perspective of the dendritic knobs (McIntyre et al., 2012; Williams this isoform localizes to cilia we expressed GFP-tagged SUMO3 in et al., 2014). Wild-type mice were injected with GFP–SUMO3 and both cultured cells and in vivo in native OSNs. In ciliated MDCKII Arl13b–mCherry, to effectively label the cilia projecting from each cells, GFP–SUMO3 was found throughout the cell cytoplasm; knob. In en face images, GFP–SUMO3 is strongest in the knobs, however it also accumulated at the base of cilia labeled with with weak signals in the cilia, seen by minimal colocalization with acetylated α-tubulin (Fig. 2A). The SUMO3 signal also colocalized Arl13b–mCherry (Fig. 2D). However, some signal is obvious, with γ-tubulin, a marker for the basal bodies of cilia (Fig. 2C). In suggesting that it is capable of entering. As the dendritic knobs contrast, SUMO1 and SUMO2 were found to be mostly localized to possess the basal bodies for each cilium, we analyzed 1-µm-thick the nucleus and not enriched at the base of cilia (supplementary images. In these it is evident that the GFP–SUMO3 signal is material Fig. S1). To test whether GFP–SUMO3 localization to the strongest as a ring around the knob, adjacent to projections of ciliary base occurred in native OSNs, wild-type mice were Arl13b–mCherry cilia (Fig. 2D, inset). Given our results intranasally injected with a GFP–SUMO3 adenovirus. After demonstrating that AC3 is a substrate for SUMO3 modification, waiting 10 days to allow for virus expression, animals were killed the lack of robustly detectable SUMO3 in the primary cilium was and coronal sections were made through the olfactory epithelium. In surprising. A possible explanation for this is that the SUMO OSNs, GFP–SUMO3 localized throughout the cell body and modification is transient and combined with a potentially limited nucleus, and appeared to be enriched at the base of cilia in the fraction of ciliated proteins that are SUMO modified, the dendritic knobs, but was not detected in the cilia layer, found along modification is difficult to detect under native conditions. Journal of Cell Science

1936 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Fig. 2. Proteins involved in the SUMOylation pathway associate with cilia. (A) Transfection of MDCK cells with GFP–SUMO (GFP: SUMO3), shows accumulation of SUMO3 at the base of cilia, marked by acetylated α-tubulin. (B) Immunostaining for the basal body protein, γ-tubulin corroborates the enrichment at the base of the cilium. (C) Ectopic expression of GFP–SUMO3 in OSNs reveals enrichment in dendritic knobs and cell bodies in fixed coronal sections of olfactory epithelium. (D) En face images of OSN knobs transduced with GFP–SUMO3 and Arl13b–mCherry (Arl13b: mCherry). The GFP signal is strongest in the knob, but can be seen in the cilia. Inset shows an image from a 1-µm-thick section, in which it is evident that the GFP–SUMO3 signal is strongest as a ring around the knob, at the base of the projecting cilia. (E) When expressed in OSNs by adenovirus transduction, HA–SUMO3 (HA: SUMO3) is enriched in the nucleus and dendritic knobs, with the HA signal not apparent in the cilia layer. (F) Non-cleavable HA– SUMO3-Q89P (HA:SUMO3Q89P), however localizes to olfactory cilia. OSNs were co-transduced with GFP adenovirus to label transduced neurons as a control. Scale bars: 10 μm.

Therefore to enrich the pool of ciliary proteins that are modified Proteolytic cleavage of SUMO alters ciliary localization AC3 by SUMOylation we used a non-cleavable SUMO protein mutant, To test for a role of SUMOylation in ciliary trafficking, we utilized SUMO3-Q89P that is resistant to removal mediated by the a truncated form of SENP2 in which the first 72 amino acids are SENP family of peptidases once conjugated to a target protein removed (ΔSENP2) (Benson et al., 2007; Hang and Dasso, 2002; (Mukherjee et al., 2009). Wild-type SUMO3 and SUMO3-Q89P Zhang et al., 2002). By truncating the N-terminus of SENP2, it is with N-terminal HA tags were then expressed in OSNs by released from its normal, nuclear localization to the cytoplasm. adenoviral transduction. Animals were co-injected with a GFP- Previous studies have therefore used overexpression of ΔSENP2 to expressing adenovirus as a secondary label for transduced OSNs. dramatically change SUMO modification of all cytoplasmic Similar to GFP–SUMO3, wild-type HA–SUMO3 was localized to proteins, as it is known to have high activity for cleaving all three the dendritic knobs and cell bodies but absent from cilia (Fig. 2E). SUMO isoforms (Benson et al., 2007). Interestingly, we found that In contrast, whereas HA–SUMO3-Q89P localizes to the cell body the truncated version was enriched at the base of cilia in both and dendrites, it was also detected in the cilia projecting from OSNs IMCD3 cells (Fig. 3A) and MDCKII cells (data not shown), (Fig. 2F). These data demonstrate that SUMO3 associates with the suggesting that it would be positioned to act on ciliary-targeted cilia of multiple cell types. Furthermore, the non-cleavable SUMO proteins. With this localization, it is possible that overexpression of mutant data suggests that endogenous SUMO-modified proteins ΔSENP2 could therefore deSUMOylate proteins that function or can exist within cilia of native cells in vivo. associate with primary cilia affecting the formation or maintenance Journal of Cell Science

1937 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Fig. 3. Overexpression of ΔSENP2 blocks ciliary localization. (A) Transduction of IMCD3 cells with GFP–SENP2 (GFP:SENP2, top) reveals SENP2 localization to the nuclear membrane, whereas expression of a N-terminus truncation (amino acids 1–82) of SENP2 (GFP: ΔSENP2; bottom) results in accumulation at the base of cilia. Neither SENP2 nor ΔSENP2 alters the presence of cilia as seen by immunostaining for acetylated α-tubulin in IMCD3 cells. (B,C) Quantification of the percentage of ciliated cells following transfection with either GFP, GFP–SENP2, or GFP–ΔSENP2 in both IMCD3 (B) or MDCKII (C) cells. Results are mean±s.e.m. (n=3 replicates per treatment). (D) AC3–GFP (AC3:GFP) localizes to cilia of MDCK cells as seen by colocalization with acetylated α-tubulin. Co-expression of AC3–GFP with ΔSENP2 prevents AC3–GFP localization in cilia. (E) Co-expression of GFP–ΔSENP2 with Arl13b– mCherry does not alter its entry to the cilium as seen by colocalization with acetylated tubulin. Scale bars: 10 μm.

of these cell organelles. To test whether global deSUMOylation already possess a cilium at the time of transfection, this result altered cilia presence, cultured cells were transfected with GFP, indicates that altering the SUMOylation state of the cell does not GFP–SENP2 or GFP–ΔSENP2. Two different cell types were used affect cilia maintenance. To determine whether ciliogenesis is to analyze these effects. MDCK cells were transfected, after ciliation altered, IMCD3 cells were transfected with the same constructs, and had occurred and then analyzed 48 hours later. No significant then induced to form cilia by serum starvation. Under this paradigm, difference was detected in the percentage of cells that possessed a we also found no change in the percentage of ciliated cells primary cilium (Fig. 3B). As MDCK cells are grown so that they indicating the SUMOylation is not crucial for the formation of cilia Journal of Cell Science

1938 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

(Fig. 3C). To determine whether overexpression of ΔSENP2 had and mutant protein localized to the cell membrane. However, whereas more subtle effects on cilia, we measured cilia lengths from IMCD3 AC3–GFP usually colocalized with the cilia marker polyglutamylated cells. Compared to GFP-transduced cells, lengths of primary cilia tubulin (64/74 cells), AC3-K465R–GFP was rarely seen in the primary transduced with GFP–ΔSENP2 were not significantly different cilium (9/90 cells) (Fig. 5A,B). Although these results indicate that the (5.23±0.243 μm and 5.100±0.243 μm respectively). SUMO motif is necessary for AC3 localization, we sought to To test whether SUMOylation is important for ciliary localization determine whether SUMOylation regulated ciliary trafficking of other of AC3, we co-expressed AC3–GFP with ΔSENP2 in MDCK cells. transmembrane proteins. The Ca2+-activated Cl− channel, ANO2 also When expressed with an empty vector, AC3–GFP was found in the localizes to cilia, and possesses two putative SUMO sites. To test the cilium of over 90% of cells (n=37 cells) (Fig. 3D). In contrast, co- role of SUMOylation in ANO2 trafficking we mutated both lysine expression of AC3–GFP with ΔSENP2 severely reduced ciliary residues of the putative SUMOylation sites, at position 876 and 888, to localization where only 7% of cilia (n=35 cells) were positive for arginine residues. Similar to AC3, the ablation of the putative AC3–GFP. As a control for specificity, we also tested ciliary SUMOylation sites in ANO2 also resulted in its loss of localization to localization of Arl13b. Co-expression of Arl13b–mCherry, with the primary cilium (Fig. 5C,D). GFP–ΔSenp2 revealed that changing the SUMOylation state of the The function of AC3 is perhaps best associated with its role in cell also did not affect its localization (Fig. 3E). This is consistent olfactory cilia, where it is necessary for the olfactory transduction with a previous report showing that mutation of SUMO sites in pathway. Therefore, to test the role of SUMOylation in ciliary Arl13b does not affect its localization (Li et al., 2012). trafficking of AC3 in vivo, wild-type mice were intranasally injected with either AC3–GFP or AC3-K465R–GFP adenovirus constructs. Disruption of SUMOylation motifs inhibits ciliary localization Following a period of 10 days to allow virus expression, animals of AC3 were killed, and the olfactory epithelium was section for As overexpression of ΔSENP2 effectively removes SUMO from all immunofluorescence analysis. AC3:GFP signal was detected proteins, mislocalization of AC3 in the presence of ΔSENP2 does along the apical layer of the olfactory epithelium and colocalized not address whether the targeting defect was directly due to altering with acetylated-tubulin (Fig. 5E), indicating ciliary localization. the SUMOylation status of AC3. It could be that removal of SUMO In contrast, whereas AC3-K465R–GFP localized to the cell from other proteins was responsible for the ciliary trafficking membrane, in a similar manner to AC3–GFP, the fluorescence defects of AC3. To determine whether direct SUMOylation of AC3 signal was not seen along the ciliary layer and instead only an is a crucial determinate of ciliary localization, we mutated its intense fluorescence of the knob was seen (Fig. 5F), indicating that predicted SUMOylation site, K465, to an arginine residue. As AC3 AC3-K465R–GFP is not enriched in the cilia. Analysis of OSN is an integral membrane protein, we hypothesized that changes to axons revealed that both constructs localized to the full length of membrane localization would alter cAMP production. To test for axons along the olfactory nerve layer and in glomeruli in the possible functional changes in cAMP, we assessed the function of olfactory bulb (Fig. 5G,H), providing further evidence that ablation AC3-K465R compared to wild-type AC3. HEK293 cells were of K465 does not affect membrane trafficking of AC3 in general. transfected with the constructs and cAMP production in response to To confirm that the loss of ciliary labeling was not due to fixation forskolin stimulation was measured. Cells expressing AC3-K465R artifacts we used live en face imaging of OSNs transduced with produced similar levels of cAMP to those expressing wild-type AC3 either AC3–GFP or AC3K465R–GFP. As a means to visualize (Fig. 4). These data provide an indication that the K465R mutation olfactory cilia and verify their presence, animals were also intranasally does not alter function of AC3, indicating proper localization to the injected with an adenovirus expressing Arl13b–mCherry. In viewing surface membrane. the cilia en face as they extend from a dendritic knob, it is readily The ability of AC3-K465R to localize to primary cilia was then apparent that AC3–GFP and Arl13b–mCherry colocalized along the tested in cultured cells. IMCD3 cells were transduced with AC3–GFP entire length of the cilium (Fig. 6A). Unlike AC3–GFP, AC3-K465R– or AC3-K465R–GFP adenovirus, and serum starved 24 hours later to GFP was restricted to dendritic knobs and was not apparent in cilia. induce ciliogenesis. Confocal imaging revealed that both the wild-type The presence of Arl13b–mCherry-expressing cilia projecting from GFP-positive knobs demonstrates that these transduced OSNs possess cilia, but that AC3-K465R–GFP is not enriched in the cilia (Fig. 6B). Line-scan analysis of the fluorescent intensity of AC3–GFP compared to AC3-K465R–GFP confirmed a significant reduction in enrichment of AC3-K465R–GFP in olfactory cilia (Fig. 6C). Interestingly, the signal intensity for AC3-K465R–GFP is greater at the base of the cilium in both cultured cells and neurons compared to AC3–GFP, suggesting that because it is restricted from entering the cilia, it accumulates in this region of the cell (Fig. 6D). Arl13b–mCherry signals were comparable between animals. Taken together, these data suggest that ciliary localization of AC3 is directly dependent upon its ability to be SUMOylated.

SUMOylation is not sufficient for ciliary localization Fig. 4. AC3-K465R mutation does not alter function of AC3. HEK293A cells Mutation of the putative SUMOylation site in AC3 demonstrates the were transfected with 1 µg of GFP, AC3–GFP (AC3:GFP) or AC3-K465R–GFP necessity for these sequences but does not address whether (AC3K465R:GFP). At 48 hours after transfection, cells were stimulated with SUMOylation is sufficient to permit entry of a transmembrane forskolin for 10 minutes at the indicated concentrations, then collected to measure cAMP levels. Mutation of the SUMOylation site did not alter forskolin- protein. However, bio-informatic analysis reveals that each of the stimulated cAMP production by AC3. Results are mean±s.e.m. (n=3 per ten type III adenylyl cyclase family members contains a putative treatment). *P<0.05; **P<0.005 (ANOVA). SUMOylation motif. In addition, several are known to localize to Journal of Cell Science

1939 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Fig. 5. SUMOylation sites are necessary for ciliary localization of transmembrane proteins. (A,B) Site-directed mutagenesis of the SUMOylated lysine (K465) residue to arginine (R) prevents AC3–GFP from localizing to cilia (arrows). Wild-type AC3–GFP colocalizes with polyglutamylated tubulin, whereas AC3- K465R–GFP accumulates at the base of cilia. (C,D) Site-directed mutagenesis of both lysine residues to arginine (K876R and K888R) prevents ANO2–GFP from localizing to cilia (arrows). Wild-type ANO2–GFP colocalizes with acetylated α-tubulin (Acet. tub), whereas ANO2-K876,888R–GFP does not. (E,F) Coronal sections through the olfactory epithelium of mice transduced with either AC3–GFP or AC3-K465R–GFP. AC3–GFP colocalizes with acetylated α-tubulin (red) along the ciliary layer, whereas AC3-K465R–GFP is restricted to the dendritic knobs. (G,H) Both AC3–GFP and AC3-K465R–GFP localize to OSN axons as GFP- expressing axons can be seen innervating glomeruli (outlined) in the olfactory bulb. ONL, Olfactory nerve layer. Scale bars: 10 μm(A–F); 50 μm (G,H). cilia, meaning that this class of proteins is inappropriate to test for colocalization was detected with ciliary markers, ANO1–GFP was sufficiency. However, another olfactory transmembrane signaling not detected in the cilia of OSNs (Fig. 7A,B) nor in those of MDCK protein, ANO2 (Billig et al., 2011; Stephan et al., 2009), contains or IMCD3 cells (supplementary material Fig. S2A,B). Using two putative SUMO consensus motifs in the intracellular C-terminal site-directed mutagenesis we added the crucial lysine residues in tail. Fortuitously, ANO2 belongs to a gene family with a naturally ANO1 to reconstitute the two putative SUMOylation sites both occurring isoform, ANO1, which lacks these predicted SUMO individually and in combination. By using dual injection with motifs. Aligning the C-terminal tail of ANO1 and ANO2 reveals Arl13b–mCherry, we analyzed ciliary localization of the GFP- two homologous regions; however, in ANO1 the crucial lysine tagged proteins en face in live OSNs. Although removal of residues to which SUMO can conjugate are not present (Fig. 7A). In SUMOylation sites had a clear effect on ciliary localization of AC3 the nasal cavity, ANO1 localizes to the microvilli of vomeronasal in vivo, the addition of putative SUMOylation sites to ANO1 did not sensory neurons and sustentacular cells, but is not detected in increase its ciliary enrichment (Fig. 7C, supplementary material ciliated OSNs (Billig et al., 2011; Maurya and Menini, 2014). Fig. S2C,D). Line-scan analysis of signal intensity further Therefore, ectopic expression of ANO1 is an ideal probe to test for confirmed the lack of ANO1 variants in the cilia of transduced the sufficiency of SUMOylation in targeting proteins to cilia. To OSNs (Fig. 7). Thus, SUMOylation might be necessary for ciliary address this question, we took advantage of the natural amino acid enrichment of some proteins, but it is not sufficient to promote differences between ANO2 and ANO1 to ask when artificial localization of an protein that is otherwise excluded from cilia. reconstitution of SUMOylation sites would allow ciliary localization. To confirm that ANO2 and ANO1 localize DISCUSSION differently to cilia we expressed GFP tagged wild-type constructs Primary cilia are crucial cellular organelles that on many cells act in cultured cells and native OSNs. Although ANO2–GFP as biological antennae detecting extracellular signals. Their Journal of Cell Science

1940 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Fig. 6. Mutation of the SUMOylation site in AC3 site abolishes ciliary localization in live OSNs. En face images of live OSNs transduced with either AC3–GFP (AC3:GFP) or AC3-K465R–GFP. To confirm to presence of cilia OSNs were co-transduced with Arl13b–mCherry (Arl13b:mCherry). (A) AC3–GFP colocalizes with Arl13b– mCherry along the full length of the cilia, whereas AC3- K465R (B) is restricted to the dendritic knob. Scale bars: 10 μm. (C,D) Quantification of signal intensity of the GFP from the base of the cilium to the tip. Results are mean±s.e.m. (n=8 cilia AC3, 13 cilia AC3-K465R).

importance is underscored by the genetic disorders, termed show that AC3 is a SUMO substrate and that mutation of the SUMO ciliopathies, associated with dysfunctional cilia. Although some site blocks SUMO3 binding. Thus, the loss of AC3 localization of these disorders are caused by mutations altering cilia structure or when the SUMO site is mutated can be assumed to be due to the loss presence, many are associated with mutations affecting function or of SUMOylation. That AC3-K465R still localizes to the OSN axons ciliary localization of signal transduction proteins. In this way, and produces cAMP in response to forskolin stimulation indicates mutations targeting specific amino acid sequences that are necessary that SUMOylation of AC3 does not affect global processes, but for specific trafficking mechanisms could cause the crucial proteins rather might be specific for aiding ciliary entry. to fail to localize to cilia, leading to disease phenotypes. How Our work demonstrates a new role fora protein modification process signaling proteins are targeted to and trafficked into cilia is therefore in the regulation of specific ciliary proteins. Ciliary enrichment is a key to understanding the mechanisms and potential treatments of complicated process for which multiple mechanisms have been ciliopathy disorders. The results of the studies presented here identified that often act in concert for the trafficking of individual provide insight into the requirements for trafficking transmembrane proteins (Berbari et al., 2008a; Bhogaraju et al., 2013; Geng et al., proteins into the cilium, and demonstrates a direct functional role for 2006; Hurd et al., 2010; Hurd et al., 2011; Jenkins et al., 2006; SUMO modification in ciliary trafficking in an in vivo system. Jenkins et al., 2009; Kee et al., 2012; McEwen et al., 2007; Sharma Although SUMOylation has been most heavily studied in et al., 2008). Additionally, these mechanisms often regulate a the context of nuclear import and transcription regulation, recent subset of ciliary-targeted proteins but are not ubiquitous in their studies have demonstrated roles for SUMOylation, both direct requirement. Furthermore, individual proteins in specific classes are and indirect, in the function and trafficking of transmembrane not necessarily regulated by the same mechanisms. For example, proteins, particularly in neurons (Luo et al., 2013). In the synapse, ciliary localization of some GPCRs is driven by a motif found in the SUMOylation of the kainate receptor GluK2 (also known as third intracellular loop; however, this motif is largely absent from GRIK2) is needed for induced internalization (Martin et al., 2007). olfactory receptors, which are ciliary localized (Berbari et al., Furthermore SUMOylation of K+ channels at the membrane is 2008a). Sequence analysis of other proteins known to be enriched in able to alter pore properties. Indirect roles for trafficking of olfactory cilia suggests that SUMOylation might regulate ciliary transmembrane proteins have also been identified. For example, localization of some, but not all of these proteins. In addition to SUMOylation of collapsing response mediator protein 2 (CRMP2, ANO2, SUMO motifs are present in CNGB1b, CNGA2 and GNG1. also known as DPYSL2) is necessary for surface localization of the However, SUMOylation motifs are not present in several other Na+ channel Nav1.7 (Dustrude et al., 2013). Similarly, an indirect proteins that localize to olfactory cilia, including olfactory marker role for SUMOylation has been postulated in the trafficking of protein (OMP), GNAL, GNB1, CNGA4 and five olfactory receptors specific transmembrane proteins into the cilia of C. elegans (Li that were analyzed. Our data suggests that disruption or mutations in et al., 2012). In that work, Arl13b was shown to harbor SUMOylation pathways within a cell could alter the localization of SUMOylation sites, although mutation of these sites did not affect ciliary proteins. Additionally, mutations in SUMOylation motifs in its localization to cilia. In Arl13b-null worms, the transmembrane signaling proteins targeted to the cilium could underlie disease proteins PKD-2 and ODR-10 are mistrafficked, showing aberrant phenotypes due to changes in trafficking. localization in dendrites. Whereas expression of wild-type Arl13b An interesting phenomenon in the localization of proteins to cilia can correct this defect, the SUMO mutant Arl13b is unable to do so. is that several mechanisms have been identified that are necessary These studies demonstrate a role for SUMOylation of Arl13b in the but not sufficient for ciliary entry. For this study, we were able to complete ciliary localization of PKD-2 and ODR-10. However, take advantage of a naturally occurring difference between two as both were still present in the cilium, these experiments do not highly homologous proteins. Although the majority of the demonstrate a requirement for SUMOylation in ciliary entry. Here, SUMOylation sequence is present in ANO1, both of its potential we have shown that direct SUMOylation of AC3 is a crucial motifs lack the crucial lysine residues that are present in ANO2. determinant of its ciliary enrichment. Immunoprecipitation studies Interestingly, ANO1 and ANO2 show differences in their Journal of Cell Science

1941 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Fig. 7. SUMOylation sites are not sufficient for localization of ANO1 to olfactory cilia. (A) Sequence alignment of the C-terminus of ANO2 and ANO1 reveals the SUMOylation motifs in ANO2 and conserved substitutions in ANO1. (B) En face images of live OSNs co-transduced with ANO2–GFP and Arl13b–mCherry reveal colocalization within the cilia. (C) Co-transduction of OSNs with ANO1–GFP and Arl13b– mCherry shows that ANO1 does not localize to cilia. (D) Mutation of H955 and R967 in ANO1 to lysine residues are not sufficient to permit ciliary entry of ANO1 as seen by lack of colocalization with Arl13b–mCherry in transduced OSNs. Quantitative results are mean±s.e.m. (n=10 cilia ANO2, 10 cilia ANO1, 11 cilia ANO1-H995KR967K). Scale bars: 10 μm.

localization patterns in the nasal cavity. ANO2, is expressed in Several mechanisms known to regulate ciliary trafficking of OSNs and localizes to cilia, whereas ANO1 is found in sustentacular transmembrane proteins, such as co-factor interaction, nuclear cells and vomeronasal sensory neurons, both of which possess import signals and phosphorylation are dispensable for Gli2 ciliary microvilli (Billig et al., 2011). The lack of SUMO sites in ANO1, localization (Santos and Reiter, 2014). Additionally, Santos and paired with its lack of ciliary localization allowed us to address Reiter found that mutation of the Gli2 SUMOylation motif did not whether these motifs are sufficient for entry. Reconstitution of the affect ciliary enrichment of the protein. Interestingly, the Gli2 SUMO motifs, both individually and combined, however, was SUMOylation mutants display increased transcriptional activity, unable to turn ANO1 into a ciliary-targeted protein. This finding is raising the possibility that SUMO modification might be important not without precedent, as several other trafficking mechanisms have for limiting Gli2 in the absence of activation and maintaining it in also been found to be necessary but not sufficient for ciliary the cilium. Santos and Reiter also reported being unable to detect targeting. For example, trafficking of CNGA2 in cilia requires components of the SUMO pathway at or near the primary cilia of interaction with CNGB1b, which possesses an RVxP motif. cultured cells. This is in contrast to our findings for SUMO3, which Insertion of this motif into CNGA2 is insufficient for targeting it was associated with cilia, and that of another study, which found to the cilium in the absence of CNGB1b. The lack of ciliary GFP-tagged Ubc9 in the cilia of C. elegans (Li et al., 2012). Thus, it localization and SUMO sites in mammalian ANO1 is interesting is likely that the use of tagged ectopically expressed proteins in our considering a recent report showing that C. elegans ANO1 localizes study aided in the identification of SUMO pathway proteins to sensory cilia (Wang et al., 2013). Analysis of ANOH-1, the associated with the cilium. The difference in results also raises the C. elegans homolog, reveals an intracellular SUMO motif in the possibility that endogenous SUMO and SENP proteins are difficult N-terminus of the protein. The presence of this SUMO site, in to detect, possibly requiring specific fixation or antigen retrieval conjunction with other unidentified mechanisms could explain the techniques to reveal their localization. These findings provide difference in ciliary localization of C. elegans and mammalian further evidence that that the diversity of mechanisms used for cilia ANO1. localization varies, likely due to the membrane interactions of The loss of AC3, but not Arl13b, targeting to cilia due to changes individual proteins and the cell type they are found in. in SUMOylation state suggests that membrane interactions of The identification that SUMOylation regulates protein entry into ciliary-targeted proteins could contribute to how SUMO affects the cilium adds to growing evidence of similarities and shared localization. It is possible that SUMOylation is important for ciliary mechanisms regulating nuclear and ciliary trafficking. Previous localization of transmembrane but not cytoplasmic proteins, studies have identified other similarities including the use of RAN particularly given that cytoplasmic proteins can vary greatly in the gradients, importin interactions, nuclear localization signals, size localization mechanisms used. This idea is supported by recent exclusion and the presence of nucleoporin proteins (NUPs) at the analysis of how the transcription factor Gli2 localizes to cilia. cilia base (Dishinger et al., 2010; Fan et al., 2011; Hurd et al., 2011; Journal of Cell Science

1942 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Kee et al., 2012; Kee and Verhey, 2013). An important role for The inserts were then recombined into the adenoviral vector pAD/V5/-dest NUPs is regulating SUMOylation of proteins trafficking through the using LR Recombinase II (Life Technologies, Carlsbad CA). Viral plasmids nuclear pore. Some NUPs, such as Nup358 are recognized SUMOP were digested with PacI and transfected into HEK293 cells. Following an E3 ligases (Pichler et al., 2002), whereas others help to keep SENP initial amplification, a crude viral lysate was produced, and used to infect proteins at the nuclear pore complex. The finding of NUPs at the confluent 60-mm dishes of HEK293 cells for amplification according to the ViraPower protocol (Life Technologies). Adenovirus was isolated with the base of cilia provides further support for SUMOylation being a Virapur Adenovirus mini purification Virakit (Virapur, San Diego, CA), mechanism that regulates cilia entry. More recently, another SUMO dialyzed in 2.5% glycerol, 25 mM NaCl and 20 mM Tris-HCl, pH 8.0, and E3 ligase, TOPORS, was found to localize to be the base of the stored at −80°C until use. connecting cilium in photoreceptors (Chakarova et al., 2011; Weger et al., 2003). The identification of SUMOylation as a regulator of Immunoprecipitations ciliary trafficking adds an additional link between nuclear and For immunoprecipitation from native olfactory epithelium, CD-1 mice ciliary similarities. were deeply anesthetized with CO2, and decapitated. All animal In summary, our results demonstrate a significant role for experiments were performed according to approved guidelines (see SUMOylation in the ciliary localization of select olfactory section on intranasal injections). Olfactory epithelium was dissected into signaling proteins. Chronic overexpression of the SUMO protease in immunoprecipitation buffer containing 10 mM Tris-HCl, pH 7.5, 1 mM SENP2 prevented the ciliary localization of AC3. This effect is EDTA, 150 mM NaCl and protease inhibitors with 1% Nonidet P-40. specific for trafficking mechanisms because overexpression of Samples with NEM contained 20 mM NEM. The tissue was then homogenized on ice and rocked for 1 hour on a nutator at 4°C in 1 ml of SENP2 did not affect the presence or length of cilia. In agreement immunoprecipitation buffer. The lysate was then centrifuged at 1000 g to with this, we found that mutation of the SUMOylation site in AC3 pellet nuclei for 5 minutes. Supernatant was then rocked with 100 µl of blocked ciliary entry in both cultured cells and in vivo in OSNs. The protein A agarose beads for 1 hour, centrifuged at 10,000 g to clear beads mutation not only blocked localization, but also appeared to cause andthenincubatedwith2µlofrabbitanti-AC3antibody(SantaCruz an accumulation of protein at the ciliary base. In addition Biotechnology, sc-588) overnight at 4°C. For immunoprecipitations from localization of AC3-K465R in the axons of OSNs was unaffected. cells, at 48 hours post-transfection, HEK293 cells were lysed on ice for Thus AC3 trafficking to the membrane in general is not altered by 30 minutes in 500 µl immunoprecipitation buffer. After douncing, cells mutation of the SUMO site, but entry into the cilium is inhibited. were centrifuged at 1000 g for 5 minutes and supernatant was incubated These results indicate that the SUMO modification is needed for overnight at 4°C with antibody-conjugated beads. 5% of the supernatant passing through the ciliary base. That SUMO itself was not found was used for starting material. Beads were washed three times for 5 minu with immunoprecipitation buffer with 1% Triton X-100 and once with accumulated in the cilium indicates that it might be cleaved from immunoprecipitation buffer without Triton X-100. Supernatant was modified proteins and rapidly removed upon moving through the analyzed by SDS-PAGE as described previously (Jenkins et al., 2006; ciliary pore. Whether SUMOylation affects trafficking of other Jenkins et al., 2009). transmembrane proteins into cilia remains to be tested. Interestingly, initial screens of odorant receptors did not identify putative SUMO Cell culture sites leaving the mechanisms regulating the trafficking of these Madin Darby canine kidney (MDCKII) cells were grown in Dulbecco’s important proteins still largely unknown. Our results clearly link modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine SUMO modification of signaling proteins to ciliary entry and serum and 1% penicillin, streptomycin and Gluta-MAX I. MDCK cells highlight a role for SUMOylation in the subcellular trafficking of were grown 7 days post-confluence on transwell filter supports (0.4 mm pore size, Corning, Tewksbury, MA) prior to fixation. IMCD3 cells were proteins in OSNs. grown in DMEM with F12 supplemented with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. To induce cilia formation, cells MATERIALS AND METHODS were serum starved for 48 hours post-transfection. For the cAMP assay, Bio-informatics HEK293 cells were transfected with 1.5mg of either AC3–GFP for Sequence analysis was performed using SUMOplotsp2.0. The following AC3K465R–GFP plasmid. At 48 hours following transfection, cells were sequences were used for analysis (unless specified all sequences are stimulated with forskolin diluted in DMSO. Cells were incubated at 37°C from Mus musculus); AC1, NP_033752.1; AC2, NP_705762.2; AC3, for 10 minutes then collected, and cAMP levels were assayed following the NP_612178.2, NP_004027.2 (Homo sapiens), NP_001156014.1 (Danio manufacturer’s instructions (Cayman Chemical, Ann Arbor, MI). rerio), XP_003641104.2 (Gallus gallus), XP_008116786.1 (Anolis Measurements were made on triplicate samples from three separate carolinensis), NP_001096154.1 (Xenopus tropicalis); AC4; NP_536683.1; experiments. AC5, NP_001012783.3; AC6, NP_031431.2; AC7, NP_001032812.2; For immunostaining, cells were fixed with 4% PFA, quenched with AC8, NP_033753.2; AC9, NP_033754.2; AC10, NP_766617.2; CNGA2, 50 mM NH4Cl in PBS, permeabilized with 0.1% Triton X-100 in PBS, and NP_031750.2; CNGA4, NP_001028489.1; CNGB1, NP_001182342.1; blocked with 0.3% goat serum (Life Technologies) in PBS. Samples were Gnal, NP_796111.2; Gng13, NP_071867.1; Gnb1, NP_001153488.1; incubated in primary antibodies for 1 hour at room temperature, washed with ANO2, NP_705817.2; Olfr15, NP_032788.2,; Olfr2, NP_035113.1; 1×PBS, incubated with fluorophore-conjugated secondary antibodies for Olfr151, NP_997547.1; Olfr73, NP_473431.1; Olfr78, NP_001161975.1; 1 hour, washed with 1×PBS, incubated with DAPI and mounted using ANO1, NP_848757.4. Prolong Gold (Life Technologies). Monoclonal antibody against acetylated α-tubulin (T6793, Sigma-Aldrich) and secondary antibodies were used at Constructs and adenovirus preparation 1:1000 dilutions. For γ-tubulin (T3559, Sigma-Aldrich) immunostaining, Plasmids containing cDNA inserts used in this study were generously cells were fixed with a 50:50 mix of methanol and acetone at −20°C for 20 provided as follows: SENP2 and SUMO3 (Jorge Iniguez-Lluhi, University minutes. Cells were then washed with PBS and treated as previously of Michigan, Ann Arbor, MI), AC3 (Randall R. Reed, Johns Hopkins described. z-series were captured on an Olympus Fluoview 500 confocal University, Baltimore, MD), Arl13b (Tamara Caspary, Emory, Atlanta, GA), microscope equipped with a 405-nm laser diode with a 430–460 nm ANO2 and ANO1 (Haiqing Zhao, Johns Hopkins University, Baltimore, bandpass filter, a 488-nm laser with a 505–525 nm bandpass filter, a 543-nm MD). Disruption of SUMOylation sites were performed using a Strategene laser with a 560-nm long-pass filter, and a 633-nm laser with a 660-nm long- site-directed mutagenesis kit. pass filter. 60×1.40 numerical aperture (N.A.) oil objectives were used. The For expression in native tissue, recombinant GFP- and mCherry-fused fluorescence signals of compressed z-stacks were quantified using NIH cDNAs were cloned into the vector p-ENTR by TOPO cloning methods. ImageJ software. Journal of Cell Science

1943 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Intranasal injections Berbari, N. F., Bishop, G. A., Askwith, C. C., Lewis, J. S. and Mykytyn, K. (2007). All animals were handled according to the guidelines for animal care and Hippocampal neurons possess primary cilia in culture. J. Neurosci. Res. 85, procedures approved by the University Committee for the Use and Care of 1095-1100. Berbari, N. F., Johnson, A. D., Lewis, J. S., Askwith, C. C. and Mykytyn, K. Animals at the University of Michigan and the Institutional Animal Care and (2008a). Identification of ciliary localization sequences within the third intracellular Use committee at the University of Florida. Neonatal animals were loop of G protein-coupled receptors. Mol. Biol. Cell 19, 1540-1547. intranasally injected with 20 µl of adenovirus at ages P7–P9. Animals Berbari, N. F., Lewis, J. S., Bishop, G. A., Askwith, C. C. and Mykytyn, K. were held in the supine position and plastic tubing was inserted into the nasal (2008b). Bardet-Biedl syndrome proteins are required for the localization of G opening for virus delivery. Animals were held in this position for 3–5 protein-coupled receptors to primary cilia. Proc. Natl. Acad. Sci. USA 105, minutes following injection. Animals were then returned to their cages for 10 4242-4246. Bhogaraju, S., Engel, B. D. and Lorentzen, E. (2013). days to allow expression of the ectopic protein(s). complex structure and cargo interactions. Cilia 2, 10. − Billig, G. M., Pál, B., Fidzinski, P. and Jentsch, T. J. (2011). Ca2+-activated Cl Tissue preparation currents are dispensable for olfaction. Nat. Neurosci. 14, 763-769. For immunostaining, mice were deeply anesthetized and then cardiac Bishop, G. A., Berbari, N. F., Lewis, J. and Mykytyn, K. (2007). Type III adenylyl J. Comp. perfused with 4% paraformaldehyde (PFA). The lower jaw, teeth and cyclase localizes to primary cilia throughout the adult mouse brain. Neurol. 505, 562-571. hair were removed from snouts prior to post fixation in 4% PFA for Bönigk, W., Bradley, J., Müller, F., Sesti, F., Boekhoff, I., Ronnett, G. V., overnight at 4°C. Snouts were then decalcified in 0.5M EDTA in 1×PBS Kaupp, U. B. and Frings, S. (1999). The native rat olfactory cyclic nucleotide- overnight, then cryoprotected in 10%, 20% and 30% sucrose for 1 hour, gated channel is composed of three distinct subunits. J. Neurosci. 19, 1 hour and overnight, respectively, at 4°C, and embedded in OCT 5332-5347. PubMed compound (Tissuetek). Coronal cryosections were sliced at a thickness of Breslow, D. K., Koslover, E. F., Seydel, F., Spakowitz, A. J. and Nachury, M. V. 10–12 µm and mounted onto Superfrost Plus slides (Fisher Scientific). (2013). An in vitro assay for entry into cilia reveals unique properties of the soluble J. Cell Biol. Immunostaining was carried out as above. diffusion barrier. 203, 129-147. Buck, L. and Axel, R. (1991). A novel multigene family may encode odorant For confocal en face imaging of transduced OSNs, animals were receptors: a molecular basis for odor recognition. Cell 65, 175-187. anesthetized with CO2, rapidly decapitated, split along the cranial Chakarova, C. F., Khanna, H., Shah, A. Z., Patil, S. B., Sedmak, T., Murga- midline, and the olfactory epithelium was dissected out in PBS. The Zamalloa, C. A., Papaioannou, M. G., Nagel-Wolfrum, K., Lopez, I., Munro, P. removed olfactory turbinates were placed with the apical (ciliary) surface et al. (2011). TOPORS, implicated in retinal degeneration, is a cilia-centrosomal against a glass coverslip and held with a tissue slice holder (Warner protein. Hum. Mol. Genet. 20, 975-987. Instruments, Hamden, CT). Turbinates were bathed in 1×PBS for imaging Choi, Y. H., Suzuki, A., Hajarnis, S., Ma, Z., Chapin, H. C., Caplan, M. J., Pontoglio, M., Somlo, S. and Igarashi, P. (2011). Polycystin-2 and on a Nikon Ti eclipse-A1R confocal microscope with a 60×1.40 N.A oil phosphodiesterase 4C are components of a ciliary A-kinase anchoring protein objective. complex that is disrupted in cystic kidney diseases. Proc. Natl. Acad. Sci. USA For imaging of tissue sections, confocal imaging was performed as 108, 10679-10684. described previously (Jenkins et al., 2006; McIntyre et al., 2012). Exposures Dishinger, J. F., Kee, H. L., Jenkins, P. M., Fan, S., Hurd, T. W., Hammond, J. W., were adjusted so that the maximal pixel intensities were at least half saturation. Truong, Y. N., Margolis, B., Martens, J. R. and Verhey, K. J. (2010). Ciliary entry Nat. Cell For adenoviral olfactory epithelium experiments, in order to minimize of the -2 motor KIF17 is regulated by importin-beta2 and RanGTP. Biol. 12, 703-710. contribution of ciliary protein localization from uninfected OSNs, we used Dustrude, E. T., Wilson, S. M., Ju, W., Xiao, Y. and Khanna, R. (2013). CRMP2 compressed z-stacks corresponding only to the thickness of the infected OSN. protein SUMOylation modulates NaV1.7 channel trafficking. J. Biol. Chem. 288, Images were analyzed with ImageJ software (NIH, Bethesda, MD), and 24316-24331. statistics were performed with Prism 5 software from Graphpad Prism Fan, S., Whiteman, E. L., Hurd, T. W., McIntyre, J. C., Dishinger, J. F., Liu, C. J., Software (San Diego, CA). Adjustments of contrast and brightness were Martens, J. R., Verhey, K. J., Sajjan, U. and Margolis, B. (2011). Induction of Mol. Biol. Cell performed using Adobe Photoshop CS5.1 (San Jose, CA). Ran GTP drives ciliogenesis. 22, 4539-4548. Feinstein, P., Bozza, T., Rodriguez, I., Vassalli, A. and Mombaerts, P. (2004). Acknowledgements Axon guidance of mouse olfactory sensory neurons by odorant receptors and the beta2 adrenergic receptor. Cell 117, 833-846. The authors thank members of the University of Florida, Center for Smell and Taste Geng, L., Okuhara, D., Yu, Z., Tian, X., Cai, Y., Shibazaki, S. and Somlo, S. for helpful comments during preparation of this manuscript. (2006). Polycystin-2 traffics to cilia independently of polycystin-1 by using an N-terminal RVxP motif. J. Cell Sci. 119, 1383-1395. Competing interests Guadiana, S. M., Semple-Rowland, S., Daroszewski, D., Madorsky, I., Breunig, The authors declare no competing or financial interests. J. J., Mykytyn, K. and Sarkisian, M. R. (2013). Arborization of dendrites by developing neocortical neurons is dependent on primary cilia and type 3 adenylyl Author contributions cyclase. J. Neurosci. 33, 2626-2638. J.C.M. and J.R.M. designed experiments. J.C.M., A.J. and L.Z. performed Hang, J. and Dasso, M. (2002). Association of the human SUMO-1 protease experiments. J.C.M., L.Z. and J.I-L. generated reagents. J.C.M. and J.R.M. wrote SENP2 with the nuclear pore. J. Biol. Chem. 277, 19961-19966. the manuscript. J.C.M. and J.I-L. made illustrations and generated figures. J.R.M. Hurd, T., Zhou, W., Jenkins, P., Liu, C. J., Swaroop, A., Khanna, H., Martens, J., directed the project. Hildebrandt, F. and Margolis, B. (2010). The retinitis pigmentosa protein RP2 interacts with and regulates cilia-mediated vertebrate development. Funding Hum. Mol. Genet. 19, 4330-4344. This work was supported by the National Institutes of Health [grant numbers Hurd, T. W., Fan, S. and Margolis, B. L. (2011). Localization of retinitis pigmentosa R01DC009606 to J.R.M., and F32DC011990 to J.C.M.]. Deposited in PMC for 2 to cilia is regulated by Importin beta2. J. Cell Sci. 124, 718-726. immediate release. Hutten, S., Flotho, A., Melchior, F. and Kehlenbach, R. H. (2008). The Nup358- RanGAP complex is required for efficient importin alpha/beta-dependent nuclear Supplementary material import. Mol. Biol. Cell 19, 2300-2310. Supplementary material available online at Jaafari, N., Konopacki, F. A., Owen, T. F., Kantamneni, S., Rubin, P., Craig, T. J., http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.164673/-/DC1 Wilkinson, K. A. and Henley, J. M. (2013). SUMOylation is required for glycine- induced increases in AMPA receptor surface expression (ChemLTP) in hippocampal neurons. PLoS ONE 8, e52345. References Jenkins, P. M., Hurd, T. W., Zhang, L., McEwen, D. P., Brown, R. L., Margolis, B., Ayaydin, F. and Dasso, M. (2004). Distinct in vivo dynamics of vertebrate SUMO Verhey, K. J. and Martens, J. R. (2006). Ciliary targeting of olfactory CNG Mol. Biol. Cell paralogues. 15, 5208-5218. channels requires the CNGB1b subunit and the kinesin-2 motor protein, KIF17. Bakalyar, H. A. and Reed, R. R. (1990). Identification of a specialized adenylyl Curr. Biol. 16, 1211-1216. Science cyclase that may mediate odorant detection. 250, 1403-1406. Jenkins, P. M., Zhang, L., Thomas, G. and Martens, J. R. (2009). PACS-1 Benson, M. D., Li, Q. J., Kieckhafer, K., Dudek, D., Whorton, M. R., Sunahara, mediates phosphorylation-dependent ciliary trafficking of the cyclic-nucleotide- R. K., Iñiguez-Lluhı,́ J. A. and MartensJ. R. (2007). SUMO modification gated channel in olfactory sensory neurons. J. Neurosci. 29, 10541-10551. regulates inactivation of the voltage-gated potassium channel Kv1.5. Proc. Natl. Johnson, E. S. (2004). Protein modification by SUMO. Annu. Rev. Biochem. 73,

Acad. Sci. USA 104, 1805-1810. 355-382. Journal of Cell Science

1944 RESEARCH ARTICLE Journal of Cell Science (2015) 128, 1934-1945 doi:10.1242/jcs.164673

Kaneko-Goto, T., Sato, Y., Katada, S., Kinameri, E., Yoshihara, S., Nishiyori, A., to olfactory dysfunction and subciliary cyclic nucleotide-gated channel trapping. Kimura, M., Fujita, H., Touhara, K., Reed, R. R. et al. (2013). Goofy coordinates J. Biol. Chem. 281, 35156-35166. the acuity of olfactory signaling. J. Neurosci. 33, 12987-12996. Mukherjee, S., Thomas, M., Dadgar, N., Lieberman, A. P. and Iñiguez-Lluhı,́ Kee, H. L. and Verhey, K. J. (2013). Molecular connections between nuclear and J. A. (2009). Small ubiquitin-like modifier (SUMO) modification of the androgen ciliary import processes. Cilia 2, 11. receptor attenuates polyglutamine-mediated aggregation. J. Biol. Chem. 284, Kee, H. L., Dishinger, J. F., Blasius, T. L., Liu, C. J., Margolis, B. and Verhey, 21296-21306. K. J. (2012). A size-exclusion permeability barrier and nucleoporins characterize Pichler, A. and Melchior, F. (2002). Ubiquitin-related modifier SUMO1 and a ciliary pore complex that regulates transport into cilia. Nat. Cell Biol. 14, 431-437. nucleocytoplasmic transport. Traffic 3, 381-387. Klein, U. R. and Nigg, E. A. (2009). SUMO-dependent regulation of centrin-2. Pichler, A., Gast, A., Seeler, J. S., Dejean, A. and Melchior, F. (2002). The J. Cell Sci. 122, 3312-3321. nucleoporin RanBP2 has SUMO1 E3 ligase activity. Cell 108, 109-120. Kulaga, H. M., Leitch, C. C., Eichers, E. R., Badano, J. L., Lesemann, A., Plant, L. D., Dementieva, I. S., Kollewe, A., Olikara, S., Marks, J. D. and Hoskins, B. E., Lupski, J. R., Beales, P. L., Reed, R. R. and Katsanis, N. Goldstein, S. A. (2010). One SUMO is sufficient to silence the dimeric potassium (2004). Loss of BBS proteins causes anosmia in humans and defects in olfactory channel K2P1. Proc. Natl. Acad. Sci. USA 107, 10743-10748. cilia structure and function in the mouse. Nat. Genet. 36, 994-998. Plant, L. D., Dowdell, E. J., Dementieva, I. S., Marks, J. D. and Goldstein, S. A. Li, Y., Zhang, Q., Wei, Q., Zhang, Y., Ling, K. and Hu, J. (2012). SUMOylation of (2011). SUMO modification of cell surface Kv2.1 potassium channels regulates the small GTPase ARL-13 promotes ciliary targeting of sensory receptors. J. Cell the activity of rat hippocampal neurons. J. Gen. Physiol. 137, 441-454. Biol. 199, 589-598. Qi, Y., Wang, J., Bomben, V. C., Li, D. P., Chen, S. R., Sun, H., Xi, Y., Reed, J. G., Lin, Y. C., Niewiadomski, P., Lin, B., Nakamura, H., Phua, S. C., Jiao, J., Cheng, J., Pan, H. L. et al. (2014). Hyper-SUMOylation of the Kv7 potassium Levchenko, A., Inoue, T., Rohatgi, R. and Inoue, T. (2013). Chemically inducible channel diminishes the M-current leading to seizures and sudden death. Neuron diffusion trap at cilia reveals molecular sieve-like barrier. Nat. Chem. Biol. 9, 83, 1159-1171. 437-443. Ren, J., Gao, X., Jin, C., Zhu, M., Wang, X., Shaw, A., Wen, L., Yao, X. and Xue, Y. Luo, J., Ashikaga, E., Rubin, P. P., Heimann, M. J., Hildick, K. L., Bishop, P., (2009). Systematic study of protein sumoylation: development of a site-specific Girach, F., Josa-Prado, F., Tang, L. T., Carmichael, R. E. et al. (2013). Receptor predictor of SUMOsp 2.0. Proteomics 9, 3409-3412. trafficking and the regulation of synaptic plasticity by SUMO. Neuromolecular Schwarzenbacher, K., Fleischer, J. and Breer, H. (2005). Formation and Med. 15, 692-706. maturation of olfactory cilia monitored by odorant receptor-specific antibodies. Maiuri, T., Woloshansky, T., Xia, J. and Truant, R. (2013). The huntingtin N17 Histochem. Cell Biol. 123, 419-428. domain is a multifunctional CRM1 and Ran-dependent nuclear and cilial export Santos, N. and Reiter, J. F. (2014). A central region of Gli2 regulates its localization signal. Hum. Mol. Genet. 22, 1383-1394. to the primary cilium and transcriptional activity. J. Cell Sci. 127, 1500-1510. Majumdar, A., Petrescu, A. D., Xiong, Y. and Noy, N. (2011). Nuclear translocation Sharma, N., Berbari, N. F. and Yoder, B. K. (2008). Ciliary dysfunction in of cellular retinoic acid-binding protein II is regulated by retinoic acid-controlled developmental abnormalities and diseases. Curr. Top. Dev. Biol. 85, 371-427. SUMOylation. J. Biol. Chem. 286, 42749-42757. Stephan, A. B., Shum, E. Y., Hirsh, S., Cygnar, K. D., Reisert, J. and Zhao, H. Martin, S., Nishimune, A., Mellor, J. R. and Henley, J. M. (2007). SUMOylation (2009). ANO2 is the cilial calcium-activated chloride channel that may mediate regulates kainate-receptor-mediated synaptic transmission. Nature 447, 321-325. olfactory amplification. Proc. Natl. Acad. Sci. USA 106, 11776-11781. Masyuk, A. I., Gradilone, S. A., Banales, J. M., Huang, B. Q., Masyuk, T. V., Lee, Sun, X., Haley, J., Bulgakov, O. V., Cai, X., McGinnis, J. and Li, T. (2012). Tubby S. O., Splinter, P. L., Stroope, A. J. and Larusso, N. F. (2008). Cholangiocyte is required for trafficking G protein-coupled receptors to neuronal cilia. Cilia 1, 21. primary cilia are chemosensory organelles that detect biliary nucleotides via PubMed P2Y12 purinergic receptors. Am. J. Physiol. 295, pG725-G734. Wang, Z., Phan, T. and Storm, D. R. (2011). The type 3 adenylyl cyclase is required Matunis, M. J., Coutavas, E. and Blobel, G. (1996). A novel ubiquitin-like for novel object learning and extinction of contextual memory: role of cAMP modification modulates the partitioning of the Ran-GTPase-activating protein signaling in primary cilia. J. Neurosci. 31, 5557-5561. RanGAP1 between the cytosol and the nuclear pore complex. J. Cell Biol. 135, Wang, Y., Alam, T., Hill-Harfe, K., Lopez, A. J., Leung, C. K., Iribarne, D., 1457-1470. Bruggeman, B., Miyamoto, M. M., Harfe, B. D. and Choe, K. P. (2013). Maurya, D. K. and Menini, A. (2014). Developmental expression of the calcium- Phylogenetic, expression, and functional analyses of anoctamin homologs in activated chloride channels TMEM16A and TMEM16B in the mouse olfactory . Am. J. Physiol. 305, R1376-R1389. epithelium. Dev. Neurobiol. 74, 657-675. Weger, S., Hammer, E. and Engstler, M. (2003). The DNA topoisomerase I binding McEwen, D. P., Koenekoop, R. K., Khanna, H., Jenkins, P. M., Lopez, I., protein topors as a novel cellular target for SUMO-1 modification: characterization Swaroop, A. and Martens, J. R. (2007). Hypomorphic CEP290/NPHP6 of domains necessary for subcellular localization and sumolation. Exp. Cell Res. mutations result in anosmia caused by the selective loss of G proteins in cilia of 290, 13-27. olfactory sensory neurons. Proc. Natl. Acad. Sci. USA 104, 15917-15922. Williams, C. L., McIntyre, J. C., Norris, S. R., Jenkins, P. M., Zhang, L., Pei, Q., McIntyre, J. C., Davis, E. E., Joiner, A., Williams, C. L., Tsai, I. C., Jenkins, P. M., Verhey, K. and Martens, J. R. (2014). Direct evidence for BBSome-associated McEwen, D. P., Zhang, L., Escobado, J., Thomas, S. ; NISC Comparative, intraflagellar transport reveals distinct properties of native mammalian cilia. Nat. Sequencing, Program et al. (2012). Gene therapy rescues cilia defects and Commun. 5, 5813. restores olfactory function in a mammalian ciliopathy model. Nat. Med. 18, Wong, S. T., Trinh, K., Hacker, B., Chan, G. C., Lowe, G., Gaggar, A., Xia, Z., 1423-1428. Gold, G. H. and Storm, D. R. (2000). Disruption of the type III adenylyl cyclase Menco, B. P., Cunningham, A. M., Qasba, P., Levy, N. and Reed, R. R. (1997). gene leads to peripheral and behavioral anosmia in transgenic mice. Neuron 27, Putative odour receptors localize in cilia of olfactory receptor cells in rat and 487-497. mouse: a freeze-substitution ultrastructural study. J. Neurocytol. 26, 691-706. Zhang, H., Saitoh, H. and Matunis, M. J. (2002). Enzymes of the SUMO Michalakis, S., Reisert, J., Geiger, H., Wetzel, C., Zong, X., Bradley, J., Spehr, modification pathway localize to filaments of the nuclear pore complex. Mol. Cell. M., Hüttl, S., Gerstner, A., Pfeifer, A. et al. (2006). Loss of CNGB1 protein leads Biol. 22, 6498-6508. Journal of Cell Science

1945