Histone Variant H2A.J Is Enriched in Luminal Epithelial Gland Cells

H2A.J is a poorly studied mammalian-specific variant of histone H2A. We used immunohistochemistry to study its localization in various human and mouse tissues. H2A.J showed cell-type specific expression with a striking enrichment in luminal epithelial cells of multiple glands including those of breast, prostate, pancreas, thyroid, stomach, and salivary glands. H2A.J was also highly expressed in many carcinoma cell lines and in particular, those derived from luminal breast and prostate cancer. H2A.J thus appears to be a novel marker for luminal epithelial cancers. Knocking-out the H2AFJ gene in T47D luminal breast cancer cells reduced the expression of several estrogen-responsive genes which may explain its putative tumorigenic role in luminal-B breast cancer.


Introduction
The diploid human genome of 6 billion base pairs of DNA is packaged into cell nuclei 10 µm in diameter by histone binding and condensation into nucleosomes. Nucleosomes comprise 145-147 bp of DNA wrapped around octamers containing two of each of the four core histone families-H2A, H2B, H3 and H4. A minimum of 20 bp of DNA bound to the linker histone H1 extends between nucleosomes. In addition to their roles in compacting DNA and organizing chromatin structure, histones are regulators of gene expression, DNA replication and DNA repair amongst others. Histones are defined as either canonical or variant based on their expression, their functional roles and genomic assignment (single gene versus several gene copies in clusters). Canonical histones are highly expressed in S phase to allow repackaging of the genome coupled to DNA replication. Transcriptional activation of the canonical histone genes and selective stabilization of their mRNAs account for this S-phase enrichment [1]. Canonical histone mRNAs typically terminate with a stem-loop sequence that is recognized by the SLBP (stem-loop binding protein) instead of a poly A sequence [2]. Binding of the stem-loop sequence by SLBP is required for maturation, stabilization, and translation of the canonical histone mRNAs. SLBP is degraded outside of S-phase and this in turn destabilizes the histone stem-loop RNAs. In contrast to the canonical histones, genes coding for histone variants are expressed independently of DNA replication and their mRNAs terminate in polyadenylation sequences. Histone variant expression is thus particularly important in differentiated post-mitotic cells. However, there have been few systematic studies of tissue-specific expression of histone variants. Histone variants often have specialized functions [3]. Mammals have eleven H1 variants, nine H2A variants, five H3 variants, at least two testis-specific H2B variants, and one H4 variant [4].
The core histone H2A has the largest number of replication-independent variants. The H2A.X, H2A.Z, and macro-H2A variants have been extensively studied in recent years, but H2A.J has received less attention and its precise roles are poorly understood. H2A.J seems to be specific to mammalian cells and differs from canonical H2A only by a valine in place of alanine at position 10 of the mature protein species, and the last seven amino acids that contain a potentially phosphorylatable SQ motif [5]. H2A.J has been implicated in the expression of inflammatory genes in senescent fibroblasts [5,6]. H2A.J over-expression has also been implicated in luminal-B breast cancer tumorigenesis [7,8] and in the chemotherapeutic resistance of colorectal [9], hepatic [10], and glioblastoma cancers [11]. We performed a global survey of H2A.J expression in human and mouse tissues to obtain initial indications regarding potential tissue-specific functions.

Cell Culture and Reagents
T47D cells were obtained from the NCI-60 cancer cell line repository. Cells were cultured in RPMI-1640 medium supplemented with GlutaMAX (Thermo-Fisher Scientific 61870036), 10% fetal bovine serum, and penicillin (50 U/mL) + streptomycin (0.2 mg/mL) (Sigma P0781, Saint-Quentin-Fallavier, France) in a 5% CO 2 incubator. For experiments involving estrogen deprivation and feedback, T47D cells were first cultivated for three days in phenol red-free RPMI-1640 medium (cat# 11835030, Life Technologies; Grand Island, NY, USA) containing 5% charcoal-dextran-treated FBS (cat# A33821-01, Life Technologies; Grand Island, NY, USA) (CDFBS). Cells were harvested after two days of treatment with 10 nM estradiol and washed twice with PBS prior to whole-cell protein extraction. βestradiol was purchased from Sigma Aldrich (cat# E8875-250MG, Saint Louis, MO, USA) and prepared as a 20 mM stock solution in DMSO. Prior to use, β-estradiol was diluted to 10 µM in ethanol (×1000 solution).

Crispr-Cas9 Knock-Out of the H2AFJ Gene
The H2AFJ gene deletion was performed using the CRISPR/Cas9 system as previously described [12]. Briefly, the H2AFJ-specific guide RNA sequence (GAGGTGATCATGTC-CGGTCG) was chosen from http://crispr.mit.edu (Synthego tool, last accessed on 18 October 2021), and inserted into the pX330 vector (Addgene, Cambridge, MA, USA). For the donor plasmid, the left and the right homology arms were PCR-amplified from human primary fibroblast genomic DNA. The two sets of primers targeting human H2AFJ are listed as follows: for the upper region, H2AFJ_trgt_FW1 (gcgttcaacattacatcacg) and H2AFJ_trgt_RV2 (gtaaggtactccaacaccgc); and for the bottom region, H2AFJ_trgt_SI_FW2 (gtctggGTCGACgcgcgtgacaacaagaagac) and H2AFJ_trgt_ApI_RV1 (gtaataGGGCCCacgactaaaggcgctgc), respectively. Both arms along with antibiotic resistance ORF (hygromycin) were inserted between EcoRI and ApaI sites of the PCR2.1 vector (Life Technologies; Grand Island, NY, USA). Selection of H2AFJ-KO clones was performed by growth in the presence of 400 µg/mL hygromycin. Expression of a cDNA encoding Flag-HA-H2A.J from the tet-ON promoter was achieved by infecting H2A.J-KO cell lines with the previously described pTRIPz lentiviral-based construct [5].

Immunohistochemical Detection of H2A.J in Mouse and Human Tissue Sections
Characterization of the H2A.J antibody and its use in immunohistochemical staining was previously described [5,13]. Similar results were obtained with an anti-H2A.J antibody commercialized by Active Motif (catalog # 61793) that was also prepared to the unique C-terminus of H2A.J.
Normal (catalog # BN242 and BN243) and breast cancer (catalog# BR963a) human tissue microarrays were obtained from US Biomax (Rockville, MD, USA). Quantification of the breast cancer microarray was done by classifying the staining on a scale from 0 (no staining) to 4 (highest staining) and then calculating the mean for ER-positive and ER-negative biopsies. Mouse tissue sections were prepared as previously described [5].

Immunofluorescent Detection of H2A.J in Cultured Cells
Cancer cells in culture were fixed with 4% paraformaldehyde and prepared for immunofluorescence with 1/500 dilution of H2A.J antibody as previously described [5,14].

Western Blots
For H2A.J Western blots, whole cell extracts were prepared by resuspending trypsinized cells in SDS-PAGE sample buffer and heating at 95 • C for 15 min. Viscosity of genomic DNA was reduced by sonication with a probe sonicator. Extracts were electrophoresed in SDS-12% polyacrylamide gels and proteins were transferred to nitrocellulose membranes for detection of proteins with the Li-Cor Odyssey imager. The rabbit H2A.J antibody was used at a dilution of 1/500 in Odyssey blocking buffer. For TFF1 Western blots, cells were washed quickly twice with PBS, directly solubilized in denaturing sample buffer supplemented with 5% β-mercaptoethanol (200 µL/1 million cells), boiled for 5 min and then subjected to SDS polyacrylamide gel electrophoresis. Proteins were electrotransferred to 0.45 µm PVDF sheets (cat# IPVH304F0, Millipore, MA, USA) for immunodetection with the following primary antibodies: TFF1, Cat# 15571S, Cell Signaling technology (Danvers, MA, USA), and β-actin (cat# A2228-200UL, Sigma-Aldrich, Saint Louis, MO, USA).

Isolation of H2A.J-KO Mice
Cyagen injected fertilized eggs of C57BL/6-N mice with a TALEN nuclease targeting the 5 end of the H2afj gene, and used these eggs to impregnate surrogate mothers. Sequencing of tail DNA from the resulting pups identified a male mouse with a heterozygous 7-nucleotide deletion at the beginning of the H2afj gene (∆12-18 starting at 1 from the ATG) that would code for a frameshifted protein with a premature stop codon. The hetererozygous mouse was mated with a B6-N female that yielded male and female mutant offspring. Mating of male and female heterozygotes led to the isolation of male and female homozygous H2afj-∆7 mutant mice. We were unable to detect H2A.J protein by Western blots, immunofluorescence, and mass spectrometry analyses of organs and MEFs derived from these mutant mice [15]. Tissue sections from these H2A.J-KO mice were used to show the specificity of our IHC detection of H2A.J in C57BL/6-N WT mice. A full phenotypic characterization of the H2A.J-KO mice will be published at a later date.
2.7. RNA-seq of T47D and T47D-H2A.J-KO Cell Lines RNA was isolated from 3 replicates each of T47D cells and 2 different Crispr-Cas9 H2A.J-KO clones (KO-1 and KO-18) under conditions of proliferation and after treatment with 5 µM tamoxifen for 7 days. RNA was purified with a Machery-Nagel Nucleospin RNA Plus kit. Poly A+ RNAs were purified and converted into cDNA for sequencing with the Illumina TruSeq stranded mRNA kit. The library preparation and 43 nucleotide paired-end DNA sequencing were performed by the I2BC high-throughput DNA sequencing platform.

Immunolocalization Shows Enrichment of H2A.J in Luminal Epithelial Cells of Multiple Glands in Mice and Humans
We previously noted some organ-specific differences in H2A.J protein levels in the mouse by mass spectrometry [5]. Here, we used immunohistochemistry (IHC) with H2A.Jspecific antibodies to visualize H2A.J expression within different cell types in adult mouse and human tissue sections ( Figure 1). We previously demonstrated the specificity of these antibodies [5,13], and we provide further confirmation by comparing tissue sections from WT and H2A.J-KO mice and showing an example of competition of the H2A.J signal by an excess of the C-terminal H2A.J-specific peptide that was used as the immunogen to prepare the antibody (Supplementary Figure S1). We observed striking cell-type specific variation in H2A.J expression (Figure 1 and Supplementary Figure S1). In particular, we saw increased levels of H2A.J in luminal epithelial cells of multiple glands, including mammary, prostate, thyroid, pancreas, stomach, bladder, and salivary glands. Similar staining was observed for the corresponding human and mouse tissues.
We next mined transcriptome databases to see whether differences in H2A.J protein levels by immunohistochemistry were correlated with differences in H2AFJ RNA levels in different tissues. First, poly A+ RNA-seq data in transcripts per million (TPM) were analyzed for 27 different tissues from the GTEx (Genotype-Tissue Expression) Portal ( Figure 2A) [29]. This subset contained TPM data for 9 to 2642 recently deceased generally non-diseased human donors depending on tissue type (Supplementary Table S1). These data were not normalized to an external RNA spike-in, and so can only be used as a qualitative indication of relative RNA expression between tissues. Genes 2021, 12, x FOR PEER REVIEW 5 of 20 We next mined transcriptome databases to see whether differences in H2A.J protein levels by immunohistochemistry were correlated with differences in H2AFJ RNA levels in different tissues. First, poly A+ RNA-seq data in transcripts per million (TPM) were analyzed for 27 different tissues from the GTEx (Genotype-Tissue Expression) Portal (Figure 2A) [29]. This subset contained TPM data for 9 to 2642 recently deceased generally non-diseased human donors depending on tissue type (Supplementary Table S1). These data were not normalized to an external RNA spike-in, and so can only be used as a qualitative indication of relative RNA expression between tissues. The testis showed the highest H2AFJ expression followed by prostate, adipose tissue, and breast. The relatively high H2AFJ RNA levels in the prostate and breast are consistent with the strong luminal epithelial staining in the prostate and breast tissue ( Figure 1A-D). Adipocyte nuclear staining was also seen in the mouse mammary tissue (Supplementary Figure S1). IHC of the mouse testis showed strong staining mainly of the spermatogonia (Supplementary Figure S1). Other tissues showing strong luminal epithelial IHC staining such as the pancreas, salivary glands, bladder, and stomach, had somewhat lower levels of H2AFJ RNA levels. Mass spectrometry indicated that the mouse liver and kidney contained significant amounts of H2A.J [5], despite containing relatively modest levels of H2AFJ RNA according to the human GTEx data (Figure 2A). IHC showed that H2A.J was expressed in hepatocyte nuclei and enriched in nuclei of the kidney cortex (Supplementary Figure  S1). Finally, tissues such as the heart and brain showed low levels of H2AFJ RNA, and IHC indicated that there were low levels of H2A.J in cardiomyocytes and the cerebrum (Supplementary Figure S2). Overall, these data suggest that there is tissue-specific variation in H2AFJ RNA expression that contributes to differences in H2A.J protein expression, but some type of post-transcriptional regulation may be required to explain why tissues such as the liver and kidney appear to contain significant levels of H2A.J protein [5] despite relatively low H2AFJ RNA levels. The testis showed the highest H2AFJ expression followed by prostate, adipose tissue, and breast. The relatively high H2AFJ RNA levels in the prostate and breast are consistent with the strong luminal epithelial staining in the prostate and breast tissue ( Figure 1A-D). Adipocyte nuclear staining was also seen in the mouse mammary tissue (Supplementary Figure S1). IHC of the mouse testis showed strong staining mainly of the spermatogonia (Supplementary Figure S1). Other tissues showing strong luminal epithelial IHC staining such as the pancreas, salivary glands, bladder, and stomach, had somewhat lower levels of H2AFJ RNA levels. Mass spectrometry indicated that the mouse liver and kidney contained significant amounts of H2A.J [5], despite containing relatively modest levels of H2AFJ RNA according to the human GTEx data ( Figure 2A). IHC showed that H2A.J was expressed in hepatocyte nuclei and enriched in nuclei of the kidney cortex (Supplementary Figure S1). Finally, tissues such as the heart and brain showed low levels of H2AFJ RNA, and IHC indicated that there were low levels of H2A.J in cardiomyocytes and the cerebrum (Supplementary Figure S2). Overall, these data suggest that there is tissue-specific variation in H2AFJ RNA expression that contributes to differences in H2A.J protein expression, but some type of post-transcriptional regulation may be required to explain why tissues such as the liver and kidney appear to contain significant levels of H2A.J protein [5] despite relatively low H2AFJ RNA levels.
For these same 27 tissues, we also compared GTEx data for H2AFJ relative to the 5 other somatic cell H2A histone variant genes (H2AFX, H2AFV, H2AFZ, H2AFY, and H2AFY2, encoding H2A.X, H2A.Z1, H2A.Z2, macroH2A.1 and macroH2A.2 respectively) (Supplementary Figure S3). The mean and standard deviation of the median TPM values for these RNAs in all tissues show that the H2AFZ gene is expressed at the highest overall levels, followed by H2AFV, H2AFJ, and H2AFX, and then the two macroH2A genes (Figure 2B). H2AFZ and H2AFJ also showed the greatest tissue-specific variation in RNA levels ( Figure 2B and Supplementary Figure S3).
Whole tissue RNA and protein determinations do not account for cell-type specific expression within tissues. The IHC images show cell-type specific variation within vari- For these same 27 tissues, we also compared GTEx data for H2AFJ relative to the 5 other somatic cell H2A histone variant genes (H2AFX, H2AFV, H2AFZ, H2AFY, and H2AFY2, encoding H2A.X, H2A.Z1, H2A.Z2, macroH2A.1 and macroH2A.2 respectively) (Supplementary Figure S3). The mean and standard deviation of the median TPM values for these RNAs in all tissues show that the H2AFZ gene is expressed at the highest overall levels, followed by H2AFV, H2AFJ, and H2AFX, and then the two macroH2A genes ( Figure 2B). H2AFZ and H2AFJ also showed the greatest tissue-specific variation in RNA levels ( Figure 2B and Supplementary Figure S3).
Whole tissue RNA and protein determinations do not account for cell-type specific expression within tissues. The IHC images show cell-type specific variation within various tissues, such as the strong labeling of luminal epithelial cells. IHC of brain slices showed that whereas most cell types including neurons expressed low levels of H2A.J, a minor cell population expressed H2A.J at much higher levels (unpublished data). We examined single-cell RNA databases for human and mouse tissues to check for cell-type specific variation in H2AFJ expression.

Single-Cell RNA-seq Data Show High H2AFJ RNA Expression in Mammary and Prostate Luminal Epithelial Cells
We used previously published single-cell RNA-seq data to examine H2AFJ expression in mammary epithelial cells in female humans and mice [23,24]. 24,280 mammary epithelial cells from 4 healthy women aged 17-36 years old were grouped into 3 clusters by UMAP (uniform manifold approximation and projection): a basal-myoepithelial cluster (labeled B) and 2 luminal cell clusters (L1 and L2) ( Figure 3A,B) [24]. The L1 cluster was characterized as secretory type luminal cells and the L2 cluster as hormone-responsive luminal cells [24].
Comparing H2AFJ expression with that of marker genes for basal (KRT14), L1 (SLPI), L2 (ANKRD30A), and both L1 and L2 (KRT18), revealed that H2AFJ is preferentially expressed in both L1 and L2 clusters compared to basal/myo-epithelial cells ( Figure 3A). This result is evident as well from a heat map of gene expression comparing H2AFJ expression with that of the top 10 marker genes for each of these clusters ( Figure 3B). The preferential expression of H2afj in luminal mammary cells is conserved in the mouse as seen with an integrated dataset of 50,407 epithelial cells from 5 independent studies leading to the definition of 6 cell clusters: a mammary stem/progenitor cell cluster (MaSc-pro), a basal cell cluster (Basal), a luminal hormone-sensitive cell cluster (L-Hor), a progenitor cluster for the L-Hor cells (LH-pro), a luminal alveolar cell cluster (L-Alv), and a progenitor cluster for the L-Alv cells (LA-pro) [23]. H2afj expression was higher in all the luminal clusters compared to the MaSc-pro and Basal clusters, with the highest expression in the hormone-sensitive luminal clusters ( Figure 3C). These data suggest that the enhanced H2A.J protein levels in luminal epithelial mammary cells ( Figure 1A,B) is due at least in part to increased levels of H2afj RNA expression in these cells.

Luminal Epithelial Cells
We used previously published single-cell RNA-seq data to examine H2AFJ expression in mammary epithelial cells in female humans and mice [23,24]. 24,280 mammary epithelial cells from 4 healthy women aged 17-36 years old were grouped into 3 clusters by UMAP (uniform manifold approximation and projection): a basal-myoepithelial cluster (labeled B) and 2 luminal cell clusters (L1 and L2) ( Figure 3A,B) [24]. The L1 cluster was characterized as secretory type luminal cells and the L2 cluster as hormone-responsive luminal cells [24]. Comparing H2AFJ expression with that of marker genes for basal (KRT14), L1 (SLPI), L2 (ANKRD30A), and both L1 and L2 (KRT18), revealed that H2AFJ is preferentially expressed in both L1 and L2 clusters compared to basal/myo-epithelial cells ( Figure 3A). This result is evident as well from a heat map of gene expression comparing H2AFJ expression with that of the top 10 marker genes for each of these clusters ( Figure  3B). The preferential expression of H2afj in luminal mammary cells is conserved in the mouse as seen with an integrated dataset of 50,407 epithelial cells from 5 independent studies leading to the definition of 6 cell clusters: a mammary stem/progenitor cell cluster (MaSc-pro), a basal cell cluster (Basal), a luminal hormone-sensitive cell cluster (L-Hor), a progenitor cluster for the L-Hor cells (LH-pro), a luminal alveolar cell cluster (L-Alv), and a progenitor cluster for the L-Alv cells (LA-pro) [23]. H2afj expression was higher in all the luminal clusters compared to the MaSc-pro and Basal clusters, with the highest expression in the hormone-sensitive luminal clusters ( Figure 3C). These data suggest that the enhanced H2A.J protein levels in luminal epithelial mammary cells ( Figure 1A,B) is due at least in part to increased levels of H2afj RNA expression in these cells. We further examined H2AFJ expression in 28,606 single normal prostate cells from 3 men aged 18-31 years old [25]. As for the mammary glands, H2AFJ RNA expression was highest in luminal epithelial cells and low in basal epithelial cells of the prostate (Supplementary Figure S4). Luminal expression of H2AFJ was similar to the KLK3 gene encoding the prostate cancer antigen that is a marker of prostate luminal epithelial cells. However, whereas KLK3 is highly restricted to luminal epithelial cells (Supplementary Figure S4D,E), there is significant expression of H2AFJ in other prostate cell clusters with epithelial club cells that have an immune-secretory transcriptome [25] showing the next highest level of expression (Supplementary Figure S4C). Single-cell RNA-seq of the mouse prostate gland [26] also showed highest expression of H2afj in luminal epithelial cells (Supplementary Figure S4F). Thus, differential expression of H2A.J in luminal mammary and prostate cells is correlated with higher levels of H2AFJ RNA in these cells.

H2A.J Is a Marker of Luminal Breast and Prostate Cancer Cells
H2AFJ was reported to be over-expressed in some cancers [7][8][9]11], and interrogation of the Cancer Genome Atlas (TCGA) database showed that prostate and breast cancer expressed high levels of H2AFJ mRNA (Figure 4, left panel). H2AFJ RNA was also highly expressed in many cancer cell lines derived from glandular tissues such as the breast, pancreas, lung, colorectum, thyroid, esophagus, stomach, and prostate according to the Broad Cancer Cell Line Encyclopedia (CCLE) (Figure 4, right panel).
We further examined H2AFJ expression in 28,606 single normal prostate cells from 3 men aged 18-31 years old [25]. As for the mammary glands, H2AFJ RNA expression was highest in luminal epithelial cells and low in basal epithelial cells of the prostate (Supplementary Figure S4). Luminal expression of H2AFJ was similar to the KLK3 gene encoding the prostate cancer antigen that is a marker of prostate luminal epithelial cells. However, whereas KLK3 is highly restricted to luminal epithelial cells (Supplementary Figure  S4D,E), there is significant expression of H2AFJ in other prostate cell clusters with epithelial club cells that have an immune-secretory transcriptome [25] showing the next highest level of expression (Supplementary Figure S4C). Single-cell RNA-seq of the mouse prostate gland [26] also showed highest expression of H2afj in luminal epithelial cells (Supplementary Figure S4F). Thus, differential expression of H2A.J in luminal mammary and prostate cells is correlated with higher levels of H2AFJ RNA in these cells.

H2A.J Is a Marker of Luminal Breast and Prostate Cancer Cells
H2AFJ was reported to be over-expressed in some cancers [7][8][9]11], and interrogation of the Cancer Genome Atlas (TCGA) database showed that prostate and breast cancer expressed high levels of H2AFJ mRNA (Figure 4, left panel). H2AFJ RNA was also highly expressed in many cancer cell lines derived from glandular tissues such as the breast, pancreas, lung, colorectum, thyroid, esophagus, stomach, and prostate according to the Broad Cancer Cell Line Encyclopedia (CCLE) (Figure 4, right panel).  Interestingly, immunofluorescence experiments showed that H2A.J was expressed at much higher levels in breast and prostate cancer cell lines of luminal cell origin [30,31] compared to basal or non-luminal cell origin [30][31][32] (Figure 5A-C), and Western blotting confirmed this result for the breast cancer lines ( Figure 5D). Since we observed high expression of H2A.J in normal breast and prostate luminal epithelial cells compared to basal cells (Figures 1 and 2), this observation suggests that luminal cancer cells retain the high H2AFJ expression of their cell of origin and that H2A.J expression might represent an additional marker of luminal epithelial cancer cells.
Interestingly, immunofluorescence experiments showed that H2A.J was expressed at much higher levels in breast and prostate cancer cell lines of luminal cell origin [30,31] compared to basal or non-luminal cell origin [30][31][32] (Figure 5A-C), and Western blotting confirmed this result for the breast cancer lines ( Figure 5D). Since we observed high expression of H2A.J in normal breast and prostate luminal epithelial cells compared to basal cells (Figures 1 and 2), this observation suggests that luminal cancer cells retain the high H2AFJ expression of their cell of origin and that H2A.J expression might represent an additional marker of luminal epithelial cancer cells. Consistent with previous reports [7], breast cancer transcriptome databases [33] show that H2AFJ RNA expression was highest in luminal-B breast cancer cells, followed by luminal-A, HER2-E, Normal breast-like, and Basal-like ( Figure 6). of the indicated cancer cell lines. ß-actin was used as a loading control. The numbers under the panels represent the intensity ratios for H2A.J/ß-actin signals normalized to the T47D ratio. Ub-H2AJ refers to a putative ubiquitinated form of H2A.J. Consistent with previous reports [7], breast cancer transcriptome databases [33] show that H2AFJ RNA expression was highest in luminal-B breast cancer cells, followed by luminal-A, HER2-E, Normal breast-like, and Basal-like ( Figure 6).  Figure S5). Luminal breast cancers are typically ER-positive whereas basal breast cancers are typically ER-negative [30]. Quantification of the H2A.J signal showed significantly higher levels of H2A.J staining for the ER-positive breast cancers that is consistent with the higher H2AFJ RNA levels for luminal versus basal breast cancers ( Figure 6). This result suggests that H2A.J represents a novel IHC marker for luminal ER-positive breast cancers.  Figure S5). Luminal breast cancers are typically ER-positive whereas basal breast cancers are typically ER-negative [30]. Quantification of the H2A.J signal showed significantly higher levels of H2A.J staining for the ER-positive breast cancers that is consistent with the higher H2AFJ RNA levels for luminal versus basal breast cancers ( Figure 6). This result suggests that H2A.J represents a novel IHC marker for luminal ER-positive breast cancers.

Single-Cell RNA Analaysis of H2AFJ Expression in Prostate Cancer Cells
A recent single-cell RNA-seq study of 36,424 cells from 12 patients with prostate cancer identified 16 epithelial cancer cell clusters [34]. Most prostate cancers show a luminaltype gene expression profile [34]. We found that H2AFJ expression in prostate cancer ep-

Single-Cell RNA Analaysis of H2AFJ Expression in Prostate Cancer Cells
A recent single-cell RNA-seq study of 36,424 cells from 12 patients with prostate cancer identified 16 epithelial cancer cell clusters [34]. Most prostate cancers show a luminal-type gene expression profile [34]. We found that H2AFJ expression in prostate cancer epithelial cells was enriched in luminal-type epithelial cells with a profile that was very similar to the KLK3 gene encoding the prostate specific antigen (PSA) that is used as a marker of luminal prostate cancer (Supplementary Figure S6). In particular, violin plots indicated that both H2AFJ and KLK3 had the highest expression levels in prostate cancer epithelial clusters 3, 4, and 9, and showed the lowest expression in the basal cell epithelial cluster (Supplementary Figure S6C,D). Plotting the mean values of H2AFJ versus KLK3 RNA levels for the prostate cancer clusters showed a high correlation coefficient of 0.88 (Supplementary Figure S6E). Thus, increased H2AFJ RNA expression is characteristic of both breast and prostate luminal-type cancers.

H2AFJ Expression in Relation to Breast Cancer Survival
Breast cancers with a high expression of H2AFJ had a slightly higher survival rate compared to breast cancers that expressed low levels of H2AFJ (Figure 8). Since luminal breast cancer cells express H2AFJ at higher levels than basal breast cancer cells, this survival difference may reflect the more aggressive nature of basal versus luminal breast cancer [35]. Separating the breast cancer survival data according to PAM50 intrinsic molecular subtypes, high H2AFJ expression was correlated with a slightly elevated survival for all subtypes except luminal A breast cancer for which it showed significant negative correlation with survival (Supplementary Figure S7).

H2AFJ Expression in Relation to Survival for Other TCGA Cancer Types
Prostate cancer patients with higher than median H2AFJ expression tended to show slightly better survival compared to patients with lower than median H2AFJ, but this difference did not reach statistical significance because there were only 10 death events in The hazard ratio (HR) and its confidence intervals (CI) are shown with the significance p-value. The curves were calculated with bc-GenExMiner v4.7 [27] using the optimal splitting criterion on breast cancer cohorts from TCGA and the Swedish Cancerome Analysis Network-Breast (SCAN-B) Initiative (GSE96058 and GSE81538).

H2AFJ Expression in Relation to Survival for Other TCGA Cancer Types
Prostate cancer patients with higher than median H2AFJ expression tended to show slightly better survival compared to patients with lower than median H2AFJ, but this difference did not reach statistical significance because there were only 10 death events in this TCGA cohort of 498 primary prostate cancer patients (Supplementary Figure S8). Interrogation of all TCGA cancers in the OncoLnc survival database [28] indicated that H2AFJ significantly affected the survival of 4/21 cancer types: high H2AFJ expression was associated with poorer survival for brain lower-grade glioma (LGG), kidney renalcell carcinoma (KIRC), and skin cutaneous melanoma (SKCM), but higher survival for bladder urothelial carcinoma (BLCA) (Supplementary Figure S8). Thus, the effect of H2AFJ expression on cancer survival depends on the cancer type.

Transcriptomic Analysis of T47D H2A.J-KO Cells
To search for a possible function for H2A.J in luminal epithelial cells, we chose to inactivate the H2AFJ gene in T47D luminal breast cancer cells using the CRISPR/Cas9 technology. Two H2A.J-KO clones (KO1 and KO18) lacking detectable expression of H2A.J by Western blotting and immunofluorescence were characterized (Supplementary Figure  S9). We sequenced polyA+ RNAs from 3 replicates of the 2 independent H2A.J-KO cell lines and the parental T47D cells under conditions of proliferation and after treatment with 5 µM tamoxifen for 7 days. Tamoxifen is a selective estrogen-receptor modulator that is used as a first-line treatment to inhibit the growth of breast cancers expressing the estrogen receptor [35]. Treatment of the parental T47D cells and the 2 H2A.J-KO cell lines induced a similar inhibition of cell growth (Supplementary Figure S10). Euclidean distance heat maps of the RNA-seq data indicated that samples were first separated by the conditions of tamoxifen-treated versus untreated proliferating cells and next by genotype (T47D versus H2A.J-KO) (Figure 9).  We first performed differential gene expression analysis for T47D cells treated with tamoxifen versus proliferating cells. 2462 genes showed differential expression (1279 upregulated and 1183 down-regulated genes with FDR < 0.01) for tamoxifen-treated versus proliferating T47D cells ( Figure 10A and Supplementary Table S2). Tamoxifen is thought to induce the proliferative arrest of mammary epithelial cells by modifying the expression of genes regulated by the estrogen receptor transcription factor ESR1 [36]. We performed gene set enrichment analyses (GSEA) using the c2 gene signature and Hallmark collections from the Molecular Signatures Database (MSigDB) [37] to identify gene sets whose We first performed differential gene expression analysis for T47D cells treated with tamoxifen versus proliferating cells. 2462 genes showed differential expression (1279 upregulated and 1183 down-regulated genes with FDR < 0.01) for tamoxifen-treated versus proliferating T47D cells ( Figure 10A and Supplementary Table S2). Tamoxifen is thought to induce the proliferative arrest of mammary epithelial cells by modifying the expression of genes regulated by the estrogen receptor transcription factor ESR1 [36]. We performed gene set enrichment analyses (GSEA) using the c2 gene signature and Hallmark collections from the Molecular Signatures Database (MSigDB) [37] to identify gene sets whose expression is affected by tamoxifen treatment. The gene sets showing the most significant enrichment scores involved down-regulation of genes activated by estradiol in MCF7 breast cancer cells [38], genes involved in cell proliferation (activated by the E2F and Myc transcription factors), and genes encoding ribosomal and protein synthesis factors ( Figure 10B and Supplementary Table S2). In particular, gene set enrichment plots showed a strong anticorrelation in the gene expression effects of estradiol and tamoxifen ( Figure 10C) in keeping with a largely antagonistic action of tamoxifen on estrogen receptor function in luminal breast epithelial cells. Thus, in agreement with previous work, our transcriptome analysis suggests that tamoxifen induces a proliferative arrest of T47D cells by interfering with estrogen receptor transcriptional activity leading to a down-regulation of genes required for proliferation and protein synthesis. , breast cancer or cell proliferation (yellow), protein synthesis (blue), and metabolic process (black). (C) Enrichment plot of tamoxifen-regulated genes compared to the c2 gene set "Dutertre_Estradiol_Response_24HR_UP" [37,38]. NES (normalized enrichment score).
We next compared the gene expression of T47D parental cells with 2 independent H2A.J-KO cell lines in proliferating conditions. We sought genes that were differentially regulated (FDR < 0.05) in a similar manner in the 2 H2A.J-KO cell lines compared to T47D in order to exclude differences in the Crispr/Cas9 H2A.J-KO clones that were unrelated to the H2A.J gene inactivation. 90 genes were down-regulated and 17 genes up-regulated ( Figure 11A,B and Supplementary Table S2). Interestingly, 39/90 (43%) of the genes downregulated in the H2A.J-KO cells corresponded to genes that were down-regulated in T47D cells by tamoxifen treatment, and 9/17 (53%) of genes that were up-regulated in the H2A.J-KO cells were up-regulated in T47D by tamoxifen treatment ( Figure 11C). Furthermore, 27 of the genes that are differentially expressed in the H2A.J-KO cells are also putative direct targets of ESR1 (ADAMTS8, APBB2, CXCL12, FAM189A2, FHL2, FOS, GAS6, GRIK3,  HES1, HSPB8, LRP4, MAPT, NDRG1, NEDD9, NEK10, OLFML3, PDZK1, RBBP8, ROBO2,   Figure 10. Genes differentially expressed upon treatment of T47D cells with tamoxifen. (A) Volcano plot of log fold-change (logFC) versus -log10 of the false discovery rate (FDR) for genes differentially expressed in proliferating versus tamoxifentreated T47D cells. Orange dots represent genes with FDR < 0.01. The indicated genes are those with log2 fold changes of +/− 3 and an average expression of greater than 1 in the limma-voom analysis. (B) Enrichment analysis indicating significant correlations of genes differentially expressed upon tamoxifen treatment with gene sets involved in estradiol response (pink), breast cancer or cell proliferation (yellow), protein synthesis (blue), and metabolic process (black). (C) Enrichment plot of tamoxifen-regulated genes compared to the c2 gene set "Dutertre_Estradiol_Response_24HR_UP" [37,38]. NES (normalized enrichment score).
We next compared the gene expression of T47D parental cells with 2 independent H2A.J-KO cell lines in proliferating conditions. We sought genes that were differentially regulated (FDR < 0.05) in a similar manner in the 2 H2A.J-KO cell lines compared to T47D in order to exclude differences in the Crispr/Cas9 H2A.J-KO clones that were unrelated to the H2A.J gene inactivation. 90 genes were down-regulated and 17 genes up-regulated ( Figure 11A,B and Supplementary Table S2). Interestingly, 39/90 (43%) of the genes down-regulated in the H2A.J-KO cells corresponded to genes that were down-regulated in T47D cells by tamoxifen treatment, and 9/17 (53%) of genes that were up-regulated in the H2A.J-KO cells were up-regulated in T47D by tamoxifen treatment ( Figure 11C). Furthermore, 27 of the genes that are differentially expressed in the H2A.J-KO cells are also putative direct targets of ESR1 (ADAMTS8, APBB2, CXCL12, FAM189A2, FHL2, FOS, GAS6, GRIK3, HES1, HSPB8, LRP4, MAPT, NDRG1, NEDD9, NEK10, OLFML3, PDZK1, RBBP8, ROBO2, SCGB2A1, SEL1L3, SLC7A5, STC1, SULF2, SULT1A2, TFF1, TNS1) according to ChIP-seq data obtained by the Encode consortium and the ChEA transcription factor database [39,40]. These results suggest that H2A.J function is linked to estrogen-regulated gene expression. We confirmed this RNA-seq differential expression analysis by reverse transcriptase coupled to quantitative PCR experiments to determine the RNA levels for five estrogenresponsive genes (PDKZ1, SPINK5, SPINK13, SULF2, and TFF1)  expression of H2AFJ RNA is correlated with better survival compared to low levels of H2AFJ expression. Since luminal breast cancers tend to have a better prognosis compared to basal breast cancers [35], this survival advantage of high H2AFJ expression when considering all breast cancers may simply reflect its association with luminal epithelial cells. We also considered survival versus H2AFJ expression levels after separating breast cancer cell types into their PAM50 molecular subtypes. In this instance, high H2AFJ expression had no significant effect on the survival of basal-like or luminal-B breast cancers, but was correlated with slightly better survival of HER2+ and normal-like breast cancer, and correlated with worse survival of luminal-A type breast cancers. Overall, luminal-B type breast cancers have intrinsically higher expression of H2AFJ and a poorer prognosis compared to luminal-A type breast cancers [7]. We speculate that the poorer survival of luminal-A cancers with the highest expression of H2AFJ might reflect a state that is more similar to the luminal-B type cancers. It is more difficult to interpret the slightly better survival of HE2+ and normal-like cancers that express higher H2AFJ RNA. The function of H2A.J in luminal epithelial cells is not clear. We obtained homozygous H2A.J-KO mice that are viable and fertile. Their phenotypic characterization is ongoing, but the ability of mutant females to successfully raise pups indicates that they do not have a strong defect in mammary gland milk production. Likewise, the male fertility suggests that H2A.J is not required for prostate gland function. Our transcriptome analysis of T47D luminal breast cancer cells indicated that knocking out the H2AFJ gene reduced the expression of several estrogen-responsive genes and genes whose expression is up-regulated in luminal-B breast cancer, as well as genes involved in the EMT and cilium movement. Thus, H2A.J appears to facilitate the expression of some genes that are likely to contribute to luminal-type breast tumorigenesis. This result is consistent with work showing that H2AFJ is in a genomic region that tends to show copy number amplification in luminal B cancers, and the H2AFJ gene itself tends to be hypo-methylated and over-expressed in these cancers, leading to the suggestion that it might be a luminal-B oncogene [7]. Further work will be necessary to experimentally test the importance of H2A.J in luminal breast cancer. The availability of the H2A.J-KO mouse will facilitate an evaluation of its role in mouse models for luminal breast cancer [43].
The mechanism by which H2A.J influences gene transcription remains to be determined. The effect of H2A.J on transcription appears to depend on the cell type. Our previous work indicated that H2A.J stimulated inflammatory gene expression in senescent human fibroblasts and others reported a similar role for H2A.J in glioblastoma cells [11]. Interestingly, over-expression of H2AFJ is associated with lower survival in Low-Grade Gliomas (Supplementary Figure S7) which perhaps may be related to its role in stimulating inflammatory gene expression in these cells [11]. However, most inflammatory genes were not highly expressed in T47D cells and CXCL12 was the only inflammatory gene whose expression was reduced by knock-out of the H2AFJ gene in T47D cells. Luminal breast cells express the ESR1 estrogen receptor transcription factor, and a subset of estrogen-responsive genes, many of which are believed to be direct targets of ESR1, showed decreased transcription in the T47D H2A.J-KO cells. It is striking that H2A.J is highly expressed in luminal epithelial cells of a diverse set of glands. These luminal gland cells presumably all have a high secretory potential, as is the case for senescent fibroblasts, but their transcriptional programs are all distinct. Further work should elucidate the regulatory elements that lead to enhanced H2AFJ RNA expression in luminal epithelial cells, and how H2A.J in turn influences the transcriptional program of these diverse glands.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/genes12111665/s1, Figure S1: Specificity of anti-H2A.J demonstrated by parallel immunohistochemical detection of H2A.J in C57BL6-N WT and H2A.J-KO tissue sections for the indicated mouse organs. Figure S2: Human heart and cerebrum tissue sections show low levels of H2A.J by immunohistochemistry. Figure S3: RNA levels (TPM) for the indicated histone variant genes in 27 human tissues from GTEx data. Figure S4: Single-cell RNA-seq data for normal human ( Figure S4A-E) and mouse ( Figure S4F) prostate epithelial cells [25] show preferential expression of H2AFJ in luminal cells as for the KLK3 prostate-specific antigen gene. Figure S5: H2A.J is expressed at higher average levels in ER-positive breast cancer biopsies compared to ER-negative biopsies. Figure S6: Single-cell RNA-seq analysis of human prostate cancer epithelial cells shows that H2AFJ (A,C) and KLK3 (B,D) are expressed in a similar set of luminal-type epithelial prostate cancer cell clusters as shown in UMAP representations and violin plots. Figure S7: Kaplan-Meier survival curves for breast cancer patients expressing the indicated levels of H2AFJ RNAs. Figure S8: Kaplan-Meier survival curves for the indicated TCGA cancer types as a function of H2AFJ RNA expression. Figure S9: Absence of H2A.J expression in T47D-H2A.J-KO1 and T47D-H2A.J-KO18 cell lines compared to parental T47D cells as determined by Western blotting (A) and immunofluorescence (B) with an anti-H2A.J antibody. Figure S10: T47D, T47D-H2A.J-KO1, and T47D-H2A.J-KO18 cells show similar sensitivity to growth inhibition by 4-hydroxy-tamoxifen. Figure S11: Expression of estrogen-responsive genes is inhibited in H2A.J-KO cell lines relative to parental T47D cells. Figure S12: Heat maps for the gene sets that are significantly enriched relative to genes differentially expressed in H2A.J-KO cells versus parental T47D cells. Table S1: Number of patient samples for each tissue type obtained from the GTEx v.8 database. Table S2: RNA-seq analysis data for T47D and H2A.J-KO derivatives.