HIB/SPOP inhibits Ci/Gli-mediated tumorigenesis by modulating the RNA Polymerase II components stabilities

Summary Hedgehog (Hh) signaling mediated by transcription factor Ci/Gli plays a vital role in embryonic development and adult tissue homeostasis in invertebrates and vertebrates, whose dysregulation leads to many human disorders, including cancer. However, till now, cofactors of Ci/Gli which can affect tumorigenesis are not well known. Here, through genetic screen, we find overexpression of active Ci alone is not sufficient to generate tumor-like eye phenotype in Drosophila, however, its overexpression combined with knockdown of hib causes a striking tumor-like big eye phenotype. Mechanistically, HIB/SPOP inhibits Ci/Gli-mediated tumorigenesis by modulating the RNA polymerase II (RNAPII) components Rpb3/Rpb7 stabilities in E3 ligase dependent manner. In addition, Ci/Gli can promote HIB/SPOP-mediated Rpb7/Rpb3 degradation. Taken together, our results indicate Ci/Gli needs to hook up with suitable RNAPII together to achieve the tumor-like eye phenotype and HIB/SPOP plays dual roles through controlling Ci/Gli and Rpb3/Rpb7 protein stabilities to temper Ci/Gli/RNAPII-mediated tumorigenesis.


Highlights
Overexpression of Ci alone is not sufficient to cause Drosophila eye tumorigenesis HIB inhibits Ci-mediated tumorigenesis in its E3 ligase dependent manner HIB degrades Rpb3/Rpb7 of RNAPII complex to inhibit Ci-mediated tumorigenesis SPOP and Gli2 play conserved roles to mediate hRpb7 degradation in mammalian cells
hib encodes Hh-induced BTB protein, which is also a Hh target gene in Drosophila wing discs. HIB together with Cul3 functions as an E3 ubiquitin ligase to modulate Ci degradation, forming a negative feedback loop

Genetic screen demonstrates that HIB inhibits ci-mediated tumor-like eye phenotype in Drosophila
Hh signaling-related tumorigenesis is mediated by transcription factor Ci/Gli, however, in some cases, only Ci/Gli is not sufficient to initiate tumorigenesis, the underlying mechanism and what cofactors of Ci/Gli affect tumorigenesis are not clear. To address this question, in Drosophila, overexpressing Ci À3P , which is an active form of Ci with three PKA kinase sites mutated, 48 we found that Ci alone was insufficient to cause eye tumorigenesis ( Figure 1A). Next, we overexpressed Ci À3P and simultaneously knocked down around 7000 conserved genes between fly and human to screen Ci/Gli cofactors by checking the eye phenotype. In this first round of genetic screen, we got two hits, hib, and warts. Since knockdown of warts alone showed the similar phenotype with co-expression of Ci and warts RNAi (data not shown), we later gave up this hit. For hib, we found its knockdown alone showed normal eye phenotype, however, its knockdown together with overexpression of Ci À3P caused very striking tumor-like eye phenotype ( Figures 1A-1C). In this condition, either knockdown of ci or co-expression of mammalian homologue of HIB, SPOP, could completely inhibit the tumor-like eye phenotype ( Figures 1D and 1E), indicating expression of Ci and inhibition of HIB are both necessary for making the tumor-like eye phenotype. Similarly, in Drosophila eye, we found that overexpressing Ci À3P in hib À/À mutant clones generated with FLP/FRT-mediated mitotic recombination also resulted in tumor-like phenotype inside hib À/À clones ( Figure 1F). Taken together, these results Inhibition of Ci-mediated tumorigenesis is dependent on HIB/SPOP E3 ligase activity HIB/SPOP usually binds with Cul3 through its BTB domain (residues 190 aa-297 aa) to form an E3 ligase complex and recognizes substrates via its MATH domain (residues 28 aa-166 aa). 49,50 To test whether HIB/SPOP E3 ligase activity is needed to inhibit the tumor-like eye phenotype, in co-overexpression of Ci À3P and hib RNAi background, we first check the rescue effects of two SPOP E3 ligase-dead truncated forms, SPOPDMATH and SPOPDBTB, respectively. As shown in Figure 2, we found SPOP but not SPOPD MATH and SPOPDBTB could rescue the tumor-like eye phenotype, SPOPDMATH and SPOPDBTB even manifested the dominant negative effect shown by boosting the tumor-like eye phenotype ( Figures 2C  and 2D), indicating that both MATH and BTB domain are needed to inhibit the eye tumorigenesis. Next, we further test whether another SPOP E3 ligase mutant, SPOPD3box, can inhibit the tumor-like eye phenotype. SPOPD3box lacks of 299 aa to 330 aa, which decreases the association with Cul3 and weakens functional SPOP-Cul3 E3 ligase. 51 We found that unlike SPOP, overexpression of SPOPD3box no longer rescued the tumor-like eye phenotype ( Figure 2E). Taken together, these results suggest that HIB/SPOP inhibits the eye tumorigenesis dependent on its E3 ligase activity.
The accumulation of ci alone is not sufficient to cause eye tumorigenesis Given HIB/SPOP modulates the tumor-like eye phenotype dependent on its E3 ligase activity, we next test whether the previous tumor-like eye phenotype caused by co-expression of Ci and hib RNAi is only due to hib RNAi-mediated upregulation of Ci level. Through modulating different copies of both Ci À3P and GMR Gal4 to increase the amount of Ci À3P , we found that expression of one and two copies of Ci À3P with one ( Figures 3A and 3C) or two copies of GMR Gal4 ( Figures 3B and 3D), none of them could cause tumorlike eye phenotype. Taken together, these results suggest that Ci À3P alone is not sufficient to generate the big eye phenotype, which needs hib RNAi together to produce the tumor-like eye phenotype.
Ci together with suitable Rpb3/Rpb7 of RNAPII complex mediates the tumorigenesis To unveil hib RNAi-affected factors which together with Ci mediate tumorigenesis, we did the secondround genetic screen in previous co-overexpression of Ci À3P and hib RNAi tumor-like eye background. By crossing with the RNAi lines of around 7000 conserved genes between fly and human, we got several hits, including rpb3, rpb7, and su(var)3-9. Because HIB did not bind SuVar3-9 and affect its stability

The tumor-like eye phenotype is related to excess interommatidial cells at pupal stage
The tumor-like eye phenotype may be generated from over proliferation at the larval stage or dampened apoptosis during the pupal stage. To determine which stage is important for formation of tumor-like eye, we use Gal80 to block the GMR-driven co-overexpression of Ci À3P and hib RNAi at different stages, the results showed that the tumor-like eye phenotype was generated at the pupal stage ( Figures 5A and 5B). Our subsequent study revealed that compared with wild type ( Figure 5C), Ci À3P ( Figure 5D) and hib RNAi (Figure 5E) alone, co-overexpression of Ci À3P and hib RNAi ( Figure 5F) presented excessive interommatidial cells, which are usually produced at the larval stage and then eliminated through apoptosis at the pupal stage for giving rise to the adult lattice. Collectively, the earlier results suggest that impaired apoptosis may be related to the tumor-like eye phenotype. Since Grim/Hid/Rpr-mediated apoptosis is involved in elimination of excessive interommatidial cells, next we check whether their mediated apoptosis was blocked upon co-overexpression of Ci À3P and hib RNAi. As shown in Figures 5G-5I%, co-overexpression of Ci À3P and hib RNAi blocked Grim/Rpr but not Hid-mediated apoptosis, implying that the tumor-like eye may be related to disability of apoptosis mediated by Grim/Rpr but not Hid in our case. Of note, overexpression of Ci alone already blocked Grim-mediated apoptosis, but that did not produce any eye phenotype, indicating that Grim is also not responsible for the tumor-like eye phenotype mediated by the cooperation of both Ci and hib RNAi. For Rpr, we surprisingly found that overexpression of Ci or hib RNAi alone could not, but they together completely blocked Rpr-mediated apoptosis, suggesting Ci and hib RNAi synergistically achieve the tumor-like eye phenotype possibly partially through inhibiting Rpr-mediated apoptosis ( Figures 5G-5I%). Consistently, TUNEL assay showed compared with Ci À3P and hib RNAi alone, co-expression of them inhibited apoptosis ( Figures 5J-5J%).
SPOP and Gli2 can substitute HIB and ci to mediate the tumor-like eye phenotype in Drosophila As mammalian homologue of HIB, SPOP is involved in various human cancers. In our previous study, we showed that SPOP function is evolutionally conserved in mediating Ci/Gli degradation. Consistently, using A B C D Figure 3. Ci À3P alone is not sufficient to generate the tumor-like eye phenotype (A and B) One copy of Ci À3P driven by one copy of GMR (A), or by two copies of GMR Gal4 (B) was not sufficient to generate the tumor-like eye phenotype.
(C and D) Two copies of Ci À3P driven by one copy of GMR (C), or by two copies of GMR Gal4 (D) were also not sufficient to generate the tumor-like eye phenotype. iScience Article SPOP to replace HIB, it totally rescued the eye phenotype, confirming SPOP function is also conserved in mediating the tumor-like eye phenotype. In addition, we generated transgenic flies of several human cancer-related mutants of SPOP, 46,52 including SPOP Y87N , SPOP F102C , SPOP S119N , SPOP F125L , SPOP W131G , which all prevented SPOP E3 ligase activity as shown by loss of the abilities to degrade both Ci ( Figures 6G-6K%) and Rpb3/Rpb7 ( Figures 7A and 7A 0 ). They no longer inhibited, instead even enhanced the tumor-like phenotype possibly due to their dominant negative effects ( Figures 6G-6K), further indicating SPOP E3 ligase activity is necessary for the rescue of tumor-like eye phenotype. Similarly, Gli2 could iScience Article substitute Ci together with knockdown of hib to cause the tumor-like eye phenotype (Figures 6A-6C). And knockdown rpb3/rpb7 could also totally rescue the Gli2-mediated tumor-like eye phenotype ( Figures 6D-6E 0 ).
Next, in mammalian 293T cells, we found SPOP could bind both hRpb3 and hRpb7, but only obviously degraded hRpb7 in its E3 ligase dependent manner ( Figures 7B and 7C). Contrastingly, cancer-related SPOP mutants no longer modulated hRpb7 stability ( Figure 7D). Except 293T cells, in A375, DaoY tumor cells, taking SPOP S119N as an example, it also inhibited hRpb7 degradation ( Figure 7E). Like Ci, Gli2 also bound hRpb7 and promoted hRpb7 degradation by SPOP ( Figures 7F and 7G). Consistently, in overexpression of Gli2 background, knockdown of spop dramatically upregulated hRpb7 levels in 293T cells and A375, DaoY tumor cells ( Figure 7H). Taken together, all these results suggest that in Drosophila, Gli2 and SPOP can substitute Ci and HIB to mediate the same tumor-like eye phenotype, in mammalian cells, hRpb7 stability is similarly regulated by Gli2 and SPOP, indicating that our model and the identified rescue genes are useful for future study of Hh-and SPOP-related tumorigenesis.

DISCUSSION
Hh signaling transduced by transcription factor Ci/Gli is essential for embryonic development and adult tissue homeostasis across species. Its dysregulation links with many kinds of human disorders, including various types of cancer, however, till now, cofactors of Ci/Gli which affect tumorigenesis are not well known. In this study, we find overexpression of active Ci alone is not sufficient to generate a tumor-like eye phenotype in Drosophila, however, its overexpression combined with knockdown of hib causes the striking tumor-like eye phenotype. However, co-overexpression of Ci and hib RNAi not only further increases Ci level but also lifts the burden on RNAPII machinery by stabilizing its subunits, therefore, leading to achieving the tumor-like eye phenotype more easily and effectively.
As mentioned, SPOP is involved in kinds of human cancer. The reason why it is related to a wide spectrum of tumors is not clear. Based on our study, dysfunction of SPOP will upregulate transcription machine RNAPII levels, leading to relatively easily reaching the thresholds of specific tumorigenesis. Additionally, except Ci/Gli, defective SPOP may also upregulate the levels of some other specific transcription factors which are usually degraded by SPOP. Taken these two aspects together, loss of function of SPOP may upregulate both specific transcription factors and general RNAPII machinery levels, leading to reaching the thresholds of various SPOP-regulated transcription factors-driven cancers easily. iScience Article As mammalian homologue of Ci, Gli includes Gli1, Gli2, and Gli3. Among them, Gli1 itself is also an Hh target gene, which and Gli2 mainly function as activators, while Gli3 as a repressor. It is reported that in some cases, loss of function of Ptch1 (Ptch1 À/À ) and gain of function of Smo (SmoM2) effectively induce MB, but overexpression of active Gli2 alone is not sufficient, the underlying reason is not clear. 53 Based on our study, any essential subunit dysfunction will impair the function of the whole RNAPII complex. We think in the context with SPOP, SPOP degrades the subunit hRpb7 of RNAPII, leading to forming low level functional RNAPII complex. Additionally, overexpressed Gli2 further promotes degradation of hRpb7, leading to forming low level functional Gli2/RNAPII complex according to limited Rpb7 level but not the overexpressed Gli2 level, subsequently causing correspondent low level Gli1/RNAPII complex. Therefore, high level overexpressed Gli2 with non-matched low level RNAPII generates low level functional Gli2/RNAPII and Gli1/RNAPII complexes, which is not easy to reach the needed threshold to effectively trigger MB. However, even Ptch1 À/À and SmoM2 activate Gli2, whose level is lower than overexpressed Gli2, therefore, it mediates relatively weak inhibition on RNAPII. Consequently, this leads to a little higher RNAPII, therefore, resulting in forming correspondent a little higher functional Gli2/RNAPII and Gli1/RNA-PII complexes. But such little increase may be still not enough to effectively trigger MB. Except this, Ptch1 À/À and SmoM2 also inhibit Gli3's repressor function, this may decrease MB threshold or further tip the balance to trigger MB. Taken together, these results imply that Gli1/2/3, all streams may function together to effectively fulfill induction of MB. In contrast to previous case, in the context without SPOP or SPOP dysfunction, RNAPII maintains relative high level, leading to forming relative high level Gli2/RNA-PII complex, which consequently causes correspondent high level Gli1/RNAPII. Therefore, they together may already reach the threshold of tumorigenesis. In this case, we predict Ptch1 À/À , SmoM2 and overexpression of active Gli2 alone all may effectively cause tumorigenesis, including MB.
As for why overexpression of Ci together with knockdown of hib antagonizes the apoptosis mediated by Rpr, but either overexpression of Ci or knockdown of hib alone not, the underlying mechanism is elusive. The specificity is possibly due to that overexpression of Ci plus knockdown of hib forms high level Ci/RNAPII complex which surpasses the specific transcription threshold needed for antagonizing Rprmediated apoptosis or high Ci/RNAPII may selectively express Ci target genes needed to fulfill the distinct roles compared with their alone. Of note, simultaneous overexpression of Ci and DIAP1 or P35 which blocks apoptosis did not cause tumor-like eye phenotype ( Figure S1), suggesting Ci and hib RNAi together-mediated tumor-like eye phenotype may be only partially through inhibiting apoptosis, except that, other unidentified process must be involved in driving the big eye phenotype.
Finally, SPOP and Gli2 can substitute HIB and Ci to mediate the same tumor-like eye phenotype in Drosophila, in mammalian cells, hRpb7 stability is similarly regulated by Gli2 and SPOP, indicating that they are functionally conserved. Our study demonstrates SPOP is a very important cofactor of Ci/Gli, which plays a dual role in regulating both Ci/Gli and RNAPII machinery for suitable Hh signaling outcome, its loss of function will unleash Ci/Gli and RNAP II machinery to cause abnormal high Hh signaling or parallel signaling, leading to various Hh-related tumorigeneses. Accordingly, this mechanism is also applied to SPOP-regulated other transcription factors-related kinds of cancer.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Qing Zhang (zhangqing@nju.edu.cn).

Materials availability
This study did not generate new unique reagents.

Data and code availability
The data reported in this paper will be shared upon request to the lead corresponding author (zhangqing@nju.edu.cn).
This paper does not report original code.
Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Cell lines
The Drosophila embryonic cell line S2 were obtained from Yun Zhao lab (CAS Center for Excellence in Molecular Cell Science, Shanghai, China) and was cultured in the Schneider's Drosophila Medium (Invitrogen) with 10% fetal bovine serum (FBS, Gibco),100 U/ml of penicillin and 100 mg/mL of Streptomycin (P/S, Gibco) in a humidified incubator at 28 C. The human embryonic kidney cell line 293T and human melanoma cell line A375 were obtained from Ying Cao lab (Nanjing university, Nanjing, China). The human medulloblastoma cell line DaoY was obtained from Chen Liu lab (Nanjing Medical University, Nanjing, China). They were cultured in Dulbecco's modified Eagle's medium (DMEM, HyClone), with 10% fetal bovine serum (FBS, Gibco),100 U/ml of penicillin and 100 mg/ml of Streptomycin (P/S, Gibco) in a humidified incubator with 5% CO 2 at 37 C.