INTRODUCTION

Acetylcholinesterase (AChE,¹ EC) is a very efficient enzyme since it hydrolyzes acetylcholine nearly as fast as allowed by diffusion (Quinn, 1987). It plays a key role in the cholinergic system by rapidly inactivating acetylcholine after its release at the synapse. Cholinesterases are also present in noncholinergic contexts. For example, AChE and butyrylcholinesterase (BChE, EC) are expressed early during embryonic development, in somites and in other parts of the developing organism, well before the appearance of cholinergic synapses (Drews, 1975; Layer et al., 1985). In this case, AChE and BChE could be involved in the formation, or the regulation, of acetylcholine gradients that could guide the growth of nerve cells (Layer et al., 1988). AChE is also present at the surface of the red cells of some vertebrates or in soluble form in the plasma, together with BChE (Toutant and Massoulié, 1988). In blood, AChE and BChE would play a detoxification role (Neville et al., 1990). It is more surprising that the venom of some snakes, belonging to the family of Elapidae, contain a high level of AChE (Kumar and Elliott, 1973).² The role of AChE in such venoms is not known, since it is neither toxic by itself nor acting in a synergistic manner with the toxic components of the venom.³

In the AChE genes of vertebrates, Torpedo, and mammals, the signal peptide and the catalytic domain are encoded by common exons, followed by alternatively spliced sequences which encode distinct types of short COOH-terminal regions, characterizing R (``read-through''), H (``hydrophobic''), or T (``tailed'') subunits. These subunits possess identical catalytic activity, but generate different molecular forms. All forms of AChE described until now are oligomers of subunits H and T (for a review, see Massoulié et al. (1993)). The COOH-terminal region of H subunit contains a glycophosphatidylinositol (GPI) addition signal, preceded by one or two cysteines. H subunits generate GPI-linked dimers (amphiphilic forms of type I). These forms are expressed in Torpedo muscles and electric organs (Massoulié et al., 1992b), as well as in embryonic mammalian muscle (Legay et al., 1995) and in hematopoietic cells of adult mammals, but seem to be lacking in birds (Massoulié et al., 1993). The T subunits terminate with a highly conserved peptide, containing a cysteine near its COOH terminus. T subunits form dimers and tetramers, and associate with structural anchoring proteins, forming collagen-tailed or hydrophobic-tailed molecules. Probably because the T peptide contains an amphiphilic -helix, the monomers and dimers of the T subunits interact with detergent or phospholipid micelles (amphiphilic forms of type II) (Bon et al., 1988a, 1991). T subunits are generally expressed in muscles and in nervous tissue. Although the third type of subunit, R, is predicted to exist from cDNA sequences, the corresponding enzyme has not been identified in vivo. The R regions of Torpedo and mammals do not show any homology (Sikorav et al., 1988; Rachinsky et al., 1990). They represent a ``read-through'' of the 3 genomic region that immediately follows the last common catalytic exon, and is normally spliced as an intron during processing of H or T in mRNAs. In all cases where they have been found, R transcripts represent only a small proportion of AChE in mRNAs, e.g. in embryonic rat liver (Legay et al., 1993) or embryonic mouse diaphragm (Legay et al., 1995), and thus may result from a splicing defect without physiological significance.

In the venom of Bungarus fasciatus, as in other Elapidae venoms, AChE is present as a soluble, hydrophilic monomer.³ This type of AChE molecule has never been described in significant proportion in other vertebrate tissues in vivo.

Pharmacological studies have shown that AChE is inhibited by ligands that do not bind at the active site, but at a ``peripheral site'' (Changeux et al., 1966; Taylor and Lappi, 1975). Extensive studies of snake venom AChEs have shown that they differ widely in their sensitivity to peripheral site ligands.<sup>2</sup> Labeling and mutagenesis experiments have shown that the peripheral site is located at the mouth of the ``catalytic gorge'' (Sussman et al., 1991), about 20 Å away from the active site (Berman et al., 1980; Kreienkamp et al., 1991; Harel et al., 1993). In particular, it has been proposed that binding of propidium to Trp-279 and surrounding residues induces an allosteric conformational change, mediated through the catalytic gorge by residues Tyr-70, Asp-72, Tyr-121, Tyr-130, and Tyr-334, eventually resulting in a displacement of Trp-84 (Ordentlich et al., 1993, 1995; Barak et al., 1994), which is involved in the positioning of the substrate in the active site (Harel et al., 1993). The peripheral site seems to be quite variable among AChEs. For example, chicken AChE is insensitive to propidium, with less than 10% inhibition at a concentration of 2.5 mM, while IC₅₀ is 13 µM for Torpedo AChE in the same conditions (Eichler et al., 1994). This is probably related to the fact that two aromatic residues involved in the binding of propidium to Torpedo AChE (Tyr-70 and Trp-279) are replaced in chicken AChE by methionine and glycine, respectively (Randall et al., 1994). AChE from Naja naja oxiana venom has also been reported to lack a peripheral site, since propidium is a poor inhibitor, K_i being 10-fold higher than for Torpedo AChE and, moreover, acts in a purely competitive manner (Kreienkamp et al., 1991). It was thus interesting to analyze the sensitivity of the Bungarus venom enzyme to propidium and other peripheral site inhibitors.

In this report, we describe cDNA clones encoding AChE from B. fasciatus venom. We show that the COOH-terminal region of this enzyme is not homologous to either the H or the T peptides, and that it undergoes a proteolytic processing. We analyzed the relationship between these unusual features and the secretion of soluble monomers by expressing the wild type venom AChE and modified COOH-terminal variants in transfected COS cells. We also examined, by site-directed mutagenesis, the role of two residues that differ between Bungarus AChE and Torpedo or mammalian AChEs, in the inhibition by peripheral site ligands.

EXPERIMENTAL PROCEDURES

B. fasciatus venom (batch 6) was from the stock of Institut Pasteur. All salts and other reagents were from Sigma (France), Merck (Darmstadt, Germany), or Prolabo (Paris, France). PI-PLC from Bacillus thuringiensis was from Immunotech (Marseilles, France); neuraminidase from Clostridium perfringens was from Worthington (Coger, France), and carboxypeptidase Y from yeast were from Boehringer (Mannheim, Germany). Fasciculin 2 was a gift from Dr. Pierre Bougis (Marseilles, France). The anti-FLAG M2 monoclonal antibody, recognizing the DYKDDDDK peptide, was from Eastman Kodak (New Haven, CT). All oligonucleotides were from Genset (Paris, France).

AChE from B. fasciatus venom was purified by affinity chromatography according to Massoulié and Bon (1976). A sample containing 0.44 mg of AChE (6.4 nmol) was dialyzed against 50 mM sodium acetate, pH 6, then mixed with 2.5 nmol of norleucine, 0.03% SDS, and 4.4 µg of carboxypeptidase Y, in a total volume of 800 µl. The mixture was incubated at 25 °C; aliquots (80 µl) were removed after 0, 1, 2, 5, 10, 20, 60, and 90 min, and added to 40 µl of 1% trifluoroacetic acid, at 0 °C. Each aliquot was centrifuged through an Ultrafree-MC filter (Millipore) with a cut-off at 10 kDa; the filter was rinsed and recentrifuged with 50 µl of 0.1% trifluoroacetic acid. The filtrates were dried in a Speed-Vac. Amino acids were derivatized to PTC according to the recommendations of Applied Biosystems, and analyzed by high pressure liquid chromatography. Residues Asp, Ala, and Arg were successively removed from the COOH terminus.

B. fasciatus snakes were kindly provided by Prof. Xiong Yu Liang and Dr. Zhang Yun (Kunming Institute of Zoology, Academia Sinica, Kunming, Yunnan, China). The heads were immediately frozen and stored in dry ice. The venom glands were dissected, and total RNA was extracted using RNAsol (Bioprobe), according to Chomczynski et al. (1987).

Polymerase chain reaction (PCR) was performed using Taq polymerase (Life Technologies, Inc. or Promega), and 10 pmol of each oligonucleotide in a final volume of 50 µl. The reaction tubes were overlaid with mineral oil (Sigma). We used a PTC150 thermocycler (MJ Research). Alternatively, the ``hot-start'' method, in which Taq polymerase was added at 85 °C after a first denaturation step, was used for RT-PCR or RACE experiments. If necessary, the PCR products were cloned with the pGEM-T kit (Promega). Double strand sequencing was performed with the Sequenase 2.0 kit (U. S. Biochemical Corp.), following the supplier's protocol.

Degenerate oligonucleotides, deduced from known partial peptidic sequences,³ were used in RT-PCR experiments. Approximately 1 µg of total RNA was reverse-transcribed using 200 units of Superscript-MMLV reverse transcriptase (Life Technologies, Inc.). cDNA was then purified using the QIAquick-spin PCR purification kit (Qiagen). The eluate was diluted to 100 µl with water, and 1 µl was used for PCR, using a combination of degenerate primers. For a second PCR, we used a second set of internal primers and 1 µl of a 1/20 dilution of the first PCR product.

RACE experiments were done essentially as described by Frohman et al. (1988). For 3 RACE, the reverse transcription was performed as described above, using the oligonucleotide (dT)₁₇-R_i-R_o (Frohman et al., 1988). 1 µl of the eluate was used for the following PCR, using a specific primer and R_o. For a second PCR, we used nested specific and R_i primers and 1 µl of a 1/20 dilution of the first PCR product.

For 5 RACE, we used the single strand ligation method (Dumas et al., 1991). The single strand DNA was obtained as for 3 RACE, except that a biotinylated specific primer was used for reverse transcription. The product was purified with Dynabeads (Dynal) coated with streptavidin. 2.5 pmol of an oligonucleotide, complementary to R_i-R_o, which had been phosphorylated in 3 and blocked in 5 by addition of a ddTTP with terminal transferase (Pharmacia Biotech Inc.), was ligated to the DNA in T4 RNA ligase buffer, 25% polyethylene glycol 8000, with 20 units of T4 RNA ligase (New England Biolabs) in a final volume of 10 µl, incubated more than 20 h at 20-22 °C. The ligated DNA was purified using Dynabeads. 2 µl of the eluate were used for PCR, as described above.

Poly(A)⁺ RNA was prepared from total RNA with Oligotex-dT columns (Qiagen). 4 µg of these mRNAs were used with the SuperScript plasmid system (Life Technologies, Inc.) with some modifications: the oligonucleotide used for the ligation step was an EcoRI adaptor (Pharmacia), and the cloning vector was pT7T3D NotI/EcoRI/BAP (Pharmacia). The library was then transformed into highly competent Escherichia coli cells (XL1-blue-MRF, Stratagene). The transformed bacteria were then fractionated in 17 pools, each containing 10⁴ clones, which were plated on LB-ampicillin plates and grown overnight. LB medium (5 ml) was poured on the plates, and bacteria were recovered. The pools were screened by PCR between a primer located in the vector and a primer located in the 5 region of the cDNA. Six pools were positive, and we selected two of them that yielded the longest 5 amplified fragments. The corresponding clones were isolated by screening sub-pools of about 200 clones and, finally, plating and screening independent colonies.

6 µg of total RNA from Bungarus venom glands were loaded on a formaldehyde-agarose gel (1.5%) gel and electrophoresed during 5 h under 100 V. The RNAs were transferred overnight to a nylon membrane (Hybond, Amersham) using 10 × SSC buffer and were fixed during 5 min by UV irradiation. The membranes were prehybridized for 1 h at 42 °C in a medium containing 50% formamide, 5 × SSC, 0.1% SDS, 2 × Denhardt's buffer. The probe was prepared with 30 ng of purified PCR product, using the Hexaprime kit (Boehringer, Mannheim, Germany) with 50 µCi of [³²P]dCTP (ICN), and purified in a S400 spin column (Pharmacia). The denatured probe and salmon sperm DNA (100 µg/ml) were added to the prehydridization solution and incubated overnight at 42 °C. The blots were first washed for 5 min in 4 × SSC at room temperature, then once in 0.5 × SSC, 0.1% SDS at 65 °C and autoradiographed overnight between two intensifying screens.

Modifications of the COOH-terminal region were performed on the wild type SARA construction, and point mutations in the catalytic domain were realized on the NAT variant (see Fig. 5). Introduction of a stop codon for truncated constructions, or insertion of a ``flag'' peptidic epitope at the COOH terminus were performed with a reverse mutagenic oligonucleotide. Two complementary mutagenic oligonucleotides were used for each point mutation. We used the cDNA library clone as a template for PCR reactions (Landt et al., 1990). The PCR products were cloned in pGEM-T and fully sequenced. The plasmid was digested with appropriate restriction enzymes (EcoRI-BstXI for the M70Y mutant, BstXI-NotI for the K285D mutant, and BglII-NotI for the COOH terminus variants), and the insert was reintroduced into the original plasmid.

For expression in COS cells, the inserts were excised from the pT7T3D vector using EcoRI and NotI (Pharmacia) and cloned in the pCDNA3 vector (Invitrogen). Plasmidic DNA was prepared with the Plasmid midi-kit (Qiagen). COS cells were generally transfected with 5-10 µg of DNA, using the DEAE-dextran method (Duval et al., 1992). Cells were grown at 37 °C. In some experiments, as indicated, the cells were transferred to 27 °C after 24 h at 37 °C.

AChE was analyzed by sedimentation in gradients containing 5% to 20% sucrose in 10 mM Tris-HCl, pH 7.0, 5 mM MgCl₂, either without detergent or in the presence of 1% Triton X-100 or Brij-96. E. coli alkaline phosphatase (6.1 S) and E. coli -galactosidase (16 S) were mixed with the sample, as standards of sedimentation coefficients. After centrifugation at 36,000 revolutions/minute for 18 h at 7 °C in a Beckman SW41 rotor, about 45 fractions were collected from the bottom of the tubes and assayed for the different enzymatic activities.

Electrophoresis in nondenaturing conditions, in horizontal 10% polyacrylamide gels, was performed as described previously (Duval et al., 1992), and AChE activity was revealed by the method of Karnovsky and Roots (1964).

COS cells culture medium containing about 5 × 10³ Ellman units of Bungarus AChE was diluted in 10 µl of 12.5 mM sodium acetate buffer, pH 6.5, CaCl₂ 5 mM, MgCl₂ 5 mM, containing 0.05 unit/ml neuraminidase and incubated overnight at 37 °C. This incubation did not modify the AChE activity. Treatment with PI-PLC was performed as described previously (Duval et al., 1992).

For the kinetic experiments, we used the culture medium containing the secreted AChE, without further purification. As a control, we used PI-PLC-treated dimeric AChE from Torpedo marmorata electric organs. AChE activity was assayed in a reaction medium containing 50 mM sodium phosphate, pH 7.4, 0.5 mM 5,5-dithiobis(2-nitrobenzoic acid), 0.01% bovine serum albumin, and various concentrations of acetylthiocholine (0.01-10 mM), at 25 °C. The reaction was monitored photometrically at 405 nm. K_m and K_ss values were determined using the Haldane equation. Alternatively, K_m was determined from Lineweaver-Burk plots, with 0.06-0.7 mM acetylthiocholine, and K_ss was determined by plotting of 1/V as a function of substrate concentration (2-10 mM). The equation fitting and graphic methods yielded the same results.

The inhibition parameters were determined as above, after incubation of the enzyme with the various inhibitors at 25 °C during 20-40 min (the incubation time had no effect, since inhibition was very rapid). For the determination of inhibition constants, K_i and K_i, as defined in Scheme I, we used five concentrations of acetylthiocholine (0.06-0.7 mM) and two or three concentrations of inhibitor. For inhibition by fasciculin, we first incubated approximately 10 pM AChE with fasciculin 2 overnight at 4 °C in a buffer containing 5 mM sodium phosphate, pH 7.4, and 0.1% bovine serum albumin, in a final volume of 900 µl. The samples were then equilibrated at 25 °C for 30 min and 100 µl of a 10-fold concentrated Ellman assay medium was added, to a final concentration of 50 mM sodium phosphate buffer, pH 7.4, 0.5 mM 5,5-dithiobis(2-nitrobenzoic acid), and acetylthiocholine (0.06-0.7 mM). The competitive inhibition constant, K_i, was calculated by plotting the relative slope defined by the Lineweaver-Burk representation (with inhibitor versus without inhibitor) as a function of inhibitor concentration; this yielded a straight line with a slope of 1/K_i. In a similar manner, the noncompetitive inhibition constant, K_i, was the inverse of the slope obtained by replotting the relative intercept (1/V_max), from the Lineweaver-Burk plots, as a function of inhibitor concentration.

For determination of the apparent first order rate constant (k_cat) by titration of active sites, we incubated the enzyme overnight at 4 °C with various concentrations of O-ethyl-S(2)-diisopropylaminoethyl methylphosphonothionate (Vigny et al., 1978). The remaining activity was then assayed as described above, with 0.8 mM acetylthiocholine.

We modeled the B. fasciatus venom enzyme, using the first approach protocol proposed by Swiss-Model (Peitsch, 1995), with the structure of Torpedo californica AChE as template (1ace.pdb; Sussman et al., 1991). The model obtained was then displayed using the RASMOL program.

RESULTS

Primary Sequence of Bungarus Venom AChE

As will be described in another report,³ we obtained partial peptidic sequences from tryptic fragments of Bungarus venom AChE, which could be unambiguously aligned with the sequence of Torpedo AChE.⁴ Using RT-PCR procedures with degenerate oligonucleotide probes based on these sequences, we obtained a cDNA fragment of 520 bp. Using RACE methods, from this fragment, we obtained the complete coding sequence, as well as the 3-untranslated sequence and a short 5 noncoding region. Because this sequence might contain errors introduced by Taq polymerase during the PCR steps, we screened a cDNA library for a complete clone. The two longest clones (2 kb) contained the same complete coding sequence, and the same 3-untranslated sequence, as shown in Fig. 1. The open reading frame containing the coding sequence is preceded by a 203-bp 5-untranslated region and followed by a 64-bp 3-untranslated region, which contains a polyadenylation site and terminates with a poly(A) tail. The cDNA clones are incomplete in their 5-untranslated region, as shown by the size of mRNAs (see below; Fig. 4). The open reading frame starts with two ATG codons, separated by 7 codons. Fig. 2 shows an alignment of the coding region with that of T. marmorata. The predicted primary sequence contains a signal peptide with two potential cleavage consensus sites at positions 28-29 and 34-35, both very close to the known NH₂ termini of the mature Torpedo, bovine, and human AChEs.

The catalytic domain of Bungarus AChE shows more than 60% identity and 80% similarity with that of Torpedo AChE (Fig. 2). It contains four potential N-glycosylation sites, all of which correspond to glycosylated positions in Torpedo or mammalian AChEs. The six cysteines that form intramolecular disulfide loops in all cholinesterases are conserved, and there is no other cysteine in the sequence of the mature protein. The three residues that compose the catalytic triad (Ser-200, Glu-327, and His-440, in Torpedo) and the tryptophan residue that binds the quaternary ammonium group of acetylcholine in the active site (Trp-84; Weise et al., 1990) are all present. In addition, 13 of the 14 aromatic residues that line the wall of the active gorge of Torpedo AChE (Sussman et al., 1991) are conserved in the Bungarus enzyme. The only exception is tyrosine 70, which is replaced by a methionine; Tyr-70 belongs to the peripheral anionic site, located at the rim of the active site gorge, together with Asp-72, Tyr-121, Tyr-130, Trp-279, and Tyr-334, which are all conserved.

The Bungarus AChE is markedly less inhibited by peripheral site ligands than Torpedo or mammalian AChEs.² We therefore examined whether significant differences in aromatic or charged residues would appear in the peripheral site. For this purpose, we used a three-dimensional model of the Bungarus venom enzyme, based on the structure determined for Torpedo by Sussman et al. (1991). Fig. 3 shows that an aspartic residue of Torpedo AChE, Asp-285, is replaced by a lysine in Bungarus AChE. Thus, the peripheral site of Bungarus AChE mostly differs from that of Torpedo by the absence of an aromatic residue at position 70, and by the replacement of Asp-285 by a positively charged residue.

In contrast with the catalytic domain, the COOH-terminal sequence of Bungarus venom AChE shows no homology with either the H or T COOH-terminal sequences of AChE from Torpedo or other vertebrate species. In order to examine the possible existence of other AChE transcripts in the Bungarus venom gland, we hybridized Northern blots with a 600-bp probe located in the catalytic domain, and with a 120-bp probe corresponding to the 3 divergent region. Both probes hybridized with the same band, at 3.5 kb (Fig. 4). It thus appears that in the venom gland, AChE transcripts are processed through a single splicing mode.

The COOH-terminal region deduced from the coding sequence (Fig. 1) is a short hydrophilic peptide of 15 residues containing two aspartic acids, six arginines, and no cysteine.

The last three COOH-terminal residues of the mature protein purified from Bungarus venom, RAD, are located near the middle of the predicted COOH-terminal peptide. This shows that a highly charged peptide of eight residues, including five arginines, is cleaved either during biosynthesis in the venom gland, in the venom itself in vivo, or during storage of the dry venom.

Expression in COS Cells: Production of Nonamphiphilic Monomers

Several variants, differing in the length of the COOH-terminal peptide, were produced in transfected COS cells. These variants are named after their last COOH-terminal amino acids (Fig. 5). In SARA-flag, we introduced a ``flag'' epitope after the SARA COOH terminus.

We determined the activity obtained in cells that were maintained at 37 °C, and in cells that were transferred to 27 °C after recovering from transfection (24 h at 37 °C), for the NAT, RAD, and SARA variants. For comparison, we analyzed r-SAT, a rat equivalent of NAT that is limited to the catalytic domain and produces a soluble monomeric form of rat AChE.⁵ All constructions produced active AChE. The activity was similar for NAT, RAD, and SARA: about 0.1 mg of active enzyme/culture dish (10⁶ cells), after 9 days at 37 °C; this activity was more than 30 times higher than for r-SAT. Both Bungarus and rat enzymes were secreted (about 95% of the total), and the level of activity was more than 3 times higher at 37 °C than at 27 °C.

The cellular and secreted recombinant NAT, RAD, and SARA enzymes, like natural venom AChE, sedimented at 4.5 S, as single homogeneous molecular species. This sedimentation was identical without detergent or in the presence of Triton X-100 or Brij-96 (Fig. 6), showing that these monomers are nonamphiphilic.

The migration in nondenaturing polyacrylamide gels of the different COOH-terminal variants was not affected by the neutral detergent Triton X-100, or by the anionic detergent sodium deoxycholate (results not shown), confirming their nonamphiphilic character.

As in the case of the venom AChE,³ the three recombinant enzymes NAT, RAD, and SARA produced electrophoretic ladder patterns with 5-6 distinct bands, although it was difficult to determine their exact number because their intensity decreased from the center toward the edges of the pattern. We observed such patterns both in the cellular (Fig. 7A) and secreted fractions (Fig. 7B). The SARA cellular extract presented an additional slower band, which will be discussed in the next section. The distances between NAT and RAD enzymes, which differ by one charge unit, and between K285D and NAT, which differ by two charges, suggest that the ladder pattern results from increments of a single charge.

Treatment with neuraminidase essentially reduced the pattern to the slowest two bands (Fig. 8), indicating that the multiplicity of bands may correspond to a variation in the number of sialic acids. In addition, the secreted SARA enzyme showed a higher proportion of the slower migrating components when the enzyme was synthesized at 27 °C rather than at 37 °C (Fig. 8), probably because sialylation is less efficient at a lower temperature.

The nonamphiphilic, monomeric form of rat AChE, r-SAT, does not show such a ladder pattern, although its structure and macromolecular properties are equivalent to those of the Bungarus NAT enzyme.

In the case of NAT and RAD, the secreted and cellular enzymes presented a similar migration (Fig. 7). The secreted SARA enzyme migrated exactly like RAD (Fig. 7B), while a fraction of the cellular enzyme migrated much more slowly than NAT or RAD (Fig. 7A). The same observation was made after treatment by neuraminidase (data not shown). The slowly migrating component probably corresponds to the complete SARA protein, in which the negative charge is reduced by the presence of five COOH-terminal arginines. Cleavage of this peptide would then occur upon secretion, generating an enzyme like the RAD variant or the natural venom enzyme.

In COS cells, the SARA-flag construction yielded a similar level of AChE activity to the other COOH-terminal variants. A fraction of the cellular enzyme was recognized by the anti-flag M2 monoclonal antibody, as shown by a sedimentation shift from 4.5 S to 7.8 S in sucrose gradients (Fig. 9) and by retardation of the active enzyme in nondenaturing electrophoresis (data not shown). The sedimentation of the secreted enzyme recovered from the culture medium was unaffected by the antibody, showing that it had lost the flag sequence. Cleavage of the hydrophilic COOH-terminal peptide therefore occurs before secretion, both in the COS cells and in the venom glands.

The Role of Residues 70 and 285 in the Peripheral Site

We constructed point mutants of the NAT variant of Bungarus AChE, containing one or both of the Torpedo residues: M70Y, K285D, and M70Y/K285D. The three mutants yielded the same level of activity secreted in the culture medium as the recombinant wild type enzyme. The wild type and mutated enzymes showed a similar excess substrate inhibition (Fig. 10). The K_m and K_ss values are reported in Table I. Compared to the wild type enzyme, K_m was slightly increased in the case of M70Y and slightly decreased in the case of K285D. The K_ss values were not markedly altered by the mutations: the largest difference, observed in the case of the double mutant, was less than 30% (Table I). In addition, titration of the active sites showed that the four enzymes possessed the same catalytic turnover number, k_cat (not shown).

Table I. Kinetic parameters Km and Kss are determined as specified in under ``Experimental Procedures.'' WT, wild type. Enzyme Km Kss µM mM WT 78.8  ± 6.3 35.8  ± 3.7 M70Y 108.6  ± 4.5 29.3  ± 1.9 K285D 58.7  ± 4.1 30.6  ± 3.2 M70Y/K285D 95.3  ± 5.9 25.5  ± 0.7 Torpedo 81.2  ± 4.2 22.0  ± 1.5

We analyzed the inhibition of these enzymes, in parallel with Torpedo AChE, by the active site ligand edrophonium, by the bis-quaternary ligands decamethonium and BW284C51, which bind at both active and peripheral sites (Berman et al., 1980), and by the peripheral site ligands propidium (Taylor and Lappi, 1975), gallamine, D-tubocurarine (Changeux, 1966), and fasciculin 2, a toxin from the venom of an Elapidae snake, Dendroaspis angusticeps (Cerveñansky et al., 1991). From Lineweaver-Burk representations (Fig. 11), we determined apparent competitive and noncompetitive inhibition constants, K_i and K_i, as defined in Scheme I (Table II).

Table II. Inhibition constants for reversible inhibition of wild type and peripheral site mutants of AChE Constants are in nM for BW284C51 and fasciculin 2, and in µM for decamethonium, edrophonium, propidium, gallamine, and D-tubocurarine. Ki is the competitive constant and Ki the non competitive constant according to Scheme I. WT, wild type. Bis-quaternary and active site inhibitors Enzyme Decamethonium BW284C51 Edrophonium Ki  Ki Ki  Ki Ki  Ki WT 8.2  ± 1.2 47.5  ± 3.5 16.9  ± 1.6 28.9  ± 0.8 0.6  ± 0.0 9.9  ± 4.2 M70Y 3.8  ± 0.4 10.2  ± 2.5 7.9  ± 1.3 17.5  ± 1.4 0.6  ± 0.0 9.6  ± 0.6 K285D 2.3  ± 0.3 14.1  ± 3.9 10.8  ± 2.4 14.4  ± 0.8 0.6  ± 0.0 10.4  ± 2.5 M70Y/K285D 1.3  ± 0.1 4.4  ± 0.7 3.8  ± 0.2 9.3  ± 0.1 0.6  ± 0.0 7.8  ± 2.0 Torpedo 1.3  ± 0.2 7.0  ± 1.3 15.1  ± 0.7 63.6  ± 8.4 0.3  ± 0.0   Peripheral site inhibitors Enzyme Propidium Gallamine D-tubocurarine fasciculin 2  Ki  Ki Ki Ki  Ki  Ki WT 46.0  ± 3.6 250.0  ± 60.0 512.0  ± 58.7 39.1  ± 2.8 66.4  ± 11.5 50.7  ± 2.6 M70Y 1.6  ± 0.0 2.3  ± 0.4 222.0  ± 23.8 92.4  ± 13.2 266.0  ± 86.0 5.6  ± 1.3 K285D 4.6  ± 0.8 27.9  ± 5.0 26.1  ± 2.9 8.6  ± 1.5 17.7  ± 3.3 3.6  ± 1.3 M70Y/K285D 0.6  ± 0.1 1.0  ± 0.2 8.0  ± 1.1 11.8  ± 1.6 26.5  ± 2.7 0.3  ± 0.2 Torpedo 2.0  ± 0.2 3.6  ± 0.4 51.6  ± 6.4 57.7  ± 15.2 208.0  ± 59.4 0.14  ± 0.04

Edrophonium acts as a competitive inhibitor, with similar K_i values for the wild type and modified Bungarus AChEs, as well as for Torpedo AChE.

The bis-quaternary inhibitors, decamethonium and BW284C51, led to a mixed type of inhibition. The Bungarus enzyme was somewhat less sensitive to decamethonium than Torpedo AChE and slightly more sensitive to BW284C51. Both the M70Y and K285D mutations increased the sensitivity of Bungarus AChE to these inhibitors, and the two mutations had a cumulative effect, affecting both K_i and K_i.

The wild type Bungarus AChE was more sensitive than Torpedo AChE to D-tubocurarine. The M70Y mutation decreased the sensitivity, mimicking the situation observed in Torpedo AChE. The K285D mutation increased it, and its effect was predominant in the double mutant, M70Y/K285D.

The Bungarus enzyme was markedly less sensitive than Torpedo AChE to the other peripheral site ligands, propidium, gallamine, and fasciculin. The individual mutations, M70Y and K285D, increased its sensitivity to these inhibitors, and the double mutant was more sensitive than Torpedo AChE to propidium and gallamine, and approximately equally sensitive to fasciculin. The two mutants, however, did not display the same efficiency for the different ligands; while K285D had a much greater effect than M70Y in the case of gallamine, the reverse was observed in the case of propidium, and their effects were equivalent in the case of fasciculin. Also, inhibition was of the competitive type for gallamine, of the mixed type for D-tubocurarine, and noncompetitive for fasciculin. In the case of propidium, the wild type and K285D enzymes were inhibited in a mixed manner, while inhibitions of M70Y and M70Y/K285D were essentially noncompetitive (Fig. 11). Comparing the inhibitions of the wild type, M70Y, K285D, and M70Y/K285D enzymes by propidium, the noncompetitive pattern appeared correlated with a high affinity of the ligand for the peripheral site.

DISCUSSION

Primary Structure of the AChE from Bungarus Venom

The venom glands of the Elapidae snake B. fasciatus abundantly secrete a soluble, monomeric form of AChE. From this tissue, we obtained several cDNA clones that possess the same coding sequence and the same short 3-untranslated region. The venom glands contain a single type of mRNA encoding AChE, with a length of about 3.5 kb.

The coding sequence possesses two in-frame methionine codons, separated by seven codons. The deduced primary sequence shows that the catalytic domain is entirely homologous to those of AChEs from other species. Conserved residues, in the active site, include the catalytic triad, Trp-84, as well as Phe-288 and Phe-290, which form an acyl pocket, defining the specificity of the enzyme for acetyl esters (Harel et al., 1992). The six cysteines that form disulfide loops in all cholinesterases are also conserved. While Torpedo AChE possesses a free cysteine (Cys-231), Bungarus AChE, like the mammalian enzymes, does not contain any additional cysteine in the catalytic domain. The Bungarus sequence contains four potential N-glycosylation sites, at positions that are glycosylated in Torpedo or in mammalian AChEs.

The cDNA sequence obtained from Bungarus venom glands clearly diverges from either H or T sequences, downstream from the common catalytic exons. The absence of hydrophobic elements and the lack of a cysteine that could form intersubunit disulfide bonds are entirely consistent with the fact that the venom enzyme is a soluble monomer.³

Processing and Secretion of Monomeric Bungarus AChE

The level of AChE activity was comparable for COOH-terminal variants of Bungarus AChE, NAT, RAD, and SARA (Fig. 5). In all cases, the transfected COS cells secreted hydrophilic monomers sedimenting as a monodisperse peak at 4.5 S. Whereas Torpedo AChE is obtained in an active form only if the cells are grown at a low temperature, e.g. 27 °C, Bungarus AChE was markedly more efficiently produced at 37 °C than at 27 °C. Bungarus AChE therefore folds into its active conformation at both temperatures, and secretion was not dramatically impaired at 27 °C.

The production of Bungarus AChE was more than 30-fold higher than that of an equivalent monomeric soluble rat enzyme, r-SAT, although the snake protein would appear more foreign to the mammalian COS cells. The reason for this large difference in biosynthetic rate might reflect intrinsic thermodynamic properties of the polypeptidic chains, since Bungarus AChE renatures more readily into its active conformation than Torpedo AChE.⁶

Like the venom AChE,³ the different COOH-terminal variants of Bungarus AChE that we expressed in COS cells presented a remarkable heterogeneity in nondenaturing electrophoresis, producing a ladder pattern of 5-6 equally spaced bands. Neuraminidase treatment reduced the electrophoretic mobility and the number of bands, indicating that sialic acids play an important role in this microheterogeneity. Hayes and Wellner (1969) observed a similar pattern in the case of another abundant enzyme of snake venom, L-amino acid oxidase, presumably resulting from variation in the carbohydrate content of the protein.

We did not observe such a ladder pattern in the case of a monomeric, secreted form of rat AChE, obtained by transfecting COS cells with a truncated subunit, r-SAT, which is equivalent to NAT and contains three N-glycosylation sites, two of which are conserved in the Bungarus sequence. Similarly, the natural monomeric G₁ form, composed of a single T subunit, which is abundant in some mammalian tissues such as embryonic brain, was never found to produce a ladder pattern in nondenaturing electrophoresis. Therefore, this property appears to singularize the Bungarus enzyme. It might be related to the high level of expression of this enzyme, if the rates of translation, folding, and transport of the Bungarus protein influence the efficiency of sialylation, both in COS cells and in the venom glands.

Analysis of the COOH terminus of the purified venom enzyme shows that it does not contain the complete SARA sequence, but undergoes a proteolytic cleavage removing about half of this peptide. This cleavage occurs upstream of a series of four consecutive arginines, and thus resembles numerous proteolytic cleavages that participate in the maturation of peptidic hormones and other proteins (Cohen, 1987).

Cells expressing the SARA construction contained a fraction of uncleaved enzyme, while the culture medium only contained cleaved molecules, identical to those obtained with the RAD construction (Fig. 7). In addition, an anti-flag antibody recognized a fraction of the intracellular active AChE produced by the SARA-flag construction, but did not recognize the secreted enzyme. Thus, the full-length SARA enzyme is already active and it is cleaved before secretion, in COS cells and probably also in the venom gland. The significance of the cleavable arginine-rich extension is not clear, since the complete or truncated enzymes produced comparable levels of secreted AChE activity, in COS cells.

The Peripheral Site of Bungarus AChE

The peripheral site of Bungarus AChE presents two major differences with those of Torpedo and mammalian AChEs: the replacement of tyrosine 70 by a methionine and of an acidic residue at position 285 by a lysine.

Tyrosine 70 is located at the entrance of the gorge and would participate in the binding of peripheral site ligands, and in the relay between the peripheral site and the orientation of Trp-84 in the active site (Taylor and Radi, 1994; Barak et al., 1994; Ordentlich et al., 1995). This residue is replaced by a methionine in chicken and a serine in Caenorhabditis elegans, but while the chicken enzyme is insensitive to propidium, C. elegans AChE is very well inhibited, with a K_i of 1.3 µM (Arpagaus et al., 1994). The lack of inhibition of the chicken enzyme by propidium, as well as gallamine, D-tubocurarine (Eichler et al., 1994), and fasciculin (Barrett and Harvey, 1979), is probably related to the absence of another aromatic residue of the peripheral site, Trp-279.

Because propidium, gallamine, and D-tubocurarine possess quaternary ammonium groups, and basic residues of fasciculin play a key role in its binding (Cerveñansky et al., 1994; 1995), acidic residues are obviously important elements of the peripheral site. Indeed, mutations modifying the charge of Asp-72 and Glu-278, located at the rim of the gorge, considerably decreased the affinity of propidium and fasciculin (Shafferman et al., 1992; Radi et al., 1993; 1994; Barak et al., 1994). Thus, the presence of a lysine at position 285 of Bungarus AChE, instead of an aspartic or glutamic acid in Torpedo and mammalian AChE sequences, appeared significant. Although the role of this residue in the peripheral site has not been discussed previously, it is replaced by neutral residues in enzymes that present low sensitivity to some peripheral site ligands: glutamine in chicken AChE and glycine in human BChE.

We mutated residues Met-70 and Lys-285 of Bungarus AChE to Tyr and Asp, as found in Torpedo. Mutations M70Y and K285D had no effect on the catalytic turnover rate of the enzyme. In the case of M70Y, the observed increase in K_m, suggests that the presence of an aromatic group at position 70, at the entrance of the narrow catalytic gorge, partially restricts the access of the substrate to the active site. On the contrary, K_m was decreased for the K285D mutant. This appears consistent with the hypothesis that the peripheral site constitutes a primary zone of contact for positively charged substrates (Haas et al., 1992).

Bungarus AChE is inhibited by substrate concentrations above 1-2 mM. Mutations M70Y and K285D had little effect on this phenomenon, although they had a considerable influence on peripheral site ligands interactions, as discussed below. This observation argues against the hypothesis that occupancy of the peripheral site is responsible for excess substrate inhibition (Radi et al., 1991); in fact Asp-72 seems to be the only residue of the peripheral site implicated in this property of the enzyme (Radi et al., 1993).

Mutations of residues Met-70 and Lys-285 had no effect on the inhibition by a specific ligand of the active site, edrophonium. In the case of bis-quaternary inhibitors, which simultaneously bind to the active and peripheral sites (Berman et al., 1980; Harel et al., 1993; Barak et al., 1994), each mutation decreased K_i less than 10-fold. Mutations affecting Tyr-70 have already been shown to decrease inhibition by bis-quaternary inhibitors (Radi et al., 1993; Barak et al., 1994). Both the crystal structure of the AChE-decamethonium complex (Harel et al., 1993) and the model of the AChE-BW284C51 complex (Barak et al., 1994) suggest a direct interaction of Tyr-70 with these ligands. Our results confirm the importance of Tyr-70 and suggest that the charge of residue 285 also affects this interaction.

The mutations of both Met-70 and Lys-285 had a pronounced effect on inhibition by peripheral site ligands. Thus, the residue located at position 285 has considerable influence on the binding of peripheral site ligands, although it is rather distant (11 Å between carbon  atoms) from Trp-279, which is considered as the core of the peripheral site.

The wild type Bungarus AChE was found to be less sensitive to propidium, gallamine, and fasciculin than Torpedo AChE, but slightly more sensitive to D-tubocurarine. For D-tubocurarine, the M70Y mutation increased K_i, while K285D decreased it. The presence of a tyrosine at position 70 may cause a steric hindrance for the binding of this polycyclic, rigid ligand.

For propidium, gallamine, and fasciculin, each single mutation increased the sensitivity of Bungarus AChE and these effects were cumulated in the double mutant, which was at least as sensitive as Torpedo AChE. However, the two mutations had distinct effects; M70Y had a larger effect for propidium, K285D for gallamine, and they were equivalent for fasciculin. Previous studies analyzed mutations of Tyr-70 on the binding of propidium (Radi et al., 1993; Barak et al., 1994) and fasciculin (Radi et al., 1994); Tyr-70 probably interacts directly with these ligands, as indicated by modelization of the AChE-propidium complex (Barak et al., 1994) and the crystal structure of the AChE-fasciculin 2 complex (Bourne et al., 1995; Harel et al., 1995). In addition, we show that an acidic residue at position 285 also plays an important role in the binding of peripheral site inhibitors. The effect of the K285D mutation on fasciculin 2 inhibition is in excellent agreement with the structure of complexes between mouse or Torpedo AChE and fasciculin, showing that the position 285 acidic residue interacts with His-29 of the toxin (Bourne et al., 1995; Harel et al., 1995). Our results suggest that Asp-285 of AChE and His-29 of fasciculin also interact in the complex formed in solution. We found that modification of the charged residue at position 285 has a similar impact as the change of methionine to tyrosine at position 70 which was shown to be directly involved in the binding. This suggests that the stability of the complex also depends on electrostatic interactions.

Thus, both the aromatic and the charged residues at positions 70 and 285 participate, to various extents, in the interactions of ligands with the peripheral site. The differences observed in the effects of the M70Y and K285D mutations on the peripheral site inhibitors illustrate the fact that each ligand interacts with a distinct subset of residues. In fact, it is likely that an inhibitor might occupy distinct positions on the peripheral sites of wild type and mutant enzymes. This also applies to enzymes from different species.

In conclusion, the monomeric soluble form of AChE from Bungarus venom is a useful molecule for detailed analyses of the catalytic mechanism, including the role of the peripheral site. The monomeric structure should prove helpful studying the influence of the electric dipole moment of the protein on the traffic of substrates and products in the catalytic gorge, without any interference between subunits (Pörschke et al., 1996). The COOH-terminal sequence of this enzyme is quite unusual, since it differs from both H and T peptides and undergoes a distinctive proteolytic cleavage. This enzyme presents a unique ladder-like pattern in nondenaturing electrophoresis, probably related to its glycosylation. It is produced, in transfected COS cells, at a much higher level than an equivalent mammalian enzyme, the rat r-SAT secreted monomeric form. Whether this efficiency of the mammalian cells in producing a foreign protein resides in an intrinsic property of the protein itself is unknown, but this feature should prove useful in obtaining large amounts of recombinant AChE.