Identification of a novel type of alternatively spliced exon from the acetylcholinesterase gene of Bungarus fasciatus. Molecular forms of acetylcholinesterase in the snake liver and muscle.

The venom of the snake Bungarus fasciatus contains a hydrophilic, monomeric species of acetylcholinesterase (AChE), characterized by a C-terminal region that does not resemble the alternative T- or H-peptides. Here, we show that the snake contains a single gene for AChE, possessing a novel alternative exon (S) that encodes the C-terminal region of the venom enzyme, located downstream of the T exon. Alternative splicing generates S mRNA in the venom gland and S and T mRNAs in muscle and liver. We found no evidence for the presence of an H exon between the last common "catalytic" exon and the T exon, where H exons are located in Torpedo and in mammals. Moreover, COS cells that were transfected with AChE expression vectors containing the T exon with or without the preceding genomic region produced exclusively AChET subunits. In the snake tissues, we could not detect any glycophosphatidylinositol-anchored AChE form that would have derived from H subunits. In the liver, the cholinesterase activity comprises both AChE and butyrylcholinesterase components; butyrylcholinesterase corresponds essentially to nonamphiphilic tetramers and AChE to nonamphiphilic monomers (G1na). In muscle, AChE is largely predominant: it consists of globular forms (G1a and G4a) and trace amounts of asymmetric forms (A8 and A12), which derive from AChET subunits. Thus, the Bungarus AChE gene possesses alternatively spliced T and S exons but no H exon; the absence of an H exon may be a common feature of AChE genes in reptiles and birds.

The venom of the snake Bungarus fasciatus contains a hydrophilic, monomeric species of acetylcholinesterase (AChE), characterized by a C-terminal region that does not resemble the alternative T-or H-peptides. Here, we show that the snake contains a single gene for AChE, possessing a novel alternative exon (S) that encodes the C-terminal region of the venom enzyme, located downstream of the T exon. Alternative splicing generates S mRNA in the venom gland and S and T mRNAs in muscle and liver. We found no evidence for the presence of an H exon between the last common "catalytic" exon and the T exon, where H exons are located in Torpedo and in mammals. Moreover, COS cells that were transfected with AChE expression vectors containing the T exon with or without the preceding genomic region produced exclusively AChE T subunits. In the snake tissues, we could not detect any glycophosphatidylinositol-anchored AChE form that would have derived from H subunits. In the liver, the cholinesterase activity comprises both AChE and butyrylcholinesterase components; butyrylcholinesterase corresponds essentially to nonamphiphilic tetramers and AChE to nonamphiphilic monomers (G 1 na ). In muscle, AChE is largely predominant: it consists of globular forms (G 1 a and G 4 a ) and trace amounts of asymmetric forms (A 8 and A 12 ), which derive from AChE T subunits. Thus, the Bungarus AChE gene possesses alternatively spliced T and S exons but no H exon; the absence of an H exon may be a common feature of AChE genes in reptiles and birds.
Acetylcholinesterase (AChE) 1 (EC 3.1.1.7) is an essential component of cholinergic synapses, in the nervous tissues and muscles of vertebrates (1). This enzyme is also found in nonsynaptic contexts, where its function is unclear. In the blood of mammals, AChE exists in the form of soluble tetramers (G 4 na ), probably originating from the liver, and of membrane-bound dimers (G 2 a ), anchored by a glycophosphatidylinositol (GPI) to the surface of erythrocytes and lymphocytes (2); these enzymes could serve as a safeguard against any diffusion of acetylcholine from synapses into the circulation. The venoms of various Elapidae from the genera Bungarus, Hemachatus, Naja, and Ophiophagus represent a particularly rich source of nonsynaptic AChE (3,4).
The presence of AChE in snake venoms is mysterious because it is nontoxic by itself and does not enhance the toxicity of other venom components. This enzyme has been characterized as a true AChE, possessing the characteristic catalytic activity of AChEs from cholinergic tissues of other species: it hydrolyses acetylcholine faster than propionylcholine or butyrylcholine and it is inhibited by eserine (5). Moreover, the primary sequences of Naja and Bungarus venom AChEs present a strong homology to those of other AChEs, as shown by analysis of partial peptidic sequences (5,6) and by analysis of the complete sequence of Bungarus AChE deduced from cDNA clones (7).
The cloning of AChE from Bungarus venom revealed, however, that this homology is limited to the catalytic domain and that the C-terminal sequence is entirely different from both C-terminal H-and T-peptides, which are encoded by alternatively spliced exons in the single AChE gene and characterize AChE H and AChE T subunits of other vertebrates (review in Ref. 1). These C-terminal peptides determine the mode of posttranslational processing and quaternary associations of AChE catalytic subunits. Thus, AChE H subunits are modified by cleavage and addition of a GPI anchor, as well as by the formation of an intersubunit disulfide bond, generating GPIanchored dimers (8). The AChE T subunits produce monomers and a variety of disulfide-linked oligomeric forms, including homo-oligomers (dimers or tetramers) and hetero-oligomers, which incorporate structural collagen subunits (collagen-tailed forms) or hydrophobic subunits (hydrophobic-tailed tetramers). These hetero-oligomeric forms are tethered to extracellular matrices at neuromuscular synapses (9) or attached to cellular membranes, particularly in the brain (10). The T-peptide may adopt an amphiphilic ␣ helical structure, thus explaining the observation that monomers and dimers of AChE T subunits can interact with detergent micelles and membranes phospholipids (1). In contrast with all molecular forms that are normally produced by AChE H and AChE T subunits, the venom AChE consists of soluble, hydrophilic monomers. This is clearly re-* This research was supported by grants from the Centre National de la Recherche Scientifique, the Direction des Recherches et Etudes Techniques, the Association Française contre les Myopathies, and the Human Capital and Mobility program of the European Community. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ Recipient of a fellowship from the Institut National de la Recherche Agronomique.
ʈ To whom correspondence should be addressed. Tel.: 33-1-45688685; Fax: 33-1-40613057; E-mail: cbon@pasteur.fr. 1 The abbreviations used are: AChE, acetylcholinesterase; A 8 and A 12 , asymmetric forms composed of two or three AChE tetramers, associated with a triple helical collagen tail; AChE H , AChE S , and AChE T , AChE subunits of type H, S and T, generated from transcripts terminating with the H, S, and T exons; AChE gT , construction containing the intron that precedes exon T; AChE ⌬ , truncated AChE subunit limited to the catalytic domain; BChE, butyrylcholinesterase; G 1 a , G 2 a , and G 4 a , amphiphilic globular monomer, dimer, and tetramer, respectively; G 1 na and G 4 na , nonamphiphilic globular monomer and tetramer; GPI, glycophosphatidylinositol; iso-OMPA, tetraisopropyl pyrophos-lated to the fact that the venom AChE possesses a specific C-terminal peptide, which we called SARA after its last four residues: this peptide is highly hydrophilic and does not contain any cysteine residue that could establish intersubunit disulfide bonds. It defines AChE S subunits, which produce only soluble monomers when expressed in COS cells (7).
The presence of the SARA sequence, replacing the H or T sequences, which are encoded by alternative exons in Torpedo and mammalian AChE genes, raises the problem of the relationship between the venom enzyme and the AChE molecules that occur in cholinergic synapses of the snake. Several hypotheses may be considered to explain the production of this unusual type of AChE in venom glands. First, although previously studied vertebrates possess a single AChE gene (11)(12)(13)(14), the snake might possess two distinct AChE genes, expressed in cholinergic tissues and the venom glands, respectively. Such a duplication would be similar to the duplication of cholinesterase genes, which generate the twin enzymes AChE and butyrylcholinesterase (BChE) (EC 3.1.1.8) in vertebrates (15). Second, the venom enzyme may derive from the same gene as AChE in other tissues. In this case, the SARA sequence could be encoded by a novel type of alternative exon or by "readthrough" transcripts. Readthrough transcripts, in which the genomic sequence following the common catalytic exons is maintained, have been characterized in Torpedo electric organs (16), in mouse MEL cells (13) and embryonic diaphragm (17), and in rat embryonic liver (18). Readthrough transcripts are expected to produce nonamphiphilic, monomeric AChE, but the corresponding proteins have never been characterized in vivo.
In the present report, we show that Bungarus possesses a single AChE gene containing a novel alternative exon, S, localized downstream of the T exon. We identified the alternative splicing of the AChE transcripts in the venom glands, the liver, and the muscles, and we characterized the resulting molecular forms in vivo, as well as in COS cells expressing various constructs.

MATERIALS AND METHODS
RNA Purification-Bungarus fasciatus snakes were kindly provided by Prof. Xiong Yu-Liang and Dr. Zhang Yun (Kunming Institute of Zoology, Academia Sinica, Kunming, Yunnan, China). They were sacrificed in China, and the tissues (venom gland, muscle and liver) were immediately frozen and transported in dry ice. Total RNA was extracted using RNAsol (Bioprobe), according to the method of Chomczynski and Sacchi (19).
Reverse Transcription and PCR Experiments-For reverse transcription-PCR experiments, 1 g of total RNA was reverse transcribed using 200 units of Superscript-Moloney murine leukemia virus reverse transcriptase (Life Technologies, Inc.) with 10 pmol of oligo-dT, 2 pmol of specific primer, or 25 pmol of hexanucleotides, as described in the legend to Fig. 4. PCR was performed essentially as described previously (7) with Taq polymerase from Promega in a PTC150 thermocycler (M. J. Research).
RNase Protection Assay-For RNase protection assays, we introduced exon 4 and the T or S exon, under control of T7 promoter to produce antisense probes. 10 g of these constructs (described in Fig. 5) were digested overnight with 20 units of EcoRI in a final volume of 30 l. The DNA was extracted with phenol and chloroform and then precipitated by ethanol in the presence of ammonium acetate. DNA was resuspended in 20 l of RNase-free water and quantified using a GeneQuant spectrophotometer (Amersham Pharmacia Biotech). For probe synthesis, 3 g of DNA were incubated with 100 Ci of [ 32 P]dUTP (800 Ci/mmol; Amersham Pharmacia Biotech) and 150 units of T7 RNA polymerase (Promega) during 1 h at 37°C and then digested using 2 units of RNase-free DNase (Ambion). The probe was purified by denaturing polyacrylamide gel electrophoresis and eluted for 2 h at 37°C in a solution containing 0.5 M ammonium acetate, 0.2% SDS, and 1 mM EDTA. Total RNA (10 g) was then co-precipitated overnight with 100,000 cpm of probe (200,000 cpm in the case of venom gland RNA, in which the AChE mRNA is more abundant) with 0.5 M ammonium acetate. The pellet was then resuspended in Hybspeed buffer (Ambion).
Further incubations and RNase digestions were performed following precisely the instructions of the manufacturer. Samples of the reaction mixtures were then loaded on a sequencing denaturing polyacrylamide gel. After electrophoresis, the gel was dried and exposed to a Fuji imaging plate, which was read after 1 h. Each band was then quantified using the TINA program. The signal was corrected according to the amount in U bases in the protected fragment.
Tissue Extraction-Tissues and transfected cells were extracted in a low salt detergent buffer (50 mM Tris-HCl, pH 7.0, 5 mM MgCl 2 , 1% Triton X-100), and in some cases the pellet was re-extracted in a high salt buffer (same as above, with 1 M NaCl).
Sedimentation and Electrophoretic Analyses-AChE and BChE were analyzed by sedimentation in 5-20% sucrose gradients containing 10 mM Tris-HCl, pH 7.0, and 5 mM MgCl 2 , either without detergent or in the presence of 0.2% Triton X-100 or 1% Brij-96. AChE and BChE activities were assayed in the presence of specific inhibitors as indicated below. For analysis of asymmetric forms, the gradients contained 0.4 M NaCl and 1% Triton X-100. E. coli alkaline phosphatase (6.1 S) and E. coli ␤-galactosidase (16 S) were included as internal sedimentation standards. After centrifugation at 36,000 rpm for 18 h at 7°C in a Beckman SW41 rotor, 45 fractions were collected from the bottom of the tubes and assayed for the different enzymatic activities. Electrophoresis in nondenaturing polyacrylamide gels was performed as described previously (24,25). The gels contained 0.25% Triton X-100 with or without 0.05% deoxycholate, and they were electrophoresed for approximately 2-3 h under 15 V/cm, with cooling at 15°C. AChE activity was revealed by the histochemical method of Karnovsky and Roots (26).
Collagenase and PI-PLC Treatment-Collagenase form III (27) was purchased from Advance Biofactures Co. (Lynnbrook, NJ). A high salt extract containing AChE was incubated with 40 units of collagenase in a buffer containing 50 mM Tris-HCl, pH 8, and 5 mM CaCl 2 , for 1 h at 26°C. Treatment with PI-PLC was performed as described previously (28).
Assays of AChE and BChE Activity-AChE and BChE were assayed by the colorimetric method of Ellman et al. (29). Acetylthiocholine was used as a substrate for both enzymes. AChE was assayed in the presence of the specific anti-BChE inhibitor iso-OMPA (10 Ϫ5 M), and BChE was assayed in the presence of the specific anti-AChE inhibitor BW284C51 (1,5-bis(4-allyldimethylammoniumphenyl)-pentan-3-one dibromide) (10 Ϫ5 M).

FIG. 1. Sedimentation analysis of AChE and BChE from Bungarus liver.
A tissue extract obtained in the presence of detergent at low ionic strength was centrifuged in the presence of Triton X-100, and equal samples of the fractions were assayed without inhibitor for total cholinesterase activity (E), in the presence of 10 Ϫ5 M iso-OMPA for AChE activity (q), and in the presence of 10 Ϫ5 BW284C51 (Ⅺ) for BChE activity. Liver cholinesterases were composed of a major G 1 na AChE and a minor G 4 na BChE, both of which were characterized as nonamphiphilic because their sedimentation was the same in the presence of Brij 96 and in the presence of Triton X-100 (not shown).
Genomic Structure and Plasmidic Constructs-Genomic DNA was extracted and isolated by a salting-out protocol (30): liver was crushed in liquid nitrogen and transferred in 10 volumes of extraction buffer (10 mM Tris-HCl, pH 8.0, 0.1 M EDTA, 20 g/ml pancreatic RNase, 0.5% SDS, and 0.5 M NaCl) and incubated at 50°C for 30 min. Proteinase K was added at a final concentration of 100 g/ml, and incubation was performed overnight at 50°C. Saturated NaCl (Ͼ6 M) was then added ( 1 ⁄4 volume) and agitated. After centrifugation (15 min at 5000 rpm), 2 volumes of cold ethanol were added to the supernatant, which was then kept on ice for 10 min. After centrifugation (15 min at 5000 rpm), the DNA pellet was redissolved in Tris-EDTA buffer.
PCR was performed with primer oligonucleotides corresponding to sequences of exons 3, 4, T, and S. To search for the presence of a putative exon H between exons 4 and T, we made several constructs by inserting various 3Ј sequences, using a unique BglII site, located in exon 4, as shown in Fig. 7.
Transfection of COS Cells-COS cells were transfected by the DEAEdextran method, as reported previously (31), using 5 g of DNA encoding the catalytic subunit AChE T with or without DNA encoding the Q N /H C binding protein (31,32), as specified. The cells were maintained at 37°C and extracted 2-4 days after transfection. The culture medium (7 ml/10-cm dish containing about 5 ϫ 10 6 cells) was collected after variable periods of time, as indicated, for analysis of released AChE activity. The extracts and culture media were stored at Ϫ80°C.

Molecular Forms of Cholinesterases in Bungarus Liver and
Muscles-Extracts from snake liver were found to hydrolyze butyrylthiocholine, as well as acetylthiocholine, indicating the presence of both AChE and BChE. As shown in Fig. 1, an AChE component sedimented at 4.5 S (about 80% of the total cholinesterase activity), and a BChE component sedimented at 10.9 S (about 20% of the total activity); only the latter component hydrolyzed butyrylthiocholine (not shown). The sedimentation patterns were not modified by incubation with PI-PLC (not shown) and were identical in the presence of Triton X-100 ( Fig.  1), in the presence of Brij-96, or without detergent (not shown), indicating that both components were nonamphiphilic, corresponding to a monomeric form of AChE (G 1 na ) and a tetrameric form of BChE (G 4 na ). The residual activity observed around 11 S in the presence of iso-OMPA, a specific inhibitor of BChE, and the fact that BW284C51, a specific inhibitor of AChE, reduced the cholinesterase activity of the same fractions suggests the presence of a small contribution of tetrameric AChE (G 4 na ). The absence of amphiphilic forms and the fact that PI-PLC had no effect on these profiles indicated that the snake liver did not produce any GPI-anchored form of AChE. Fig. 2 shows that detergent-soluble extracts from Bungarus muscles contained a much smaller proportion of BChE, repre-  Fig. 1, except that here we used a 10-fold larger sample for the assay of BChE than of AChE. We observed that the main activity was due to AChE. Amphiphilic forms G 1 a and G 4 a are characterized by a shift in sedimentation between Triton X-100 (TX-100) and Brij-96, whereas hydrophilic monomer G 1 na sediments identically in both conditions.

FIG. 3. Existence of collagen-tailed AChE in
Bungarus muscle. The pellet obtained after low salt extraction of muscle was re-extracted with a high salt buffer. In the resulting extract, globular forms were still major components; for this reason, we used a 5-fold enlarged scale for the left part of the graph, which shows the minor collagen-tailed components, compared to the right part of the graph, which corresponds to the G 1 a and G 4 a forms. These molecules are characterized as collagentailed A 8 and A 12 by collagenase treatment, which increased their sedimentation coefficient. E, untreated; OE, collagenase-treated.
senting less than 2% of the total cholinesterase activity. The sedimentation patterns, obtained in the presence of Triton X-100 ( Fig. 2A) or of Brij-96 (Fig. 2B), showed three AChE components (a major one, corresponding to amphiphilic tetramers (G 4 a ) sedimenting at 9.5 S in Brij-96 and 10.9 S in Triton X-100, and two minor ones, corresponding to amphiphilic monomers (G 1 a ) sedimenting at 3.5 S in Brij-96 and 4.5 S in Triton X-100) and nonamphiphilic AChE monomers (G 1 na ) sedimenting at 4.5 S in both conditions. The same muscle extract contained a small proportion of BChE, corresponding to amphiphilic tetramers (G 4 a ) sedimenting at 9.8 S in Brij-96 and 11 S in Triton X-100. The AChE and BChE G 4 a peaks presented a small but definite difference in their sedimentation when assayed in the same gradient fractions, showing that they really correspond to distinct molecules. There was no indication of the presence of a dimeric AChE form, and treatment with PI-PLC did not modify the sedimentation profiles.
To examine whether snake muscles also contained collagentailed AChE forms, which aggregate at low ionic strength, the pellet obtained after two successive extractions in the low salt detergent buffer was re-extracted in a high salt detergent buffer. The AChE activity that was solubilized under these conditions was analyzed by sedimentation in a gradient containing 0.4 M NaCl and 0.2% Triton X-100, with or without prior digestion by collagenase. Fig. 3 shows that in addition to major G 1 a and G 4 a components, the high salt extract contained a small proportion of asymmetric A 8 and A 12 forms, which were modified by collagenase in a characteristic manner: their sedimentation coefficients were increased from 13.4 to 15.1 S and from 17.3 to 18.5 S, respectively. Taken together, these two molecular forms represent less than 3‰ of the total AChE activity in the muscle extracts.
The presence of G 1 a , G 4 a , A 8 , and A 12 forms clearly demonstrates that AChE T subunits are expressed in Bungarus muscles. The absence of GPI-anchored dimers indicate that AChE H subunits are not produced in Bungarus liver or muscles. Therefore, it appears that only AChE T and AChE S are produced in these tissues. To confirm this, we analyzed the structure of AChE transcripts and of the AChE gene.
AChE Transcripts Are Generated by a Single Gene in Bungarus Muscles, Liver and Venom Glands-A Southern blot of digested Bungarus genomic DNA was hybridized with a probe FIG. 4. Cloning of the 3 end of muscle AChE cDNA. A, description of the oligonucleotides used for reverse transcription. The hexa(N) part of the primer annealed randomly with RNA. After reverse transcription, we amplified the 3Ј region of AChE cDNA by PCR, using a forward primer specific for the catalytic domain and a reverse primer situated on the Ri-Ro part of the hexa(N) primer (20). B, sequence of a cDNA clone obtained by this method. The arrowhead indicates the limit between the catalytic domain, which is identical to that of venom gland AChE S , and a C-terminal T-peptide. C, alignment of the protein sequence deduced from the Bungarus cDNA fragment with sequences of T-peptides from different species. The symbols indicate fully conserved aromatic residues (*), an aromatic residue that is conserved only in vertebrate T-peptides (#), and the conserved cysteine (OE). References are as follows: Torpedo (11), mouse (13), rat (18), human (21), chicken (22), and Caenorhabditis elegans (23).

FIG. 5. Analysis of AChE transcripts in Bungarus venom gland and muscle by RNase protection assay.
Total RNA from venom gland (G) and muscle (M) from B. fasciatus was protected using two probes containing the last catalytic exon (named exon 4, by analogy with the mammalian AChE gene) (167 bp), followed by the coding region of exon T (125 bp) (A) or of exon S (50 bp) (B) and by a short fragment of the vector (52 bp). The YϪ and Yϩ lanes are controls with yeast RNA, without (Ϫ) and with (ϩ) RNase digestion. The signal obtained with liver RNA was not strong enough to be shown on the same figure. The product of a sequence reaction was loaded on the same gel to determine the size of protected fragments (not shown).
corresponding to nucleotides 1610 -1727 of the AChE cDNA from venom glands, within the coding region of the catalytic domain. We obtained a single labeled band after digestion by EcoRI and BamHI, suggesting that Bungarus possesses a single gene for AChE (not shown).
Experiments involving 3Ј rapid amplification of cDNA ends were unsuccessful to identify the 3Ј region of the coding sequence of AChE cDNA in muscle. To obtain the complete coding sequence, we amplified a cDNA fragment encoding the C-terminal region of muscle AChE, by reverse transcription-PCR. Reverse transcription of mRNA was performed with random hexanucleotides associated with Ri and Ro sequences for PCR priming (20) (Fig. 4A). PCR was then performed with Ri and Ro reverse primers and a forward primer corresponding to a frag-ment from the AChE cDNA previously cloned from venom gland (7). We thus amplified a fragment of about 200 bp, which was subcloned and sequenced (Fig. 4B). According to this sequence, the end of the catalytic domain is identical to that of the venom cDNA clone, but it is associated with a different C-terminal region. In agreement with our analysis of AChE forms in muscles, this region corresponds to a T-peptide, as shown by its alignment with sequences from Caenorhabditis, Torpedo, avian, and mammalian AChEs (Fig. 4C). The strict conservation of eight aromatic residues and of a cysteine residue at position Ϫ4 of the C terminus is particularly noticeable. Thus, the AChE gene that produces the venom enzyme also generates T transcripts in muscles.
Reverse transcription-PCR experiments showed that both S  Fig. 7) is indicated by a dotted line. Note that the coding sequence of the S-peptide is preceded by a classical consensus splicing acceptor site and that the preceding untranslated region contains GC-rich domains. The end of the presented sequence corresponds to the beginning of the poly(A) tail of S transcripts (7). and T transcripts exist in venom glands, liver, and muscle (not shown). RNase protection assays indicated that the venom glands contain mostly S transcripts, with only a small fraction (Ͻ5%) T transcripts, whereas the other tissues contain approximately two-thirds S transcripts and one-third T transcripts (Fig. 5).
Genomic Structure of the 3Ј Region of the Bungarus AChE Gene-We explored the structure of the Bungarus AChE gene by PCR amplification of genomic DNA, using primers corresponding to various exonic sequences. The results are shown in Fig. 6. The sequence encoding the catalytic domain is interrupted by an intron (about 1.3 kb) at the level of amino acid 475 (according to the Torpedo numbering), as in other vertebrates. The last common exon (encoding 35 residues) is followed by a genomic region of 1741 base pairs, preceding the T exon. The region encoding the C-terminal part of the venom AChE (SARA, or S) is located about 300 nucleotides downstream of the stop codon of the T exon. Therefore, the SARA region does not derive from a readthrough sequence but from a novel type of exon, which we call S (Fig. 7). It is interesting to note that the 300-nucleotide-long sequence located between the sequence encoding the T-peptide and exon S contains GC-rich domains.
In Torpedo and mammalian AChE genes, the H exon is located between the last common catalytic exon and exon T (1). The 1741 bp-long corresponding region in Bungarus AChE gene was fully sequenced: its analysis did not reveal the presence of any H-like open reading frame that might encode a GPI-addition C-terminal signal peptide (33). Therefore, the Bungarus AChE gene does not appear to contain an H exon.
Expression of Bungarus AChE in Transfected COS Cells: Existence of an H Exon?-We transfected COS cells with expression vectors containing either the cDNA sequence encoding the AChE T subunit or a partial genomic construct (AChE gT ), which included the 1741-bp intron preceding exon T, where a putative H exon would be expected to be localized (see Fig. 7). The total AChE activity and the proportion that was secreted in the culture medium were the same in both cases: 30 -45% of the activity was recovered in the medium, 2 days after transfection. As illustrated in the case of the AChE T construct (Fig.  8, A and B), sedimentation analysis showed that the cells and the culture medium contained G 1 a , G 2 a and G 4 na forms, as well as heavy polydisperse aggregates. These aggregates, which have not been observed in other AChE species, accounted for as much as 80% of the total activity in cell extracts. The culture medium contained the same type of molecules, including aggregates, with a higher proportion of G 2 a than in the cells. We obtained the same results with the AChE gT construct (not shown). We could not detect any PI-PLC-sensitive AChE, showing that only AChE T subunits were produced. This demonstrates the absence of any functional exon H, that would generate AChE H subunits.
In co-transfection with the binding protein Q N /H C (31, 32), were analyzed by sedimentation in sucrose gradients containing Brij-96 (E) or Triton X-100 (q). In the case of co-transfection with Q N /H C , a sample of the cell extract (C) was treated with PI-PLC (‚). The presence of aggregates sedimenting above 15 S was clearly visible in all gradients, but their proportion was markedly decreased by Q N /H C , which induced the association of monomers and dimers into GPI-anchored tetramers (G 4 a ) and the subsequent production of lytic nonamphiphilic tetramers (G 4 na ) and monomers (G 1 na ) (31). A small proportion of G 4 a was recovered in the culture medium (D), possibly attached to membrane fragments. the proportion of monomers, dimers, and aggregates was markedly decreased; AChE T subunits were largely assembled into GPI-anchored tetramers characterized by their sensitivity to PI-PLC, both in the case of the cDNA (Fig. 8, C and D) and in the case of the AChE gT construct (not shown). These results were confirmed by electrophoretic analyses in nondenaturing polyacrylamide gels (Fig. 9). The GPI-anchored G 4 a form was converted by PI-PLC to a hydrophilic derivative, which migrated faster in nondenaturing electrophoresis (Fig. 9). The production of such heteromeric molecules is a further confirmation that Bungarus AChE T subunits can form the same types of quaternary associations as those of other species.
When we varied the quantity of vector DNA used for transfection, the production of secreted AChE activity increased at low doses, and reached a plateau for about 5 g of DNA/dish. Although the saturating dose was approximately the same, the yields obtained varied widely with the different constructions: the secreted AChE activity was about 10-fold lower in the case of AChE T than of AChE S or of AChE ⌬ . In the case of AChE S or AChE ⌬ , the cells only produced monomers (G 1 na ), and the activity was at least 85% secreted (7). We also compared the production of AChE activity with corresponding rat AChE T and AChE ⌬ . As shown previously (7), the yield of secreted AChE was about 30-fold higher for Bungarus AChE S or AChE ⌬ than for rat AChE ⌬ . Similarly, the yield was higher for Bungarus AChE T than for rat AChE T , but in this case the ratio was only 2-fold. DISCUSSION Bungarus Possesses a Single AChE Gene, with a Novel Type of Alternative Exon-Analyses of genomic DNA and of AChE cDNA from muscle showed that Bungarus, like other vertebrates, possesses a single AChE gene, and that AChE S and AChE T subunits are produced by alternative splicing. The Speptide is not encoded by a readthrough sequence but by a bona fide alternative exon, called S, which is located 3Ј of the T exon.
The peptidic sequence encoded by the T exon is highly homologous to the C-terminal T-peptides of other AChEs and BChEs. Among the conserved residues, it is interesting to note the presence of a cysteine residue near the C terminus, involved in intersubunit disulfide bridges, as well as of several aromatic residues, probably involved in the hydrophobic char-acter of an amphiphilic ␣ helix, in the N-terminal part of the peptide (1). The capacity of Bungarus AChE T subunits to form heteromeric quaternary structures with a proline-rich attachment domain (PRAD) (31,34) is demonstrated by the presence of collagen-tailed AChE forms in Bungarus muscle and the formation of GPI-anchored tetramers with the Q N /H C chimeric protein in transfected COS cells.
The 3Ј untranslated region of the muscle transcripts encoding AChE T subunits contains GC-rich domains, so that it was not possible to define its extremity by the rapid amplification of cDNA ends 3Ј method. Three putative polyadenylation sites are located 190 bases 3Ј of the T exon stop codon. The T transcripts may terminate at such sites or include the S coding sequence, in the same manner as the T sequence is included in the 3Ј region of H transcripts of mouse AChE (14).
In the AChE genes of Torpedo and mammals, the H exon is located upstream of the T exon. Several lines of evidence indicate that the Bungarus AChE gene does not possess an H exon: (a) we could not find any GPI-anchored AChE dimers, which would be generated from AChE H subunits, in Bungarus tissues (this is particularly significant in the liver, because this organ is rich in GPI-anchored AChE in rat) (18,35,36); (b) the genomic region separating the last common exon from the T exon does not contain any sequence that might encode an H-peptide; and (c) COS cells transfected with a construct, AChE gT , in which this region was included produced only AChE T subunits, showing that it does not contain any alternative exon. In the case of human AChE, a similar construction led to the production of both H and T subunits (37).
It is interesting that H exons have not been found in the genes of mammalian BChE, quail AChE (1), and Electrophorus AChE (38). In fact, among vertebrates, only Torpedo and mammals have been shown to possess a GPI-anchored AChE form.
Expression in Transfected Cells-It has been shown that the C-terminal peptides of AChE subunits determine the fate of the enzyme in a tissue-specific manner, and in particular its metabolic stability: thus, the rat RBL cells express rat AChE H subunits and expose them at their surface much more efficiently than AChE T subunits, despite the fact that the two proteins seem to be synthesized at equivalent levels (39). We found that AChE activities were systematically higher in transfected COS cells expressing Bungarus AChE than in cells expressing rat AChE. This was true for AChE T subunits and was even more marked in the absence of the T-peptide (AChE ⌬ or AChE S ). In all cases, we observed a similar influence of the amount of vector DNA used for transfection, with saturation at approximately the same dose (5 g/dish), so that the difference could not be ascribed to transcription; translation is also very unlikely to differ between constructions such as AChE T and AChE ⌬ , which differ only by the presence or absence of the C-terminal 40-amino acids T-peptide. Comparisons of AChE activities obtained with the different constructions clearly showed that the presence of a C-terminal T-peptide reduced the yield of active enzyme and that the catalytic domains of Bungarus and rat AChEs present intrinsic differences. The snake enzyme may be able to fold more efficiently into its active conformation, as suggested by its capacity to renature after exposure to guanidinium hydrochloride. 2 Whereas the truncated Bungarus or rat AChE ⌬ subunits remained exclusively monomeric, the AChE T subunits of both species generated monomers, dimers, and tetramers, as expected (31). However, the major part of active Bungarus AChE T formed heavy polydisperse aggregates, which were not observed in the case of rat AChE T . We do not know whether active Bungarus AChE T subunits aggregate with incorrectly folded inactive subunits or with other cellular proteins, such as chaperons (40). In any case, in the presence of the binding protein Q N /H C , GPI-anchored heteromeric tetramers were assembled at the expense of the heavy aggregates.
Expression of Cholinesterases in Bungarus Liver and Muscle-The liver and muscles of B. fasciatus contain both AChE and BChE activities, demonstrating that Bungarus possesses two distinct cholinesterase genes, like other vertebrates.
Reverse transcription-PCR and RNase protection assays showed that AChE transcripts in the venom gland are predominantly of type S, with less than 5% type T, in agreement with the production of soluble monomers derived from AChE S subunits. Liver and muscle contain both types of transcripts: about two-thirds type S and one-third type T. The liver contains mostly soluble AChE monomers, which may correspond to AChE S subunits, as in the venom, and a minor proportion of tetrameric AChE, which probably consists of AChE T subunits. The proportions of these two molecular forms do not correspond to those of the S and T transcripts, either because AChE T subunits are partially converted into soluble monomers, by removal of their amphiphilic C-terminal region, or because AChE S subunits are more efficiently produced than AChE T subunits, as observed in transfected COS cells.
In Bungarus muscle, the situation is opposite because the major AChE forms (G 1 a , G 4 a , and collagen-tailed molecules) derive from AChE T subunits, despite the fact that this tissue also contains more S than T transcripts. The production of AChE S subunits in muscle may be underestimated because of their rapid secretion, as observed in transfected COS cells.
In any case, the contrast between the AChE transcripts and molecular forms in liver and muscle clearly illustrates the fact that the expression of a protein cannot be simply deduced from the level of its mRNA but also critically depends on cellular specificity of posttranslational processing.
Collagen-tailed forms of AChE in Bungarus Muscles-The small proportion of collagen-tailed AChE forms obtained in extracts from Bungarus muscle raises the question of their functional role: muscles differ widely in the proportions of collagen-tailed and globular AChE forms, depending on the species and on their physiological slow or rapid type (1); an extreme case is that of Torpedo muscle, which only contains GPI-anchored G 2 a AChE. On the other hand, our results may reflect the nonextractability of collagen-tailed molecules rather than their low abundance. In quail muscle, collagen-tailed AChE exists in extractable and nonextractable states (41); such molecules may be linked by disulfide bonds to other components of the extracellular matrix through the C-terminal cysteine-rich region of the collagen tail (42).
Evolutionary Significance of the S Exon-The presence of a novel alternatively spliced exon, S, in Bungarus, raises interesting evolutionary questions. The S exons may have originated independently of the production of AChE in venoms. In fact, S transcripts are also expressed in the liver and muscles. In addition, AChE does not appear to contribute to the toxicity of the venom (7). It will therefore be interesting to examine whether the presence of an S exon is correlated with expression of AChE in the venom, in particular in Dendroaspis snakes (mambas), which do not contain AChE in their venom (4), and whether S exons also exist in other reptiles. The evolutionary significance of the absence of H exons and the presence of S exons in Elapidae snakes clearly deserves more detailed studies.