Mutagenesis of Human Acetylcholinesterase IDENTIFICATION OF RESIDUES INVOLVED IN CATALYTIC ACTIVITY AND IN POLYPEPTIDE FOLDING*

Evidence for the involvement of Ser-203, His-447, and Glu-334 in the catalytic triad of human acetylcholinesterase was provided by substitution of these amino acids by alanine residues. Of 20 amino acid positions mutated so far in human acetylcholinesterase (AChE), these three were unique in abolishing detectable enzymatic activity (less than 0.0003 of wild type), yet allowing proper production, folding, and secretion. This is the first biochemical evidence for the involvement of a glutamate in a hydrolase triad (Schrag, J.D., Li, Y., Wu, M., and Cygler, M. (1991) Nature 351, 761-764), supporting the x-ray crystal structure data of the Torpedo californica acetylcholinesterase (Sussman, J.L., Harel, M., Frolow, F., Oefner, C., Goldman, A., Toker, L. and Silman, I. (1991) Science 253, 872-879). Attempts to convert the AChE triad into a Cys-His-Glu or Ser-His-Asp configuration by site-directed mutagenesis did not yield effective AChE activity. Another type of substitution, that of Asp-74 by Gly or Asn, generated an active enzyme with increased resistance to succinylcholine and dibucaine; thus mimicking in an AChE molecule the phenotype of the atypical butyrylcholinesterase natural variant (D70G mutation). Mutations of other carboxylic residues Glu-84, Asp-95, Asp-333, and Asp-349, all conserved among cholinesterases, did not result in detectable alteration in the recombinant AChE, although polypeptide productivity of the D95N mutant was considerably lower. In contrast, complete absence of secreted human AChE polypeptide was observed when Asp-175 or Asp-404 were substituted by Asn. These two aspartates are conserved in the entire cholinesterase/thyroglobulin family and appear to play a role in generating and/or maintaining the folded state of the polypeptide. The x-ray structure of the Torpedo acetylcholinesterase supports this assumption by revealing the participation of these residues in salt bridges between neighboring secondary structure elements.

ergic transmission. By rapid hydrolysis of the transmitter acetylcholine, the enzyme effectively terminates the chemical impulse, allowing rapid and repetitive responses (for recent reviews see Chatonnet and Lockridge (1989), Taylor (1991), and Hucho et al. (1991)). AChE' is a serine hydrolase, reacting with its natural substrate at close to diffusion-controlled rate (Bazelyansky et al., 1986). Sequence conservation analysis relates the cholinesterases to a superfamily of polypeptides (Myers et al., 1988), including enzymes such as microsomal carboxylesterase (Korza and Ozols, 1988), cholesterol esterase (Kyger et al., 1989), lysophospholipase (Han et al., 1987), Geotrichum lipase (Shimada et al., 1989) and Drosophila esterase-6 (Oakeshott et al., 1987), as well as noncatalytic polypeptides such as the C-terminal domain of thyroglobulin (Mercken et al., 1985) and the Drosophila cell adhesion proteins neurotactin and glutactin (De La Escalera et al., 1990;Olson et al., 1990). AChE is the best characterized enzyme in this superfamily. Kinetic studies have indicated that the active site of AChE consists of two subsites: an anionic subsite, to which the trimethylammonium group of acetylcholine binds, and an esteratic subsite which interacts with the ester-bond region and mediates catalysis.
The esteratic subsite of AChE resembles active sites of serine hydrolases of other enzyme families. The active-site serine, which is the target for organophosphate interaction (MacPhee-Quigley et al., 1985) was shown by sequence comparison to reside within an amino acid stretch conserved in many hydrolases including serine peptidases (Gibney et al., 1990). A histidine residue appears to be part of the active site as well, as indicated both by the pH dependent kinetics (Krupka, 1966) and by inhibition experiments with the histidine-specific reagent diethyl pyrocarbonate (Roskoski, 1974). Moreover, active-site serine and histidine residues were identified in Torpedo californica AChE as Ser-200 and His-440 by site-directed mutagenesis (Gibney et al., 1990).
All these observations suggest that AChE-driven hydrolysis, like protease-driven hydrolysis, depends on a charge relay mechanism in which a carboxylic amino acid stabilizes the tautomer of the imidazole necessary for proton transfer. Yet, existence of such a Ser-His-acid catalytic triad in AChE has been subjected to controversy on mechanistic grounds (Quinn, 1987). Various Asp residues have been suggested as putative triad-related carboxylic residues on the basis of evolutionary conservation (Soreq and Prody, 1989;Doctor et al., 1989;Krejci et al., 1991). The recently resolved x-ray structure of TcAChE (Sussman et al., 1991) suggests, however, that the active site of AChE differs from that of the prototypic serine hydrolases, such as chymotrypsin and subtilisin, both in containing a Glu rather than an Asp in its catalytic triad and in the relative spatial configuration of the three components of the triad.
Site-directed mutagenesis of recombinant AChE could be instrumental in identifying amino acid residues essential for enzyme function. An efficient expression system for human AChE cDNA was recently developed and employed to produce soluble, secretable enzyme in the 293 human cell line (Velan et al., 1991a;Kronman et al., 1992). This expression system is easily amenable to site-directed mutagenesis and was used to examine the involvement of disulfide bonds in assembly of AChE subunits (Velan et al., 1991b). In the present study we use site-directed mutagenesis to analyze the catalytic triad of recombinant human AChE (rHuAChE). We confirm the participation of ser and His residues in the active site and examine several conserved carboxylic moieties for their possible involvement in the enzymatic activity. We demonstrate that presence of Glu at position 334 (analogous to Glu-327 of TcAChE) is essential for hydrolysis, while many other conserved carboxylic residues are most likely required for maintaining structural integrity.

EXPERIMENTAL PROCEDURES
Construction of Expression Vectors for AChE and Its Muteins-Plasmid construction, isolation of DNA fragments, cloning, and bacterial transformation were performed essentially as described (Ausubel et al., 1987). pEwCAT, the bipartite vector expressing both human ache and cat (Fig. 1A) was constructed as follows. ( a ) The DNA fragment containing the SV40 early promoter, the CAT coding sequence and SV40 poly(A) signal were isolated from pSV2CAT (Gorman et al., 1982) as an AccI-BamHI fragment and protruding ends were "filled-in." (b) pLINK5, a pGEM derived polylinker-vector (Velan et al., 1991a) was digested with BglII, "filled-in," and ligated to the cat DNA cassette to create pL5SCAT.
( c ) The HuAChE cassette, which includes also a promoter and a polyadenylation signal, was isolated from pL5CA (Velan et al., 1991a) as a KpnI-XbaI DNA fragment and introduced between the unique KpnI-XbaI site of pL5SCAT to generate pEwCAT (Fig. 1A). Expression vectors for the various HuAChE mutants were generated by replacing the wild type HuAChE cassette by the mutated cassettes (see below) using the unique KpnI and SpeI sites of pEwCAT.
Site-directed Mutagenesis-HuAChE sequence variants which conserve the wild type coding specificity, but carry new unique restriction sites (Fig. 2B), were created by replacement of wild type HuAChE cDNA fragments (carried on pL5CA) with synthetic DNA duplexes. In HuAChE-w3 the sequence between Bsu36I and the distal BamHI was replaced by synthetic DNA to generate unique AccI, BamHI, and BstBI sites. HuAChE-w4 was derived from HuAChE-w3 by replacement of sequences between AccI and BstEII by a DNA containing NruI, NarI, and XhoI as new sites. For HuACbE-w5 construction, the HuAChE-w3 sequences between PmlI and Bsu36I were replaced by a DNA fragment carrying BclI, NarI, and BglII sites. The native HuAChE sequence as well as AChE-w3, AChE-w4, and AChE-w5 sequences were used to generate the various mutated AChE sequences described in this study (Table I). The cDNA spanning the amino acid targeted for mutagenesis was excised by cutting at the nearest restriction sites on the appropriate AChE cDNA variant and then replaced with a synthetic DNA duplex carrying the mutated codon ( Fig. 1, Table I).
All synthetic DNA oligodeoxynucleotides (-60 nucleotides long) were prepared using the automatic Applied Biosystems DNA synthesizer (Foster City, CA) and then assembled by ligation to generate the final DNA fragments. The sequences of all cloned synthetic DNAs were verified by the dideoxy sequencing method (Sequenase kit; U. S. Biochemical Corp.).
Transient Transfection-CsC1-purified plasmid preparations were used to transfect 293 cells using the calcium phosphate method (Wigler et al., 1977). At least two different clone isolates were tested for each plasmid construct. Transfection was carried on as described previously (Velan et al., 1991a) except that cells were transferred 24 h after transfection to medium containing 10% AChE-depleted serum. AChE depletion was carried out by batch adsorption of fetal calf serum (Bet Haemek, Israel) to procainamide-conjugated Sepharose (Ralston et al., 1985) for 18 h, 4 "C, at a serum/resin ratio of 30:l (v/ v). Cells were incubated in 2 ml of medium/lOO-mm plate for 48 h. Medium was collected, concentrated 10-fold by vacuum centrifugation, and assayed for AChE; medium of mock-transfected 293 cells served as control. Cell lysates (Velan et al., 1991b) were assayed for intracellular AChE and CAT activities (Gorman et al., 1982).
Determination of AChE Actiuity-AChE activity in medium and in cellular fractions of transfected cells was assayed according to Ellman et al. (1961) in the presence of 0.5 mM acetylthiocholine, 50 mM sodium phosphate buffer, pH 8.0, 0.1 mg/ml bovine serum albumin, and 0.3 mM 5,5'-dithiobis(2-nitrobenzoic acid). The assay was performed at 27 "C and monitored by a Thermomax microplate reader (Molecular Devices; Menlo Park, CAI. One unit is defined as the amount of enzyme hydrolyzing 1 pmol of acetylthiocholine/min. AChE activity of individual mutants (average of four transfections) was calculated by normalizing to coexpressed CAT activity. Activity of rHuAChE muteins was also monitored by the radiometric method (Johnson and Russell, 1975) using ["Hjacetylcholine (Du Pont-New England Nuclear) as substrate, allowing the hydrolysis to proceed for up to 18 h at room temperature. In both colorimetric and radiometric AChE assays, medium collected from mock-transfected 293 cells was used as negative control. Activity associated with this control never exceeded a value of 1 milliunit/ml.
Antibodies and Sera-Hyperimmune antisera to full-length Hu-AChE catalytic subunits were generated by immunization with purified recombinant enzyme derived from 293 transfected clones (Kronman et al., 1992). Rabbits were injected at 2-week intervals, with 100 pg HuAChE per injection. The first two intramuscular injections were administered with complete Freund's adjuvant, while the last subcutaneous injection was with incomplete Freund's adjuvant. BALB/c mice were injected with four doses of 10 pg of rHuAChE. The first dose was injected into the footpad with complete Freund's adjuvant, and the second was injected subcutaneously 4 weeks later with incomplete Freund's adjuvant and was followed by two similar injections at 2-week intervals. Animals were bled 2 weeks after the last injection; ELISA antibody titers in immunized animals were about 1 X lo6.
Quantitation of HuAChE by ELISA-A HuAChE specific antigencapture ELISA was developed. Ninety-six-well microtiter plates (Nunc; Denmark) were coated (37 "C overnight) with rabbit anti rHuAChE antiserum, diluted 1:lOO in 0.05 M carbonate buffer, pH 9.6. Blocking was performed with 2% bovine serum albumin in phosphate-buffered saline containing 0.05% Tween 20 and 0.05% sodium azide. Aliquots (50 pl) of cell growth medium containing secreted AChE were added to wells and incubated at 37 "C overnight. Plates were washed five times with saline.
Mouse anti-HuAChE hyperimmune serum (50 pl/well) diluted 1:4,000 in phosphate buffered saline was used as second antibody; incubation and washing were performed as above. Rabbit anti-mouse IgG (50 pl) conjugated to alkaline phosphatase (Sigma; diluted 1:500) were added and allowed to incubate for 2 h at 37 "C. Plates were washed and developed by a colorimetric reaction (Sigma 104; Sigma) at room temperature for 60 min. The ELISA was calibrated by standard curves of 2-fold dilutions (320-0.6 ng/ml) of purified rHuAChE (specific activity -6,000 units/mg; Kronman et al., 1992). Plates were read (405 nm) in a Tbermomax (Menlo Park, CA) microplate reader using the SOFTmax program for curve fitting. AChE-protein production levels of individual mutants (average of four transfections) were calculated by normalization to CAT activity. Specific activity was derived from this value, and from the CATnormalized AChE activity value.
In alternative protocols the various monoclonal anti-HuAChEs were used to replace the mouse polyclonal serum as second antibody. All monoclonals used were diluted 1:1,000 in growth medium.
Structure Analysis and Molecular Graphics-Analysis of the threedimensional structure of AChE was performed on an E & S PS390 system using the FRODO program (Jones, 1978;Pflugrath et al., 1984) and on a Silicon Graphics Personal IRIS model 35, using the INSIGHT-I1 program (Biosym Corp., San Diego, CA). The coordi-nates of TcAChE structures (Sussman et al., 1991) used are those deposited in the Brookhaven Protein Data Bank, entry code lACE (Bernstein et al., 1977).

Generation of AChE Mutants-
The amino acids targeted for mutagenesis are given in Fig, 1 and Table 1. The strategy used for introduction of site-directed mutations is based on DNA cassette replacement (Shafferman et al., 1987) of wild type HuAChE cDNA coding sequences (Soreq et al., 1990) by synthetic DNA duplex cassettes carrying the mutated codon. For this purpose, we have generated a series of HuAChE vectors (AChE-w3, AChE-w4, and AChE-w5; Fig. lB), in which HuAChE amino acid sequence is conserved, yet degeneracy of genetic code is used to create new unique restriction sites separated from each other by 60-120 base pairs. I t should be noted that vectors carrying the three new cDNA variants expressed the HuAChE with efficiencies essentially comparable with that of wild type cDNA vector.
The various mutations were introduced into the appropriate vector by DNA cassette replacement as described in Table I.
This replacement method appears to be of higher fidelity than methods involving replication by polymerases (For review see Hedstrom et al. (1991)). Nevertheless, two independent (sequenced) plasmid isolates were always used for analysis of each specific mutation. The mutated HuAChEs were expressed in the context of bipartite expression vectors based on pEwCAT (Fig. 1A). The major element in these plasmids besides the HuAChE cassette (consisting of cytomegalovirus promoter, HuAChE or mutant coding sequence, and SV40 early polyadenylation signak Velan et al., 1991a) is the CAT reporter cassette (Gorman et aL, 1982). The CAT cassette serves as an internal reference in the various transfection experiments for determining efficiencies of transfection and expression.
The putative three-dimensional locations (see "Discussion") of the HuAChE residues, targeted for mutagenesis, were deduced from the x-ray structure of TcAChE (Sussman et al., 1991). HuAChE and TcAChE are likely to have a similar fold as indicated by sequence homology (Soreq et al., 1990) and comparative modeling studies (Barak et al., 1992).

Analysis of M u t a~t AChEs: I~e n t~f i c u t~n of the Actiue-s~~e Serine and Histidine in
HuAChE-Secretion of AChE by 293 cells transfected with either wild type or mutated cDNAs was examined. The methodology utilized for analyzing the recombinant AChEs in cell growth medium is demonstrated in Table If. To correct for variations in efficiency of transfection, experiments were performed in quadruplicates, and all measurements were normalized to the coexpressed CAT activity. Enzymatic activity and protein mass of the secreted AChE were measured independently, by the Ellman reaction and by ELISA based on polyclonal antibodies to rHuAChE, respectively (see "Experimental Procedures"). These procedures allow calculation of AChE-specific activity directly in the transfection medium, without resorting to enzyme purification. Indeed, the 6.3 milliunits/ng average specific activity value obtained for the wild type AChE cDNA transfection (Table 11) corresponds very well to values obtained previously (Kronman et al., 1992) by active-site titration (5.7 milliunitsl ng) and by protein quantitation of purified rHuAChI3 (6.0 milliunits/ng).
The efficient interaction of the nonactive mutant polypeptides with polyclonal antibodies (raised against native Hu-AChE) in the antigen-capture ELISA suggests that loss of activity is not the consequence of a major conformational change. This assumption was further substantiated by examining the interaction of the mutated polypeptides with four monoclonal antibodies (AE-1 and AE-2, Fambrough et al. (1982); HR-5 and A123, Brimijoin et al. (1983)) directed toward conformational epitopes. Indeed, both S203A and H447A AChE polypeptides interacted with all four monoclonal antibodies as efficiently as the wild type enzyme (see below).
Trace background activity (1 milliunit/ml) observed in the mock-transfection medium sets a limit to the determination of residual specific activities. Only mutants maintaining an activity higher than 0.002-0.005 milliunit/ng (1-3 X of wild type) will be identified as active. Nevertheless, it is safe to assume that Ser-203 and His-447 of human AChE, in analogy to Ser-200 and His-440 of TcAChE (Gibney et al., 1990) are essential for catalysis and cannot be substituted by alanine residues. The conservation of His-432, on the other hand, seems to be related to requirement for proper folding and structure stabilization.
Asp-95, Asp-175, and Asp-404 were proposed in the past (Doctor et al., 1989;Soreq and Prody, 1989;Krejci et al., 1991) as candidate triad-carboxylates, based on their high conser- Two independent plasmid isolates, each used in two independent transfection assays were used for quantitation of each AChE type.
Amounts recovered from 1.5 X lo6 cells submitted to transfection. CAT activity was determined in cell lysates and AChE protein and activity in cell supernatant. Activity equivalent or lower than 1 milliunit/ml was considered background level (see "Experimental Procedures").
' AChE ELISA and enzymatic activity (kS.D.) were determined in 10-fold concentrated cell growth medium as described under "Experimental Procedures." Normalized to 1 nmol of CAT.
The apparent values obtained for specific activity in these AChE mutants are derived from background activity detectable in concentrated mock-transfected 293 cells.
To preserve maximal steric resemblance, the carboxylic residues were substituted by their corresponding amides (i.e. D to N and E to Q). The mutated cDNAs were introduced into 293 cells and AChE secretion was monitored by both ELISA and enzymatic activity (Table 111).
Three of the mutated HuAChEs, E84Q, D333N, and D349N, which are located at positions of less stringent conservation, were expressed at levels comparable with wild type HuAChE and yielded fully active HuAChE polypeptides. In contrast, mutagenesis at positions where evolutionary conservation is more stringent (Asp-95, Asp-175, Asp-404) had a far more significant effect on AChE production. Substitution of Asp-95 resulted in 90% decrease in AChE productivity, although enzyme molecules retained wild type specific activity. Substitutions of the aspartates at positions 175 and 404 resulted in loss of detectable enzymatic activity and AChEprotein secretion. Both our standard antigen-capture ELISA (directed mainly to native enzyme) and immunoblots, developed with antibodies to linear AChE epitopes (bacteria-derived antigens; Velan et al., 1991b) failed to detect AChE from cells transfected with the D175N and the D404N mutants.
The failure to detect these secretable mutant AChE products led us to examine possible intracellular accumulation of the enzyme. Low levels of intracellular activity were associated with the D404N mutant (8 milliunits/transfection over 3 milliunits of background, compared to 20 milliunits/transfection in wild type) but not with the D175N mutant. This result suggests that, although Asp-404 is not essential for catalysis, substitution at this position, and also at Asp-175, probably results in aberrant folding, which in turn may lead to intracellular degradation of the mutated polypeptide (Hurtly and Helenius, 1989). Intracellular events may also contribute to the low productivity of the D95N mutant. Indeed, 25% of the expressed D95N AChE activity is retained ' Sensitivity limit to AChE activity (1 milliunit/ml) determination pressed CAT activity and then averaged as described in Table 11. is set by the background values in mock transfections.
in the cellular fraction, compared with only 3-6% in wild type rHuAChE.
Mutations at positions 74 and 334 appear to affect catalytic activity.
Substitution of the Asp-74 by Asn did not impair AChEprotein production, yet the mutated polypeptide exhibited a decrease in specific activity (Table 111). This decrease need not be an indication of a true catalytic defect and may result from an altered enzymatic profile of the HuAChE (see below).
The E334Q mutation did not affect levels of AChE-protein production (Table 111) but abolished all detectable enzymatic activity (secreted as well as intracellular). Radiometric AChE assays, using the natural AChE substrate acetylcholine (as opposed to acetylthiocholine in the standard assay) also failed to detect any enzyme activity. It appears therefore, that of all eight conserved HuAChE carboxylates examined, only Glu at position 334 may be clearly defined as essential for catalytic activity.
To further substantiate the involvement of Glu-334 in HuAChE catalysis, an additional AChE mutant (E334A) was generated. Once again, enzymatic assays could not detect esterase activity, whereas efficient secretion of AChE-protein was revealed by ELISA (Table 111). The high levels of E334A polypeptide secretion allowed calculation of an upper level value for specific activity of c0.002 milliunit/ng. The effective interaction of the E334A AChE polypeptide with the array of four monoclonal antibodies (Table IV) indicates that, as in the S203A and H447A AChE mutants, loss of enzyme activity is not a consequence of major conformational distortions.
In summary, the observed resemblance in phenotype of E334A AChE to S203A and H447A AChEs (compare Tables  I1 and I11 with IV), is consistent with the TcAChE x-ray structure, suggesting that these three amino acids constitute a functional AChE catalytic triad.
Enzyme Engineering by Mutagenesis at Positions Asp-74, Ser-203, and Glu-334"Sequence alignment of cholinesterase superfamily members (Krejci et al., 1991) indicates a requirement for a carboxylate at a position analogous to 334 of HuAChE. This carboxylate is most frequently a glutamate (cholinesterases, rat carboxylesterase, microsomal esterases, and in Geotrichum lipase) but can also be an aspartate. Aspartate exists in non-catalytic (thyroglobulin and glutactin) but also in catalytic proteins (lysophospholipase and Drosophila esterase 6 and esterase P). To test whether substitution of Glu to Asp at position 334 would result in active HuAChE, the relevant mutant (Table I) was engineered and subjected to analysis. The E334D HuAChE mutant was detected (Table  111) in the cell supernatant by ELISA (10% of wild type), but lacked enzymatic activity. Specific activity calculations indicate that Asp-334 is at least 3 x 102-fold less effective than Glu in the AChE triad (Table 111).
In another attempt to engineer a modified hydrolase triad a Concentration of each AChE type was 40 ng/ml. ELISA was performed as described under "Experimental Procedures." In all assays the first antibody layer consisted of a rabbit polyclonal hyperimmune serum, and the second antibody consisted of mouse polyclonals or of the indicated monoclonal antibodies. into the AChE backbone the nucleophilic hydroxyl was replaced by a sulfhydryl group by generating a S203C HuAChE mutant (Table I). The resulting S203C mutant was produced and secreted (60 ng/ml AChE-protein, which is 15% of wild type) but did not manifest activity above background (1 milliunit/ml), thus placing an upper limit of the hydrolysis of acetylthiocholine at 1/350 of wild type AChE. Interestingly, attempts to replace Ser-203 by Thr which could also provide a nucleophilic hydroxyl led to decreased AChE-polypeptide production (5% of wild type) with no detectable activity.
A natural mutation of Asp-70 (McGuire et al., 1989;Neville et al., 1990) in human BuChE (analogous to Asp-74 in Hu-AChE) was implicated in the "atypical" phenotype (D70G). The atypical variant is characterized by reduced affinity to charged ligands, leading to the assumption that this aspartate resides within the anionic subsite. We have, therefore, generated a D74G HuAChE in addition to the D74N HuAChE (Table 11) and examined the two mutants in representative atypical assays: resistance to succinylcholine and resistance to dibucaine (La Du et al., 1990). Both the D74N and the D74G HuAChE mutations confer loss of inhibition by succinylcholine and a marked decrease in inhibition by dibucaine (Fig. 2). It appears therefore that the atypical phenotype can be "recreated" in an AChE, suggesting that, in both HuBuChE and HuAChE, Asp-70/74 assumes an analogous role in the configuration of the enzyme.

DISCUSSION
Compilation of data on enzymatic activity, physical quantitation and relative productivity of the various HuAChE mutants (Tables 11-IV) allowed differentiation between activity-related and production/folding-related effects. Assessment of productivity and structural integrity of the mutated AChE polypeptide relied on quantitative immunological assays based on polyclonal antibodies to native HuAChE, monoclonals to conformation-dependent epitopes, and antibodies to denatured HuAChE epitopes. All values were normalized by quantitation of a coexpressed cat reporter gene. This type of analysis reveals (Fig. 3) four major categories of AChE mutations: A , mutations that affect neither production levels nor enzymatic activity; E , mutations that cause significant reduction in production ( B 1 ) without affecting activity of residual polypeptide, or those that cause complete loss ( E 2 ) of AChE protein production; C, mutations that allow normal I . , .   Tables I1 and 111). Mutants are gathered into four major groups according to phenotype, as discussed in text. Note that no detectable secreted AChE-protein is associated with the D175N and D404N mutations; AChE activity above background was not detected in the D175N, D404N, E334Q, E334A, S203A, and H447A mutants. synthesis of AChE polypeptide but abolish catalytic activity; D , mutations that affect but do not abolish catalytic activity, and do not impair productivity.
Type A mutations include replacement of Glu-84, Asp-333, Asp-349, which are conserved in all sequenced cholinesterases (Gentry and Doctor, 1991), with their corresponding amides. The HuAChE three-dimensional model based on the TcAChE structure indicates that residues Glu-84 and Asp-349 are positioned on the outer surface of the TcAChE molecule. It is not surprising, therefore, that mutation of these residues does not affect catalytic activity, and one will have to seek for other reasons for their conservation. One may argue that conservation of these residues is due to their involvement either in quaternary structure determination or in subtle modulations of catalytic activity. However, oligomerization patterns as determined by sucrose gradients, as well as reactivity with a variety of substrates and inhibitors (data not shown), did not indicate any functional change, in either the E84Q or the D349N mutants. Modification of charge on the side chain of Asp-333 (Asp to Asn), an amino acid adjacent to the active-site Glu-334, would be expected to affect the catalytic profile. The failure to do so is, nevertheless, consistent with the HuAChE three-dimensional model, indicating that the charged moiety of Asp-333 is pointed away from the catalytic triad. Thus, the role of the charged residues at position 84, 333, and 349, which are conserved in cholinesterase, yet refractive to modification, remains to be determined.
The common feature of type B mutations is the "destabilizing effect" on the molecule. This phenomenon is most pronounced in mutants of residues Asp-175 and Asp-404 which appear to abrogate production of the rHuAChE polypeptide. The high resolution structural analysis of TcAChE provides some insight into the molecular events that could lead to these destabilizing effects. Asp-404 (HuAChE numbering) can be implicated in a salt bridge with Arg-525 and in a hydrogen bond with Tyr-382 (Fig. 4). These interactions may be crucial for stabilizing the interaction between the three helices (helix-aF, helix-aH, and helix-aF'3; Sussman et al., 1991) and could play a role in bringing together Cys-409 (helix-aF) and Cys-529 (helix-aH) which form a disulfide bridge (Fig. 4). Both Arg-525 and Tyr-382 are completely conserved among the cholinesterases. A similar type of structural requirement might also explain the conservation of Asp-175 and the effect of its substitution on production. Asp-175 can form a salt bridge with Arg-152, bringing together an a-helix (aC) and a @-strand (04) as depicted in Fig. 5. Both Asp-175 and Arg-152 are conserved not only among cholinesterases but also in every sequenced member of the cholinesterase superfamily (Krejci et aL, 1991). Salt bridges analogous to those involving Asp-175 and Asp-404 in HuAChE were also revealed by x-ray structure of the distantly related enzyme Geotrichum lipase.' M. Cygler, personal communication. This provides additional support for the structural importance of the ion pairs discussed above.
Asp-95 is strictly conserved in the small cysteine loop of cholinesterase superfamily polypeptides, but as shown here, it can be replaced by an Asn without affecting activity. The analogous TcAChE residue is located on the surface of the molecule, -20 A away from the catalytic triad. A straightforward explanation for the low production of the Asp-95 substitution mutant is not readily provided by the x-ray structure. It is worth noting, however, that to date the D95N substitution is the only HuAChE mutation which affected the ratio of secreted/intracellular enzyme. The amount of secreted D95N AChE activity is only 3-fold higher than that retained Involved in Activity and Folding c 17647 in the cell. In wild type rHuAChE as well as in other mutants analyzed here (E84Q, D333N, and D349N, as well as the production impaired H432A mutant) the amount of secreted AChE is about 15-30-fold higher than the amount of cellassociated activity. One is tempted, therefore, to assign to the Asp-95 residue a protein traffic-related function. Nevertheless, functional interpretation of type B mutations that lead to partial instability (such as D95N and H432A) should be considered with care. There is always the possibility that, in a given position, certain side chains substitution can be better tolerated then others (cf. the difference in productivity between E334A and E334D, Table 111). Of over 20 HuAChE positions mutated to date only Ser-203, Glu-334, and His-447 display the "C phenotype": effective production of a folded polypeptide devoid of catalytic activity. The involvement of two of these residues, Ser-203 and His-447, in catalysis is in accordance with previous mutagenesis studies in TcAChE (Gibney et al., 1990). Mutations a t position Glu-334 provide the first biochemical confirmation of the xray structural models that infer a Glu residue in the catalytic triad of hydrolases (Sussman et al., 1991;Schrag et al., 1991). It is of interest that alanine substitutions of any of the HuAChE-triad residues are tolerated by the molecule with no major effect on folding or secretion. This seems to be in accordance with the overall fold of the a/P hydrolase enzyme group to which the cholinesterases belong (Ollis et al., 1992). The catalytic residues of these enzymes are protruding from turns located on three different loops brought together to form the catalytic triad.
These observations, which suggest that triad positions may eventually accommodate different amino acids, led us to generate various catalytic triad permutations in the HuAChE mold. Engineering of triads such as Cys-His-Glu or Ser-His-Asp resulted in reduced productivity and in non-detectable (less than 1% of wild type) cholinesterase activity. Substitution of active-site Ser by Cys in TcAChE (Gibney et al., 1990) was reported previously to yield AChE with detectable activity, yet the catalytic efficiency of the mutant was -100-fold lower than that of wild type. In contrast, replacement by Ser of the Cys residue which serves as nucleophile in the dienelactone hydrolase triad (Pathak and Ollis, 1990) did not lead to activity loss beyond 10% of wild type (Pathak et al., 1991). This could indicate a shift toward a catalytic mechanism of higher stringency during evolution of the cholinesterases, which is not observed in the dienelactone hydrolase of Pseudomonas.
Type D mutants, having an altered AChE activity, are represented by D74N and D74G AChEs. Substitution of Asp-74 by either Asn or Gly created an AChE with the phenotype of the atypical BuChE, rendering the rHuAChE less sensitive to inhibition by succinylcholine and dibucaine and thus suggesting a common role for Asp-70/74 in both enzymes. It would be tempting to speculate that in AChE, as previously suggested for BuChE, this Asp residue constitutes a component of the anionic subsite involved in choline binding ((Neville et al., 1990). However, the analogous Asp in TcAChE appears to be located at the entrance of the aromatic gorge (Sussman et al., 1991), with the carboxyl group protruding into the gorge cavity. The distance between this residue ayd the active-site serine near the bottom of the gorge is -16 A, far greater than the expected 4.7-A distance between the esteratic and anionic subsites (Rosenberry, 1975;Berman and Decker, 1986). Kinetic studies, in progress, with other AChE mutants, as well as x-ray structure analysis of TcAChE/ inhibitor complexes, indeed, suggest allosteric interactions between residues at the gorge entrance and residues close to the catalytic center.