Biotinylation of histones represses transposable elements in human and mouse cells and cell lines, and in Drosophila melanogaster^,

Abstract

Transposable elements such as long terminal repeats (LTR) constitute about 45% of the human genome; transposition events impair genome stability. Fifty-four promoter-active retrotransposons have been identified in humans. Epigenetic mechanisms are important for transcriptional repression of retrotransposons, preventing transposition events and abnormal regulation of genes. Here, we demonstrate that the covalent binding of the vitamin biotin to lysine-12 in histone H4 (H4K12bio) and lysine-9 in histone H2A (H2AK9bio), mediated by holocarboxylase synthetase (HCS), is an epigenetic mechanism to repress retrotransposon transcription in human and mouse cell lines and in primary cells from a human supplementation study. Abundance of H4K12bio and H2AK9bio at intact retrotransposons and a solitary LTR depended on biotin supply and HCS activity, and was inversely linked with the abundance of LTR transcripts. Knockdown of HCS in Drosophila enhances retrotransposition in the germline. Importantly, we demonstrated that depletion of H4K12bio and H2AK9bio in biotin-deficient cells correlates with increased production of viral particles, transposition events, and ultimately decreases chromosomal stability. Collectively, this study reveals a novel diet-dependent epigenetic mechanism that could affect cancer risk.

INTRODUCTION

Type I transposable elements constitute about 42% of the human genome; they predominantly fall into two categories, the long terminal repeat(LTR)-containing retrotransposons and the non-LTR long interspersed nucleotide elements (LINE) (1-3). LTR-containing retrotransposons and LINE contain coding information for reverse transcriptase which catalyzes a critical step during transposition events (2). Mammalian genomes contain two types of LTR elements, intact retrotransposon LTR and solitary LTR (Fig. 1). In intact retrotransposons, the viral genes gag, pol, and env are flanked by two repeat regions: 5′-LTR and 3′-LTR. The expression of retroviral genes is regulated by promoters in the 5′-LTR (4). Transcription of intact retrotransposons by a reverse transcriptase and subsequent translocation impairs genomic stability and is associated with various disease states, such as cancer and autoimmunity (1, 5). In solitary LTR, the retroviral genes have been deleted by recombination between the LTR (1, 4, 6, 7). While solitary LTR cannot produce viral proteins, they may have promoter activity and can cause abnormal patterns of host gene expression (6, 7). Most retrotransposons are inactive, but 54 promoter-active retrotransposons were identified in human testes (4). Repression of intact retrotransposons and solitary LTR is important to prevent abnormal gene activity, and to decrease the incidence of retrotranspositions.

Fig. 1.

Structures of long terminal repeats (LTR). In intact LTR, the viral genes gag, pol, and env are flanked by two repeat regions, 5′-LTR and 3′-LTR. The expression of viral genes is regulated by regions in the 5′-LTR. In solitary LTR, the viral genes have been deleted by recombination events between LTR. Solitary LTR may affect the expression of host genes bi-directionally (arrows). LTR contain the following regions: (i) the U3 = a non-coding region of 200 - 1200 bp; U3 in the 5′-LTR contains the promoter elements responsible for transcription of both viral genes and host genes; (ii) R region = a short (18-250 bp) repeat sequence; (iii) U5 = non-coding region of 75-250 bp where reverse transcription of the transposon originates.

Various epigenetic mechanisms impact transcription. For example, LINE-specific siRNA, K9-dimethylation of histone H3 (H3K9me2), and methylation of cytosine residues in DNA are associated with transcriptional repression, whereas K4-trimethylation of histone H3 (H3K4me3) is associated with active genes (8-11). Methylation of cytosine and H3K9me2 are associated with repression of retrotransposons (2, 12, 13). Likewise, hyperacetylation of histones (14, 15) causes transcriptional activation of retrotransposons. Notwithstanding the importance of the above modifications, evidence suggests that other unknown chromatin modifications are also critical for silencing of retroviral elements, and preserving chromosomal stability and human health (15). One of these modifications is biotinylation of histones. The covalent binding of the vitamin biotin to histones is mediated by holocarboxylase synthetase (HCS) (16, 17). The following biotinylation sites have been identified: K9, K13, K125, K127, and K129 in histone H2A (18); K4, K9, and K18 in histone H3 (19); and K8 and K12 in histone H4 (20). Importantly, K12-biotinylation of histone H4 (H4K12bio) is a mark for repeat regions and heterochromatin, and plays a role in gene repression (21); H4K12bio co-localizes with the repression marker H3K9me2 (21). Here we tested the hypothesis that biotinylation of histone represses retrotransposon expression and transposition events.

MATERIALS AND METHODS

Cell culture

Human T lymphoma Jurkat cells and mammary carcinoma T47D cells, and murine mammary carcinoma Mm5MT cells were obtained from American Type Culture Collection. HCS-deficient Jurkat cells were generated by siRNA as described (22). Cells were cultured using the following biotin-defined media (23) for the indicated periods of time: 0.025 nmol/L of biotin, 0.25 nmol/L of biotin, and 10 nmol/L of biotin, representing concentrations observed in plasma from biotin-deficient individuals, normal individuals, and users of biotin supplements (23). In some experiments, cells were treated with 750 nmol/L trichostatin A for up to 17 hours to stimulate histone acetylation and transcriptional activity (15). Where indicated, DNA methyltransferase and G9A histone methyltransferase were inhibited by treatment with 0.25 μmol/L 5-aza-2′-deoxycytidine (AZA) for 4 days (24, 25). Reduction in DNA methylation was confirmed by bisulfite sequencing and methylation-specific PCR (24). Cell viability was monitored by Trypan blue exclusion at timed intervals (26).

Biotin-dependent carboxylases

The biotin content in holocarboxylases is a reliable marker for biotin supply (27); biotin in holocarboxylases from whole cell extracts was probed using gel-electrophoresis and streptavidin blotting (22, 23). Loading of lanes was normalized by total protein. Biotinylation of carboxylases was dose-dependent upon the levels of biotin in culture media, in all cell lines (Supplemental Fig. 1), confirming that treatment was effective.

ChIP assay

ChIP assays were conducted as described (21), using the following ChIP-grade antibodies: rabbit anti-human H4K12bio serum (20); rabbit anti-human K8-biotinylated histone H4 (H4K8bio) serum (20); rabbit anti-human K9-biotinylated histone H2A (H2AK9bio) serum (18); rabbit anti-human K13-biotinylated histone H2A (H2AK13bio) serum (18); rabbit antiserum to the C-terminus in human histone H3 (H3-C) (ab1791; Abcam); rabbit anti-human H3K9me2 serum (ab7312; Abcam); and rabbit anti-human H3K4me3 serum (ab8580; Abcam). The specificities of antibodies to biotinylated histones have been confirmed in an extensive series of testing (18, 20-22). In previous ChIP studies, results obtained with polyclonal antibodies were confirmed by using a set of monoclonal antibodies (21).

In previous studies, 54 promoter-active LTR were identified in human testes, including LTR22 and LTR15 (4). Here, LTR22 and LTR15 were used as models for solitary and intact retrotransposon LTR, respectively. The relative abundance of LTR in immunoprecipitated DNA and input DNA was quantified by quantitative real-time PCR (qRT-PCR) with sequence-specific primers (Supplemental Table 1), using 10 ng of DNA as template. ABsoluteTM QPCR SYBR Green fluorescein mix (ABgene) was used for qRT-PCR. Amplification of GAPDH sequences was used to control and normalize for qRT-PCR efficiency. The relative enrichment of genomic sequences was expressed as ratio of antibody-precipitated DNA to an equal amount of input DNA. Relative enrichment was normalized to nucleosomal density at LTR loci by chromatin precipitated with anti-H3-C. In previous studies we demonstrated that chromatin precipitation with bulk rabbit IgG yielded only minute amounts of DNA and can be neglected (21, 22).

Transcript abundance

Transcription of retrotransposons is the first step in retrotransposition events. Total RNA was isolated from cells using the Illustra RNAspin Mini kit (GE Healthcare) and reverse transcribed for subsequent quantification of transcripts by qRT-PCR. Mammalian LTR transcripts originate in the U5 region or at the R/U5 boundary (4, 7). PCR primers were designed to quantify the abundance of both transcripts (Supplemental Table 1). Data were normalized for qRT-PCR efficiency by using GAPDH mRNA as a control.

Human supplementation study

Ten healthy adults (n = 4 males, 6 females), 19-51 years old, participated in this study, which was approved by the Institutional Review Board for Human Studies at the University of Nebraska-Lincoln. Apparently healthy non-supplement users were eligible for participation. All subjects were supplemented with 600 μg of biotin per day for 4 wk to achieve a new steady state of biotin (28). Lymphocytes were isolated from peripheral blood by density gradient centrifugation before and after biotin supplementation (26). Holocarboxylases in lymphocyte extracts were probed with streptavidin peroxidase and quantified by gel densitometry (see above) to test for study compliance. One male subject (27 years old) was excluded from this study due to an abnormally high (82% above average) holocarboxylase abundance before biotin supplementation, suggesting the subject used biotin supplements before study begin. For the remaining subjects, the abundance of holocarboxylases in lymphocytes increased by 88 ± 64% (n = 9; P < 0.05) after 4 wk of biotin supplementation compared with pre-supplementation values.

Viral particle production

Mm5MT cells were cultured in media containing 0.25 nmol/L biotin for 4 days before transferring into designated biotin-defined media. On day 9, the cells were labeled with 150 mCi/L 35S Trans Label (MP Biochemicals) for 45 min. At the end of the labeling period, biotin-defined media were added and the labeled cultures were incubated for 16 hours. Culture supernatants were harvested and filtered through 0.45 μm cellulose acetate syringe filters. The filtrate was underlain with a 20% (wt/wt) sucrose cushion and the radiolabeled mouse mammary tumor virus (MMTV) particles were collected by centrifugation at 10,000 x g. Viral pellets were lyzed using 10 ml nonidet P-40 per L, 3.5 mmol/L sodium lauryl sulfate, and 12 mmol/L sodium deoxycholate in phosphate-buffered saline. The lysates were immunoprecipitated with rabbit anti-MMTV capsid p27 serum. MMTV particles were quantified by gel electrophoresis and phosphoimager scanning (Cyclone, Perkin Elmer).

ovoD1 reversion assay

The HCS RNAi knockdown transgene (pUAST-IRsp-HCS) was described previously (17). The ovo^D1 v²⁴ stock was a gift from Dr. Dr. Mariano Labrador (University of Tennessee-Knoxville). An insert of pUAST-IRsp-HCS on the X chromosome was used here. ovo^D1 v²⁴ males were crossed to females of the following genotypes: y w pUAST-IRsp-HCS; Actin5C-Gal4/ CyO y⁺; and y w. Cy⁺ female progeny from each cross were mated to two y w males to test fertility. Absence of eggs or larvae after two wk resulted in a score of sterile.

Chromosomal abnormalities

Jurkat cells in metaphase were stained with Giemsa as described (29) and analyzed for aberrations such as translocations, fragments, breaks, dicentrics, rings, triradial, and quadriradial formations.

Statistics

Homogeneity of variances among groups was tested using Bartlett's test; if variances were heterogeneous, data were log transformed before further statistical analysis (30). Significance of differences among groups was tested by one-way ANOVA. Fisher's Protected Least Significant Difference procedure was used for posthoc testing (30). Data from the human supplementation study (before vs. after biotin supplementation) were analyzed by using the unpaired Wilcoxon rank-sum test. The Chi-square and the Fisher Exact tests were used for analysis of chromosomal abnormalities in Jurkats cells and transpositions in flies, respectively. StatView 5.0.1 (SAS Institute) was used to perform calculations. Differences were considered significant if P < 0.05. Data are expressed as mean ± SD.

RESULTS

Histone biotinylation is enriched at LTR

To test whether biotinylated histones are enriched at LTR, chromatin immunoprecipitation (ChIP) assays were performed using antibodies to specific biotinylated histone isoforms. H4K12bio was enriched at LTR, and the level of enrichment depended on biotin concentrations in culture media. For example, the relative enrichment of H4K12bio at a solitary LTR (LTR22) was 95% greater in Jurkat cells cultured in medium containing 10 nmol/L biotin compared with 0.25 nmol/L biotin (Fig. 2A); the relative enrichment of H4K12bio at LTR22 decreased by 50% in cells cultured in medium containing 0.025 nmol/L biotin compared with 0.25 nmol/L biotin. H4K12bio enrichment was disrupted in HCS-deficient Jurkat cells. The concentration of biotin in culture media specifically affected the enrichment of H4K12bio in LTR, as opposed to globally affecting H4K12bio abundance in bulk histones. If bulk extracts of nuclear histones from biotin-defined Jurkat cells were probed with anti-H4K12bio, the relative abundance of H4K12bio was similar in cell cultured with 0.025, 0.25, and 10 nmol/L biotin (data not shown); this is consistent with previous observations in our laboratory (23, 31). The overall pattern of H4K12bio enrichment was paralleled by that of H3K9me2, a marker for repressed chromatin, although the magnitude of change in histone methylation was smaller than that seen for histone biotinylation (Figure 2A). Interestingly, the increased methylation was also absent in the HCS-deficient cells, suggesting that histone biotinylation could be essential for histone methylation at LTR. In contrast, the euchromatin marker H3K4me3 was not enriched at LTR22 and no link between H3K4me3 and biotin concentrations in culture media was evident (data not shown). The relative enrichment of sequences from an intact retrotransposon LTR (LTR15) in chromatin precipitated using antibodies to H4K12bio, H3K9me2 (Fig. 2B), and H3K4me3 (data not shown), were similar to those observed for the solitary LTR22. Here, too, the range of H4K12 biotinylation was greater than for H3K9 methylation and enrichment for both modifications required HCS. If Jurkat cells were incubated with 0.25 nmol/L of biotin, the enrichment of H4K12bio at LTR38 and LTR6, which are known to be transcribed at low levels (4), was similar to the enrichment observed for LTR22 and LTR15; enrichment was 2.3 ± 0.7 and 2.4 ± 0.9 times input DNA for LTR38 and LTR6, respectively (n = 3). Collectively, this is consistent with a role of histone biotinylation in repression of LTR.

Fig. 2.

Biotin supplementation leads to enrichment of H4K12bio and H3K9me2 at LTR in human T lymphoma Jurkat cells. A = solitary LTR (LTR22); B = intact retrotransposon LTR (LTR15). The relative enrichment of genomic sequences was expressed as ratio of antibody-precipitated DNA to an equal amount of input DNA. Relative enrichment was normalized to nucleosomal density at LTR loci by chromatin precipitated with anti-H3-C. Values are means ± SD, n=3. Means for each variable without a common letter differ, P < 0.05.

Biotin-dependent enrichment of H4K12bio at LTR22 and LTR15 was not a tissue-dependent phenomenon given that both human T lymphoma Jurkat cells and mammary carcinoma T47D cells (Supplemental Fig. 2) produced similar enrichment profiles. However, the enrichment of biotinylated histones at LTR was unique to certain biotinylation sites in histones. When chromatin from Jurkat cells was precipitated with antibodies to H2AK9bio, H2AK13bio, and H4K8bio, only H2AK9bio precipitation produced an enrichment pattern similar to that of H4K12bio in qRT-PCR (Supplemental Fig. 3). Chromatin immunoprecipitation with anti-H4K8bio and anti-H2AK13bio did not enrich for LTR22 and LTR15 (data not shown). H2AK9bio might play a role in gene repression in analogy to observations made for H4K12bio (21), based on the following findings. The relative enrichment of sequences from the interleukin-2 promoter (a repressed gene in Jurkat cells not stimulated with phorbol esters and mitogens) by anti-H2AK9bio precipitation exceeded the enrichment achieved with anti-H3K4me3: 2.4 ± 0.3 for H2AK9bio and 0.8 ± 0.03 for H3K4me3 (n = 3, P < 0.05). In contrast, the relative enrichment of sequences from the aldehyde dehydrogenase 5 promoter (a constitutively transcribed housekeeping gene) precipitated with anti-H2AK9bio was significantly less than the enrichment achieved with anti-H3K4me3: 0.9 ± 0.1 for H2AK9bio and 1.4 ± 0.2 for H3K4me3 (n = 3, P < 0.05). Thus, levels of two biotinylated histone isoforms implicated in transcriptional repression are correlated with biotin availability in an HCS-dependent fashion.

Histone biotinylation depends on cytosine methylation in LTR

Cytosine methylation in DNA is an epigenetic mark frequently associated with gene repression in higher eukaryotes. In some cases, DNA methylation is mechanistically linked to histone modification. We found that enrichment of H4K12bio at LTR was disrupted by treatment with the DNA methylation inhibitor 5-aza-2′-deoxycytidine (AZA). The relative enrichment of LTR22 sequences for Jurkat cells cultured in medium containing 0.25 nmol/L biotin by anti-H4K12bio decreased from 1.6 ± 0.1 in AZA-free cells to 0.8 ± 0.1 in AZA-treated cells (n = 3, P < 0.05); enrichment decreased from 1.2 ± 0.2 to 0.6 ± 0.05 if chromatin was precipitated with anti-H3K9me2 (n = 3; P < 0.05). Effects of AZA treatment were similar for LTR15 (data not shown). Efficacy of AZA treatment was confirmed by demonstrating decreased methylation of the progesterone receptor(Supplemental Fig. 4).

In contrast, cytosine methylation in LTR22 and LTR15 did not depend on biotin in Jurkat cells. For LTR22, the following levels of cytosine methylation were observed (units = % methylation in 12 putative methylation sites): 74 ± 25 in biotin-deficient cells; 84 ± 29 in biotin-normal cells; and 68 ± 29 in biotin-supplemented cells (n = 10 clones sequenced per treatment group; not statistically significant). For LTR15, the following levels of cytosine methylation were observed (units = % methylation in 11 putative methylation sites): 42 ± 24 in biotin-deficient cells; 63 ± 39 in biotin-normal cells; and 74 ± 32 in biotin-supplemented cells (n = 10 clones sequenced per treatment group; not statistically significant).

Biotin supplementation inhibits transcriptional activity of LTR

To test whether histone biotinylation at LTR is correlated with transcriptional activity, LTR-mediated transcript levels were determined in cells cultured at different biotin concentrations. Transcription of LTR may begin in region U5 or at the R/U5 boundary (7). U5 transcripts would be unlikely to mediate reverse transcription and therefore retrotransposition. Here, both U5 and R/U5 transcripts were quantified. The abundance of LTR mRNA originating in the U5 region increased by 85% in Jurkat cells cultured in medium containing 0.025 nmol/L biotin and decreased by 24% in cells supplemented with 10 nmol/L of biotin, respectively, compared with 0.25 nmol/L biotin (Fig. 3). When biotinylation of histones was disrupted by HCS knockdown, the abundance of LTR mRNA originating in the U5 region increased by 118% compared with biotin-matched wild-type controls (Fig. 3). Effects of biotin concentrations and HCS knockdown on the abundance of LTR mRNA originating at the R/U5 boundary were similar to those observed for the U5 transcript (Fig. 3). The abundance of U5 and R/U5 transcripts increased by 105 ± 80% and 411 ± 71%, respectively, when global transcriptional activity in Jurkat cells was enhanced by treatment with the histone deacetylase inhibitor trichostatin A (positive control; 0.25 nmol/L biotin). Finally, effects of biotin on the abundance of R/U5 transcripts in T47D cells were comparable to those seen in Jurkat cells; U5 transcripts showed a similar trend but the effects of biotin were not significant (Supplemental Fig. 5).

Fig. 3.

Biotin supplementation leads to a reduction in LTR transcription in Jurkat cells. Mammalian LTR transcripts originating in the U5 region and at the R/U5 boundary were quantified by qRT-PCR. Values are means ± SD, n=3. Means for each variable without a common letter differ, P < 0.05.

Biotin supplementation enhances histone biotinylation at retroelement LTR in healthy adults

Biotin is an essential vitamin in humans. Biotin deficiency and mutations in the biotin-conjugating enzyme HCS result in a suite of seemingly unrelated physiological disturbances. Previous studies have implicated HCS-dependent histone biotinylation in gene regulation. To test whether these effects extend to transcriptional control of retrotransposons, a significant source of genomic instability, we examined the level of histone biotinylation at retrotransposon LTR in lymphocytes of individuals before and after biotin supplementation. The relative enrichment of LTR22 sequences by ChIP with anti-H4K12bio increased from 0.7 ± 0.3 (= background noise, equaling input DNA) before biotin supplementation to 1.8 ± 2.0 after biotin supplementation (n = 9 subjects; P < 0.05); a similar pattern was observed for LTR22 if chromatin was precipitated using anti-H3K9me2: 0.8 ± 0.3 before supplementation vs. 1.8 ± 1.6 after supplementation (P = 0.05). Likewise, the relative enrichment of LTR15 by precipitation with anti-H4K12bio increased from 0.8 ± 0.1 before biotin supplementation to 2.3 ± 3.6 (n = 9; P < 0.05) after supplementation; the relative enrichment of LTR15 increased from 0.9 ± 0.4 before biotin supplementation to 1.9 ± 1.4 after supplementation (P = 0.10) if chromatin was precipitated using anti-H3K9me2. The increased enrichment of H4K12bio at LTR15 and LTR22 coincided with a decrease in LTR transcripts (arbitrary units, by qRT-PCR): 1.0 ± 0.01 before biotin supplementation vs. 0.7 ± 0.2 after biotin supplementation for the U5 transcript (n = 9; P < 0.05); and 1.0 ± 0.02 before biotin supplementation vs. 0.6 ± 0.2 after biotin supplementation for the R/U5 transcript (P < 0.05). These measurements point to a modest, but statistically significant, enrichment in biotinylated histones at retroelement LTR with biotin supplementation and a concomitant modest reduction in LTR transcription, suggesting that histone biotinylation could play a role in down-regulation of retroelement transposition.

Biotin suppresses endogenous retroviral particle formation

Murine Mm5MT mammary carcinoma cells are known to produce particles of MMTV (32). We demonstrated that production of viral particles by Mm5MT cells is a biotin-dependent process. First, we demonstrated that biotinylated histones are enriched at exogenous MMTV LTR in Mm5MT cells (units = relative enrichment compared with an equal amount of input DNA): H4K12bio = 1.4 ± 0.1; H2AK9bio = 1.4 ± 0.1; and H3K9me2 (control) = 1.2 ± 0.1. Production of viral particles, as assessed by immunoprecipitation of radiolabeled culture supernatants, decreased by >90% if cells were cultured at a biotin concentration of 10 nmol/L compared with 0.025 nmol/L biotin(Fig. 4).

Fig. 4.

Production of viral (MMTV) particles in murine mammary carcinoma Mm5MT cells depends on biotin status in cell culture media. Values are means ± SD, n=3. Means for each variable without a common letter differ, P < 0.05. Inserts depict representative examples from one gel.

HCS knockdown increases the frequency of retrovirus transposition in Drosophila

To test whether the role of biotinylation in repressing retrotransposons is evolutionarily conserved, we measured the effect of HCS knockdown on the rate of retrotransposition in D. melanogaster. In Drosophila, ovoD1 females are sterile due to a point mutation that resulted in dominant arrest of oogenesis (33). ovoD1 females revert spontaneously to fertility due to the insertion of the endogenous retrovirus gypsy or the retrotransposon copia (34). Thus, the frequency of fertile ovoD1 females reflected the rate of retroelement transposition in the germ-line. The incidence of transposition events was greater in HCS knockdown, ovoD1 mutant Drosophila than in ovoD1 mutant females with normal HCS levels. In HCS knockdown, ovoD1 mutants, fertility was restored in 3.3% of females compared with 0.8% in HCS wild-type controls (n = 600; P < 0.05). Thus, the transposition rate of retroelements in HCS-deficient flies was four times that in HCS wild-type flies. These observations demonstrate that HCS normally constrains retroelement transposition and is part of a mechanism to maintain genome stability in the germ line.

Biotin depletion is associated with increased frequency of chromosomal abnormalities

To test whether biotin levels contribute to genome stability in somatic cells, we examined the frequency of microscopically visible chromosome aberrations in cells cultured at different biotin concentrations. Biotin depletion increased the incidence of chromosomal abnormalities in Jurkat cells. When cells were cultured in biotin-deficient medium, the frequency of chromosomal abnormalities per metaphase cell was 0.71 ± 0.16; no abnormalities were detected in cells cultured in medium containing 0.25 nmol/L and 10 nmol/L biotin (n = 125 - 150 metaphase cells; P < 0.05).

DISCUSSION

Here we provide evidence for the existence of a novel diet-dependent epigenetic mechanism that represses retrotransposons. Importantly, we demonstrate depletion of biotinylated histones in biotin-deficient cells increases LTR transcript levels, production of viral particles, retrotransposition events, and ultimately decreases chromosomal stability. Both biotin deficiency and supplementation are prevalent in the U.S. For example, moderate biotin deficiency has been observed in up to 50% of pregnant women (35, 36). About 20% of the U.S. population reports taking biotin supplements (37), producing supra-physiological concentrations of vitamin in tissues and body fluids (23, 28, 35). The findings presented here suggest that altered biotin status in these population subgroups might affect chromosomal stability and cancer risk.

In this study, models were selected so to permit an evaluation of effects of biotin status on human health. First, mammalian cell lines were used as primary models and were cultured in biotin-defined media representing concentrations observed in plasma from biotin-deficient individuals, normal individuals, and users of biotin supplements (23). Second, biotinylated histones were found to be enriched at both a solitary LTR and in the LTR of an intact retrotransposon, suggesting that we might have identified a regulatory mechanism common to most if not all LTR. Third, data were validated in primary cells in a human biotin supplementation study. Fourth, effects of abnormal histone biotinylation on the incidence of transposition events were demonstrated in a whole organism, Drosophila melanogaster. Fifth, the frequency of chromosomal abnormalities in human lymphoid cells increased in response to biotin depletion.

Previous studies suggest that H4K12bio is enriched in repeat regions in pericentromeric chromatin and also mediates transcriptional repression (21). Here, we demonstrated that H4K12bio is also enriched at LTR, suggesting that H4K12bio is a ubiquitous mark for repressed chromatin in eukaryotes. The biological functions of biotinylation marks other than H4K12bio are unclear. In this study, we provided evidence that H2AK9bio is enriched at LTR to an extent similar to that observed for H4K12bio; moreover, we provide evidence that H2AK9bio is also enriched in other repressed regions of chromatin. In contrast, H2AK13bio and H4K8bio are not enriched at LTR, suggesting that enrichment at LTR is specific for certain species of biotinylated histones.

The N-terminal tails of histones are exposed at the nucleosomal surface and amino acid residues in these tails are targets for numerous posttranslational modifications (9, 38, 39). There is substantial cross-talk among these modifications. For example, H3K9me2 prevents K4-methylation of histone H3 by inhibiting the activity of a methyltransferase that specifically targets lysine-4 in histone H3 (40). Here, we demonstrated that biotinylated histones participate in this cross-talk. The enrichment of H3K9me2 at LTR coexists with H4K12bio. When histone biotinylation was disrupted in HCS-deficient cells, enrichment of H3K9me2 was abolished. This is consistent with ongoing studies in our laboratory suggesting that histone H3 K9-methyltransferases physically interact with HCS (41). Moreover, we provided evidence for cross-talk between histone biotinylation and another mark related to repression, methylation of cytosines. Enrichment of H4K12bio at LTR was abolished if DNA methylation was decreased by AZA treatment. We cannot formally exclude the possibility that inhibition of the G9A histone methyltransferase by AZA (25) contributed to the decreased enrichment of H3K9me2 in LTR in AZA-treated cells. In contrast, methylation of cytosine residues was not affected by biotin status in the cell culture medium, suggesting that histone biotinylation at LTR depends on DNA methylation but not vice versa. Finally, we demonstrated that H3K4me3 is not enriched at de-repressed LTR in biotin-deficient cells. This is consistent with previous studies suggesting that H3K4me3 might not be an unambiguous mark for de-repressed sequences(42).

Further studies on biotin's effects on retrotransposition will require transgenic mouse models. Also, we are currently in the process of quantifying the turnover rates of biotinylated histones during the course of biotin depletion in mammalian systems.

Abbreviations

AZA

5-aza-2′-deoxycytidine

HCS

holocarboxylase synthetase

HERV

human endogenous retrovirus

H3K4me3

K4-trimethylated histone H3

H4K8bio

K8-biotinylated histone H4

H2AK9bio

K9-biotinylated histone H2A

H3K9me2

K9-dimethylated histone H3

H4K12bio

K12-biotinylated histone H4

H2AK13bio

K13-biotinylated histone H2A

LTR

long terminal repeat

MMTV

mouse mammary tumor virus

PBS

primer binding site

qRT-PCR

quantitative real-time PCR