Critical Role of Asparagine 1065 of Human α2-Macroglobulin in Formation and Reactivity of the Thiol Ester*

It has been shown that the relative reaction preference of the C4 thiol ester toward oxygen and nitrogen nucleophiles upon activation by proteinase depends on whether residue 1106 is aspartate or histidine (Dodds, A. W., Ren, X.-D., Willis, A. C., and Law, S. K. A. (1996) Nature 379, 177–179). To determine if the equivalent residue in the related thiol ester-containing protein human α2-macroglobulin (α2M), asparagine 1065, plays a similar role, we have expressed and characterized four α2M variants in which this asparagine has been replaced by aspartate, alanine, histidine, or lysine. The change from asparagine resulted in an altered ability to form the thiol ester. This ranged from failure to form the thiol ester (Asn → Asp) to a maximum extent of formation of about 50% (Asn → Ala). For the three variants that were able to form the thiol ester, the rates of thiol ester cleavage by a given amine were found to be different from one another and slower in nearly all cases than plasma α2M, but with the same relative reactivity of methylamine > ethylamine > ammonia. The rate of conformational change that follows cleavage of thiol esters in a functional half-molecule was also found to differ between the variants and to be slower than plasma α2M. TNS emission spectra indicated that the conformations of the transformed variants differed measurably from transformed plasma α2M. These findings suggest that residue 1065 plays a critical role in human α2M, for formation of the thiol ester, for its subsequent reaction with nucleophiles, and for the conformational change induced by this reaction. By analogy with C4, where this residue influences the nucleophile preference through direct interaction with the thiol ester, residue 1065 in α2M is expected to be located in or very close to the thiol ester region in α2M.

The abundant human plasma protein ␣ 2 -macroglobulin (␣ 2 M) 1 shares a number of properties with the two complement proteins C3 and C4. All three proteins arose from a common ancestral gene (1) and contain a reactive internal thiol ester that becomes even more reactive toward nucleophiles follow-ing limited proteolytic cleavage of the protein. In the case of the monomeric proteins C3 and C4, the proteolytic cleavage occurs as part of complement activation. The consequent activation of the thiol ester results in a fraction (ϳ10%) of the activated C3 or C4 forming covalent cross-links to nucleophiles in the vicinity. C3 shows a preference for oxygen nucleophiles (2), whereas C4 shows a preference for nitrogen nucleophiles, but with an increase in reactivity toward oxygen for the B isotype (3,4). In the case of human ␣ 2 M, activation of the thiol ester toward nucleophiles and its subsequent cleavage results in a major conformational change of the protein that traps, and thereby inhibits, the proteinase that caused the activation (5).
Given the key role that the thiol ester plays in the functioning of both the two complement proteins and of ␣ 2 M, it is important to understand both how the thiol ester is formed and what determines its reactivity toward nucleophiles. For ␣ 1 inhibitor 3, a rodent protein of the ␣-macroglobulin family, it has been shown that thiol ester formation is very dependent on the conformation of the protein and only occurs after folding and subsequent conformational rearrangement that involves a disulfide isomerization (6). Similarly, time-dependent conformation-specific reformation of the thiol ester from the aminecleaved residues that initially formed it has been demonstrated for C3, C4, and ␣ 2 M (7-9), although without an elucidation of the specific structural requirements for the reformation. More definitive information on the molecular basis for the nucleophile preference of the thiol ester in C4 has been provided by an examination of the effect of single site mutations at position 1106 of C4. Although this is a position 115 residues C-terminal from the cysteine that is involved in forming the thiol ester, it is one of only four residues, occurring in a hexapeptide, that differ between the C4A and C4B isotypes of C4. Law and co-workers (10) showed that histidine at this position (C4B) resulted in enhanced reactivity toward oxygen, whereas aspartate (C4A) resulted in overwhelming preference for reaction with nitrogen nucleophiles. The molecular basis for this was elegantly demonstrated to be the formation of an acyl-histidine intermediate in C4B upon activation, that had enhanced reactivity toward oxygen nucleophiles (11).
The equivalent residue in human ␣ 2 M is asparagine 1065, which occurs in the hexapeptide SGSLLN. Comparison of all available sequences of ␣-macroglobulins (13 total) showed that asparagine occurs in 11 and histidine in 2 ( Table I). Because of this high conservation of asparagine and the presence of histidine as the only alternative, as in C4B, we sought to determine whether this residue also influenced the properties of the thiol ester in human ␣ 2 M. Four variants of human ␣ 2 M, in which residue 1065 was changed from asparagine to alanine, aspartate, histidine, or lysine, were therefore expressed in a baby hamster kidney cell expression system (12), and their properties were determined. We report here the characterization of these variants and demonstrate that this residue was critical for normal formation and functioning of the thiol ester.

MATERIALS AND METHODS
Creation and Expression of ␣ 2 M Variants at Position 1065-Plasmid p1167 (8.76 kilobases) (13), a generous gift from Dr. Esper Boel (Novo Nordisk), contains the cDNA for human ␣ 2 M under the control of the adenovirus 2 major late promoter. Single site mutagenesis was carried out directly on the intact plasmid using the polymerase chain reactionbased QuikChange site-directed mutagenesis protocol (Stratagene, La Jolla, CA). Complementatry pairs of oligonucleotides were used. The sequences for the antisense oligonucleotides were 5Ј-CTT TAT GGC ATT XXX GAG CAG TGA CCC-3Ј, where XXX was GGC for the mutation to alanine, GTC for the mutation to aspartate, GTG for the mutation to histidine, and CTT for the mutation to lysine. The mutations were confirmed by sequencing of a stretch of at least 40 nucleotides centered on the codon for residue 1065. Baby hamster kidney cells were transfected with plasmids containing cDNA for ␣ 2 M, dihydrofolate reductase (pSVdhfr), and Escherichia coli amino glycoside 3Ј-phosphotransferase (pRMH140). Methotrexate and neomycin were used to select for stable transfectants, as described (12). Stable transfectants were grown to confluence in roller bottles and cycled between serumcontaining and serum-free medium. ␣ 2 M was isolated from serum-free cycles of medium. Levels of secreted ␣ 2 M were in the range of 3-13 mg/liter for stable transfectants. ␣ 2 M was purified by metal-chelate chromatography, as described previously (14). Protein concentrations were determined spectrophotometrically from the absorbance at 280 nm, using the same extinction coefficient of 640,000 M Ϫ1 cm Ϫ1 for plasma ␣ 2 M (15) and the four recombinant variants.
Polyacrylamide Gel Electrophoresis-Polyacrylamide gels were run under nondenaturing conditions in 5% acrylamide slabs according to the procedure of Davis (16) or under denaturing conditions in 7.5% gels according to the procedure of Laemmli (17). Protein bands were visualized by staining with Coomassie Brilliant Blue.
Quantitation of Thiol Esters and Kinetics of Cleavage-Determination of both the number and kinetics of cleavage of thiol esters present in the variants was carried out by reaction of the free cysteine derived from the thiol ester with DTNB and spectrophotometric measurement of the TNB Ϫ released. TNB Ϫ was quantitiated from the absorbance at 410 nm using an extinction coefficient of 13,500 M Ϫ1 cm Ϫ1 (18). For all measurements, a concentration of DTNB of 100 M was used. The reaction of DTNB with free SH groups under these conditions is essentially instantaneous, so that the rate of this reaction does not influence the measured rate of thiol ester appearance (19). The concentration of free SH groups initially present in samples of plasma, recombinant wild type, and variant ␣ 2 Ms was determined from the immediate change in absorbance at 410 nm upon the addition of ␣ 2 M to the sample cuvette and of an equivalent volume of buffer to the reference cuvette. Both cuvettes contained DTNB at the same concentration and had been zeroed relative to one another prior to the addition of ␣ 2 M. For determination of intact thiol esters, sample and reference cuvettes were set up containing both DTNB and ␣ 2 M (0.5 M final concentration) and zeroed relative to one another. In this way, free SH already present reacted immediately with DTNB to release TNB Ϫ to give an absorbance change that was then zeroed and was thus eliminated from further measurements. Cleavage of thiol esters was then initiated by the addition of amine hydrochloride stock solution (5 M, pH 8.0) to both reference and sample cuvettes to give the desired final concentration. Quantitation of thiol esters was then based only on the time-dependent change in absorbance due to the reaction of DTNB with newly formed SH groups. The whole time course was fitted to a monoexponential function to obtain the best fit for the total change in absorbance. All measurements were made in a dual beam Shimadzu 2100PC spectrophotometer. Data were fitted by nonlinear least squares analysis using the program Scientist (MicroMath, Salt Lake City, UT). The pseudofirst order rate constant was converted to a second order rate constant (k 1 ) by dividing by the free base concentration of the amine at that pH.
Kinetics of Conformational Change-The rate constant for the conformational change step (k c ) was determined directly by following the change in TNS fluorescence after prior rapid cleavage of the thiol esters. ␣ 2 M was prereacted for a short time (20 -60 s) with 2.5 M amine hydrochloride, pH 9.0. The high pH and high amine concentration ensured that all thiol esters had been cleaved within the time of reaction, as calculated from the rate constants for thiol ester cleavage determined by the DTNB assay. This was confirmed by the invariance of the fitted rate constant obtained for experiments with longer preincubation time. The reaction mixture was diluted into 50 mM Hepes buffer, pH 9.0 containing 150 mM NaCl and 50 M TNS. The rate of conformational change was determined from the change in intensity of the TNS fluorescence emission spectrum, monitored at 410 nm, with excitation at 316 nm. Measurements were made on an SLM8000 spectrofluorimeter in thermostatted acrylic cuvettes. Slits of 4 nm (excitation) and 16 nm (emission) were used. Data were fitted by nonlinear least squares analysis to a monoexponential using Scientist (MicroMath).
TNS Fluorescence Emission Spectra-TNS emission spectra were recorded on an SLM8000 spectrofluorimeter, with excitation at 316 nm and emission monitored in 2-nm steps from 360 to 600 nm. Excitation and emission slits were both set to 4 nm. An integration time of 10 s/point was used. Samples were 1.2 ml in 3-ml thermostatted acrylic cuvettes.

Lower Than Expected Thiol Ester Content of 1065
Variants-An immediate demonstration that residue 1065 influences thiol ester function in human ␣ 2 M was the presence of higher electrophoretic mobility species (fast form) for the variants on a polyacrylamide gel run under nondenaturing conditions ( Fig. 1B, odd numbered lanes), indicative of some species without thiol esters (20). The N1065D variant was an extreme case with 100% fast form, indicating no thiol esters. The other three variants showed mixtures of slow, intermediate, and fast electrophoretic forms, with the fast form component ranging from ϳ53% for the N1065K variant through ϳ42% for the N1065H variant to a low of ϳ33% for the N1065A variant, as estimated from scanning of the Coomassie-stained gel and integration. The effect of reaction of each of these samples with methylamine confirmed that the slow and intermediate forms represented a species containing intact thiol esters (full or partial complement, respectively) and the fast form a species lacking intact thiol esters (Fig. 1B, even numbered lanes). Thus, any slow or intermediate form was converted to the fast form through cleavage of the thiol ester by methylamine, whereas the fast form species was unaffected, since there were no thiol esters to be cleaved.
The heterogeneity of the variant species was shown not to be due to the conditions used for production of the BHK-derived recombinant ␣ 2 Ms (Fig. 1A). Thus, both recombinant wild-type ␣ 2 M and a "revertant" wild type, in which we had converted the cDNA of the N1065K variant back to wild-type ␣ 2 M and expressed the resulting wild-type ␣ 2 M, showed a single slow form species with the same mobility as plasma ␣ 2 M (Fig. 1A, odd numbered lanes). These two recombinant wild-type ␣ 2 Ms were also completely transformed to the fast form by reaction with methylamine ( Fig. 1A, even numbered lanes). A more quantitative estimate of the stoichiometry of thiol esters was obtained from DTNB assay of each of the variants upon complete reaction with methylamine, measuring the appearance of new thiol groups, as described under "Materials and Methods." The quantitation of thiol esters was based upon the change in absorbance during the reaction and therefore did not include any free cysteine already present in the variants. Based on this assay, the N1065D variant contained no thiol esters, whereas the N1065A, N1065H, and N1065K variants contained 1.9, 1.6, and 1.7 thiol esters, respectively, based on at least five separate determinations and with an uncertainty of no more that Ϯ 0.2 thiol esters. Recombinant wild-type ␣ 2 M showed the presence of 3.5 Ϯ 0.3 thiol esters.
Basis for Submaximal Thiol Ester Content of Variants-The complete absence of thiol esters in the N1065D variant and the approximately 50 -60% less than maximal thiol ester content of the other three variants could be due to several causes. One is that the thiol esters were originally present but were rapidly hydrolyzed. The second is that the thiol esters did not form or formed to a smaller extent than in wild-type ␣ 2 M. In either case, this should have resulted in the presence of free SH groups in the variant ␣ 2 Ms, as isolated. This was confirmed by the DTNB assay, which showed that N1065D had 1.9 mol of free SH/tetramer, N1065H had 1.7 mol/tetramer, N1065A had 1.1 mol/tetramer, and N1065K had 0.7 mol/tetramer. Recombinant wild-type ␣ 2 M showed the presence of only 0.3 free SH groups.
One way in which the thiol esters might have been cleaved is through exposure of the ␣ 2 M to proteinase, with consequent bait region cleavage and thiol ester activation to nucleophilic attack. If this were the case, SDS-polyacrylamide gel electrophoresis should show the presence of a large fraction of ␣ 2 M chains cleaved in the bait region. The overwhelming appearance of only intact chains indicated that this was not the case for any of the variants or for control plasma ␣ 2 M (Fig. 2, odd numbered lanes), indicating that proteinase activation was not the cause of low thiol ester content. The possibility of a greatly enhanced reactivity toward water, such as in C4B, as the cause of loss of thiol esters was tested for the three variants that contain some thiol esters (N1065A, N1065H, and N1065K) by incubating samples under conditions of buffer, temperature, and pH identical to those of the BHK cells during expression of the variants. Aliquots were withdrawn after 24, 48, and 72 h and run under nondenaturing conditions to determine if there had been any reduction in slow form species as an indicator of further loss of thiol ester through hydrolysis. No such reduction was seen for any of the three variants even for the longest time point (data not shown), indicating that any increased reactivity toward water must be so small as to be unable to account for the reduction in thiol ester content seen in the variants. It is therefore most likely that any fast or intermediate form species contain subunits in which the thiol ester never formed.
Ability of Slow Form Variant Species to React with Proteinase-We examined, in that fraction of each of the variants that formed thiol esters, the ability of proteinase to cleave the bait region and induce the conformational change that results in a change in electrophoretic mobility from slow to fast. All of the slow mobility components were rapidly converted to fast mobility by reaction with trypsin (data not shown, but gel was almost identical to Fig. 1), through the normal cleavage of the bait region (Fig. 2, even numbered lanes) that activates the thiol ester and results in its rapid cleavage. It should be noted that even in methylamine-produced fast form ␣ 2 M, the bait region remains somewhat accessible to proteinase (14). This explains the presence of a fraction of cleavage for the N1065D variant.
Normal Ability of Slow Form Variants to Undergo Heat Fragmentation-One of the characteristic properties of ␣ 2 M, C3, and C4 is that, when heated, each undergoes autolysis of the peptide backbone at the glutamate that forms the thiol ester. This is thought to result from attack of the peptide nitrogen on the carbonyl of the thiol ester followed by hydrolysis. Such autolysis therefore requires an intact thiol ester and the appropriate positioning of the thiol ester group and the intervening peptide backbone. To determine if the three thiol estercontaining variants behaved normally in this regard, samples were heated to 95°C for 35 min, and the products were examined by SDS-polyacrylamide gel electrophoresis. For all three variants, the characteristic heat fragmentation bands of ϳ120 and 60 kDa were observed in about the same amounts relative to uncleaved subunits as was seen for plasma ␣ 2 M (data not shown), suggesting that residue 1065 is not involved in this process. The N1065D variant showed no heat fragmentation.
Conformation of Variants in Their "Fast Form" Conforma-tion Probed by TNS-TNS, noncovalently bound to ␣ 2 M, is a sensitive monitor of conformational change induced by methylamine treatment or reaction with proteinase (21-23). ␣ 2 M has a single class of relatively weak binding sites for TNS (K d ϭ 100 -200 M) (23) that changes both affinity and environment upon change of conformation (21,23). For human plasma ␣ 2 M, this results in a large fluorescence enhancement and a blue shift from 450 to 410 nm upon reaction with either methylamine or proteinase. We therefore used TNS fluorescence to compare the conformations of the four variants. Since the N1065D variant was entirely in fast form, with residue 1065 as a glutamine, and the other three variants were mixtures of slow, intermediate, and fast, we first converted N1065A, N1065H, and N1065K completely into the fast form by reaction with ammonia (for consistency, the N1065D was also treated with ammonia, although no thiol esters were initially present and thus no further reaction was possible). The TNS emission spectra of fast forms of N1065A, N1065H, and N1065K were almost identical (Fig. 3B), differing only slightly in wavelength maximum, which was centered at about 430 nm. The spectrum for N1065D, however, showed a much lower intensity, although with similar wavelength maximum. In contrast, the spectrum for ammonia-treated plasma ␣ 2 M showed a very different position for the wavelength maximum and a lower fluorescence intensity. Because of the major differences between plasma ␣ 2 M and the variants and the less dramatic, but reproducible, differences between the variants, we also compared the spectra of amine-cleaved plasma ␣ 2 M for cleavage by ammonia and methylamine, since this would result in different chemical groups becoming part of the environment of the thiol ester site. This comparison showed that even the small differences in these groups have significant effects on the TNS spectra (Fig.  3A), with both the intensity and wavelength maximum changing for cleavage by ammonia or methylamine. This suggests that the TNS binding site either is very close to the thiol ester site or is exquisitely sensitive to even minor conformational differences in ␣ 2 M.

Kinetics of Thiol Ester Cleavage by Different Amines-
The second order rate constants for cleavage of the thiol esters in the three variants that contain some thiol esters were determined by DTNB assay carried out under pseudo-first order conditions for ammonia, methylamine, and ethylamine and compared with the rate constants for plasma ␣ 2 M determined under identical conditions (Table II). Reactivity toward Tris was also examined, since the low pK a of 8.08 at 25°C (24) enabled high concentrations of the free base form to be achieved at relatively low total amine concentration. Reactivity with Tris was so low as to be unmeasurable, even using a final Tris concentration of 0.5 M. Each of the three variants showed  a Second order rate constant for cleavage reaction of ␣ 2 Ms by different amines. This was determined under pseudo-first order conditions by assay with DTNB. The second order rate constant was calculated from the measured slope of a plot of ln (A max Ϫ A obs )/A max against time, using the free base concentration of amine at the pH of measurement (pH 8.0). b First order rate constant for conformational change in ␣ 2 M, determined from change in TNS fluorescence after initial rapid and complete cleavage of the thiol esters by preincubation of the ␣ 2 M with 2.5 M amine at pH 9.0 for 20 -40 s before dilution into the assay buffer containing 50 M TNS. the same rank order of reactivity as is found for plasma ␣ 2 M, namely methylamine Ͼ ethylamine Ͼ ammonia, with approximately the same relative values for the rate constants, suggesting that accessibility to the thiol ester in each variant is very similar to that in plasma ␣ 2 M. The mutations at position 1065 did, however, have an adverse effect in nearly all cases on the actual rate constants (Table II). Only histidine was able to confer similar reactivity to plasma ␣ 2 M for two of the amines studied, ammonia and ethylamine, although the rate of reaction with methylamine was slower with this variant than with plasma ␣ 2 M. The magnitude of the rate reduction did not correlate with the size of the replacement residue, as might result if accessibility were being affected, since it was greatest for the smallest replacement, alanine.
Effect of Mutations on the Kinetics of Conformational Change-The major conformational change(s) induced in ␣ 2 M by thiol ester cleavage occur cooperatively only after both thiol esters within a half-molecule have been cleaved. To determine whether the mutations at residue 1065 had affected the rate of this conformational change, we determined the rate constant for this step for each of the three variants that contains slow or intermediate form species and for plasma ␣ 2 M under identical conditions. From knowledge of the rate of thiol ester cleavage by amines (see above), we calculated that, by incubating the ␣ 2 M at very high methylamine concentration at pH 9.0 for 20 -60 s, Ͼ95% of the thiol esters could be cleaved. By then diluting the sample into the TNS assay buffer, the kinetics of the conformational change step alone could be followed from the change in TNS fluorescence. We have previously used the same approach for determining this rate constant in plasma ␣ 2 M after chemical modification of the thiol group (12). The changes in TNS fluorescence showed simple monoexponential behavior, with a rate constant that did not depend on the length of prereaction with methylamine, as long as sufficient time of reaction had been allowed to give nearly complete cleavage of the thiol esters. The rate constants were different both from one another and from the rate constant for plasma ␣ 2 M under the same conditions (Table II). As was found for most of the second order rate constants for thiol ester cleavage, the rate constants for conformational change were slower than for plasma ␣ 2 M, again by a maximum of about 4-fold. DISCUSSION We set out to determine whether residue 1065 of human ␣ 2 M plays a role in the reactions of the mechanistically important thiol ester by creating four ␣ 2 M variants with single site mutations at this position and characterizing the properties of the expressed proteins. The choice of residue 1065 was suggested by the demonstrated importance of residue 1106 of human C4 in the mechanism of thiol ester cleavage (10,11) and the equivalent positions represented by these two residues based on sequence alignment of the two related proteins (1,10). We found that residue 1065 of human ␣ 2 M not only influenced the reactivity of the thiol ester but also greatly affected the ability of the thiol ester to be formed in the first place. Residue 1065 also influenced the kinetics of the conformational change that results from thiol ester cleavage and resulted in conformations of the fast form of the ␣ 2 M variant that were significantly different from those of fast form plasma ␣ 2 M, as detected by TNS fluorescence emission spectra. For the kinetics of thiol ester cleavage, it was found that the native asparagine at position 1065 gave the fastest rates when compared with the two variants of alanine or lysine and comparable rates for two of the amines with the histidine variant. For the rate of conformational change, asparagine gave the fastest rate compared with the three variants. Together, these alterations in the properties of human ␣ 2 M resulting from a single site mutation at position 1065 indicate that the wild-type asparagine at this position is very important for achieving the correct conformation of the protein to ensure thiol ester formation and is probably close enough to the thiol ester, once formed, to ensure optimal reactivity toward nucleophiles, probably through direct interaction with the thiol ester. This presumably accounts for the nearly complete conservation of this residue in the 13 macroglobulins sequenced (Table I).
Although the sequence CGEQ is conserved in macroglobulins, in C4, and in C3 as the tetrapeptide that gives rise to the thiol ester (cysteine and glutamine underlined), there is a body of evidence to support the need for additional specific residues (25) and protein conformations in the vicinity of the thiol esterforming residues for efficient formation or reformation of the thiol ester in these proteins. The requirement for a specific conformation is presumably to bring about the correct constellation of residues to align and activate the thiol and glutaminyl carbonyl for mutual interaction. The requirement for specific flanking residues has been nicely demonstrated in C3 by sitedirected mutagenesis (25) and explains in part why the occurrence of the sequence CGEQ, which occurs in some other unrelated proteins, is not sufficient to bring about thiol ester formation. In rat ␣ 1 inhibitor 3, it has been shown by pulsechase experiments that the thiol ester forms only after the folded protein undergoes a gross conformational change that follows slow formation of an interdomain disulfide (6). In human ␣ 2 M, rat ␣ 1 inhibitor 3, and complement proteins C4 and C3, it has been shown that it is possible to reform a fraction of the thiol esters following cleavage by ammonia or methylamine (7)(8)(9). For reformation of the thiol ester in C3, this is only possible from a transient intermediate conformation (8). Our present findings on the TNS-detected conformational differences between the variants and the different abilities of the variants to form thiol esters are in keeping with these previous results. Thus, the TNS binding sites of the fast forms of N1065A, N1065H, and N1065K variants are all very similar, but they are different from that of plasma ␣ 2 M in both wavelength of maximum emission and fluorescence intensity. Each of these three variants forms thiol esters, but less efficiently than wild-type ␣ 2 M. The N1065D variant has a TNS binding site qualitatively similar to the other variants, based on wavelength maximum of fluorescence emission, but quantitatively different in intensity of the fluorescence, probably reflecting a weaker affinity for TNS. This is paralleled by an inability to form the thiol ester at all. Since alanine, histidine, and lysine have little in common, it suggests that the wild-type asparagine is required for a specific interaction with the thiol esterforming residues, such as hydrogen bonding to the glutaminyl carbonyl, that occurs less effectively or not at all with other replacements. It should be noted, however, that different classes of these proteins may bring about this activation in different ways. Thus, in C3 the known sequences (human, mouse, chicken, guinea pig, trout, Xenopus, lamprey, sea urchin, and cobra) all have histidine at the position corresponding to 1065 in human ␣ 2 M.
Thiol ester reactivity toward different nucleophiles has only been examined in detail for human ␣ 2 M (19). It has been shown that the rate of cleavage of the thiol ester by nitrogen nucleophiles is a function of the nucleophility of the nitrogen and, most importantly, the size of the side chain substituent. The importance of size of the side chain is presumed to result from a relatively restricted access of nucleophiles to the thiol ester in the native state. Together these factors result in a nucleophile reactivity for human ␣ 2 M of methylamine Ͼ ethylamine Ͼ ammonia, with branched chain amines such as Tris or dimethylamine very much less reactive than any of these (19). This is the same order of reactivity found here for all three of the variants that form thiol esters, suggesting that access to the thiol ester is not affected in the variants by the substitution at position 1065. However, the lower reactivity in nearly all cases again suggests an important role for asparagine 1065 in promoting attack on the carbonyl, which is not as well replicated by alanine, histidine, or lysine. Since histidine and lysine are more effective than alanine, with histidine as effective as asparagine for reaction with ammonia and ethylamine, this suggests that the role of residue 1065 may be to hydrogen bond to the thiol ester carbonyl and make the carbonyl carbon more electrophilic. In this context, it should be realized that the thiol ester in ␣ 2 M is indeed activated compared with the thiol ester in the model compound ethyl thioactetate, with relative rates of reaction of the thiol ester 37-700-fold higher in ␣ 2 M than in the model compound for amines without bulky side chains (19).
In conclusion, we have shown, through the multiple effects that amino acid substitutions at position 1065 of human ␣ 2 M have on the thiol ester-dependent properties of the protein, that this asparagine is probably in contact with or very close to the thiol ester-forming residues despite being over 100 residues away in the primary structure. It plays an important role in determining the reactivity of these thiol ester-forming residues both for the initial formation of the thiol ester and in its reactivity to nucleophiles once formed. There are thus parallels with the earlier studies on C4 that prompted the present study, but there are also additional effects that were not seen in C4 that suggest that each of the families of ␣-macroglobulins and complement proteins may use different constellations of residues to tailor the properties of the thiol ester to the requirements of the protein.