Redundant function of DNA ligase 1 and 3 in alternative end-joining during immunoglobulin class switch recombination

Significance

DNA ligase IV (Lig4) is essential for nonhomologous end-joining (NHEJ), the major pathway for repairing DNA double-strand breaks in mammalian cells. An ill-defined alternative end-joining (A-EJ) pathway can also mediate end-joining in cells deficient in NHEJ, although A-EJ is kinetically slower and less accurate than NHEJ. Here, we report that either Lig1 or Lig3 can mediate alternative end-joining in Lig4 knockout cells. Cells having only one ligase (Lig1 or nuclear Lig3) are capable of DNA replication and DNA repair after exposure to a wide range of genotoxins. These results demonstrate the remarkable (and unexpected) plasticity of DNA ligases in mammalian cells.

Abstract

Nonhomologous end-joining (NHEJ) is the major DNA double-strand break (DSB) repair pathway in mammals and resolves the DSBs generated during both V(D)J recombination in developing lymphocytes and class switch recombination (CSR) in antigen-stimulated B cells. In contrast to the absolute requirement for NHEJ to resolve DSBs associated with V(D)J recombination, DSBs associated with CSR can be resolved in NHEJ-deficient cells (albeit at a reduced level) by a poorly defined alternative end-joining (A-EJ) pathway. Deletion of DNA ligase IV (Lig4), a core component of the NHEJ pathway, reduces CSR efficiency in a mouse B-cell line capable of robust cytokine-stimulated CSR in cell culture. Here, we report that CSR levels are not further reduced by deletion of either of the two remaining DNA ligases (Lig1 and nuclear Lig3) in Lig4^−/− cells. We conclude that in the absence of Lig4, Lig1, and Lig3 function in a redundant manner in resolving switch region DSBs during CSR.

DNA double-strand breaks (DSBs) are one of the most severe forms of DNA damage that can result from pathological conditions such as replication stress, exposure to ionizing radiation (IR), free radicals, or other DNA-damaging drugs or because of failed single-strand break repair (SSBR) (1, 2). In developing lymphocytes, programmed DSBs are essential intermediates for antigen receptor gene rearrangements, including V(D)J recombination and Ig heavy chain class switch recombination (CSR) (1, 2). Homologous recombination (HR) and nonhomologous end-joining (NHEJ) are the two major pathways for DSB repair. Whereas HR is restricted to the S/G₂ phase of the cell cycle, NHEJ is active throughout the cell cycle and is generally considered the major pathway for DSB repair in mammals (1, 2).

The NHEJ pathway has been extensively studied. The core components include the Ku70/Ku86 heterodimer, DNA-dependent protein kinase, X-ray cross complementation factor 4 (XRCC4), and DNA ligase IV (Lig4) (1, 2). Additional NHEJ factors include the Artemis nuclease, XRCC4-like factor (XLF) (or Cernunnos), Paralog of XRCC4 and XLF, and Polymerases µ and λ. Missing any of these factors results in various degrees of DSB repair deficits that are highly context-dependent. In general, cells lacking core components of NHEJ are hypersensitive to IR and abolished for V(D)J recombination but are only partially defective for CSR and competent for circulation of transfected linearized plasmids, suggesting that there exists an “alternative” way to join at least some types of DSBs. This alternative end-joining (A-EJ) pathway has recently become a focal area of research because of its implications in oncogenic chromosomal translocations (3), which are rare in NHEJ-proficient cells but much more frequent when NHEJ is compromised. Little is known about A-EJ other than it is kinetically slow and uses an increased level of microhomology (nucleotide overlaps that can be assigned to either of the two DNA ends) during joining (2, 4). A number of DNA repair factors, many of which are involved in SSBR, have been implicated in A-EJ (5), but the overall composition of A-EJ remains elusive. It is still unclear whether A-EJ is a distinct pathway, consists of multiple subpathways, or is merely an aberrant form of NHEJ with missing components substituted by compatible but less efficient factors. It is also unclear whether A-EJ contributes to DSB repair in NHEJ-proficient cells at all or is only active when NHEJ is compromised.

Much of our understanding of mechanistic details of DSB repair has derived from studies of V(D)J recombination and CSR; both involving DSB intermediates (1). V(D)J recombination is initiated by the recombination-activating genes (RAGs) that bind and cleave at specific DNA sequences flanking the V, D, and J segments to assemble an exon encoding the variable (antigen binding) domain of the B- and T-cell receptors. CSR is initiated by activation-induced cytidine deaminase (AID) in antigen-stimulated B cells that changes the IgH constant (C) region to a different isotype. AID catalyzes DNA cytosine deamination (converting cytosines to uracils) at switch regions preceding each C region (6, 7). Processing of AID-generated uracils, through a mechanism still not fully characterized, leads to DSB formation. Although both processes use NHEJ to join DSBs, in cells missing any of the core components of NHEJ, CSR is only partially defective, whereas V(D)J recombination is completely abolished. It has been reported that the RAG complex holds the four broken ends in a postcleavage complex and directs VDJ-associated DSBs into the NHEJ pathway (8, 9). In contrast, significant levels of CSR can occur in the absence of any core NHEJ factors (10, 11–14), suggesting that switch region breaks are more accessible to alternative DSB repair pathways.

Regardless of how broken DNA ends are processed, at least one DNA ligase is required to ligate the two ends. Vertebrates have three ATP-dependent DNA ligases (Lig1, Lig3, and Lig4) (15). Lig1 and Lig4 are conserved in all eukaryotes, whereas Lig3 is only present in vertebrates (15). Lig4 is a core component of the NHEJ pathway and functions exclusively in NHEJ. Cells deficient for Lig4, or its cofactor XRCC4, display the most severe phenotypes of NHEJ deficiency. In the absence of Lig4, A-EJ must rely on Lig1 or Lig3 (or both). It is generally accepted that the major role of Lig1 is to join Okazaki fragments during DNA replication, and this function is mediated by an interaction with the proliferating cell nuclear antigen (PCNA) (16). Lig3 is produced in somatic cells in two forms (mitochondrial and nuclear) via alternative translation initiation (17). It was recently shown that mitochondrial, but not nuclear, Lig3 is essential for cell viability (18, 19). Nuclear Lig3 stably interacts with the X-ray complementation factor 1 (XRCC1), a scaffold protein that is essential for base excision repair. For this reason, Lig3 is regarded as the primary DNA repair ligase for SSBR, although Lig1 has also been implicated in DNA repair (15).

We have previously reported the disruption Lig4 and Lig1 in a mouse B-cell line (CH12F3) capable of robust cytokine-induced CSR (10). Lig4^−/− CH12F3 cells undergo kinetically slow but significant levels of CSR [∼50% of wild-type (WT) level at day 3] (10). Switch junctions isolated from Lig4^−/− CH12F3 cells show increased microhomology and no direct joins (10). In the present study, we focused on determining which of two remaining ligases (Lig1 and Lig3) is responsible for A-EJ during CSR in Lig4^−/− cells. To that end, we disrupted Lig1 and Lig3 (nuclear) individually in Lig4^−/− CH12F3 cells. We found that Lig1 and Lig3 have redundant functions in DNA repair in response to a variety of DNA-damaging agents and during repair by A-EJ during CSR.

Results

Generation of a Lig1^−/−Lig4^−/− Cell Line.

We have previously established Lig4^−/− and Lig1^−/− cell lines (10, 20) by somatic gene targeting in a mouse mature B-cell line (CH12F3) capable of robust cytokine-inducible CSR in cell culture. There is no clear difference between Lig1^−/− and WT cells with regard to cell proliferation, sensitivity to a variety of DNA-damaging agents, or CSR (20). Lig4^−/− cells are, as expected, hypersensitive to Zeocin (a DSB-inducing drug) (10). CSR is reduced, but not abolished, in Lig4^−/− cells (10). To determine whether Lig1 is responsible for the residual CSR activity in Lig4^−/− cells, we performed gene targeting of Lig1 in Lig4^−/− cells to generate cells that are deficient for both Lig1 and Lig4 (Lig1^−/−Lig4^−/−). Disruption of Lig1 was carried out by two rounds of gene targeting as previously described (20). Successful disruption of Lig1 was confirmed by Southern blot (Fig. S1) and immunoblotting (Fig. 1A). Proliferation of Lig1^−/−Lig4^−/− cells is similar to that of Lig4^−/− cells (Fig. 1B), which is slightly slower than WT cells when the cells are not stimulated, but markedly slower upon cytokine stimulation. These data indicate that Lig3 (the only DNA ligase left in Lig1^−/−Lig4^−/− cells) is sufficient to support DNA replication (i.e., Okazaki fragment ligation).

Fig. S1.

Gene targeting of Lig1 in Lig4^−/− cell line. (A) Genomic organization of the WT and targeted Lig1 allele. Exons are indicated by numbered boxes. Arrows indicate transcription orientations of expression cassettes of the puromycin-resistant gene (Puro) and diphtheria toxin A chain (DTA), respectively. “B” indicates BamHI restriction sites. The probe used in Southern blot analysis is depicted at the top. The plus symbol indicates the WT allele. “∆” indicates targeted alleles after Puro cassette excision. Triangles indicate loxP sites. (B) Southern blot analysis of BamHI-digested genomic DNA from WT (+/+) and targeted (∆/∆) cells. Genotypes, sizes of bands, and probes are indicated.

Fig. 1.

Generation and proliferation of Lig1^−/−Lig4^−/− cells. (A) Immunoblot of Lig1, Lig3, Lig4, and β-actin proteins in WT and Lig1^−/−Lig4^−/− cells; the asterisk indicates a nonspecific band. (B) Live cell counts of WT, Lig1^−/−, Lig4^−/−, and Lig1^−/−Lig4^−/− cells in unstimulated (-CIT) or stimulated (+CIT) cultures (CIT: anti-CD40, interleukin 4, and TGFβ1). Error bars indicate SE of three independent experiments.

DNA Repair and CSR in Lig1^−/−Lig4^−/− Cells.

Previously, we have shown that Lig1^−/− cells display no obvious DNA repair defects in response to exposure to a variety of DNA-damaging agents [although these cells display modest sensitivity to methyl methanesulfonate (MMS)] (20). Lig1^−/−Lig4^−/− cells are similarly hypersensitive to Zeocin as Lig4^−/− cells (Fig. 2A). However, the modest MMS-sensitivity of Lig1^−/− cells is exacerbated in Lig1^−/−Lig4^−/− cells (Fig. 2A). Lig1^−/−Lig4^−/− cells show no hypersensitivity to cisplatin, hydroxyurea, or camptothecin, indicating Lig3 is sufficient to repair DNA damage induced by these drugs. No additional CSR defect was observed in Lig1^−/−Lig4^−/− cells beyond that observed in Lig4^−/− cells (Fig. 2B), indicating that Lig1 is not essential for A-EJ–mediated CSR, although this result does not rule out Lig1’s participation in A-EJ.

Fig. 2.

Drug sensitivity and CSR analysis of Lig1^−/−Lig4^−/− cells. (A) Sensitivity of WT, Lig1^−/−, Lig4^−/−, and Lig1^−/−Lig4^−/− cells to DNA-damaging agents. (B) CSR efficiency (percentage of IgA-positive cells) in WT, Lig1^−/−, Lig4^−/−, and Lig1^−/−Lig4^−/− cells. Error bars indicate SE of three independent experiments.

Generation of Lig3^nuc−/− and Lig3^nuc−/−Lig4^−/− Cell Lines.

Disruption of Lig3 was initially attempted with conventional gene targeting (Fig. S2). Only Lig3^+/− and Lig3^+/−Lig4^−/− cells, but no Lig3^−/− or Lig3^−/−Lig4^−/− cells, were obtained. The failure to generate Lig3-null cells is consistent with recent reports showing that mitochondrial Lig3 is essential for cell viability (18, 19). Thus, we designed a strategy to selectively ablate nuclear Lig3 without disrupting mitochondrial Lig3. In Lig3^+/− or Lig3^+/−Lig4^−/− cells (Fig. 3 A and B), clustered regularly interspaced short palindromic repeats (CRISPR)-mediated editing was performed to target the second “ATG” codon of Lig3 (Fig. 3B). It has been shown that mutating this ATG effectively eliminates nuclear-targeted Lig3 while preserving the mitochondrial Lig3 (17). Several cell clones with in-frame deletions at this ATG on the “+” allele in Lig3^+/− and Lig3^+/−Lig4^−/− cells were obtained. The absence of nuclear Lig3 in these cell clones (hereafter termed Lig3^nuc−/− and Lig3^nuc−/−Lig4^−/−, respectively) was confirmed by cell fractionation and immunoblotting analyses (Fig. 3 C and D). Proliferation of Lig3^nuc−/− cells is comparable to that of WT cells, and proliferation of Lig3^nuc−/−Lig4^−/− cells is comparable to that of Lig4^−/− cells (Fig. 3E), indicating that ablation of nuclear Lig3 does not affect cell growth.

Fig. S2.

Gene targeting of Lig3. (A) Genomic organization of the WT and targeted Lig3 allele. Exons are indicated by numbered boxes. Arrows indicate transcription orientations of expression cassettes of the puromycin-resistant gene (Puro) and diphtheria toxin A chain (DTA), respectively. “H” indicates Hind III restriction sites. The probe used in Southern blot analysis is depicted at the top. The plus symbol indicates the WT allele. “P” indicates targeted alleles containing the Puro cassette. Triangles indicate loxP sites. (B) Southern blot analysis of Hind III-digested genomic DNA from WT (+/+) and targeted (+/P) cells. Genotypes, sizes of bands, and probes are indicated.

Fig. 3.

Gene targeting of Lig3 in Lig4^−/− cells. (A) Mitochondrial and nuclear forms of Lig3 produced via alternative translation initiation. M1 and M2, initiation methionine residues for the mitochondrial and nuclear Lig3, respectively. MLS, mitochondrial localization signal. (B) CRISPR-mediated disruption of nuclear ATG results in production of only mitochondrial Lig3. (C) Immunoblot of Lig1, Lig3, Lig4, and β-actin proteins from whole-cell extracts of WT and Lig3^nuc−/−Lig4^−/− cells; the asterisk indicates a nonspecific band. (D) Immunoblot of Lig3 and Lamin B proteins from nuclear extracts of WT and Lig3^nuc−/−Lig4^−/− cells. (E) Live cell counts of WT, Lig3^nuc−/−, Lig4^−/−, and Lig3^nuc−/−Lig4^−/− cells in unstimulated (-CIT) or stimulated (+CIT) cultures (CIT: anti-CD40, interleukin 4, and TGFβ1). Error bars indicate SE of three independent experiments.

DNA Repair and CSR in Lig3^nuc−/− and Lig3^nuc−/−Lig4^−/− Cells.

Lig3^nuc−/− CH12F3 cells display no hypersensitivity to a variety of DNA-damaging agents (Fig. 4A), including Zeocin, MMS, cisplatin, or hydroxyurea, consistent with previous studies in mouse embryonic stem cells (19). Interestingly, Lig3^nuc−/− and Lig3^nuc−/−Lig4^−/− cells are hypersensitive to a topoisomerase I inhibitor camptothecin (Fig. 4A), which induces SSBs. This is the first observed DNA repair defect associated with Lig3 deficiency after the surprising discovery by Simsek et al. that Lig3 is mostly dispensable for nuclear DNA repair (19). In addition, whereas Lig3 deficiency alone does not cause sensitivity to MMS, Lig3^nuc−/−Lig4^−/− cells are hypersensitive to MMS (Fig. 4A). No CSR defect was observed in Lig3^nuc−/− cells (Fig. 4B). CSR efficiency is reduced in Lig3^nuc−/−Lig4^−/− cells, but to the same level as Lig4^−/− cells (Fig. 4B), indicating that Lig1 alone is sufficient to support A-EJ. Therefore, Lig1 and Lig3 have redundant function in A-EJ during CSR, and expression of either is sufficient to support efficient A-EJ.

Fig. 4.

Drug sensitivity and CSR of Lig3^nuc−/−Lig4^−/− cells. (A) Sensitivity of WT, Lig3^nuc−/−, Lig4^−/−, and Lig3^nuc−/−Lig4^−/− cells to DNA-damaging agents. (B) CSR in WT, Lig3^nuc−/−, Lig4^−/−, and Lig3^nuc−/−Lig4^−/− cells. Error bars indicate SEMs of three independent experiments.

Switch Junctions from Lig1^−/−Lig4^−/− and Lig3^nuc−/−Lig4^−/− Cells.

Switch junctions from Lig1^−/−Lig4^−/− and Lig3^nuc−/−Lig4^−/− cells were PCR-amplified from individual switched cell clones and sequenced. We have shown previously that switch junctions from Lig4^−/− cells display increased microhomology and no direct joins (10). Switch junctions from Lig1^−/−Lig4^−/− or Lig3^nuc−/−Lig4^−/− cells (Fig. S3) are similar to those from Lig4^−/− cells (increased microhomology and no direct joins) (Fig. 5), indicating that either Lig1 or Lig3 is sufficient for microhomology-mediated end-joining during CSR.

Fig. S3.

Sμ-Sα junction sequences. Germ-line Sμ and Sα sequences (gray) are listed on the top and bottom, respectively, of each switch junction. Microhomologies (boxes) are identified as the largest perfect match to both Sμ and Sα germ-line sequences. Nucleotide insertions are underlined. Small vertical lines indicate identity between the junction and germ-line sequences. “∆” indicates a portion of germ-line sequences that have been deleted.

Fig. 5.

Switch Junctions from Lig1^−/−Lig4^−/− and Lig3^nuc−/−Lig4⁻ cells. Number of switch junctions with the indicated nucleotide overlap or insertions. No significant difference was noted between Lig1^−/−Lig4^−/− and Lig3^nuc−/−Lig4^−/− cells based on Mann-Whitney test (P = 0.76).

Discussion

DNA ligases play critical roles in nearly all aspects of DNA metabolism, including DNA replication, repair, and recombination (15). The previous established paradigm delineates Lig1 as the replicative ligase, Lig3 as the SSB repair ligase, and Lig4 as the dedicated DSB repair ligase. However, recent discoveries have revealed an astonishing functional overlap among the DNA ligases and greatly challenged the present ascribed function for each DNA ligase. The finding that Lig3 deficiency has no effect on DNA repair (18, 19) and Lig1 deficiency has no effect on DNA replication (20, 21) was unexpected. Data from this study are consistent with the emerging evidence that Lig1 and Lig3 are functionally redundant in DNA replication and many facets of DNA repair. We were not able to obtain any Lig1^−/−Lig3^−/− cells, possibly because of a synthetic lethality between the two as has been reported in chicken DT40 cells (21). It is interesting to postulate that cells must have either Lig1 or Lig3 to join Okazaki fragments during DNA replication and perhaps to repair spontaneous DNA damages. The catalytic core of each ligase is very similar, although there are substantial differences in the other domains that likely mediate distinct protein–protein interactions and target each ligase to different pathways in normal cells. The striking functional overlap between Lig1 and Lig3 raises an important question: How does a “foreign” ligase cooperate with a set of very different cofactors in a given process? The interaction between XRCC1 and PCNA might help explain the flexibility to use either Lig1 or Lig3 in DNA replication and repair. Lig1 has a PCNA-binding motif at its N terminus that is thought to be important for Lig1's recruitment to the replication fork (16). In the absence of Lig1, Lig3 may access replication forks by an indirect interaction with PCNA bridged by XRCC1 (22). Likewise, it is conceivable that in the absence of Lig3, Lig1 may access DNA damage via an indirect interaction with XRCC1 via PCNA.

Our primary interest in this study is to determine which ligase (Lig1 or Lig3) is responsible for joining switch region breaks in the absence of Lig4. We have shown previously that deletion of XRCC1 in Lig4^−/− cells does not further reduce CSR (23), suggesting that A-EJ of switch region breaks is XRCC1-independent and not a simple form of XRCC1/Lig3-mediated SSB ligation. Now, we have generated a complete panel of viable ligase-deficient cell strains (Lig1^−/−, Lig3^nuc−/−, Lig4^−/−, Lig1^−/−Lig4^−/−, and Lig3^nuc−/−Lig4^−/−). Because neither Lig1 nor Lig3 deficiency further reduces CSR efficiency in Lig4^−/− cells, we argue that Lig1 and Lig3 are functionally redundant in A-EJ. However, it remains unclear whether Lig1 or Lig3 can be incorporated into the NHEJ machinery (albeit as lesser substitutes) for Lig4, as reviewed by Pannunzio et al. (24), or whether these enzymes function in a separate (parallel) pathway. Removing additional NHEJ factors (e.g., XRCC4, XLF, etc.) or expressing a catalytically inactive Lig4 in the double ligase-deficient cells may provide clues to this question.

It is interesting that Lig1^−/−Lig4^−/− and Lig3^nuc−/−Lig4^−/− cells are hypersensitive to MMS even though any single ligase deficiency only displays a mild sensitivity to this DNA-damaging agent. MMS is a DNA base modifier that most frequently methylates guanines at the N7 position (25). This modification is not considered to directly affect DNA replication or transcription or to miscode the mRNA; instead, this modification triggers depurination, which can produce abasic sites (and subsequently SSB) (25). We speculate that when used at a similar level of cell toxicity, MMS may induce more SSBs than other SSB-inducing agents (e.g., UV, free radicals, etc.) because MMS-induced damages are more DNA-specific. An excessive number of SSBs may lead to DSBs (e.g., during DNA replication) that are toxic to NHEJ-deficient cells.

Materials and Methods

Reagents.

Lig1 antibody (18051-1-AP) was purchased from Proteintech Group. Lig1 (sc-20222) and β-actin (sc-47778) antibodies were purchased from Santa Cruz Biotechnology. Lig4 antibody was kindly provided by David Schatz, Yale University, New Haven, CT. Lig3 antibody (611876) was purchased from BD Biosciences. Oligonucleotides were purchased Sigma-Aldrich.

Cell Culture and CSR Assay.

CH12F3 cells were cultured in RPMI 1640 medium supplemented with 10% (vol/vol) FBS and 50 µM β-mercaptoethanol. For CSR assay, cells (viability, >95%) were seeded at 5 × 10⁴ cells/mL in the presence of 1 µg/mL anti-CD40 antibody (16-0402-86; eBioscience), 5 ng/mL of IL-4 (404-ML; R&D Systems), and 0.5 ng/mL TGF-β1 (R&D Systems 240-B) and grown for 72 h. Cells were stained with a FITC-conjugated anti-mouse IgA antibody (BD Biosciences 559354) and analyzed on a LSR II flow cytometer (BD Biosciences). CSR efficiency is determined as the percentage of IgA-positive cells.

Gene Targeting and CRISPR.

Disruption of Lig1 was carried out in Lig4^−/− cells by two rounds of gene targeting with the same targeting vector used in our previous study (20). The resulting Lig1^−/−Lig4^−/− genotype was confirmed by Southern blot analysis as previously described (10, 20). Gene targeting of Lig3 in WT and Lig4^−/− cells was carried out using a targeting vector containing two homology blocks amplified from CH12F3 genomic DNA (Fig. S2). Exons 5–19 were deleted on the targeted allele. Targeted mutation of the initiation codon for the nuclear Lig3 was carried out by CRISPR technology (26). Briefly, 1 µg of hCas9 vector (no. 41815; Addgene) and 1µg of modified gRNA vector (backbone from Addgene no. 41824) were cotransfected into 2 × 10⁶ cells using Amaxa nucleofector (kit V, program K-005). Transfected cells were allowed to recover for 48 h before seeding into single cells per well in a 96-well plate by limited dilution. Cell clones were picked and analyzed by PCR, followed by Sanger sequencing of the PCR products to confirm the insertions/deletions generated by CRISPR.

Drug Sensitivity and MTT Assay.

Cells were seeded at 1 × 10⁵ cells/mL/well in a 24-well plate, and various DNA-damaging agents were added at different concentrations. After 48 h of growth, cell viability was determined by a colorimetric assay that measures the metabolic conversion of thiazolyl blue tetrazolium bromide (MTT) in the mitochondria of living cells, as described previously (10).

Cell Proliferation Analysis.

Cells were seeded at 1 × 10⁵ cells/mL/well in a 24-well plate in regular growth medium or in the presence of 1 µg/mL anti-CD40 antibody (16-0402-86; eBioscience), 5 ng/mL of IL-4 (404-ML; R&D Systems), and 0.5 ng/mL TGF-β1 (240-B; R&D Systems). Viable cells were counted every 24 h.

Switch Junction Analysis.

Individual IgA-positive clones were isolated by limiting dilutions of cytokine-stimulated cultures in 96-well plates. Switch junctions were amplified from individual IgA+ clones with primers KY761 5′-AACTCTCCAGCCACAGTAATGACC-3′ and KY743 5′-GAGCTCGTGGGAGTGTCAGTG-3′, as previously described (10). PCR products were sequenced at Molecular Cloning Laboratories.

Cell Fractionation/Nuclear Isolation.

Ten million cells were washed with 10 mL of ice cold 1× PBS and suspended in 1 mL of ice cold lysis buffer [15 mM KCl, 10 mM Hepes (pH7.6), 2 mM MgCl₂, 0.1 mM EDTA, 1 mM DTT, 0.1% Nonidet P-40). The cell suspension was incubated for 10 min on ice, and nuclei were collected by centrifugation at 700 × g for 5 min at 4 °C.