Activity of DNA ligase IV stimulated by complex formation with XRCC4 protein in mammalian cells

Abstract

Mutation of the XRCC4 gene in mammalian cells^1,2 prevents the formation of the signal and coding joints in the V(D)J recombination reaction³, which is necessary for production of a functional immunoglobulin gene, and renders the cells highly sensitive to ionizing radiation⁴. However, XRCC4 shares no sequence homology with other proteins, nor does it have a biochemical activity to indicate what its function might be². Here we show that DNA ligase IV (ref. 5) co-immunoprecipitates with XRCC4 and that these two proteins specifically interact with one another in a yeast two-hybrid system. Ligation of DNA double-strand breaks in a cell-free system by DNA ligase IV is increased fivefold by purified XRCC4 and seven- to eightfold when XRCC4 is co-expressed with DNA ligase IV. We conclude that the biological consequences of mutating XRCC4 are primarily due to the loss of its stimulatory effect on DNA ligase IV: the function of the XRCC4–DNA ligase IV complex may be to carry out the final steps of V(D)J recombination and joining of DNA ends.

Main

We immunoprecipitated XRCC4 from mammalian cells in order to identify factors that might be associated with it. We used the XRCC4-deficient Chinese hamster ovary cell line XR-1 (ref. 1), which has been stably transfected with an expression vector encoding human XRCC4 fused to a 9× histidine tag and a 3× Myc epitope at the carboxy terminus. The tagged XRCC4 protein was detectable by anti-Myc western blotting as a double band, with the apparent relative molecular mass (M_r) of each band being 60K and 65K (Fig.1). The existence of a double band of this size for XRCC4 suggests that there must be extensive post-translational modification, because the calculated M_r of the tagged XRCC4 protein is 44K. Four different stable XRCC4 transfectants showed full reconstitution to wild-type V(D)J recombination (not shown) and resistance to X-rays (Fig. 1), indicating that human XRCC4 complementary DNA functionally compensates for the deleted hamster XRCC4 gene.

Figure 1: Functional characterization of stable XRCC4 transfectants.

Stable XRCC4–His₉/Myc₃ transfectants were generated in the CHO cell line XR-1, which is deficient for XRCC4 (ref. 2). X-ray sensitivity for four different stable XRCC4 transfectants (designated (TF) 1-1, 1–17, 3-2, 3-23) as well as the wild-type parental cell line of XR-1, 4364A, are shown. Expression of the stably transfected XRCC4 protein assayed by anti-Myc western blotting is shown in the insert.

Immunoprecipitation of each of the four stable XRCC4 transfectants gave a protein band of M_r ∼95K that co-immunoprecipitated with epitope-tagged XRCC4 protein (Fig. 2: XAP is XRCC4-associated protein). This XAP band was not co-immunoprecipitated using XR-1 cell lysates or other His/Myc tagged proteins (such as Rag-1 and -2; data not shown). The XRCC4/XAP protein complex remained stable up to a salt concentration of 1.2 M KCl, indicating that this protein–protein interaction was very stable.

Figure 2: Immunoprecipitation (IP) of epitope-tagged XRCC4.

Preparative anti-Myc IPs from total cell lysates (10⁸ cells per lane) using XR-1 and XRCC4 transfectant cell line 3-2. Immunoprecipitated proteins were visualized using SYPRO orange stain. IPs from XRCC4 transfectant cells show the 60K/65K XRCC4 protein bands and co-immunoprecipitate of one additional band, XAP, migrating at ∼95K. IP controls were carried out with an IgG₁ control antibody (mAb MR5-2, specific for mouse TCR Vβ8). Bands migrating at 50K and 25K–30K are the heavy and light chains of the Myc mAb, respectively.

XAP was isolated from silver-stained gels and microsequenced as described⁶. The sequences of three tryptic peptides for XAP were determined, which unequivocally identified XAP as DNA ligase IV: one octapeptide originated from the amino terminus (amino acids 2–9), one octapeptide had an identical sequence to that between residues 447–454, and one nonapeptide was from the C-terminal tail (amino acids 771–779). Two of these three regions are not well conserved among ligases I, III and IV (ref. 5), and yet all 25 amino-acid residues were identical to the sequence of human DNA ligase IV, suggesting a high degree of homology between rodent and human DNA ligase IV. Single-strand-nick ligase activity was reproducibly detectable in immunoprecipitates from XRCC4-transfected cells but was absent in immunoprecipitates from XR-1 control cells (data not shown). This activity was independent of the addition of exogenous ATP, which is a feature of human DNA ligase IV (ref. 7).

We next tested whether the interaction between XRCC4 and DNA ligase IV could be reproduced in the yeast two-hybrid system. For this and all further experiments, human DNA ligase IV cDNA was cloned from Reh cDNA using the polymerase chain reaction (PCR). DNA ligase IV was found specifically to interact with XRCC4, but not with two selected control proteins (the 70K and 86K subunits of the DNA-end-binding Ku heterodimer⁸; Table 1). This finding independently demonstrates that DNA ligase IV interacts with XRCC4 and that this occurs in a eukaryotic cell nucleus.

Table 1 Interaction of XRCC4 with DNA-ligase IV in yeast two-hybrid system

All known eukaryotic DNA ligases form covalent AMP-enzyme adducts upon addition of ATP⁹. However, in contrast to other DNA ligases DNA ligase IV is pre-adenylated in mammalian cells⁷. To determine whether XRCC4 participates in this pre-adenylation, human DNA ligase IV was expressed as a His/Myc-tagged protein in XR-1 cells using a vaccinia virus overexpression system. Its adenylation in vitro was compared with that of endogenous DNA ligase IV in complex with XRCC4. Both the recombinant and endogenous ligases could be efficiently adenylated only after pretreatment with inorganic pyrophosphate (PP_i), which strips the AMP–ligase adduct (Fig. 3). This indicates that recombinant DNA ligase IV is pre-adenylated as extensively as endogenous DNA ligase IV and that XRCC4 is therefore not needed for pre-adenylation of DNA ligase IV.

Figure 3: Adenylation of recombinant and endogenous DNA ligase IV.

Recombinant, 9× His/3 × Myc-tagged human DNA ligase IV was transiently expressed in XR-1 cells using the vaccinia virus expression system. Endogenous DNA ligase IV was isolated by co-immunoprecipitation from stable XRCC4 transfectants. Both the endogenous and recombinant DNA ligase IVs were adenylated with [α-³²P]ATP, either with or without pyrophosphate pre-treatment (±PP_i, as indicated). In vitro adenylated commercial T4 DNA ligase was used as a positive control.

Purified human DNA ligase IV was previously believed to be able to seal only single-strand DNA breaks and not double-strand breaks⁷. Using an oligonucleotide-based intramolecular ligation assay (Fig. 4a), we tested the activity of the XRCC4–DNA ligase IV complex under different conditions. Nick-ligation activity (on single-strand breaks) was quite insensitive to variations in salt concentration, divalent cations or temperature. In contrast, duplex ligations required specific conditions that were optimal in 100 mM KCl at 37 °C with Mn²⁺ present (data not shown). Under these conditions, linearized plasmids with compatible ends could also be religated efficiently (data not shown). Duplex ligation by XRCC4–DNA ligase IV was detectable at blunt DNA ends, but was much more efficient at sticky ends (Fig. 4a). As the repair of DNA double-strand breaks frequently involves short regions of microhomology (as in V(D)J recombination¹⁰), our duplex ligation assay with compatible ends should mimic non-homologous DNA-end-joining events in vivo.

Figure 4: XRCC4 stimulates DNA ligase IV activity in vitro.

a, DNA oligonucleotide substrates for blunt-end and compatible-end ligations were created by annealing a DNA 80-mer to two short oligomers that were complementary to each end of the long oligomer. One of the short oligomers was 5′-labelled (black dot), resulting in incorporation of the label upon ligation of the two double-stranded arms. Left, analysis by denaturing PAGE of serial 10-fold dilutions of T4 DNA ligase starting at 1U per reaction. Right, ligation with nicked (N), sticky-end (S) or blunt-end (B) substrates catalysed by either 5× 10⁻³ U T4 DNA ligase or DNA ligase IV/XRCC4 co-immunoprecipitated from 10⁷XRCC4 transfectants. Lanes with no protein indicate the position of each substrate. b, Sticky-end ligation catalysed by DNA ligase IV alone (L4-MYC) or by DNA ligase IV in complex with XRCC4 (L4-MYC/X4-HA, and L4-MYC + rX4), analysed by denaturing PAGE. In one experiment, DNA ligase IV was expressed in XR-1 cells either in the absence (lanes 5–7) or presence (lanes 8–10) of the XRCC4 expression vector. Proteins were isolated by anti-Myc IP. Ligations with serial 5-fold dilutions are shown. In a second experiment, DNA ligase IV expressed in the absence of XRCC4 (lane 11) was supplemented with increasing amounts of recombinant XRCC4 protein (lanes 12–15), ending at a molar ratio at 1:1 (lanes 12–15). Control ligations were carried out with anti-Myc IPs from vaccinia-virus-infected (lane 2) or XRCC4–HA-expressing XR-1 cells (lane 3), as well as with identical amounts of recombinant XRCC4 protein alone (lanes 16–19). A possible nonspecific protein effect on DNA ligase IV activity was assayed by addition of BSA (lanes 20, 21). c, Anti-Myc (top/left) and anti-HA western blot (bottom/left) analysis of serial dilutions of immunoprecipitated proteins used in ligation assays, and SYPRO orange staining of the proteins (right). GST-tagged recombinant XRCC4 from insect cells is designated as GST-X4. Note that HA-tagged XRCC4 migrates close to the heavy chain of the antibody used for IP.

Using this system, we compared DNA ligase IV activity on sticky ends in the presence and absence of XRCC4 (Fig. 4b). Co-expression with XRCC4 reproducibly increased the activity of DNA ligase IV 7–8-fold, as determined by phosphorimager analysis (Fig. 4b). This effect is partly due to a consistent slight increase in the steady-state level of DNA ligase IV (∼2-fold; data not shown) that carries over in immunoprecipitation (Fig. 4c), but the difference in activity always exceeded the observed differences in DNA ligase IV levels. We therefore tested whether adding purified recombinant XRCC4 could stimulate DNA ligase IV and found that recombinant XRCC4 at a molar ratio of 1 : 1 consistently gave a 5-fold increase in sticky-end ligation (Fig. 4b). XRCC4 protein itself had no ligase activity, and a nonspecific protein (bovine serum albumin) had no effect on DNA ligase IV.

Our results clarify the role of XRCC4 in DNA metabolism. They indicate XRCC4 is needed to stimulate DNA ligase activity, primarily through direct structure interaction with DNA ligase IV. This interaction distinguishes XRCC4 from the DNA-end-binding autoantigen Ku70/Ku86 (refs 11, 12) and from DNA-PK (ref. 13), a kinase that is mutated in severe combined immunodeficiency^14,15, because these activities appear to be required earlier on for recognition (Ku70/86) and processing of DNA ends (DNA-PK)¹⁶.

A logical extension of this is that DNA ligase IV, in complex with XRCC4, is the joining activity in V(D)J recombination and generally in DNA-end joining. Previous work with mammalian cell lines with deficiencies in DNA ligase I and III indicates that they are entirely normal for V(D)J recombination³,^17,18,19. The DNA ligase II cDNA has not been identified, but it may represent an alternatively spliced form of DNA ligase III or a start-codon variant⁵,²⁰. Thus, DNA ligases I, II and III are unlikely to be the ligases involved in V(D)J recombination and DNA-end joining. Our conclusion that DNA ligase IV is the end-joining activity is supported by results in an accompanying paper²¹, in which the deletion of the homologue of DNA ligase IV in yeast severely impairs non-homologous DNA-end joining. It is likely that a deficiency in either XRCC4 or DNA ligase IV results not only in an increased sensitivity of cells to X-rays, but may also give rise to immunodeficiency in animals.

Methods

Expression vectors. Human XRCC4 cDNA (gift from F. Alt) was cloned into the mammalian expression vector pCDNA3 and fused to a sequence encoding a C-terminal His₉/Myc₃ tag (plasmid pUG14). HA-tagged XRCC4 was generated by replacing the Myc epitope in pUG14 with a linker encoding the HA epitope (plasmid pUG23). Human DNA ligase IV was PCR-cloned from Reh cDNA into pCDNA3 also containing a C-terminal His/Myc tag (plasmid pUG19). The XRCC4-His₉/Myc₃ gene was cloned into baculovirus vector pAcG2T (Pharmingen) from pUG14 in-frame with an N-terminal glutathione-S-transferase (GST) tag (pUG10).

Stable XRCC4 transfectants and XRCC4 baculovirus. Stable puromycin-resistant colonies were generated by co-transfecting plasmid pUG14 with a puromycin-resistance-marker-containing plasmid LXSP. Transfectants were selected at 4 µg ml⁻¹ puromycin. Recombinant XRCC4 baculovirus was generated by co-transfecting pUG10 with Baculo Gold transfection module (Pharmingen), as recommended by the manufacturer.

Immunoprecipitation. Cells were lysed in immunoprecipitation (IP) buffer (25 mM HEPES/KOH, 0.25% Tween-20, 10 mM MgCl₂, 300 mM KCl, pH 7.9, supplemented with 1 mM PMSF, 2 µg ml⁻¹ aprotinin, 2 µg ⁻¹ leupeptin and 1 µg ml⁻¹ pepstatin A). Immunoprecipitates were prepared with mAb MYC 1-9E10.2 (ATCC CRL1729) and protein G-Sepharose and were washed 5 times with 25 mM HEPES/KOH, 0.1% NP40, 10 mM MgCl₂, 650 mM KCl, pH 7.9.

Proteins were fractionated by SDS–PAGE, developed with SYPRO orange stain and analysed on a STORM phosphorimager (Molecular Dynamics). ECL western blots have been described²². Anti-Myc and anti-HA blots were developed with mAb 1-9E10.2 and mAb 3F10 (Boehringer Mannheim), respectively, followed by horseradish-peroxidase-conjugated secondary reagents.

X-ray sensitivity and V(D)J recombination assays. Cells grown to confluency were γ-irradiated and plated into 96-well plates under limiting dilution conditions. The percentage surviving was calculated by dividing the number of colonies at a given dilution of irradiated cells by the number of colonies at the same dilution of non-irradiated cells. Transient cellular V(D)J recombination was assayed as described^23,24. Recombination substrates retaining either coding or signal joints upon V(D)J recombination were pGG51 and pGG49, respectively²⁵.

Protein microsequencing. Microsequencing of tryptic peptides from silver-strained protein bands (500–1,000 ng XAP) was done by nano-electrospray tandem mass spectrometry⁶. From peptide sequences, DNA ligase IV was uniquely identified in a non-redundant database containing more than 200,000 protein sequences without using species or protein size restrictions²⁶.

Ligation assay and in vitro adenylation. Nick ligation and adenylation were done in 60 mM Tris 10 mM MgCl₂, 5 mM DTT, 5 µg ml⁻¹ acetylated BSA, pH 7.9. Ligations were done with 0.1 pmol nick-ligation substrate (a 50-nucleotide bottom strand annealed to a complementary 17-mer and a 5′-phosphorylated 33-mer forming the nicked top strand). Ligations were incubated for 30 min at room temperature, denatured at 95 °C in formamide, and fractionated on denaturing 20% polyacrylamide gels. Adenylation was done with 10 µCi [α-³²P]ATP for 15 min at room temperature. Pyrophosphate pretreatment was for 15 min at RT with 5 mM Na₂P₂O₇.

Duplex ligations were carried out in 50 mM HEPES, 100 mM KCl, 5 mM MnCl₂, 2 mM 2-mercaptoethanol, pH 7.9. Primer sequences for duplex ligation substrates: 80-mer UG301: 5′-GGGTGGATTCGTCCGCTTTCCTTCCCTTT TCTCTCTTTTCCCTTTCTCCTTCCTTCCTTCCCTCCCCTCGTCTAGATCCC-3′, two 15-mers UG302: 5′-CGGACGAATCCACCC-3′ and 5′-[³²P]UG303: 5′-GGGATCTAGACGAGG-3′ (for blunt-end substrate) and two 18-mers UG304: 5′-CGGACGAATCCACCCGGG-3′ and 5′-[³²P]UG305: 5′-ATCTAGACGAGGGGAGGG-3′ (for sticky-end substrate). Duplex ligations were incubated for 1 h at 37 °C.

Expression and purification of recombinant proteins. Recombinant GST-tagged XRCC4 was purified from Hi-5 insect cells infected with baculovirus bvUG7-3. Cells were lysed in IP buffer containing 20 mM imidazole. Proteins were bound to Ni-NTA agarose (Qiagen) and eluted with buffer containing 100 mM EDTA. Supernatants were then incubated with glutathione–Sepharose 4B (Pharmacia) which was eluted with buffer containing 10 mM reduced glutathione. Proteins were dialysed overnight at 4 °C against 50 mM HEPES, 100 mM KCl, pH7.9.

Recombinant Myc-tagged DNA ligase IV was expressed for 24 h in XR-1 cells by liposome-mediated transfection of the appropriate expression vector (10 µg plasmid per 3× 10⁷ cells) and co-infection with vaccinia virus vTF7-3 (ref. 27).

Yeast-two-hybrid system. Protein interactions were analysed using the yeast-two-hybrid system as described^28,29,30. The Ku70 and Ku86 fusion constructs have been reported⁸.