Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

The venom of the green mamba, Dendroaspis angusticeps, contains a number of trace isotoxins of m1-toxin (Max et al., 1993a,b,c), of which m1-toxin1 (Fig. 1; the renamed original m1-toxin) is the most prevalent (Carsi and Potter, 2000). m1-Toxin1 is a remarkably useful ligand for studying genetically defined M₁ muscarinic receptors. It binds specifically to otherwise unliganded M₁ receptors in vitro and in vivo (Max et al., 1993a,c; Liang et al., 1999; Carsi and Potter, 2000). Binding of the toxin completely blocks the subsequent binding of the antagonist [³H]N-methylscopolamine ([³H]NMS) for several hours at 4-37°C (Max et al., 1993b; Carsi and Potter, 2000), and prevents the action of muscarinic agonists on M₁ receptors (Max et al., 1993c; Cuevas et al., 1997; Marino et al., 1998; Rouse et al., 1999, 2000). When applied after [³H]NMS, the toxin forms toxin-NMS-receptor complexes that are quite stable in membranes and completely stable in digitonin-solution (Max et al., 1993b). The toxin must therefore bind allosterically to the unique extracellular loops and/or N terminus of M₁ receptors, which are located outside the receptor pocket that contains the binding site for NMS. ¹²⁵I-m1-Toxin1 dissociates slowly from toxin-receptor complexes at 37°C, indicating that toxin binding is not covalent (Potter, 2001).

View larger version (26K): [in this window] [in a new window]   Fig. 1.   Amino acid alignment of seven toxins from D. angusticeps and D. polyepsis that bind to muscarinic receptors. Black shading shows consensus homologies, and the C-C bonds that form the three larger loops in each toxin are indicated. The sequence diversity among these toxins is primarily in these loops. m1-Toxin1 binds specifically and irreversibly to M1 receptors (Carsi and Potter, 2000) and m4-toxin binds with 102-fold higher affinity to M4 than M1 receptors (IC50 values of 0.95 and 95 nM; Liang et al., 1996). The other toxins bind reversibly and with nearly equal affinity to at least two receptor subtypes (Jolkkonen, 1996). MT is unusual because it binds with 3 to 5 nM affinity to M3-M5 receptors. Other differences between these toxins include the fact that m1-toxin1 and m4-toxin are antagonists (Max et al., 1993a; Cuevas et al., 1997; Cuevas and Adams, 1997; Marino et al., 1998), whereas MT1 behaves as an agonist (Jerusalinsky et al., 1995).

The binding properties of m1-toxin1 indicate an exceptionally good molecular fit between the loops and/or ends of the toxin, and the extracellular loops and/or N terminus of M₁ receptors (Max et al., 1993b). The present studies are an initial step toward characterizing this fit, using site-directed mutants of m1-toxin1 to establish toxin residues that are essential for toxin binding. Future studies of interactions between mutants of both proteins should reveal a detailed picture of the toxin-receptor complex, and a mirror image of the receptor may then emerge from knowledge of the structure of the toxin. Several lines of information facilitate this approach. First, m1-toxin1 belongs to a large family of relatively rigid spoon-shaped short-chain toxins. NMR studies have shown that muscarinic toxin MT2, like other members of this family, has a dense core containing four disulfide bonds and four loops (Figs. 1 and 2; Ségalas et al., 1995). Most of the sequence diversity among short-chain toxins resides in their three larger loops (Fig. 1). Several residues in the corresponding loops of the similar toxin erabutoxin a have been mutated, yielding information about residues that perform structural and functional roles (Pillet et al., 1993; Tremeau et al., 1995). Second, the toxins that bind to muscarinic receptors differ only modestly in their sequences (Fig. 1), despite large differences in their binding characteristics. Only seven of the amino acids of m1-toxin1 are not conserved between other toxins that bind to muscarinic receptors, although only the m1-isotoxins bind specifically, irreversibly, and allosterically to M₁ receptors (Max et al., 1993a,b,c; Carsi and Potter, 2000). Two of these seven residues (Phe³⁸, Lys⁶⁵) were mutated to the corresponding residue that is conserved in the other muscarinic toxins (Ile³⁸, Glu⁶⁵; Fig. 1). The structural ramifications of making F38I and K65E were predicted from computer modeling of the atomic coordinates of MT2 (Ségalas et al., 1995). Ile³⁸ exhibits multiple interactions with surrounding residues in the middle loop of MT2 (Fig. 2), so F38I is likely to have a slightly different conformation than m1-toxin1, but still be a stable molecule. The C-terminal Glu⁶⁵ is exposed on the convex face of MT2 and interacts minimally with surrounding residues. So K65E should have a protruding C terminus with an opposite charge to that of the native toxin, yet no change in conformation. Finally, only 13 of the 77 extramembranous amino acids of M₁ receptors are not conserved in M₂-M₅ receptors, so the most likely sites in M₁ receptors for the specific binding of m1-toxin1 are limited (Hulme et al., 1990; Max et al., 1993b).

View larger version (91K): [in this window] [in a new window]   Fig. 2.   Solution NMR structure of MT2. Atomic coordinates were provided by Ségalas et al. (1995). In the picture on the left, the molecule has been rotated to show the "three-fingered" folding pattern found in all short-chain neurotoxins. The position of Ile38 in the ascending portion of the middle loop is indicated with an asterisk. This region is stabilized by four C-C bonds in the core, by a double-stranded antiparallel -sheet in the left loop shown here, and by a triple-stranded antiparallel -sheet formed by the middle loop and the ascending portion of the right loop. In the picture on the right, the molecule has been rotated to show that the C-terminal region is exposed on the convex face of the toxin. Mutations of the C-terminal residue (Lys65 in m1-toxin1) are obviously much less likely to modify the conformation of the whole molecule than mutations near Ile38 (Phe38 in m1-toxin1).

Recombinant m1-toxin1 and F38I and K65E were tested for four properties: 1) the ability of each toxin to block the binding of [³H]NMS to cloned M₁ receptors in membranes, 2) the selectivity of the toxin for M₁ versus M₂-M₅ receptors, 3) the reversibility of M₁-blockade, and 4) the ability of the toxin to slow the dissociation of [³H]NMS from M₁ receptors.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials. An adult Dendroaspis angusticeps was obtained from Natural Selections (Miami, FL), guanidinium isothiocyanate and primers and a TA cloning kit from from Invitrogen (Carlsbad, CA), [³H]N-methylscopolamine (82 Ci/mmol) and [³²P]ATP (3000 Ci/mmol) from PerkinElmer Life Science Products (Boston, MA), and restriction enzymes from New England Biolabs (Beverly, MA). Unless specified, all other reagents were from Sigma (St. Louis, MO). DNA sequencing was performed using a Thermal Cycling Sequencing kit (PerkinElmer). Sequences were aligned by the Clustal method (DNASTAR, Madison, WI). Protein assays were performed using a Micro-BCA kit from Pierce (Rockford, IL). Antiserum to m1-toxin1 was generated to a synthetic peptide (PKAEYRDVIN) by Tana Labs (Houston, TX).

Cloning the cDNA for m1-Toxin1. An adult D. angusticeps was chilled in a freezer and decapitated. Its venom glands were homogenized in 4.0 M guanidinium isothiocyanate, the homogenate was centrifuged over 5.7 M cesium chloride for 16 h at 214,000g_max, and the RNA pellet was resuspended in water treated with diethylpyrocarbamate (Chirgwin et al., 1979). Messenger RNA was isolated from total RNA using a PolyATract mRNA isolation kit (Promega, Madison, WI). Two degenerate oligonucleotides were designed to amino acid sequences unique to m1-isotoxins (Max et al., 1993a; Carsi and Potter, 2000). Reverse transcription was performed using a downstream primer, DEG2 (5'CAYTTRTCIGTICCRCARCARTTIATIAC3'), and 250 ng of venom gland mRNA. The resultant first strand cDNA was then used as a template for PCR, using an upstream primer, DEG1 (5'GARGAYTGYCCIGAYGGICARAAY3'), in combination with DEG2. PCR was performed with 25 cycles of 1 min at 95°C, 1 min at 65°C, and 1 min at 72°C. An adapter-ligated cDNA library was prepared from 1 µg of venom gland mRNA using a Marathon RACE kit (CLONTECH, Palo Alto, CA). RACE was initiated using 10 µl of a 1:25 dilution of the adaptor-ligated cDNA mix. The 50-µl reaction mixture contained 2.0 mM Mg, 0.3 mM dNTP, 0.5 µM each of primer AP1 (CLONTECH) and GSP-1 (5'AACAGTTAATGACGTCACGGTATTCC3'), and 2.5 units of Primezyme (Biometra, Inc., Tampa, FL). PCR was performed with 30 cycles of 30 s at 94°C, 2.5 min at 68°C, and a final 15-min incubation at 68°C. Identical conditions were used for 3'-RACE, except that GSP-2 (5'TTCTCCAAGAATGTATGACTTCACCAGG3') replaced GSP-1. Both RACE products were purified on a 1% agarose gel, ligated into pCRII (Invitrogen) and used to transform DH5 cells (Invitrogen). Plasmids were purified using a Wizard Mini-Prep kit (Promega). Both strands were sequenced to check their fidelity. Restriction mapping of both PCR products identified a unique AatII site present in the region of overlap between the 5' and 3' RACE clones. This site was used to ligate the RACE products into pCRII to reconstruct the full-length cDNA. Two primers were designed to remove the untranslated region and to add restriction sites, 5'BamHI (5'GCGGATCCGCCACCATGAAAACTCTGC3') and 3'EcoRI (5'CGCGCGAATTCGAGCTATTTGTTGC3'); the underlined regions indicate the restriction sites. These facilitated cloning of the cDNA into a Pichia pastoris expression vector pPIC3.5 (Invitrogen), resulting in the m1-toxin1 expression plasmid pm1tx1. The GenBank accession number for m1-toxin1 cDNA is AF241871.

Site-Directed Mutagenesis. Site-directed mutants of m1-toxin1 were prepared by primer mismatch PCR. The primers for F38I were F38Iup (5'ACATCCCCTGGTGATGTCATACATTCTTGGAGAAATGTACTG3') and F38Idown (5'AGAATGTATGACATCACCAGGGGATGTGCTGC3'); the underlined portions include the mismatch. Because K65 is a C-terminal amino acid, only one primer was designed for K65E, (5'TATATATGAATTCTGGAGCTATTCGTTGCATTTGTCTGTTCCGC3'). F38I was created using two separate PCR reactions and an outside primer, in this case the AOX1 primer to the pPIC3.5 vector, F38Iup for the 5' side of the cDNA, and F38Idown for the 3' side of the cDNA. The reaction conditions were a 2-min incubation at 94°C followed by 30 cycles of 1 min at 94°C, 1 min at 50°C, and 30 s at 72°C. Each PCR product was purified and used as a template in a third PCR reaction with only the AOX1 primers. The conditions for the PCR extension were a 2-min incubation at 94°C, followed by five cycles of 1 min at 94°C, 1 min at 55°C, and 1 min at 72°C. Then the PCR reaction incubation was changed to 25 cycles of 1 min at 94°C, 1 min at 65°C, and 1 min at 72°C. The PCR product was subcloned into the pPIC3.5 vector and sequenced to ensure PCR fidelity. The expression vector with the F38I mutant cDNA cloned downstream of the AOX1 promoter was called pF38I. K65E was generated by primer mismatch. Using the upstream primer AOX1 and the K65E primer, the PCR product was generated with the following protocol: 2 min at 94°C followed by 30 cycles of 1 min at 94°C, 1 min at 50°C, and 1 min at 72°C with a 10-min extension at 72°C. The K65E PCR product was subcloned into the pPIC3.5 vector downstream the AOX1 promoter, and named pK65E.

Pichia Growth Media. Minimal media contained 1.5% yeast nitrogen base, 75 mM ammonium sulfate, and 0.005% d-biotin. Minimal glycerol media (MG) was minimal media plus 5% glycerol. Minimal methanol media (MM) was minimal media plus 0.5% methanol. Buffered minimal glycerol media was MG plus 100 mM potassium phosphate at pH 6.0. Buffered glycerol complex media was buffered minimal glycerol media plus 2% peptone and 1% yeast extract. Buffered minimal methanol media was MM plus 0.5% methanol and 100 mM potassium phosphate at pH 6.0.

Expression and Purification of Recombinant m1-Toxin1. Ten micrograms of the expression vector with the cDNA for m1-toxin1 was linearized with SacI and used to transform P. pastoris GS115 cells, using the Easy-Comp kit (Invitrogen). Recombinant cells were screened for their ability to grow on histidine-deficient media. An immunoblot method was used for the initial screening of recombinant toxin production (Wung and Gascoigne, 1996). Recombinant cells were grown on a nitrocellulose disk placed on MG agar plates for 2 days at 30°C. The disk was transferred to a fresh plate containing MM agar and incubated at 30°C. Fresh methanol was added to the lid every 24 h to replenish that lost by metabolism and evaporation. After 48 h a fresh nitrocellulose disk was placed on top of the cells for 3 h at 30°C. The top disk was removed and the cells were washed off with phosphate-buffered saline. The nitrocellulose was blocked with 5% dry milk in 0.1% Tween 20 for 1 h, treated with rabbit antiserum to m1-toxin1 (1:100), and then exposed to a secondary anti-rabbit antibody coupled to horseradish peroxidase (Sigma) at a dilution of 1:20,000. Detection of recombinant expression was performed with an ECL-Plus kit (Amersham Pharmacia Biotech, Inc., Piscataway, NJ). Recombinant P. pastoris clones were grown in buffered glycerol complex media overnight at 30°C at 300 rpm to an A₆₀₀ of 2.0 to 6.0. Cells were pelleted at 4000g_max for 10 min at room temperature, resuspended in buffered minimal methanol media to an A₆₀₀ of 2.0, and incubated at 30°C and 300 rpm. The level of expression was assessed after 24 to 72 h. Media from expressing cells were clarified by centrifugation at 10,000g_max for 10 min. The media were then filtered through a 0.45-µm cellulose acetate filter (Corning, Palo Alto, CA) and diluted 10-fold with sodium acetate buffer at pH 4.5 (buffer A) to a final concentration of 25 mM. The sample was pumped onto a 5.0-ml cation-exchange column (SP Hi-Trap; Amersham Pharmacia Biotech) at a flow rate of 5.0 ml/min. The column was washed with buffer A and then eluted with 100 mM sodium chloride. The anti-M₁ activity of fractions was determined with a radioligand binding assay (see below). The most active fractions were pooled, the pH was adjusted to 6.0 with NaOH, BSA was added to 0.5%, and the sample was lyophilized. The dry sample was dissolved in 10% acetonitrile-0.1% trifluroacetic acid and loaded on a C₁₈ reversed phase column (10 × 250 mm; Vydac). The toxin was eluted with a gradient of 10 to 90% acetonitrile at 5.0 ml/min over 80 min, and fractions were assayed for activity. The active peak was lyophilized with 0.5% BSA at pH 6.0. The sample was resuspended in buffer A plus 20% acetonitrile, loaded on a cation-exchange HPLC column (7.5 by 50 mm; Vydac) and eluted with 0-500 mM NaCl at 1.0 ml/min during one hour. The pure toxin was desalted on a reversed-phase column (4.5 × 250 mm, Vydac), and samples of the single peak of toxin protein were taken for a Western immunoblot and for mass spectrometry. The remainder was lyophilized with 0.5% BSA at pH 6.0, resuspended in 50 mM sodium phosphate-1.0 mM EDTA at pH 7.41, and stored at 4°C.

Western Blot of Recombinant m1-Toxin1. One hundred nanograms of native and recombinant m1-toxin1 was loaded on a 10 to 20% Tricine gel and subjected to electrophoresis under denaturing conditions. The proteins were transferred to nitrocellulose, probed with antiserum to m1-toxin1, and detected using the ECL-Plus kit (Amersham Pharmacia Biotech).

Mass Spectrometry of Recombinant m1-Toxin1. Matrix-assisted laser desorption ionization/time of flight mass spectral analysis was performed at the Macromolecular Resources Center at the University of Colorado.

Expression and Purification of F38I and K65E. Ten micrograms of pF38I and pK65E was linearized with SacI and used to transfect competent P. pastoris GS115 cells. Colonies were screened for mutant toxin production using the same immunoblot method as for wild-type toxin. The clones that secreted the most toxin were chosen for further studies. The purification procedures for F38I and K65E were identical to those for m1-toxin1, except that the sodium acetate buffer used for cation-exchange steps had a pH of 4.0.

Radioligand Binding Assays. The antimuscarinic activity of recombinant toxins was determined by measuring their ability to block the binding of 1.0 nM [³H]NMS to membranes from CHO cells transfected with the cDNA for human M₁-M₅ receptors (Max et al., 1993a). Membranes were first incubated in 0.2 ml of 50 mM phosphate/1.0 mM EDTA buffer at pH 7.4 (phosphate-EDTA) and various concentrations of each toxin for 30 min at 25°C. [³H]NMS was added to 1.0 nM and a final volume of 1.0 ml (0.25 ml for F38I), and incubation was continued for 1 h. Membranes were collected on filters, dried, and radioactivity was counted. Nonspecific binding was determined by including 1 µM (±)quinuclidinyl benzilate and omitting the toxin during the preincubation step. The amount of membranes used was adjusted so that the amount of specific [³H]NMS binding was 1900 to 2800 cpm in each assay. Binding curves were analyzed with Prism 3.0 (GraphPad, San Diego, CA).

Measurement of Toxin Dissociation. Membranes from CHO cells expressing M₁ receptors were incubated in 0.1 ml of phosphate-EDTA buffer for 30 min at 25°C with sufficient amounts of each toxin to occupy 90% of M₁ receptors, as determined from competition assays. One nanomolar [³H]NMS was added to a final volume of 5.0 ml (50-fold dilution) and incubation was continued for 5 to 180 min. The rate of association of [³H]NMS to M₁ receptors was used to indicate the rate of dissociation of each toxin. Nonlinear regression was analyzed with Prism 3.0. The dissociation time given is the mean from three experiments ± S.E.M.

Allosteric Binding of Toxins. Membranes from CHO cells expressing M₁ receptors were incubated in phosphate-EDTA buffer with 1 nM [³H]NMS for 1 h at 25°C to saturate all M₁ receptors with NMS as determined by competition assays. The rate of dissociation of [³H]NMS from M₁ receptors was then determined after the addition of 1 µM atropine. To measure the effects of toxins on the dissociation of [³H]NMS, 250 to 1000 nM toxin was added with 1 µM atropine. Nonlinear regression was analyzed with Prism 3.0.

NMR Structure of MT2. The atomic coordinates of the solution NMR structure of MT2 (Ségalas et al., 1995) were kindly provided by Isabelle Ségalas (C.E. Saclay, Gif-Sur-Yvette, France). The molecule was modeled using an Insight II program on a Silicon Graphics Indigo² computer.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Cloning of cDNA for m1-Toxin1. Reverse transcription-PCR was carried out with mRNA purified from venom glands, using primers degenerate to portions of the m1-toxin1 protein sequence. These experiments yielded a PCR product of 93 base pairs, which was used to design primers specific to m1-toxin1 for use in RACE. The 5'- and 3'-RACE products were ligated together to form the full-length cDNA for m1-toxin1 (Fig. 3). This cDNA has 518 base pairs, of which 258 are the open reading frame. The deduced amino acid sequence of the open reading frame has a 21 amino acid signal peptide upstream of the mature amino acid sequence of m1-toxin1. The coding region of the cDNA for m1-toxin1, including its signal peptide, was subcloned into the expression vector pPIC3.5, placing it under control of the yeast alcohol oxidase 1 promotor.

View larger version (37K): [in this window] [in a new window]   Fig. 3.   Full-length cDNA and deduced amino acid sequence for m1-toxin1. The predicted site of signal peptide cleavage is indicated by the triangle (). Amino acid sequences that were used to design the primers DEG1 and DEG2 are numbered 1 and 4, and the gene specific primers GSP1 and GSP2 are numbered 3 and 2, respectively. The asterisk denotes the stop codon.

Expression and Purification of Recombinant m1-Toxin1, F38I, and K65E. Recombinant clones were screened on MG agar deficient in histidine, and screened for toxin production using a modified immunoblot assay. Once a suitable clone was selected, the time course for toxin expression was examined. Maximal expression of m1-toxin1 was achieved in 48 to 60 h (Fig. 4). Similar results were observed for F38I and K65E (data not shown).

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Time course of expression of recombinant m1-toxin1. Aliquots of yeast media (100 µl) were assayed for their ability to block the binding of 1.0 nM [3H]NMS to cloned M1 receptors (n = 3). Media from cells transfected with parent plasmid pPIC3.5 served as controls (n = 3). Maximal expression was at 48 to 60 h.

View larger version (12K): [in this window] [in a new window]   Fig. 5.   Cation-exchange purification of recombinant m1-toxin1. This figure shows the third step in the purification of recombinant m1-toxin1. The lyophilized sample obtained after reversed phase purification was resuspended in buffer A plus 20% acetonitrile, loaded on a cation-exchange HPLC column and eluted with 0 to 500 mM NaCl at 1.0 ml/min during 1 h. The peak fractions of the peak of protein marked with an asterisk were used for all further studies (Western blot, mass spectrometry, desalting by reversed phase HPLC).

View larger version (98K): [in this window] [in a new window]   Fig. 6.   Immunoblot of recombinant m1-toxin1. Aliquots of recombinant m1-toxin1 (lane 1) and m1-toxin1 purified from venom (lane 2) were loaded on a 10 to 20% Tricine gel and subjected to electrophoresis under denaturing conditions. The protein was transferred to nitrocellulose and probed with antisera to m1-toxin1. The recombinant toxin migrated at the same rate as the native toxin.

View larger version (17K): [in this window] [in a new window]   Fig. 7.   Mass spectrometry of recombinant m1-toxin1. The parent peak measured 7470.6 Da, in agreement with a theoretical mass for m1-toxin1 of 7472 Da. The lower mass peaks are double and triple mass species, respectively.

Binding of Recombinant m1-toxin1, F38I, and K65E to Muscarinic Receptors. Preincubation of M₁ receptors with recombinant m1-toxin1, F38I, and K65E all blocked the binding of 1.0 nM [³H]NMS to cloned M₁ receptors (Fig. 8). Similar amounts of m1-toxin1 and K65E were required to block similar amounts of M₁ receptors, indicating the presence of similar amounts of the active toxins. Nonetheless, the molar ratio of each toxin to receptor protein far exceeded a stoichiometric ratio (see Discussion). No blockade of binding was observed with M₂-M₅ receptors at toxin concentrations up to 2.2 µM for m1-toxin1 and K65E, and up to 5.0 µM for F38I (data not shown).

View larger version (24K): [in this window] [in a new window]   Fig. 8.   Binding of recombinant toxins to muscarinic receptors. Membranes containing cloned human M1 receptors were exposed first to recombinant m1-toxin1, F38I, or K65E, and then to 1.0 nM [3H]NMS (n = 3 for each toxin). The total specific binding in each assay was 1900 to 2800 cpm. IC50 values for m1-toxin1 and K65E were calculated on the basis of the volume of buffer used during the preincubation step, because these toxins bound irreversibly (Fig. 9). The IC50 value for F38I was calculated on the basis of the volume used during exposure to [3H]NMS, because this toxin bound reversibly. IC50 values for m1-toxin1, F38I, and K65E were 3.0 ± 0.2, 7.3 ± 0.26, and 1.7 ± 0.16 nM, respectively. No blockade of binding was observed with M2-M5 receptors at toxin concentrations up to 2.2 µM for m1-toxin1 and K65E, and 5.0 µM for F38I (data not shown).

Duration of Binding of m1-Toxin1, K65E, and F38I to M₁ Receptors. The stability of M₁ receptor-toxin complexes was examined by observing the ability of [³H]NMS (1.0 nM) to bind to receptors preincubated with a saturating amount of each toxin. Figure 9 shows that M₁ receptors preincubated with m1-toxin1 or K65E did not bind [³H]NMS in 3 h at 25°C. However, M₁ receptors preincubated with F38I showed a rapid increase in [³H]NMS binding sites (Fig. 9). Thus, recombinant m1-toxin1 and K65E bound irreversibly, whereas F38I bound reversibly. Because F38I bound reversibly, its affinity for M₁ receptors must be lower than that of m1-toxin1 and K65E.

View larger version (21K): [in this window] [in a new window]   Fig. 9.   Stability of toxin-receptor complexes. M1 receptors were exposed first to a IC90 concentration of m1-toxin1 (15 nM), F38I (100 nM), or K65E (11 nM) and then to 1.0 nM [3H]NMS (n = 3 for each toxin). No toxin was used for the controls. The rate of appearance of binding sites for [3H]NMS at 25°C was taken as the rate of dissociation of each toxin. After exposure of M1 receptors to F38I the binding of [3H]NMS was near control levels at 45 min, indicating that F38I bound reversibly. In contrast, receptors exposed with m1-toxin1 or K65E showed no increase in [3H]NMS binding sites throughout the experiment, indicating that [3H]NMS binding sites were still occluded by these toxins.

Allosteric Binding of Toxins to M₁ Receptors. The ability of recombinant toxins to stabilize the binding of [³H]NMS to M₁ receptors was tested by measuring the rate of dissociation of [³H]NMS from toxin-NMS-receptor complexes. [³H]NMS had a half-life of 2.74 ± 0.14 min with no toxin (Fig. 10). In the presence of m1-toxin1 the half-life increased 6-fold to 15.98 ± 0.31 min, in keeping with prior results for the native toxin (Max et al., 1993b; Carsi and Potter, 2000). In contrast, neither F38I nor K65E had an observable effect on the dissociation of NMS (Fig. 10). These results indicate that K65E and F38I do not bind to NMS-M₁ receptor complexes with the same affinity as m1-toxin1.

View larger version (24K): [in this window] [in a new window]   Fig. 10.   Stability of toxin-NMS-M1 receptor complexes. M1 receptors were incubated with 1.0 nM [3H]NMS at 37°C and then exposed to a saturating concentration of atropine (1.0 µM) with or without m1-toxin1 (50 nM), K65E (250 nM), or F38I (1.0 µM). The rate of dissociation of 3H- NMS from M1 receptors at 37°C with atropine alone was taken as the half-life for [3H]NMS (2.74 ± 0.14 min). The half-life increased to 15.98 ± 0.31 min in the presence of m1-toxin1, a 6-fold increase (p < 0.001). K65E and F38I did not slow the dissociation of [3H]NMS from M1 receptors, indicating that these toxins were unable to bind to receptors occupied with [3H]NMS (n = 3 for all toxins)..

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

The amino acid sequence of m1-toxin1 is highly homologous to those of cloned cDNAs for more than 100 short-chain toxins from members of families Elapidae and Hydrophiidae (GenBank), despite the fact that the snakes live in environments as diverse as trees and oceans, and the toxins bind to targets as diverse as nicotinic and muscarinic receptors, acetylcholinesterase, calcium channels, and voltage-gated potassium channels (Harvey, 1991). The signal peptide sequences of these toxins are 100% conserved at the amino acid level and 98% identical at the nucleotide level, implying a common ancestral origin for their genes (Ohno et al., 1998). The coding sequence of cloned m1-toxin1 established the correct full sequence of this toxin for the first time (Potter et al., 1996), because the original sequence for m1-toxin contained two errors (His²⁸ and Trp²⁹; Max et al., 1993a). Trp²⁸ and Gln²⁹ have since been found in all the m1-isotoxins (Carsi and Potter, 2000). We did not obtain any full-length cDNAs corresponding to other m1-toxins. The full sequence of MT7 has been determined recently and was found to be the same as that of m1-toxin1 (Näsman et al., 2000).

The cDNAs for m1-toxin1, F38I, and K65E were expressed in P. pastoris so that the toxins would be correctly folded, processed, and secreted. The use of methylotropic yeast also meant that each toxin was secreted into media containing only salts and methanol, simplifying the purification of each toxin. Recombinant m1-toxin1 had the same linear mass as native m1-toxin1, and it bound specifically, irreversibly, and allosterically like the native toxin, demonstrating that our yeast expression system was satisfactory for producing m1-toxin1. The yeast also produced useful amounts of F38I and K65E.

Given the remarkable selectivity of m1-toxin1 for M₁ receptors, it was anticipated that any mutation of this toxin would alter its selectivity for M₁-M₅ receptors. To our surprise, we found that recombinant m1-toxin1, F38I, and K65E were all selective for M₁ versus M₂-M₅ receptors. So, positions 38 and 65 do not play a critical role in determining selective binding, and further experiments are necessary to determine which amino acids do confer subtype selectivity. Jolkkonen (1996) suggested that the amino acids at the tip of loop 2 might be critical, because MT4 and MT are identical except for three amino acids at the tip of this loop (Fig. 1), and these two toxins differ dramatically in their specificity (Fig. 1). But the corresponding three amino acids in m1-toxin1 and m4-toxin differ relatively little, despite the markedly different specificities of these toxins.

m1-Toxin1 is also unique because of its irreversible binding to M₁ receptors (Max et al., 1993a,b,c; Carsi and Potter, 2000; Näsman et al., 2000; Olianas et al., 2000). Recombinant m1-toxin1 bound irreversibly as expected. K65E also bound irreversibly, showing that position 65 is not critical for irreversible binding. Because recombinant m1-toxin1 and K65E each bound irreversibly to unoccupied M₁ receptors, it is not possible to calculate an affinity for either toxin. F38I dissociated rapidly from M₁ receptors, so the Phe³⁸ of m1-toxin1 must be important for irreversible binding. F38I may prove useful for functional studies when reversible M₁-blockade is desirable. We cannot calculate a reliable K_d for F38I at this time because of evidence of major losses of toxins during their lyophilization (see last paragraph).

Some inferences can be made about the role of Phe³⁸ in m1-toxin1, based on the present results with F38I and the solution NMR structure of MT2 (Ségalas et al., 1995). The homologous residue to Phe³⁸ in MT2 is Ile³⁸. Ile³⁸ is in the middle of a -sheet in loop 2 of MT2 (Fig. 2), and it has the second highest number of contacts with adjacent residues of any amino acid in the entire MT2 sequence. Assuming that m1-toxin1 and MT2 have a similar overall structure, mutation of Phe³⁸ to Ile³⁸ in m1-toxin1 should significantly alter the structure of the toxin in loop 2. The altered binding characteristics of F38I thus indicate that the structure of loop 2 is important for a tight molecular fit between m1-toxin1 and M₁ receptors.

Both native m1-toxin1 (Max et al., 1993b) and recombinant m1-toxin1 slow the dissociation of [³H]NMS from M₁ receptors 6-fold. F38I did not slow the dissociation of [³H]NMS, which is not surprising given its reversible binding to unliganded M₁ receptors. However, K65E also did not slow the dissociation of [³H]NMS, even though K65E bound irreversibly to unliganded M₁ receptors. Thus, the positive C-terminal Lys⁶⁵ of m1-toxin1 seems to be critical for the formation of ternary complexes of toxin, [³H]NMS and M₁ receptors. Again, some inferences can be made about the role of Lys⁶⁵ in m1-toxin1 for its binding to M₁ receptors, based on the solution structure of MT2. The homologous C-terminal residue to Lys⁶⁵ in m1-toxin1 is Glu⁶⁵ in MT2. Glu⁶⁵ protrudes from the convex face of MT2 and makes few contacts with its adjacent residues (Fig. 2). The C-terminal regions of all the toxins shown in Fig. 1 are nearly identical, so it is unlikely that the C-terminal residue of any of these toxins (including Lys⁶⁵ in m1-toxin1) is important for their structural integrity. Nonetheless, and despite its irreversible binding to unliganded M₁ receptors, K65E did not block the dissociation of [³H]NMS from M₁ receptors. Our interpretation of this result is that there is an interaction between Lys⁶⁵ on the convex face of m1-toxin1, and an extracellular part of M₁ receptors. Because the only difference between m1-toxin1 and K65E is a negative C-terminal, it is possible that the highly basic C-terminal Lys of m1-toxin1 interacts with an acidic residue in one of the outer loops or the N terminus of M₁ receptors. Although any of the acidic residues on M₁ receptors might be involved, Glu¹⁷⁰ in the middle of outer loop 2, Asp³⁹³ in the middle of loop 3, and Glu⁴⁰¹ are of particular interest, because they are not found in M₂-M₅ receptors (Hulme et al., 1990). Glu⁴⁰¹ is especially intriguing because it lies just outside the membrane surface next to Trp⁴⁰⁰, which is essential for the allosteric effect of gallamine, and Glu⁴⁰¹ just overlies Tyr⁴⁰⁴ and Tyr⁴⁰⁸ in transmembrane segment 7, which are important for the binding of NMS (Matsui et al., 1995). Site-directed mutagenesis of Glu⁴⁰¹ in the M₁ receptor is therefore in progress, to evaluate whether Lys⁶⁵ on the toxin, and Glu⁴⁰¹ on the receptor, may interact.

It is well known that native m1-toxin1 (Max et al., 1993b) and synthetic MT7 (synthetic m1-toxin1; Olianas et al., 2000) do not block the specific binding of [³H]NMS to M₁ receptors completely, and recombinant m1-toxin1 showed the same effect (Fig. 8). There are a number of possible explanations for this effect, including negative cooperativity between the binding of toxin and NMS at different sites (Olianas et al., 2000), reversible blockade of NMS by partly denatured toxin, and different access of NMS and toxin to intravesicular sites in membrane preparations.

We experienced two significant problems with the amounts of recombinant toxins that were recovered after expression. The first was that the levels of expression were inadequate for structural studies by NMR or circular dichroism. This problem might be ameliorated in future studies by using synthetic yeast codons instead of native cDNA codons for expression. The second problem was that it was necessary to use much larger quantities of recombinant m1-toxin1 and K65E for irreversible blockade (by a factor of 100 or more) than should be required for stoichiometric blockade. By comparison, M₁ receptors can be irreversibly blocked with synthetic m1-toxin1 or radioiodinated synthetic m1-toxin1 using only a slight molar excess of toxin over receptor (Potter, 2001). This second problem could have appeared in the present studies because of incorrect folding of m1-toxin1 and K65E. However, the fact that we obtained a single toxin protein by HPLC and by mass spectrography makes it unlikely that there was variable folding or impure protein. Moreover, the problem of nonstoichiometric binding has been present in every study in this laboratory with native m1-toxin1 (Max et al., 1993a; Liang et al., 1996; Carsi and Potter, 2000; Potter, 2001), and we think it is due to extensive surface denaturation of toxins when they are lyophilized. We attempted to avoid the inactivation of recombinant toxins by adding BSA before lyophilizations, but were unable to prevent major losses of the active toxins. We now know that losses can be avoided by concentrating each toxin by centrifugation over molecular cutoff filters, but this was not done in the present studies. It may be noted that the partial denaturation of K65E does not explain the inability of the active toxin to bind to [³H]NMS-receptor complexes and to slow the dissociation of [³H]NMS, because we used concentrations of this toxin for dissociation rate experiments (Fig. 10) that were capable of causing full blockade of unoccupied M₁ receptors.