Histone deacetylase inhibitors suppress ACE2 and ABO simultaneously, suggesting a preventive potential against COVID-19

Coronavirus disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has spread worldwide as a pandemic throughout 2020. Since the virus uses angiotensin-converting enzyme 2 (ACE2) as a receptor for cellular entry, increment of ACE2 would lead to an increased risk of SARS-CoV-2 infection. At the same time, an association of the ABO blood group system with COVID-19 has also been highlighted: there is increasing evidence to suggest that non-O individuals are at higher risk of severe COVID-19 than O individuals. These findings imply that simultaneous suppression of ACE2 and ABO would be a promising approach for prevention or treatment of COVID-19. Notably, we have previously clarified that histone deacetylase inhibitors (HDACIs) are able to suppress ABO expression in vitro. Against this background, we further evaluated the effect of HDACIs on cultured epithelial cell lines, and found that HDACIs suppress both ACE2 and ABO expression simultaneously. Furthermore, the amount of ACE2 protein was shown to be decreased by one of the clinically-used HDACIs, panobinostat, which has been reported to reduce B-antigens on cell surfaces. On the basis of these findings, we conclude that panobinostat could have the potential to serve as a preventive drug against COVID-19.


Results
HDACIs such as sodium butyrate and panobinostat suppress ACE2 expression in KATOIII cells. We have previously reported that HDACIs such as sodium butyrate and panobinostat suppress ABO expression in the gastric cancer cell line KATOIII 19 . To examine whether the HDACIs also decrease ACE2 expression, we performed quantitative real-time PCR (qPCR) on KATOII cells treated with or without 1 mM sodium butyrate or 25 nM panobinostat for 6 or 24 h, targeting ABO, ACE2 and TMPRSS2 transcripts as well as β-actin (ACTB) as endogenous control. As we had shown before, these HDACIs suppressed ABO expression (Fig. 1A). In addition, it was also clarified that they suppressed the expression of ACE2 in a time-dependent manner (Fig. 1B). On the other hand, the HDACIs did not suppress TMPRSS2 (Fig. 1C), suggesting that the HDACI-related suppression was gene-specific.

Various HDACIs suppress ACE2 expression in KATOIII cells.
To further evaluate the effect of HDA-CIs on ABO, ACE2 and TMPRSS2 expression, KATOIII cells were treated with several concentrations of various HDACIs including sodium valproate, vorinostat and trichostatin A for 24 h, and the relative amounts of these transcripts were evaluated by qPCR ( Fig. 2A-O). In addition, the cell proliferations and viabilities were evaluated in each condition to reveal potential cytotoxicity of the HDACIs (Fig. 2P-T). As expected, sodium butyrate and panobinostat reduced the amount of both ABO and ACE2 transcripts simultaneously in a dose-dependent manner. In addition, sodium valproate and vorinostat caused similar suppression of both ABO and ACE2, except that a low concentration (0.5 mM) of sodium valproate increased the amount of ACE2 (Fig. 2H). On the other hand, trichostatin A rarely reduced the ABO and ACE2 transcripts (Fig. 2M,N), while moderately reducing the cell proliferations and viabilities (Fig. 2T). Those observations suggested the considerable cytotoxicity of trichostatin A. None of the HDACIs had a suppressive effect on TMPRSS2, except for a high concentration (50 nM) of panobinostat, which suppressed TMPRSS2 (Fig. 2F). On this basis, we concluded that the HDACIs used in the present study, except for trichostatin A, had the potential to suppress ABO and ACE2 concurrently on KATOIII cells, whereas such suppression was rarely observed for the TMPRSS2 transcript. and TMPRSS2 (C) transcripts in KATOIII cells treated with or without HDACIs such as sodium butyrate and panobinostat. Clear bars indicate the basal expression levels in the absence of HDACIs, the gray bars represent the relative amounts of transcripts in the presence of 1 mM sodium butyrate, and the solid bars denote those in the presence of 25 nM panobinostat. The graphs express mean fold values relative to those without HDACI. In each panel, the left-hand bars show the results obtained 6 h after incubation with or without HDACIs, while the right-hand bars show data 24 h after treatment. The asterisks represent a significant reduction compared to the values without HDACI (p < 0.05).  (Table 1). As a result, we found that, in addition to KATOIII cells, the gastric cancer cell line NUGC-4 also expressed considerable amounts of ABO, ACE2 and TMPRSS2 transcripts, while all the other cell lines including a lung derived cell line, HMVEC-L, expressed an insufficient amount of at least one of the three transcripts. Therefore, the NUGC-4 cell line was deemed relevant for further evaluation of the HDACI-related suppression of ABO and ACE2. Against this background, we performed similar qPCR experiments on NUGC-4 cells incubated with or without various concentrations of sodium butyrate or panobinostat (Fig. 3), resulting in similar suppression of ABO and ACE2 by the HDACIs, while TMPRSS2 was not suppressed. Thus, it was clarified that HDACIs such as sodium butyrate and panobinostat had the ability to suppress both ABO and ACE2 simultaneously in several epithelial cell lines. www.nature.com/scientificreports/ HDACIs decrease the ACE2 protein in cell lysates of KATOIII and NUGC-4. Whether or not the HDACI-related suppression of ACE2 would lead to a reduced amount of ACE2 protein in cultured cell lines was considered an intriguing issue. Notably, we had shown previously that panobinostat reduced the amount of B-antigens on KATOIII cells. Therefore, among the various HDACIs, we decided to focus on panobinostat, considering that it might serve as a preventive drug against COVID-19 by simultaneously diminishing A-or B-antigens and ACE2 proteins on the cell surface.
To this end, we performed enzyme-linked immunosorbent assays (ELISA) using cell lysates prepared from KATOIII and NUGC-4 cells incubated with 0, 25 or 50 nM panobinostat for 24 or 48 h (Fig. 4). For KATOIII cells, we confirmed that the amount of ACE2 was reduced at 48 h after incubation with 25 nM and 50 nM panobinostat (Fig. 4A). For NUGC-4 cells, ACE2 reduction was observed after both 24-and 48-h treatment with   www.nature.com/scientificreports/ 25 nM and 50 nM panobinostat (Fig. 4B). There was no significant difference in the amount of ACE2 between the two concentrations of panobinostat. Furthermore, we carried out additional Western blot analysis of whole-cell lysates prepared from KATOIII and NUGC-4 cells cultured under the same conditions as those for the above ELISA. For KATOIII cells, however, we rarely detected ACE2 in the lysates even though no panobinostat was added (Fig. 4C,D), probably because of the relatively low level of ACE2 expression in KATOIII cells (Table 1). On the other hand, for NUGC-4 cells, it was clarified that panobinostat at both 25 nM and 50 nM decreased the amount of ACE2 after treatment for both 24 and 48 h (Fig. 4E-G). In conclusion, panobinostat was able to reduce the amount of ACE2 in cultured epithelial cells.

Discussion
We revealed that several HDACIs suppressed ABO and ACE2 transcripts concurrently in KATOIII or NUGC-4 cells, while TMPRSS2 expression was rarely repressed. Notably, among the HDACIs used in the present study, panobinostat caused drastic suppression at the lowest concentrations, whereas trichostatin A barely reduced those transcripts probably because of its cytotoxicity. Finally, panobinostat reduced the amount of ACE2 proteins in both KATOIII and NUGC-4 cells. Considering together the findings that panobinostat decreases B-antigen on the KATOIII cells 19 , that non-O individuals have a higher risk of COVID-19 18 and that higher expression of ACE2 is a risk factor for COVID-19 7-10 , it seems plausible that panobinostat could have the potential to serve as a preventive drug against COVID-19. Currently, the association of panobinostat with ACE2 expression is also being investigated by two other research groups 22,23 . Those two groups adopted very similar approaches; they re-analyzed the same publically available data, comprising gene expression profiles for thousands of perturbagens at a variety of time points, doses, and cell lines 24 in order to identify drugs that could significantly modify ACE2 expression. Although their interpretations of the results differed, to our surprise, both of their analyses suggested that panobinostat might up-regulate ACE2 expression, contrary to our findings. Although the reason for this contradiction is unclear, we speculate that it might be attributable to differences in experimental conditions between our data and those publically available. For example, the former group, He and Garmire, analyzed expression profiles in the presence of 10 µM panobinostat, which was more than 200 times higher than the concentration we employed 22 . In addition, neither KATOIII nor NUGC-4 were featured in the publically available data 24 . As shown in Table 1, few cell lines seem to express a sufficient amount of ACE2, and thus most cultured cells are unsuitable for investigating regulation of the gene 4,25 . Therefore, the effects of HDACIs on gene expression in vitro need to be evaluated carefully. On the other hand, Xu et al. recently reported a novel approach whereby conventional molecular docking was computationally accelerated in combination with generative artificial intelligence, resulting in the identification of potential drug-repurposing candidates for COVID-19 26 . Surprisingly, though their approach was completely different from ours, panobinostat was identified as one of the six candidate drugs. Considering this accumulating evidence to suggest the preventive potential of panobinostat against COVID-19, further evaluations, including clinical trials, seem warranted. Notably, the in vitro potencies of valproic acid, panobinostat and vorinostat as therapeutics for various cancers were reported to be in the mM, µM and nM order, respectively 27 , whereas our data revealed that panobinostat might have preventive potential against COVID-19 at concentrations as low as the nM order. Although future in vivo or clinical studies will be needed to clarify the possible tolerable doses of HDACIs for potential prevention of COVID-19, our present data suggest that panobinostat in particular could be employed clinically at especially low doses, leading to less adverse effects.
In some experiments, we observed slight up-regulation of TMPRSS2 in KATOIII cells when the cells were treated with 1 mM sodium butyrate for 6 h (Fig. 1C) or 1 mM sodium valproate for 24 h (Fig. 2I). In addition, the NUGC-4 cells showed an apparent increase in TMPRSS2 expression in the presence of HDACIs such as sodium butyrate and panobinostat (Fig. 3C,F). While these results suggest potential activation of the host protease TMPRSS2 by HDACIs, this might be irrelevant to the risk of viral infection considering that TMPRSS2 is involved in the process of viral entry only after the virus has bound to ACE2 on epithelial cells 3,4 . Whether HDA-CIs actually inhibit the cellular entry of the virus will need to be clarified by future experiments using live virus.
An increasing number of reports have suggested an association between the ABO system and COVID-19 11,12,[14][15][16][17][18] . This is not surprising, since involvement of the ABO system in viral infection, including SARS, has been documented 28 , although the precise mechanism remains unclear. It is controversial whether the association of the ABO system with COVID-19 could be attributable to the amount of A-or B-antigens on cells or anti-A or -B antibodies in serum 11,12,14 . Interestingly, Ladikou et al. have reported that patients with severe COVID-19 who developed venous thromboembolism had highly elevated levels of von Willebrand factor (vWF) and coagulation factor VIIIc 29 , whose serum levels are correlated with the ABO system and are higher in non-O individuals 30,31 . Considering that the presence of A-or B-antigens in vWF N-linked oligosaccharides plays a role in vWF levels 32,33 , reduction of the A-or B-antigen might be a reasonable approach for reducing plasma vWF levels as well as the risk of thrombopoietic symptoms of COVID-19. Notably, the ABO system is also associated with a number of other factors, including ACE plasma activity 34 and interleukin levels 35,36 . Thus, the influence of the ABO system seems to be complex, highly diversified and much more significant than has been clarified 33 . Further investigations focusing on the ABO blood group system should help to reveal the hidden roles of this system that could significantly impact human health, disease and biology.

Materials and methods
Cell culture with or without HDACIs. The  www.nature.com/scientificreports/ (TKG 0605), KK47 (TKG 0663) and T24 (TKG 0443) were from Cell Resource Center for Biomedical Research (Miyagi, Japan) and LAN-5 (RCB0485) was from RIKEN BioResource Research Center (Ibaraki, Japan). The KATOIII and K562 cells were cultured as described previously 19 . The NUGC-4, MKN1, SW480, 5637, KK47 and LAN-5 cell lines were cultured in RPMI1640 medium containing 10% FCS, 100 U/ml penicillin and 100 µg/ml streptomycin. The culture medium for SV-HUC, T24, HMVEC-L or SH-SY5Y was Ham's F12K, MEM, EGM-2MV Microvascular Endothelial Cell Growth Medium-2 or 1:1 mixture of MEM and F12, each with 10% FCS, 100 U/ml penicillin and 100 µg/ml streptomycin. For treatment with the HDACIs, the cells were seeded at a density of 2.5 × 10 5 /ml one day before the experiment. On the following day, the cells were re-seeded at a density of 2.5 × 10 5 /ml in new medium with or without HDACIs. The medium was not changed thereafter until harvest of the cells. The HDMCIs we used included sodium butyrate (#303410; Sigma-Akdrich), panobinostat (#13280; Cayman Chemical Company), sodium valproate (#13033; Cayman Chemical Company), vorinostat (#10009929; Cayman Chemical Company) and trichostatin A (#89730; Cayman Chemical Company). The solvent used for sodium butyrate and sodium valproate was deionized-distilled water, and that used for panobinostat, vorinostat and trichostatin A was dimethyl sulfoxide. KATOIII cell proliferation and viability were evaluated by both manual and automatic counting of live or dead cells after trypan blue staining in each condition where the cells were treated with various concentrations of each HDACI for 24 h. The automatic counting was performed with a Countess II FL Automated Cell Counter (#AMQAF1000, Invitrogen). Every count was conducted at least twice.
Quantitative real-time PCR (qPCR). RNA purification, cDNA preparation and quantification of ABO and ACTB transcripts were performed as described previously 19 . qPCR of the ACE2 and TMPRSS2 transcripts was performed with the specific primer sets "ACE2 Primer 2" and "TMPRSS2 Primer 2", respectively, as described by Ma et al. 37 , under the following conditions: 95 °C for 3 min and 40 cycles at 95 °C for 3 s and at 60 °C for 30 s. Every assay was conducted at least twice, and the absolute amount of each transcript determined by qPCR was standardized by the amount of ACTB transcript.
Enzyme-linked immunosorbent assay (ELISA). ELISA was performed using a Human ACE2 ELISA Kit (#ab235649, Abcam) following manufacturer's instructions. Briefly, KAKTOIII or NUGC-4 cells were harvested 24 or 48 h after incubation with or without HDACIs, and solubilized in 1 × cell extraction buffer PTR. After centrifugation, the concentration of total protein in the supernatant was measured using a DC protein assay kit (#5000112JA, Bio-Rad), and 250 or 100 ng of total protein derived from KATOIII or NUGC-4 cell lysates, respectively, was applied to each well of a ready-to-use microplate provided in the kit. Then the ACE2 antibody cocktail was added to each well, followed by 1-h incubation at RT on a plate shaker set to 400 rpm. After the incubation, each well was rinsed three times, TMB Development Solution was added, and incubation was performed at RT on a plate shaker. Ten minutes later, the Stop Solution was added to each well, and the end point reading of OD at 450 nm was recorded using a iMark microplate absorbance reader (#168-1135, Bio-Rad). The concentration of the ACE2 protein in the sample was determined by interpolating the blank control subtracted absorbance values against the standard curve. Every assay was conducted in duplicate.
Western blotting. Whole-cell lysates were prepared from KATOIII and NUGC-4 cells incubated with 0, 25 or 50 nM panobinostat for 24 or 48 h. One hundred or 40 µg of total protein was applied to each lane for the KATOIII or NUGC-4 assays, respectively. The transferred membrane was reacted with a recombinant rabbit anti- www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.