Site-specific protein modification to identify the MutL interface of MutH

Abstract

We have mapped the region for the protein interaction site of the Escherichia coli mismatch repair protein MutH for its activator protein MutL by a site-specific protein modification approach. For this purpose we generated a cysteine-free variant of MutH and 12 variants thereof, each containing a single cysteine residue at surface positions selected on the basis of available structural and sequence information for MutH. All MutH variants displayed wild type activity both in vivo and in vitro. These variants were then site-specifically modified at their cysteine residues with thiol-specific reagents and then tested for their ability to be stimulated in their DNA cleavage activity by the activator protein MutL. Thereby we were able to identify a defined region in the MutH protein that is important for interaction with MutL, and most likely represents the MutL binding site of MutH.

INTRODUCTION

Correction of replication errors by mismatch repair systems present in most organisms enhances the fidelity of DNA replication by up to a factor of 1000 (1–5). The paradigm of such a system is the MutHLS system from Escherichia coli (6). Mismatch repair by MutHLS is initiated by binding of a MutS homodimer to the mismatch. After recruitment of MutL, this complex searches for a strand discrimination signal in an ATP-dependent manner. In the case of E.coli, this signal is the transiently unmethylated dam site of the newly replicated daughter strand, which is recognized by the endonuclease MutH. After activation by MutL, MutH cleaves the unmethylated daughter strand (7) and thereby marks it for removal and repair-synthesis involving a variety of other proteins (8).

Recently, the structures of E.coli MutH (9), the N-terminal domain of E.coli MutL (10,11), and E.coli as well as Thermus aquaticus MutS (12,13) have been determined. The analysis of the crystal structures of MutH revealed three different forms of the MutH protein indicating that there is an inherent conformational flexibility in the MutH protein (9). The conformations of MutH that do not have a properly formed catalytic center, therefore, might be transformed into the active conformation upon interaction with MutS and MutL. In vitro, the activation of MutH can occur in a mismatch-dependent manner involving MutS and MutL (14,15) or in a mismatch-independent manner by MutL alone (11,16). Physical interaction between MutH and MutL has been demonstrated by two-hybrid assays (16,17), by the ability of MutH to bind to a MutL column (16) as well as by the MutL-mediated binding of MutH to a MutS column (18). The mechanism of activation by MutL involves stimulation of both DNA binding and cleavage by MutH (19). Models for the complex of MutHLS have been proposed (12,20). However, the protein–protein interaction sites have not been verified experimentally so far. It has been proposed that the non-conserved C-terminal tail of MutH is important for the interaction with MutL (9). However, deletion of the last five amino acid residues of MutH did not affect its function in vivo and in vitro (21). Moreover, complementation of a mutH^– mutator phenotype in E.coli by MutH from related bacteria indicated that the non-conserved C-terminal tail of MutH is unlikely to be the interaction site for MutL (22). Hence, the interaction site of MutH for MutL remains elusive.

In order to identify the MutL binding site of MutH, we used a mapping strategy in which single amino acid residues on the surface of MutH were changed to cysteine residues and, after site-specific modification of these cysteines with maleimide derivatives of different size, we determined whether the modifications disrupted activation of MutH by MutL. Similar strategies have already been successfully used for the mapping of the binding interaction between gp45 and gp44/62 (23) and the sonic hedgehog–receptor interaction (24). It has advantages over other mapping strategies, e.g. alanine scanning mutagenesis, not relying on the exact identification of hot spots in binding interfaces (25) and should be applicable for the mapping of the MutL–MutS interaction as well.

MATERIALS AND METHODS

Plasmids and reagents

Escherichia coli K12 strains CC106 {P90C [araΔ(lac-pro)XIII F′lacIZ proB⁺]} (26), TX2928 (CC106 but mutH471::Tn5;Kan^r) (27) and the pET-15b (Novagen) derived plasmids pTX417 and pTX418 containing the mutH and mutL genes, respectively, under control of the T7 promotor were kindly provided by Dr M. Winkler (27). Plasmid pMQ402 (His₆-MutH), a pBAD18 derivative, was a kind gift of Dr M. Marinus (21). For protein expression, the E.coli strains HMS174(λDE3) from Novagen and XL1 blue MRF′ from Stratagene were used. The E.coli strain JM110 (rpsL (Str^r) thr leu thi-1 lacY galK galT ara tonA tsx dam dcm supE44 Δ(lac-proAB) [F′ traD36 proAB lacI^qZ Δ M15] (28) was used for preparation of the unmethylated plasmid pUC8 (29). Luria–Bertani (LB) medium (1 l) was made with 10 g tryptone, 5 g yeast extract, 5 g NaCl and 1 mM NaOH. Ampicillin, kanamycin and rifampicin were added to the media at a final concentration of 75, 25 and 100 µg/ml, respectively.

Site-directed mutagenesis

Site-directed mutagenesis was carried out using a modification of the Quikchange protocol (Stratagene) (30) essentially as described before using the plasmid pMQ402 as template (31). Escherichia coli XL1-blue MRF′ were transformed with the full-length PCR product. Marker positive clones were inoculated in LB medium containing ampicillin for overnight growth. Plasmid DNA was isolated using the QIAprep Spin Miniprep (Qiagen) and the whole mutH gene was sequenced.

Complementation of a mutH mutator phenotype

The complementation assay is based on the mutator phenotype, which becomes apparent when E.coli lacks MutH. The mutation phenotype is monitored by the frequency of mutations in the polB gene, causing resistance to rifampicin (21). Single colonies of CC106 (wild type) or TX2928 (mutH^–) transformed with the indicated plasmids were grown in 3 ml LB cultures containing 100 µg/ml ampicillin, overnight at 37°C. Aliquots of 50 µl from the undiluted or 10^–6 diluted culture were plated on LB agar containing 25 µg/ml ampicillin with or without 100 µg/ml rifampicin, respectively. Colonies were counted after overnight incubation at 37°C.

Purification of recombinant MutH, MutL and MutS proteins

Recombinant His₆-tagged MutH and MutL proteins were expressed in XL1 blue MRF′ and HMS174(λDE3), respectively, and purified by Ni-NTA chromatography essentially as described before (27,31). Protein concentrations were determined using the theoretical extinction coefficients (32).

Site-specific protein modification

The following thiol-specific reagents were used: N-ethylmaleimide (NEM, M_r = 125.1) and benzophenone-4-maleimide (M_r = 277.3) were from Sigma, fluorescein-5-maleimide (M_r = 498.4) was from Molecular Probes and mPEG-MAL (M_r = 5000) was from Shearwater Polymers, Inc. (Fig. 1). Stock solutions of NEM and mPEG-MAL were prepared in water, of benzophenone-4-maleimide and fluorescein-5-maleimide in dimethylformamide. For modification MutH at a concentration of 10–50 µM protein was incubated in 20 mM HEPES, pH 6.5, 200 mM NaCl for 30 min at room temperature with a 2- to 100-fold molar excess of maleimide reagent. Reactions were stopped by adding a 5-fold molar excess of DTT over maleimide reagent for 10 min at room temperature. The extent of modification in experiments with varying excess of modification reagent over protein was monitored by SDS–PAGE followed by fluorescence imaging in the case of fluorescein-5-maleimide modification or Coomassie staining in the case of mPEG-MAL. The extent of modification by NEM was controlled by inhibition of subsequent modification with excess fluoresceine-5-maleimide. All but the benzophenone-4-maleimide modified proteins were purified from excess reagent and, in the case of the PEG modification from unmodified protein, by gel filtration on a Superdex75 column.

Figure 1.

Thiol specific reagents used in this study. The smallest compound N-ethylmaleimide (NEM; M_r = 125) has a radius of action below 0.5 nm, followed by fluorescein-5-maleimide (M_r = 427) with ∼1 nm and then mPEG-MAL 5000 with ∼5 nm.

MutH endonuclease assay

Endonuclease activity of MutH proteins was determined by analyzing the cleavage of unmethylated plasmid pUC8. MutH (50 nM) was incubated with 20 ng/µl pUC8 in 10 mM Tris–HCl, pH 7.5, 10 mM MgCl₂, 0.75 mM ATP and 0.05 mg/ml bovine serum albumin at 37°C in the presence of either 10% DMSO or 2 µM E.coli MutL. Cleavage reactions were stopped by removing 10-µl aliquots from the reaction mixture and adding 2 µl of 250 mM EDTA, pH 8.0, 25% sucrose, 0.1% bromphenol blue and 0.1% xylene cyanol. The reaction products were analyzed by 1.2% agarose gel electrophoresis. The ethidium bromide-stained gels were analyzed with a video documentation system (Bio-Rad). The intensity of the stained DNA bands was quantified using the program Total Lab (Phoretix). Initial rates were calculated from the disappearance of the supercoil form of the plasmid and the appearance of the open circular and linear forms of the plasmid, respectively.

Site-specific UV crosslinking

For cross-linking studies MutH variants were modified with benzophenone-4-maleimide as described above. Modified variants were incubated at 5 µM with 5 µM MutL in 20 mM Tris–HCl, 5 mM MgCl₂, 1 mM ATP on ice and then kept in the dark or irradiated at 354 nm with a handheld UV lamp for 20 min. Samples were then subjected to SDS–PAGE analysis.

RESULTS

Strategy for site-directed protein modification

Strategies for mapping protein interaction sites fall into three broad categories: cross-linking, protection and interference (33). We chose a strategy that in principle allows performing any of these approaches in a site-directed manner and show here our results for an interference analysis. First, we generated a cysteine-free variant of MutH by changing the single non-conserved Cys-96 to serine. This variant was active both in vivo and in vitro indicating that the sulfhydryl group is not important for function (Tables 1 and 2). Next, we used the alignment of all available MutH sequences and the structural information for E.coli MutH to identify potential surface residues that could serve as targets for modification. Ten strategically positioned sites were selected for an initial study. None of these residues is conserved in the MutH protein family, and together they cover the entire surface of the molecule. After the results of this first study (see below) two additional sites were selected, Val-166 and Leu-167, which are conserved in the MutH proteins. The positions of the residues on the surface of the MutH protein are shown in Figure 2. Together these 12 residues cover 9% of the total protein surface.

Table 1. In vivo complementation screen for MutH proteins in a mutH^– strain.

Plasmida	Mutation rateb (× 108)	nc
Vector control	360 (28)	35
Wild type	1.6 (0.7)	15
E77A	280 (68)	4
C96S	1.6 (0.8)	10
S9C;C96S	1.2 (0.7)	10
T40C;C96S	2.3 (1.0)	5
S85C;C96S	1.1 (0.7)	10
S104C;C96S	3.9 (2.0)	5
H115C;C96S	5.9 (2.7)	5
S145C;C96S	1.5 (0.8)	10
E156C;C96S	1.4 (0.7)	10
V166C;C96S	4.1 (2.1)	5
L167C;C96S	4.1 (1.5)	10
R172C;C96C	2.1 (1.0)	10
A190C;C96S	2.4 (1.0)	10
E202C;C96S	1.6 (1.2)	5

^aEscherichia coli TX2928 (mutH^–) was transformed with the indicated plasmid containing no mutH gene (vector control) or wild type or variants of the mutH gene with the indicated amino acid exchange(s). The variant E77A is an active site mutant of MutH which is unable to cleave the DNA and was included here as a negative control. For details see Materials and Methods.

^bMutation rates were calculated by the method of median (38). Standard deviations are given in parentheses.

^cn is the number of independent experiments.

Table 2. Analysis of MutH variants for endonuclease activity and stimulation by MutL.

	MutL-unstimulateda (kuD) (1/h)	MutL-stimulated (kuL) (1/h)	Stimulation by MutL (kuL/kuD)
Wild type	1.70	14.8	9
E77A	n.d.c.b	n.d.c.	–
C96S	1.70	15	9
S9C;C96S	0.17	10	59
T40C;C96S	2.04	15	7
S85C;C96S	1.67	13	8
S104C;C96S	0.45	12	30
H115C;C96S	0.10	22	220
S145C;C96S	0.09	10	110
E156C;C96S	1.70	19	11
V166C;C96S	0.39	7	18
L167C;C96S	0.70	20	29
R172C;C96C	1.30	13	10
A190C;C96S	0.39	7	18
E202C;C96S	0.10	4	40

^aNote that the unstimulated cleavage reaction was carried out in the presence of 10% DMSO, which increases the MutH activity roughly by a factor of 10. Reaction rates are given for the conversion plasmid pUC8 from supercoil form to the linear form.

^bn.d.c. = no detectable cleavage.

Figure 2.

Selection of single cysteine variants. The structure of MutH (pdb code 2azo chain B) was checked for potential surface sites where modifications could serve as probes for the MutL interaction site. Residues selected for mutagenesis are indicated by yellow spheres. The positions of the active site residues (Asp-70, Glu-77 and Lys-79) are indicated by red spheres.

Single cysteine variants of MutH are active in vivo and in vitro

A prerequisite for the mapping study is that the mutation is not affecting the function of the protein. In rare cases, the mutation directly identifies the protein–protein interface. Such residues are called ‘hot spots’, since they are of key importance for complex formation (25,34). We, therefore, tested MutH variants first for their in vivo function in mismatch repair. All variants were able to complement a mutH mutator phenotype (Table 1). The results indicate that none of the mutated residues is a hot spot for the function in mismatch repair and, thus, also for the interaction with MutL. Subsequently, the proteins were purified and tested for their ability to cleave unmethylated plasmid DNA in the presence of 10% DMSO. It should be noted that the endonuclease activity of MutH in the presence of 10% DMSO is ∼10-fold higher than in its absence, a phenomenon which is well known for many restriction endonucleases (35). In the absence of MutL, some of the unmodified MutH variants (u) showed a significantly reduced endonuclease activity (k^u_D) compared to the wild type MutH protein (Table 2). For example the variants H115C;C96S and S145C;C96S have <10% activity compared with the C96S variant. However, in the presence of the activator protein MutL all variants displayed a near wild type activity (k^u_L). This indicates that some mutation might affect the conformational equilibrium between the inactive and active forms of MutH but that this effect can be overcome by the MutL-stimulation, which might have a chaperone-like function on MutH. Note that MutL is structurally related to the chaperone Hsp90 (11). In summary, all variants show a stimulation in endonuclease activity (k^u_L/k^u_D) of approximately 10 or greater (i.e. >100-fold compared with the MutH activity in the absence of DMSO). This together with the in vivo results suggests that none of the positions selected is a hot spot for the interaction with MutL.

Site-specific modification of MutH single cysteine variants

We modified the MutH variants with thiol-specific reagents ranging in size from <0.4 nm (NEM), ∼1 nm (fluorescein-5-maleimide) up to 5 nm (mPEG-MAL 5000), based on the hydrodynamic radius of mPEG-MAL 5000) (Fig. 1). The extent of modification was monitored by SDS–PAGE and fluorescence imaging or Coomassie staining and proved to be >90% at ratios of reagent:protein >5-fold (NEM and fluoresceine-5-maleimide) or >25-fold (mPEG-MAL) (Fig. 3). In the case of the MutH variants S104C;C96S and S145C;C96S a 100-fold excess of modification reagent over protein was needed to achieve >90% modification (data not shown). The modified proteins were purified from excess reagent by gel filtration chromatography. In the case of the PEG-modified MutH variants we were able to separate the modified from unmodified variants since the PEG modification (R_h = 2.7 nm, M_r^app = 29 000) almost doubles M_r^app of the modified variants (R_h = 3.3 nm, M_r^app = 56 000) compared to the unmodified variants (R_h = 2.4 nm, M_r^app = 23 000) (Fig. 4A). The pegylated proteins were >95% pure as checked by SDS–PAGE (Fig. 4B).

Figure 3.

Site-directed protein modification. SDS–PAGE analysis of titrations of MutH variant S9C;C96S with increasing excess of thiol specific modification reagent. Similar results were obtained for the other variants. Modification of MutH variant with increasing molar ratios of (A) mPEG-MAL; (B) fluorescein-5-maleimide; (C) N-ethylmaleimide (NEM). The extent of modification was quantified after Coomassie staining or fluorescence imaging. In the case of NEM the extent of modification was monitored by inhibiton of a subsequent modification with 10-fold exess fluoresceine-5-maleimide.

Figure 4.

Site-directed protein modification of MutH single cysteine variants with PEG. (A) The gel filtration analysis of MutH variants, mPEG-MAL 5000 and pegylated MutH variants was performed on a Superdex75 column. The following markers (open circles) were used: β-amylase (M_r = 200 kDa), bovine serum albumin (M_r = 66 kDa), carbonic anhydrase (M_r = 29 kDa) and cytochrome C (M_r = 12 kDa) with V_e corresponding to the peak elution volume of the protein and V₀ representing the void volume of the column determined with blue dextran 2 000 000. The average elution volumes (V_e) were 12.4 ± 0.2 ml and 10.4 ± 0.2 ml for the unmodified and pegylated MutH variants, respectively. (B) SDS–PAGE analysis of unmodified or modified MutH variants after purification by gel filtration. Proteins were stained with Coomassie Brilliant Blue. Note that the cysteine-free variant MutH C96S is not modified by PEG and that the single cysteine variants are modified only at a single position.

Endonuclease activity of modified MutH single cysteine variants

We determined the endonuclease activity in the presence of DMSO but in the absence of the activator protein MutL (k^NEM_D, k^FLU_D, k^PEG_D) in order to check whether the modification has changed the catalytic activity of the modified MutH variants per se. Interestingly, the pegylated variants fall into different classes (see Table 3): (i) variants that are inhibited upon modification (k^u_D/k^PEG_D > 1), (ii) variants that are almost not affected by the modification (k^u_D/k^PEG_D ≈ 1), and (iii) variants that are activated upon modification (k^u_D/k^PEG_D < 1). This has to be taken into consideration when investigating the effect of the modification in the MutL-stimulated endonuclease activity of the modified MutH variants.

Table 3. Effect of modification on the unstimulated and MutL-stimulated MutH endonuclease activity.

	Unstimulateda	MutL stimulated
	(kuD/kNEMD)	(kuD/kFLUD)	(kuD/kPEGD)	(kuL/kNEML)	(kuL/kFLUL)	(kuL/kPEGL)
S9C;C96S	1.3	1.3	1.9	1.2	3.3	1.4
T40C;C96S	1.5	1.5	>10	1.4	1.6	2.8
S85C;C96S	0.6	1.2	1.0	0.7	0.9	1.1
S104C;C96S	0.8	2.5	0.5	1.7	4.1b	>10
H115C;C96S	2.3	2.1	<0.5	1.4	1.1	2.2
S145C;C96S	0.7	0.7	<0.5	0.9	1.1	0.7
E156C;C96S	n.d.c	2.5	1.5	1.6	1.8	2.3
V166C;C96S	1.9	3.2	0.6	1.1	5.3	>10
L167C;C96S	0.6	0.8	0.6	1.8	>10	>10
R172C;C96C	1.0	7.1	1.2	0.9	0.9	8.3
A190C;C96S	1.9	2.9	4.4	1.4	1.6	1.1
E202C;C96S	0.7	0.7	<0.5	0.5	0.5	1.1

^a For cleavage rate constants (unstimulated (k^u_D); MutL-stimulated: (k^u_L)) of the unmodified variants see Table 2.

^bReduction in endonuclease activity of >4 (statistically significant with p < 0.01) are highlighted in bold

^cn.d. not determined.

Subsequently, we determined the MutL-stimulated cleavage activity of the modified MutH variants (k^NEM_L, k^FLU_L, k^PEG_L). None of the modified variants showed a significantly increased MutL-stimulated cleavage activity compared to the unmodified variants. About half of the modified MutH variants did not show any changes in MutL-stimulated cleavage activity upon modification with N-ethylmaleimide, fluorescein-5-maleimide or mPEG-MAL 5000 indicating that these sites of the protein are unlikely to be part of the MutH–MutL interface. However, the variants S104C;C96S, V166C;C96S or L167C;C96S showed a reduction in MutL stimulation upon modification with fluorescein-5-maleimide (k^u_L/k^FLU_L > 4) or mPEG-MAL 5000 (k^u_L/k^PEG_L > 10), respectively, compared with the unmodified variants.

Structure–activity map of modified MutH variants

We represented the activity data on the surface of the MutH protein (Fig. 5) in order to obtain a structure–activity relationship (SAR). It is obvious that all residues which upon modification show a strong reduction in MutL-stimulated endonuclease activity (e.g., k^u_L/k^PEG_L > 10) cluster in one region of the protein surface. This suggests that, based on our steric interference analysis, the MutL interface of MutH is located in this area of the protein.

Figure 5.

SAR of modified MutH variants. (A–C) The surface of MutH (pdb code 2azo chain B) was colored according to the results of our interference analysis (see panel D). Residues not modified are indicated in grey. Shown are the front view (A) similar as in Figure 2, a top view (B) and a side view (C). (D) The interference with MutL-stimulation of MutH endonuclease activity (k^u_L/k^PEG_L) upon pegylation is plotted against the inhibition of the unstimulated MutH endonuclease activity (k^u_D/k^PEG_D) upon pegylation. Note that variants which upon pegylation show a high interference of the MutL-stimulated endonuclease activity cluster in one region of the protein.

A comparison of the structure–activity map with an evolutionary trace map of the MutH branch of the MutH/Sau3AI sequence family reveals that several conserved residues cluster in the proposed interaction region (data not shown). Notably one of these residues is Leu-167. The variant L167C;C96S was one of the most severely affected variants in MutL-stimulation upon modification with either fluorescein-5-maleimide or mPEG-MAL. This further supports our proposition that this area represents the MutH–MutL interface of MutH.

Mapping the MutH–MutL interaction by cross-linking

Since the interference analysis cannot prove the absence of an interaction between the modified MutH variant and MutL we performed site-specific UV crosslinking to probe the proximity of amino acids in MutH to MutL. Therefore we modified the MutH variants with the photoreactive cross-linker benzophenone-4-maleimide and subsequently incubated the modified variant with or without irradiation by UV light (354 nm). The reaction mixtures were analyzed on SDS–PAGE followed by Coomassie staining. Whereas the MutH variant R172C;C96S was able to form a crosslink with MutL to form a species of M_r ≈ 100 000 which corresponds to a 1:1 complex between MutL (M_r = 70 000) and MutH (M_r = 27 700) (Fig. 6), the variants S104C;C96S; V166C;C96S and L167C;C96S were not able to do so. No crosslinks were observed without irradiation, with MutH in the absence of MutL, or with MutL in the absence of MutH (data not shown).

Figure 6.

Site-specific UV crosslinking of MutL to variant MutH R175C;C96S. SDS–PAGE analysis of UV-crosslinking MutL to the benzophenone-5-maleimide modified MutH variant R172C;C96S as described under Materials and Methods. The position of the crosslink is indicated by an asterisk.

DISCUSSION

The identification of specific amino acid residues involved in protein–protein interaction is fundamental to understanding structure–function relationships. In this study we wanted to map the binding site of MutH for its activator protein MutL by introducing more or less bulky modifications at selected positions on the surface of MutH and determining their interference with MutL activation of MutH. Guided by the structure of MutH and a sequence alignment of MutH proteins we generated 12 variants of MutH containing a single cysteine residue located at strategic positions on the surface of the protein (Fig. 2). All variants displayed a normal activity in vivo (Table 1). Due to the fact that the number of proteins expressed from the plasmid are probably much higher than in a wild type cell, a 10-fold reduction in activity of MutH is unlikely to be detected (21). Our in vitro data on the MutL-stimulated activity of MutH, however, revealed that none of the variants was reduced in activity more than by a factor of 1.5 (Table 2). Therefore, we conclude that the amino acid exchanges did not interfere with the function of the protein in mismatch repair and that our selection of positions to be modified is reasonable.

Next we modified all the MutH variants with thiol specific reagents, i.e. N-ethylmaleimide, fluorescein-5-maleimide and mPEG-MAL 5000. The pegylated MutH variants could be quantitatively separated from excess reagent and unmodified MutH protein by gel filtration chromatography as shown by a SDS–PAGE analysis (Fig. 4). It should be noted that the degree of pegylation shows some dependence on the site of the cysteine residues that were modified and, therefore, the purification by gel filtration chromatography was crucial to get >95% modified variants as judged by SDS–PAGE. The modified variants of MutH were tested for endonuclease activity in the absence and presence of the activator protein MutL (Table 3). Most of the variants displayed only small changes (<2-fold) in endonuclease activity upon modification. However, there are a few exceptions: the pegylated variants S104C;C96S, V166C;C96S, L167C;C96S and to a lesser extent R172C;C96S showed a significant interference with MutL-stimulation of cleavage activity compared with the unmodified variants. When modified with the smaller probe fluorescein these variants showed interferences in the order L167C;C96S > V166C;C96S > S104C;C96S > R172C;C96S ≈ C96S. It should be noted that the variants V166C;C96S and L167C;C96S were made as a result of an initial interference study with the other 10 variants. Since the endonuclease activity of these modified variants in the absence of MutL was not changed, we concluded that these residues are part of the protein interaction site for MutL.

This conclusion is strengthened by our structure–activity analysis of the modified variants (Fig. 5). The positions of the modified amino acid residues with the highest effect on MutL-stimulated endonuclease activity map to one region of MutH, a region that only slightly overlaps with the hitherto proposed protein interaction site of MutH for MutL based on a docking model (12,20). Moreover, the UV crosslink between the benzophenone-4-maleimide modified MutH variant R172C:C96S with MutL, indicates proximity between Arg-172 of MutH and residues of MutL. The absence of UV crosslinks between MutL and other benzophenone-4-maleimide-modified variants can be interpreted either as due to an absence of proximity of these residues to residues in MutL or due to an interference of the modification with interaction to MutL. The second possibility could be true for the three variants S104C;C96S, V166C;C96S and L167C;C96S since these variants displayed interference in activation by MutL upon modification with fluorescein-5-maleimide, which has a similar radius of action as benzophenone-4-maleimide.

Our analysis will be extended to study also the interaction between MutH and MutS. Furthermore, we are currently generating a cysteine-free variant of MutL and will conduct a similar analysis with MutL as described here for MutH. Since MutL has been shown to interact with several other proteins beside MutH, e.g., MutS (36), UvrD (16) and Vsr (37), such an analysis will further our understanding of the molecular basis of mismatch repair.