Histone H1.2 promotes hepatocarcinogenesis by regulating signal transducer and activator of transcription 3 signaling

Abstract Linker histone H1.2 (H1.2), encoded by HIST1H1C (H1C), is a major H1 variant in somatic cells. Among five histone H1 somatic variants, upregulated H1.2 was found in human hepatocellular carcinoma (HCC) samples and in a diethylnitrosamine (DEN)‐induced HCC mouse model. In vitro, H1.2 overexpression accelerated proliferation of HCC cell lines, whereas H1.2 knockdown (KD) had the opposite effect. In vivo, H1.2 insufficiency or deficiency (H1c KD or H1c KO) alleviated inflammatory response and HCC development in DEN‐treated mice. Mechanistically, H1.2 regulated the activation of signal transducer and activator of transcription 3 (STAT3), which in turn positively regulated H1.2 expression by binding to its promoter. Moreover, upregulation of the H1.2/STAT3 axis was observed in human HCC samples, and was confirmed in mouse models of methionine‐choline‐deficient diet induced nonalcoholic steatohepatitis or lipopolysaccharide induced acute inflammatory liver injury. Disrupting this feed‐forward loop by KD of STAT3 or treatment with STAT3 inhibitors rescued H1.2 overexpression‐induced proliferation. Moreover, STAT3 inhibitor treatment‐ameliorated H1.2 overexpression promoted xenograft tumor growth. Therefore, H1.2 plays a novel role in inflammatory response by regulating STAT3 activation in HCC, thus, blockade of the H1.2/STAT3 loop is a potential strategy against HCC.


| INTRODUC TI ON
Liver cancer is the fourth leading cause of death worldwide, with HCC the most dominant type. 1 Over the past decade, neither surgery nor chemotherapy has proven appreciably effective in improving HCC outcomes. 2 Hence, understanding the mechanisms behind HCC and identifying effective therapeutic targets are urgently required.
Genetic and epigenetic alterations all contribute to the development and progression of HCC. 3 Compared to epigenetic alterations, such as DNA methylation and histone modifications, linker histone H1 alterations are less investigated in HCC. 4 The linker histone H1 family, well known as chromatin structural proteins, consists of five somatic variants (H1.1-H1.5) in mammals. 5 In most somatic cells, histone H1.2 coded by H1C is one of the predominant variants. 5 Although KO of H1.2 in mouse shows no anatomic or histological abnormality, 6 its roles in apoptosis, 7 cell cycle progression, 8 autophagy, 9 and DNA damage repair 10  Signal transducer and activator of transcription 3, a pivotal transcription factor that regulates inflammation and innate immunity, plays critical roles in multiple physiological processes, including cell growth, survival, metastasis and angiogenesis. 13 Activation of STAT3 requires phosphorylation on tyrosine 705 (Y705), which mediates its homodimerization, nuclear translocation, and transcription of downstream target genes, including IL-6, FOS, and SOCS3. 14 Nuclear trafficking of STAT3 is essential to its function, and its nuclear translocation has been reported to be regulated by various importins, including KPNA2. 15 A Y640F mutation that causes constitutive Y705 phosphorylation by stabilizing STAT3 homodimerization (constitutive activation) is found in multiple cancers, especially in inflammation-associated tumorigenesis, such as HCC. 16,17 Therefore, elucidation of new factors that regulate STAT3 activation will reveal insights on HCC development.
Here, we reported a H1.

| Mice and experimental design
The Hist1h1c (H1c) KO mice were generated by Biocytogen using a CRISPR/Cas9 based system. Two single guide RNAs were designed to target the upstream of the 5′-UTR and 3′-UTR of Hist1h1c exon 1, respectively, using the CRISPR design tool (http://www.sanger. ac.uk/htgt/wge/) ( Figure S1). Genotyping was undertaken by PCR as we previously described 18,19 with the primers listed in Table S1.
Representative genotyping results are shown in Figure S1B. Male BALB/c nude mice were obtained from Hunan SJA Laboratory Animal Co. Ltd. Mice were maintained in a specific pathogen-free, temperature controlled (22 ± 1°C) animal facility with a 12/12-h light/dark cycle, and free access to water and food. Animals were For the DEN-induced HCC mouse model, male mice were intraperitoneally injected with 25 mg/kg DEN (Sigma) at 2 weeks old and killed at 40 weeks old. 20,21 Tumors with diameter of 3 mm or more were counted and measured, and liver and serum were collected.
Liver samples of mice treated with LPS, and HCC samples of high fat diet plus DEN-treated male offspring from a multigenerational maternal obesity model were collected as we previously reported. 18,21 For the nonalcoholic steatohepatitis model, mice were fed an MCD diet (HFK Bioscience) for 4 weeks.

| Histological, immunohistochemical and immunofluorescent studies
Paraffin-embedded mouse liver samples were sectioned and stained with H&E. 22 For immunohistochemical studies, mouse liver sections were incubated overnight at 4°C with primary Abs for H1.2, Ki-67, F4/80, CD3, or Ly6G; a human HCC tissue microarray comprising 15 pairs of tumor and paratumor tissues (Outdo Biotech), was incubated overnight at room temperature with Ab for H1.2 or p-STAT3 Y705 .
Detailed information regarding the Abs used is provided in Table S2.
Positive staining was visualized by 3,3'-diaminobenzidine substrate following the ABC kit (both from Vector Laboratories). For mouse liver sections, positively stained areas or cells were quantified using ImagePro Plus software (Media Cybernetics) based on four to six different randomly selected fields per sample. For human HCC tissue microarray, a semiquantitative analysis was carried out to evaluate H1.2 or p-STAT3 Y705 levels in a double-blinded fashion and scored from 0 to 5, which correspond to the percentage of positively stained cells per field (0%, 1%-20%, 20%-40%, 40%-60%, 60%-80%. and 80%-100%, respectively). Pearson's correlation coefficient analysis between H1.2 and p-STAT3 Y705 levels was carried out using GraphPad Prism version 8.0.
Immunofluorescent staining was undertaken as we previously reported. 23 Mouse liver sections or HepG2 cells were incubated with primary Abs for H1.2, or TIM23 (a protein located in the inner mitochondrial membrane 24 ), or p-STAT3 Y705 , with images taken by a TCS SP8 confocal microscope (Leica). Detailed information on Abs used is provided in Table S2.

| RNA sequencing
Total RNA of tumor samples from different groups was isolated and prepared for RNA sequencing as we previously described. 19,21 Sequencing was undertaken by Genewiz, with details provided in the supporting information. To knock down or overexpress STAT3, cells were transiently transfected with two different shRNAs targeting STAT3 (Table S1), WT STAT3 (STAT3-Flag), a constitutive activated STAT3 (STAT3 Y640F -Flag), or a transactivation-deficient STAT3 (STAT3 Y705F -Flag), respectively. Primary hepatocytes were isolated from male mice as we described previously. 25 For IL-6 treatment, cells were starved overnight in medium without FBS, then treated with 50 ng/ml recombinant human IL-6 (Peprotech) for the indicated times.

| Colony formation and MTT assays
Cells were transiently transfected with indicated plasmids; 48 h later, cells were stimulated with or without 50 μM S3I (Selleck) or 10 μM BP (TargetMol) for another 72 h. Standard MTT and colony formation assays were carried out as we described previously. 21,26

| Tumor xenograft
Both flanks of nude mice (6 weeks old) were subcutaneously injected with 1 × 10 7 LM3 cells transfected with empty vector or pLVX-H1.2, and counted as day 0. S3I (5 mg/kg body weight) was injected intraperitoneally daily from day 9 after cell injection. Tumor sizes were measured every other day. Mice were killed at day 17, tumors were removed and weighed, and volumes were calculated as described. 21,27

| Western blot analyses and qPCR
Freshly isolated liver or cultured cells were sonicated in ice-cold RIPA buffer (Beyotime) and protein concentrations were determined. Western blots were probed with respective primary Abs (Table S2), visualized by enhanced chemiluminescence (Advansta), and quantitated using Quantity One software (Bio-Rad). The relative levels of targeted proteins were quantitated to the respective loading control in the same sample.
RNA from each sample was reverse transcribed into cDNA using an M-MLV First Strand Kit (Invitrogen). Actb for human cell lines or Rn18s for mouse samples was used as internal control. Primers used are provided in Table S1.

| Coimmunoprecipitation and ChIP
Coimmunoprecipitation and ChIP assays were carried out as we previously described. 18,28,29 Primers and Abs used are provided in Tables S1 and S2, and three different regions of human H1C or mouse H1c promoter ranging from −2000 bp to the TSS were chosen.

| Statistical analysis
Results were expressed as the mean ± SD. All the cell experiments were repeated at least three times. Data were analyzed using the Kruskal-Wallis test followed by the Mann-Whitney test for more than two-group comparison, and the Mann-Whitney test was used for two-group comparison. Differences were considered statistically significant at p < 0.05.

| Histone H1.2 is upregulated in liver of HCC patients and mice
H1.2 has the highest transcriptional level among somatic H1 variants in human or mouse liver ( Figure S2A Figure 1B). Similarly, significantly increased H1.2 levels were found in the tumors of DEN-treated mice, compared to that in the liver of age-and sex-matched controls; most of H1.2 staining was found in the nuclei but not in the mitochondria (suggested by costaining with DAPI or TIM23, respectively) ( Figures 1C and S2D,E).
Furthermore, in DEN-treated mice, significantly higher H1.2 levels were observed in tumors compared with the paratumor tissues ( Figure 1D).

| Downregulation of H1.2 mitigates HCC development in DEN-treated mice
To explore whether H1.2 is an oncogenic factor for HCC, we challenged the H1c knockout (H1c KO) or insufficient (H1c KD) mice, as well as their WT littermates, with DEN ( Figure 2A

| Downregulation of H1.2 suppresses hepatic inflammation and STAT3 signaling in DENtreated mice
To investigate how H1.2 promotes HCC development, RNA sequencing was carried out using tumor samples. Compared to WT mice, 4973 and 2342 differentially expressed genes (>1.5-fold change) were identified in H1c KD and H1c KO mice, respectively ( Figure 3A, Tables S3 and S4); among 1789 overlapped genes, most genes were associated with inflammatory response (Figures 3A,B and S4A, Table S5). Therefore, two classic inflammation-related signaling pathways, STAT3 and nuclear factor-κB, were further examined.
Knockout or KD of H1c significantly inhibited phosphorylation of STAT3 Y705 but not total STAT3 level in tumor and paratumor tissues ( Figure 3C,D), and showed no effect on the level of phosphorylated p65 or IκBα ( Figure S4B,C). Furthermore, the transcriptional levels of several STAT3 targeted genes, such as Il-6, Socs3, Fos, and Mcl-1 (MCL1 apoptosis regulator), were decreased in H1c KD or H1c KO mice compared to those of WT mice ( Figure 3E). Consistently, the cell number of positively stained F4/80 (a macrophage marker 32 ), or CD3 (a T cell marker 33 ), or Ly6G (a neutrophil marker 34 ), was significantly lower in the liver of H1c KD or H1c KO mice compared to that of the WT mice following DEN stress ( Figure 3F). Collectively, these results indicated that downregulated STAT3 activation and associated inflammation could contribute to tumor suppression in H1.2 insufficient and deficient mice.
To explore how H1.2 regulates STAT3, we first investigated whether H1.2 might regulate some secretory factor(s) that affect STAT3 phosphorylation. Medium from shScram or shH1.2 transfected HepG2 cells, with or without IL-6 treatment, were collected and used to treat HepG2 cells ( Figure S5A). Compared with the observation that H1.2 KD downregulated p-STAT3 Y705 level ( Figure 4A), no obvious effect on p-STAT3 Y705 level was found in cells treated with medium from H1.2 KD cells ( Figure S5B). Next, we found the effects of H1.2 KD on the p-STAT3 Y705 level were retained after inhibiting protein translation by cycloheximide ( Figure S5C). Coimmunoprecipitation studies indicated no interaction between H1.2 and TC-PTP, which regulates STAT3 dephosphorylation in the nucleus 35 (data not shown), while no significant alteration in total TC-PTP level was found following H1.2 KD, with or without IL-6 treatment ( Figure S5D). As nuclear localization of STAT3 has been reported to protect p-STAT3 from dephosphorylation, 36 we then investigated whether H1.2 might regulate STAT3 by affecting its nuclear translocation. H1.2 KD increased STAT3 level in cytoplasm and reduced nuclear STAT3 level ( Figure 4I), and coimmunoprecipitation showed interaction between H1.2 and KPNA2, an importin mediating nuclear translocation of STAT3 ( Figure 4J). 15 Together, these results suggested that H1.2 could regulate STAT3 activation by affecting its nuclear translocation, but not by a noncell-autonomous manner, or by regulating new protein synthesis or TC-PTP level.

| HIST1H1C is a target of activated STAT3
We also noticed that IL-6 treatment significantly upregulated the H1.2 protein levels in HepG2 cells or mouse primary hepatocytes ( Figure 4A,B), which made us wonder how IL-6 affects H1.2 levels.
In addition to STAT3 activation, a time-dependent upregulation of H1.2 was also observed in IL-6-treated primary mouse hepatocytes ( Figure 5A). Significantly increased protein level of H1.2 was also observed after IL-6 treatment in HepG2 cells, and most of H1.2 was found in the nuclei but not in the mitochondria (Figures 5B and S6A).
As STAT3 is a transcription factor regulating multiple genes, 14 we hypothesized that H1C might also be regulated by STAT3. By analyzing the promoter region (from −2000 bp to the TSS) using JASPAR software ( Figure 5C), 40 potential STAT3 binding sites were suggested. A ChIP assay further confirmed accumulation of STAT3 on the promoters of human H1C and mouse H1c, while IL-6 stimulation increased bindings ( Figures 5D-G and S6B,C). Luciferase reporter assays showed that STAT3 KD greatly reduced human H1C promoter activity by 63%-71%, while mutating six putative STAT3-binding sites similarly reduced the activity of H1C promoter ( Figure 5H). Further STAT3 KD only showed mild effects on mutant H1C promoter activity (16%-24%) ( Figure 5H). These results suggested H1C as a direct target of STAT3.
The mRNA and protein levels of H1.2 were upregulated in 293T cells (a cell line with high transfection efficiency) overexpressing WT STAT3 or STAT3 Y640F (Figure 5I-L). Furthermore, either mRNA or protein levels of H1.2 were significantly downregulated in the STAT3 KD 293T cells ( Figure 5M,N). Consistently, similar results were found in HepG2 cells when overexpressing STAT3 Y640F or knocking down STAT3 ( Figure S5D-G).

| H1.2 is positively correlated with p-STAT3 level in hepatic inflammation
To confirm upregulation of a H1.2/STAT3 axis in hepatic inflammatory stresses, we further examined the correlation between H1.2 and p-STAT3 in MCD diet-induced nonalcoholic steatohepatitis or LPS-induced acute inflammatory liver injury. Significantly upregulated nuclear H1.2 levels, indicated by immunohistochemistry, and significantly upregulated H1.2 and p-STAT3 Y705 levels, indicated by western blots, were found in the liver of MCD-or LPS-treated mice ( Figure 6A-D). Moreover, we investigated H1.2 and p-STAT3 Y705 levels in tumor samples from DEN-induced male offspring from a multigenerational maternal obesity model, which has shown gradually increased hepatic inflammation over generations, as we previously reported. 21,37,38 Under DEN stress, compared with normal chow or high-fat diet treated male offspring from lean mothers (NCD or HFD1D), the H1.2 and p-STAT3 Y705 levels of high-fat diet treated male offspring from obese mothers or obese grandmothers (HFD2D or HFD3D) were both significantly upregulated ( Figure 6E). Importantly, in human HCC samples, a positive correlation between the levels of nuclear H1.2 and p-STAT3 was found ( Figure 6F).

| DISCUSS ION
In this study, we identified that H1.2 was the most abundant linker histone among five somatic H1 variants in human and mouse liver ( Figure S2). H1.2, rather than other variants, upregulated and aggravated hepatocarcinogenesis (Figures 1 and 2). Here, we identified that H1.2 suppressed inflammatory response in the liver by regulating the level of p-STAT3 Y705 and its downstream genes ( Figure 3). Previously, we have reported that upregulation of H1.2 is associated with increased expression of inflammatory factors in diabetic retinopathy 9 ; however, we were unclear whether H1.2 directly regulated inflammation. In this study, we showed that H1.2 directly affected inflammatory response in hepatocytes by regulating the protein level and nuclear distribution of activated form of STAT3 (p-STAT3 Y705 ), subsequently influencing its transcriptional activity (Figure 4). Similar pro-inflammatory roles of hepatic STAT3 have also been found in other liver diseases, such as viral hepatitis, nonalcoholic steatohepatitis, and chemical-induced liver injury. [43][44][45] Consistently, upregulation of the H1.2/STAT3 axis was also found in MCD-induced nonalcoholic steatohepatitis and LPSinduced acute inflammatory liver injury ( Figure 6). Considering the commonly found constitutive STAT3 activation in many cancers, 46 the existence and impacts of the H1.2/STAT3 axis in other types of cancer will be interesting for further exploration.
As H1.2 locates and is upregulated in the nuclei of HCC tissues and cell lines examined ( Figures 2C,D and S6A), it is plausible that H1.2 might regulate STAT3 activity by affecting some nuclear events.
In conclusion, we found that, following DEN induction, upregulated H1.2 drives HCC development by regulating STAT3 activation to promote inflammatory response and cell proliferation.