Selective molecular inhibition of the HDAC6 ZnF-UBP binding domain impairs multiple myeloma cell function.

Introduction

Multiple myeloma (MM) is an incurable plasma cell malignancy accounting for about 10% of hematological malignancies. The uncontrolled growth of monoclonal plasma cells in the bone marrow leads to complications such as anemia and destructive bone lesions, followed by hypercalcemia and renal failure [ 1 ]. Considered a late-onset disease, MM has a median age of about 70 years at diagnosis and a global incidence of 4.5–6 per 100,000 per year, which is continuously increasing due to population growth and aging [ 2 ]. During the last decades, overall survival and quality of life of MM patients has notably improved following the introduction of proteasome inhibitors (e.g., bortezomib, ixazomib and carfilzomib), monoclonal antibodies against MM cell surface antigens (e.g., daratumumab and elotuzumab), immunomodulatory agents (e.g., thalidomide, lenalidomide and pomalidomide) and autologous hematopoietic stem cell transplantation, when applicable [ 3 ]. However, nearly all patients relapse and become refractory to treatment after a few years of initial therapy, and treatment regimens have to be adapted [ 4 ].

A few years ago, histone deacetylase (HDAC) inhibitors emerged as a possible new therapeutic strategy against MM. HDACs are essential epigenetic regulators in the cell and they are involved in the maintenance of gene expression, chromatin stability, and cell homeostasis. Dysregulation of HDACs is involved in a large variety of cancers affecting crucial processes such as cell proliferation, cell differentiation, angiogenesis, autophagy, and apoptosis [ 5 ]. Previous studies have shown that some HDACs are overexpressed in MM cell lines, MM primary cells from patients, and MM bortezomib-refractory models when compared to their normal plasma cell counterparts and that they are indicators of poor prognosis [ 6 , 7 ]. The pan-HDAC inhibitor panobinostat showed promising overall response rates in combination with bortezomib and dexamethasone for bortezomib-refractory MM patients in the PANORAMA 1 clinical trial, which led to its FDA accelerated approval in 2015 [ 8 ]. However, panobinostat treatment leads to important adverse events, such as diarrhea, fatigue, thrombocytopenia, and anemia, and it was withdrawn from the market by the FDA in 2022 due to the lack of the required post-approval data [ 9 ]. Isoform-selective inhibition of HDAC6 has been proposed as a therapeutic strategy against MM with a better safety profile than pan-HDAC inhibitors. HDAC6 acts mainly in the cytoplasm due to its nuclear export signal (NES) and its Ser-Glu tetradecapeptide repeat domain (SE14) for cytoplasmic stabilization [ 10 ]. It contains two catalytic domains, CD1 and CD2, which deacetylate client proteins such as α-tubulin and Hsp90, among others [ 11 ]. Its cytoplasmic deacetylase role is thought to be involved in important processes such as microtubule regulation and chaperone-induced stress response [ 12 ]. In addition, HDAC6 contains a C-terminal zinc finger ubiquitin-binding protein (ZnF-UBP) domain, which can bind to ubiquitinated motifs. Therefore, HDAC6 can recognize the unanchored C-terminal diglycine motifs of ubiquitin that are generated in the protein aggregates by deubiquitinase ataxin-3 [ 13 ]. This leads to the transport of ubiquitinated misfolded proteins via the microtubule network to the aggresome for further degradation. Given this, HDAC6 has been proposed as a main actor for aggresome formation and autophagy activation, which are necessary for the elimination of cytotoxic protein aggregates and regulation of cell death [ 14 ]. This mechanism of protein clearance is thought to be responsible for the resistance to proteasome inhibitors in MM patients, and preclinical studies have shown that HDAC6 inhibitors induced synergistic anti-tumor activity with bortezomib in bortezomib-resistant cell models [ 15 , 16 ]. Moreover, phase I/II clinical trials showed that the HDAC6 inhibitor ricolinostat was better tolerated than panobinostat and that its combination with bortezomib displayed synergistic effects [ 17 , 18 ]. However, it still caused noticeable adverse events that could be explained by the destabilization of the microtubule network. Like other HDAC inhibitors, ricolinostat is a hydroxamate-based small molecule designed to bind the catalytic CD2 domain of HDAC6 and, therefore, it does not target the ZnF-UBP binding domain [ 19 ]. The available X-ray crystallographic structures of the HDAC6 ZnF-UBP binding domain prompted researchers to design inhibitors that could mimic the terminal glycine-glycine repeat of ubiquitin (GG) that binds to the ZnF-UBP cavity. In their work, Ferreira de Freitas et al. produced a series of propanoic acids bearing various aryle or biaryle groups, with one of the best compounds being a quinazolinylpropanoic acid [ 20 ]. This compound was active at the micromolar level, and X-ray crystallographic data (PDB number: 6CED) revealed how the propanoic chain mimicked the ubiquitin GG repeat in the binding domain. More recently, the same group achieved the synthesis of another inhibitor based on the same scaffold, and X-ray data (PDB number: 8G45) showed that a conveniently selected N 3 -substituent could improve the docking capability [ 21 ].

In the current study, the molecular and chemical inhibition of the HDAC6 ZnF-UBP binding domain was explored in an MM cell line model to understand the role this domain plays in MM pathogenesis. Its specific inhibition may provide a rationale for a new approach to block the aggresome pathway – alone or in combination with proteasome inhibitors – that may limit the side effects that pan-HDAC or HDAC6 inhibitors induce in patients.

Results

R1155A-Y1156A mutation in the HDAC6 ZnF-UBP binding domain impaired its interaction with ubiquitin in a cell-free assay

A two-residue mutation was proposed to inhibit the binding of ubiquitin to the HDAC6 ZnF-UBP domain. Previous crystallography data of the human HDAC6 ZnF-UBP binding domain bound to the ubiquitin C-terminal pentapeptide RLRGG showed that residues R1155 and Y1156 were essential to anchor these last amino acids of ubiquitin to the ZnF-UBP cavity [ 22 ]. The polar side chains of R1155 and Y1156 were oriented towards the inside of the cavity and established hydrogen bonds with RLRGG (Fig. 1 ).

Therefore, a ZnF-UBP peptide (HDAC6 1109 – 1215 ), in which residues R1155 and Y1156 were transformed into alanines, was generated through site-directed mutagenesis and peptide purification. The original backbone of the peptide was maintained, and the polar side chains where RLRGG pentapeptide was able to interact were removed. Peptide expression was confirmed through SDS-polyacrylamide gel electrophoresis (SDS-PAGE). The binding affinity of the wild-type (ZnF-UBP WT ) and mutated (ZnF-UBP RY ) peptides to the RLRGG FITC-labeled ubiquitin pentapeptide was assessed using a fluorescence polarization assay. The RLRGG pentapeptide was incubated at 50 nM with increasing concentrations of the ZnF-UBP WT and ZnF-UBP RY peptides, and results showed that only the ZnF-UBP WT peptide could interact with RLRGG (Fig. 2 ). ZnF-UBP RY did not interact with the pentapeptide, suggesting that the residues R1155 and Y1156 of HDAC6 are essential for its ubiquitin-binding capacity.

R1155A-Y1156A mutation in the HDAC6 ZnF-UBP binding domain altered its surface electrostatic charge potential

To rationalize the structural role of the R1155A-Y1156A mutation, molecular dynamics simulations of the ZnF-UBP WT peptide (apo), in complex with RLRGG and the ZnF-UBP RY peptide, were carried out. The backbone conformation of the apo ZnF-UBP WT and the RLRGG pentapeptide complex is well conserved in all simulations. The side chains of residues R1155 and Y1156 are highly dynamic and act as gatekeepers to the ubiquitin binding site. This is consistent with previously reported studies [ 13 , 20 ]. The principal components analysis identified that mutating R1155 and Y1156 side chains have a negligible effect on the overall dynamics of the ubiquitin binding site (Fig. 3A ). This observation did not explain why the ubiquitin pentapeptide was unable to bind to the ZnF-UBP RY peptide. Peptide impairment was then investigated by running Poisson-Boltzmann solvation calculations. These calculations revealed a significant decrease in the electrostatic potential, suggesting a substantial reduction in charged and polar interactions that are crucial for the stability of the ZnF-UBP WT -RLRGG complex. The change can be visually appreciated in Fig. 3B , which highlights the electrostatic charged surfaces of the ubiquitin pentapeptide and the complementary charged surfaces of the ZnF-UBP WT and the ZnF-UBP RY mutant. A clear diminished electrostatic charge complementarity between the pentapeptide and the ZnF-UBP RY mutant was observed. Furthermore, the R1155A-Y1156A mutation also led to the loss of hydrogen bonds between the mutated side chains and the backbone/side chain of the pentapeptide, further undermining the structural integrity of the interaction. Following the analysis of the atomistic interactions between the pentapeptide and ZnF-UBP showed an altered contact frequency of residues of the pentapeptide when comparing the wild-type and the mutated forms (Fig. 3C ).

Molecular inhibition of the HDAC6 ZnF-UBP binding domain did not impair HDAC6 expression or its catalytic activity but impaired aggresome formation in RPMI 8226 cells

The evidence of ZnF-UBP molecular inhibition through a two-residue mutation was the rationale for generating an MM cell line with a non-functional HDAC6 ZnF-UBP binding domain. Two RPMI 8226 cell lines were generated through CRISPR-Cas9 technology: a full HDAC6 knockout cell line (HDAC6 KO ) and a ZnF-UBP-mutated cell line (HDAC6 RY ). After delivery of the Cas9 ribonucleoprotein (RNP) complex through electroporation and subsequent clonal expansion, the efficiency of the CRISPR-Cas9 edition was verified at the DNA level for clone selection. Tracking of indels by decomposition (TIDE) analysis showed that the HDAC6 KO RPMI 8226 cells successfully acquired indels of various lengths in exon 3, a modification that is normally deleterious for protein expression due to reading frame shifting. Sequencing of the genomic region encoding for ZnF-UBP binding domain showed that reference codons CGT TAC (R1155, Y1156) were mutated into GCT GCC (A1155, A1156) in HDAC6 RY RPMI 8226 cells.

At the protein level, HDAC6 KO cells did not express HDAC6 and consistently, acetylated tubulin expression increased with the loss of HDAC6, as it is one of its main substrates [ 23 ]. HDAC6 RY cells showed a similar HDAC6 expression than wild-type cells despite the change in residues R1155 and Y1156 into nonpolar alanines. No changes in tubulin acetylation were observed, suggesting that this two-residue mutation in the ZnF-UBP binding domain did not affect HDAC6 expression or tubulin deacetylase activity (Fig. 4A ). Total HDAC catalytic activity was also measured using a UHPLC-MS method in which the ratio between the deacetylated and acetylated MAL, a pan-HDAC substrate, was measured. The total HDAC catalytic activity was maintained in HDAC6 RY cells compared to the wild-type cells, while a reduction in the deacetylation of MAL substrate was observed in HDAC6 KO , which is consistent with the loss of HDAC6 expression (Fig. 4B ). The aggresome formation was then measured in the different RPMI 8226 cell lines to assess the role that HDAC6 plays in this process. Cells were stimulated with the proteasome inhibitor MG-132 (5 µM) for 18 h to induce aggresome formation in response to proteasome blockade. A decreasing trend in aggresome activation was observed in HDAC6 KO cells compared to wild-type cells, even if they were still able to activate the aggresome pathway to some extent. However, there was no statistically significant increase in aggresome formation after MG-132 treatment in HDAC6 RY cells (Fig. 4C ). This indicates that a non-functional HDAC6 ZnF-UBP binding domain may affect aggresome formation.

Proliferation of HDAC6 KO and HDAC6 RY RPMI 8226 cells was reduced, and cell cycle was dysregulated compared to wild-type cells

Cell growth and cell cycle were assessed in the different RPMI 8226 cell lines to determine whether the loss or mutation of HDAC6 could impact cell proliferation. Cell growth was decreased in HDAC6 KO and HDAC6 RY cells compared to wild-type cells with statistically significant differences after 72 h (Fig. 5A ). In agreement with these findings, significant changes in the cell cycle were observed, with a decrease of cells in the G0/G1 phase and an increase of cells in the S and G2/M phases in HDAC6 KO and HDAC6 RY cells compared to wild-type cells (Fig. 5B ). These results suggest that factors such as cell cycle arrest or extended cell cycle phase durations may affect the progression of the cell cycle in HDAC6 KO and HDAC6 RY cells. Despite the differences in cell proliferation, bortezomib did not display different antiproliferative activity between the three cell lines (data not shown), which suggests that a selective inhibition of the ZnF-UBP domain may be more relevant in bortezomib-resistant cells in which the aggresome pathway is upregulated.

Chemical synthesis of quinazolinones 1a-g as potential HDAC6 ZnF-UBP inhibitors

Results obtained with the engineered cell lines suggested that a specific inhibition of the HDAC6 ZnF-UBP binding domain might be a valid target for MM therapy. Therefore, selective ZnF-UBP inhibitors with no affinity for the catalytic domains were designed and synthesized. Alternative substitutions of the quinazolinone scaffold were developed, focusing on the benzene part of the quinazolinone. The synthesis of the substituted quinazolinones 1a - g was based on the procedure published by Ferreira de Freitas et al. for unsubstituted quinazolinone ( 1a , R = H) (Scheme 1 ) [ 20 ].

The preparation of compounds 4a - g was optimized compared to the previous two-step procedure with a more sustainable single-step preparation under microwave conditions using pinane as solvent [ 24 ]. Then, an esterification led to compounds 5a - g , and the subsequent N-methylation afforded compounds 6a - g in variable yields. In some cases, an alternative approach was found more productive by performing in one step the esterification as methyl ester and the N-methylation using 3 eq. of CH 3 I and K 2 CO 3 . A final ester hydrolysis gave the expected quinazolinylpropanoic acids 1a - g . All these new compounds were characterized by NMR and exact mass, and all tested compounds were 95% or more pure as determined by HPLC.

Molecular docking of quinazolinones 1a-g

The structural binding of compounds 1a - g to the HDAC6 ZnF-UBP binding domain was assessed using automated docking. Previous structural studies by Ferreira de Freitas et al. (PDB number: 6CE6) were used as a starting point for chemical exploration around the ZnF-UBP binding site [ 20 ]. All compounds ( 1a - g ) docked in a similar orientation, sandwiched between R1155 and W1182. The pyrimidine ring in the quinazolinone made π-π stacking interactions with W1182, and the carbonyl group was oriented towards Y1189 (Fig. 6 ). The propanoic acid tail made hydrogen bonding interactions with the side chains of R1155 and Y1184 and the backbone nitrogen of G1154. Compound 1f was also able to make a hydrogen bond with the side chain of Q1187. The RTCNN scores, used to rank the compounds, ranged between -31.33 and -36.66 (Table 1 ).

ZnF-UBP inhibitor candidates 1a-g disrupted the HDAC6 ZnF-UBP interaction with ubiquitin in a cell-free assay, but did not show significant effects in RPMI 8226 cells

Compounds 1a - g were tested for their ubiquitin-displacing activity towards the HDAC6 ZnF-UBP binding domain. Compound 1a exerted a ubiquitin-displacing activity similar to what was previously published by Ferreira de Freitas et al. [ 20 ]. Compounds 1b, c, e and g displayed a slightly improved ubiquitin-displacing activity (Table 1 ). These results were consistent with RTCNN docking scores, which allowed to identify the strongest binders. The HDAC6 inhibitors ricolinostat, tubacin and panobinostat did not exert any ubiquitin-displacing activity at 20 µM, showing their specificity towards the catalytic domains of HDAC6 and validating this assay as a useful tool to assess the specific ZnF-UBP domain inhibitory capacity of small molecules.

Then, the aggresome formation was measured in RPMI 8226 cells treated with compound 1b , as it showed the strongest ubiquitin-displacing activity. Cells were treated with 1b for 24 h with concentrations up to 100 µM prior to adding the proteasome inhibitor (MG-132, 5 µM). No significant differences were observed between the treated and control samples, suggesting that the chemical approach to inhibit the aggresome pathway did not lead to comparable effects than what was obtained with the CRISPR-Cas9-edited HDAC6 RY RPMI 8226 cells (data not shown). Moreover, compounds 1a - g did not exert any anti-proliferative activity or deregulation of the cell cycle in RPMI 8226 cells up to 100 µM (data not shown). Furthermore, a 24 h treatment with 1b did not induce the expression of cell stress-related proteins (data not shown). Finally, combination treatment of 1b with bortezomib did not display any synergistic effect in RPMI 8226 cells (data not shown). Taken together, these results indicate that the synthetized compounds were able to bind to the ZnF-UBP domain and displace ubiquitin, but no effect could be identified in RPMI 8226 cells.

Non-functional HDAC6 ZnF-UBP binding led to a decrease in signaling pathways involved in cell adhesion and in the immunological function

To gain insight into pathways affected by the absence of HDAC6 or a non-functional HDAC6 ZnF-UBP binding domain, bulk RNA-seq analysis of the three RPMI 8226 cell lines was performed. Out of 21,394 genes analyzed, 3498 transcripts (16.4%) corresponded to differentially expressed genes (DEG) in HDAC6 KO cells compared to wild-type cells, among which 2237 (10.5%) were downregulated and 1261 (5.9%) were upregulated. Greater changes were observed in the HDAC6 RY cells, where there were 14,334 (67.0%) DEGs compared to wild-type cells, among which 7506 (35.1%) were downregulated and 6828 (31.9%) were upregulated. This shows that the absence of HDAC6 protein led to less changes in gene expression than the two-residue mutation of its ZnF-UBP binding domain (Fig. 7A ). Gene ontology (GO) enrichment analysis showed that in both HDAC6 KO and HDAC6 RY cell lines, pathways linked to cell-cell adhesion and to immune responses were among the most significantly dysregulated processes (Fig. 7B ). Interestingly, the number of DEGs involved in these processes was higher in HDAC6 RY cells than in HDAC6 KO cells, and the fold changes were also globally higher in HDAC6 RY cells. For example, the transcripts of the three cell surface proteins desmoglein-2 ( DSG2 ), N-cadherin ( CDH2 ), and ICAM-1 ( ICAM1 ) were significantly down-regulated in the HDAC6 RY cell line compared to wild-type cells, but not in the HDAC6 KO line. The same trend was observed for light chain immunoglobulin genes ( IGLV2-14 , IGLC3 , IGLC2 ) and immune receptors and transcription factors ( TLR7 , TLR4 , IRF4 ) (Fig. 7C ).

To assess whether HDAC6 KO and HDAC6 RY cells showed differences in immune function compared to wild-type cells, we measured by ELISA the amount of lambda light chain immunoglobulins secreted by the three RPMI 8226 cell lines but observed no differences between the lines (data not shown). Further, we stimulated all three cell lines with lipopolysaccharide (LPS), a ligand for the innate immune receptor TLR4, and measured the production of the cytokines IL-6 and IL-10 by ELISA. There was no difference in viability between unstimulated cells and cells stimulated by LPS. While LPS induced high amounts of IL-6 in wild-type cells, production was significantly decreased in HDAC6 RY cells, and it was completely absent in HDAC6 KO cells (Fig. 7D ). These results suggest that both the catalytic domains and the ZnF-UBP binding domain of HDAC6 may cooperate to support IL-6 production. In contrast, IL-10 production was decreased in a similar manner in HDAC6 KO and HDAC6 RY cells compared to wild-type cells (Fig. 7E ), suggesting that the ZnF-UBP binding domain, rather than the catalytic domains, is involved in the IL-10 inducing function of HDAC6.

Discussion

HDAC6 inhibition is currently proposed as a possible therapeutic approach for MM. The versatile and essential role that HDAC6 displays in the cell cytoplasm through its multi-domain structure makes it a promising target for drug development pipelines. Despite the extensive studies focusing on its deacetylase activity towards multiple protein substrates, the knowledge about the role that HDAC6 plays in MM progression is still limited.

In this research article, the role of the HDAC6 ZnF-UBP binding domain was explored as a possible target in MM. To do so, the effect of a two-residue mutation in the ZnF-UBP binding domain was shown in a cell-free assay and in silico to be a potential disruptor of the HDAC6-ubiquitin interaction. Then, an MM cell line containing this mutation in the ZnF-UBP domain of HDAC6 was developed to obtain insights into the role this domain plays in MM pathogenesis. Impairment of the ubiquitin-binding capacity of the HDAC6 ZnF-UBP domain caused a decrease in aggresome formation in RPMI 8226 cells, indicating a direct role of HDAC6 in this pathway through this domain. These results bring new evidence on the role that the ZnF-UBP binding domain plays in a cancer cell model. However, partial aggresome activation was still observed in HDAC6 KO cells, suggesting that this process might not exclusively depend on HDAC6, but also on other described aggresome activator proteins able to recognize ubiquitin, such as p62 or USP10 [ 25 , 26 ].

In addition to aggresome inhibition, slower cell growth and an increase in the S and G2/M phases of the cell cycle were observed. Anti-cancer drugs such as kinesin-5 inhibitor BRD9876 or Aurora A inhibitor ENMD-2076 have shown anti-proliferative effects in MM cell models and a cell arrest in the G2/M phase [ 27 , 28 ]. These findings suggest that the HDAC6-ubiquitin interaction could also play a role in the regulation of MM cell division and that this inhibition could induce similar effects than the inhibition of mitotic checkpoints. Aggresome formation, cell growth, and the cell cycle were surprisingly more affected in cells with a non-functional HDAC6 ZnF-UBP binding domain than in cells with a full HDAC6 knockout. Moreover, higher transcriptional changes were observed in HDAC6 RY cells than in HDAC6 KO cells. The impairment of the ubiquitin-binding capacity of HDAC6 due to the mutation of the ZnF-UBP domain may, therefore negatively interfere with the aggresome formation induced by other proteins and impair the ubiquitination process, hence negatively interfering with the aggresome formation process. This leads to the hypothesis that an inactivating mutation of the ZnF-UBP domain acted in a dominant-negative fashion, leading to a stronger aggresome inhibition than when HDAC6 was entirely missing. Based on this assumption, which still has to be confirmed, a specific inhibition of the ZnF-UBP binding domain, allowing HDAC6 catalytic activity to be unaffected, may be a suitable approach against MM.

At the transcriptional level, a gene ontology analysis indicated that the HDAC6 ZnF-UBP domain may be related to the regulation of cell adhesion molecules. Integrins, cadherins, selectins and syndecan-1/CD138 have previously been shown to play a role in the adhesion and communication between MM and bone marrow stromal cells (BMSC) [ 29 ]. In this regard, some anti-MM treatments designed to target cell adhesion molecules are being developed. Daratumumab, an anti-CD38 antibody, was approved in 2016 by the FDA for MM treatment, as a single agent or in combination with bortezomib or lenalidomide [ 30 ]. Natalizumab, an anti-integrin subunit α4 antibody, inhibited the adhesion of MM cells to non-cellular and cellular components of the microenvironment and reduced tumor growth, VEGF secretion and angiogenesis in an immunodeficient murine model [ 31 ]. Therefore, it will be interesting to further study adhesion and migration to characterize the impact that a non-functional ZnF-UBP domain has in MM cells.

Moreover, a negative regulation of the immunological function was suggested at the transcriptomic level for HDAC6 KO and HDAC6 RY cells. At the protein level, the production of IL-10 induced by TLR4 activation was decreased in both HDAC6 KO and HDAC6 RY cell lines. Previous publications showed that HDAC6 deletion could inhibit IL-10 production in macrophages and in antigen-presenting cells [ 32 , 33 ]. Similarly, LPS-induced IL-6 production was absent with the loss of HDAC6. This finding is in agreement with a previous study, in which ricolinostat induced an anti-inflammatory response in LPS-stimulated macrophages through the suppression of the TLR4-MAPK/NF-κB pathways, which are essential for the induction of IL-6 [ 34 ]. Although the association between HDAC6 and immunological functions was already described in myeloid cells, this is to our knowledge the first time it is shown in MM cells. IL-6 reduction may be of clinical value, as IL-6 was reported to play a central role in MM progression by mediating the communication between MM cells, osteoblasts and stromal cells in the bone marrow [ 35 ]. Interestingly, the production of IL-6 was reduced in the HDAC6 RY cells, but to a lesser extent than in HDAC6-deficient cells. This suggests an additive activity between the ZnF-UBP binding domain and the catalytic domains of HDAC6 for the induction of IL-6. Taken together, these findings demonstrate that the catalytic and ZnF-UBP domains of HDAC6 may have their own independent functions for certain cellular processes – such as tubulin deacetylation or aggresome formation –, but they may also cooperate for others, such as those involved in immune responses. In this regard, it would be of interest to generate RPMI 8226 cell lines harboring inactivating mutations of the catalytic domains CD1 and CD2 to assess if the aggresome formation would be impacted, as well as to explore the downstream effect of each mutation on the cells.

The strong impact of a non-functional HDAC6 ZnF-UBP binding domain on several biological processes in MM cells highlights its potential as a drug target. Substituted quinazolinones were synthetized to mimic the ubiquitin glycine-glycine motif bound to the ZnF-UBP domain. Despite their ubiquitin-displacing activity and their predicted binding in silico, the lack of aggresome inhibitory activity of these first designed quinazolinones showed that the cellular context must be considered for an efficient drug design. Besides displaying cell membrane permeability, HDAC6 ZnF-UBP inhibitors should have higher affinity in order to compete with abundant ubiquitin substrate inside the cell. Moreover, the versatile interactions that HDAC6 establishes with other partner proteins, such as the Hsp90-HSF1 complex or cortactin could impede an optimal accessibility or conformation for ZnF-UBP domain druggability [ 36 ]. Cell-effective ZnF-UBP inhibitors will need to be tested to validate the effects that were observed in cells containing an inactivating mutation of the ZnF-UBP domain. Possible future improvements to obtain a pharmacological effect in cells could consist in the esterification of the carboxylate tail of the quinazolinones, which would lead to prodrugs. These prodrugs would have enhanced cell permeability, and the active form could be released intracellularly. Also, the addition of polar groups, such as amides, or lipophilic moieties, such as alkyl chains, could help to obtain small molecules with an optimized lipophilicity that would increase cell permeability and reduce binding energy towards the ZnF-UBP cavity. For example, the addition of a methyl substituent to compound 1b enhanced the interaction with residue L1162 and improved the in silico RTCNN score (data not shown).

In conclusion, the strong impact of a non-functional HDAC6 ZnF-UBP binding domain on several biological processes in MM cells highlights its potential as a drug target. To further explore the role of this domain in MM pathogenesis and in the resistance to conventional therapies, it will be important to translate this mutation to other bortezomib-sensitive and resistant MM cell lines. These new cell models will hopefully reinforce the current findings and will lead to a broader characterization of the impact of a ZnF-UBP inhibition alone and in combination with proteasome inhibitors. Newly designed ZnF-UBP inhibitors may be of interest to address some of the challenges in MM treatment and to propose an innovative strategy against MM.

Material and Methods

Site-directed mutagenesis, ZnF-UBP WT and ZnF-UBP RY peptide expression and purification

The plasmid encoding for residues 1190 – 1215 of HDAC6 wild-type ZnF-UBP domain (ZnF-UBP WT ) was obtained from Addgene (Watertown, MA, USA; plasmid #25297). The mutated plasmid encoding for R1155A-Y1156A mutant ZnF-UBP domain (ZnF-UBP RY ) was generated by site-directed mutagenesis using the wild-type plasmid as DNA template for PCR (forward primer: 5’ – AGGTCTACTGTGGT GC T GC CATCAATGGCCACATGCTCCAACACCATGGA – 3’; reverse primer: 3’ – TCCAGATGACACCA CG A CG GTAGTTACCGGTGTACGAGGTTGTGGTACCT – 5’). PCR products were incubated with 20 U/µL DpnI restriction enzyme (ThermoFisher, Waltham, MA, USA) for 1 h to digest parental non-mutated DNA. TOP10 chemically competent E. coli cells (ThermoFisher) were transformed, and plasmid DNA was purified with NucleoSpin Plasmid Mini kit (Macherey-Nagel, Düren, Germany) according to the manufacturer instructions. Sanger sequencing of plasmid DNA was performed to verify the nucleotide substitutions (Microsynth, Balgach, Switzerland).

The peptides were then obtained as follows. Wild-type and mutated plasmids were transformed into chemically competent BL21 Star (DE3) E. coli cells (Invitrogen, Waltham, MA, USA), which were then grown at 37 °C under agitation (140 rpm) in 2xTY medium supplemented with 50 µg/mL kanamycin until OD 600 = 0.8. After that, the culture medium was cooled down to 12 °C, and 1 mM isopropyl ß-D-1-thiogalactopyranoside (IPTG, Sigma-Aldrich, St. Louis, MO, USA) was added to induce protein expression. After 48 h of incubation, bacteria were harvested by centrifugation at 5000 rpm for 30 min at 4 °C. The supernatant was discarded, and the pellet was resuspended in buffer (20 mM Tris-HCl (pH 8), 500 mM NaCl, 20 mM imidazole (pH 8), 10% glycerol, 10 mM β-mercaptoethanol, 1 mM PMSF). Bacterial suspensions were physically lysed by liquid homogenization using a French press (ThermoFisher) and then centrifuged at 15,000 g for 30 min at 4 °C. Lysates were passed through a 0.2 µm filter to remove cell debris and loaded on a HisTrap column (Cytiva, Marlborough, MA, USA) for peptide purification. Purified protein samples were fractionated by FPLC with maximal elution levels at 150 mM of imidazole. The purity of eluted fractions was assessed by 20% SDS-polyacrylamide gel electrophoresis and Coomassie detection.

Fluorescence polarization assay

ZnF-UBP WT and ZnF-UBP RY peptides were serially diluted 1:2 from a starting concentration of 10 µM in a total volume of 60 µL of assay buffer (10 mM HEPES (pH 7.4), 150 mM NaCl) in a 96-well black polystyrene half-area plate (Corning, Corning, NY, USA). Then, the FITC-labeled ubiquitin pentapeptide RLRGG (GenScript, Piscataway, NJ, USA) was added to a final working concentration of 50 nM. Plates were incubated for 5 min under agitation (230 rpm) at RT. Fluorescence polarization was measured on a CLARIOstar plate reader (BMG Labtech, Ortenberg, Germany) at excitation and emission wavelengths of 482 and 530 nm, respectively. When compounds were tested, they were serially diluted 1:2 in assay buffer from a starting concentration of 12.5 µM, and wild-type HDAC6 ZnF-UBP domain was added at a working concentration of 2.5 µM, corresponding to approximately 80% of its maximum FP, together with the FITC-labeled ubiquitin pentapeptide RLRGG at 50 nM.

Molecular dynamic simulations

The crystal structure of human HDAC6 zinc finger domain in complex with ubiquitin C-terminal peptide RLRGG (PDB number: 3GV4) served as the template for all molecular dynamic simulations. The proposed R1155A-Y1156A mutation was introduced using the mutagenesis function of Molsoft ICM ( http://www.molsoft.com/ ). Both the wild-type and mutant proteins were simulated in their apo and complexed states for 500 ns. Each system was simulated as triplicates. Parameters for protein atoms were obtained from the standard AMBER FF14SB libraries, while customized zinc parameters from the Zinc AMBER Force Field (ZAFF) were employed to accurately represent the zinc ion coordination environment and ion interactions [ 37 , 38 ]. Each system was solvated in a TIP3P water box with a 10 Å buffer from the protein and neutralized with 150 mM K + and Cl − [ 39 ]. Systems underwent an initial 500-step steepest descent energy minimization, with constraint applied on the ligand peptide to allow its side chains to relax and form the stabilizing interaction.

Simulations were performed using the ACEMD package starting with 5 ns equilibration [ 40 ]. Systems were then heated from 0 K to 300 K over 100 ps and equilibrated at 300 K and 1 atm for 1 ns using the NPT ensemble. Production runs extended for 500 ns at a timestep of 4 fs employing hydrogen mass partitioning. The coordinates were saved every 100 ps. Long-range electrostatics were computed with Particle Mesh Ewald summation, and the van der Waals interactions were cut off at 9 Å. Analyses focused on conformational changes were conducted using MDANALYSIS and custom scripts [ 41 ]. The Poisson-Boltzmann electrostatic calculations were carried out using APBS biomolecular solvation software suite version 3.4.1, and the figures were generated using UCSF ChimeraX [ 42 , 43 ].

CRISPR-Cas9 edition

The human multiple myeloma RPMI 8226 cells (ATCC CCL-155, Manassas, VA, USA) were cultured in complete Roswell Park Memorial Institute (RPMI) 1640 medium (Gibco, Waltham, MA, USA), supplemented with 10% fetal bovine serum (Biowest, Nuaillé, France), 50 U/mL penicillin, 50 µg/mL streptomycin (Gibco), 2 mM L-glutamine (Gibco) at 37 °C in a humidified atmosphere with 5% CO 2 . Cells were used at passage numbers below 30 for all the experiments to ensure consistency.

Knockout HDAC6 (HDAC6 KO ) and ZnF-UBP mutant (HDAC6 RY ) RPMI 8226 cell lines were generated through CRISPR-Cas9 editing. Single-guide RNAs (sgRNA) and the single-stranded DNA donor template (ssODN) for the HDAC6 RY RPMI 8226 cells were designed using design tools from Integrated DNA Technologies (IDT, Coralville, IA, USA) (Table S1 ).

Cas9 ribonucleoprotein complexes (RNP) were assembled by incubating the sgRNA of interest with Alt-R CRISPR-Cas9 tracrRNA ATTO 550 (IDT) at equimolar concentrations at 95 °C for 5 min to ensure hybridization. After cooling down for 10 min at RT, Alt-R HiFi Cas9 Nuclease V3 (IDT) was added, and the complex was incubated for 20 min at RT. RPMI 8226 cells (130,000 cells) were washed three times with PBS and resuspended in a final volume of 15 µL of buffer containing the RNA complex (1.8 µM sgRNA, 1.8 µM tracrRNA, 1.5 µM HiFi Cas9) and 1.8 µM electroporation enhancer (IDT). ssODN was also added (3 μM) to obtain the ZnF-UBP mutant cells. Ten µL of this cell suspension were electroporated using the Neon Transfection System (ThermoFisher) with optimized conditions (1150 V, 30 ms, 2 pulses) and immediately diluted in 100 µL of prewarmed RPMI 1640 medium without antibiotics, supplemented with 1 µL HDR enhancer (IDT, to obtain the ZnF-UBP mutant cells). Electroporated cells were then incubated for 24 h at 37 °C, washed three times with PBS and resuspended in sorting buffer (PBS with 1% BSA). Live ATTO 550-positive cells were sorted and individually added to a well containing 100 µL of conditioned medium. Medium was regularly added to the wells to avoid evaporation, and clonal populations were obtained after about 4 weeks of growth. Then, their genomic DNA was extracted using the Wizard Genomic DNA Purification Kit (Promega, Madison, WI, USA), according to the manufacturer instructions. Regions of interest were amplified through polymerase chain reaction (PCR) using designed primers (IDT) (Table S1 ) and PCR products were sent for Sanger sequencing (Microsynth). For HDAC6 KO RPMI 8226 cells, chromatograms were analyzed by the TIDE (Tracking of Indels by DEcomposition) algorithm ( http://tide-calculator.nki.nl/ , Netherlands Cancer Institute) to obtain the editing efficiency of Cas9.

Western blot

Cells (wild-type, HDAC6 KO and HDAC6 RY RPMI 8226, 2 × 10 6 ) were lysed using NP-40 lysis buffer (ThermoFisher) supplemented with protease inhibitor cocktail (Sigma-Aldrich) and 0.2 M phenylmethylsulfonyl fluoride (ThermoFisher) for 30 min at 4 °C. Lysates were centrifuged at 16,000 g for 10 min at 4 °C and supernatants containing soluble proteins were collected. Proteins were quantified using Coomassie Protein Assay Reagent (ThermoFisher). Protein samples (30 μg) were mixed with loading buffer (50 mM Tris pH 6.8, 6.5% glycerol, 1% sodium dodecyl sulfate, 50 mM dithiothreitol, 0.012% bromophenol blue) and heated to 95 °C for 5 min. Then, proteins were separated in sodium dodecyl sulfate – 8% polyacrylamide gel by electrophoresis at 100 V for 2 h (Bio-Rad, Hercules, CA, USA). A wet transfer to a PVDF membrane (Bio-Rad) was performed in a transfer buffer (48 mM Tris, 39 mM glycine, 10% methanol) at 100 V for 90 min. Membranes were blocked in TBS-T (50 mM Tris pH 7.5, 150 mM NaCl, 0.1% Tween 20) containing 5% milk for 1 h, and primary antibodies were incubated overnight at 4 °C. The following primary antibodies were used: HDAC6 (Cell Signaling Technology, Danvers, MA, USA, #7558, 1:500), acetylated tubulin (Cell Signaling Technology #3971, 1:1000), GADPH (Merck, Burlington, MA, USA, MAB374, 1:500). After washing the membranes three times with TBS-T, species-specific HRP-conjugated secondary antibodies (Cell Signaling Technology, all at 1:1,000) were incubated in TBS-T containing 5% milk for 1 h. Proteins were revealed with SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher) and then imaged using an iBright CL1500 Imaging System (ThermoFisher).

Histone deacetylase catalytic activity

Cells were seeded in a 96-well plate at a density of 15,000 cells per well in 50 µL of medium containing the HDAC substrate N-(4-methyl-7-aminocoumarinyl)-Nα-(t-butoxycarbonyl)-Nω-acetyllysineamide (21 μM, MAL, Sigma-Aldrich) and incubated for 8 h at 37 °C. Following the incubation, the reaction was stopped by adding 10 µL of 6X RIPA buffer, supplemented with protease inhibitor (Sigma-Aldrich). Then, 160 µL of cold acetonitrile were added, and the plate was placed at -80 °C for 10 min. The plate was centrifuged at 5000 g for 10 min at 4 °C, and the supernatants were transferred to a UHPLC-adapted 96-well plate (Nunc 96-well V-bottom polypropylene plate, ThermoFisher) and sealed. The substrate (MAL) and its deacetylated product (dMAL) were measured using a UHPLC-MS method as previously described [ 44 ].

Aggresome formation assay

Cells were seeded in a 6-well plate at a density of 3 × 10 5 cells/mL in 2 mL of medium, and assays were performed 48 h later during the log phase of cell growth. Cells were treated with 5 µM MG-132 for 18 h to stimulate aggresome formation. After, they were fixed with 4% paraformaldehyde (Sigma-Aldrich) for 30 min at RT and then permeabilized with a solution of 0.5% Triton X-100 (Sigma-Aldrich) and 3 mM EDTA (pH 8) (Sigma-Aldrich) for 30 min at 4 °C. After the incubation, cells were stained with Aggresome Detection Reagent (1:5000, #ab139486, Abcam, Cambridge, UK) for 30 min in the dark, and then analyzed by flow cytometry (NovoSampler Pro, Agilent, Santa Clara, CA, USA). Aggresome formation levels were detected at excitation and emission wavelengths of 488 and 600 nm, respectively, and the mean of fluorescence intensity (MFI) was calculated using FlowJo v10.10 analysis software (Ashland, OR, USA).

Cell growth assay

Cells were seeded in a 96-well plate at a density of 75,000 cells/mL in 200 μL of medium and incubated at 37 °C for 0, 6, 12, 24, 36, 48, 60 or 72 h before adding 50 µL of a solution of 1 mg/mL XTT (ThermoFisher) and continuing the incubation for 4 h. The absorbance was measured at 450 nm using a Cytation 3 plate reader (BioTek, Winooski, VT, USA).

Cell cycle assay

Cells were collected from culture during the log phase of cell growth (500,000 cells), washed once with PBS and fixed in cold 70% ethanol for 30 min at 4 °C. Cells were then washed twice with PBS and incubated with propidium iodide staining solution (2.5 μg/mL PI (BD Biosciences, Franklin Lakes, NJ, USA), 100 µg/mL RNAse A (Promega)) for 2 h at RT. Cells were acquired (NovoSampler Pro), and propidium iodide signal was detected at excitation and emission wavelengths of 488 and 600 nm, respectively. The percentages of cells in G0/G1, S, and G2/M phases were obtained using FlowJo v10.10.

Chemical synthesis

Starting building blocks 2a - g and 3 were purchased at the highest commercial quality from Enamine (Kyiv, Ukraine) and Acros (ThermoFisher), respectively. Solvents for solution phase reactions: N,N’-dimethylformamide (DMF), acetone, acetonitrile (MeCN), methanol (MeOH), ethanol (EtOH), and toluene (PhMe) were purchased in anhydrous grade from Sigma-Aldrich and used without further purification. Ethyl acetate (EtOAc), petroleum ether (PET), cyclohexane, chloroform (CHCl 3 ), dichloromethane (CH 2 Cl 2 ) and diethyl ether (Et 2 O) were used as received. HPLC-quality acetonitrile (MeCN), HPLC-quality MeOH and MilliQ water were used for HPLC analyses and purifications. Evaporations were conducted under reduced pressure at temperatures less than 40 °C unless otherwise noted. Reactions were monitored by analytical thin layer chromatography (TLC) carried out on ALUGRAM Xtra SIL G/UV254 silica gel plates (Macherey-Nagel). Compounds were visualized with a UV lamp (λ = 254 nm) and/or stained with a solution of potassium permanganate. Determination of purity of compounds was evaluated by analytical HPLC (VWR Hitachi Chromaster, Radnor, PA, USA) on a reversed-phase column (4.6 × 250 mm, 5 µm, 80 A, ZORBAX Extend-C18, Agilent), using MeOH/H 2 O or MeCN/H 2 O as the mobile phase (UV detection on a range from 200 to 800 nm). Column chromatography was carried out under positive pressure using 15-40 µm silica gel (Macherey-Nagel) and the indicated solvents. Melting points were recorded in open capillary tubes on a Buchi Melting Point M-560 apparatus (Flawil, Switzerland) and are uncorrected. IR spectra were recorded on an Agilent Cary 630 FT-IR spectrometer. Optical rotations were measured at room temperature on an Applied Biosystems 2720 Thermal Cycler polarimeter (Waltham, MA, USA) using a solution cell of 10 cm optical path. NMR spectra of samples in the indicated solvent were recorded on a Bruker Advance 400 or 500 MHz spectrometer (Billerica, MA, USA) and were calibrated using residual solvent as internal reference. The following abbreviations were used to indicate multiplicities: s = singlet, bs = broad singlet, d = doublet, t = triplet, q = quartet, quint = quintuplet, m = multiplet, and AB = AB quartet. Carbon multiplicities were determined by DEPT-135 experiments. Diagnostic correlations were obtained by two-dimensional COSY, HSQC and HMBC experiments. Low resolution (ESIMS) and/or high-resolution mass spectrometric analyses (HRMS) were obtained by electrospray ionization on a Bruker Q-TOF Impact HD apparatus and performed at the PLATeforme INstrumentale d’Analyses (PLATINA) mass spectrometry facility of the Institut de Chimie des Milieux et Matériaux de Poitiers (IC2MP, CNRS-UMR 7285, 86000 Poitiers, France).

Compound 5b

According to the procedure described in the literature [ 20 ], compound 4b (2.1 g, 8.3 mmol, 1 equiv) was suspended in absolute EtOH (80 ml) in a flask equipped with a condenser. A catalytic amount of concentrated H 2 SO 4 (0.5 ml) was carefully added, and the mixture was refluxed until a turbid solution was obtained (17 h). After completion of the reaction (monitored by TLC), the reaction mixture was allowed to cool to 0 °C and was filtered to afford the desired product 5b as white crystals (1.97 g, 84%) (Fig. S1 – S3 ). R f = 0.48 (EtOAc/cyclohexane 50:50); mp 229-231 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 10.97 (bs, 1H), 8.22 (d, J = 2.4 Hz, 1H), 7.68 (dd, J = 8.7, 2.4 Hz, 1H), 7.59 (d, J = 8.7 Hz, 1H), 4.19 (q, J = 7.1 Hz, 2H), 3.06-3.03 (m, 2H), 2.93-2.89 (m, 2H), 1.27 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 173.0, 162.1, 155.1, 147.5, 135.1, 132.4, 128.9, 125.7, 121.9, 61.2, 30.7, 30.2, 14.2; HRMS calcd for C 13 H 14 ClN 2 O 3 [M + H] + 281.0693, found 281.0686.

Compound 6b

According to the procedure described in the literature [ 20 ], K 2 CO 3 (608 mg, 4.4 mmol, 1.2 equiv) and intermediate 5b (1.03 g, 3.7 mmol, 1 equiv) were mixed in dry DMF (18.5 ml) under an N 2 atmosphere. Iodomethane (0.45 ml, 7.4 mmol, 2 equiv) was added, and the reaction was stirred at RT for 22 h. The resulting mixture was diluted with water (30 ml) and extracted with Et 2 O (4 × 25 ml). The combined organic layers were washed with brine (3 × 30 ml), dried over MgSO 4 , filtered, and evaporated. The crude mixture was purified by column chromatography (Et 2 O/PET 10:90 → 45:55) to afford the desired product 6b as white crystals (1.06 g, 98%) (Fig. S4 – S6 ). R f = 0.41 (Et 2 O/PET 60:40); mp 56-58 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 8.20 (d, J = 2.4 Hz, 1H), 7.62 (dd, J = 8.7, 2.4 Hz, 1H), 7.51 (d, J = 8.7 Hz, 1H), 4.18 (q, J = 7.1 Hz, 2H), 3.64 (s, 3H), 3.10 (t, J = 6.6 Hz, 2H), 2.92 (t, J = 6.7 Hz, 2H), 1.28 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 172.6, 161.2, 155.1, 145.4, 134.3, 132.0, 128.6, 126.0, 121.2, 60.6, 30.1, 30.0, 29.5, 14.2; HRMS calcd for C 14 H 16 ClN 2 O 3 [M + H] + 295.0849, found 295.0841.

Quinazolinone 1b

According to the procedure described in the literature [ 20 ], a solution of the ester 6b (233 mg, 0.8 mmol, 1 equiv) in a THF/MeOH 1:1 mixture (18 ml), 1 N KOH (3.2 ml, 3.2 mmol, 4 equiv) was added. The mixture was stirred at RT for 22 h. The reaction was diluted with water, and the organic solvents were evaporated in vacuo . The residue was carefully acidified with aq. 1 N HCl to pH 6-7. The resulting white solid was filtered, washed with wate,r and dried (176 mg, 83%) (Fig. S7 - S11 ). R f = 0.10 (Et 2 O); mp 219-221 °C; IR (neat) ν max 1674, 1588, 1413, 1334, 1215, 1139 cm -1 ; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 12.23 (bs, 1H), 8.02 (d, J = 2.5 Hz, 1H), 7.79 (dd, J = 8.7, 2.5 Hz, 1H), 7.57 (d, J = 8.7 Hz, 1H), 3.55 (s, 3H), 3.08 (t, J = 6.7 Hz, 2H), 2.76 (t, J = 6.6 Hz, 2H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 173.8, 160.3, 157.2, 145.3, 134.3, 130.4, 129.0, 125.1, 120.8, 29.9, 29.8, 29.2; HRMS calcd for C 12 H 12 ClN 2 O 3 [M + H] + 267.0536, found 267.0526; HPLC R t = 19.61 min, purity 98.3%.

Compound 5c

According to the procedure described in the literature [ 20 ], compound 4c (1 g, 4 mmol, 1 equiv) was suspended in absolute EtOH (40 ml) in a flask equipped with a condenser. A catalytic amount of concentrated H 2 SO 4 (0.2 ml) was carefully added and the mixture was refluxed until a turbid solution was obtained (18 h). After completion of the reaction (monitored by TLC), the reaction mixture was allowed to cool to 0 °C and was filtered to afford the desired product 5c as white crystals (1.06 g, 95%) (Fig. S12 – S14 ). R f = 0.70 (MeOH/CH 2 Cl 2 5:95); mp 168-170 °C; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 7.59 (d, J = 8.8 Hz, 1H), 7.51 (d, J = 2.9 Hz, 1H), 7.47 (dd, J = 8.8, 3.0 Hz, 1H), 4.06 (q, J = 7.1 Hz, 2H), 3.88 (s, 3H), 2.98-2.95 (m, 2H), 2.87-2.84 (m, 2H), 1.15 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 171.3, 159.8, 158.8, 158.5, 125.1, 123.2, 120.8, 107.0, 60.5, 56.0, 30.2, 27.7, 14.1; HRMS calcd for C 14 H 17 N 2 O 4 [M + H] + 277.1188, found 277.1181.

Compound 6c

According to the procedure described in the literature [ 20 ], K 2 CO 3 (541 mg, 3.9 mmol, 1.2 equiv) and intermediate 5c (900 mg, 3.26 mmol, 1 equiv) were mixed in dry DMF (16 ml) under N 2 atmosphere. Iodomethane (0.4 ml, 6.5 mmol, 2 equiv) was added, and the reaction was stirred at RT for 24 h. The resulting mixture was diluted with water (25 ml) and extracted with Et 2 O (4 × 20 ml). The combined organic layers were washed with brine (3 × 25 ml), dried over MgSO 4 , filtered, and evaporated. The crude mixture was purified by column chromatography (Et 2 O/PET 20:80 → 75:25) to afford the desired product 6c as white microcrystals (542 mg, 57%) (Fig. S15 – S17 ). R f = 0.35 (Et 2 O/PET 75:25); mp 125-127 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 7.60 (d, J = 3.0 Hz, 1H), 7.51 (d, J = 8.9 Hz, 1H), 7.29 (d, J = 8.7 Hz, 1H), 4.18 (q, J = 7.1 Hz, 2H), 3.64 (s, 3H), 3.10 (t, J = 6.6 Hz, 2H), 2.92 (t, J = 6.7 Hz, 2H), 1.28 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 172.7, 162.2, 158.0, 152.4, 141.6, 128.5, 124.3, 120.8, 105.8, 60.5, 55.7, 30.3, 30.0, 29.3, 14.2; HRMS calcd for C 15 H 19 N 2 O 4 [M + H] + 291.1345, found 291.1342.

Quinazolinone 1c

According to the procedure described in the literature [ 20 ], a solution of the ester 6c (290 mg, 1 mmol, 1 equiv) in a THF/MeOH 1:1 mixture (24 ml), 1 N KOH (4 ml, 4 mmol, 4 equiv) was added. The mixture was stirred at RT for 23 h. The reaction was diluted with water, and the organic solvents were evaporated in vacuo . The residue was carefully acidified with aq. 1 N HCl to pH 6-7. The resulting white solid was filtered, washed with water, and dried (115 mg, 44%) (Fig. S18 – S22 ). R f = 0.38 (Et 2 O); mp 204-206 °C; IR (neat) ν max 2934, 1702, 1666, 1491, 1212, 1019, 849, 795 cm -1 ; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 12.16 (bs, 1H), 7.51 (d, J = 8.9 Hz, 1H), 7.47 (d, J = 2.9 Hz, 1H), 7.38 (dd, J = 8.9, 3.0 Hz, 1H), 3.86 (s, 3H), 3.56 (s, 3H), 3.06 (t, J = 6.7 Hz, 2H), 2.76 (t, J = 6.6 Hz, 2H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 173.9, 161.1, 157.4, 154.2, 141.3, 128.5, 123.8, 120.4, 106.0, 55.6, 29.9, 29.7, 28.9; HRMS calcd for C 13 H 15 N 2 O 4 [M + H] + 263.1032, found 263.1023; HPLC R t = 17.77 min, purity 100%.

Compound 5 d

According to the procedure described in the literature [ 20 ], compound 4d (711 g, 3 mmol, 1 equiv) was suspended in absolute EtOH (30 ml) in a flask equipped with a condenser. A catalytic amount of concentrated H 2 SO 4 (0.2 ml) was carefully added and the mixture was refluxed until a turbid solution was obtained (24 h). After completion of the reaction (monitored by TLC), the reaction mixture was allowed to cool to 0 °C and was filtered to afford the desired product 5d as white crystals (794 mg, 99%) (Fig. S23 - S25 ). R f = 0.82 (MeOH/CH 2 Cl 2 10:90); mp 214-215 °C; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 12.13 (bs, 1H), 7.86 (s, 1H), 7.57 (dd, J = 8.3, 1.8 Hz, 1H), 7.44 (dd, J = 8.2 Hz, 1H), 4.04 (q, J = 7.1 Hz, 2H), 2.88-2.85 (m, 2H), 2.80-2.76 (m, 2H), 2.41 (s, 3H), 1.14 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 172.1, 161.6, 155.0, 146.7, 135.6, 135.6, 126.7, 125.1, 120.6, 59.9, 29.8, 28.9, 20.8, 14.1; HRMS calcd for C 14 H 17 N 2 O 3 [M + H] + 261.1239, found 261.1233.

Compound 6 d

According to the procedure described in the literature [ 20 ], K 2 CO 3 (116 mg, 0.8 mmol, 1.2 equiv) and intermediate 5d (182 mg, 0.7 mmol, 1 equiv) were mixed in dry DMF (4 ml) under N 2 atmosphere. Iodomethane (0.1 ml, 1.4 mmol, 2 equiv) was added and the reaction was stirred at RT for 19 h. The resulting mixture was diluted with water (10 ml) and extracted with Et 2 O (4 × 10 ml). The combined organic layers were washed with brine (2 × 15 ml), dried over MgSO 4 , filtered and evaporated. The crude mixture was purified by column chromatography (Et 2 O/PET 50:50 → 90:10) to afford the desired product 6d as a white solid (157 mg, 82%) (Fig. S26 – S28 ). R f = 0.46 (Et 2 O); mp 97-99 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 8.03 (s, 1H), 7.51 (dd, J = 8.4, 2.0 Hz, 1H), 7.47 (d, J = 8.2 Hz, 1H), 4.18 (q, J = 7.1 Hz, 2H), 3.63 (s, 3H), 3.10 (t, J = 6.8 Hz, 2H), 2.92 (t, J = 6.7 Hz, 2H), 2.46 (s, 3H), 1.27 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 172.7, 162.3, 153.8, 144.9, 136.4, 135.4, 126.7, 126.0, 119.9, 60.5, 30.3, 29.9, 29.4, 21.2, 14.2; HRMS calcd for C 15 H 19 N 2 O 3 [M + H] + 275.1396, found 275.1391.

Quinazolinone 1 d

According to the procedure described in the literature [ 20 ], a solution of the ester 6 d (100 mg, 0.36 mmol, 1 equiv) in a THF/MeOH 1:1 mixture (4 ml), 2 N KOH (0.7 ml, 1.4 mmol, 4 equiv) was added. The mixture was stirred at RT for 20 h. The reaction was diluted with water, and the organic solvents were evaporated in vacuo . The residue was carefully acidified with aq. 1 N HCl to pH 6-7. The resulting white solid was filtered, washed with water, and dried (82 mg, 91%) (Fig. S29 - S33 ). R f = 0.22 (Et 2 O); mp 207-209 °C; IR (neat) ν max 2575, 1709, 1650, 1567, 1406, 1037, 830, 782 cm -1 ; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 7.93 (bs, 1H), 7.69 (dd, J = 8.4, 1.8 Hz, 1H), 7.63 (d, J = 8.3 Hz, 1H), 3.57 (s, 3H), 3.19 (t, J = 7.0 Hz, 2H), 2.81 (t, J = 7.1 Hz, 2H), 2.45 (s, 3H); 13 C NMR (125 MHz, DMSO- d 6 ) δ (ppm): 173.1, 160.2, 159.1, 139.8, 137.7, 136.6, 126.0, 122.8, 118.7, 30.7, 30.2, 28.2, 20.9; HRMS calcd for C 13 H 15 N 2 O 3 [M + H] + 247.1083, found 247.1076; HPLC R t = 18.45 min, purity 100%.

Compound 5e

According to the procedure described in the literature [ 20 ], compound 4e (2.1 g, 8.3 mmol, 1 equiv) was suspended in absolute EtOH (80 ml) in a flask equipped with a condenser. A catalytic amount of concentrated H 2 SO 4 (0.5 ml) was carefully added and the mixture was refluxed until a turbid solution was obtained (19 h). After completion of the reaction (monitored by TLC), the reaction mixture was allowed to cool to 0 °C and was filtered to afford the desired product 5e as yellow crystals (1.72 g, 74%) (Fig. S34 – S36 ). R f = 0.41 (EtOAc/cyclohexane 50:50); mp 164-166 °C; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 12.37 (bs, 1H), 8.04 (d, J = 8.5 Hz, 1H), 7.55 (dd, J = 1.7 Hz, 1H), 7.46 (dd, J = 8.5, 1.9 Hz, 1H), 4.04 (q, J = 7.1 Hz, 2H), 2.88 (t, J = 6.8 Hz, 2H), 2.79 (t, J = 6.7 Hz, 2H), 1.14 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 172.0, 161.1, 157.8, 149.7, 139.0, 127.9, 126.4, 125.9, 119.7, 60.0, 29.7, 29.1, 14.1; HRMS calcd for C 13 H 14 ClN 2 O 3 [M + H] + 281.0693, found 281.0689.

Compound 6e

According to the procedure described in the literature [ 20 ], K 2 CO 3 (829 mg, 6 mmol, 1.2 equiv) and intermediate 5e (1.4 g, 5 mmol, 1 equiv) were mixed in dry DMF (18.5 ml) under N 2 atmosphere. Iodomethane (0.45 ml, 7.4 mmol, 2 equiv) was added and the reaction was stirred at RT for 16 h. The resulting mixture was diluted with water (35 ml) and extracted with Et 2 O (4 × 30 ml). The combined organic layers were washed with brine (3 × 40 ml), dried over MgSO 4 , filtered and evaporated. The crude mixture was purified by column chromatography (Et 2 O/PET 10:90 → 60:40) to afford the desired product 6e as white crystals (1.37 g, 93%) (Fig. S37 – S39 ). R f = 0.35 (Et 2 O/PET 60:40); mp 115-116 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 8.17 (d, J = 8.6 Hz, 1H), 7.57 (dd, J = 2.0 Hz, 1H), 7.37 (d, J = 8.6, 2.0 Hz, 1H), 4.19 (q, J = 7.1 Hz, 2H), 3.63 (s, 3H), 3.10 (t, J = 6.4 Hz, 2H), 2.92 (t, J = 6.7 Hz, 2H), 1.29 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 172.6, 161.6, 156.2, 147.8, 140.0, 128.2, 126.9, 126.4, 118.6, 60.6, 30.0, 29.9, 29.5, 14.2; HRMS calcd for C 14 H 16 ClN 2 O 3 [M + H] + 295.0849, found 295.0850.

Quinazolinone 1e

According to the procedure described in the literature [ 20 ], a solution of the ester 6e (1.19 g, 4 mmol, 1 equiv) in a THF/MeOH 1:1 mixture (48 ml), 2 N KOH (8 ml, 16 mmol, 4 equiv) was added. The mixture was stirred at RT for 17 h. The reaction was diluted with water, and the organic solvents were evaporated in vacuo . The residue was carefully acidified with aq. 1 N HCl to pH 6-7. The resulting white solid was filtered, washed with water, and dried (1.08 g, quant.) (Fig. S40 - S44 ). R f = 0.10 (Et 2 O); mp 250-252 °C; IR (neat) ν max 1722, 1640, 1590, 1419, 1174, 872 cm -1 ; 1 H NMR (500 MHz, DMSO- d 6 ) δ (ppm): 12.24 (bs, 1H), 8.05 (d, J = 8.5 Hz, 1H), 7.54 (d, J = 1.9 Hz, 1H), 7.47 (dd, J = 8.5, 1.9 Hz, 1H), 3.53 (s, 3H), 3.07 (t, J = 6.7 Hz, 2H), 2.76 (t, J = 6.7 Hz, 2H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 173.7, 160.7, 158.3, 147.6, 138.8, 128.3, 126.5, 125.7, 118.4, 29.7, 29.7, 29.2; HRMS calcd for C 12 H 12 ClN 2 O 3 [M + H] + 267.0536, found 267.0531; HPLC R t = 19.63 min, purity 100%.

Compound 5 f

According to the procedure described in the literature [ 20 ], compound 4f (395 mg, 1.5 mmol, 1 equiv) was suspended in absolute EtOH (14 ml) in a flask equipped with a condenser. A catalytic amount of concentrated H 2 SO 4 (0.1 ml) was carefully added and the mixture was refluxed until a turbid solution was obtained (17 h). After completion of the reaction (monitored by TLC), the reaction mixture was allowed to cool to 0 °C and was filtered to afford a crude mixture. The resulting solid was purified by column chromatography with EtOAc as eluent to yield compound 5 f as a white powder (354 mg, 81%) (Fig. S45 – S47 ). R f = 0.60 (Et 2 O); mp 165-167 °C; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 12.67 (bs, 1H), 8.31 (d, J = 8.7 Hz, 1H), 8.25 (dd, J = 2.1 Hz, 1H), 8.19 (dd, J = 8.7, 2.2 Hz, 1H), 4.07 (q, J = 7.1 Hz, 2H), 2.95 (t, J = 6.6 Hz, 2H), 2.84 (t, J = 6.5 Hz, 2H), 1.17 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 172.0, 160.6, 158.7, 151.1, 148.9, 128.0, 125.2, 121.6, 119.7, 60.0, 29.6, 29.0, 14.1; HRMS calcd for C 13 H 14 N 3 O 5 [M + H] + 292.0933, found 292.0930.

Compound 6 f

According to the procedure described in the literature [ 20 ], K 2 CO 3 (152 mg, 1.1 mmol, 1.2 equiv) and intermediate 5f (270 mg, 0.9 mmol, 1 equiv) were mixed in dry DMF (5 ml) under N 2 atmosphere. Iodomethane (0.11 ml, 1.8 mmol, 2 equiv) was added, and the reaction was stirred at RT for 17 h. The resulting mixture was diluted with water (15 ml) and extracted with Et 2 O (4 × 30 ml). The combined organic layers were washed with brine (3 × 20 ml), dried over MgSO 4 , filtered, and evaporated. The crude mixture was purified by column chromatography (Et 2 O/PET 10:90 → 75:25) to afford the desired product 6f as yellow crystals (171 mg, 60%) (Fig. S48 – S50 ). R f = 0.30 (Et 2 O/PET 75:25); mp 119-121 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 8.40 (s, 1H), 8.39 (d, J = 11.7 Hz, 1H), 8.17 (dd, J = 8.8, 2.2 Hz, 1H), 4.19 (q, J = 7.1 Hz, 2H), 3.67 (s, 3H), 3.15 (t, J = 6.2 Hz, 2H), 2.96 (t, J = 6.7 Hz, 2H), 1.30 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 172.6, 161.1, 157.4, 151.4, 147.3, 128.7, 124.3, 122.7, 120.0, 60.8, 30.3, 29.9, 29.6, 14.3; HRMS calcd for C 14 H 16 N 3 O 5 [M + H] + 306.1090, found 306.1088.

Quinazolinone 1 f

According to the procedure described in the literature [ 20 ], a solution of the ester 6f (100 mg, 0.36 mmol, 1 equiv) in a THF/MeOH 1:1 mixture (4 ml), 2 N KOH (0.7 ml, 1.4 mmol, 4 equiv) was added. The mixture was stirred at RT for 20 h. The reaction was diluted with water, and the organic solvents were evaporated in vacuo . The residue was carefully acidified with aq. 1 N HCl to pH 6-7. The resulting yellow solid was filtered, washed with water, and dried (103 mg, 89%) (Fig. S51 – S55 ). R f = 0.17 (Et 2 O); mp 225-227 °C; IR (neat) ν max 3093, 1721, 1650, 1594, 1527, 1340, 1174, 884 cm -1 ; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 12.25 (bs, 1H), 8.33 (d, J = 8.6 Hz, 1H), 8.23 (d, J = 2.0 Hz, 1H), 8.20 (dd, J = 8.6, 2.1 Hz, 1H), 3.59 (s, 3H), 3.14 (t, J = 6.6 Hz, 2H), 2.79 (t, J = 6.5 Hz, 2H); 13 C NMR (100 MHz, DMSO- d 6 ) δ (ppm): 173.7, 160.3, 159.2, 151.0, 146.8, 128.6, 123.8, 121.6, 119.8, 30.0, 29.7, 29.3; HRMS calcd for C 12 H 12 N 3 O 5 [M + H] + 278.0777, found 278.0776; HPLC R t = 18.83 min, purity 98.0%.

Compound 5 g

According to the procedure described in the literature [ 20 ], compound 4g (1.86 g, 8 mmol, 1 equiv) was suspended in absolute EtOH (80 ml) in a flask equipped with a condenser. A catalytic amount of concentrated H 2 SO 4 (0.5 ml) was carefully added and the mixture was refluxed until a turbid solution was obtained (22 h). After completion of the reaction (monitored by TLC), the reaction mixture was diluted with water and extracted with EtOAc (3 × 50 ml). The combined organic layers were washed with brine (3 × 100 ml), dried over MgSO 4 , filtered and evaporated. The crude mixture was purified by column chromatography using EtOAc as eluent to afford the desired product 5g a white powder (1.16 g, 56%) (Fig. S56 - S58 ). R f = 0.84 (MeOH/CH 2 Cl 2 10:90); mp 184-186 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 12.06 (bs, 1H), 8.14 (d, J = 8.1 Hz, 1H), 7.44 (s, 1H), 7.25 (dd, J = 8.0, 1.1 Hz, 1H), 4.16 (q, J = 7.2 Hz, 2H), 3.10 (t, J = 7.1 Hz, 2H), 2.93 (t, J = 7.1 Hz, 2H), 2.47 (s, 3H), 1.24 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 172.6, 163.9, 154.9, 149.3, 145.7, 128.0, 127.0, 126.1, 118.2, 60.8, 30.7, 30.0, 21.9, 14.1; HRMS calcd for C 14 H 17 N 2 O 3 [M + H] + 261.1239, found 261.1237.

Compound 6 g

According to the procedure described in the literature [ 20 ], K 2 CO 3 (664 mg, 4.8 mmol, 1.2 equiv) and intermediate 5g (1.05 g, 4 mmol, 1 equiv) were mixed in dry DMF (20 ml) under N 2 atmosphere. Iodomethane (0.5 ml, 8 mmol, 2 equiv) was added, and the reaction was stirred at RT for 16 h. The resulting mixture was diluted with water (100 ml) and extracted with Et 2 O (4 × 20 ml). The combined organic layers were washed with brine (2 × 50 ml), dried over MgSO 4 , filtered, and evaporated. The crude mixture was purified by column chromatography (Et 2 O/PET 50:50 → 90:10) to afford the desired product 6g as white microcrystals (1.0 g, 91%) (Fig. S59 – S61 ). R f = 0.56 (Et 2 O); mp 114-116 °C; 1 H NMR (400 MHz, CDCl 3 ) δ (ppm): 8.12 (d, J = 8.1 Hz, 1H), 7.37 (s, 1H), 7.24 (dd, J = 8.3, 1.1 Hz, 1H), 4.18 (q, J = 7.1 Hz, 2H), 3.63 (s, 3H), 3.10 (t, J = 6.8 Hz, 2H), 2.92 (t, J = 6.8 Hz, 2H), 2.47 (s, 3H), 1.28 (t, J = 7.1 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ (ppm): 172.7, 162.2, 154.7, 147.0, 144.8, 127.9, 126.6, 126.4, 117.8, 60.5, 30.3, 29.8, 29.5, 21.7, 14.2; HRMS calcd for C 15 H 19 N 2 O 3 [M + H] + 275.1396, found 275.1392.

Quinazolinone 1 g

According to the procedure described in the literature [ 20 ], a solution of the ester 6 g (912 mg, 3.6 mmol, 1 equiv) in a THF/MeOH 1:1 mixture (42 ml), 2 N KOH (7 ml, 14 mmol, 4 equiv) was added. The mixture was stirred at RT for 4 h. The reaction was diluted with water, and the organic solvents were evaporated in vacuo . The residue was carefully acidified with aq. 1 N HCl to pH 6-7. The resulting white solid was filtered, washed with water, and dried (699 mg, 85%) (Fig. S62 - S66 ). R f = 0.11 (Et 2 O); mp 256-258 °C; IR (neat) ν max 2991, 1724, 1640, 1619, 1592, 1420, 1181, 785 cm -1 ; 1 H NMR (400 MHz, DMSO- d 6 ) δ (ppm): 12.18 (bs, 1H), 7.98 (d, J = 8.1 Hz, 1H), 7.36 (s, 1H), 7.29 (d, J = 8.1 Hz, 1H), 3.54 (s, 3H), 3.07 (t, J = 6.6 Hz, 2H), 2.76 (t, J = 6.6 Hz, 2H), 2.44 (s, 3H); 13 C NMR (125 MHz, DMSO- d 6 ) δ (ppm): 173.8, 161.2, 156.5, 146.7, 144.6, 127.7, 126.3, 126.1, 117.3, 29.8, 29.5, 29.0, 21.3; HRMS calcd for C 13 H 15 N 2 O 3 [M + H] + 247.1083, found 247.1081; HPLC R t = 18.01 min, purity 97.8%.

Molecular docking

The structure of the HDAC6 ZnF-UBP domain in complex with inhibitors (PDB number: 6CE6) was downloaded from Protein Data Bank [ 20 ]. The structures of compounds 1a - g were sketched in the Ligedit module implemented in the MolSoft ICM-Pro package ( http://www.molsoft.com/ ). The ZnF-UBP binding site was defined based on the ligand present in the reference PDB and enclosed 6 Å cubic grid. The compounds were then docked using a thoroughness value of 10. A maximum of 10 docked conformations per compound was generated. A radial convolutional neural net (RTCNN) score was used to rank the docked compounds [ 45 ]. The RTCNN score is an abstract value that employs trained features to separate binders from non-binders.

RNA isolation and RNA-seq analysis

RNA was isolated from each cell line using Aurum total RNA mini kit (Bio-Rad) and following manufacturer instructions. In brief, 2 × 10 6 cells of each line were washed in PBS and then resuspended in 350 μL lysis solution (supplemented with 1% β-mercaptoethanol (Sigma-Aldrich)). 350 μL of 70% ethanol were added and the mixture was transferred to an RNA binding column. After centrifugation, the column was washed with 700 μL of wash solution, and 80 μL of DNAse I solution were added to digest DNA. The column was washed again twice, and RNA was eluted by adding 50 μL of elution buffer. The RNA quality control of all samples was checked prior to sequencing (RNA integrity number: 10). Sequencing was performed by an Illumina NovaSeq 6000 System using the library TruSeq HT stranded (100 bp single-end reads). After sequencing quality control (FastQC), the reads were aligned to the GRCh38 Ensembl human genome reference with STAR v.2.7.4 [ 46 ]. The aligned reads were assessed for gene features with HTSeq v0.9.1 and the matrix of counts per gene was produced. The differential expression analysis was performed with the statistical analysis R/Bioconductor package edgeR 1.34.1 [ 47 ]. After normalization, the poorly detected genes were filtered out and out of 60,666 total genes, 21,394 genes with a count above 10 were kept. Among them, only those genes with a fold change 2 threshold and a p -value < 0.05 (with a false discovery rate of 5%) were considered as differentially expressed. The results are the average of 2 independent RNA extractions. Gene ontology enrichment analysis was performed through the R package clusterProfiler [ 48 ].

IL-6 and IL-10 production measurement

Cells were seeded in a 96-well plate at a density of 5 × 10 5 cells/mL in 200 μL of medium with 100 ng/mL LPS (InvivoGen, San Diego, CA, USA) and incubated at 37 °C for 24 h. Cell viability was evaluated using Trypan Blue. Then, the plate was centrifuged at 200 g for 2 min at RT, and supernatants were transferred to another plate for IL-6 and IL-10 quantification. The concentrations of the cytokines were measured by ELISA using two commercially available kits from BioLegend (#430501 for IL-6 and #430601 for IL-10, San Diego, CA, USA). The detection limits for IL-6 and IL-10 were 7.8 pg/mL and 3.9 pg/mL, respectively.

Statistical analysis

Data representation and statistical analyses were performed using GraphPad Prism 9 (GraphPad, La Jolla, CA, USA). Sample distribution was tested for normality (Shapiro-Wilk, α = 0.05) prior to group comparison. The details of the tests used and levels of significance are mentioned in the figure legends.

Supplementary information

Supplementary information

Original data

Acknowledgements

We thank Aurélie Gouiller, Olivier Petermann, and the iGE3 Genomics Platform for their technical support and La Ligue Contre le Cancer (Comité Charente-Maritime) for their financial support.

Author contributions

Conceptualization, PB and MC; methodology, RR, IFC, AR, FA, MU, SH, CB, PB and MC; biological evaluation, RR, IFC and MU; synthesis of compounds, AR and PB; writing—original draft preparation, RR and MC; writing—review and editing, RR, SH, CB, PB and MC; funding acquisition, CB, PB and MC; supervision, PC and MC; project administration, MC. All authors have read and approved the final manuscript.

Funding

This project was supported by the Swiss National Science Foundation grants #310030_184790 (to MC) and #182317 (to CB).

Data availability

All the RNA-Seq data have been deposited in the Yareta repository with the DOI 10.26037/yareta:tc2l7x5xxza4rfi2wrz5vg3c5q. Other data supporting the findings of this study will be made available upon reasonable request.

Competing interests

The authors declare no competing interests.

Ethics

All the methods were performed in accordance with the relevant guidelines and regulations. The in vitro experiments (recombinant proteins and cells) performed in this work did not use material that requires ethical approval, and no experiments were conducted on animals or human patients’ samples.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

The online version contains supplementary material available at 10.1038/s41420-025-02465-1.

References

1. Rajkumar SV. Multiple myeloma: Every year a new standard? Hematol Oncol. 2019;37:62–5.

2. van de Donk N, Pawlyn C, Yong KL. Multiple myeloma. Lancet. 2021;397:410–27.

3. Cowan AJ, Green DJ, Kwok M, Lee S, Coffey DG, Holmberg LA, et al. Diagnosis and management of multiple myeloma: a review. JAMA. 2022;327:464–77.

4. Rajkumar SV. Multiple myeloma: 2020 update on diagnosis, risk-stratification and management. Am J Hematol. 2020;95:548–67.

5. Liang T, Wang F, Elhassan RM, Cheng Y, Tang X, Chen W, et al. Targeting histone deacetylases for cancer therapy: Trends and challenges. Acta Pharm Sin B. 2023;13:2425–63.

6. Mithraprabhu S, Kalff A, Chow A, Khong T, Spencer A. Dysregulated class I histone deacetylases are indicators of poor prognosis in multiple myeloma. Epigenetics. 2014;9:1511–20.

7. Cheng T, Kiser K, Grasse L, Iles L, Bartholomeusz G, Samaniego F, et al. Expression of histone deacetylase (HDAC) family members in bortezomib-refractory multiple myeloma and modulation by panobinostat. Cancer Drug Resist. 2021;4:888–902.

8. Richardson PG, Hungria VT, Yoon SS, Beksac M, Dimopoulos MA, Elghandour A, et al. Panobinostat plus bortezomib and dexamethasone in previously treated multiple myeloma: outcomes by prior treatment. Blood. 2016;127:713–21.

9. U.S. Food and Drug Administration. Withdrawal of approval of new drug application for Farydak (panobinostat) capsules, 10 milligrams, 15 milligrams, and 20 milligrams. Fed Reg. 2022;87:16742–3.

10. Liu Y, Peng L, Seto E, Huang S, Qiu Y. Modulation of histone deacetylase 6 (HDAC6) nuclear import and tubulin deacetylase activity through acetylation. J Biol Chem. 2012;287:29168–74.

11. Brindisi M, Saraswati AP, Brogi S, Gemma S, Butini S, Campiani G. Old but gold: Tracking the new guise of histone deacetylase 6 (HDAC6) enzyme as a biomarker and therapeutic target in rare diseases. J Med Chem. 2020;63:23–39.

12. Aldana-Masangkay GI, Sakamoto KM. The role of HDAC6 in cancer. J Biomed Biotechnol. 2011;2011:875824.

13. Ouyang H, Ali YO, Ravichandran M, Dong A, Qiu W, MacKenzie F, et al. Protein aggregates are recruited to aggresome by histone deacetylase 6 via unanchored ubiquitin C termini. J Biol Chem. 2012;287:2317–27.

14. Boyault C, Zhang Y, Fritah S, Caron C, Gilquin B, Kwon SH, et al. HDAC6 controls major cell response pathways to cytotoxic accumulation of protein aggregates. Genes Dev. 2007;21:2172–81.

15. Huang FI, Wu YW, Sung TY, Liou JP, Lin MH, Pan SL, et al. MPT0G413, a novel HDAC6-selective inhibitor, and bortezomib synergistically exert anti-tumor activity in multiple myeloma cells. Front Oncol. 2019;9:249.

16. Lee SW, Yeon SK, Kim GW, Lee DH, Jeon YH, Yoo J. HDAC6-selective inhibitor overcomes bortezomib resistance in multiple myeloma. Int J Mol Sci.2021;22:1341.

17. Yee AJ, Bensinger WI, Supko JG, Voorhees PM, Berdeja JG, Richardson PG, et al. Ricolinostat plus lenalidomide, and dexamethasone in relapsed or refractory multiple myeloma: a multicentre phase 1b trial. Lancet Oncol. 2016;17:1569–78.

18. Vogl DT, Raje N, Jagannath S, Richardson P, Hari P, Orlowski R, et al. Ricolinostat, the first selective histone deacetylase 6 inhibitor, in combination with bortezomib and dexamethasone for relapsed or refractory multiple myeloma. Clin Cancer Res. 2017;23:3307–15.

19. Porter NJ, Mahendran A, Breslow R, Christianson DW. Unusual zinc-binding mode of HDAC6-selective hydroxamate inhibitors. Proc Natl Acad Sci USA. 2017;114:13459–64.

20. Ferreira de Freitas R, Harding RJ, Franzoni I, Ravichandran M, Mann MK, Ouyang H, et al. Identification and structure-activity relationship of HDAC6 zinc-finger ubiquitin binding domain inhibitors. J Med Chem. 2018;61:4517–27.

21. Harding RJ, Franzoni I, Mann MK, Szewczyk MM, Mirabi B, Ferreira de Freitas R, et al. Discovery and characterization of a chemical probe targeting the zinc-finger ubiquitin-binding domain of HDAC6. J Med Chem. 2023;66:10273–88.

22. Dong A, Ravichandran, M., Loppnau, P., Li, Y., MacKenzie, F., Kozieradzki, et al. Structural Genomics Consortium (SGC). Crystal structure of human HDAC6 zinc finger domain and ubiquitin C-terminal peptide RLRGG 10.2210/pdb3GV4/pdb (2009).

23. Hubbert C, Guardiola A, Shao R, Kawaguchi Y, Ito A, Nixon A, et al. HDAC6 is a microtubule-associated deacetylase. Nature. 2002;417:455–8.

24. Richieu A, Bertrand P. Convergent micro-wave assisted synthesis of quinazolinone and its precursor using the bio-sourced solvent pinane. RSC Adv. 2023;13:20646–50.

25. Anisimov S, Takahashi M, Kakihana T, Katsuragi Y, Kitaura H, Zhang L, et al. G3BP1 inhibits ubiquitinated protein aggregations induced by p62 and USP10. Sci Rep. 2019;9:12896.

26. Zhou L, Wang H, Chen D, Gao F, Ying Z, Wang G. p62/sequestosome 1 regulates aggresome formation of pathogenic ataxin-3 with expanded polyglutamine. Int J Mol Sci. 2014;15:14997–5010.

27. Chattopadhyay S, Stewart AL, Mukherjee S, Huang C, Hartwell KA, Miller PG, et al. Niche-based screening in multiple myeloma identifies a kinesin-5 inhibitor with improved selectivity over hematopoietic progenitors. Cell Rep. 2015;10:755–70.

28. Wang X, Sinn AL, Pollok K, Sandusky G, Zhang S, Chen L, et al. Preclinical activity of a novel multiple tyrosine kinase and aurora kinase inhibitor, ENMD-2076, against multiple myeloma. Br J Haematol. 2010;150:313–25.

29. Bou Zerdan M, Nasr L, Kassab J, Saba L, Ghossein M, Yaghi M, et al. Adhesion molecules in multiple myeloma oncogenesis and targeted therapy. Int J Hematol Oncol. 2022;11:IJH39.

30. Sanchez L, Wang Y, Siegel DS, Wang ML. Daratumumab: a first-in-class CD38 monoclonal antibody for the treatment of multiple myeloma. J Hematol Oncol. 2016;9:51.

31. Podar K, Zimmerhackl A, Fulciniti M, Tonon G, Hainz U, Tai YT, et al. The selective adhesion molecule inhibitor natalizumab decreases multiple myeloma cell growth in the bone marrow microenvironment: therapeutic implications. Br J Haematol. 2011;155:438–48.

32. Wang Y, Wang K, Fu J. HDAC6 mediates poly (I:C)-induced TBK1 and Akt phosphorylation in macrophages. Front Immunol. 2020;11:1776.

33. Cheng F, Lienlaf M, Perez-Villarroel P, Wang HW, Lee C, Woan K, et al. Divergent roles of histone deacetylase 6 (HDAC6) and histone deacetylase 11 (HDAC11) on the transcriptional regulation of IL10 in antigen presenting cells. Mol Immunol. 2014;60:44–53.

34. Zhang WB, Yang F, Wang Y, Jiao FZ, Zhang HY, Wang LW, et al. Inhibition of HDAC6 attenuates LPS-induced inflammation in macrophages by regulating oxidative stress and suppressing the TLR4-MAPK/NF-kappaB pathways. Biomed Pharmacother. 2019;117:109166.

35. Harmer D, Falank C, Reagan MR. Interleukin-6 interweaves the bone marrow microenvironment, bone loss, and multiple myeloma. Front Endocrinol (Lausanne). 2018;9:788.

36. Li T, Zhang C, Hassan S, Liu X, Song F, Chen K, et al. Histone deacetylase 6 in cancer. J Hematol Oncol 2018;11:111.

37. Maier JA, Martinez C, Kasavajhala K, Wickstrom L, Hauser KE, Simmerling C. ff14SB: Improving the accuracy of protein side chain and backbone parameters from ff99SB. J Chem Theory Comput 2015;11:3696–713.

38. Yu Z, Li P, Merz KM Jr. Extended zinc AMBER force field (EZAFF). J Chem Theory Comput 2018;14:242–54.

39. Jorgensen WL, Chandrasekhar J, Madura JD, Impey RW, Klein ML. Comparison of simple potential functions for simulating liquid water. J Chem Phys 1983;79:926–35.

40. Harvey MJ, Giupponi G, Fabritiis GD. ACEMD: Accelerating biomolecular dynamics in the microsecond time scale. J Chem Theory Comput 2009;5:1632–9.

41. Michaud-Agrawal N, Denning EJ, Woolf TB, Beckstein O. MDAnalysis: a toolkit for the analysis of molecular dynamics simulations. J Comput Chem. 2011;32:2319–27.

42. Jurrus E, Engel D, Star K, Monson K, Brandi J, Felberg LE, et al. Improvements to the APBS biomolecular solvation software suite. Protein Sci. 2018;27:112–28.

43. Meng EC, Goddard TD, Pettersen EF, Couch GS, Pearson ZJ, Morris JH, et al. UCSF ChimeraX: Tools for structure building and analysis. Protein Sci. 2023;32:e4792.

44. Simoes-Pires CA, Zwick V, Cretton S, Cuendet M. Simultaneous measurement of HDAC1 and HDAC6 activity in HeLa cells using UHPLC-MS. J Vis Exp. 2017;e55878.

45. Su M, Yang Q, Du Y, Feng G, Liu Z, Li Y, et al. Comparative assessment of scoring functions: The CASF-2016 update. J Chem Inf Model. 2019;59:895–913.

46. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: Ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29:15–21.

47. Robinson MD, McCarthy DJ, Smyth GK. edgeR: A Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–40.

48. Yu G, Wang LG, Han Y, He QY. clusterProfiler: An R package for comparing biological themes among gene clusters. OMICS. 2012;16:284–7.

Associated Data

Supplementary Materials

Supplementary information

Original data

Data Availability Statement

All the RNA-Seq data have been deposited in the Yareta repository with the DOI 10.26037/yareta:tc2l7x5xxza4rfi2wrz5vg3c5q. Other data supporting the findings of this study will be made available upon reasonable request.