Latrotoxin Binding and ␣-Latrotoxin-Stimulated Secretion a STUDY with CALCIUM-INDEPENDENT RECEPTOR of ␣-LATROTOXIN (CIRL) DELETION MUTANTS*

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 274, No. 6, Issue of February 5, pp. 3590–3596, 1999 © 1999 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A. Structural Requirements for ␣-Latrotoxin Binding and ␣-Latrotoxin-stimulated Secretion A STUDY WITH CALCIUM-INDEPENDENT RECEPTOR OF ␣-LATROTOXIN (CIRL) DELETION MUTANTS*

(Received for publication, October 26, 1998, and in revised form, November 25, 1998)

Valery Krasnoperov‡, Mary A. Bittner§, Ronald W. Holz§, Oleg Chepurny‡, and Alexander G. Petrenko‡¶ʈ From the ‡Departments of Pharmacology, ¶Physiology and Neuroscience, and ¶Environmental Medicine, New York University Medical Center, New York, New York 10016 and the §Department of Pharmacology, University of Michigan Medical School, Ann Arbor, Michigan 48109

Stimulation of neurotransmitter release by ␣-latro- ␣-latrotoxin (CIRL)1 (5, 6). CIRL is thought to be more impor- toxin requires its binding to the calcium-independent tant for ␣-latrotoxin effects in neurons than neurexin I␣ be- receptor of ␣-latrotoxin (CIRL), an orphan neuronal G cause ␣-latrotoxin can stimulate neurotransmitter release from protein-coupled receptor. CIRL consists of two nonco- neurons in Ca2ϩ-free media (7, 8). CIRL, also called latrophilin, valently bound subunits, p85, a heptahelical integral belongs to a family of closely related orphan G protein-coupled membrane protein, and p120, a large extracellular receptors (GPCRs) homologous to the secretin receptor family polypeptide with domains homologous to lectin, olfacto- (9, 10). In this family of three closely homologous proteins, medin, mucin, the secretin receptor family, and a novel CIRL-1 is a brain-enriched high affinity ␣-latrotoxin receptor, structural motif common for large orphan G protein- whereas CIRL-2 is a ubiquitously expressed low affinity recep- coupled receptors. The analysis of CIRL deletion mu- tor of the toxin (11). tants indicates that the high affinity ␣-latrotoxin-bind- The CIRL receptors have an unusual structure for GPCRs. ing site is located within residues 467–891, which First, they are significantly larger (about 200 kDa) than most of comprise the first transmembrane segment of p85 and the GPCRs. Second, they consist of two heterologous subunits the C-terminal half of p120. The N-terminal lectin, olfactomedin, and mucin domains of p120 are not required because of endogenous proteolytic processing of the precursor for the interaction with ␣-latrotoxin. Soluble p120 and protein. The site of this cleavage is located 18 residues up- all its fragments, which include the 467–770 residues, stream from the first transmembrane segment. As a result, bind ␣-latrotoxin with low affinity suggesting the impor- mature CIRL consists of two noncovalently bound subunits, tance of membrane-embedded p85 for the stabilization p120 and p85. p120 is a hydrophilic protein that is soluble and of the complex of the toxin with p120. Two COOH-termi- secreted if expressed separately from p85, whereas p85 has nal deletion mutants of CIRL, one with the truncated structural features typical of a generic GPCR although with an cytoplasmic domain and the other with only one trans- unusually large cytoplasmic tail (9, 11). membrane segment left of seven, supported both ␣-lat- There is ample evidence that ␣-latrotoxin receptors are crit- rotoxin-induced calcium uptake in HEK293 cells and ically required for the effects of ␣-latrotoxin (1, 12, 13). How- ␣-latrotoxin-stimulated secretion when expressed in ever, the mechanism of signaling downstream of the receptors chromaffin cells, although with a different dose depend- is not known. The heptahelical structure of CIRL suggests its ence than wild-type CIRL and its N-terminal deletion function as a regulator of a G protein pathway. However, no mutant. Thus the signaling domains of CIRL are not coupling of CIRL to any G protein has been convincingly critically important for the stimulation of exocytosis in shown. Moreover, no direct data are currently available to ␣ intact chromaffin cells by -latrotoxin. prove that ␣-latrotoxin acts as an agonist or antagonist of its receptors. To analyze the structural requirements for ␣-latrotoxin bind- ␣ -Latrotoxin, a potent natural stimulator of secretion from ing and ␣-latrotoxin stimulatory function, we generated three neurons and secretory cells, has two structurally and pharma- series of CIRL deletion mutants. Soluble fragments of the ex- cologically distinct classes of high affinity receptors (1). The tracellular region of CIRL were used to map the ␣-latrotoxin- ␣ ␣ calcium-dependent receptor of -latrotoxin or neurexin I is a binding site. On the basis of this information, N-terminally large (160–220 kDa) cell surface membrane protein existing in truncated membrane-bound forms of CIRL were produced, multiple isoforms (2, 3). It has one transmembrane segment which were shown to retain high affinity ␣-latrotoxin binding and structurally resembles cell adhesion proteins (4). A second activity. Finally, deletions in the COOH-terminal region of high affinity receptor is the calcium-independent receptor of CIRL were produced to remove the domains potentially in- volved in receptor signaling. The constructs lacking the CIRL * This study was supported by public health service Grants cytoplasmic tail and six of its seven transmembrane segments R01NS35098 and R01NS34937 from the NINDS, National Institutes of appeared to be fully functional in terms of ␣-latrotoxin binding, Health (to A. G. P.) and Grant R01DK27959 from the NIDDK, National Ca2ϩ influx in HEK293 cells, and coupling of the toxin to Institutes of Health (to R. W. H.). The costs of publication of this article secretion in transfected chromaffin cells. Our data suggest that 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. ʈ To whom correspondence should be addressed: Dept. of Pharmacol- 1 The abbreviations used are: CIRL, calcium-independent receptor of ogy, New York University Medical Center, 550 First Ave., MSB-202, ␣-latrotoxin; GPCR, G protein-coupled receptor; PCR, polymerase chain New York, NY 10016. Tel.: 212-263-5969; Fax: 212-263-7133; E-mail: reaction; GPS, GPCR proteolysis site; PSS, physiological salt solution; [email protected]. STP domain, Ser, Thr, and Pro-rich domain.

3590 This paper is available on line at http://www.jbc.org Interaction of ␣-Latrotoxin with CIRL 3591

G protein-mediated signaling is not critically important for the blotting with anti-p120 antibody. ␣ ␣-latrotoxin-stimulated secretion in chromaffin cells. -Latrotoxin Binding Analysis of Cell Membranes—Three days after transfection, the COS cells were harvested on ice in physiological saline and centrifuged. The pellet was resuspended in a 50 mM Tris-HCl, 150 EXPERIMENTAL PROCEDURES mM NaCl, 2 mM EDTA, and 1% bovine serum albumin, pH 8.0, incuba- ␣-Latrotoxin was purified from lyophilized black widow spider tion buffer, and 10% of the cell material harvested from one 100-mm 125 glands and radioactively labeled with I using the chloramine T pro- Petri dish was incubated with 5 nM 125I-␣-latrotoxin in the final volume cedure. The toxin was immobilized on BrCN-Sepharose as described (2). of 200 ␮l for 15 min. The binding reaction suspensions were diluted Soluble Deletion Mutants of the Extracellular Region of CIRL—The with the ice-cold incubation buffer and immediately centrifuged at pCDR7N construct encoding the extracellular region of CIRL (residues 14,000 rpm in the Eppendorf centrifuge for 10 min. The pellets were 1–856) with COOH-terminal His6 tag was described previously (9). The counted for radioactivity in a ␥-counter. The nonspecific binding was pCDR120 construct encoding the p120 subunit of CIRL precisely was measured in the presence of 100 nM cold toxin. prepared by ligating the AgeI/XbaI-digested PCR product obtained with Chromaffin cell and HEK293 cell transfections and functional anal- primers ACATCTAGAGGTGGCTGCAGGCACATGTGGTA and ACA- ysis of CIRL mutants were performed as described previously (9, 14). GGCCCAGCCGGCCAACACCATCAAGCAGAACAGCC on the 87-7 CIRL cDNA clone as a template into pCDR7 plasmid cut with AgeI/ RESULTS XbaI. The structure of the PCR-derived region of the final plasmid was Domain Structure of CIRL—Computer-assisted analysis of verified by sequencing. The recombinant DNA fragments encoding other deletion mutants were prepared by high fidelity PCR with Pfu the CIRL protein sequence reveals a number of distinct struc- polymerase and synthetic oligonucleotide primers containing SfiI tural domains (Fig. 1). A central region of CIRL (residues (sense) or XbaI (antisense) restriction sites. The expression constructs 850–1100) shows significant homology to the members of the were prepared by ligating the SfiI/XbaI-digested PCR products into secretin family of GPCRs (9). According to the hydrophobicity SfiI/XbaI-digested pSecTag plasmid (Invitrogen) in frame with the His 6 plot of CIRL, this region contains seven long hydrophobic tag. Thus prepared constructs encoded the following residues of CIRL: stretches, a hallmark of GPCRs. In GPCRs, these hydrophobic pSTR7-1, residues 25–598; pSTR7-2, residues 25–856; pSTR7-3, resi- ␣ dues 25–631; pSTR7-4, residues 25–705; pSTR7-5, residues 25–770; sequences are -helical rods that form a compact oval-shaped pSTR7-6, residues 128–856; pSTR7–9, residues 538–856; pSTR7-16, integral transmembrane cluster (15). Similar to other GPCRs, residues 467–705; pSTR7-20, residues 185–856. The plasmids were three intracellular and three extracellular hydrophilic loops in transfected into COS-7 cells using the LipofectAMINE method accord- between transmembrane helices can be identified in CIRL. ing to Life Technologies, Inc. protocol. After 3 days, the conditioned The intracellular COOH-terminal region of CIRL (residues media and cells were harvested and analyzed for the presence of the 1100–1471) is unusually large for an average GPCR. It con- recombinant protein by precipitation with nickel-agarose followed by Western blotting with anti-p120 antibody. tains a pair of vicinal Cys residues typical for the GPCR palmi- N-terminal Deletion Mutants of CIRL—To generate the N-terminal toylation site and several proline reach clusters. It has no deletion mutants anchored to the membrane-bound fragment of CIRL, significant homology to any known protein except for two other AgeI/XbaI-digested pSTR7-6, -7, -8, and -9 plasmids were ligated with a members of the CIRL family, CIRL-2 and -3 (11). DNA fragment of 3,060 base pairs obtained by digesting pCDR7 with In contrast, several domains of the extracellular N-terminal AgeI/XbaI. This insert encoded a short COOH-terminal region of p120 region of CIRL show significant homology with various recep- and the entire p85 subunit. These constructs encoded the following residues of CIRL: pSTR7-6M, residues 128–1471; pSTR7-7M, residues tor and nonreceptor proteins. The signal peptide sequence is 394–1471; pSTR7-8M, residues 467–1471; pSTR7-9M, residues 538– followed by a cysteine-rich domain (residues 30–120) homolo- 1471. COS cells were transfected with these plasmids, harvested in 3 gous to sea urchin egg D-galactoside-specific lectin (GenBank days, and analyzed for ␣-latrotoxin binding activity as described below. number P22031) and plant ␤-galactosidase (GenBank number Purification of the Recombinant Extracellular Domain of CIRL and Z99708). The same domain is found in a Caenorhabditis el- ␣ the Analysis of Its Affinity to -Latrotoxin—COS cells were transfected egans homolog of CIRL (GenBank number Z54306). The adja- with an expression plasmid pCDR7N encoding the entire N-terminal cent domain (residues 120–400) is similar to olfactomedin extracellular region of CIRL (1–837 residues) with a His6 tag at the COOH terminus. In 2 days, the cell media were collected, and 10 ml of (GenBank number AF028740), a major structural block in the media were incubated with 300 ␮l of nickel-agarose overnight at 4 °C extracellular matrix of the olfactory neuroepithelium and sev- with gentle agitation. The matrix was washed 4 times with 50 mM eral structurally related proteins including the pancortin fam- Tris-HCl and 500 mM NaCl, pH 7.5, and eluted with 1 ml of 100 mM ily (GenBank number Q62609, D78264, D78262, Q99784, imidazole in the same buffer. The binding activity of the eluted proteins AB006688, AF049796, AF035301, Q99972, AF039869). Their was measured by a solid-phase method on a 96-well plate (5). 125I-␣- physiological role is unclear. Latrotoxin was added to a final concentration of 0.5–60 nM. Specific binding was measured as the difference between total binding and The next structural domain (residues 400–470) is enriched nonspecific binding in the presence of 900 nM unlabeled ␣-latrotoxin. in Ser, Thr, and Pro residues and shows insignificant homology COOH-terminal Deletion Mutants of CIRL—Two COOH-terminal to mucin and other Pro-rich proteins. We will therefore refer to deletion mutants of CIRL were prepared. The first one, pCDR-7TMR, this domain as the STP domain. The STP domain is followed by included the p120 subunit, the seven transmembrane region, and 49 a short region (residues 470–540) with a Cys-rich motif CX - N-terminal amino acid residues of the COOH-terminal cytoplasmic tail 9–10 WX CX CX WX C identified by ⌿-Blast search in (residues 1–1149 of CIRL). The second one, pCDR-1TMR, included 9–12 10–17 4–6 8–16 p120, the first transmembrane region, and the first intracellular loop of the extracellular domains of GPCRs, which belong to the se- p85 (residues 1–891). To generate a pCDR-7TMR expression vector, cretin receptor family. Interestingly, in “normal” receptors of two complementary 5Ј-phosphorylated oligonucleotides, 5Ј-pCCGGAG- this family (e.g. vasoactive intestinal peptide receptor, pitui- CGGCCGCTGAT and 5Ј-pCTAGATCAGCGGCCGCT, were used in the tary adenylate cyclase-activating polypeptide receptor, secretin ligation reaction with the BspEI/XbaI-digested pCDR7 (9) vector. To receptor, calcitonin receptor, corticotropin-releasing factor re- generate a pCDR-1TMR expression vector, the 7313-base pair product ceptor, growth hormone-releasing hormone receptor, glucagon- of AgeI/XbaI digestion of pCDR7 was ligated with a 421-base pair fragment obtained by AgeI/XbaI digestion of a 1267-base pair PCR like peptide receptor, etc.) this motif is located very close to the product where pCDR7 was used as a template and the primers ACAG- transmembrane core. In the CIRL family and in a number of GCCCAGCCGGCCAATCTGCATGTGTCCCCTGAGCT and TTTCTA- other large orphan GPCRs (GenBank numbers U39848, GATCAGCGGTCGGTCTGCAG were used. Z54306, D87469, AB011529, AB011528, AB005297, AB011536, ␣ -Latrotoxin Binding Analysis of Soluble Proteins—0.7 ml of media AB011122), this motif is located several hundred amino acid from transfected COS cells was cleared by centrifugation and incubated residues from the membrane segments, and therefore we may with 15 ␮lof␣-latrotoxin-Sepharose overnight at 4 °C. The matrices were washed 3 times with 1.25 ml of ice-cold 50 mM Tris-HCl, 150 mM assume that the large orphan receptors represent a separate NaCl, 2 mM EDTA, pH 8.0, and eluted with SDS electrophoresis sample subfamily within the family of the secretin receptor. buffer. The eluates were electrophoresed and analyzed by Western The region between residues 541 and 800 has low homology 3592 Interaction of ␣-Latrotoxin with CIRL

FIG.1. Domain structure of CIRL. The domains with homology to lectin and olfactomedin are labeled as such. The STP domain is the region weakly homologous to mucin and enriched in Ser, Thr, and Pro. Cys-rich motif or SR, a signature conserved in the N-terminal extracellular regions of the secretin receptor family members. An ellipse indicates the location of GPS (GPCR proteolysis site), a structural motif common for large orphan GPCRs homologous to CIRL. An arrow on the top indicates the cleavage site of the signal peptide. All cysteine/cysteine residues are numbered and marked as SH. PM, plasma membrane.

to brain-specific angiogenesis inhibitor-3 (GenBank number cloned into a pSecTag eucaryotic expression plasmid (Invitro- AB005299), another large orphan GPCR that among all known gen), which contained a signal peptide sequence that allowed large GPCRs is most similar to the members of the CIRL extracellular secretion of soluble proteins and a COOH-termi- family. The recently discovered brain-specific angiogenesis in- nally fused His6 tag that allowed easy purification of the re- hibitor family was implicated in tumor angiogenesis regula- combinant protein. tion; brain-specific angiogenesis inhibitor-1 expression is reg- The resulting constructs were transfected into COS cells. ulated by p53 (16, 17). The cells and media were harvested and analyzed for the pres- Finally, in the COOH terminus of p120 immediately adja- ence of recombinant proteins by the adsorption onto nickel- cent to the putative site of CIRL proteolysis a novel structural agarose followed by Western blotting with anti-p120 antibody. motif is found that is characteristic for about a dozen large Among tested constructs, pCDR-120, pCDR7N, and pSTR7-2, orphan GPCRs of the secretin receptor family (GenBank num- -6, -7, and -20 were expressed and secreted in the medium. bers U39848, U76764, P48960, Q61549, Q14246, AC004262, Proteins encoded by pSTR7-1, -3, -4, -5, -9, and -16 were ex- D87469, AB011529, AB011528, AB011536, AB005297, X81892, pressed well but accumulated inside the cells (data not shown). AB005298, AF006014, AB011122). We propose to name this It is interesting to note that most of the secreted proteins were motif GPS for GPCR proteolysis site. The characteristic feature N-terminally truncated, whereas most of the nonsecreted dele- of the GPS domain is a cysteine signature including CXC, two tion mutants were COOH-terminally truncated. This finding additional cysteine residues, and two tryptophan residues at raises a possibility that the extracellular region of CIRL con- fixed positions. When conserved residues of the GPS motif were tains in its COOH-terminal part a signal sequence that regu- 2 mutated, CIRL could no longer be proteolyzed endogenously. lates intracellular sorting and trafficking of the protein. ␣ Mapping the -Latrotoxin-binding Site in CIRL—We showed To analyze the ␣-latrotoxin binding activity of the recombi- earlier that the recombinant N-terminal extracellular region of nant proteins, the conditioned media or cell extracts were ad- ␣ CIRL binds efficiently to immobilized -latrotoxin (9). To iden- sorbed onto ␣-latrotoxin-Sepharose followed by Western blot- ␣ tify the -latrotoxin binding domain(s) of CIRL more precisely, ting with anti-p120 antibody (Fig. 3). The constructs pCDR- we have generated a series of deletion mutants of its extracel- 120, pCDR7N, and pSTR7-2, -5, -6, -7, and -20 specifically lular region. The desired DNA fragments were prepared by interacted with the toxin, whereas pSTR7-1, -3, -4, -9, and -16 high fidelity PCR with synthetic oligonucleotide primers that did not. The shortest N-terminally truncated mutant that still were designed according to the putative domain borders of the bound the toxin was pSTR7-7. A shorter construct pSTR7-9 extracellular region of CIRL (Fig. 2). The PCR fragments were (residues 538–856) failed to interact with the toxin. The analysis of ␣-latrotoxin binding activity of COOH-terminally trun- 2 V. Krasnoperov, K. Ichtchenko, and A. G. Petrenko, manuscript in cated mutants demonstrated that the downstream border of preparation. the toxin-binding site may be located close to the membrane. Interaction of ␣-Latrotoxin with CIRL 3593

FIG.2. Deletion mutants of CIRL. The structure of recombinant proteins is shown with domains labeled. The numbers on the left identify the borders of the deletion mutants according to the CIRL protein sequence. SR, sarcoplasmic retic- ulum; SP, signal peptide.

FIG.3. Localization of the ␣-latrotoxin-binding site. 1mlof conditioned media of COS cells transfected with CIRL deletion mutants (constructs pSTR7-2, -6, -7, -16, -20, and pCDR7N) or 50 ␮l of cells (pSTR7-1, -3, -4, -5, and -9) extracted with 1 ml of a buffer with 2% Triton X-100 was incubated with 10 ␮lof␣-latrotoxin-agarose overnight with gentle agitation. After three washes, the pellets were eluted in 50 ␮l of SDS-sample buffer/dithiothreitol, boiled for 3 min, and analyzed by Western blotting with anti-p120 antibody (L). To verify the expression, 2 ␮l of transfected cells were analyzed directly (E).

The pCDR7N protein, which is the entire extracellular region of CIRL (residues 1–856) and recombinant p120 (residues FIG.4.High affinity interaction of ␣-latrotoxin with N-termi- 1–837, construct pCDR-120), bound to the toxin quite well. A nal deletion mutants of CIRL. A, COS cells transfected with pCDR7 shorter protein pSTR7-5 (residues 25–770) bound to ␣-latro- (wild-type CIRL or WT), salmon sperm DNA (SS) as a negative control, or pSTR7-6M, -7M, -8M, -9M were harvested and incubated with 5 nM toxin much less efficiently, whereas pSTR7-4 (residues 25–705) 125I-␣-latrotoxin. The total binding of the labeled toxin is shown as did not interact with the toxin at all. Together, these data black bars. The nonspecific binding measured in the presence of a suggest that the residues critical for ␣-latrotoxin binding are 20-fold excess of unlabeled toxin is shown as white bars. Each experi- located in the COOH-terminal half of p120 with a significant ment was done in duplicate, and the average is shown (deviation was less than 3% for each measurement shown). The ␣-latrotoxin binding site of interaction around residue 770. affinity of the deletion mutants pSTR7-7M (B) and pSTR7-8M (C) was Soluble Fragments of the Extracellular Domain of CIRL Are compared with the affinity of CIRL by the Scatchard plot analysis. Low Affinity ␣-Latrotoxin-binding Proteins—To further analyze the interaction of ␣-latrotoxin with the extracellular do- well to immobilized ␣-latrotoxin, we assumed that they were main of CIRL and its fragments, we tested the ability of these ␣-latrotoxin-binding proteins. However, their affinity to the soluble proteins to inhibit the binding of iodinated ␣-latrotoxin toxin was lower than the affinity of endogenous CIRL, and to brain membranes. Surprisingly, no significant inhibition therefore they could not compete effectively in the concentra- was detected with any protein tested including the entire un- tion range tested. To estimate the affinity of p120 and its cleaved extracellular domain encoded by pCDR7N, p120 (these deletion mutants to ␣-latrotoxin, we used the solid-phase assay two constructs were not based on PCR-generated sequences), developed earlier for the analysis of ␣-latrotoxin binding activ- and their fragments (data not shown). The concentration of ity of detergent-solubilized CIRL (5). p120 solution was applied CIRL fragments in the binding mixtures was typically in the to the bottoms of 96-well plates by drying at ambient temper- range of 20–60 nM, whereas the concentration of labeled ␣-la- ature. After washing and blocking, solutions with various con- 125 trotoxin was 2 nM. centrations of I-␣-latrotoxin were added to the wells. Follow- Because these same soluble CIRL fragments bound quite ing multiple washes, the absorbed labeled toxin was eluted 3594 Interaction of ␣-Latrotoxin with CIRL with SDS-containing buffer and counted for radioactivity. Scat- ment also demonstrates that the low affinity of the recombi- chard plot analysis revealed that the affinity of ␣-latrotoxin nant soluble CIRL fragments was not because of a PCR or binding to p120 was about 25 nM, which was approximately 2 cloning artifact. orders of magnitude lower than the affinity of CIRL binding To test the importance of different regions of p85 in the (data not shown). interaction with ␣-latrotoxin, we generated two mutants of High Affinity Binding of ␣-Latrotoxin to N-terminally and CIRL with deleted portions of the p85 subunit by introducing COOH-terminally Truncated Membrane-associated CIRL— stop codons (Fig. 5). The first mutant pCDR-7TMR did not Low affinity interaction of ␣-latrotoxin with soluble p120 raised contain most of the large COOH-terminal cytoplasmic tail of the possibility that the formation of high affinity complex re- p85. In the second mutant pCDR-1TMR, the N-terminal region quires participation of p85. To test this, we recombined DNA of p85 with only one transmembrane segment was preserved. It fragments encoding soluble deletion mutants with the se- appeared that both deletion mutants expressed very well and quence of p85 so that the resulting constructs encoded N- both showed ␣-latrotoxin binding activity (Fig. 6A). The Scat- terminally truncated forms of CIRL. In the 5Ј-region of chard plot analysis of the one transmembrane mutant demon- pSTR7-6, -7, -8, and -9, a fragment of CIRL cDNA was sub- strated high affinity binding indistinguishable from the wild- cloned, which restored the junction of p120 and p85 identically type receptor (Fig. 6B). These results rule out the involvement to wild-type CIRL. The resulting expression constructs encoded of the extracellular loops and the cytoplasmic tail of p85 in the the following residues of the CIRL protein sequence: pSTR7- stabilization of the ␣-latrotoxin complex with p120. 6M, residues 128–1471; pSTR7-7M, 394–1471; pSTR7-8M, Deletion Mutants of CIRL Are Functional in Coupling CIRL 467–1471; pSTR7-9M, 538–1471. The analysis of the trans- to Exocytosis—We determined whether deletion constructs that fected COS cells demonstrated that all mutants except possessed high affinity ␣-latrotoxin binding (pCDR-1TMR, ␣ pSTR7-9M bound to -latrotoxin specifically (Fig. 4A). This pCDR-7TMR, and pCDR7-8) supported ␣-latrotoxin-stimu- result was in good agreement with the analysis of soluble lated calcium influx when expressed in HEK293 cells (Fig. 7). deletion mutants of CIRL and suggests that lectin-like, olfac- All three proteins increased the uptake of 45Ca2ϩ, although the tomedin-like, and STP (mucin-like) domains in the N-terminal pCDR-1TMR mutant was less effective than pCDR-7TMR, half of p120 are not important for ␣-latrotoxin binding. pCDR7-8, or CIRL itself and had little effect at 50 pM ␣-latro- ␣ We further quantitated the -latrotoxin binding activity of toxin. The data indicate that the COOH-terminal part of CIRL ␣ pSTR7-7M and pSTR7-8M, the two shortest -latrotoxin-bind- is not required in mediating the effects of ␣-latrotoxin on cal- ing mutants, by Scatchard plot analysis. It appeared that both cium permeability and suggest that high affinity ␣-latrotoxin ␣ membrane-bound deletion mutants interacted with -latro- binding is sufficient for calcium influx. toxin with the same high affinity as wild-type CIRL (Fig. 4, B We noted that at higher concentrations of ␣-latrotoxin cells ␣ and C). We can therefore conclude that the primary -latro- transfected with a control plasmid (neo) or nontransfected cells toxin-binding site is located in the COOH-terminal half of (not shown) also exhibited some ␣-latrotoxin-stimulated p120. However, the high affinity interaction requires complex- ϩ 45Ca2 influx. This increase in calcium permeability is proba- ing of p120 with membrane-bound p85. Because the plasmids bly because of the ␣-latrotoxin interaction with an endogenous for membrane-bound mutants were prepared on the basis of HEK293 cell protein, because it is completely blocked by pre- the constructs encoding soluble CIRL fragments, this experi- incubation of the cells with concanavalin A (data not shown). Transient expression of CIRL in chromaffin cells increases their sensitivity to ␣-latrotoxin (9). We asked whether the increase in calcium permeability seen with pCDR-1TMR, pCDR-7TMR, and pCDR7-8 was coupled to the secretory response in chromaffin cells. Because the transfection efficiency of primary cultures is low, the various mutants were co-expressed with human growth hormone, which is stored in secretory granules and serves as a reporter for secretion from the transfected cells. Each of the three constructs increased the FIG.5.Structures of the COOH-terminal deletion mutants of sensitivity of transfected cells to low concentrations of ␣-latro- CIRL. Wild-type CIRL (WT), the mutant with deleted COOH-terminal toxin (2.5 and 10 pM), and the magnitude of the effect was cytoplasmic region up to residue 1149 (7TMR) encoded by plasmid similar to that of wild-type CIRL (Fig. 8A). Catecholamine pCDR-7TMR, and the mutant with only one transmembrane segment (1TMR, residues 1–891 of CIRL) encoded by plasmid pCDR-1TMR are release (a measure of secretion from all the cells, the majority shown schematically. PM, plasma membrane. of which were not successfully transfected) was similar for all

FIG.6. High affinity interaction of ␣-latrotoxin with COOH-terminal deletion mutants of CIRL. A, COS cells transfected with pCDR7 (wild-type CIRL or WT), salmon sperm DNA (SS)asa negative control, pCDR-7TMR, and pCDR-1TMR on day 3 were harvested and incubated with 5 nM 125I-␣-latrotoxin. The total binding of the labeled toxin is shown as black bars. The nonspecific binding measured in the presence of a 20-fold excess of unlabeled toxin is shown as gray bars. Each experiment was done in duplicate and the average is shown (deviation was less than 3% for each measurement shown). B, the ␣-latrotoxin binding affinity of the deletion mutant pCDR-1TMR was compared with the affinity of CIRL by Scatchard plot analysis. Interaction of ␣-Latrotoxin with CIRL 3595 the groups (Fig. 8B). We conclude that the COOH-terminal deletion mutants that were examined in binding experiments, part of CIRL is unnecessary in mediating ␣-latrotoxin-stimu- 45Ca2ϩ influx assays in HEK293 cells and secretion experi- lated secretion from intact chromaffin cells. Interestingly, at ments in chromaffin cells. We found that 1) small segments of higher concentrations of ␣-latrotoxin, some inactivation of the the extracellular and membrane domains of CIRL are required secretory response was seen with CIRL and pCDR7-8 that did for high affinity binding to latrotoxin and for functional effects not occur with the COOH-terminal deletion mutants 1TMR of latrotoxin, and 2) CIRL-coupled G protein signaling is not and 7TMR, raising the possibility that the COOH-terminal critically important for ␣-latrotoxin-stimulated Ca2ϩ influx in cytoplasmic tail of CIRL plays an additional role in modulating HEK293 cells or secretion from chromaffin cells. The evidence secretion. in support of these conclusions is discussed below. High Affinity Binding of Latrotoxin Requires a Short Extra- DISCUSSION cellular Segment and the First Transmembrane Domain— ␣ -Latrotoxin acts extracellularly by binding to endogenous CIRL consists of two noncovalently bound subunits, p120, membrane receptors that belong to the neurexin and CIRL which is extracellular, and p85, which contains seven trans- families. The formation of receptor-toxin complexes is followed membrane segments and the COOH-terminal cytoplasmic do- ␣ by cation influx through -latrotoxin-induced channels and by main. The analysis of a series of soluble recombinant fragments as yet unidentified signaling that eventually results in massive of CIRL by precipitation with ␣-latrotoxin-agarose showed that spontaneous exocytosis. Part of the effects of ␣-latrotoxin can the ␣-latrotoxin-binding site lies within the COOH-terminal be explained by Ca2ϩ entry through the toxin-induced pores. half of p120. However, none of the recombinant soluble frag- However, the toxin is also active when applied in nominally ments of CIRL (including p120 and the nonprocessed full-size Ca2ϩ-free buffers (7, 8, 18) or when cation fluxes are controlled, extracellular region, pCDR7N) that bound effectively to immo- e.g. in permeabilized cells (14, 19). A possible explanation of the bilized ␣-latrotoxin were able to compete with the binding of calcium-independent effect of ␣-latrotoxin is that it activates ␣-latrotoxin to endogenous CIRL in brain membranes. The its membrane receptors, which results in intracellular signal- Scatchard plot analysis of the ␣-latrotoxin binding activity of the ing, through a G protein-linked pathway (19, 20). full-length extracellular region indicated that its affinity was To analyze the interaction of ␣-latrotoxin with CIRL and about 2 orders of magnitude lower than the affinity of CIRL. subsequent functional effects, we have generated a set of CIRL In contrast, two N-terminally truncated fragments of CIRL, which contained p85 in addition to the ␣-latrotoxin binding domain of p120, bound the toxin with the same high affinity as wild-type CIRL. The extracellular loops of p85 are apparently not important for the stabilization of the complex with the toxin because a deletion mutant, which contained only the first transmembrane segment, bound to ␣-latrotoxin with high affinity. Therefore we propose that the interaction of ␣-latrotoxin with CIRL consists of two sequential steps. At first, ␣-latrotoxin binds with medium affinity to the extracellular region of CIRL. Following this binding, the toxin interacts with the first transmembrane segment of CIRL, with the membrane lipids, or with both and penetrates into the lipid bilayer as a result.

؉ This second step increases affinity of the interaction and may FIG.7.Deletion mutants of CIRL support Ca2 entry elicited by ␣-latrotoxin. HEK293 cells were transfected with plasmids for require a longer time, which would explain a known delay in pCDR7 (filled circle), pCDR-7TMR (filled square), pCDR-1TMR (open ␣-latrotoxin effects within a minute after its application. square), pSTR7-8M (triangle), or pCMVneo (open circle) as a control by Only a Single Transmembrane-spanning Domain Is Re- LipofectAMINE. Two days later, the cells were incubated with various 2ϩ ϩ quired to Support Latrotoxin-induced Ca Influx and Secre- concentrations of ␣-latrotoxin (␣-LTX) in PSS containing 2.2 mM Ca2 , ϩ ϩ tion—CIRL deletion mutants that bound ␣-latrotoxin with high 0.5 mM Mg2 , and 3 mCi/ml 45Ca2 . After 5 min, the cells were immediately rinsed 3 times in PSS, and the amount of 45Ca2ϩ in the cells was affinity were also effective in coupling ␣-latrotoxin to calcium determined by liquid scintillation spectrometry. n, 4 wells/group. influx into HEK293 cells and exocytosis in chromaffin cells,

FIG.8.Expression of CIRL deletion mutants increases the sensitivity of intact chromaffin cells to stimulation by ␣-latrotoxin. Chromaffin cells were transfected with plasmids for pCDR7 (filled circle), pCDR-7TMR (filled square), pCDR-1TMR (open square), pSTR7-8M (triangle), or pCMVneo (open circle) as a control by calcium phosphate precipitates. Five days later, cells were incubated with the indicated ϩ ϩ concentrations of ␣-latrotoxin in PSS without Ca2 or Mg2 and with 0.2 mM EGTA. After 4 min, the toxin was removed and the cells were ϩ ϩ incubated for an additional 5 min in PSS containing 2.2 mM Ca2 and 0.5 mM Mg2 . The amounts of human growth hormone (A) and catecholamine (B) released into the medium and the amounts remaining in the cells were determined as described. n, 4 wells/group. 3596 Interaction of ␣-Latrotoxin with CIRL although the 1TMR mutant was noticeably less effective than REFERENCES wild-type CIRL and other mutants in the 45Ca2ϩ uptake assay. 1. Petrenko, A. G., and Krasnoperov, V. G. (1998) in Cellular and Molecular Mechanisms of Toxin Action Secretory Systems (Linial, M., and Grasso, A., Most importantly, a mutant receptor lacking six of seven trans- eds) pp. 185–214, Harwood Academic Publishers, Amsterdam membrane segments was similar to wild-type CIRL in the 2. Petrenko, A. G., Kovalenko, V. A., Shamotienko, O. G., Surkova, I. N., Tarasyuk, T. A., Ushkaryov, Y. A., and Grishin, E. V. (1990) EMBO J. 9, secretion experiments. Because this mutant receptor would be 2023–2027 unable to activate G proteins, these experiments indicate that 3. Petrenko, A. G., Lazaryeva, V. D., Geppert, M., Tarasyuk, T. A., Moomaw, C., Khokhlatchev, A. V., Ushkaryov, Y. A., Slaughter, C., Nasimov, I. V., and the receptor does not mediate the stimulatory effect of ␣-latro- Sudhof, T. C. (1993) J. Biol. Chem. 268, 1860–1867 toxin in intact chromaffin cells by direct coupling to G proteins. 4. Ushkaryov, Y. A., Petrenko, A. G., Geppert, M., and Sudhof, T. C. (1992) Science 257, 50–56 The shape of the dose-effect curves for overexpressed wild- 5. Krasnoperov, V. G., Beavis, R., Chepurny, O. G., Little, A. R., Plotnikov, A. N., type CIRL and its N-terminally truncated mutant was signifi- and Petrenko, A. G. (1996) Biochem. Biophys. Res. Commun. 227, 868–875 6. Davletov, B. A., Shamotienko, O. G., Lelianova, V. G., Grishin, E. V., and cantly different from those of the COOH-terminal mutants Ushkaryov, Y. A. (1996) J. Biol. Chem. 271, 23239–23245 (Fig. 8A). The dose dependence of the ␣-latrotoxin-stimulated 7. Misler, S., and Falke, L. C. (1987) Am. J. Physiol. 253, C469–C476 8. Capogna, M., Gahwiler, B. H., and Thompson, S. M. (1996) J. Neurophysiol. exocytosis in chromaffin cells via either endogenous receptors, 76, 3149–3158 overexpressed CIRL, or its N-terminal mutant was bell-shaped 9. Krasnoperov, V. G., Bittner, M. A., Beavis, R., Kuang, Y., Salnikow, K. V., Chepurny, O. G., Little, A. R., Plotnikov, A. N., Wu, D., Holz, R. W., and (14) (Fig. 8). In contrast, for both COOH-terminally truncated Petrenko, A. G. (1997) Neuron 18, 925–937 mutants, the dose dependence did not show “inactivation” at 10. Lelianova, V. G., Davletov, B. A., Sterling, A., Rahman, M. A., Grishin, E. V., Totty, N. F., and Ushkaryov, Y. A. (1997) J. Biol. Chem. 272, 21504–21508 higher concentrations of ␣-latrotoxin. It is therefore possible 11. Ichtchenko, K., Bittner, M. A., Krasnoperov, V., Little, A. R., Chepurny, O., that ␣-latrotoxin binding to CIRL results in multiple effects, Holz, R. W., and Petrenko, A. G. (1999) J. Biol. Chem., in press 12. Rosenthal, L., and Meldolesi, J. (1989) Pharmacol. Ther. 42, 115–134 one of which is dependent upon the integrity of the COOH 13. Geppert, M., Khvotchev, M., Krasnoperov, V., Goda, Y., Missler, M., Hammer, terminus of the p85 subunit. R. E., Ichtchenko, K., Petrenko, A. G., and Sudhof, T. C. (1998) J. Biol. Chem. 273, 1705–1710 What may be the physiological importance of CIRL? Because 14. Bittner, M. A., Krasnoperov, V. G., Stuenkel, E. L., Petrenko, A. G., and Holz, it serves to target ␣-latrotoxin, a potent stimulator of secretion, R. W. (1998) J. Neurosci. 18, 2914–2922 15. Dohlman, H. G., Thorner, J., Caron, M. G., and Lefkowitz, R. J. (1991) Annu. it is likely that this receptor is positioned appropriately for the Rev. Biochem. 60, 653–688 regulation of exocytosis. CIRL has been shown to co-purify with 16. Nishimori, H., Shiratsuchi, T., Urano, T., Kimura, Y., Kiyono, K., Tatsumi, K., Yoshida, S., Ono, M., Kuwano, M., Nakamura, Y., and Tokino, T. (1997) syntaxin, a t-SNARE (9). In permeabilized chromaffin cells Oncogene 15, 2145–2150 overexpression of CIRL inhibits Ca2ϩ-stimulated secretion 17. Shiratsuchi, T., Nishimori, H., Ichise, H., Nakamura, Y., and Tokino, T. (1997) Cytogenet. Cell Genet. 79, 103–108 (14). Together, these findings suggest that CIRL may serve as 18. Rosenthal, L., Zacchetti, D., Madeddu, L., and Meldolesi, J. (1990) Mol. a physiological regulator of exocytosis. Because of the presence Pharmacol. 38, 917–923 19. Lang, J., Ushkaryov, Y., Grasso, A., and Wollheim, C. B. (1998) EMBO J. 17, of cell adhesion structural modules in its extracellular domain, 648–657 it is possible that direct physical contacts of cells modulate 20. Davletov, B. A., Meunier, F. A., Ashton, A. C., Matsushita, H., Hirst, W. D., Lelianova, V. G., Wilkin, G. P., Dolly, J. O., and Ushkaryov, Y. A. (1998) secretion via CIRL-mediated signaling. EMBO J. 17, 3909–3920