Multisite interaction with Sufu regulates Ci/Gli activity through distinct mechanisms in Hh signal transduction

Yuhong Hana,1, Qing Shia,1, and Jin Jianga,b,2

Departments of aDevelopmental Biology and bPharmacology, University of Texas Southwestern Medical Center at Dallas, Dallas, TX 75390

Edited by Gary Struhl, Columbia University College of Physicians and Surgeons, New York, NY, and approved April 16, 2015 (received for review November 11, 2014) The tumor suppressor Suppressor of fused (Sufu) plays a respectively. Furthermore, we provide evidence that binding conserved role in the Hedgehog (Hh) signaling pathway by inhibiting of Sufu to Ci impedes the recruitment of the transcriptional Cubitus interruptus (Ci)/Glioma-associated oncogene homolog (Gli) coactivator CBP. factors, but the molecular mechanism by which Sufu inhibits Ci/Gli activity remains poorly understood. Here we show that Results Sufu can bind Ci/Gli through a C-terminal Sufu-interacting site (SIC) in Sufu Can Inhibit a Full-Length Ci Lacking the N-Terminal Sufu-Binding addition to a previously identified N-terminal site (SIN), and that both Site. Previous studies indicated that Ci/Gli binds Sufu through its SIC and SIN are required for optimal inhibition of Ci/Gli by Sufu. N-terminal domain containing an SYGH core motif (Fig. 1A) We show that Sufu can sequester Ci/Gli in the cytoplasm through (11, 12, 23, 24), which we named SIN (Sufu-interacting site in the binding to SIN while inhibiting Ci/Gli activity in the nucleus depend- N-terminal region). We were surprised to observe that the tran- Δ ing on SIC. We also find that binding of Sufu to SIC and the middle scriptional activity of a Ci variant (Ci-PKA N) lacking SIN was still region of Ci can impede recruitment of the transcriptional coactiva- inhibited by Sufu (Fig. S1). Of note, Ci-PKA has three PKA sites tor CBP by masking its binding site in the C-terminal region of Ci. mutated to Ala and is no longer processed into a truncated re- Indeed, moving the CBP-binding site to an “exposed” location can pressor (CiR) (25). By using Ci-PKA as a backbone for structure– render Ci resistant to Sufu-mediated inhibition in the nucleus. Hence, function study, we could focus on the regulation of the activator our study identifies a previously unidentified and conserved Sufu- form of Ci (CiA) by Sufu. When SIN was deleted in CiGA1, in binding motif in the C-terminal region of Ci/Gli and provides mech- which the Ci sequence C-terminal to its Zn-finger DNA-binding anistic insight into how Sufu inhibits Ci/Gli activity in the nucleus. domain was replaced by the Gal4 activation domain (GA), the resulting CiGA1ΔN was less inhibited by Sufu (Fig. S1), sug- Hedgehog | Sufu | Ci | Gli | CBP gesting that Sufu can inhibit Ci through a region C-terminal to its Zn-finger DNA-binding domain. he Hedgehog (Hh) signaling pathway controls embryogenesis Tand adult tissue homeostasis by regulating the Cubitus Sufu Interacts with a Conserved C-Terminal Site in Ci/Gli. To deter- interruptus (Ci)/Glioma-associated oncogene homolog (Gli) mine whether Sufu could interact with Ci through a domain(s) family of zinc-finger transcription factors (1, 2). Initially identi- other than SIN, we divided Ci into three fragments—Ci1–439 fied as a suppressor of the segmentation defects caused by the (amino acids 1–439), Ci440–1160 (amino acids 440–1160), and loss of the serine/threonine kinase Fused (Fu) in (3), Ci1161–1397 (amino acids 1161–1397)—and examined their inter- Suppressor of fused (Sufu) plays a conserved negative role in Hh action with Sufu by coimmunoprecipitation (CoIP) assay. When signal transduction by inhibiting the Ci/Gli transcription factors coexpressed with a Flag-tagged Sufu (Fg-Sufu) in S2 cells, all (4–6). Moreover, mutations in Sufu predispose to me-

dulloblastoma and meningioma (7, 8). Much of the attention in Significance BIOLOGY

the past has been given to the role of Sufu in the cytoplasmic DEVELOPMENTAL – sequestration of Ci/Gli (5, 9 14). In addition, Sufu is also re- Hedgehog (Hh) signaling controls embryonic development and quired for the production of the form of Gli in mam- adult tissue homeostasis by regulating the Cubitus inter- mals (15–17), a function carried out by the kinesin-like protein Drosophila ruptus (Ci)/Glioma-associated oncogene homolog (Gli) fam- Costal2 (Cos2) in (18). However, several studies have ily of transcription factors. Abnormal Hh pathway activity suggested that Sufu may also function in the nucleus to inhibit causes congenital disease and cancer. As a conserved negative Drosophila cos2 the activator form of Ci/Gli. For example, in regulator of the Hh signaling pathway, the tumor suppressor mutant wing discs, full-length Ci accumulated in the nucleus in a protein Suppressor of fused (Sufu) binds and inhibits Ci/Gli, but latent form inhibited by Sufu (18, 19). In cultured mammalian how Sufu contacts Ci/Gli and how Sufu–Ci/Gli interaction in- cells, overexpression of a truncated Sufu could inhibit Gli activity hibits Hh signaling activity remain poorly understood. Here we without sequestering it in the cytoplasm (20). Sufu can interact identified a conserved Sufu-binding site in the C-terminal with several nuclear , including the Drosophila myelodys- region of Ci/Gli. Further characterization of this Sufu-binding plasia/myeloid leukemia factor and transcriptional corepressor site provided insight into how Sufu blocks Ci/Gli activation in the complex Sin3–SAP18 (21, 22); however, a role for these nuclear nucleus. Understanding the mechanism by which Sufu regulates factors in the regulation of Ci/Gli activity has not been demon- Ci/Gli activity is important for developing therapeutic treat- strated by a loss-of-function study (17). ment of cancers caused by abnormal Hh pathway activation. In this study, we observed that Sufu could still inhibit a full- length Ci lacking the previously identified N-terminal Sufu-binding Author contributions: Y.H., Q.S., and J.J. designed research; Y.H. and Q.S. performed motif. Following up this unexpected observation, we identified a research; Y.H., Q.S., and J.J. analyzed data; and Q.S. and J.J. wrote the paper. previously unidentified and conserved Sufu-binding motif located The authors declare no conflict of interest. at the C terminus of Ci/Gli. We show that both the N- and C-ter- This article is a PNAS Direct Submission. minal Sufu-interacting sites are required for optimal binding of 1Y.H. and Q.S. contributed equally to this work. Ci/Gli to Sufu as well as for effective inhibition of Ci/Gli by Sufu. 2To whom correspondence should be addressed. Email: [email protected]. We find that the N- and C-terminal sites can mediate cyto- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. plasmic retention and nuclear inhibition of Ci/Gli by Sufu, 1073/pnas.1421628112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1421628112 PNAS | May 19, 2015 | vol. 112 | no. 20 | 6383–6388 Downloaded by guest on September 25, 2021 Fig. 1. Sufu binds Ci through both the N- and C-terminal domains. (A) A diagram of Ci domain structure and sequence alignment of SIN (blue bar) and SIC (green bar) within Ci and human (h) Gli proteins. ZF and CBP indicate the positions of the zinc-finger DNA-binding domain and CBP-binding domain, re- spectively. The SYGH motif is highlighted by a dashed box, and conserved residues in SIN and SIC are colored in blue and green, respectively. (B) Western blots (Left) and quantification (Right) of coimmunoprecipitation experiments from lysates of S2 cells coexpressing Fg-Sufu and various Myc-tagged Ci fragments (asterisks). The arrow indicates IgG. Data are means ± SD from two independent experiments. IB, immunoblotting; IP, immunoprecipitation. (C) Western blots of GST pull-down experiments. One or 5 μg of GST or GST fusion proteins was used as bait and coincubated with equal amounts of lysates from S2 cells expressing Fg-Sufu. (D) Western blots (Top) and quantification (Bottom) of coimmunoprecipitation experiments from lysates of S2 cells coexpressing Fg-Sufu and the indicated Myc-tagged Ci proteins. TCL, total cell lysates. Data are means ± SD from two independent experiments. (E) Western blots of a competition

experiment. Equal amounts of immunopurified Fg-Sufu were incubated with cell extracts from S2 cells expressing a fixed amount of Myc-Ci1161–1397 and increasing amounts of Myc-Ci1–439. The arrow indicates IgG. (F) Western blots of coimmunoprecipitates from lysates of S2 cells coexpressing Fg-Sufu or D154R Fg-Sufu with Myc-Ci1–439 or Myc-Ci1161–1397.

three Ci fragments coimmunoprecipitated Fg-Sufu, with Ci1–439 suggest that Sufu may simultaneously interact with both SIN and exhibiting 10-fold higher affinity than Ci440–1160 or Ci1161–1397 SIC, and that this multisite interaction greatly increases the overall (Fig. 1B). Sequence alignment revealed a C-terminally conserved binding affinity. sequence motif present in Ci (amino acids 1370–1397), Gli2, and Gli3 (Fig. 1A). To determine whether this Both SIN and SIC Contribute to Sufu-Mediated Inhibition of Ci. Having mediates the interaction between Ci and Sufu, we deleted it from established that Sufu can bind Ci through both SIN and SIC, we Ci1161–1397 to generate Ci1160–1370, and found that Ci1160–1370 next explored the functional significance of these Sufu-binding failed to bind Sufu (Fig. 1B). GST pull-down experiments domains in mediating Ci inhibition. We first compared the activity -PKA -PKAΔN -PKAΔC -PKAΔNΔC revealed that GST-Ci1370–1397 pulled down Fg-Sufu derived from of Ci ,Ci ,Ci ,andCi in the absence or S2 cell extracts but 10-fold less effectively compared with GST- presence of coexpressed Fg-Sufu using the ptc-luc reporter assay in -PKA – C – S2 cells. As shown in Fig. 2B,Ci was suppressed by Sufu more Ci1 439 (Fig. 1 ), suggesting that Ci1370 1397 contains a low- Δ Δ Δ Δ affinity Sufu-binding site, which we name SIC (Sufu-interacting site effectively than Ci-PKA N and Ci-PKA C, whereas Ci-PKA N C in the C terminus). exhibited little if any suppression by Sufu, suggesting that both SIN and SIC contribute to the Sufu-mediated inhibition of Ci. Simultaneous Binding to SIN and SIC Promotes Strong Sufu–Ci To determine the relative contribution of SIN and SIC to Association. We next determined how SIN and SIC contribute to Sufu-mediated inhibition of Ci in vivo, UAS transgenes expressing the overall binding of Sufu to Ci. Accordingly, we generated Myc- individual Ci variants were introduced into flies at the same geno- Δ tagged Ci-PKA and its variants lacking either SIN (Ci-PKA N), SIC mic using the phiC31 integration system to ensure similar Δ Δ Δ (Ci-PKA C), or both (Ci-PKA N C) (Fig. 2A), and compared their levels of transcription of the transgenes (28). Wing discs expressing binding affinity toward Sufu in S2 cells using the CoIP assay. As these transgenes using MS1096, a wing-specific Gal4 driver, in the shown in Fig. 1D, deletion of either SIN or SIC resulted in ap- absence or presence of a UAS-Sufu transgene were immunostained proximately fivefold reduction in Sufu binding, whereas their with Ci and Patched (Ptc) antibodies to monitor the levels of full- combined deletion nearly abolished Sufu binding. In a competition length Ci (derived from both transgenic and endogenous expres- assay, we found that increasing the amount of Ci1–439 did not affect sion) and Hh pathway activity, respectively. Quantification of -PKA the association between Sufu and Ci1161–1397; instead, Sufu coim- full-length Ci levels in control and Ci –expressing wing discs munoprecipitated both Ci fragments (Fig. 1E), suggesting that SIN indicated that exogenously derived Ci reached levels four- to and SIC may bind different regions of Sufu instead of competing fivefold higher than the endogenous levels in anterior (A)- for the same binding pocket. Consistent with this notion, a Sufu compartment cells distant from the anterior–posterior (A/P) variant (SufuD154R), which contains a point mutation at D154 boundary after subtracting the background signal, whereas in implicated in contacting the SYGH motif (26, 27), exhibited di- A-compartment cells near the A/P boundary, Ci-PKA–expressing minished binding to Ci1–440 but interacted with Ci1161–1397 similar wing discs contained full-length Ci at levels approximately two- to the wild-type Sufu (Fig. 1F). Taken together, these observations fold higher than the control discs (Fig. S2 A–C). Real-time

6384 | www.pnas.org/cgi/doi/10.1073/pnas.1421628112 Han et al. Downloaded by guest on September 25, 2021 truncated Ci-PKA induced ectopic expression of ptc along the A/P axis (Fig. 2 D′–G′). Coexpression of Sufu completely suppressed the ectopic ptc expression induced by Ci-PKA in both A- and P-compartment cells (Fig. 2H′ compared with Fig. 2D′). Co- expression of Sufu completely suppressed the ectopic ptc expression Δ Δ induced by Ci-PKA N or Ci-PKA C in A-compartment cells but only Δ partially suppressed ectopic ptc expression induced by Ci-PKA N or Δ Ci-PKA C in P-compartment cells (Fig. 2 I′ and J′ compared with Fig. 2 E′ and F′). By contrast, coexpression of Sufu did not Δ Δ block the ectopic ptc expression induced by Ci-PKA N C in ei- ther A- or P-compartment cells (Fig. 2K′ compared with Fig. 2G′). Hence, both SIN and SIC can contribute to Sufu-mediated inhibition of Ci. Δ Δ We noted that Ci-PKA N C induced ectopic ptc expression at lower levels compared with other Ci transgenes (Fig. 2G′ com- pared with Fig. 2 D–F′), likely due to its instability, because wing Δ Δ discs expressing Ci-PKA N C exhibited weaker Ci staining than those expressing other Ci transgenes (Fig. 2G compared with Fig. 2 D–F). Previous studies indicated that Sufu protects Ci/Gli from Δ Δ HIB/SPOP-mediated degradation (17, 29). Because Ci-PKA N C did not bind Sufu, it could be degraded by HIB more rapidly than other Ci variants. In support of this notion, we found that in- Δ Δ activation of HIB by RNAi stabilized Ci-PKA N C in wing discs, leading to elevated ptc expression at levels similar to those induced by Ci-PKA (Fig. S4).

SIN and SIC Mediate Sufu Inhibition Through Distinct Mechanisms. We next determined how SIN and SIC mediate Ci inhibition by Sufu. Consistent with previous findings that blocking nuclear export promotes Ci nuclear accumulation (11, 30), when expressed alone Δ Δ in S2 cells, Myc-Ci-PKA, Myc-Ci-PKA N, and Myc-Ci-PKA C were accumulated mainly in the nucleus after the transfected cells were treated with the nuclear export inhibitor LMB (Fig. 3 A–C). Δ Coexpression of Fg-Sufu retained Myc-Ci-PKA and Myc-Ci-PKA C Δ but not Myc-Ci-PKA N in the cytoplasm (Fig. 3 E–G′). Furthermore, Fg-SufuD154R, which failed to bind SIN but exhibited normal binding to SIC (Fig. 1F), failed to retain Myc-Ci-PKA in the cytoplasm (Fig. 3 H and H′). Quantification of nuclear–cyto- plasmic distribution of the various Ci constructs is shown in Δ Fig. 2. Both the N- and C-terminal Sufu-binding sites can contribute to Ci Fig. S5.Whenexpressedinwingdiscs,Myc-Ci-PKA N exhibited inhibition. (A) A diagram of full-length (FL), ΔN, ΔC, and ΔNΔCvariantsof more nuclear localization than Myc-Ci-PKA in A-compartment cells Ci-PKA. SIN and SIC are indicated by blue and green bars, respectively. (B) ptc- -PKAΔC -PKA Δ Δ Δ Δ away from the A/P boundary, whereas Myc-Ci behaved luc reporter assays in S2 cells transfected with Ci FL, N, C, and N C -PKA BIOLOGY alone or with Sufu. Data are means ± SD from three independent experi- similarly to Myc-Ci (Fig. S6). Taken together, these ob- – DEVELOPMENTAL ments. (C–K′) Late third-instar wing discs of wild type (C and C′) or expressing servations suggest that Sufu SIN interaction is essential for the indicated UAS-Ci transgenes either alone (D–G′) or together with a UAS- Sufu-mediated cytoplasmic retention of Ci. -PKA Sufu transgene (H–K′) under the control of the MS1096 Gal4 driver were Despite failing to sequester Myc-Ci in the cytoplasm, D154R immunostained with anti-Ci (red) and anti-Ptc (green) antibodies to monitor Fg-Sufu still inhibited its transcriptional activity (Fig. 3J), the levels of full-length Ci derived from both transgenic and endogenous suggesting that SufuD154R could inhibit Ci activity in the nucleus. expression, and Hh pathway activity, respectively. Arrowheads and arrows To confirm this, we generated a constitutively nuclear form of Ci, indicate A and P compartments, respectively (H′–K′). HA-Nuc-Ci-PKA, in which an SV40 nuclear localization signal (NLS) was inserted at the N terminus of Ci (19) and a major nuclear export signal (NES) was mutated (31). When expressed quantitative PCR revealed that Sufu mRNA levels in Sufu- in S2 cells, HA-Nuc-Ci-PKA was accumulated largely in the nu- overexpressing wing discs were approximately threefold higher cleus even in the presence of Fg-Sufu (Fig. 3 D, I, and I′ and Fig. than in control discs (Fig. S2D). CoIP experiments using wing -PKA -PKAΔN -PKAΔC S5); however, HA-Nuc-Ci activity was still sensitive to Fg- disc extracts indicated that both Ci and Ci bound Sufu–mediated inhibition (Fig. 3K). Whereas deleting SIN (ΔN) approximately fivefold less endogenous Sufu compared with D154R -PKA -PKA -PKAΔNΔC did not affect Sufu -mediated inhibition of Ci or Sufu- Ci , whereas Ci exhibited no detectable binding to mediated inhibition of Nuc-Ci-PKA, deletion of either SIC (ΔC) or less endogenous Sufu (Fig. S3). both SIN and SIC (ΔNΔC) abolished these inhibitions (Fig. 3 J In control wing discs, Hh is expressed in the posterior (P) and K), suggesting that Sufu can inhibit Ci activity in the nucleus compartment whereas Ci is expressed in the A compartment. Hh depending on SIC. moves to the A compartment to stabilize Ci and activate target such as ptc in A-compartment cells near the A/P boundary Interaction Between Sufu and the C-Terminal Half of Ci Interferes (Fig. 2 C and C′). When Ci is ectopically expressed in P-com- with CBP Recruitment. We next determined the mechanism by partment cells using the MS1096 Gal4 driver, it can ectopically which Sufu inhibits Ci in the nucleus. It has been shown that activate downstream Hh target genes, and this ectopic activity Ci activates its target genes by binding to the Drosophila CBP is subject to suppression by Sufu, providing an additional place (dCBP) through a C-terminal binding domain between amino to assay Sufu–Ci regulatory interactions. Both full-length and acids 1020 and 1160 (32). Because of the proximity between the

Han et al. PNAS | May 19, 2015 | vol. 112 | no. 20 | 6385 Downloaded by guest on September 25, 2021 Ci621–1020, but not to either Ci440–1370 or Ci1020–1397 (Fig. 4B), suggesting that simultaneous binding of Sufu to SIC and Ci621–1020 is likely to be required to impede dCBP recruitment. If Sufu inhibits Ci in the nucleus by impeding dCBP recruitment, we reasoned that moving the dCBP-binding domain (dCBP-BD) to an “exposed” location in Ci should overcome the inhibition by Sufu in the nucleus. Therefore, we fused a dCBP-BD to the C terminus of either Ci-PKA or Nuc-Ci-PKA after addition of a flexible linker protein, MBP, to separate SIC from the exogenously added dCBP- BD (Fig. 4C). In the meantime, the endogenous dCBP-BD was Δ deleted from these fusion proteins to generate Ci CBP-MBP-CBP Δ or Nuc-Ci- CBP-MBP-CBP, respectively (Fig. 4C). We also gener- Δ ated several Ci constructs, including Ci-MBP, Ci CBP-MBP, Nuc- Δ Ci-MBP, and Nuc-Ci CBP-MBP as controls (Fig. 4C). As shown in Fig. 4, addition of MBP to the C terminus of either Ci-PKA or Nuc- Ci-PKA did not affect their activity as well as Sufu- or SufuD154R- mediated inhibition (Fig. 4C, compare groups 2 and 6 with 1 and 5). Δ Δ Both Ci CBP-MBP-CBP and Nuc-Ci CBP-MBP-CBP were as active as Ci-PKA//Ci-PKA-MBP and Nuc-Ci-PKA/Nuc-Ci-PKA-MBP whereas Δ Δ neither Ci CBP-MBP nor Nuc-Ci CBP-MBP was active, suggesting Δ Δ that the activity of Ci CBP-MBP-CBP and Nuc-Ci CBP-MBP-CBP Δ depends on the exogenously added dCBP-BD. Importantly, Ci CBP- Δ MBP-CBP and Nuc-Ci CBP-MBP-CBP were no longer suppressed by SufuD154R (Fig. 4C, groups 4 and 8). Furthermore, although Sufu Δ still inhibited Ci CBP-MBP-CBP, likely by sequestering it in the Δ cytoplasm (Fig. 4C, group 4), Sufu failed to inhibit Nuc-Ci CBP- MBP-CBP (Fig. 4C, group 8). Taken together, these results further support the notion that Sufu can inhibit Ci in the nucleus by im- peding dCBP recruitment.

Fig. 3. SIN and SIC can regulate Ci by distinct mechanisms. (A–I′) Represen- A Conserved Role of SIC in Sufu-Mediated Inhibition of Gli2. In the Δ tative confocal images of S2 cells transfected with Myc-tagged Ci-PKA,Ci-PKA N, vertebrate Hh pathway, Gli2 and Gli3 are the primary tran- -PKAΔC -PKA or Ci or HA-tagged Nuc-Ci either alone or together with Flag-tagged scription factors. We noticed that the sequence of SIC is highly Sufu or SufuD154R and immunostained with antibodies against Myc or HA (red) and Flag (green) and phalloidin for cell membrane (blue). Of note, transfected cells were treated with 20 ng/mL LMB for 3 h before immunostaining. (J and K) ptc-luc reporter assays in S2 cells transfected with the indicated Ci and Sufu constructs. Data are means ± SD from three independent experiments.

dCBP-binding domain and SIC, we speculated that binding of Sufu to SIC may impede the recruitment of dCBP. Therefore, we carried out CoIP experiments using nuclear extracts prepared from S2 cells transfected with HA-Nuc-Ci-PKA and Myc-tagged dCBP (Myc-dCBP) with or without cotransfection of Fg-Sufu. In the absence of Fg-Sufu, Myc-dCBP formed a complex with HA-Nuc-Ci-PKA; however, the association between Myc-dCBP and HA-Nuc-Ci-PKA was blocked by Fg-Sufu (Fig. 4A, lanes 1 Δ and 2). Deletion of SIN (Nuc-Ci-PKA N) only slightly affected this blockage (Fig. 4A, lanes 3 and 4). By contrast, deletion of Δ Δ Δ either SIC (Nuc-Ci-PKA C) or both SIN and SIC (Nuc-Ci-PKA N C) abolished the Sufu-mediated inhibition of dCBP–Ci association (Fig. 4A, lanes 5–8), suggesting that binding of Sufu to SIC may be essential for preventing dCBP–Ci association. Because SIC is an intrinsically weak Sufu-binding site (Fig. 1C), we speculated that SIC might cooperate with a Sufu-binding site located in the middle region of Ci to bind Sufu when the function of SIN was compromised. Indeed, Ci440–1397, which contains both SIC and the middle Sufu-binding site (Fig. 1B), exhibited approximately fivefold higher binding affinity to Sufu than Ci440–1160 and Ci1161–1397, both of which contain only one Sufu-binding site (Fig. S7A). Further mapping revealed that Sufu interacted with a Ci fragment (Ci621–1020) between the zinc-finger DNA-binding domain and the CBP-binding domain (Fig. S7B), Fig. 4. Interaction of Sufu with the C-terminal half of Ci can impede CBP binding. (A and B) Western blots of coimmunoprecipitation experiments from consistent with the finding of a previous study showing that a Ci – lysates of S2 cells expressing the indicated constructs. Asterisks in B indicate fragment containing amino acids 832 1187 interacted with Sufu HA-tagged Ci fragments. (C) ptc-luc reporter assays (Left) in S2 cells transfected in a yeast two-hybrid assay (33). Importantly, Fg-Sufu inhibited with the indicated Ci constructs (diagrams are shown; Right) without or with D154R the binding of dCBP to Ci440–1397, which contains both SIC and Sufu or Sufu . Data are means ± SD from three independent experiments.

6386 | www.pnas.org/cgi/doi/10.1073/pnas.1421628112 Han et al. Downloaded by guest on September 25, 2021 Intriguingly, Zhang et al. reported that mSufuD159R failed to inhibit Gli1 activity (26), which we confirmed in this study (Fig. S9). Although a previous study indicated that Sufu could interact with the C-terminal half of Gli1, the binding site was not mapped (34). Sequence alignment revealed that the C-terminal sequence of Gli1 diverges significantly from those of Gli2/3 and Ci (Fig. S9), suggesting that Gli1 may contain a less potent SIC, which could explain why mSufuD159R failed to inhibit Gli1. Discussion As a conserved negative regulator of intracellular Hh signaling, Sufu binds and inhibits Ci/Gli. A conserved Sufu-binding site with an SYGH core motif (SIN) was identified in the N-terminal region of Ci/Gli, and structural information has recently been obtained on how Sufu binds this conserved site (26, 27); however, how Sufu interacts with full-length Ci/Gli remains a mystery, because the crystal structures of the Sufu–Gli complexes only contain a small peptide flanking the SYGH motif. Here we identified a conserved Sufu-binding site (SIC) at the C terminus of Ci/Gli. We provided evidence that Sufu may contact SIN and SIC simultaneously, which could be essential for effective Sufu– Ci/Gli association under physiological conditions. In addition, both SIN and SIC appear to be required for optimal inhibition of Ci/Gli but may regulate Ci/Gli through distinct mechanisms, with SIN primarily mediating cytoplasmic retention and SIC con- Fig. 5. A conserved role of SIC in mediating Sufu binding and nuclear in- tributing to nuclear inhibition of Ci/Gli. hibition of Gli2. (A) Western blots of coimmunoprecipitation experiment Previous studies have suggested that Sufu can impede Ci/Gli from lysates of NIH 3T3 cells expressing the indicated Gli2 and Sufu con- nuclear localization and inhibit Ci/Gli transcriptional activity in structs. (B) Gli-luc reporter assays in NIH 3T3 cells transfected with the in- – ± the nucleus (5, 9 12, 19, 20); however, how Sufu exerts this dual dicated Gli2 and Sufu constructs. Data are means SD from three regulation still remains poorly understood. Our recent study independent experiments. (C) Representative confocal images of NIH 3T3 revealed that binding of Sufu to SIN may inhibit the binding of cells transfected with an the indicated mouse Myc-Gli2 and Sufu constructs β and immunostained with an anti-Myc antibody (Left) and DNA dye Hoechst Kap 2 to a NLS of the PY family located in the N-terminal re- that labels the nucleus (Nuc) (Right). gion of Ci/Gli (13). Another recent study also provided evidence that binding of Sufu to Gli1 can preclude the binding of importin β1 (14). Taken together, these studies suggest that binding of conserved in both Gli2 and Gli3 (Fig. 1A). Because Gli2 con- Sufu to SIN may inhibit Ci/Gli nuclear import by masking its tributes mostly to Gli activator activity, we focused on Gli2 using NLSs. Here we provide evidence that binding of Sufu to SIC can the approaches similar to what we have applied to Ci. Accord- inhibit Ci transcriptional activity by impeding the recruitment of ingly, we generated Myc-tagged full-length Gli2 and its variants dCBP (Fig. 6). Indeed, deletion of SIC but not SIN abolished with either SIN (Gli2ΔN), SIC (Gli2ΔC), or both (Gli2ΔNΔC) Sufu-mediated inhibition of Ci in the nucleus (Fig. 3K). By deleted. These constructs were cotransfected with a Flag-tagged contrast, deletion of SIN but not SIC affected Sufu-mediated mouse Sufu (mSufu) into NIH 3T3 cells, followed by CoIP ex- cytoplasmic retention of Ci (Fig. 3 E–G and Figs. S5 and S6). BIOLOGY periments or Gli-luc reporter assays. Similar to Ci, deletion of We found that both SIN and SIC were required for optimal either SIN or SIC from Gli2 greatly reduced its binding to binding of Sufu to Ci, because deletion of either site in the full- DEVELOPMENTAL mSufu, and their combined deletion nearly abolished the Gli2– length Ci background resulted in approximately fivefold re- mSufu association (Fig. 5A). Furthermore, both SIN and SIC duction in Sufu binding affinity (Fig. 1 and Fig. S3). Therefore, contributed to mSufu-mediated inhibition of Gli2 activity, as we speculate that in anterior-compartment cells distant from the deletion of either SIN or SIC only partially alleviated whereas A/P boundary, where there is no Hh and the levels of full-length F F their combined deletion almost completely blocked the mSufu- Ci (Ci ) are low, Sufu may bind Ci by simultaneously contacting F mediated inhibition (Fig. 5B). Immunostaining of the transfected both SIN and SIC to prevent nuclear import of Ci . In A-com- cells with an antibody against Myc and a nuclear marker revealed partment cells 5–10 cells away from the A/P boundary, where low to intermediate levels of Hh block Ci processing to increase the that Myc-Gli2 was primarily localized in the nucleus when F F expressed alone but was sequestered in the cytoplasm when levels of Ci and promote nuclear translocation of Ci (11, 30), mSufu was cotransfected (Fig. 5C and Fig. S8). However, cyto- plasmic retention of Gli2 by mSufu was abolished by the deletion of SIN (ΔN + mSufu; Fig. 5C and Fig. S8) but was unaffected by the deletion of SIC (ΔC + mSufu; Fig. 5C and Fig. S8). Con- sistent with a previous study (26), mSufuD159R, which is the counterpart of Drosophila SufuD154R, exhibited reduced binding to Gli2 (Fig. 5A) and failed to sequester Gli2 in the cytoplasm (Fig. 5C and Fig. S8). Nevertheless, mSufuD159R still inhibited the activity of Gli2 (Fig. 5B). Furthermore, this inhibition was abolished by deletion of SIC (ΔCorΔNΔC) but was unaffected by deletion of SIN (ΔN). Taken together, these results suggest that Sufu can sequester Gli2 in the cytoplasm through binding to Fig. 6. Diagrams of Ci bound by Sufu or CBP. The N- and C-terminal Sufu- SIN but inhibit Gli2 activity in the nucleus through binding binding sites (SIN and SIC) and the CBP-binding domain are indicated by to SIC. blue, green, and red bars, respectively. See text for details.

Han et al. PNAS | May 19, 2015 | vol. 112 | no. 20 | 6387 Downloaded by guest on September 25, 2021 SIN–Sufu interaction may be compromised, likely due to Hh- through additional mechanisms such as recruiting a corepressor induced modification of Ci–Sufu. Under such conditions, SIC complex. Indeed, while our manuscript was under review, Lin may cooperate with the middle region of Ci (Ci621–1020) to bind et al. reported that Sufu inhibited Gli in the nucleus by recruiting Sufu to mask the dCBP-binding site (Fig. 6). This could explain p66β (38). We speculate that Sufu may inhibit Ci/Gli more ef- why low to medium levels of Hh can promote CiF nuclear fectively in the nucleus by simultaneously excluding coactivators translocation but fail to fully activate CiF. In A-compartment and recruiting corepressors. cells immediately abutting the A/P boundary, where peak levels of Hh convert CiF into CiA, both N- and C-terminal interactions Experimental Procedures F may be compromised, allowing dCBP to be recruited to Ci Standard procedures for Drosophila genetics and tissue-culture experiments (Fig. 6). were used. Drosophila stocks, transgenes, and DNA constructs and detailed In the mammalian Hh pathway, full-length Gli3 interacts with procedures for cell culture, transfection, cell fractionation, immunostaining, CBP and the MED12–Mediator complex through its C-terminal immunoprecipitation, Western blot analysis, GST pull-down assays, and transactivation domain (35, 36). Gli1 and Gli2 also contain ex- luciferase reporter assays are described in SI Experimental Procedures. tended transactivation domains in their C-terminal regions (37). ACKNOWLEDGMENTS. We thank Bing Wang for assistance and the De- Therefore, it is conceivable that binding of Sufu to Gli proteins velopmental Studies Hybridoma Bank for reagents. This work was supported may also preclude coactivator binding. However, our study does by grants from the NIH (GM061269, GM067045) and Welch Foundation not exclude the possibility that Sufu inhibits Ci/Gli in the nucleus (I-1603) (to J.J.).

1. Jiang J, Hui CC (2008) Hedgehog signaling in development and cancer. Dev Cell 15(6): 20. Barnfield PC, Zhang X, Thanabalasingham V, Yoshida M, Hui CC (2005) Negative 801–812. regulation of Gli1 and Gli2 activator function by Suppressor of fused through mul- 2. Briscoe J, Thérond PP (2013) The mechanisms of Hedgehog signalling and its roles in tiple mechanisms. Differentiation 73(8):397–405. development and disease. Nat Rev Mol Cell Biol 14(7):416–429. 21. Fouix S, et al. (2003) Over-expression of a novel nuclear interactor of Suppressor of fused, 3. Préat T (1992) Characterization of Suppressor of fused, a complete suppressor of the the Drosophila myelodysplasia/myeloid leukaemia factor, induces abnormal morphogen- – fused segment polarity of Drosophila melanogaster. Genetics 132(3):725–736. esis associated with increased apoptosis and DNA synthesis. Genes Cells 8(11):897 911. 4. Ohlmeyer JT, Kalderon D (1998) Hedgehog stimulates maturation of Cubitus inter- 22. Cheng SY, Bishop JM (2002) Suppressor of Fused represses Gli-mediated transcription ruptus into a labile transcriptional activator. Nature 396(6713):749–753. by recruiting the SAP18-mSin3 corepressor complex. Proc Natl Acad Sci USA 99(8): – 5. Ding Q, et al. (1999) Mouse Suppressor of fused is a negative regulator of Sonic 5442 5447. 23. Monnier V, Dussillol F, Alves G, Lamour-Isnard C, Plessis A (1998) Suppressor of fused hedgehog signaling and alters the subcellular distribution of Gli1. Curr Biol 9(19): links fused and Cubitus interruptus on the Hedgehog signalling pathway. Curr Biol 1119–1122. 8(10):583–586. 6. Svärd J, et al. (2006) Genetic elimination of Suppressor of fused reveals an essential 24. Méthot N, Basler K (2000) Suppressor of fused opposes Hedgehog signal transduction repressor function in the mammalian Hedgehog signaling pathway. Dev Cell 10(2): by impeding nuclear accumulation of the activator form of Cubitus interruptus. De- 187–197. velopment 127(18):4001–4010. 7. Taylor MD, et al. (2002) Mutations in SUFU predispose to . Nat Genet 25. Wang G, Wang B, Jiang J (1999) Protein kinase A antagonizes Hedgehog signaling by – 31(3):306 310. regulating both the activator and repressor forms of Cubitus interruptus. Genes Dev 8. Aavikko M, et al. (2012) Loss of SUFU function in familial multiple meningioma. Am J 13(21):2828–2837. – Hum Genet 91(3):520 526. 26. Zhang Y, et al. (2013) Structural insight into the mutual recognition and regulation 9. Méthot N, Basler K (1999) Hedgehog controls limb development by regulating the between Suppressor of Fused and Gli/Ci. Nat Commun 4:2608. activities of distinct transcriptional activator and repressor forms of Cubitus inter- 27. Cherry AL, et al. (2013) Structural basis of SUFU-GLI interaction in human Hedgehog ruptus. Cell 96(6):819–831. signalling regulation. Acta Crystallogr D Biol Crystallogr 69(Pt 12):2563–2579. 10. Kogerman P, et al. (1999) Mammalian Suppressor-of-Fused modulates nuclear- 28. Bischof J, Maeda RK, Hediger M, Karch F, Basler K (2007) An optimized transgenesis cytoplasmic shuttling of Gli-1. Nat Cell Biol 1(5):312–319. system for Drosophila using germ-line-specific phiC31 integrases. Proc Natl Acad Sci 11. Wang G, Amanai K, Wang B, Jiang J (2000) Interactions with Costal2 and Suppressor USA 104(9):3312–3317. of fused regulate nuclear translocation and activity of Cubitus interruptus. Genes Dev 29. Zhang Q, et al. (2006) A Hedgehog-induced BTB protein modulates Hedgehog 14(22):2893–2905. signaling by degrading Ci/Gli . Dev Cell 10(6):719–729. 12. Dunaeva M, Michelson P, Kogerman P, Toftgard R (2003) Characterization of the 30. Wang QT, Holmgren RA (2000) Nuclear import of Cubitus interruptus is regulated physical interaction of Gli proteins with SUFU proteins. J Biol Chem 278(7):5116–5122. by Hedgehog via a mechanism distinct from Ci stabilization and Ci activation. – 13. Shi Q, Han Y, Jiang J (2014) Suppressor of fused impedes Ci/Gli nuclear import by Development 127(14):3131 3139. opposing Trn/Kapβ2 in Hedgehog signaling. J Cell Sci 127(Pt 5):1092–1103. 31. Seong KH, et al. (2010) Inhibition of the nuclear import of Cubitus interruptus by 14. Szczepny A, et al. (2014) Overlapping binding sites for importin β1 and suppressor of Roadkill in the presence of strong Hedgehog signal. PLoS ONE 5(12):e15365. 32. Akimaru H, et al. (1997) Drosophila CBP is a co-activator of cubitus interruptus in fused (SuFu) on glioma-associated oncogene homologue 1 (Gli1) regulate its nuclear hedgehog signalling. Nature 386(6626):735–738. localization. Biochem J 461(3):469–476. 33. Croker JA, Ziegenhorn SL, Holmgren RA (2006) Regulation of the Drosophila tran- 15. Kise Y, Morinaka A, Teglund S, Miki H (2009) Sufu recruits GSK3beta for efficient scription factor, Cubitus interruptus, by two conserved domains. Dev Biol 291(2): processing of Gli3. Biochem Biophys Res Commun 387(3):569–574. 368–381. 16. Humke EW, Dorn KV, Milenkovic L, Scott MP, Rohatgi R (2010) The output of 34. Merchant M, et al. (2004) Suppressor of fused regulates Gli activity through a dual Hedgehog signaling is controlled by the dynamic association between Suppressor of binding mechanism. Mol Cell Biol 24(19):8627–8641. – Fused and the Gli proteins. Genes Dev 24(7):670 682. 35. Dai P, et al. (1999) -induced activation of the Gli1 promoter is me- 17. Chen MH, et al. (2009) Cilium-independent regulation of Gli protein function by Sufu diated by GLI3. J Biol Chem 274(12):8143–8152. – in Hedgehog signaling is evolutionarily conserved. Genes Dev 23(16):1910 1928. 36. Zhou H, Kim S, Ishii S, Boyer TG (2006) Mediator modulates Gli3-dependent Sonic 18. Zhang W, et al. (2005) Hedgehog-regulated Costal2-kinase complexes control phos- hedgehog signaling. Mol Cell Biol 26(23):8667–8682. phorylation and proteolytic processing of Cubitus interruptus. Dev Cell 8(2):267–278. 37. Hui CC, Angers S (2011) Gli proteins in development and disease. Annu Rev Cell Dev 19. Wang G, Jiang J (2004) Multiple Cos2/Ci interactions regulate Ci subcellular localiza- Biol 27:513–537. tion through microtubule dependent and independent mechanisms. Dev Biol 268(2): 38. Lin C, et al. (2014) Regulation of Sufu activity by p66β and Mycbp provides new insight 493–505. into vertebrate Hedgehog signaling. Genes Dev 28(22):2547–2563.

6388 | www.pnas.org/cgi/doi/10.1073/pnas.1421628112 Han et al. Downloaded by guest on September 25, 2021