Competitive

Protein Expression and Purification 21, 401–411 (2001) doi:10.1006/prep.2001.1391, available online at http://www.idealibrary.com on

Expression, Purification, and Biophysical Characterization of the BRCT Domain of Human DNA Ligase III␣ Kevin H. Thornton, V. V. Krishnan, Mary G. West, Jennifer Popham, Melissa Ramirez, Michael P. Thelen, and Monique Cosman1 Molecular and Structural Biology Division, Biology and Biotechnology Research Program, Lawrence Livermore National Laboratory, Livermore, California 94551

Received September 18, 2000, and in revised form December 19, 2000

The C-terminal regions of several DNA repair and cell cycle checkpoint proteins are homologous to the breast-cancer-associated BRCA-1 protein C-terminal region. These regions, known as BRCT domains, have been found to mediate important protein–protein interactions. We produced the BRCT domain of DNA ligase III␣ (L3[86]) for biophysical and structural characterization. A glutathione S-transferase (GST) fusion with the L3[86] domain (residues 837–922 of ligase III␣) was expressed in Escherichia coli and purified by glutathione affinity chromatography. The GST fusion protein was removed by thrombin digestion and further purification steps. Using this method, 15N-labeled and 13 C/15N-double-labeled L3[86] proteins were prepared to enable a full determination of structure and dynamics using heteronuclear NMR spectroscopy. To obtain evidence of binding activity to the distal BRCT of the repair protein XRCC1 (X1BRCTb), as well as to provide insight into the interaction between these two BRCT binding partners, the corresponding BRCT heterocomplexes were also prepared and studied. Changes in the secondary structures (amount of helix and sheet components) of the two constituents were not observed upon complex formation. However, the melting temperature of the complex was significantly higher relative to the values obtained for the L3[86] or X1BRCTb proteins alone. This increased thermostability imparted by the interaction between the two BRCT domains may explain why cells require XRCC1 to maintain ligase III␣ activity. 䉷 2001 Academic Press

1 To whom correspondence and reprint requests should be addressed. Fax: (925) 424-3130. E-mail: [email protected].

1046-5928/01 $35.00 Copyright 䉷 2001 by Academic Press All rights of reproduction in any form reserved.

BRCT2 domains (BRCA1 C-terminal repeats) are segments of approximately 100 amino acids in length found within proteins that function in DNA transcription, repair, replication, and cell cycle checkpoints (1–3). This superfamily of domains is structurally similar to the two C-terminal regions of the breast cancer suppresser protein, BRCA1 (4), as deduced from multiple sequence alignments, hydrophobic clustering, and secondary structure predictions. Deletions and mutations within these domains of BRCA1 have been identified as a high risk factor for breast and ovarian cancers (5, 6). BRCT domains have also been found to recruit and bind other BRCT domains or unknown protein folds with high affinity and specificity. Thus, mutations within these domains might deleteriously affect normal cell functions, such as DNA repair, that rely on protein–protein interactions. DNA ligase III and XRCC1 are proteins that participate in the DNA base excision repair pathway, the process that corrects modified DNA bases generated by endogenous cellular metabolic processes and by exposure to environmental alkylating agents or ionizing radiation (7). These proteins have been shown to form a 1:1 complex in vitro that is resistant to 2M NaCl (8–10). In addition, xrcc1 mutant Chinese hamster ovary cell lines contain significantly reduced amounts of both the 2 Abbreviations used: BRCT, BRCA1 C-terminal repeats; GST, glutathione S-transferase; LB, Luria–Bertani; IPTG, isopropylthiogalactopyranoside; DTT, dithiothreitol; S-75, Superdex 75; Ni–NTA, nickel–nitriloacetic acid; ␤ME, ␤-mercaptoethanol; CD, circular dichroism; DSC, differential scanning calorimetry; NMR, nuclear magnetic resonance; HSQC, heteronuclear single quantum correlation; DSS, dimethyl silapentane sulfonate; GuHCl, guanidine hydrochloride; BPTI, basic pancreatic trypsin inhibitor; SEC, size-exclusion chromatography.

401

402

THORNTON ET AL.

XRCC1 (11) and the DNA ligase III proteins, suggesting that the coordinated expression of XRCC1 and ligase III may be required to maintain normal levels of DNA ligase III activity (12, 13). Two different forms of DNA ligase III are produced by the same gene and have identical catalytic properties, but they differ as a result of a tissue-specific alternative splicing mechanism entailing exons coding for the carboxyl terminus of the protein (14–16). The 96kDa (862 amino acids) form (ligase III␤) was shown to occur only in testes during the latter stages of meiotic prophase, whereas the 103-kDa (922 amino acids) form (ligase III␣) is present at similar levels in several somatic tissues (16). The C-terminal 148 amino acids of DNA ligase III␣ are sufficient to bind to the C-terminal 96-amino-acid BRCTb domain of XRCC1, whereas the 96-kDa ligase III␤ that is 60 residues shorter (at the C-terminal end) does not interact with XRCC1 (16, 17). This observation indicates that the C-terminal 60amino-acid residues of ligase III␣ are critical for specifying the interaction with the XRCC1 BRCTb domain. This region of ligase III␣ comprises most of its BRCT domain. XRCC1 contains two BRCT domains, designated here as X1BRCTa (aa 315–403) and X1BRCTb (aa 538–629), that are separated by an acidic, proline-rich region. The proximal X1BRCTa domain interacts with poly(ADPribose)polymerase and negatively regulates its activity following DNA damage (18), while the BRCT domain of DNA ligase III␣, designated here as L3BRCT, interacts with the distal X1BRCTb domain (17). Mutations in the X1BRCTa domain disable the function of XRCC1 in the repair of DNA base damage (11), while alterations in the X1BRCTb domain disrupt interactions with ligase III␣ (19) and abolish the XRCC1-dependent repair of DNA strand breaks during the G1 phase of the cell cycle (20). The X1BRCTb domain isolated as a separate polypeptide retains the ligase III␣-binding function (21), and the crystal structure of this domain (residues 538–629) has been previously determined (http://www.rcsb.org/ pdb/ and pdb access code 1cdz) (22). The protein forms a homodimer containing a four-stranded parallel ␤-sheet surrounded by three ␣-helices (22). Within each monomeric unit, the order of the strands in the sheet is ␤2␤1␤3␤4, with ␣1 and ␣3 on one side of the sheet and ␣2 on the other. Here we describe the overexpression and biophysical characterization of the L3BRCT domain and the L3BRCT:X1BRCTb complex, the first step needed to understand the properties of these domains and their association in the complex of XRCC1 and ligase III␣.

EXPERIMENTAL PROCEDURES Cloning, Overexpression, and Purification of Ligase III␣ BRCT Domains The cloning and purification of the BRCT domain of XRCC1 (X1BRCTb, amino acid residues 533–633) has been previously described (21). The X1BRCTb domain obtained contains a Met residue at the N-terminus, as confirmed by mass spectroscopy analysis (SynPep, Dublin, CA). The calculated pI is 4.77, the molecular weight is 12,061 Da, and the extinction coefficient is 17,200 M⫺1 cm⫺1 at 280 nm (assuming all cysteines are reduced) (http:/www.expasy.ch). To obtain soluble ligase III␣ BRCT domain (L3BRCT), two different constructs were made, one that could be used to express the domain containing residues 843–922 (L3[80]) and another for expressing residues 837–922 (L3[86]). The expression vector for L3[86] was generated by amplifying the region containing nucleotides 2844–3102 in a DNA ligase III␣ pGSTag expression vector generously provided by A. Tomkinson (16). DNA primers (Genosys) used for amplification by PCR were forward, 5⬘-TATAGGATCCGCTGATGAGACGCTGTGCCA-3⬘; and reverse, 5⬘-TATAGAATTCAGCAGGGAGCTACCAGTCT-3⬘. The underlined nucleotides of the forward and reverse primers denote the BamHI and EcoRI sites, respectively, used in cloning the 278-bp product. PCR was performed using Platinum Taq polymerase (Gibco BRL) and the reaction protocol 94⬚C (5 min), followed by 25 cycles of 94⬚C denaturation (30 s), 55⬚C annealing (45 s), and 72⬚C extension (45 s), and was completed by a 72⬚C (7 min) extension. The product was TA cloned using standard techniques into the pCR2.1 vector (Invitrogen). The Escherichia coli strain INV␣F⬘ (Invitrogen) was transformed to ampicillin resistance, plasmid DNA was isolated from the transformants, and the entire sequence of the insert and flanking regions was verified by DNA sequencing. The insert was released by restriction digest with BamHI and EcoRI and ligated into a pGEX-2T vector (Pharmacia) using the BamHI and EcoRI sites in the plasmid. L3[86] was overexpressed as a glutathione S-transferase (GST) fusion protein in E. coli BL21 grown in LB medium, or for nuclear magnetic resonance (NMR) studies, in M9 minimal medium supplemented with 1 g/L 15NH4Cl and 5 g/L [13C]glucose. Typically, an overnight culture was diluted 1:100 into LB or M9 medium with each 2-L flask containing 500 ml of culture. The cultures were grown at 37⬚C with shaking at 300 rpm until A600 ⫽ 0.3–0.4 at which time the heater was turned off and the shaker allowed to cool to shaker temperature (29–30⬚C). Protein expression was induced at A600 ⫽ 0.6 by addition of IPTG to a final concentration of 0.4 mM. The bacteria were harvested the following morning by centrifuging at 4000g. Bacteria were lysed after suspending in 50 mM Tris, pH 8.0, 10

CHARACTERIZATION OF THE BRCT DOMAIN OF LIGASE III␣

mM DTT and passing them through a French press twice at 20,000 psi. After centrifuging the lysate at 14.4g for 20 min, the supernatant was clarified using a Beckman TL ultracentrifuge at 100,000g for 30 min. The supernatant was loaded onto a column containing glutathione–agarose and the column was washed with 5 column vol of 50 mM Tris, pH 8.0, 1 mM DTT. The protein was eluted with 10 mM glutathione, 50 mM Tris, pH 8.0, 1 mM DTT. After quantitation using a BioRad protein microassay, bovine thrombin was added at a ratio of 80 units of thrombin:12 mg of GST-L3[86]. The solution was left at room temperature overnight for complete digestion. Further purification of L3[86] were carried out using a Superdex 75 (S-75) size-exclusion column (2.5 ⫻ 100 cm) attached to a Biologic FPLC (Bio-Rad). The column was equilibrated in 50 mM sodium phosphate, pH 7.3, 150 mM NaCl, 1 mM DTT and elution of the protein was monitored by absorbance at 280 nm. Fractions containing L3[86] were combined and concentrated using an Amicon 8MC microultrafiltration system (YM10 membrane). Following the S-75 step L3[86] was again incubated with glutathione–agarose to remove any remaining GST-tag. The calculated pI of L3[86] is 7.94, the molecular weight is 9,973 Da, and the extinction coefficient is 13,940 M⫺1 cm⫺1 at 280 nm (assuming all Cys residues are reduced) (23). For L3[80], the DNA primers used to amplify the sequence by PCR were forward 5⬘TATACATATGCAAACAAAGGTATTGCTGGA3⬘ and reverse 5⬘AGTGCGGCCGCGCAGGGAGCTACCAGTCT3⬘. The ligase III cDNA sequence corresponding to residues 843–922 was ligated into pET16b (Novagen). L3[80] protein was not soluble when expressed alone in the E. coli host. However, transformation of E. coli BL21(DE3) to both kanamycin and ampicillin resistance with pET29a(X1BRCTb) (described in (21)) and pET16b(L3[80]), followed by induction with IPTG, resulted in the production of a soluble L3[80]:X1BRCTb complex. The complex was expressed using conditions similar to that described for L3[86]. However, neither X1BRCTb nor L3[80] contained an N-terminal GST or other fusion partner, but L3[80] contained a C-terminal His-tag (including residues AAALEHHHHHH). The complex was purified using the L3[80] His-tag on a Ni–NTA agarose column (Qiagen). To keep the cysteines reduced, 0.1% ␤ME was used instead of DTT prior to size-exclusion chromatography. The resulting L3[80] protein has a calculated pI of 8.77, molecular weight of 10,606 Da, and an extinction coefficient of 13,940 M⫺1 cm⫺1 at 280 nm. For the L3[80]:X1BRCTb complex, the calculated pI is 6.12, the molecular weight is 22,667 Da, and the extinction coefficient is 31,720 M⫺1 cm⫺1 at 280 nm. The L3[80]:X1BRCTb complex is soluble up to 10 mg/ ml but is most stable at lower concentrations for long term storage.

403

SDS–Polyacrylamide Gel Electrophoresis Protein concentrations were determined using the Bio-Rad microassay protocol or by measuring the absorbance at 280 nm. All SDS–PAGE protein samples were prepared using a 2⫻ Tris–glycine sample buffer (Novex) with added DTT. The samples were heated for 2–3 minutes at 90⬚C prior to electrophoresis on a 4–20% polyacrylamide Tris–glycine gel (Novex) and stained with Gelcode blue stain (Pierce). Purity was assessed by analyzing the stained polyacrylamide gel photograph image, using 1D image analysis software (Eastman Kodak Co.) to capture the image and calculate band intensities. Size-Exclusion Chromatography To estimate the molecular masses, each protein/complex sample was mixed with the following standards and run on the S-75 column: aprotinin (6500 g/M), cytochrome C (12,400 g/M), carbonic anhydrase (29,000 g/ M), and bovine albumin (66,000 g/M) (Sigma). The column buffer used was 50 mM NaH2PO4, 150 mM NaCl, pH 7.3, or 50 mM KH2PO4, 500 mM KCl, pH 7.3. The void volume (Vo) for the column was 187.5 ml as determined by the elution of blue dextran (MW ⫽ 2,000,000 Da). The molecular weights of the proteins were determined from a plot of Ve /Vo ⫽ log(MW) of the standards, where Ve is the elution volume of the protein and Vo is the void volume. The resulting correlation coefficient (R2) for the standards was 0.998. Circular Dichroism (CD) Spectra were acquired on a Jasco 715 CD spectropolarimeter using a 0.1-cm cell at 25⬚C unless indicated otherwise. The system was purged with nitrogen at 50 L/min during data collection. Data acquisition parameters were as follows: sensitivity, 50 mdeg; resolution, 0.5 nm; bandwidth, 1 nm; response, 8 s; scan speed, 20 nm/min; and number of accumulations, 4. Background spectra were acquired using the same buffer as used for the protein samples. After background subtraction, the spectra were analyzed by a backpropagation neural network algorithm using the program Circular Dichroism Deconvolution (24) (http://bioinformatik. biochemtech.uni-halle.de/cdnn/). Differential Scanning Calorimetry (DSC) DSC measurements were performed on a Nano IIdifferential scanning calorimeter (Calorimetry Sciences Corp.) using a 0.299-ml capillary cell at three atmospheres pressure. Prior to data acquisition, the proteins, in 100 mM DTT, were dialyzed (8000 MWCO dispodialyzer, Spectrum Laboratories) three times (1:500) against degassed 50 mM NaH2PO4, 150 mM

404

THORNTON ET AL.

NaCl, pH 7.3, at 4⬚C. The dialysate was used for all conditioning and baseline runs as well as in the reference cell. Protein concentrations were 1 mg/ml for all runs. Data analysis was performed using the program CPcalc (Calorimetry Sciences Corp.). NMR One dimensional proton and gradient and sensitivity enhanced 15N-1H HSQC (heteronuclear single quantum correlation) (25) spectra of 0.6 mM L3[86] dissolved in 50 mM NaH2PO4, 150 mM NaCl, 1 mM DTT, pH 6.7, and 5% D2O at 15⬚C were obtained using a 600-MHz Varian INOVA NMR spectrometer. 1H and 15N axes were referenced with respect to the dimethyl silapentane sulfonate (DSS) signal at 15⬚C and indirectly to DSS using the ratio of ␥ s (H␥/N␥), respectively (26). The one dimensional experiment was obtained after presaturation of the water resonance (1.5 s with radio frequency strength of 95 Hz), with an acquisition time of 0.65 s over a spectral width of 12.5 kHz. In the 1H15 N HSQC experiment, the spectral widths were 7200 and 1775 Hz for 1H and 15N dimensions, respectively. The acquisition times along the t2 and t1 dimensions were 71 and 36 ms, respectively, for each complex point, while States-TPPI was used for quadrature detection in the indirect dimension. Coherence selection was achieved by Rance–Kay-type sensitivity enhancement (25) and the recycling delay between the scans was 1.5 s while the signal was averaged over 64 transients. 15N hard pulses were applied at a field strength of 10.2 kHz, while the decoupling was achieved by the multipulse sequence GARP1 (27) at 1.5 kHz. Spectra were processed and analyzed using the program FELIX (MSI, Inc.) and NMRPIPE (28). Self-Diffusion Coefficient Measurements Self-diffusion coefficient measurements of L3[86] dissolved in 50 mM NaH2PO4,150 mM NaCl, 1 mM DTT, pH 6.7, and 5% D2O at 15⬚C were obtained using a Varian INOVA 600 MHz NMR spectrometer and the bipolar pulsed-field gradient-selective echo dephasing sequence (2931). The experimental parameters were as follows: acquisition time, 0.328 s; spectral width, 12,500 Hz; signal averaging, 256 scans; recycling delay, 3 s; and water-selective pulse, 4 ms. Gradients were varied from 1 to 32 Gcm⫺1 in units of 1.0 Gcm⫺1, while the other gradients were applied at a strength of 30 Gcm⫺1 for 1 ms each, yielding a total echo time (␶1 ⫹ ␶2) of 14.026 ms. Phase cycling was used to advantageously utilize the radiation damping effects for water suppression as previously reported (32). Time domain self-diffusion coefficient data were zero filled once and a cosine bell apodization applied prior to complex Fourier transformation. The area under each spectrum from 5 to ⫺1 ppm was integrated and a nonlinear least square fit of

equation (S(q) ⫽ S(0)exp(⫺Dsq2(⌬ ⫺ ␦/3 ⫺ ␶/2)) was used to estimate the self-diffusion coefficients (30). S(q) is the measured integral value as a function of q and S(0) is the value at q ⫽ 0. q is the effective area of the gradient pulse given by (␥gz␦), where ␥ is the gyromagnetic ratio of a proton (2.6752 ⫻ 108 s⫺1 T⫺1) and gz and ␥ are the amplitude and duration of the gradient pulse, respectively. Ds is represented in units of m2 s⫺1, while ⌬ and ␶ are delays in seconds employed in the pulse sequence. RESULTS AND DISCUSSION Expression and Purification of the Ligase III␣ BRCT Domain, L3[86] To characterize the BRCT domain of ligase III␣ (L3BRCT) and its interaction with the distal BRCT domain of XRCC1 (X1BRCTb), these proteins were overexpressed in bacteria and purified. This allowed us to study the biophysical properties of isolated L3BRCT and determine how they change upon binding with X1BRCTb. We observed that in the absence of a fusion partner during expression, the L3[80] protein (residues 843–922 of ligase III␣) was insoluble. The slightly longer L3[86] protein (residues 837–922), on the other hand, was soluble when expressed as a GST fusion protein. After thrombin cleavage, the L3[86] protein has two additional nonnative residues (Gly-Ser) at the N-terminus. The purification protocol, consisting of glutathione affinity and size exclusion chromatography steps, yielded 14 mg of highly homogeneous L3[86] from 1 liter of bacterial culture. The protein is soluble up to 24 mg/ml at pH 7.3 in our preparations, and the final L3[86] fraction was 98% pure as judged by densitometric evaluation of proteins separated by SDS–gel electrophoresis (Table 1 and Fig. 1). For NMR structure and dynamic studies, the 15Nand 15N,13C-labeled L3[86] proteins were prepared by conducting the expression in M9 minimal medium supplemented with 1 g/L 15NH4Cl and either 5 g/L unlabeled glucose or 4 g/L [13C]glucose. In general, the expression yields in M9 medium were approximately half that obtained in LB medium. TABLE 1 Purification of Overexpressed L3[86] Step Lysate (soluble) GST-affinity (before thrombin digestion) Superdex 75 purified L3(86)

Protein (mg) 558 83.8 14.1

Yield (%)

Purification (fold)

13.2

1

89 98

6.7 7.4

Note. Yield and purification were determined by measuring the quantity of L3[86] in an SDS–polyacrylamide gel using densitometric scanning, compared to the total protein determined in each corresponding sample.

CHARACTERIZATION OF THE BRCT DOMAIN OF LIGASE III␣

405

Estimation of L3[86] Protein Mass Using folded globular proteins as standards, data from size-exclusion chromatography indicate that the mass of L3[86] is approximately 21,300 Da (Fig. 2). This result is in agreement with the electrospray mass spectrometry analysis, in which the protein was observed to have a mass of 19,983 Da. Two other components were also observed by mass spectrometry, corresponding to masses of 19,962 and 20,003. These minor components are likely to be L3[86] dimers with different numbers of bound counterions, such as sodium, or tightly associated water molecules. Therefore, both the size-exclusion chromatography and the mass spectrometry results indicate that L3[86] exists as a dimer. Homology with the BRCT Superfamily

FIG. 1. Purification of L3[86]. Gelcode blue-stained SDS– polyacrylamide gel of L3(86) purification fractions. Protein corresponding to 100 ␮l of the original bacterial culture was loaded in each lane: Lane 1, lysate (56 ␮g); lane 2, fusion protein from GST-affinity column prior to thrombin digestion (8.4 ␮g); lane 3, thrombin-digested fusion protein (8.4 ␮g); lane 4, Superdex 75-purified L3(86) (1.4 ␮g).

The amino acid sequences for X1BRCTb (residues 533–633 of XRCC1) and for L3[86] (residues 837–922 of ligase III␣) are aligned in Fig. 3. The X1BRCTb secondary structural elements identified from the crystal structure of residues 538–633 (22) are designated above the amino acid sequence in Fig. 3. This sequence alignment was obtained by matching amino acid identities and conserved substitutions between L3[86] and those in the structured segments of X1BRCTb to obtain the greatest agreement. Thus this alignment differs slightly from the multiple alignment reported in Zhang et al. (22) in that it contains two single amino gaps in L3 corresponding to the ␣1 region of X1BRCT. In Zhang et al., these two gaps are located in the ␤1– ␣1 loop,

FIG. 2. Mass estimation of L3[86] Superdex 75 size-exclusion chromatography of 3 mg/ml of L3[86] (dark line) shows that the MW of the protein is 21.3 kDa. The calculated molecular weights of L3[86] are 9973 Da and 19,946 Da for a monomer and dimer, respectively. The elution profiles (dashed lines) and the MWs for several standards are also shown.

406

THORNTON ET AL.

FIG. 3. Sequence alignment of X1BRCTb and L3[86]. Consensus amino acid residues between L3[86] and its binding partner X1BRCTb are shown in bold underlined (identities) and italic underlined (conserved substitutions). Gaps in the alignment between the sequences are designated by dashes. The secondary structure elements identified from the crystal structure of X1BRCTb (22) are shown above the sequence of X1BRCTb.

which is a highly variable region among the BRCT family. Positioning the two gaps within the ␣1-helix yields 100% alignment between all the conserved residues (bold underlined, Fig. 3) and suggests that the homologous ␣1-helix will be two residues shorter in the structure of L3[86] compared to that in the X1BRCTb structure. The L3[86] protein used in our studies is 9 amino acid residues longer at the N-terminal end and eight residues shorter at the C-terminal end compared to the X1BRCTb protein used in the crystallographic studies. The alignment also results in a single amino acid gap in the X1BRCTb ␤2– ␤3 loop, which is another highly variable region, and the same large 10- amino-acid gap in L3[86] as previously reported (22) that encompasses the entire X1BRCTb ␣2-helix. The ␣2-helix in X1BRCTb is the least conserved and the most variable among the BRCT family members (4). Based on our alignment (Fig. 3), there is approximately 43% sequence homology between the X1BRCTb(538–590) and L3[86](846–897) segments and 68% between the X1BRCTb(601–625) and L3[86](898–922) segments. The overall sequence homology (number of underlined residues divided by the 75 residues in the most highly conserved regions) between these X1BRCTb and L3[86] segments is ⬃51%. It is noteworthy that while X1BRCTb contains one cysteine residue at position 615, L3[86] contains 3 (C842, C912, and C922) (boxed residues, Fig. 3). The highly conserved cysteine (C615 in X1BRCTb and C912 in L3[86], Fig. 3) is buried in the crystal structure of X1BRCTb (22) and also in the solution structure of L3[86] (manuscript in preparation); thus it is not available to participate in disulfide bond formation. The remaining two L3[86] cysteines would not be expected to

form intramolecular disulfide bonds in the folded protein based on structural homology to the X1BRCTb structure, in which the analogous positions of the L3[86] C842 and C922 residues would place them in the N- and C-terminal ends, respectively, of the X1BRCTb ˚ apart from each molecule and thus more than 40 A other (22). However, in the absence of reducing agent, the L3[86] C842 and C922 residues might participate in the formation of intermolecular disulfide bonds, but this would be expected to produce a very complex mixture. For example, for nonspecific homodimer formation, three possible configurations are C842A to C842B, C842A to C922B, and C922A to C922B, where A and B designate different monomeric L3[86] molecules. In addition, higher order aggregates would also be able to form as only one of these two cysteines would participate in dimerization, while the second cysteine could interact with another monomer, dimer or higher order aggregate to form an additional disulfide bond. Since this could lead to undesirable precipitation of the protein, we found it was important to keep reducing agent present during purification and data collection. In the presence of 1 mM DTT, we saw no evidence of a mixture of higher ordered aggregates, and this has been confirmed by the NMR structure of the protein. Additional NMR data also show that these cysteines do not participate in forming intermolecular disulfide bonds. It has been reported that the 13C NMR chemical shift of each cysteine’s C␤ carbon is extremely sensitive to its redox state (33). In reduced cysteine, the observed C␤ shifts fall within 25.4 to 31.7 ppm, whereas for oxidized cysteine they are located within 35.2 to 47.2 ppm (33). In a very few cases, C␤ shifts for oxidized and reduced states can also overlap in a narrow range of 33.0 to 34.0

CHARACTERIZATION OF THE BRCT DOMAIN OF LIGASE III␣

ppm. However, under the reducing conditions used for our NMR studies, all the cysteines in L3[86] clearly exhibit C␤ chemical shifts (C842, 27.4 ppm; C912, 27.5 ppm; and C922, 29.0 ppm) well within the range (25.4 to 31.7 ppm) expected for the reduced state. Secondary Structure Analyses of L3[86] The far UV CD spectrum of L3[86] (Fig. 4A, top) provides information about the secondary structure of this BRCT domain. The percentages of helix, sheet, and coil present were determined by fitting the far UV CD spectra (190–250 nm) to a theoretical model (24). Using this method, L3[86] is predicted to have 20% ␣-helix and 25% ␤-sheet. The X-ray structure of X1BRCTb has 27.08% helix and 13.54% sheet (22). The smaller amount of helical secondary structure in L3[86] is not surprising, considering that 10 residues comprising ␣2 that are present in X1BRCTb would most likely not be present in the structure of L3BRCT (Fig. 3). The near UV CD spectrum (250–310 nm, Fig. 4A, bottom) of L3[86] also provides information about whether the aromatic residues are held rigidly in an asymmetric environment, as would be the case in the hydrophobic core of a structured protein, and whether disulfide bonds are present in the protein. However, the contribution of the disulfide chromophore is difficult to distinguish and is often buried under that of the aromatic phenylalanine, tyrosine, and tryptophan residues. In addition, the cysteine residues in the protein

407

were kept in a reduced state using DTT prior to acquiring the CD spectra and the sample was continuous purged with nitrogen gas during acquisition. Therefore, any contribution present in the spectrum from disulfide bonds would be minimal. L3[86] contains five phenylalanines (F851, F865, F872, F875, F883), two tyrosines (Y857, Y871), and two tryptophans (W908, W910) (Fig 3). The presence of signals for these aromatic residues in the near UV CD spectra of L3[86] confirm that significant structured regions are present, which is a prerequisite for tertiary structure determination by either NMR spectroscopy or X-ray diffraction studies. L3[86] Denaturation Experiments DSC experiments were carried out on L3[86] to determine its thermal stability (Fig. 4B). Thermal denaturation was observed to be irreversible, with L3[86] precipitating out of solution. In spite of the difficulties this presented in the DSC analysis, the data indicate that L3[86] undergoes a two-state single unfolding transition with a melting temperature at 49 ⫾ 0.1⬚C. The DSC data confirm the CD secondary structure analyses that L3[86] has a structured core. A control experiment was also performed to determine the percent of unfolded and folded protein by titrating the protein with different concentrations of the chemical denaturant guanidine hydrochloride (GuHCl) (Fig. 4C). In agreement with CD and DSC, L3[86] exhibits an unfolding transition at 2 M GuHCl, indicative of a protein that has tertiary structure in solution.

FIG. 4. (A) CD, DSC, and GuHCl denaturation of L3[86]. The far (190–250 nm, top) and near (250–310 nm, bottom) UV CD spectra, (B) differential scanning calorimetry data, and (C) GuHCl denaturation curve of L3(86) monitored by far UV CD. DSC experiments were acquired at 1.0 mg/ml purified L3(86) in 50 mM NaH2PO4, 150 mM NaCl, pH 7.3. Far and near UV CD experiments were acquired at 0.1 and 1.0 mg/ml, respectively, in 10 mM NaH2PO4, pH 7.0.

408

THORNTON ET AL.

Evidence for L3[86] Purity Determined by NMR Both one-dimensional 1H and two-dimensional 1H– N-HSQC NMR spectra of L3[86] exhibit good signal dispersion as expected for homogeneous, structured proteins (Fig. 5). The spectra show reasonably sharp lines, indicating that L3[86] is a good candidate for more detailed structural and dynamic studies, which will be reported elsewhere. 15N and 15N,13C-labeled protein was used to assign all the proton, nitrogen (as shown for the backbone amide protons and nitrogens in Fig. 5), and carbon chemical shifts in the NMR data. 15

Homodimer Formation of L3[86] Supported by NMR Diffusion Coefficient Measurements Since BRCT domains appear to participate in both homodimer and heterodimer association interactions, self-diffusion coefficient (Ds) measurements were undertaken to further characterize the oligomeric state of L3[86] under NMR sample conditions (Fig. 6). The Ds value of a protein, which measures the rate at which it diffuses through aqueous solutions, is dependent on the size and shape of the protein. The Ds of L3[86] was determined to be 11.2 ⫾ 0.3 ⫻ 10⫺11 m2 s⫺1 (Fig. 6), and this value was compared to that obtained for lysozyme

FIG. 5. NMR spectra. (A) One-dimensional 1H NMR spectrum of the unlabeled L3[86] and (B) 1H– 15N gradient sensitivity enhanced HSQC (25) of 15N-labeled L3[86].

CHARACTERIZATION OF THE BRCT DOMAIN OF LIGASE III␣

FIG. 6. Self-diffusion coefficient measurements plots of the selfdiffusion coefficient measurements of L3[86] (circles), lysozyme (squares), and basic pancreatic trypsin inhibitor (BPTI, diamonds). The continuous lines correspond to the fit to the diffusion data. The error bars were obtained using duplicate experimental measurements.

(0.5 mM, T ⫽ 25⬚C, Ds ⫽ 13.4 ⫾ 0.2 ⫻ 10⫺11 m2 s⫺1) and basic pancreatic trypsin inhibitor (BPTI, concentration 0.5 mM, T ⫽ 25⬚C, Ds ⫽ 19.6 ⫾ 0.2 ⫻ 10⫺11 m2 s⫺1) (34)). The results show that L3[86], a 10-kDa protein, diffuses through solution at a slower rate than either lysozyme (14.2 kDa) or BPTI (6.59 kDa), proteins that are known to exist as monomers. Thus, the effective mass of L3[86] is greater than that expected for a monomer, and the measured Ds value of 11.2 ⫻ 10⫺11 m2 s⫺1 for L3[86] is within a range expected for a L3[86]2 homodimer, in agreement with the size-exclusion chromatography and mass spectrometry data. Evidence for Binding between the BRCT Domains of Ligase III␣ and XRCC1 We also examined how the biophysical characteristics of the ligase III␣ BRCT domain changed in the presence of its binding partner, X1BRCTb. Two different L3BRCT:X1BRCTb complexes were compared: the complex formed by mixing L3[86] and X1BRCTb in a 1:1 ratio and a coexpressed and purified L3[80]:X1BRCTb complex. One advantage of preparing the L3[80]:X1BRCTb complex by coexpression is that the dynamics of complex formation may more closely resemble that present in vivo. Complexes formed by direct mixing of the individual protein components, on the other hand, may be influenced by the presence of preexisting multimers that formed during the isolation, purification or storage of the two BRCT domains. Also, the L3[80] protein in the L3[80]:X1BRCTb complex is several amino acid residues shorter at the N-terminus than L3[86] (see Fig. 3). This allows us to determine how

409

subtle changes in the protein sequence, such as differences in the polypeptide chain length, affect the secondary structure and thermal stability of the L3BRCT:X1BRCTb complex. The oligomeric states of the L3[80]X1BRCTb and L3[86] ⫹ X1BRCTb heterocomplexes were assessed at concentrations used for our NMR studies by sizeexclusion chromatography (Figs. 7A and 7B). The L3[80]X1BRCTb coexpressed complex (9 mg/ml) elutes as a single chromatographic peak with a molecular weight of 39.6 kDa, suggesting that it is a tetramer (calculated MW is 45.3 kDa) (Fig. 7A). The L3[86] ⫹ X1BRCTb complex (3 mg/ml) exhibits a major elution peak at 36.5 kDa and a broad shoulder at approximately 300 ml corresponding to about a 19.5-kDa molecular weight species (Fig. 7B). While the 36.5-kDa peak corresponds most closely to a tetramer (calculated MW is 44.1 kD), the small 19.5-kDa peak suggests the presence of excess amounts of either L3[86]2 (calculated MW is 19.9 kDa) or X1BRCT2 (calculated MW is 24.1 kDa). The SEC results confirm that both constructs of the BRCT domain of ligase III␣ that we have produced exhibit similar binding activity to the distal BRCT domain of XRCC1. The CD spectrum obtained for the L3[80]:X1BRCTb complex was compared with that generated by adding the CD spectra of L3[86] with X1BRCTb (Fig. 7C). Despite the differences in sequences of the two forms of L3BRCT, the CD spectra indicate that they adopt nearly identical secondary structures. The fit to the CD spectrum of L3[80]:X1BRCTb indicates that the complex contains 32% ␣-helix and 17% ␤-sheet. Based on the CD spectra of the individual X1BRCTb and L3[86] domains, we would predict the L3[86] ⫹ X1BRCTb heterocomplex to contain approximately 27% ␣-helix and 16% ␤-sheet. These values fall well within the error limits of the calculated secondary structure values for the L3[80]:X1BRCTb complex and indicate that no significant changes in the secondary structures of X1BRCTb and L3BRCT occur upon formation of the complex. Single melting transitions are observed in the differential scanning calorimetry scans of the complexes. The melting transition for the L3[86] ⫹ X1BRCTb is 63.7⬚C, 3.5⬚C higher than the value, 60.2⬚C, obtained for the coexpressed L3[80]:X1BRCTb complex (Fig. 7D). One possible explanation for this variance is that the subtle differences between L3[86] and L3[80] in one or more N-terminal residues may stabilize interactions with X1BRCTb. The presence of the L3[80] His-tag might also destabilize interactions with X1BRCTb. The melting transitions for L3[86] and X1BRCTb occur at 48.8 ⫾ 0.1⬚C (Fig. 4) and 56.1 ⫾ 0.1⬚C (data not shown), respectively. Thus, both L3[80]:X1BRCTb and L3[86] ⫹ X1BRCTb complexes exhibit a significant 4 to 15⬚C increase in thermal stability over that of the individual

410

THORNTON ET AL.

FIG. 7. Complexes. (A) Size-exclusion chromatography (SEC) of L3[80]:X1BRCTb at 9 mg/ml, 39.6 kDa; and (B) L3[86] ⫹ X1BRCT 3 mg/ ml, 36.5 kDa. The arrow indicates the presence of a minor component with a molecular weight of approximately 19.5 kDa. (C) CD spectrum of 0.1 mg/ml L3[80]X1BRCTb (solid line) compared to that obtained from adding the CD spectrum of 0.1 mg/ml L3[86] and 0.1 mg/ml X1BRCTb and dividing by 2 (dashed line). (D) DSC spectrum of L3[80]:X1BRCTb (solid line) compared to that of a 1:1 mixture of L3[86] and X1BRCTb (dashed line).

components L3[86] and X1BRCTb. This result provides direct experimental evidence that L3BRCT and X1BRCTb interact with one another to form a stable complex. CONCLUSION Using BRCT domains produced by overexpression in E. coli, we have demonstrated that the ligase III␣ BRCT domain, L3[86], is autonomously folded and that it consists of both helical and ␤-sheet segments. Size-exclusion chromatography and NMR diffusion coefficient studies show that L3[86] exists as a homodimer in solution. We have prepared 15N- and 15N,13C-labeled L3[86] protein and the NMR data acquired are of sufficient quality to determine the structure and dynamics of the protein in the solution state. Circular dichroism studies show that neither L3[86] nor X1BRCTb undergo a significant change in secondary structure upon binding to one another. However, calorimetry studies show that L3[86] is stabilized by binding to the X1BRCTb domain, as evidenced by the 15⬚C increase in the melting transition of the complex relative to that of L3[86] alone. The increased thermostability of the heterocomplex may explain, in part, why the presence of XRCC1 is critical in maintaining ligase III␣ levels and activity in vivo.

ACKNOWLEDGMENTS The authors are grateful to Dr. Alan Tomkinson for the GSTagligase III␣ expression vector and to Dr. Rod Balhorn for critical reading and comments on the manuscript. This work was performed under the auspices of the U.S. Department of Energy by the Lawrence Livermore National Laboratory under Contract W-7405-ENG-48, supported by the Lawrence Livermore National Laboratory LDRD Director’s Initiative Grant 99-LW-031 to M.C., and partly funded by NIH Grant CA76116 to M.P.T. REFERENCES 1. Bork, P., Hofmann, K., Bucher, P., Neuwald, A. F., Altschul, S. F., and Koonin, E. V. (1997) A superfamily of conserved domains in DNA damage responsive cell cycle checkpoint proteins. FASEB J. 11, 68–76. 2. Callebaut, I., and Mornon, J. P. (1997) From BRCA1 to RAP1: A widespread BRCT module closely associated with DNA repair. FEBS Lett. 400, 25–30. 3. Huyton, T., Bates, P. A., Zhang, X., Sternberg, M. J. R., and Freemont, P. S. (2000) The BRCA1 C-terminal domain: Structure and function. Mutat. Res. 460, 319–332. 4. Koonin, E. V., Altschul, S. F., and Bork, P. (1996) Functional motifs. Nature Genet. 13, 266–268. 5. Futreal, P. A., Liu, Q. Y., Shattuckeidens, D., Cochran, C., Harshman, K., Tavtigian, S., Bennett, L. M., Haugenstrano, A., Swensen, J., Miki, Y., Eddington, K., McClure, M., Frye, C., Weaverfeldhaus, J., Ding, W., et al. (1994) BRCA1 mutations in primary breast and ovarian carcinomas. Science 266, 120–122. 6. Friedman, L. S., Ostermeyer, E. A., Szabo, C. I., Dowd, P., Lynch,

CHARACTERIZATION OF THE BRCT DOMAIN OF LIGASE III␣

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

E. D., Rowell, S. E., and King, M. C. (1994) Confirmation of BRCA1 analysis of germline mutations linked to breast and ovarian cancer in ten families. Nature Genet. 8, 399–404. Parikh, S. S., Mol, C. D., and Tainer, J. A. (1997) Base excision repair enzyme family portrait: Integrating the structure and chemistry of an entire DNA repair pathway. Structure 5, 1543– 1550. Caldecott, K. W., McKeown, C. K., Tucker, J. D., Ljungquist, S., and Thompson, L. H. (1994) An interaction between the mammalian DNA repair protein XRCC1 and DNA ligase-III. Mol. Cell. Biol. 14, 68–76. Caldecott, K. W., Tucker, J. D., Stanker, L. H., and Thompson, L. H. (1995) Characterization of the XRCC1–DNA ligase III complex in vitro and its absence from mutant hamster cells. Nucleic Acids Res 23, 4836–4843. Ljungquist, S., Kenne, K., Olsson, L., and Sandstrom, M. (1994) Altered DNA ligase III activity in the CHO EM9 mutant. Mutat. Res. 314, 177–186. Shen, M. R., Zdzienicka, M. Z., Mohrenweiser, H., Thompson, L. H., and Thelen, M. P. (1998) Mutations in hamster single-strand break repair gene XRCC1 causing defective DNA repair. Nucleic Acids Res. 26, 1032–1037. Cappelli, E., Taylor, R., Cevasco, M., Abbondandolo, A., Caldecott, K., and Frosina, G. (1997) Involvement of XRCC1 and DNA ligase III gene products in DNA base excision repair. J. Biol. Chem. 272, 23970–23975. Wilson, S. H., and Singhal, R. K. (1998) Mammalian DNA repair and the cellular DNA polymerases in “DNA Damage and Repair: DNA Repair in Higher Eukaryotes,” pp. 161–180, Humana Press, Totowa, NJ. Chen, J., Tomkinson, A. E., Ramos, W., Mackey, Z. B., Danehower, S., Walter, C. A., Schultz, R. A., Besterman, J. M., and Husain, I. (1995) Mammalian DNA ligase III: Molecular cloning, chromosomal localization, and expression in spermatocytes undergoing meiotic recombination. Mol. Cell. Biol. 15, 5412–5422. Wei, Y. F., Robins, P., Carter, K., Caldecott, K., Pappin, D. J. C., Yu, G. L., Wang, R. P., Shell, B. K., Nash, R. A., Schar, P., Barnes, D. E., Haseltine, W. A., and Lindahl, T. (1995) Molecular cloning and expression of human cDNAs encoding a novel DNA ligase IV and DNA ligase III, an enzyme active in DNA repair and recombination. Mol. Cell. Biol. 15, 3206–3216. Mackey, Z. B., Ramos, W., Levin, D. S., Walter, C. A., McCarrey, J. R., and Tomkinson, A. E. (1997) An alternative splicing event which occurs in mouse pachytene spermatocytes generates a form of DNA ligase III with distinct biochemical properties that may function in meiotic recombination. Mol. Cell. Biol. 17, 989–998. Nash, R. A., Caldecott, K. W., Barnes, D. E., and Lindahl, T. (1997) XRCC1 protein interacts with one of two distinct forms of DNA ligase III. Biochemistry 36, 5207–5211. Masson, M., Niedergang, C., Schreiber, V., Muller, S., MenissierdeMurcia, J., and deMurcia, G. (1998) XRCC1 is specifically associated with poly(ADP-ribose) polymerase and negatively regulates its activity following DNA damage. Mol. Cell. Biol. 18, 3563– 3571.

411

19. Taylor, R. M., Wickstead, B., Cronin, S., and Caldecott, K. W. (1998) Role of a BRCT domain in the interaction of DNA ligase III-a with the DNA repair protein XRCC1. Curr. Biol. 8, 877–880. 20. Taylor, R. M., Moore, D. J., Whitehouse, J., Johnson, P., and Caldecott, K. W. (2000) A cell cycle-specific requirement for the XRCC1 BRCT II domain during mammalian DNA strand break repair. Mol. Cell. Biol. 20, 735–740. 21. Thornton, K., Forstner, M., Shen, M. R., West, M. G., Rupp, B., and Thelen, M. P. (1999) Purification, characterization, and crystallization of the distal BRCT domain of the human XRCC1 DNA repair protein. Protein Express. Purif. 16, 236–242. 22. Zhang, X. D., Morera, S., Bates, P. A., Whitehead, P. C., Coffer, A. I., Hainbucher, K., Nash, R. A., Sternberg, M. J. E., Lindahl, T., and Freemont, P. S. (1998) Structure of an XRCC1 BRCT domain: A new protein–protein interaction module. EMBO J. 17, 6404–6411. 23. Pace, C. N., Vajdos, F., Fee, L., Grimsley, G., and Gray, T. (1995) How to measure and predict the molar absorption coefficient of a protein. Protein Sci. 4, 2411–2423. 24. Bohm, G., Muhr, R., and Jaenicke, R. (1992) Quantitative analysis of protein far UV circular dichroism spectra by neural networks. Protein Eng. 5, 191–195. 25. Kay, L. E., Keifer, P., and Saarinen, T. (1992) Pure absorption gradient enhanced heteronuclear single quantum correlation spectroscopy with improved sensitivity. J. Am. Chem. Soc. 114, 10663–10665. 26. Wishart, D. S., Bigam, C. G., Yao, J., Abildgaard, F., Dyson, H. J., Oldfield, E., Markley, J. L., and Sykes, B. D. (1995) H-1, C13 and N-15 chemical shift referencing in biomolecular NMR. J. Biomol. NMR 6, 135–140. 27. Shaka, A. J., Barker, P. B., and Freeman, R. (1985) Computer optimized decoupling scheme for wide-band applications and lowlevel operation. J. Magn. Res. 64, 547–552. 28. Delaglio, F., Grzesiek, S., Vuister, G. W., Zhu, G., Pfeifer, J., and Bax, A. (1995) NMRPipe—A multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293. 29. Bax, A., and Subramanian, S. (1986) J. Magn. Reson. 67, 565–569. 30. Wu, D. H., Chen, A. D., and Johnson, C. S. (1995) An improved diffusion-ordered spectroscopy experiment incorporating bipolargradient pulses. J. Magn. Reson. Ser. A 115, 260–264. 31. Mescher, M., Tannus, A., Johnson, M. O., and Garwood, M. (1996) Solvent suppression using selective echo dephasing. J. Magn. Reson. Ser. A 123, 226–229. 32. Krishnan, V. V., Thornton, K. H., and Cosman, M. (1999) An improved experimental scheme to measure self-diffusion coefficients of biomolecules with an advantageous use of radiation damping. Chem. Phys. Lett. 302, 317–323. 33. Sharma, D., and Rajarathnam, K. (2000) C-13 NMR chemical shifts can predict disulfide bond formation. J. Biomol. NMR 18, 165–171. 34. Teller, D. C., Swanson, E., and de Hae¨n, C. (1979) The translational friction coefficient of proteins. Methods Enzymol. 61, 103–124.