Monoubiquitinated H2B, a main chromatin target of formaldehyde, is important for S-phase checkpoint signaling and genome stability

Abstract

Formaldehyde (FA) is a human carcinogen with ubiquitous environmental exposures and significant endogenous formation. Genotoxic activity of FA stems from its reactivity with DNA-NH₂ groups. Histone lysines are another source of aldehyde-reactive amino groups in chromatin, however, chromatin/histone damage responses to FA and their biological significance are poorly understood. We examined histone posttranslational modifications in FA-treated human lung cells and found that the majority of the most prominent small lysine modifications associated with active or inactive chromatin were unchanged. FA moderately decreased H3K9 and H3K27 acetylation and H2A-K119 monoubiquitination but caused surprisingly severe losses of H2B-K120 monoubiquitination, especially in primary and stem-like cells. H2Aub1 decreases reflected its slower ubiquitination linked to a lower ubiquitin availability due to K48-polyubiquitination of FA-damaged proteins. Depletion of H2Bub1 resulted from its rapid deubiquitination in part by ATXN7L3-associated deubiquitinases and was independent on DNA damage signaling, indicating a direct chromatin damage response. Manipulations of H2Bub1 abundance showed that it was important for robust ATM and ATR signaling, efficient S-phase checkpoint and suppression of mitotic transmission of unreplicated DNA and formation of micronuclei. Our findings identified H2B deubiquitination as a major FA-induced chromatin damage response which regulates S-phase checkpoint signaling and genome stability.

Introduction

Formaldehyde (FA) is a firmly established animal and human carcinogen with widespread population exposures.^1,2 Ubiquitous in-door air presence of FA results from its offgassing from numerous household products, paints and building materials. FA is also frequently found in the ambient air in many occupational settings³ and it is abundantly produced and inhaled during smoking of tobacco products⁴ and e-cigarettes.⁵ Mouse genetic models with deficiencies in FA detoxification enzymes have also provided strong evidence for the importance of FA as one of the endogenous sources of genotoxic damage.^6,8 A major genotoxic activity of FA involves the formation of DNA-protein crosslinks (DPC) that act as potent inhibitors of DNA replication due to stalling of the replicative helicase.⁹ The very large size of DPC prevents their direct removal by nucleotide excision repair^10–12 which is the main cellular process for elimination of DNA adducts formed by bulky carcinogens. Repair of DPC involves their initial proteolytic degradation by proteasomes^10,12,13 or the metalloprotease SPRTN,^14,15 yielding small DNA-peptide crosslinks as substrates for nucleotide excision repair to complete DPC removal.¹¹ Translesion DNA synthesis across unrepaired peptide-DNA crosslinks can be mutagenic (16,17).^16,17 Consistent with its genotoxic properties, FA caused activation of two apical DNA damage-responsive kinases ATM and ATR which triggered cell cycle checkpoints and promoted cell survival.^18,19 However, these studies also pointed to the existence of some significant DNA damage-independent factors impacting checkpoint signaling in FA-treated cells.

DNA damage is not the only type of chemical injury by carcinogens that can impact genome stability and functions. For DNA amino group-reactive electrophiles such as FA, ε-amino group of histone lysines represents even a more favorable conjugation site in chromatin. In addition to the formation of lysine monoadducts, FA also produces lysine-lysine crosslinks leading to the formation of intra-histone, histone-histone and histone-nonhistone protein crosslinks.^20,21 FA adduction at the ε-amino group of histone lysines is expected to block the formation of posttranslational modifications (PTMs) as shown for acetylation.^22,23 Histone-histone crosslinking can impair conformational mobility of nucleosomes and prevent nucleosome disassembly/eviction during transcription, replication or DNA repair. Chromatin remodeling and specific histone PTMs are known to play a major role in DNA damage responses as part of the orchestrated signaling aimed to recruit repair factors and provide access to damaged DNA.^24,25 In contrast to DNA damage-induced signaling, our understanding of direct chromatin damage responses and their impact on genome stability mechanisms and cell fitness is much more limited.

In this work, we examined levels of histone PTMs in several human cell lines, including three primary lines, after treatments with mildly cytotoxic doses of FA. We unexpectedly found that the loss of K120-monoubiquinated histone H2B (H2Bub1) was the most prominent chromatin change in all FA-treated cells. The depletion of H2Bub1 resulted from its rapid deubiquitination which was independent on DNA damage signaling, indicating a bone fide chromatin damage response. We further determined that H2Bub1 was necessary for robust DNA damage responses and a fully functional S-phase checkpoint after FA exposures. H2Bub1 was also important for suppression of cytotoxicity and genome instability in FA-treated cells. Our work identified depletion of H2Bub1 as a major form of chromatin damage by FA, which adversely affected DNA damage/checkpoint responses and cell fitness. These findings also suggest that a frequent loss of H2Bub1 in cancers could in part result from selection for the attenuated replication checkpoint promoting unrestrained proliferation under chronic genotoxic conditions.

Materials and methods

Cells and treatments

All cells were obtained from ATCC which authenticated these cells. New batches of cells were used every six months. Cells were grown under the supplier’s recommended conditions. FA (Sigma, F8775) treatments were done in complete growth media.

Immunoblotting

Cells were treated at 50–60% confluence and collected by trypsinization. Histones were extracted from nuclei using 0.25 N H₂SO₄ followed by precipitation with 4 volumes of cold ethanol.²⁶ Whole cell lysates were prepared by boiling cells for 10 min in a 2% SDS solution (2% SDS, 50 mM Tris-HCl pH 6.8, 10% glycerol) containing Halt Protease and Phosphatase Inhibitors (ThermoFisher Scientific). Insoluble nuclear proteins (crude chromatin fraction) were dissolved with 2% SDS lysis buffer after removal of soluble proteins using a cell lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM MgCl₂, 0.5% NP40, 2.5% glycerol and protease/phosphatase inhibitors, 15 min on ice, then 10 000g for 10 min at 4°C). Small proteins were separated on 12% SDS-PAGE gels and electrotransferred onto PVDF or nitrocellulose membranes using PierceG2 Fast Blotter and 12% ethanol-containing buffer. Large proteins were separated on 8% or 10% gels and transferred onto PVDF membranes overnight at 18V using cold Bjerrum Schafer-Nielsen Buffer. Histone antibodies: H2Bub1 (Cell Signaling, 5546), H2Aub1 (Cell Signaling, 8240), H2B (Cell Signaling, 12364), H2A (Cell Signaling, 2578), H2AX (Cell Signaling, 7631), γH2AX (Cell Signaling, 9718), H2AZ (Cell Signaling, 2718), H3 (Cell Signaling, 9715), H3K9ac (Cell Signaling, 9649), H3K27ac (Cell Signaling, 8173), H3K4me3 (Millipore, 07–473), H3K9me3 (Abcam, ab8898), H3K27me3 (Cell Signaling, 9733), H3K36me3 (Cell Signaling, 4909), H3K79me3 (EpigenTek, A-4045–050), H4 (Cell Signaling, 13919), H4K16ac (Cell Signaling, 13534), H4K12ac (Active Motif, 39165), H4K20me2 (Abcam, ab9052), and H4K20me3 (Abcam, ab9053). Other primary antibodies: K48-polyubiquitin (Cell Signaling, 4289), ubiquitin (Cell Signaling, 58395), RNF20 (Cell Signaling, 11974), RNF40 (Cell Signaling, 12187), ATXN7L3 (Bethyl, A302–800A), phospho-S15-p53 (Cell Signaling, 9284), p21 (Abcam, ab109199), phospho-S317-CHK1 (Cell Signaling, 2344), CHK1 (Cell Signaling, 2360), phospho-T68-CHK2 (Cell Signaling, 2344), CHK2 (Cell Signaling, 3440), phospho-S1981-ATM (Cell Signaling, 13050), phospho-S326-HSF1 (Abcam, ab76076), DNAPK (Cell Signaling, 12311), phospho-S2056-DNAPK (Abcam, ab18192), phospho-S824-KAP1 (Bethyl, A300–767A), anti-PAR (Trevigen, 4335-MC-100), LMP7 (Cell Signaling, 13635), lamin B1 (Cell Signaling, 12586), fibrillarin (Cell Signaling, 2639), L7A (Cell Signaling, 2415) and γ-tubulin (Sigma, T6557). Horseradish peroxidase-conjugated goat anti-mouse IgG (Millipore, 12–349) and goat anti-rabbit IgG (Cell Signaling, 7074) were used as secondary antibodies. Protein bands were detected using ECL Western Blotting Detection Reagent (GE Life Sciences, RPN2232). Densitometry analysis of bands within a linear response range was performed using ImageJ.

siRNA transfections

Knockdowns of RNF20 and ATXN7L3 were generated using the ON-TARGETplus SMARTpool human siRNA from Dharmacon (RNF20: L-007027–00-0020, ATXN7L3: L-023237–01-0020). Controls were transfected with the ON-TARGETplus Control Pool Non-Targeting pool siRNA (Dharmacon, D-001810–10-10). Mixtures of siRNA (50 nM final concentration in media) and Lipofectamine RNAiMAX (Invitrogen, 13778150) were added to cells in 100-mm dishes for 6 h. The next day, cells were transfected again and seeded for FA treatments delivered 24 h later.

Fluorescence-activated cell sorting (FACS)

Cells were seeded on dishes at ~40–50% confluence and treated with FA on the next day. For the assessment of rates of DNA synthesis,²⁷ cells were labeled with 10 μM EdU during the last hour of FA treatments. Trypsinized cells were fixed overnight in 80% ethanol at 4°C, washed with PBS and then permeabilized with 0.5% Triton X-100 in PBS for 30 min at room temperature. After a wash with PBS, cells were incubated for 30 min in a Click-iT reaction mixture (Invitrogen, C10420) to fluorescently label DNA-incorporated EdU. FACS analysis was performed using FACSCalibur (BD Biosciences) and data were analyzed by the CellQuest Pro software.

Microscopy

Cells were seeded at approximately 60% confluence on human fibronectin-coated coverslips (Electron Microscopy Sciences, 72297–08). Next day, cells were treated with FA for 3 h, washed with PBS and grown in fresh media for 24 h (53BP1 staining). Cells were fixed with 4% paraformaldehyde in PBS for 15 min at room temperature followed by permeabilization with 0.5% Triton X-100 in PBS for 15 min. Non-specific binding was blocked with 2% fetal bovine serum in PBS at 37°C for 30 min. Cells were incubated for 2 h at 37°C with 53BP1 antibody (Cell Signaling, 4937) diluted 1:100 in PBS containing 1% BSA and 0.5% Tween-20. Secondary antibodies (Life Technologies, A11034) were incubated at room temperature for 1 h in the dark. Coverslips were mounted on glass slides with the Vectashield DAPI mounting media (H-1200). Cells were viewed on the Nikon Ti2-E Fluorescence Microscope (40× magnification). At least 50 cells in five or more randomly selected fields were scored per each slide. For scoring of micronuclei, cells were treated with FA for 3 h, allowed to grow for 72 h and then fixed with 2% paraformaldehyde for 15 min followed by permeabilization with 1% Triton X-100 in PBS for 15 min. Coverslips were stained and mounted on glass slides with the Vectashield DAPI mounting media. At least 400 cells were scored per slide.

Cytotoxicity and colony formation

Cell viability was measured using the CellTiter-Glo luminescent assay (Promega, G7571). Cells were seeded (1000 per well) into optical bottom cell culture plates (ThermoFisher Scientific, 165305) and grown overnight before the addition of FA. Cytotoxicity was assayed at 48–72 h post-treatment. For colony formation assay, H460 cells were seeded onto 6-well plates (200 cells/well), grown overnight and then treated with FA for 24 h. Plates were stained with Giemsa solution and colonies with >30 cells were scored.

Statistics

Statistical significance between two independent groups was assessed by two-tailed, unpaired t-test. Data in multiple comparisons were evaluated by one-way ANOVA with Tukey’s post-hoc test. Differences were considered statistically significant when p-value was <0.05.

Results

Small histone PTMs after FA treatments

FA is a histone lysine adducts-forming chemical which impairs the formation of histone PTMs in cell-free reactions.²² Prolonged exposures to FA²⁸ or acetaldehyde²⁹ diminished histone H3K9/K14 and H4K12 acetylation in the cytosol, leading to a lower incorporation of these histones into new chromatin. Short treatments of cells with FA, mimicking exposure from smoking tobacco or e-cigarettes, triggered a chromatin damage-dependent activation of the apical DNA damage-responsive kinase ATM.¹⁹ To identify primary/initial changes in histone modifications elicited by FA damage, we tested 2 h treatments with 200 and 300 μM FA concentrations that approximately corresponded to IC₂₅ and IC₄₀ in primary IMR90 cells (Figure 1A). TERT-immortalized HBEC3 lung epithelial cells were moderately more sensitive (~IC₄₀ for 200 μM and ~IC₆₀ for 300 μM FA), probably reflecting their growth in serum-free medium. Shorter 1 h treatments with FA produced no significant changes in the long-term viability of IMR90 or HBEC3 cells (Figure 1A). Histone PTMs include different chemical forms, with the most abundant being Lys methylation and acetylation.^30,31 We examined the most prominent methylation and acetylation sites in histone H3 and H4 in HBEC3 and two primary cell lines. None of the three cell lines showed noticeable changes in the levels of euchromatin-specific (K4me3, K36me3, K79me3) or heterochromatin-specific (K9me3, K27me3) histone H3 trimethylation after FA (Figure 1B). Several histone H4 modifications, such as the constitutive heterochromatin-specific K20 trimethylation, a widespread mark K20 dimethylation and the euchromatin-specific K12 and K16 acetylation, displayed no appreciable changes in any of the three tested lines (Figure 1C). However, chromatin levels of histone H3K9ac and H3K27ac were diminished in all three cell lines with stronger decreases seen in more sensitive HBEC3 stem-like cells (Figure 1D,E). A moderate decrease in histone H3K9ac has previously been detected in A549 lung carcinoma cells with a higher (1 mM) FA dose.³² Overall, the main H3/H4 lysine trimethylation PTMs, which are firmly associated with either euchromatin or heterochromatin, were not affected by moderately toxic FA treatments. Euchromatic histone acetylation showed divergent responses: no appreciable changes in H4 acetylation and decreases in H3 acetylation.

Fig. 1. Histones H3 and H4 methylation and acetylation in FA-treated human cells.

Immunoblots were done with chromatin fractions of cells treated with FA for 2 h. Fibrillarin and individual histones were used as loading controls. Cell viability was determined at 72 h after FA treatments. (A) Cytotoxicity of FA in IMR90 and HBEC3 human cells (means+/−SD, n=3, *-p<0.05, **-p<0.01 relative to 0 μM FA). (B) Immunoblots for histone H3 and (C) histone H4 posttranslational modifications in TERT-immortalized (HBEC3) and primary human lung cells (IMR90, WI38). (D) Immunoblots for histone H3 acetylation at Lys9 and Lys27 in different human lung cell lines. (E) Quantitation of H3K9 and H3K27 acetylation in FA-treated cells (means+/−SD, n=2 independent sets of cell lysates, *-p<0.05, **-p<0.01, ***-p<0.001 relative to untreated cells).

H2Bub1 and H2Aub1 depletion by FA

Next, we assessed protein damage by FA in the nuclei by measuring protein polyubiquitination. Immunoblotting of isolated histones for K48-linked polyubiquitin showed a higher abundance of high-molecular products in FA-treated primary cells (Figure 2A). In contrast, the amount of the nonproteolytic Lys63-linked polyubiquitin was only minimally increased. Immunoblotting with anti-ubiquitin antibodies, which recognize both mono- and polyubiquitinated proteins, detected the formation of a large polyubiquitin (>250 kDa) and steep losses of mono- and diubiquitinated bands of histones H2A/B (Figure 2B). H2A/B-ub1 did not return to its normal levels even at 4 h after FA removal. H2AK119ub1 and H2BK120ub1 are nonproteolytic modifications that play major roles in chromatin structure and functions, such as transcription.^33,34 H2Aub1 and H2Bub1 have similar molecular weights and cannot be resolved as separate bands on SDS-PAGE gels. To examine their individual levels, we ran immunoblots with specific antibodies for each ub1 form. Three human cell lines, including two primary lines, showed dramatic losses of H2Bub1, with the IC_40–50 dose of 300 μM FA causing >10-fold reduction in its abundance (Figure 2C,D). The same dose of FA also decreased the amounts of H2Aub1 across different cell lines but these changes were more moderate. For example, 200 μM FA caused 3–4-fold reduction in H2Bub1 vs statistically insignificant 10–30% decreases for H2Aub1. The differences in depletion of H2Aub1 and H2Bub1 were consistent among different human cell lines. A preferential depletion of H2Bub1 vs H2Aub1 was further observed in primary cells treated for 24 h with lower FA concentrations (Figure 2E) that were only mildly cytotoxic (Figure 2F). H2Aub1 was more resistant to depletion by FA after 24 h relative to 2 h exposures. All preceding immunoblots analyzed histones in the NP40-insoluble cellular fraction (crude chromatin). It was possible that the observed losses in H2Aub1 and/or H2Bub1 reflected eviction of these histones from chromatin. To evaluate this possibility, we ran immunoblots with whole cell lysates prepared by boiling of cells in 2% SDS, which also irreversibly inactivates cellular deubiquitinases.³⁵ Immunoblots with total cell lysates from two FA-treated cells again showed severe losses of H2Bub1 and more moderate decreases of H2Aub1 (Figure 2G). Cells were slow to restore H2Bub1 levels after FA removal and remained depleted of this PTM at 4 h post-exposure (Figure 2H), a time point when polyubiquitin products already completely disappeared from nuclei (Figure 2A,B). Specificity of antibodies that we used for detection of H2Aub1 and H2Bub1 was confirmed by the loss of bands in cells with inactivated E3 ubiquitin ligases forming these modifications: PRC1 for H2Aub1 (Fig. 2I) and RNF20 for H2Bub1 (Fig. 5A,B). Overall, we found that the loss of H2Bub1 was the most dramatic change in histone PTMs by FA in three human cell lines, including two primary cell lines. Doses of FA corresponding to IC₄₀-IC₅₀ caused >90% loss of H2Bub1.

Fig. 2. Deubiquitination of histones H2A and H2B by FA.

Unless indicated otherwise, cells were treated with FA for 2 h. Immunoblots were done with the insoluble cellular fraction except for panel G where whole cell lysates were used. Fibrillarin and individual histones were used as nuclear loading controls. (A) Immunoblots for Lys48- (K48-ub) and Lys68-linked (K63-ub) polyubiquitin and (B) ubiquitin (Ub) in histones isolated from IMR90 cells. (C) Immunoblots for monoubiquitinated histones H2A and H2B in three human cell lines treated with FA. (D) Quantitation of immunoblots for H2Aub1 and H2Bub1 in three human cell lines (means±SD, n=3 independent sets of cell samples, *-p<0.05, **-p<0.01, ***-p<0.001 relative to 0 μM FA). (E) Immunoblots for H2Aub1 and H2Bub1 in IMR90 cells treated with FA for 24 h. (F) Viability of IMR90 cells assayed at 48 h after 24 h FA treatments (means±SD, n=3, *-p<0.05, **-p<0.01 relative to 0 μM FA). (G) Immunoblots of monoubiquitinated H2A and H2B using whole cell lysates of IMR90 and HBEC3 cells. (H) Amounts of H2Bub1 in IMR90 cells at different times after FA removal. (I) Loss of H2Aub1 signal in IMR90 cells treated with the PRC1 inhibitor PRT4165 (PRC1i, 50 μM, 3 h).

Independence of H2Bub1 losses on other stress responses

FA activates DNA damage responses initiated by two apical kinases ATM and ATR.^24,33 To examine a potential involvement of genotoxic stress signaling in H2Bub1 depletion, we tested the impact of inactivation of three main DNA damage-responsive kinases ATM, ATR and DNAPK using a set of their inhibitors which we have previously validated in human lung cells.³⁶ Inactivation of ATM, ATR or DNAPK by these inhibitors did not prevent the loss of H2Bub1 by FA in primary IMR90 cells (Figure 3A). The use of another set of inhibitors of ATM, ATR and DNAPK again did not affect the losses of H2Bub1 by FA in another (WI38) primary cell line (Figure 3B). These inhibitors were confirmed to be effective towards their known targets in validation experiments (Fig. 3C). Inhibition of PARP, a multifunctional regulator of DNA damage responses through the formation of poly-ADP-ribose on histones and other nuclear proteins,³⁷ also had no apparent effects on the dose- and time-dependent losses of H2Bub1 by FA (Figure 3D). The employed concentration of the PARP inhibitor olaparib was confirmed to be very effective in elimination of the poly-ADP-ribose formation by its canonical inducer H₂O₂ (Fig. 3E). Next, we examined chromatin amounts of H2Bub1 in cells treated with several mechanistically distinct genotoxic agents: topoisomerase I inhibitor camptothecin, topoisomerase II inhibitor etoposide, radiomimetic bleomycin and DNA adducts-forming chromate and cisplatin. Unlike FA, none of these genotoxicants caused detectable protein damage in cells, as evidenced by the absence of proteolytic K48-polyubiquitination and activating HSF1-S326 phosphjorylation³⁸ (Figure 3F). These DNA-damaging agents also failed to significantly alter chromatin levels of H2Bub1 (Figure 3F,G). Pretreatments of cells with the histone deacetylase inhibitor trichostatin A before the addition of FA increased the overall levels of histone acetylation and prevented decreases in H3K9ac and H3K27ac (Figure 3H). However, this increase in histone acetylation did not affect the ability of FA to deplete H2Bub1 or H2Aub1. Addition of trichostatin A after FA exposures also strongly elevated levels of H3K9 and H3K27 acetylation but it had no apparent impact on recovery of H2Bub1 (Fig. 3I). Thus, H2B deubiquitination by FA is a major chromatin change that is independent on genotoxic stress signaling or histone acetylation.

Fig. 3. Independence of H2B deubiquitination by FA on DNA damage responses.

Primary human cells were pretreated with inhibitors of DNA damage responses for 1 h prior to the addition of FA. Cells were incubated with FA or DNA-damaging chemicals for 2 h. H2Bub1 was measured in the chromatin fraction. Histone H3 and fibrillarin were used as loading controls. (A) Chromatin levels of H2Bub1 in IMR90 cells pretreated with the set #1 of DNA damage response inhibitors (ATMi1 – 10 μM KU55933, DNAPKi1 – 30 μM NU7026, ATRi1 – 3 μM ETP46464). (B) H2Bub1 in chromatin of WI38 cells pretreated with the set #2 of DNA damage response inhibitors (ATMi2 – 2 μM KU60019, DNAPKi2 – 3 μM NU7441, ATRi2 – 10 μM VE821). (C) Validation of the set #2 inhibitors in cells treated with 5 μM bleomycin for 1 h. Cells were pretreated with inhibitors for 1 h before the addition of bleomycin. (D) H2Bub1 in chromatin of IMR90 cells treated with FA in the presence of the PARP inhibitor olaparib (5 μM). (E) Inhibition of poly-ADP-ribose (PAR) formation in IMR90 cells preincubated for 30 min with 5 μM olaparib and then treated with H₂O₂ for 1 h. (F) Immunoblots of IMR90 cells treated with various DNA-damaging chemicals (CPT- camptothecin, Eto – etoposide, Bleo - bleomycin, Cr – K₂CrO₄, Pt – cisplatin). Whole cell lysates were analyzed with antibodies against Lys48-linked polyubiquitin (K48-ub), S326-phosphorylated HSF1 (p-HSF1) and S139-phosphorylated histone H2AX (γH2AX). H2Bub1 was measured in the insoluble nuclear fraction. (G) Quantitation of H2Bub1 from immunoblots of the chromatin fraction of IMR90 cells treated as in panel F (means±SD, n=3, ***-p<0.001). (H) Immunoblots of IMR90 cells pretreated with 2 μM trichostatin A (TSA) for 1 h before the addition of FA for 2 h. (I) Recovery of H2Bub1 post-FA in the absence and presence of 2 μM trichostatin A (TSA).

Mechanisms of H2Bub1 and H2Aub1 decreases by FA

The heterodimer of RNF20 and RNF40 is the main E3 ligase for the formation of H2Bub1.^33,34 We found no marked changes in RNF20 or RNF40 abundance after 2 or 24 h FA treatments in primary cells (Figure 4A). The loss of H2Aub1 can occur under conditions of a limited availability of free ubiquitin.³⁹ The pool of free ubiquitin is decreased upon the accumulation of polyubiquitinated proteins and the resulting trapping of ubiquitin. Inhibition of HSP90 led to the hyperaccumulation of polyubiquitinated proteins in FA-treated cells due to increased instability of damaged proteins.⁴⁰ Enhancement of the cellular demand for ubiquitin by HSP90 inhibition did not change H2Bub1 depletion by FA but strongly exacerbated the loss of H2Aub1 (Figure 4B), demonstrating a higher sensitivity of H2Aub1 to a free ubiquitin deficiency. Inhibition of immunoproteasomes, which remove FA-damaged proteins via 19S/ubiquitin-independent proteolysis,⁴¹ did not affect FA-induced deubiquitination of H2A or H2B (Figure 4C), indicating that protein polyubiquitination, not the overall burden of damaged proteins, increases the loss of H2A-conjugated ubiquitin. Next, we examined kinetics of their deubiquitination by FA in the presence of the E1 ubiquitin ligase inhibitor TAK243. H2Bub1 was a stable modification in TAK243-treated primary cells but it was rapidly lost upon the addition of FA (Figure 4D), demonstrating activation of deubiquitination. H2Aub1 was less stable and its deubiquitination rate was not affected by FA (Figure 4E), indicating that depletion of this PTM in FA-treated cells reflected its decreased formation. The loss of H2Aub1 but not that of H2Bub1 tightly correlated with FA-induced decreases in the cellular pool of free ubiquitin (Figure 4F,G). These findings are consistent with a faster turnover of H2Aub1 making its levels more sensitive to the availability of ubiquitin. All cells express two quantitatively minor but physiologically important variants of histone H2A: H2AZ and H2AX. H2AZub1 and H2AXub1 were more unstable than the canonical H2Aub1 and their rates of deubiquitination were also unaffected by FA (Figure 4H). Thus, depletion of H2Bub1 and H2Aub1 in response to FA results from different mechanisms: a more rapid deubiquitination for H2Bub1 and a slower ubiquitination for H2A.

Fig. 4. Dynamics of H2A and H2B deubiquitination.

(A) Protein abundance of RNF20 and RNF40 in primary human cells treated with FA for 2 or 24 h. (B) Impact of HSP90 inhibition on H2Aub1 and H2Bub1 in IMR90 cells (2 h FA alone or in the presence of 2 μM DMAG added 1 h earlier). (C) Immunoblots of IMR90 cells pretreated for 1 h with 0.3 μM ONX-0914 (LMP7i) prior to the addition of FA for 2 h. LMP7 was analyzed in the soluble extracts and histones in the insoluble nuclear fractions. Shifted bands of LMP7 in LMP7i-treated cells was caused by a covalent attachment of the drug. (D) H2B and (E) H2A deubiquitination rates in control and 300 μM FA-treated IMR90 cells in the presence of the E1 ubiquitin ligase inhibitor TAK-243 (1 μM, added 15 min before FA). Graphs show means±SD for two independent sets of samples (**-p<0.01 relative to controls). (F) Immunoblots of free ubiquitin (Ub), H2Aub1 and H2Bub1 in control and FA-treated IMR90 cells. (G) Dose-dependent changes in free ubiquitin, H2Aub1 and H2Bub1 in IMR90 cells treated with FA for 2 h. Data are means±SD for 3 independent sets of cell lysates (**-p<0.01, ***-p<0.01 relative to free ubiquitin). (H) Deubiquitination rates for histones H2AZ and H2AX. Top bands at ~23 kDa in immunoblots for H2AZ and H2AX correspond to monoubiquitinated forms of these histones. Cells were treated as in panel F.

DNA damage responses and FA tolerance in cells with RNF20 knockdown

To understand the role of H2Bub1 in cellular responses to FA, we examined cells with siRNA-depleted RNF20 which is essential for H2B monoubiquitination.^33,34 DNA damage responses linked to the activation of ATR (assayed by phospho-CHK1) and ATM (assayed by phospho-CHK2) were repressed in primary cells with depleted RNF20/H2Bub1 (Figure 5A). A weaker phosphorylation of CHK1 and CHK2 in RNF20 knockdown cells was also found at 4 h post-FA (Figure 5B). We next examined rapidly proliferating H460 lung cells which display a more robust S-phase checkpoint signaling. Similar to primary cells, H460 cells with depleted H2Bub1 were deficient in the stimulation of ATM- (CHK2) and ATR-dependent phosphorylation (CHK1) (Figure 5C,D). Activation of the transcription factor p53, measured by its abundance, S15 phosphorylation and the upregulation of its target protein p21, was also impaired by RNF20 knockdown (Figure 5D). Upregulation of p53 by FA in H460 and primary cells is mediated by ATR kinase.¹⁸ Depletion of H2Bub1 by RNF20 knockdown also suppressed γH2AX monoubiquitination by FA (Figure 5E). In contrast to H2B ubiquitination, histone acetylation did not modulate checkpoint signaling as the treatment of cells with the histone deacetylase inhibitor trichostatin A, which strongly increased histone H3 acetylation and prevented its loss by FA (Figure 3H), produced no noticeable effects on CHK1 or CHK2 phosphorylation (Figure 5F).

Fig. 5. Checkpoint responses and tolerance of FA in cells with RNF20 knockdown.

(A) Immunoblots of primary WI38 cells treated with FA for 2 h. Insoluble nuclear proteins (H2Bub1 and lamin B1) and soluble proteins (phospho-CHK1, phospho-CHK2, tubulin) were examined. (B) DNA damage signaling in WI38 cells treated with FA for 3 h and collected 4 h later for preparation of whole cell extracts. (C) ATM-related signaling in H460 cells treated with FA for 3 h and collected for whole cell lysates 4 h later. (D) CHK1 phosphorylation and activation of p53 in H460 cells treated as in panel C. (E) Impaired γH2AXub1 formation by FA in H460 cells with RNF20 knockdown. Long and short exposures of the same γH2AX immunoblot are shown. Cells were treated as in panel C. (F) Phosphorylation of CHK1 and CHK2 in primary cells pretreated with 2 μM trichostatin A for 1 h before the addition of FA for 2 h. (G) Representative FACS profiles of EdU incorporation in control and RNF20 knockdown H460 cells treated with FA for 3 h. (H) Viability of H460 cells treated with FA for 3 h and assayed after 48 h recovery (means±SD, n=3). (I) Viability of H460 cells treated with FA for 24 h and assayed after 24 h recovery (means±SD, n=3). (J) Viability of primary WI38 cells treated with FA for 24 h and assayed after 48 h recovery (means±SD, n=6). Statistics: * - p<0.05, ** - p<0.01, *** - p<0.001.

DNA damage responses by ATM and ATR are initiated by nuclear damage in S-phase of FA-treated cells.^18,19 In light of repressive effects of RNF20 knockdown on checkpoint signaling by FA, we next examined DNA synthesis by EdU labeling. We found that a large fraction of cells with depleted H2Bub1 by RNF20 knockdown showed a continuing DNA synthesis after 100 μM FA, which caused a complete cessation of DNA replication in control cells (Figure 5G). Viability measurements in H460 and primary cells using short and subchronic treatments all showed a more severe cytotoxicity of FA in populations with RNF20 knockdown (Figure 5H,I,J). A long-term survival of H460 cells assessed by the colony formation assay was also decreased by RNF20 knockdown (24 h treatments with 30 μM FA: 81.7±2.0% vs 66.2±3.8% for si-control and si-RNF20, respectively (p=0.0033, n=3), 50 μM FA: 18.5±3.4% vs 10.2±1.6% (p=0.019, n=3). Thus, a normal H2Bub1 availability promotes the establishment of a more efficient S-phase checkpoint and a better survival of cells damaged by FA.

Checkpoint signaling in cells with high H2Bub1

To test the impact of high H2Bub1, we depleted cells of ATXN7L3 which is a component of three H2B deubiquitinases: USP22, USP51 and USP27X.⁴² ATXN7L3 knockdown in primary cells strongly increased abundance of H2Bub1 (>10-fold) and delayed but did not prevent the H2Bub1 loss in response to FA (Figure 6A). Formation of proteolytic K48-polyubiquitin by FA in the insoluble nuclear proteins was also similar in cells with normal and elevated H2Bub1. In contrast, checkpoint signaling assayed by CHK1 and ATM phosphorylation was clearly enhanced in cells with high H2Bub1 (Figure 6B). ATXN7L3 knockdown in H460 carcinoma cells also strongly elevated basal H2Bub1 but did not affect its decrease by FA (Figure 6C), which was much smaller than in primary cells. Levels of the genotoxic stress marker γH2AX and especially, its monoubiquitinated form were elevated in FA-treated cells with increased H2Bub1. Next, we tested the impact of high H2Bub1 on S-phase checkpoint by measuring EdU incorporation by FACS. Knockdown of ATXN7L3 had no appreciable effect on the basal rate of DNA synthesis but it diminished EdU incorporation after FA (Figure 6D), indicating activation of a more robust S-phase checkpoint. To determine the role of checkpoint signaling in the observed differences in DNA synthesis, we treated cells with FA in the presence of ATM and ATR inhibitors to suppress main DNA damage responses. We found that inactivation of ATM/ATR in control and ATXN7L3-depleted cells eliminated their differences in rates of DNA synthesis by FA (Figure 6E). Suppression of checkpoint signaling did not rescue DNA synthesis completely which is also inhibited directly by FA-induced DNA damage, especially by replicative helicase-blocking DPCs.^9,18 Efficiency of S-phase checkpoint under stress conditions affects the completion of DNA replication prior to entry of cells into mitosis. Mitotically transmitted unreplicated DNA is sequestered in G1 phase in protective nuclear structures known as 53BP1 bodies.⁴³ Our examination of S-phase checkpoint showed its dependence on the abundance of H2Bub1 in cells. To test whether these differences translate into different amounts of unreplicated DNA in G1 cells, we scored the frequency of cells with 53BP1 bodies. In agreement with checkpoint efficiency results, FA induced a significantly higher formation of 53BP1 bodies in cells with RNF20 knockdown (low H2Bub1, weak checkpoint) and a significantly smaller number of these structures in cells with depleted ATXN7L3 (high H2Bub1, strong checkpoint) (Figure 6F). Scoring of micronuclei, a well-established marker of chromosomal instability, further confirmed a protective role of H2Bub1 in suppression of genome abnormalities by FA in primary WI38 cells as evidenced by a significantly higher frequency of micronuclei in cells with depleted RNF20 (low H2Bub1) and a smaller number of micronuclei in ATXN7L3 knockdown (high H2Bub1) after FA treatments (Figure 6G).

Fig. 6. FA-induced DNA damage responses in cells with elevated H2Bub1.

(A) Immunoblots of insoluble nuclear proteins from WI38 cells treated with 300 μM FA. (B) Immunoblots of soluble proteins from WI38 cells treated with 300 μM FA. (C) Immunoblots of soluble (ATXN7L3, L7A) and insoluble proteins (other blots) from H460 cells treated with 300 μM FA. Long and short exposures of the same γH2AX blot are shown. (D) Representative FACS profiles of EdU labeling of control and ATXN7L3-depleted H460 cells treated with FA for 3 h. EdU was added for the last 1 h of FA incubations. (E) Representative FACS profiles of DNA synthesis in H460 cells treated with FA for 3 h in the presence of ATM (10 μM KU55933, ATMi) and ATR (1 μM AZD6738, ATRi) inhibitors. (F) Representative image of 53BP1 (green) and DAPI (blue)-stained H460 cells and a graph with the frequency of FA-treated cells containing two or more 53BP1 bodies. Cells were treated with FA for 3 h and fixed for microscopy 24 h later. Background-subtracted data are shown (means±SD, n=4). (G) Frequency of micronuclei in primary WI38 cells with normal (si-ctrl), depleted (si-RNF20) or overexpressed (si-ATXN7L3) H2Bub1 after FA damage. Cells were treated with 0 or 150 μM FA for 3 h and micronuclei were scored after 72 h recovery (*- p<0.05, **-p<0.01, n=3).

Discussion

FA conjugation at histone lysines was shown to interfere with enzymatic acetylation of these sites in vitro²² and the presence of N⁶-formyllysine-histone adducts was also detected in cellular chromatin.²³ FA adduction is also expected to inhibit Lys methylation. We did not detect noticeable changes in trimethylation levels of several H3-Lys and H4-Lys residues associated with transcriptionally active or inactive chromatin. Euchromatic H4K12 and H4K16 acetylation was also unaffected in all FA-treated cells. However, H3K9ac and H3K27ac, that are also euchromatin-specific PTMs, showed dose-dependent decreases by FA. Unchanged levels of the majority of examined histone PTMs indicate that the frequency of FA adducts at histone lysines at toxicologically moderate doses (IC₂₀-IC₅₀) probably much lower than the normal frequency of these PTMs. These considerations would also suggest that decreases in H3K9ac and H3K27ac likely resulted from an active response to FA damage rather than a direct inhibition of acetylation by Lys-conjugated FA. Stability of H3K9ac and H3K27ac in cells with inactivated deacetylase activity pointed to a faster deacetylation as a cause of lower H3 acetylation by FA.

Our examination of histone ubiquitinated products led to the unexpected discovery of a severe depletion of H2B-K120ub1 by FA. A far more abundant monoubiquitination of H2A-K119 was also decreased by FA but to a much lesser extent. After 300 μM FA treatments in primary cells, losses of H2Aub1 were ~10-times less severe than those for H2Bub1. Decreases in H2Aub1 are likely linked to its function as a source of ubiquitin under conditions of proteotoxic stress depleting free nuclear ubiquitin.³⁹ FA was previously found to cause accumulation of K48-linked polyubiquitin in the nucleus⁴⁰ and here we detected these proteasome-targeting products in the histone and chromatin preparations from FA-treated cells. The main mechanism for the loss of H2Aub1 after FA was its diminished ubiquitination, which is consistent with the sensitivity of H2Aub1 to levels of free ubiquitin. The loss of H2Bub1, which was a more stable modification than H2Aub1 under normal conditions, was mechanistically different and resulted from its rapid deubiquitination upon addition of FA. FA appeared to activate redundant H2Bub1 deubiquitinases, as evidenced by a slowdown but not prevention of H2Bub1 losses in primary cells with depleted ATXN7L3 which is a cofactor for three H2B deubiquitinating enzymes.⁴² Depletion of H2Bub1 by FA was independent on the main apical DNA damage-activated kinases, histone acetylation, PARP or HSP90 activity. Thus, the FA-induced loss of H2Bub1 represents a major chromatin damage response that is induced separately from other stress responses linked to DNA damage or accumulation of polyubiquitinated proteins. H2Bub1 is preferentially localized to the coding regions of actively transcribed genes where it promotes RNA polymerase II elongation.^33,34,44 Therefore, rapid and severe losses of H2Bub1 are expected to decelerate ongoing transcription and diminish the incidence of stalled RNA pol II on FA-damaged chromatin. Alternatively, transcription inhibition due to loss of H2Bub1 could enhance FA cytotoxicity by limiting protective gene expression responses. Parallel FA-induced decreases in H3K9ac and H3K27ac, histone modifications boosting the activity of gene promoters and enhancers,^30,31 represent a second mechanism for the transcription attenuation by limiting initiation rates. A potential trigger for H2B deubiquitination in FA-damaged chromatin is a mechanical/topological stress due to protein-protein crosslinking between histones and higher-order chromatin architectural proteins such as CTCF,⁴⁵ inhibiting conformational dynamics in the transcriptionally active chromatin loops.

Our work showed that H2Bub1 was also a critical regulator of DNA damage signaling and S-phase checkpoint after FA (summarized in Figure 7). Both ATM and ATR branches of DNA damage responses were impaired by depletion of H2Bub1 and enhanced by its upregulation. It seems unlikely that these effects were induced by chromatin decompaction, as histone hyperacetylation did not alter activity of ATM or ATR pathways. A more plausible explanation for the enhancement of checkpoint signaling by H2Bub could be its ability to recruit specific DNA damage response factors.⁴⁶ One of these proteins is the E3 ubiquitin ligase RNF168 catalyzing H2A/γH2AX-K15 ubiquitination,⁴⁷ a histone modification that is important for the recruitment of BRCA1-BARD1 complex⁴⁸ and DNA damage response amplification.^33,47 FA-treated cells showed low levels of γH2AXub1 in RNF20 knockdowns and higher levels of this form in ATXN7L3 knockdowns although it is currently unclear whether these changes in monoubiquitination involved K15 or some other site. The BRCA1-RAP80 complex also showed a high affinity for binding nucleosomal H2Bub1,⁴⁶ which points to the positive impact of H2Bub1 on many functions of BRCA1. In addition to its role in homologous recombination, BRCA1 is also known to promote checkpoint signaling,⁴⁹ including p53-S15 phosphorylation and p21 induction⁵⁰ and S-phase checkpoint,⁵¹ which we found to be H2Bub1-sensitive responses to FA. A stronger ATM activation could be linked to another H2Bub1-binding protein – MRG15.⁵² MRG15 is a component of KAT5 acetylase whose chromatin recruitment was required for the acetylation-mediated ATM activation by FA.¹⁹ H2Bub1 levels are much lower in various malignancies,³⁴ including in the most common form of lung cancer – adenocarcinomas.⁵³ Our findings point to a possibility that the loss of H2Bub1 in cancers could in part occur as a result of selection for diminished checkpoint signaling to allow unrestrained proliferation under chronic replication stress.

Fig. 7.

Flow chart summarizing the response of H2Bub1 to chromatin damage and its role in checkpoint signaling and genome instability by FA. The loss of H2Bub1 after FA damage was caused by its accelerated deubiquitination. Transcription-repressive effects of H2Bub1 depletion are assumed based on the known properties of this PTM in transcription elongation.^34,44

Abbreviations

DPC

DNA-protein crosslink

FA

formaldehyde

H2Aub1

monoubiquitinated histone H2A

H2Bub1

monoubiquitinated histone H2B

PTM

posttranslational modification

Data Availability.

All data supporting the findings of this study are available within the article.