The Hira complex regulates Gli3R-dependent transcription in Hedgehog signaling and medulloblastoma cell growth and migration

Abstract

Regulation of the Hedgehog pathway activity may be supported by coactivators and corepresors of its main effectors- Gli transcription factors. While activation processes are well studied, repression mechanisms remain elusive. We identified chromatin remodelling complex Hira to interact with Gli3R protein, showed that its loss-of-function changes Hh pathway activity, and examined possible mechanism behind the observed effect. We also established that Hira influences the viability and migratory abilities of Hh-dependent medulloblastoma Daoy cells. Our study paves the way for a better understanding of processes involved in Hh pathway regulation and Hh-dependent carcinogenesis.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-83468-3.

Introduction

The Hedgehog (Hh) signaling pathway plays a major role in embryonic development by regulating cell fate, proliferation and differentiation, leading to various tissue patterning and organ growth^1–3. Uncontrolled Hh activation results in the development of numerous cancers such as skin, muscle, cerebellum, pancreas, or aggressive prostate tumors⁴. The main effectors of Hh pathway are Gli proteins from Glioma Associated Oncogene Family. In the active pathway, full-length Gli proteins, mainly Gli1 and Gli2, act as activators (GliA), and when the pathway is inactive Gli proteins, especially Gli3, become C-terminally truncated and act as transcriptional repressors (GliR)⁵.

So far, little is known about the mechanisms of Gli3-dependent gene repression, however, there are some proteins described as Gli3 corepressors that are involved in this process. Among them is Suppressor of Fused (Sufu), which was shown to control the process of Gli3R production⁶ but also has corepressor function as it may recruit to the chromatin repressor complexes like HDAC1 (Histone Deacetylase Complex 1) or NuRD (nucleosome remodeling and histone deacetylase)⁵. Another protein is SAFB-like transcription modulator (SLTM) which takes part in Gli3FL binding to the chromatin, increases the level of Gli3R, and promotes repressive chromatin environment by implementing H3K27me3 mark on Gli1 gene promoter⁷. Also Ski protein was described as one of Gli3R corepressors and Gli3FL negative regulators⁸. N-terminal domain of Gli3R protein has an important role in the repression process. It recruits corepressors like Ski or HDAC complex, also Sufu mediated interaction with corepressors occurs at the Gli3R N-terminal domain⁵.

Recent genome-wide CRISPR/Cas9 loss of function screens made it possible to identify novel negative regulators of the Hh pathway and potential Gli3R corepressors^9,10. One of the negative regulators identified in one of the screens is Hira, which we also identified as an interactor of Gli3 in a co-IP/MS proteomic screen¹¹. The Hira complex is composed of Hira, Cabin1, Asf1a, and Ubn1/2 proteins. It is responsible for cell cycle-independent deposition of H3.3/H4 histones to the chromatin, which allows for gene expression regulation. Two independent and functionally partially redundant, complexes Ubn1-Hira and Ubn2-Hira are present in the cell¹². It was shown that Hira complex may repress gene transcription during stem cells differentiation process; Hira knock out results in genome-wide reduction in gene repression¹³. Repressive role of the complex was shown in erythroid differentiation of adult hematopoietic stem cells. By recruiting H3.3 histone Hira complex creates an environment sufficient for polycomb-dependent gene repression¹⁴. It was also shown that Hira complex is involved in the formation of SAHF (senescence-associated heterochromatin foci) at genes connected with cell proliferation in senescent human cells which leads to the repression of gene expression¹⁵.

In this manuscript we study the interaction between Gli proteins and the Hira complex, and further characterize the role of Hira as a potential Gli corepressor. In loss-of-function experiments we show that the Hira complex affects Gli-mediated gene expression, as well as the proliferation and viability of medulloblastoma cells. We also show that Hira interacts with Sufu and explore the possible mechanism of Hira complex dependent gene repression.

Results

Hira complex interacts with Gli3 repressor

Among nuclear Gli3 interactors that we had identified in a proteome-wide screen using co-IP/MS¹¹, we found components of Hira complex (Hira, Ubn1, Ubn2, and Cabin1) (Fig. 1A). To determine if Hira preferentially interacts with activator or repressor forms of Gli3, we compared the strength of the interaction between the different Gli3 isoforms (Fig. 1B) and either Hira alone or Hira in the presence of other complex components using co-IP/western blot. Hira alone interacts with full-length wild-type Gli3 and with constitutively active full-length Gli3(P1-6 A)¹⁶ however, the repressor form Gli3R only interacts with Hira in the presence of the other Hira complex components (Fig. 1C; D). To confirm Gli3R-Hira interaction on an endogenous level, we used cells with doxycycline inducible HA-Gli3R. Similar to experiments on overexpressed Hira, HA-Gli3R co-immunoprecipitates endogenous Hira (Fig. 1E). To further understand the Hira complex-Gli3R interaction, we generated HA-tagged Gli3R constructs consisting of the repressor domain, the Sufu binding site, and the zinc finger domain (Fig. 1B), and tested if any of them will preferentially interact with Hira complex. We found the interaction in all tested domains, however the strongest one was with the repressor domain (Fig. 1F; G).

Fig. 1.

Hira complex interacts with Gli proteins. (a) Hira complex components identified as Gli3 interactors in mass-spectrometry analysis as described in Niedziolka et al., 2024. (b) Schematic representation of HA-tagged Gli3 and Gli3R constructs used in co-immunoprecipitation experiments. REP-represor domain, SF- Sufu binding site, ZF- Zing finger domain, A1/A2- activator domain. (c) Co-immunoprecipitation of overexpressed HA-tagged Gli3 constructs and Hira protein (d) or Hira complex in HEK293T cells using anti-HA beads. Original western blots are presented in Supplementary Fig. S4-1. (e) Endogenous co-immunoprecipitation of Hira complex and HA-Gli3R in Gli3R-HA-Tet-On NIH/3T3 cells using anti-HA beads. Original western blots are presented in Supplementary Fig. S4-2. (f) Co-immunoprecipitation of overexpressed HA-tagged Gli3R domains and Hira complex in HEK293T cells using anti-HA beads. Representative results choosen from three independent biological replicates of experiment. Original western blots are presented in Supplementary Fig. S4-2, 4-3, 4-4. (g) Plot showing ratio of immunoprecipitated vs. input level of Hira protein in three independent biological replicates of experiment. Results were normalized to the tubulin level in input samples. Dox -doxycycline treatment 300 μg/ml for 24 h. Student’s t-test * p < 0.05.

Because Gli3R only interacts with Hira in the presence of Ubn2, we wanted to know what domains of Ubn2 are responsible for the binding with Gli proteins. Specific Hpc2-related domain (HRD) localized at the N-terminus of Ubn1/2 is responsible for histone H3.3 interactions¹⁷. The middle domain is involved in both H3.3 and DNA binding¹⁸. We performed co-immunoprecipitation of Gli3R and Hira complex containing specific Ubn2 domains (Supplementary Fig. S1A). We identified two of them to interact with Gli3R: 168-368aa domain, containing HRD domain and 737-1349aa domain, a C-terminal part of Ubn2, whose function is unknown (Supplementary Fig. S1B; C;D). We show that Ubn2 construct lacking the 168-368aa domain interacts with Gli3R less strongly than full-length Ubn2 protein or Ubn2 lacking the 737-1349aa domain (Supplementary Fig. S1E). Interestingly, Gli3R interacts strongly with Hira in the absence of exogenous Ubn2 or Cabin1 (Supplementary Fig. S1F; G), but not when both components are missing (Fig. 1C).

Hira loss-of-function increases Hedgehog pathway activity

Hira is a scaffold protein that’s essential for Hira complex function and it was identified among negative Hedgehog pathway regulators in CRISPR/Cas9 based screen⁹. To examine the effect of Hira loss of function on Hedgehog signaling, we knocked it out in NIH/3T3 cells using the CRISPR/Cas9 method and tested Hedgehog pathway activity by measuring the level of expression of the Hh target gene Gli1. To be sure that we obtained the full knockout we isolated single clone in which Hira frame-shift mutation was confirmed both by western blot (Fig. 2A) and sequencing (Supplementary Fig. S2). Upon Hira knockout, the basal level of Gli1 is elevated and cells are more sensitive to pathway activation by the Smo agonist SAG, we did not observe any changes in the level of Gli3R and Gli3FL protein (Fig. 2A; B). We were able to reverse these effects by rescuing Hira expression in KO cells (Fig. 2C; D).

Fig. 2.

Effect of Hira loss-of-function on Hedgehog pathway activity. (a) Western blot analysis of Gli1, Gli3FL, and Gli3R protein levels in Hira knock out and Control NIH/3T3 cells upon SAG treatment. Original western blots are presented in Supplementary Fig. S4-5. (b) mRNA expression level of Gli1 gene in Hira knock out and Control NIH/3T3 cells upon SAG treatment (n = 3). (c) Western blot analysis of Gli1 protein level in Control, Hira knockout, and Rescue NIH/3T3 cells upon SAG treatment. Original western blots are presented in Supplementary Fig. S4-5. (d) mRNA expression level of Gli1 gene in Control, Hira knockout, and Rescue NIH/3T3 cells upon SAG treatment (n = 3). (e–j) Western blot analysis of Gli1 and Hira protein levels in Ubn1, Ubn2, and Ubn1/2 knockout NIH/3T3 cells upon SAG and Dox treatment and corresponding mRNA expression level of Gli1 gene upon SAG treatment. Original western blots are presented in Supplementary Fig. S4-5, 4-6. (k) Ciliary localization of Gli3 protein in Control and Hira knockout NIH/3T3 cells. Cells per variant > 100. Violin plot present log-transforrmed ratios of Gli3 fluorescence intensity at cilia tips to the surrounding background. (l) Percent of ciliated cells in Control and Hira knockout NIH/3T3 groups. SAG – SAG treatment 200nM for 24 h, Dox - doxycycline treatment 300 µg/ml for 24 h, Student’s t-test *** p < 0.001; ** p < 0.01; * p < 0.05.

Because Ubn1 and Ubn2 interact with Gli3 and because they form alternative complexes with Hira, Cabin1, and Asf1, we wanted to know which of the two Ubn isoforms plays a major role in the Hira complex function in Hh signaling. To that end, we performed single and double knockouts of Ubn1 and Ubn2 proteins and examined the activity of the Hedgehog pathway by measuring the level of Gli1. Upon single knockouts, we did not observe any significant differences (Fig. 2E; F;G; H), but in double knockout cells, we observed increased sensitivity to SAG treatment (Fig. 2I; J). Interestingly, we observed decreased Hira protein levels in both single and double knockout cells, suggesting that Ubn proteins stabilize Hira in cells.

The mechanism of Hira-dependent Hh pathway repression

Because Hira knockout induces an increase in Hh-dependent gene expression, we wanted to know what mechanism is responsible for this increase. One of the mechanisms by which Gli protein function is regulated is their transport to primary cilia, which both affects Gli3R formation and Gli3FL activation. However, Hira KO cells did not have impaired Gli3 ciliary transport – both the baseline Gli3 localization and the SAG-induced accumulation of Gli3 at cilia were slightly elevated in Hira KO cells (Fig. 2K). We also did not observe a difference in the number of ciliated cells between control and Hira KO cells (Fig. 2L). This is consistent with the repressor being formed at normal levels in Hira KO cells (Fig. 2A) and suggests that ciliogenesis and ciliary function are no responsible for the effects we observed on Gli-dependent gene expression.

Next, we checked the interaction between Hira and Sufu, a protein that is a well-established Gli protein inhibitor⁶. We were able to co-immunoprecipitate HA-tagged Sufu and the Hira complex, showing that these proteins interact with each other (Fig. 3A). However, Gli3R-HA co-immunoprecipitates with Hira even in Sufu KO cells, albeit with less potency than in control cells (Fig. 3B). This suggests, that while Sufu may facilitate the interaction between Gli proteins and the Hira complex, it is not required for it. Additional studies are needed to understand the role of this interaction in Hedgehog pathway repression processes.

Fig. 3.

Hira complex may interact with Gli3R protein in the absence of Sufu. (a) Co-immunoprecipitation of overexpressed HA-tagged Sufu and Hira complex in HEK293 cells using anti-HA beads. Original western blots are presented in Supplementary Fig. S4-6. (b) Co-immunoprecipitation of doxycycline-induced HA-tagged Gli3R and endogenous Hira in control and Sufu KO NIH/3T3 cells using anti-HA beads. Original western blots are presented in Supplementary Fig. S4-7. (c) Western blot analysis of Gli3FL and doxycycline-induced HA-tagged Gli3R protein level after treatment with 30,100–300 µg/ml doxycycline for 24 h in control and Hira KO NIH/3T3 cells. Original western blots are presented in Supplementary Fig. S4-7. (d) mRNA expression level of Gli3 gene and (e) Gli1 gene after treatment with 30,100–300 µg/ml doxycycline for 24 h in control and Hira KO NIH/3T3 cells, n = 3. Dox-doxycycline treatment 300 µg/ml for 24 h; Student’s t-test *** p < 0.001; ** p < 0.01; * p < 0.05.

To further investigate the mechanism of Hira-dependent Hh-pathway repression, we checked if Hira KO affects the function of Gli3R. We treated control and Hira KO cells with doxycycline-dependent Gli3R expression with different doses of doxycycline to test if Hira knockout may impact externally induced Gli3R-HA level or activity. Although at lower doses of doxycycline Gli3R seems to be induced stronger than in the control, in the highest concentration of doxycycline (300 µg/ml) we did not observe differences in the expression of Gli3R-HA (Fig. 3C; D). Moreover, the strength of Gli1 repression by equivalent amounts of Gli3R appears identical in control and Hira KO cells, suggesting that Gli3R repressive function remains intact in the absence of Hira (Fig. 3E).

Next, to determine how Hira affects Gli3R chromatin occupancy, we performed a CUT&RUN experiment in wild type and Hira KO Gli3R-HA Tet-ON cells with Gli3R expression fully induced by 24 h doxycycline treatment at 300 µg/ml. We found 3618 Gli3R peaks in Hira KO cells and 2445 of them were identified also in the control group (Fig. 4A). Among peaks common for both the control and the Hira KO we identified regions of Hedgehog-dependent genes Gli1 and Ptch1, which were enriched in knockout cells (Fig. 4B). We observed also overall elevated promotor occupancy in Hira KO cells (Fig. 4C). However, KEGG Pathway Enrichment Analysis showed no significant changes between gene sets occupied by Gli3R in control and Hira KO cells (Fig. 4D).

Fig. 4.

Hira knockout increases Gli3R chromatin binding at target genes. (a) Schematic representation of the number of Gli3R specific peaks identified in CUT&RUN experiment in control and Hira knockout Gli3R-HA-Tet-On NIH/3T3 cells. (b) Gli3R occupied regions of Hedgehog-dependent genes Gli1 and Ptch1 enriched in Hira knockout Gli3R-HA-Tet-On NIH/3T3 cells. (c) Distribution of Gli3R occupied regions of chromatin in control and Hira knockout Gli3R-HA-Tet-On NIH/3T3 cells. (d) Gli3R occupancy enrichment analysis of pathway specific genes performed with the use of the KEGG database in control and Hira knockout Gli3R-HA-Tet-On NIH/3T3 cells, p-value Cutoff = 0.05, p.adjust method = Benjamini-Hochberg. (e) Volcano plot showing genes identified in RNAseq analysis as Gli3R-dependent (p-value Cutoff = 0.05, log2FoldChange Cutoff = 0.5) in Gli3R-HA-Tet-On NIH/3T3 cells. Genes downregulated by Gli3R that were also identified as enriched Gli3R targets in CUT&RUN Hira knockout group are labeled on the plot. (f) Integrated genomics viewer tracs of genes identified in RNAseq as downregulated upon Gli3R and enriched in Hira knockout cells in the CUT&RUN experiment.

To identify additional Gli3R target genes, we performed RNAseq analysis of gene expression in cells with inducible Gli3R expression. Among genes downregulated by Gli3R we identified genes that were enriched in our CUT&RUN in Hira KO group (Supplementary Fig. S3, Fig. 4E). Two of these genes Adamts1 and Cdsn were previously described as upstream Hedgehog pathway regulators or Hh-target genes^19,20. We identified also Acot2, Isoc1, Cyb5r1, Car6, and Dlk1 genes that had not been examined so far in the context of Hedgehog pathway. For all these Gli3R target genes, we saw Gli3R chromatin binding peaks, and in all cases the binding of Gli3R to chromatin was enhanced in Hira KO cells (Fig. 4F). These results suggest that Hira does not impair the binding of Gli3R to chromatin, as might be suggested by the effect of Hira KO on gene expression. Further studies will be necessary to determine the precise mechanism of Hira-mediated suppression of Hh target genes.

Hira loss of function influence viability and migratory abilities of Daoy medulloblastoma cells

To determine the potential role of Hira in Hh-dependent carcinogenesis, we tested the effect of Hira knockout in the Hh-dependent human medulloblastoma cell line Daoy. First, we performed CRISPR/Cas9 knockout experiment and examined Hh pathway activity by measuring the GLI1 protein and mRNA levels. Daoy Hira KO cells have elevated basal GLI1 level and are less sensitive to external pathway activation via SAG treatment (Fig. 5A; B).

Fig. 5.

Hira knockout decreases proliferation and viability of Daoy cells. (a) Western blot analysis of Gli1 protein level in control and Hira knockout Daoy cells. (b) mRNA expression level of Gli1 gene in control and Hira knockout Daoy cells. Original western blots are presented in Supplementary Fig. S4-8. (c) Crystal violet staining of colonies from colony formation assay in control and Hira knockout DAOY cells. (d) Flow cytometry analysis of combined Annexin V + PI staining in control and Hira knockout Daoy cells and barplot representation of a summary of three independent repetitions of the experiment. Cells were labeled with Annexin V-FITC and directly before analysis PI was added to label live cells. Then, gates for live, necrotic, late, and early apoptosis cells were set (n > 20000). (e) Immunofluorescent staining of Ki-67 proliferation marker in control and Hira knockout Daoy cells (n > 130) (f) BrdU incorporation assay staining of control and Hira knockout Daoy cells (n > 130). SAG - SAG treatment 500 nM for 48 h, Student’s t-test *** p < 0.001; ** p < 0.01; * p < 0.05.

To determine how this affects tumor cell growth and metastatic potential, we measured cell viability and mobility in control and Hira KO cells. Hira knockout results in decreased size and number of colonies in a colony formation assay (Fig. 5C). To understand why Hira KO Daoy cells form fewer and smaller colonies, we examined cell proliferation and death. Combined annexin V + PI flow cytometry analysis shows an increase among early and late apoptotic cells in Hira KO group (Fig. 5D). To test whether cell proliferation was affected by Hira KO we performed Ki-67 staining and the BrdU incorporation assay. Hira KO cells have lower expression of the proliferation marker Ki67 and incorporate BrdU at a lower rate than control cells (Fig. 5E; F). Next, we tested the migratory potential of Daoy Hira KO cells in the scratch assay and the transwell migration assay. In both tests, we observed that knockout cells migrate at a slower rate than the controls (Fig. 6A; B ).

Fig. 6.

Hira knockout changes expression profiles of genes involved in epithelial to mesenchymal transition (EMT) in Daoy cells. (a) Barplot showing the numer of migrated cells in transwell migration test in control and HIRA knockout Daoy group (n = 3). (b) Representative photos of scratch assay showing migration ability of control and HIRA knockout Daoy cells (n = 4). (c) GeneOntology analysis of genes upregulated upon HIRA knockout, identified in RNAseq analysis., p-value Cutoff = 0.05, p.adjust method = Benjamini-Hochberg. (d) GeneOntology analysis of genes downregulated upon HIRA knockout identified in RNAseq analysis p-value Cutoff = 0.05, p.adjust method = Benjamini-Hochberg. (e) Volcano plot of differentially expressed genes in Hira knockout cells identified in RNAseq analysis (p-value Cutoff = 0.05, log2FoldChange Cutoff = 0.5). (f) Western blot analysis of E-cadherin and Vimentin protein levels in control and HIRA knockout Daoy cells. Original western blots are presented in Supplementary Fig. S4-8. Student’s t-test ** p < 0.01.

To explore the molecular underpinnings of the Hira KO-driven phenotype in Daoy cells, we performed RNA-seq analysis of gene expression in control and Hira KO Daoy cells. Consistent with the results of the cell proliferation and migration assays, Hira KO causes upregulation among genes connected with cadherin binding and cell adhesion, and downregulation of genes involved in regulation of locomotion and epithelial cell proliferation in Daoy cells (Fig. 6C; D). Consistent with our earlier results, we also found upregulation of Hedgehog target genes GLI1 and HHIP expression upon Hira knockout (Fig. 6E). The decrease of the migratory potential, as well as increase in genes associated with cell adhesion, is a hallmark for the inhibition of epithelial to mesenchymal transition (EMT). To confirm that Hira KO blocks EMT in Daoy cells, we checked the expression of epithelial and mesenchymal markers. Cells without Hira have elevated level of the epithelial marker E-CADHERIN and decreased level of the mesenchymal marker VIMENTIN (Fig. 6F), suggesting that Hira is required for the transition of the cells to a more aggressive metastatic phenotype.

Discussion

One of the more intriguing features of Gli proteins is their ability to function as both transcriptional activators and repressors^5,21. Whereas Gli-mediated transcriptional activation is relatively well studied, little is known so far about possible mechanisms of Gli repression. One of the proposed mechanisms is the competition of Gli repressor forms with transcriptional activators binding. However, this mechanism is unlikely to explain the entirety of Gli repressor function. There are studies indicating an important role of the repressor domain, localized on the N-terminal part of the Gli protein, in the process of gene repression. Cells lacking the repressor domain was shown to have reduced repression of Hh-target gene Gli1²². An alternative mechanism of Gli repression is the recruitment of corepressors, which happens mainly by the protein N-terminal repressor domain⁵. Corepressors form functional complexes with GliR, which modulate the chromatin environment to inhibit gene expression^6–8. The role of corepressors in gene expression regulation is often challenging to study because of the complexity of the repression process and a wide variety of possible mechanisms from simple on/off switches to more subtle repression that appears only for a short time and modestly decreases gene expression level²³. In the Hedgehog pathway this aspect of regulation is poorly studied despite its relevance for disease mechanisms^24–26.

In search for proteins that enhance repression of Gli proteins, we combined Gli3 co-IP/MS proteomic screen performed in our laboratory¹¹ and genome-wide CRISPR/Cas9 loss of function screens^9,10. The Hira protein appeared in both proteomic and functional screens. Using Hira knockout cells, we confirmed the role of Hira as a negative regulator of Hh target gene expression.

Hira had previously been found to interact with several transcription factors, but this interaction has predominantly been found to increase rather than decrease gene expression, despite the documented role of Hira homologs as transcriptional repressors in yeast. Interaction of Hira with EKLF is essential for β-globin gene expression²⁷, its interaction with RUNX1 and BRG1, positively regulates the expression of hematopoietic genes²⁸, and its interaction with PHB complex positively modulates the transcription of EMT-associated genes²⁹. The specific effects of Hira on gene expression may depend on the co-factors that it interacts with, specifically Asf1a, Cabin1, and Ubn1/2. Studies on specific chromatin regions occupied by Hira complex components show regions specific only for Hira, for Hira and Ubn1/Asf1 proteins, and for all components³⁰. Our data suggest that while activator forms of Gli interact with Hira on its own, the repressor form Gli3R requires the expression of additional complex components Ubn1 and Cabin1 to stabilize the interaction. Importantly, it appears that the Ubn1/2 proteins are essential for the downregulation of Hh target genes. We show that while the single Ubn1 or Ubn2 knockout does not affect Hedgehog pathway function, the double Ubn1/2 knockout increases the sensitivity of the pathway to SAG treatment. This suggests that the apparent repressive effect of Hira on the Hh pathway requires Ubn1/2.

We considered several mechanisms on how Hira knockout may contribute to the derepression of Hh target genes. First, we checked if the Gli3 repressor formation was impaired in Hira KO cells, but we saw no difference in the level of Gli3 full length and repressor between control and knockout cells. We also wanted to know if the binding of Gli3R to chromatin might be impaired, but paradoxically we noticed an increased occupancy of Gli3R at Hh target genes in Hira KO cells. Moreover, the ability of overexpressed Gli3R to block Hh the expression of Hh target genes was not affected by the absence of Hira. These observations, coupled with the fact that Hira binds both repressor and activator forms of Gli proteins, suggest that the mechanism of the action of Hira in Hh target gene regulation is likely to be more indirect. For instance, the knockout of Hira may contribute to an overall less compact chromatin structure at Gli target genes which both helps Gli3R bind, but also partially derepresses the genes in the absence of forced Gli3R expression. Hira KO was shown to both increase and decrease chromatin accessibility depending on the genomic locus⁷. Hira complex is known mostly for regulating gene expression by incorporating the H3.3 histone variant to change the chromatin state. This histone was originally described to be involved in gene activation process³¹ but may have diverse functions in distinct genomic regions. Recent studies indicate its function in polycomb repressive complex (PRC2) related gene repression³². Hira complex- dependent deposition of H3.3 is necessary for incorporation of H3K27me3 repressive mark by PRC2 complex during hematopoietic stem cells development and upon Hira knockout genes involved in this process become more accessible¹⁴. Future work will provide more clarity on the exact mechanism of Hira involvement in the negative regulation of Hh target genes.

Elevated Hedgehog pathway activity is typical for many Hedgehog-dependent cancers and associated with tumor progression, increased cell proliferation, and survival. Among Hh pathway proteins, Smo has been the most fruitful anti-cancer drug target. However, anti-Smo drugs fail when Smo mutations appear, conferring resistance³³. Moreover, Smo-independent Gli-driven cancers may arise de novo from non-canonical pathway activation or Sufu mutations³³. For these reasons, a more promising therapeutic approach seems to be directly inhibiting Gli proteins. GANT61, one of the few direct Gli inhibitors was shown to inhibit Gli1 and Gli2 which leads to DNA damage induction, caspase cleavage, and cell death of human colon carcinoma³⁴. Also, treatment with a combination of GANT61 and RITA inhibitors gives a reduction in rhabdomyosarcoma tumor growth in mice³⁵. However, despite their efficacy in preclinical models, Gli inhibitors have not successfully cleared clinical trials. Therefore, novel Gli-related drug targets are urgently needed.

In our study, we knocked out Hira in Hedgehog-dependent medulloblastoma Daoy cell line and observed an increase in GLI1 protein basal level, as we would expect for GLI repression deficient cells. However, this Hh pathway upregulation did not result in elevated cell survival or migratory abilities. On the contrary, we observed increased apoptosis rate, decreased colony formation and proliferation abilities in Hira KO cells. What is more, RNAseq analysis shows upregulation of genes related to cadherin binding and downregulation of genes related to proliferation and locomotion. This is consistent with functional migration and colony formation assays we performed on these cells. The Hira complex, in addition to regulating the Hh pathway, is also involved in other processes connected with gene activation like EMT induction in breast cancer²⁹ and enhancement of invasive and proliferative abilities³⁶. On the other hand, Hira KO may lead to cancer cell death by limiting the lengthening of telomers³⁷, and to cell growth slowdown³⁸. Our transcriptomic investigation of Hira KO cells implies broad effects on gene expression, unlikely to be dependent only on Hedgehog pathway upregulation. Targeting the Hira complex may therefore be a promising strategy in Hh-dependent cancers despite its repressive effect on the Hedgehog pathway.

Materials and methods

Cell culture

NIH/3T3 (Thermo fisher), Daoy (ATCC), and HEK293T (ATCC) cells were cultured in complete media DMEM (high glucose, Biowest) supplemented with 10% fetal bovine serum(EurX), stable glutamine (Biowest), non-essential amino acids (Thermo Fisher), sodium pyruvate (Thermo Fisher), and penicillin/streptomycin solution (Thermo Fisher).

Gli3R-HA TetOn NIH/3T3 stable cell line was generated with the use of specially designed LT3GEPIR-Gli3R-HA vector and 3rd generation lentiviral system. Cells were reselected under the puromycin on every other passage.

Hedgehog pathway was induced by SAG (Biomibo, No#ALX-270-426-M001)treatment 200nM for 24 h (NIH/3T3 cells) or 500nM for 48 h (Daoy cells) in culture media containing 0.5% of fetal bovine serum. To induce Gli3R-HA expression stable cell lines were treated with doxycycline 300ug/ml for 24 h in culture media containing 0.5% of fetal bovine serum.

Lentiviral transfection

To produce lentiviruses HEK cells were transfected with pRSV-rev, pMDLg/pRRE, pMD2.G packaging vectors and vector containing construct of interest (Table 1). After 48 h media containing lentiviruses was taken from HEK cells, filtered and added to the NIH/3T3 or Daoy cells. After another 48 h neomycin, puromycin or blastocidin was added for selection.

Table 1.

Plasmids used for lentiviral transfections.

LentiCas9-Blast (#52962)	Addgene
LentiGuide-neo (#139449)	Addgene
LentiGuide-puro (#52963)	Addgene
LentiGuide-hygro(#139462)	Addgene
pRSV-Rev (#12253)	Addgene
pMDLg/pRRE (#12251)	Addgene
pMD2.G (#12259)	Addgene

CRISPR-Cas9 knockout

sgRNA sequences were designed using the Benchlink sgRNA designed tool (Table 2) and cloned into pLentiGuide vector (Addgene). Daoy and NIH/3T3 cells were transdued with lentiviruces carring sgRNA of interest and reselected with proper antibiotics.

Table 2.

Sequences used for CRISPR edition and RT-qPCR primers.

sgRNA-Ubn2 (Ms)	GGTGGTACCTACACTTCCAG
sgRNA-Ubn1 (Ms)12	TAGCCATGTCGGAGCCCCAC
sgRNA-Hira (Ms, Hs)	ATGACAAACTGATTATGGTG
sgRNA-Sufu (Ms)	ATACCAGTACTTGACGATAG
qPCR-GAPDH (Ms) F	GGCCTTCCGTGTTCCTAC
qPCR-GAPDH (Ms) R	TGTCATCATACTTGGCAGGTT
qPCR-Gli1 (Ms) F	CCAAGCCAACTTTATGTCAGGG
qPCR-Gli1 (Ms) R	AGCCCGCTTCTTTGTTAATTTGA
qPCR-GAPDH (Hs) F	ACATCGCTCAGACACCAT
qPCR-GAPDH (Hs) R	TGTAGTTGAGGTCAATGAAGGG
qPCR-Gli1 (Hs) F	TCTGGACATACCCCACCTCCCTCTG
qPCR-Gli1 (Hs) R	ACTGCAGCTCCCCCAATTTTTCTGG

Co-immunoprecipitation

Co-immunoprecipitation was performed with the use of Pierce Anti-HA Magnetic Beads (Life Technologies).

Cells were lysed at 4^oC in lysis buffer (50mM Tris pH7.4, 420mM NaCl, 0.5% NP-40 [v/v], 0.25% sodium deoxycholate [v/v], protease inhibitor cocktail [1× EDTA-free protease inhibitors, Sigma], 1mM sodium orthovanadate, 10mM sodium fluoride).

1/10 of the lysate was saved as an input sample and the rest was subjected to overnight immunoprecipitation with beads at 4^oC.

Next day beads were washed 3 × 5 min at RT in washing buffer (50mM Tris pH7.4, 420mM NaCl, 0.5% NP-40 [v/v], 0.25% sodium deoxycholate [v/v]) to remove unbound proteins. Then, proteins were eluted from beads in 2xSDS in 37 ^oC for 30 min and analyzed by SDS-PAGE and Western blot method.

SDS-PAGE and Western blot

Samples were denatured for 30 min in 65 ^oC and subjected to SDS-PAGE. Next proteins were transferred onto a nitrocellulose membrane using a fast-dry electrotransfer. Membranes were blocked for 1 h at RT in 5% low-fat milk in TBST buffer (0.5% Tween 20 in Tris buffer saline, pH 7.5) with gentle rocking. Then, membranes were incubated with primary antibody (Table 3) dissolved in blocking buffer overnight at 4 ^oC with gentle rocking. The next day membranes were washed 3 × 10 min in TBST buffer, incubated with secondary antibody (Table 3) dissolved in blocking buffer for 1 h at RT, and washed again 3 × 10 min. Prior to incubation with primary antibody membranes were cut based on the expected molecular weight of the proteins of interest.

Table 3.

Antibodies used for western blotting, CUT&RUN, and immunofluorescence stainings.

Antibody	Dilution
Anti-HA (Biolegend, 901501)	WB 1:1000 IF:1:500 C&R: 1 µg
Anti-c-Myc (DSHB, 9E10-s)	WB 1:25
Anti-Ubn2 (Novus, NBP2-9416)	WB 1:1000
Anti-Hira (Sigma, 04-1488)	WB 1:1000
Anti-α-Tubulin (Sigma, T6199)	WB 1:1000
Anti-β-actin (Sigma, A1978-100UL)	WB 1:1000
Anti-Lamin A/C (Thermo Fisher, MA5-35284)	WB 1:1000
Anti-FLAG (Thermo Fisher, MA1-91878)	WB 1:1000
Anti-Gli1 (Cell signaling, 2643 S)	WB 1:1000
Anti-Gli2 (R&D System, AF3635-SP)	WB 1:1000 IF 1:500
Anti-Gli3 (R&D System, AF3690)	WB 1:500 IF 1:300
Anti-SufU (Cell Signaling, C81H7)	WB 1:1000
Anti-Arl13b (Proteintech, 17711-1-AP)	IF 1:2000
HRP anti-mouse (BioLegend, 405306)	WB 1:3000
HRP anti-rabbit (BioLegend, 406401)	WB 1:3000
HRP anti-goat (Sigma, A5420)	WB 1:3000
Anti-goat Alexa 488 (Biokom, 705-545-147)	IF 1:1000
Anti-rabbit Alexa 594 (Biokom, 711-545-152)	IF 1:1000
Anti-mouse Alexa 488 (Biokom, 715-605-150)	IF 1:1000
Anti-Ki-67 (Thermo Fisher, MA5-14520)	IF 1:250
Anti-BrdU (Sigma, B8434)	IF 1:1000

Enhanced chemiluminescence detection system (Clarity and Clarity Max, Bio-rad) and Amercham Imager 800 were used to detect immunocomplexes. To estimate the molecular weight of proteins pre-stained protein marker (Bio-rad) was used.

RT-qPCR

RNA was isolated with the use of Trizol method according to manufacture’s protocol. Reverse transcription was performed with the use of High-Capacity cDNA Reverse Transcription Kit (Thermo). Real-Time qPCR analysis was performed with the use of the Real-Time 2xHS-PCR Mix Sybr B (A&A Biotechnology) on a LightCycler 480 II qPCR System (Roche) with specific primers (Table 2).

Immunofluorescence staining

NIH/3T3 cells were cultured on glass coverslips in starv media for 24 h with 200 nM SAG or 300ug/ml doxycycline. Next, fixed in 4% [w/v] paraformaldehyde solution in PBS for 15 min at room temperature, washed in PBS 3 × 10 min, permeabilized in 0.2% Triton X-100 in PBS for 15 min at room temperature, and blocked in 5% [w/v] donkey serum in 0.2% [w/v] Triton 100-X in PBS. Next, cells were incubated with primary antibodies (Table 3) diluted in blocking buffer, overnight at 4 ^oC. The next day coverslips were washed 3 × 10 min with 0.05% Triton 100-X in PBS, incubated with secondary antibody (Table 3) dissolved in blocking buffer for 1 h at room temperature, and washed again 3 × 10 min with 0.05% Triton 100-X in PBS. Then, coverslips were mounted with the fluorescent mounting medium with DAPI (ProLong Diamond, Thermo Fisher). Slides were analyzed on the fluorescent, inverted microscope Olympus IX-73 under the 63x uPLANAPO oil objective and the Photometrics Evolve 512 Delta camera.

BrdU staining

Daoy control and Hira KO cells were cultured on glass coverslips with 10 μm BrdU solution for 24 h. Next, cells were washed in PBS, fixed and permeabilized as for other stainings. Prior to blocking DNA hydrolysis was performed. Cells were incubated in 1 M HCL for 1 h, than in 0.1 M sodium borate pH8.5 for 30 min, washed 3 times in PBS. Next, immunostaining and analysis was performed as for other antibodies (Table 3).

Migration assays

30k of Daoy control and Hira KO cells were seeded onto Corning^® Transwell^® polycarbonate inserts (pore 8.0 μm) in starve medium. 500 ul of complete media was added to each well, outer the insert. After 24 h media was removed and the inner side of the inserts was cleaned with a cotton swab. Migrated cells on the outer side of the insert were washed with PBS and fixed with 4% [w/v] paraformaldehyde for 15 min at room temperature. Next, membranes were cut from the insert and mounted with the fluorescent mounting medium with DAPI (ProLong Diamond, Thermo Fisher). Migrated cells were counted on the fluorescent, inverted microscope Olympus IX-73 under the 40x uPLANAPO objective and the Photometrics Evolve 512 Delta camera.

Colony formation assay

Daoy control and Hira KO cells were seeded at a density of 300 cells/plate in complete media and left to form colonies for 1 week. After that time cells were fixed with 4% [w/v] paraformaldehyde for 15 min at room temperature, washed with PBS, and stained with crystal violet solution for 20 min at room temperature. Colonies were counted under the light microscope.

Flow cytometry analysis

Daoy control and Hira KO cells were harvested, washed in PBS and resuspended in 1x binding buffer at a density of 2 × 10⁵/ml. Annexin V staining was performed according to manufacture’s protocol with the use of eBioscience™ Annexin V-FITC Apoptosis Detection Kit (Invitrogen). Cells were analyzed on BD LSRFortessa flow cytometer.

CUT&RUN

Gli3R-HA Tet-On control, Gli3R-HA Tet-On Hira KO, and Hira-HA Gli3R TetOn cells were starved for 24 h with 300ug/ml doxycycline to induce Gli3R/Gli3R-HA expression. 200k cells were taken and subjected to cut and run protocol with the use of Cell Signaling CUT&RUN Assay Kit (#86652). Cells were immobilized on concavanilin A magnetic beads then permeabilized and anti-HA primary antibody (Table 3) was added for overnight incubation. Next day enzymatic cleavage with pAGMNase was performed and desired chromatin fragments were isolated with the Mag-Bind Total Pure NGS Kit (Omega BIO-TEK, #M1378-01).

DNA libraries were constructed using Kapa HyperPrep reagent kit (KAPA Biosciences, cat. no. KK8504) and KAPA UDI Adapters Kit 15 μm (KAPA Biosystems, cat. no. 8861919702). The manufacturer’s protocol was followed, with the following modifications: the incubation temperature during end repair (ERAT) was lowered from 65 °C to 50 °C, the concentration of adapters used in the ligation reaction was lowered to 300 nM, increased the reagent-to-sample ratio during polygenic purification to 1.1x, and the annealing and extension times during the PCR reaction were lowered to 15 s. The procedure used 25 ng each of input samples and 15 ul each of CUT&RUN samples. Nine cycles of enrichment by amplification were used. A final selection of fragment lengths in a two-step purification at a sample: reagent volume ratio: (1) stage − 1:0.55, (2) stage 1:1 (expected fragment length range 100–700 bp). Libraries were pooled equimolar and purified from adapter dimers. The resulting libraries and pools were subjected to fragment length control using an Agilent TapeStation 2200 analyzer and a set of High Sensitivity D1000 Screen Tape (Agilent, cat. no. 5067–5584) and High Sensitivity D1000 Reagents (Agilent, cat. no. 5067–5585). Library concentrations were then determined by qPCR using the Kapa Library Quantification kit (Kapa Biosciences, cat. no. KK4824).

Sequencing was performed on an Illumina NovaSeq 6000 instrument using NovaSeq 6000 S1 Reagent Kit v 1.5 (200 cycles) reagents (Illumina, cat. no. 20028318), in pair-end read mode 2 × 100 cycles using the standard procedure recommended by the manufacturer with 1% addition of the Phix control library (Illumina, cat. no. FC-110-3001).

RNAseq

Daoy control and Hira KO cells were starved for 48 h with 500 nM SAG. Next RNA was isolated with the use of Universal RNA Purification Kit (EURx). RNA 3’ mRNA-seq libraries were constructed using QuantSeq 3’ mRNA-Seq Library Prep Kit reagents FWD for Illumina (Lexogen, cat. no. 015.96) and Lexogen i7 6 nt Index Set adapters (7001–7096; Lexogen, cat. no. 044.96). The manufacturer’s protocol was followed, starting with 500 ng of supplied RNA and using 11 cycles of enrichment by amplification. The resulting 3’ mRNA-seq libraries were subjected to fragment length control using an Agilent Bioanalyzer 2100 and High Sensitivity DNA kit reagents (Agilent, cat. no. 5067 − 4626). Library concentration was then determined by qPCR using the Kapa Library Quantification kit (Kapa Biosciences, cat. no. KK4824). The manufacturer’s protocols were followed. Sequencing was carried out on an Illumina NovaSeq 6000 instrument using NovaSeq 6000 S1 Reagent Kit (200 cycles) reagents (Illumina, cat. no. 20012864), in pair-end read mode 2 × 100 cycles using the manufacturer’s recommended standard procedure with 1% addition of Phix control library (Illumina, cat. no. FC-110-3001).

Data analysis

Data analysis was performed using RStudio and Microsoft Excel. Immunofluorescence images were analyzed by the Fiji/ImageJ software. Statistical significance was calculated using Student’s t-test for experiments involving two experimental groups.

CUT&RUN raw sequencing reads were aligned using Bowtie2³⁹, peaks were called using MACS2⁴⁰, and annotated using R package ChIPseeker⁴¹. Enriched motifs were found using Homer (Heinz et al., Molecular Cell, 2010). Functional enrichement analysis and visualization were performed with the use of ChIPseeker⁴¹ and clusterProfiler⁴² R packages.

RNASeq raw data was analysed using salmon⁴³ in the unmapped mode. Differential expression was quantified using tximport⁴⁴ and DSeq2⁴⁵.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Author contributions

W.S: Designed the study, performed experiments, analyzed data, wrote the initial manuscript, established methods, B.B: Performed experiments. P. N: Designed the study, performed preliminary experiments, analyzed data, supervised the project, revised the manuscript. All the authors read and approved the final manuscript.

Data availability

The datasets generated during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.