Activation of gene expression by histone deubiquitinase OTLD1

ABSTRACT

One of the main mechanisms of epigenetic control is post translational modification of histones, and one of the relatively less characterized, yet functionally important histone modifications is monoubiquitylation, which is reversed by histone deubiquitinases. In Arabidopsis, only two of such enzymes are known to date. One of them, OTLD1, deubiquitylates histone 2B and functions as a transcriptional repressor. But, could the same deubiquitinase act both as a repressor and an activator? Here, we addressed this question. Using gain-of-function and loss-of-function Arabidopsis alleles, we showed that OTLD1 can promote expression of a target gene. This transcriptional activation activity of OTLD1 involves occupation of the target chromatin by this enzyme, deubiquitination of monoubiquitylated H2B within the occupied regions, and formation of the euchromatic histone acetylation and methylation marks. Thus, OTLD1 can play a dual role in transcriptional repression and activation of its target genes. In these reactions, H2B ubiquitylation acts as both a repressive and an active mark whereas OTLD1 association with and deubiquitylation of the target chromatin may represent the key juncture between two opposing effects of this enzyme on gene expression.

Introduction

Epigenetic regulation of gene expression involves imposition of two altering functional states, i.e., active, euchromatic and repressive, heterochromatic, on the target chromatin. One of the main mechanisms of epigenetic control is post translational modification of histones, which includes over a hundred of specific modification sites, the studies of many of which are still in their infancy.^1-3 Among such relatively less characterized, yet functionally important histone modifications, is monoubiquitylation.^4-6 In plant chromatin, monoubiquitylation is found both on histones 2A (H2A) and 2B (H2B),⁷ with H2B monoubiquitylation being better characterized and shown to affect diverse aspects of the plant life cycle, such as cell cycle, photomorphogenesis, root growth, leaf shape, flowering time, and seed dormancy.^7-12 H2B is thought to be ubiquitylated by plant homologs of the Rad6-Bre1 ubiquitin ligase¹³ whereas monoubiquitylated histone H2B (H2Bub) presumably is deubiquitylated by histone deubiquitinases, only two of which are known in plants to date: UBP26^7,14-16 and OTLD1.^17,18

Both UBP26 and OTLD1 have been shown to function in gene repression.^15,17,18 But, could the same plant deubiquitinase act both as a transcriptional repressor and as a transcriptional activator? The first suggestion of this possibility came from the observations that, while UBP26 repressed one of its target genes, PHE1,¹⁵ it activated another target gene, FLC¹⁶; yet, because association of UBP with the target chromatin has not been examined, one of these opposing effects of UBP26 on target gene expression was proposed to be indirect.¹⁶ Here, we addressed the potential dual function of plant histone deubiquitinases using OTLD1.^17,18 OTLD1 belongs to the ovarian tumor (OTU) deubiquitinase family^7,14; these enzymes differ from the ubiquitin-binding protein (UBP) deubiquitinases in that they are linkage-specific¹⁹ whereas UBPs cleave most ubiquitin chain types indiscriminately.²⁰ OTLD1 is known as a transcriptional repressor that associates with its target chromatin, deubiquitylates it, and depletes euchromatic histone modifications.¹⁷ Using gain-of-function and loss-of-function Arabidopsis alleles, we demonstrate that OTLD1 can also promote expression of a target gene. This transcriptional activation activity of OTLD1 involves occupation of the target chromatin by the enzyme, H2Bub deubiquitination in the occupied regions, and formation of the euchromatic histone acetylation and methylation marks.

Results

Transcriptional activation of the AN3 gene in OTLD1 gain-of-function plants

Our studies of OTLD1, which belongs to a 12-member family of OTU deubiquitinases^14,21 (Fig. S1) with potentially partially redundant activities,²² utilized two independent gain-of-function transgenic lines OTLD1 OE-1 and OTLD1 OE-2.¹⁷ Our previous RT-qPCR analysis of these plant lines showed that they were moderate expressors, accumulating ca. 2.5-fold higher amounts of OTLD1 transcripts than the wild-type plants.¹⁷ Based on their overall phenotypes of enhanced growth, we analyzed expression of 42 genes that included representatives of major gene families involved in control of plant growth. Quantitative RT-PCR (RT-qPCR) analysis demonstrated that, although most of the tested genes showed no meaningful changes in their expression, five displayed significant reduction in expression, indicating the repressive function of OTLD1.¹⁷ However, one tested gene, ANGUSTIFOLIA3 (AN3/GIF1), exhibited statistically significant (p<0.05) increase in expression in both transgenic lines. Specifically, overexpression of OTLD1 elevated, with statistical significance (p<0.05), expression of AN3 by 5- to 3-fold of the wild-type levels in the OTLD1 OE-1 and OTLD1 OE-2 plants, respectively (Fig. 1A). Only slight and statistically insignificant changes were observed in the expression level of a control ABI2 gene (Fig. 1B), which does not respond to OTLD1.¹⁷ In the internal control experiments, expression levels of two reference genes ACT7 and UBQ10 in both gain-of-function plant lines were similar to the wild-type plants (Fig. 1C).

Figure 1.

Transcriptional activation of the AN3 gene in the OTLD1 OE-1 and OTLD1 OE-2 plants. (A) Expression of AN3. (B) Expression of OTLD1-independent control gene ABI2. (C) Expression of the reference genes ACT7 and UBQ10. Relative expression levels of OTLD1 in wild-type (black bars), OTLD1 OE-1 (dark gray bars), and OTLD1 OE-2 plants (light gray bars) was analyzed by RT-qPCR. The expression level in the wild-type plants is set to 1.0. Error bars represent s. e. m. of three independent biological replicates, N = 3; p<0.05 for statistical significance of differences between the OTLD1 OE-1 or OTLD1 OE-2 plants and wild-type plants in (A) whereas differences between all tested plants in (B) and (C) were not statistically significant.

OTLD1 associates with the AN3 chromatin, deubiquitylates it and enhances its euchromatic histone marks

Understanding whether or not OTLD1 associates with the AN3 chromatin is important to provide insight into whether the transcriptional activation of AN3 occurs directly or indirectly, for example, by suppressing expression of a putative AN3 repressor. Thus, we used quantitative chromatin immunoprecipitation (qChIP) to detect the presence of OTLD1, tagged with an His₆ epitope, in several overlapping regions of the putative promoter sequence of AN3 in the OTLD1 OE-1 and OTLD1 OE-2 plants.²³ Fig. 2A shows that OTLD1 associated in a statistically significant fashion (p<0.05) with two distinct regions, located within 1.2 kb upstream of the translation initiation codon, in the AN3 chromatin of both plant lines; this is in line with the notion that most intergenic distances and, by implication, promoter sequences in the Arabidopsis genome are around 2 kb.^23,24 Consistent with similar levels of OTLD1 overexpression in the OTLD1 OE-1 and OTLD1 OE-2 lines,¹⁷ the extent of OTLD1 association with the AN3 chromatin was comparable in both lines. In negative control experiments, no signal was observed in non-transgenic, wild-type plants (Fig. 2A); also, previous experiments did not detect association of the same His₆-tagged OTLD1 with OTLD1-independent genes ABI2 and GRF5.¹⁷ These observations correlated between the presence of OTLD1 in the AN3 chromatin and the role of OTLD1 in transcriptional activation of AN3.

Figure 2.

OTLD1 associates with the AN3 chromatin and alters its histone modification marks in the OTLD1 OE-1 and OTLD1 OE-2 plants. (A) Association of His-tagged OTLD1 with the AN3 chromatin. Locations of sequences upstream of the translation initiation site used for qChIP analyses are indicated. A. U., arbitrary units. (B) Hypoubiquitination of the AN3 chromatin. (C) H3 hyperacetylation in the AN3 chromatin. (D) Increased H3K4 trimethylation in the AN3 chromatin. Relative levels of H2Bub, H3ac, and H3K4me3 were determined by qChIP. Wild-type plants, black bars; OTLD1 OE-1, dark gray bars; OTLD1 OE-2, light gray bars. For panel A, the absolute levels of signal are shown; for panels B and C, the levels of signal are shown relative to the signal level in the wild-type plants, which is set to 1.0. Error bars represent s. e. m. of three independent biological replicates, N = 3; p<0.05 for statistical significance of differences between the OTLD1 OE-1 or OTLD1 OE-2 plants and wild-type plants.

Next, we examined whether the AN3 chromatin occupied by OTLD1 becomes deubiquitylated. Our qChIP analysis (Fig. 2B) showed statistically significant (p<0.05) levels of hypoubiquitylation of H2Bub in the same regions of the AN3 chromatin that were physically associated with OTLD1 (Fig. 2A). In both OTLD1 OE-1 and OTLD1 OE-2 lines, the AN3 chromatin exhibited levels of monoubiquitylation that were 2-8-fold lower than those detected in the same chromatin regions in the wild-type plants (Fig. 2B).

Epigenetic activation of gene expression is reflected in increased occurrence of euchromatic marks, such as H3 acetylation (H3ac) and trimethylation of H3K (H3K4me3). Does the AN3 chromatin display the same dynamics in the OTLD1 OE-1 and OTLD1 OE-2 plants? Fig. 2 shows that it does. Specifically, the H3ac levels in the AN3 chromatin in OTLD1 OE-1 and OTLD1 OE-2 were elevated by 2-3.5 fold in comparison to the wild type plants (Fig. 2C). In one specific region of the AN3 chromatin, this hyperacetylation of the AN3 chromatin was paralleled by the 1.7-2.3-fold increase in the levels of H3K4me3 (Fig. 2D) in the OTLD1 OE-1 and OTLD1 OE-2 plants relative to the wild-type plants (Fig. 2D). Together, the changes in the amounts of the euchromatic marks that occurred with statistical significance (p<0.05) in the AN3 chromatin of the OTLD1 gain-of-function lines support the role of OTLD1 as transcriptional activator of AN3.

Transcriptional repression of the AN3 gene and epigenetic modification of the AN3 chromatin in OTLD1 loss-of-function plants

What are the effects of the loss of OTLD1 function on the AN3 gene expression and on the levels of H2Bub, H3ac, and H3K4me3 in the AN3 chromatin? First, we analyzed two previously characterized^17,18 Arabidopsis OTLD1 loss-of-function alleles otld1-1 and otld1-2 for the expression of AN3. Our RT-qPCR analysis of the accumulation of the AN3 transcripts revealed a statistically significant (p<0.05) 2-fold decrease in both mutants (Fig. 3A). Neither of the mutant lines showed alterations in expression of the negative control ABI2 gene (Fig. 3B) or of the internal reference genes ACT7 and UBQ10 (Fig. 3C). This incomplete suppression of the AN3 gene expression in the absence of OTLD1 most likely is due to the presumed partial functional redundancy between OTLD1 and other members of the Arabidopsis OTU family.¹⁷ Furthermore, that the expression of AN3 was reduced, but not blocked, in both otld1-1 and otld1-2 alleles, may explain the lack of clear phenotypic changes in these plants as the AN3-specific phenotypes have been observed with the AN3 loss-of-function mutants that lacked any detectable AN3 expression.²⁵

Figure 3.

Transcriptional suppression of the AN3 gene in the otld1-1 and otld1-2 mutant plants. (A) Expression of AN3. (B) Expression of OTLD1-independent control gene ABI2. (C) Expression of the reference genes ACT7 and UBQ10. Relative expression levels of OTLD1 in wild-type (black bars), otld1-1 (dark gray bars), and otld1-2 plants (light gray bars) was analyzed by RT-qPCR. The expression level in the wild-type plants is set to 1.0. Error bars represent s. e. m. independent of three biological replicates, N = 3; p<0.05 for statistical significance of differences between the otld1-1 or otld1-2 plants and wild-type plants in (A) whereas differences between all tested plants in (B) and (C) were not statistically significant.

Next, we analyzed the AN3 chromatin of the otld1-1 and otld1-2 mutants for the presence of ubiquitylation H2Bub marks and euchromatic H3ac and H3K4me3 marks. Generally, these analyses (Fig. 4) revealed, with statistical significance (p<0.05), a picture reciprocal to that observed in the gain-of-function plants (Fig. 2). Specifically, in both loss-of-function lines, the AN3 chromatin occupied and deubuquitylated by OTLD1 in the gain-of-function lines (Fig. 2A, B), was hyperubuquitylated by 1.8-2.6 fold as compared to the wild-type lines (Fig. 4A). Similarly, the H3ac and H3K4me3 marks of the active chromatin were reduced by 5-1.7 fold (Fig. 4B) and 1.7-2 fold (Fig. 4C), respectively, relative to the wild-type plants. Collectively, our data suggest that OTLD1, in addition to its known role as a transcriptional repressor,^17,18 can function as a transcriptional activator.

Figure 4.

OTLD1 alters the AN3 chromatin histone modification marks in the otld1-1 and otld1-2 mutant plants. (A) Hyperubiquitination of the AN3 chromatin. Locations of sequences upstream of the translation initiation site used for qChIP analyses are indicated. (B) H3 hypoacetylation in the AN3 chromatin. (C) Reduced H3K4 trimethylation in the AN3 chromatin. Relative levels of H2Bub, H3ac, and H3K4me3 were determined by qChIP. Wild-type plants, black bars; otld1-1, dark gray bars; otld1-2, light gray bars. The corresponding signal level in the wild-type plants is set to 1.0. Error bars represent s. e. m. of three independent biological replicates, N = 3; p<0.05, for statistical significance of differences between the otld1-1 or otld1-2 plants and wild-type plants.

Discussion

In eukaryotic cells, H2B ubiquitylation associates with both active transcription and transcriptional gene silencing whereas its reversal by deubiquitylation correspondingly suppresses and activates transcription.^4,5,26,27 This deubiquitylation is mediated by histone deubiquitinase enzymes which can, therefore, act as both repressors and activators of transcription of their target genes. So far, these opposing effects on the target chromatin have been largely relegated to distinct histone deubiquitinases. For example, in mammals, UBP44, a member of the UBP/USP family of deubiquitinases, deubiquitylates H2Bub at its target gene promoter and represses transcription²⁸ whereas H2Bub deubiquitylation at the target gene promoter by a different member of this enzyme family, UBP42, activates transcription.²⁹ Also in mammals, on the genome-wide level, H2Bub negatively correlates gene induction.³⁰ In yeast, H2Bub deubiquitylation by Ubp10 is involved in transcriptional repression,²⁷ and H2Bub deubiquitylation by Ubp8 is required for transcriptional activation,³¹ and these histone deubiquitinases target H2Bub in different chromatin domains.³² Also, in yeast, H2B ubiquitylation in quiescent promoters inhibits transcription and in active gene bodies promotes transcription elongation.³³

The notion that the same histone deubiquitinase can associate with the chromatin of its target genes and directly control their transcription via both activating and repressive regulatory pathways has not been examined. We addressed this question using OTLD1, one of the two known plant histone deubiquitinases. Our data suggest that OTLD1 can play a dual role in transcriptional repression and activation. This idea is illustrated in Fig. 5 using OSR2 as an example of an OTLD1-repressed gene¹⁷ and AN3 as an example of an OTLD1-activated gene. OTLD1 can associate with the chromatin of the promoter regions of both genes and deubiquitylate H2Bub in both promoters. After these two “common” steps, the effects of OTLD1 on both genes diverge, and the OSR2 chromatin loses its euchromatic histone modification marks whereas the AN3 chromatin acquires them, ultimately resulting in transcription repression or activation, respectively. In this scenario, H2B ubiquitylation would act as both a repressive and an active mark. Thus, OTLD1 association with and deubiquitylation of the target chromatin may represent the key juncture between two opposing effects of this enzyme on gene expression. So far, AN3 is the sole known gene that can be transcriptionally activated by OTLD1; yet, irrespective whether AN3 represents a specific or a more general case of such regulation, the present and previous^17,18 data indicate that OTLD1 is capable of both gene repression and activation.

Figure 5.

Schematic illustration of the dual role of OTLD1 in transcriptional repression and activation. OTLD1-repressed gene is exemplified by OSR2,¹⁷ and OTLD1-activated gene is exemplified by AN3. Association of OTLD1 with the OSR2 promoter chromatin deubuquitylates H2Bub and reduces the levels of euchromatic histone marks H3ac and H3K4me3, culminating in transcriptional repression of OSR2.¹⁷ Association of OTLD1 with the AN3 promoter chromatin also deubuquitylates H2Bub, but it elevates the levels of euchromatic histone marks H3ac and H3K4me3, promoting transcription of AN3. For further details, see text.

To date, there is no clear evidence that such dual-function H2Bub deubiquitinases exist in other eukaryotes. The mammalian USP22 has been shown to activate transcription of SAGA-dependent genes^34-36 and repress transcription of the Sox2 locus.³⁷ Yet, these activities of USP22 have been attributed to its deubiquitylation of different histone molecules: by deubiquitylating H2Aub USP22 might act as a transcriptional activator and by deubiquitylating H2Bub USP22 might act as a transcriptional repressor.³⁷ In plants, on the other hand, H2Bub deubiquitylation by UBP26 has been reported to correlate with accumulation of euchromatic histone marks within the promoters of the targets genes, including the flower timing regulator FLC gene.¹⁶ UBP26, however, also has been shown to promote accumulation of the repressive marks at different transgenes³⁸ as well as at the PHE1 locus involved in seed and embryo development.¹⁵ Because none of these studies had assessed association of UBP26 with the target promoters, one of the opposing influences of UBP26-mediated H2Bub deubiquitylation on transcription was attributed to indirect effects,¹⁶ for example, via repressing another repressor, resulting in apparent transcriptional activation. Based on our findings of the dual repressor/activator activity of OTLD1, it is possible that UBP26 also fulfills such dual function toward its different target genes. In this case, it would be interesting that the only two known plant histone deubiquitinases-which belong to different deubiquitinase families but display the same H2Bub substrate specificity-function both as transcriptional repressors and activators. Such dual functionality of H2Bub-specific deubiquitinases has not been described in non-plant systems, suggesting that it might be more prevalent or, perhaps, unique in plants.

Materials and methods

Plants

Wild-type Arabidopsis thaliana (ecotype Col-0) plants and SALK_028707 and SALK_037047 lines, representing the otld1-1 and otld1-2 T-DNA insertion mutants of OTLD1 were obtained and grown at 22°C under long-day conditions (16-h light/8-h dark cycle at 140 μEsec⁻¹m⁻² light intensity) as described previously.¹⁷ Production of the transgenic Arabidopsis lines OTLD1 OE-1 and OTLD1 OE-2 that over-express OTLD1 also has been described.¹⁷

Quantitative RT-PCR (RT-qPCR)

RT-qPCR was performed as described.¹⁷ Briefly, cDNA (2 µl) reverse transcribed from a total RNA (1 µg) extracted from whole aerial parts of 25-days-old plants was amplified using Maxima SYBR Green qPCR Master Mix (Thermo Fisher Scientific Inc.) and specific primers described in Table S1 in a StepOnePlus real-time PCR system (Applied Biosystems) for 1 cycle at 95°C for 5 min and 40 cycles each at 95°C for 10 s, 57°C for 10 s, and 72°C for 15 s. Each sample was analyzed in three biological replicates. Two validated constitutive reference genes UBQ10 (At4g05320) and ACT7 (At5g09810)¹⁷ were used for normalization of RT-qPCR data by the comparative C_t method, using ΔC_t obtained by subtracting the C_t value of the tested transcript from the C_t value of UBQ10 and ACT7 transcripts in each sample, and the relative gene expression levels were calculated by the cycle threshold (CT) 2^−ΔΔCt method.³⁹ All quantitative data were analyzed by the Student's t-test with p<0.05, corresponding to the statistical probability of >95%, considered statistically significant. Standard error of the mean (s. e. m.) and t-test calculations were performed using Excel 2010 (Microsoft Inc.).

Quantitative chromatin immunoprecipitation (qChIP)

ChIP was performed as described.¹⁷ Briefly, cell nuclei were isolated from whole areal parts (10 g) of 25-days-old plants cross–linked by 1% formaldehyde (v/v). Chromatin was sheared by sonication an average size of 400–1,000 bp fragments, pre-incubated at 4°C for 1 h with pre-equilibrated protein A agarose beads (16–157, Millipore), centrifuged, and the supernatant was incubated at 4°C for overnight with the appropriate antibody [anti-acetyl-histone H3 (06–599, Millipore, 1:100 dilution); anti-monoubiquityl-histone H2B (Lys-120) (5546S, Cell Signaling Technology, Inc., 1:100 dilution); anti-trimethyl H3K4 (8580, Abcam, 1:100 dilution); or anti-penta-His (34660, Qiagen, 1:100 dilution)], combined with protein A agarose (75 µl) and further incubated at 4°C for 2 h. The immunocomplexes immobilized on the protein A agarose beads were washed sequentially with low and high salt buffers [20 mM Tris-HCl pH 8.0, 2 mM EDTA, 0.1% SDS, 1.0% Triton X-100 supplemented with 0.15 M NaCl (low salt) or 0.5 M NaCl (high salt)], LiCl buffer (250 mM LiCl, 10 mM Tris-HCl pH 8.0, 1.0 mM EDTA, 1% NP-40, 1.0% deoxycholate), and twice with TE (10 mM Tris-HCl pH 8.0, 1.0 mM EDTA), and eluted at room temperature for 15 min in the elution buffer (0.1 M NaHCO₃, 0.5% SDS). Following incubation to reverse cross-linking (0.2 M NaCl at 65°C for overnight) and digestion with Proteinase K (20 mg/ml, 45°C for 90 min), the recovered DNA (10 ng) was analyzed by qPCR using the appropriate primers (Table S1). Non-specific, background signal was evaluated using protein A agarose incubated with chromatin samples in the absence of antibody; no signal was obtained in any of these experiments.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

The work in the V.C. laboratory was supported by grants from NIH, NSF, USDA/NIFA, and BARD to V.C.

Author contributions

I.K. designed, performed, analyzed, and discussed all the experiments and data in this study and wrote the manuscript. V.C. provided conceptual guidance, secured required funding, and reviewed and edited the manuscript.