HDAC6 modulation influences the growth rate of cancer cells.
Several cancer cell lines were treated at a range of ITF3756 from 250 nM to 5 µM. Cell growth was monitored over 72h (Fig. S1 A-D). At 48h, crystal violet staining showed a marked reduction in HC1806 triple-negative breast cancer and B16F10 melanoma cell proliferation (Fig. S1 E-F) at concentrations equal to or greater than 1 µM. HDAC6 inactivation by CRISPR-Cas9 in B16F10 cells confirmed the inhibitory effect of HDAC6 inhibition on cell proliferation. Fig. S2 A highlights a 20% growth decrease in HDAC6_KO cells compared to WT. Furthermore, the HDAC6-deprived cells exhibited a dampened and scattered growth pattern concerning WT cells and enhanced response to the 5 µM ITF3756 treatment (Fig. S2 B).
HDAC6 modulation influences lysine acetylation dynamics.
Remarkable changes in several acetylation markers were observed when HCC1806 cells were treated with 1 µM ITF3756 for 16h. Immunofluorescence staining established a 2- to 6-fold increase in the acetylation levels of H3K27, H3K9, and H3K14. In contrast, the levels of H4K16Ac decreased (Fig. 1A-B, Fig. S3 A-B). These observations were validated by capillary electrophoresis analyses, which demonstrated an enrichment in the acetylation levels of H3K4Ac post-treatment. However, H3K23Ac was not modulated (as seen in Fig. S3 C). This pattern of acetylation after ITF3756 treatment was consistent across other cell lines, including human and mouse triple-negative breast cancer cell lines MDA-MB-231 (Fig. S4 A) and 4T1 (Fig. S4 B), as well as the human leukemia T-lymphocyte Jurkat cell line (Fig. S4 C). It is worth noting that the non-tumorigenic pulmonary Wi38 fibroblasts exhibited a different acetylation pattern (Fig. S4 D). Different results emerged in the HDAC6_KO cells. Immunofluorescence staining revealed a 2-fold enhancement in the acetylation levels of H3K27, H3K9, and H3K14 in the HDAC6-inactivated cells compared to their wild-type counterparts. Meanwhile, a decline in the H4K16Ac signal was recorded (Fig. 1. C-D, Fig. S5 A-B). These findings were supported by capillary electrophoresis analyses, which showed a similar enrichment in acetylation markers, including H3K4Ac, H3K9Ac, H3K14Ac, and H3K27Ac, while H3K23Ac and H4K16Ac remained relatively stable (Fig. S5 C). When the B16F10 wild-type cells were exposed to 1 µM ITF3756 for 16h, they mirrored the acetylation profiles observed in the HDAC6-inactivated cells, with significant increases, especially in H3K4Ac, H3K9Ac, H3K14Ac, and H3K27Ac histone markers (Fig. S5 D). The effects of HDAC6 modulation are not limited to histone lysine modifications. They also extend to the acetylation of K40 on α-tubulin, a well-known target of HDAC6. Indeed, ITF3756 induced a pronounced increase in α-tubulin acetylation, as determined by both immunofluorescence and capillary electrophoresis, in several cell lines (Fig. S6 A-E).
HDAC6 inactivation impacts lysine acetylation in vivo.
We took advantage of HDAC6_KO mice to validate what was observed in vitro. Specific acetylation markers, including α-tubulin K40, H3K9, H3K14, H3K27, and H4K16, were investigated in both WT and HDAC6_KO mice by using capillary electrophoresis performed on tissue extracts from various organs, including liver, spleen, brain, and heart. There was a significant increase in K40 acetylation in liver, spleen, and heart samples. However, the same result did not occur in the brain (Fig. S6 F). In livers from HDAC6_KO mice, an increased acetylation of H3 on lysine 9 and 27 was detected (Fig. 1E). Conversely, H3K14 and H4K16 did not change in HDAC6_KO animals. (Fig. S7 A). In the spleen, acetylation of K27 on histone H3 was higher in HDAC6_KO mice (Fig. 1F), while H3K9, H3K14, and H4K16 did not change (Fig. 1F and S7 B). Neither the brain nor the heart showed any considerable modification in histone acetylation levels (Fig. S7 C-D).
HDAC6 inhibition/inactivation increases global HAT activity and elevated P300 protein levels.
To delve deeper into the effects of ITF3756, we explored whether any changes were detectable in total HDAC and total HAT activities upon HDAC6 inhibition/inactivation. HCC1806 cells treated with ITF3756 1 µM for 16h showed no alteration in total HDAC activity compared to the control (Fig. S8 A, left panel). This result indicates that other HDACs might compensate for the loss of HDAC6 function or that a reduction in HDAC6 activity is insufficient to determine appreciable changes in global HDAC activity. However, there was a remarkable increase in global HAT activity in the ITF3756 conditions compared to the control solvent (Fig. 2A). Interestingly, global levels of acetyl-CoA (AcCoA), which serves as the substrate for HATs, remained unchanged (Fig. S8 A, right panel), suggesting that the effect of HDAC6 inhibition on total HAT activity was not dependent on an altered metabolism of AcCoA. To get more insights, we evaluated whether P300, one of the most abundant HAT within cells and known to interact with HDAC6 (28), was affected by ITF3576 treatment. When we assessed the mRNA expression levels of HDAC6 and P300, we found no significant difference upon ITF3756 treatment (Fig. S8 B), hinting that ITF3756 did not transcriptionally impact HDAC6 and P300 functions. However, capillary electrophoresis experiments showed increased P300 protein levels, as depicted in Figs. 2B and 2C, whereas HDAC6 protein levels were not modulated. A similar result was observed when HDAC6 was inactivated using siRNAs, leading to a decrease in HDAC6 expression by around 70% and a 2-fold increase of P300 protein levels (Fig. 2D). Noteworthy, P300 inactivation, using siRNAs, did not change HDAC6 levels (Fig. S9 A), while treating si_P300 HCC1806 cells with ITF3756 1 µM for 16h, P300 protein levels increased, once again, significantly (Fig. 2E), leading to relatively higher levels of H3K9Ac and H3K27Ac (Fig. S9 B). Additionally, total HAT activity and P300 protein levels were notably higher in HDAC6_KO than in WT cells (Fig. 2F and G). Nevertheless, total HDAC activity, AcCoA levels, and P300 mRNA levels did not change (Fig. S8 C-D).
Comparable results were obtained in ITF3756-treated Jurkat (Fig. S8 E-F) and B16F10 cells (Fig. S8 G-H).
Of note, P300 inhibition by the small molecule EML425 (38), lead to a significant decrease in H3K9Ac and H3K27Ac levels in HCC1806 cells before and after HDAC6 inactivation by siRNAs (Fig. S9 C)
HDAC6 inhibition promotes P300 stabilization.
HEK293T cells were transfected with pcDNA3.1+-HDAC6-Flag and control pcDNA3.1+ (empty vector, EV). After 24h, cells were exposed to 1 µM ITF3756 for an additional 16h, and co-immunoprecipitation experiments, analyzed by capillary electrophoresis, were performed to assess whether a putative HDAC6/P300 complex could dissociate in the presence of ITF3576 (Fig. 3A). We found that in this cellular system, HDAC6-Flag content did not change after treatment, and the transfected protein was mainly localized in the cytoplasm in both conditions (Fig. 3B). On the contrary, there was an evident increase in P300 levels in treated samples (Fig. 3A, B and Fig. S9 D). Furthermore, the enzyme appeared to be predominantly localized in the nucleus of ITF3576-treated cells than in controls. Moreover, we detected an HDAC6/P300 complex in untreated cells, significantly reduced after ITF3756 administration.
Proximity Ligation Assays (PLA) further confirmed these findings (Fig. S9 D). Indeed, in pcDNA3.1+-HDAC6-Flag-transfected MRC5 cells, a significant interaction was observed between the endogenous P300 and transfected HDAC6-Flag. This interaction was more pronounced than in the control cells expressing EV alone. Moreover, P300 levels increased upon ITF3576 treatment in pcDNA3.1 + and pcDNA3.1+-HDAC6-Flag transfected cells, as previously observed. A decrease in PLA signal was detected upon normalization of PLA foci on P300 content in pcDNA3.1 + and pcDNA3.1+-HDAC6-Flag-transfected cells upon ITF3576 treatment. (Fig S9 D, bottom right graph). These experiments support the evidence that P300 can be associated with HDAC6, whereas HDAC6 inhibition reduces this interaction while increasing P300 protein levels.
We then investigated which mechanism could support the increase in P300 protein levels detected upon ITF3576 treatment. Since P300 might undergo polyubiquitination and HDAC6 was reported to serve a role in Ub-signaling mechanisms, we evaluated whether HDAC6 inhibition could change the rate of P300 ubiquitination. To this aim, HCC1806 cells were exposed to 1 µM ITF3756 for 16h, and immunoprecipitation with an anti-polyubiquitin antibody was performed (Fig. 3C). We observed a significant reduction in the polyubiquitinated species of P300 in the ITF3756-treated cells, compared to controls, despite the marked increase in P300 (Fig. 3C, bottom right).
To corroborate further the specificity of this observation and to rule out that P300 stabilization was not caused by global alterations of proteasome or Ub-signalling induced by ITF3756, canonical parameters of UPS activity were assayed in HCC1806 treated with 1 µM ITF3756 for 16h and compared to controls (i.e., solvent-treated).
Crude cell extracts (i.e., the soluble fraction of cytosol) were isolated under non-denaturing conditions from ITF3756- and solvent-treated cells and assayed for the rate of cleavage of the fluorogenic probe LLVY-amc by the chymotrypsin-like activity of intact proteasome particles. The spectrofluorimetric assay is commonly called proteasome assay (see Methods). The slopes of each sample were then calculated over a linear interval. The data highlighted that the rate of cleavage of LLVY-amc (expressed as µmol/min) was fully comparable between ITF3756- and solvent-treated cells (Fig. S10A), suggesting that the drug did not alter the basal proteasome activity.
To strengthen this finding, the same extracts were further analyzed by Wb. Filters were assayed for the overall content of i) representative proteasome subunits, either belonging to the 20S (i.e., PSMA7, PSMB6) and the 19S (i.e., PSMD4); ii) Ub-proteins and those conjugated with Ub-K:48 linkages, which more truthfully identify proteasome substrates (i.e., the K:48 topology of Ub-chains is the preferential configuration for substrate processing through the 26S proteasome) (Fig. S10B).
Also, in this case, the data obtained after normalization suggested that the overall content in proteasome subunits and Ub-proteins was fully comparable between ITF3756- and solvent-treated cells.
Therefore, the overall data set suggests that ITF3756 treatment, reducing HDAC6/P300 association, might inhibit P300 ubiquitination through its stabilization.
HDAC6 inactivation affects chromatin accessibility.
Transposase-accessible chromatin sequencing (ATAC-seq) assay and H3K27Ac chromatin immunoprecipitation sequencing (ChIP-seq) were performed on B16F10 HDAC6_KO and WT cell lines. Indeed, Principal Component Analysis (PCA) shows HDAC6_KO and WT cells well separated in both analyses (Fig. S11-S12 A).
ATAC-seq and H3K27Ac-ChIP-seq experiments revealed a significant number of differentially accessible regions (DARs) with a -0.5 ≤ shrunken LogFC ≤ 0.5 and p adjusted-value ≤ 0.05 (Fig. S11-S12 B). Specifically, 4835 DARs were identified from the ATAC-seq and 9650 DARs from the ChIP-seq.
For what concern the genomic features, the majority of different ATAC and H3K27Ac peaks between HDAC6_KO and WT samples occurred at introns and distal intergenic regions (Fig. S11-S12 C), red and green bars, respectively). The GO analysis revealed putative targets of HDAC6 inactivation (Fig. S11-S12 D). These analyses highlighted a different regulation of genes involved in cell proliferation (GO-ID: 0008283), cell adhesion (GO-ID: 0007155), and migration (GO-ID: 0016477), which were down-modulated due to inaccessible chromatin (Fig. S11-S12 D, left panel, blue bars). Meanwhile, genes with a role in apoptotic pathways (GO-ID: 0043065) were up-regulated due to increased chromatin accessibility (right graphs, red bars, in Fig. S11-S12 D). Specifically, by combining the HDAC6-associated DARs, identified by ATAC-seq and ChIP-seq analyses, two protein networks related to down- and up-regulated regions, respectively, were generated, pointing out a series of genes associated with an altered survival program including Akt1, Itgb3, Gas6, Sox9, Nf1, Tgfb2, and Casp7 (Fig. 4A and B).
Similar results were obtained by analyzing Jurkat cells treated with 1 µM ITF3756 for 16h compared to solvent controls by ATAC-seq and H3K27ac-ChIP-seq. The PCA showed that the two samples were well separated (Fig. S13-S14 A); the ATAC-seq revealed 1399 DARs and the ChIP-seq 11450 DARs with a -0.5 ≤ shrunken LogFC ≤ 0.5 and p adjusted-value ≤ 0.05 (Fig. S13-S14 B). Again, the most affected genomic features were the introns and the distal intergenic regions (Fig. S13-S14 C, red and green bars, respectively). Once more, the GO analyses showed a different regulation in the biological processes related to cell proliferation, cell adhesion, and apoptosis (Fig. S13-S14 D).
The effect of HDAC6 inactivation in B16F10 cells has been further assessed by comparing the KO_HDAC6 and the WT condition by transcriptome analysis. Genes with − 1 ≤ Log2FC ≤ 1 and p adjusted-value ≤ 0.1 were considered differentially expressed (DE) and retained for further analysis. Through DESeq2 analysis, it was possible to identify 554 up- and 778 down-regulated genes in the KO_HDAC6 condition. The PCA analysis distinguishes the two conditions (Fig. S15 A). The volcano plot highlighted all statistically significant DE genes in the two subgroups; the expression difference is considered significant for a -1 ≤ Log2FC ≤ 1 and p adjusted-value ≤ 0.1. Red dots represent significantly up- and downregulated genes with 1 ≤ Log2FC ≤ 1 (x-axis) and a p adjusted-value ≤ 0.1 (y-axis; Fig. S15 B). The 60 most modulated genes were selected and represented in the heatmap, in which the cluster genes, separated according to their degree of modulation, could be observed. KO_HDAC6 samples appeared to have different gene expression profiles than WT controls (Fig. S15 C). To further explore the mechanisms underlying the differences observed between the two groups, GO term enrichment analysis for DE genes DE_KO_HDAC6_vs_WT was performed for both up-regulated and down-regulated genes. In the dot-plot graph, results are represented according to the criterion for which the red dots are the most significantly differentially regulated; while the point size indicates the number of genes involved in the process, the dot position represents the expression fold change. The biological processes, the cell components, the molecular functions, and the KEGG pathways analyses of KO_HDAC6 samples revealed a significant modulation in genes correlated with the extracellular matrix, cell adhesion, cell motility, regulation of protein stability, channel activity, signal transduction, and receptor complexes (Fig. S15 D-G).
Of note, the change in the expression of some of the genes associated with an altered survival program (Casp7, Sox9, Smad 3, Itgb3, Nf1, and Tgfb2) was validated by RT-qPCR (Fig. S15 H).
Furthermore, proteomic analyses were performed on HCC1806 scramble and HDAC6-silenced cells. Among the 3010 proteins identified (Fig. 16A, B), the gene ontology analysis of upregulated or downregulated proteins between HDAC6-silenced and scrambled cells revealed a significant modulation (Log2FC = 0.58 and p adjusted-value ≤ 0.05 (Benjamini-Hochberg correction). Specifically, also in this condition, from the GO analysis, we observed modulation in proteins related to the process of cell proliferation (GO-ID:0008283), cell adhesion(e.g. GO-ID: 0007155), cell migration(GO-ID: 0016477), extracellular matrix organization (GO-ID: 0030198), channel activity and integrin binding (e.g. GO-ID: 0005178, Fig. S16 C-H) confirming the data provided by the transcriptomic analyses on HDAC6_KO B16F10 cells.