Cleavage of Vasoactive Intestinal Peptide at Multiple Sites by ...

THEJOURNALOF

Vol. 266, No. 24, Issue of August 25, PP.

BIOLOGICAL CHEMISTRY

0 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Cleavage of Vasoactive Intestinal Peptide at Multiple Sites by Autoantibodies* (Received for publication, February 11, 1991)

Sudhir Paul$, Sun Mei, Bina Mody, Steven H. Eklund, Carol M. Beach#, Richard J. Masseyll, and Frederick Hamel From the Departments of Pharmacology, Biochemistry, and InternalMedicine, University of Nebraska Medical Center, Omaha, Nebraska 68198-6260. TIGEN. Inc.. Rockuille. Marvland 20852, and the §University of Kentucky MacromolecularAnalysis Facility, Lexington, Kentucky 40536 I

I

Vasoactive intestinal peptide (VIP) fragments generated by autoantibodiespurified from the blood of two human beings were separated and sequenced. Based on the identity of these fragments, seven peptide bonds cleaved by the antibodies were identified. Six of the seven scissile bonds are clustered in the region of VIP spanning residues 14-22 and were cleaved by antibodies from both human subjects. The seventh scissilebond is located at residues 7-8 and was cleaved by antibodies from one of the subjects. The scissile bonds link amino acid residues with different size, charge, and hydrophobicity. The hydrolytic activity of the antibodies wasselective in that they failed to hydrolyze polypeptides unrelated in sequence to VIP (insulin and atrial natriuretic peptide). These observations demonstrate substrate specific hydrolysis by naturally occurring antibodies and expand the range of peptide bonds hydrolyzed by these antibodies.

Human vasoactive intestinal peptide (VIP)’ is a neuropeptide with a broad profile of biological roles in the central and peripheral nervous systems. VIP is believed to mediate neurotransmission and neuromodulation, relax the smooth muscle of the respiratory, gastrointestinal and reproductive tracts, stimulate exocrine and endocrine secretion (1,2), increase the survival time of neurons (3), and promote the proliferation of untransformed and cancer cells (4,5). Autoantibodies to VIP are found in humans (6-8). We have described previously the catalytic cleavage of a Gln-Met bond in VIP by one such autoantibody (9-12). Antibodies that catalyze peptide bond cleavagecould serve as efficient and specific mediators of immunological defence against microbes andtumorsand, conversely, of the pathophysiology of autoimmune disorders. It is of considerable interest, therefore, to determine the catalytic properties, biological functions, and frequency of * This work was supported in part by Grants 40348 and 44126 and a Research Career Development Award from the National Institutes

of Health (to S. P.) and a contract with IGEN, Inc. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18U.S.C. Section 1734 solelyto indicate this fact. $ T o whom correspondence should be addressed. Tel.: 402-5594044; Fax: 402-559-7495. ’ The abbreviations used are: VIP, vasoactive intestinal peptide; ANP,atrialnatriuretic factor; BSA, bovine serum albumin; Fab, fragment antigen binding; HPLC, high performance liquid chromatography; SDS, sodium dodecyl sulfate; MES, 2-(N-morpho1ino)ethanesulfonic acid; bis-Tris, 2-[bis(2-hydroxyethyl)amino]-2(hydroxymethyl)-propane-1,3-diol.

incidence of naturally occurring peptide bond cleaving antibodies. The susceptibility of a peptide bond to hydrolysis is dependent, in part, on the nature of the amino acids that it links. Asn, and to a lesser degree, Gln, can undergo spontaneous uncatalyzed deamidation, one of the consequences of which may be slow hydrolysis of Asn-X and Gln-X peptide bonds (13). In regard to peptide bond hydrolysis catalyzed by conventional proteases, an important determinantis the recognition of side chains of amino acids at or in the vicinity of the scissile bond (14, 15). VIP, like other polypeptides, is a stable molecule. It is possible, nevertheless, that thesensitivity of the Glr~’‘-Met’~bond to antibodies arises from a destabilizing effect of the amide side chain of Gln. To investigate the potential generality of peptide bond cleaving antibodies, we searched for human VIP autoantibodies capable of cleaving other types of peptide bonds. Here, we report evidence for: (i) VIP-cleaving autoantibodies in two additional humansubjects (designated HS-2 and HS-3),*(ii) the susceptibility of several peptide bonds in VIP to antibody-mediated hydrolysis, and (iii) failure of the antibodies to hydrolyze peptides unrelated to VIP. EXPERIMENTALPROCEDURES

Preparation of Antibodies-Autoantibodies to VIP were identified in eight human subjects by measuring saturable [Tyr’o-’Z51]VIP binding (8).The antibody titers, defined as the saturable [Tyr’o-12SI]VIP binding at a 1:lO dilution of plasma, ranged from 5 to 63%. IgG was purified from plasma by ammonium sulfate precipitation and protein G-Sepharose chromatography (9). Specific antibodies were enriched by affinity chromatography of IgG on immobilized VIP (10). For this purpose, about 90 mg of IgG was shaken with 4.5 ml of VIP-Sepharose for 2 h at 4 “C in 0.05 M Tris-HC1, pH 8, 0.5 M NaCl, the gel poured into a column, washed until the effluent AzW(sensitivity 0.05 absorbance units at full scale) had returned to the base-line value, bound antibody eluted with 0.1 M glycine-HC1, pH 2.7,0.5 M NaCl, and neutralized immediately with 1M Tris-HC1, pH 9. Fab was generated by treatment of IgG with immobilized papain and purified by protein A-agarose chromatography (9) and high performance gel filtration on a Superose-12 column (Pharmacia LKB Biotechnology Inc.) in 0.05 M sodium phosphate, pH 7.0, containing 0.15 M NaCl (flow rate 0.5 ml/min). IgG and Fab concentrations were estimated spectrophotometrically (Azs0)assuming that 0.8 mgIgG/ml corresponds to 1optical density unit or by scanning of silver-stained nonreducing SDS-gels (A562), using authentic IgG as standard. Increasing IgG concentrations (2-20 ng/lane) showed a linear increase in A562 values. Electrophoresis-IgG or Fab treatedwith 2.5% SDS (5 min)in the absence or presence of 20 mM 2-mercaptoethanol were electrophoresed on Phast SDS-polyacrylamide gradient gels (8-25%) (Pharmacia). Silver staining and immunostaining of nitrocellulose blots with rabbit antibodies to human heavy chains (IgG) (1:lOOO) or light chains ( K and X; 1:2, 500) (Axell) were as described (10).

16128

These antisera are available on request.

16129

Peptide Bonds Hydrolyzed by Antibodies Iodinated Peptides-Human VIP-(1-28) (Bachem) was labeled with Iz5I, and [Tyr'0-'251]VIPwas purified by reverse phase HPLC and identified by amino acid sequencing (9). Human ANP (Peninsula) was iodinated and purified as described for VIP. Porcine insulin labeled with lZ5Iat TyrI4of the A-chain was from E. Lilly. The specific activity of the iodinated peptides was 2.2 Cilpmol. Hydrolysis and Binding A~says-[Tyr'~-'~~1]VIP (30-100 pM) was treated with antibody preparations (3 h, 38 "C) in 200 pl of 0.05 M Tris-HC1, 0.1 M glycine, pH 7.7, containing 0.025% Tween 20 and 0.1% BSA. Hydrolysis of the peptide was estimated as theamount of radioactivity that: (i) was soluble in trichloroacetic acid (10%) or (ii) exhibited retention timesdifferent from that of intact [Tyr'0-'251]VIP on a reverse phase HPLC column (C-18) (9). Thelevels of hydrolysis of VIP determined by the two methods were similar at each of three concentrations ofIgG purified from HS-3 ( r = 0.99, slope = 1.16) (Fig. 1).In conformity with previous findings (9), the proportions of radioactivity insoluble in trichloroacetic acid following incubation for increasing lengths of time (0-3 h) in assay diluent were essentially identical (92% at 0 h and 94% a t 3 h), indicating lack of spontaneous hydrolysis of [Tyr'o-'251]VIP.This conclusion was confirmed by reverse phase HPLC analysis on a C-18 column (12), which showed an equivalent recovery of radioactivity in the peak of intact [Tyr'0-'251] VIP at 0 and 3 h of incubation in diluent (94 and 93%, respectively, of the total amounts of radioactivity recovered from the column). There was little orno trichloroacetic acid-soluble radioactivity (5% of total radioactivity were characterized further. To remove low molecular weight inhibitors of the hydrolytic activity that were potentially present in the IgG preparations, thelatter were subjected to two cycles of ultrafiltration on a 10-kDa cutoff membrane (YM-10, Amicon) prior to assay (11). Dialysis of chromatography fractions, performed to :?move inhibitory salts, was against 0.05 M Tris-HCI, 0.1 M glycine, ; 1 R 7.7, 0.025% Tween 20 using a 28-well apparatus (GIBCO-Beh d a Research Laboratories). The relationship between pH (4.5'.'I and hydrolytic activity was examined in two different buffers (38 ?cei acid, 19 mM bis-Tris, 31 mM Tris, 25 mM glycine, 0.025% Tween ZC),O.l% BSA; 39mM MES, 31 mM Tris, 19 mM ethanolamine, 25 mM glycine, 0.025% Tween 20, 0.1% BSA; the pH of the diluent was varied using HCI or NaOH as needed) (16). The VIP hydrolytic '

'2 L

3 I

In"

40

I

RESULTS

/

I

activity of IgG from HS-3 was greatest at pH 6.9-7.7. Bell-shaped curves were observed, with low levels of activity detected at the extreme pH values. The activity curves in the two buffers were virtually superimposable over the entire pH range tested. To determine saturability, the hydrolytic rates were measured at increasing concentrations of unlabeled VIP mixed with a fixed amount of [Tyr'O'251]VIP(30 PM) (9). Data were analyzed by means of the ENZFITTER program (Elsevier). Substrate specificity was examined by incubation of antibodies or diluent with '251-labeledANP, '251-labeled insulin, and control [Tyr'o-'251]VIPunder equivalent experimental conditions, followedby reverse phase HPLC on a C-18 Novapak column (Waters) in the case of ANP and VIP (12) and a C-18 Zorbax column (Dupont) in the case of insulin (17). The insulin reaction mixtures were treated with 0.1 M 2-mercaptoethanol (15 min) prior toHPLCto facilitate release of peptide fragments anchored by disulfide bonds (17). Binding assays and Scatchard analysis were as described (8). Kd was estimated as the slope of the Scatchard plot and the antibody binding capacity (BmaX) as the x intercept. Under the conditions employed in binding assays, hydrolysis of VIP was undetectable, judged by observations that thelevels of trichloroacetic acid-soluble radioactivity after incubation of [Tyr'o-'z51]VIPwith the antibody preparations or assay diluent were comparable (5 and 7% of total trichloroacetic acid-insoluble radioactivity, respectively). Adsorption of Antibodies on Immobilized Anti-IgG-Rabbit antibodies to human IgG ( K and X light chains) (103 mg; Axell)were purified by protein G-Sepharose chromatography (8) and covalently immobilized by covalent coupling to 5 g CNBr-Sepharose (Pharmacia) in a 0.1 M sodium bicarbonate buffer, pH 8.5, containing 0.5 M NaCl for 3.5 h, followed by treatment with 0.2 M glycine to block the remaining NH,-reactive sites on the gel (18).The coupling efficiency was 94%, based on protein assay by the bicinchoninic acid method (Pierce). Control Sepharose preparationsused were CNBr-Sepharose blocked with glycine and CNBr-Sepharose on which nonimmune rabbit IgG (Sigma) was immobilized. Specific HS-2 antibodies enriched by VIP-Sepharose chromatography (96 ng) or total HS-3 IgG purified by protein G-Sepharose chromatography (94 pg) weretreated (18 h, 4 C) with the Sepharose gels at a 50-fold excess of the latter (settled vo1ume:protein mass) in 0.05 M Tris-HC1, 0.1 M glycine, pH 7.7, containing 0.025% Tween 20 with end-to-end shaking. The gels were centrifuged (1000 X g, 5 min), and the supernatants were assayed for [Tyr'o-'251]VIPhydrolytic activity. Identification of VIP Fragments-Unlabeled VIP (50 pg)was treated with IgG or antibodies purified by affinity chromatography as described, except that BSA was absent in the reaction mixture. Control incubations consisted of VIP incubated in assay diluent without antibodies, VIP incubated with nonimmune IgG, and IgG incubated without VIP. The reaction mixtures were extracted on C18 cartridges (Extractclean, Alltech) and analyzed by reverse phase HPLC (Novapak C-18 column) using gradients of acetonitrile in 0.1% trifluoroacetic acid as described (9). Peptidepeaks that were present in the antibody-treated VIP preparations, and were absent in the control reaction mixtures, were sequenced with a pulsed liquid phase sequenator with on-line phenylthiohydantoin-derivative detection (Applied Biosystems, model 477A).

2oL

.-i N 1010

20

30

40

50

I2%-VIP Hydrolyzed (TCA), % FIG. 1. Estimation of [Tyr'o-'Z61]VIPhydrolysis by HS-3 IgG. Plotted are the values of radioactivity that were soluble in trichloroacetic acid (x axis) or eluted at retention times distinctfrom that of intact [Tyr'o-'251]VIP( y axis). Hydrolysis of [Tyr'o-'251]VIP (expressed as percent of total radioactivity; 125,690 cpm and 20,211 cpm, respectively, for analyses performed by HPLC and trichloroacetic acid precipitation) were estimated a t three concentrations of IgG(4.5, 15, and 45 pg/digest). HPLC of [Tyr10-'251]VIPwas as described (9). The retention time of the intact peptide was 20 min. Following incubation with antibody, a broad early eluting peak corresponding to fragments of VIP was observed a t 5-14 min. See "Experimental Procedures" for the trichloroacetic acid precipitation method.

VIP Cleaving Antibodies-Autoantibodies t o VIP are present in some healthyhumans (7) and patients with obstructive airway disorders(8).We have described previously(9-12) the hydrolysis of a Gln-Met bond in VIP by an autoantibody from a human subject (designatedHS-1). In the present study, IgG

samples were purified from eight human subjects positive for VIP-binding antibodies by ammonium sulfate precipitation and chromatography on immobilized protein G, a bacterial IgG binding protein. These IgG preparations (40 pg each) were screened initially for their ability to hydrolyze [Tyr'O'251]VIP, measured as t h e a m o u n t of radioactivity rendered soluble in trichloroaceticacid. The IgG from twosubjects exhibited VIP hydrolytic activity (HS-2, a patient with obstructiveairwaydisease, and HS-3, a healthyindividual). Linearincreasesinhydrolysis of[Tyr'o-'251]VIP were observed with increasing concentrations of IgG (2-100 pg) from HS-2 and HS-3. Analysis of antibody binding as a function of increasing VIP concentration (Fig. 2) suggested the pres-

Bonds Peptide

16130

Hydrolyzed by Antibodies

a, a, i c

\ V

2

m

FIG. 2. Scatchard plots of VIP binding by the IgG from human subject HS-2 (0)and HS-3 (0).Values are means of three replicates each.

A

1

2

3

67 43 30 20 14

ies retained by the column were 1154-fold (HS-2) and 100fold (HS-3) greater than those of corresponding IgG preparations (Fig. 4A).The affinity-purified antibodies also exhibited binding of [Tyr'"-'251]VIP displaceable by excess unlabeled VIP (HS-2, 14,037 cpm/pg protein; HS-3, 3819 cpm/pg protein). Electrophoresis of the affinity-purified antibodies a 61-kDa heavy under reducing conditions revealed two bands: chain band and a 26-kDalight chain band, visualized by silver staining and immunostaining with anti-human heavy and light chain antibodies(Fig. 3B). Treatment of the HS-2 and HS-3 antibody preparations with immobilized antibodies to human L-chains resulted in near-complete depletion of the hydrolytic activity (Fig. 4B), providing further evidence that the hydrolytic agent is anantibody. To prepare Fab fragments, the IgG was cleaved with immobilized papain and chromatographed on a protein A-agarose column. The unbound fraction was fractionated further by high performance gel filtration. The protein peak eluting at 25.5 min from this column was identified as Fab based on its behavior on nonreducing and reducing electrophoresisgels.

B

B -

1

2

I

3

61

L

26

1

FIG. 3. A, nonreducing SDS-gel electrophoresis of silver-stained IgG purifiedfrom HS-3 by proteinG-Sepharosechromatography (lane 2 ) , Fab prepared by papain digestion of HS-3 IgG, protein Aagarose chromatography and gel filtration on a Superose-12 column (lane I ) , and marker proteins (lane 3; mass in kilodaltons is indicated); B, Reducing SDS-gel electrophoresis of HS-3 antibodies enriched by VIP-Sepharose chromatographyof IgG: lane I , stained with silver; lanes 2 and 3, nitrocellulose blots stained with anti-human heavy chain (IgG) antibody and anti-human light chain (K/X) antibody.

2

3

4

FIG.4. A, specific hydrolytic activities of IgG purified by protein G-Sepharosechromatographyand specific antibodies purified by VIP-Sepharose chromatographyfrom HS-2 (columns 3 and 4, respectively) and HS-3 (columns I and 2, respectively); B, reduced VIP hydrolysis by supernatants obtained by incubation of specific HS-2 antibodies (96 ng; prepared by VIP-Sepharose chromatography) or HS-3 IgG (94 pg; prepared by protein G-Sepharose chromatography) with immobilized rabbit anti-human Ig (KIAlight chains) (columns 2 and 4, respectively). Values of VIP hydrolysis measured in control supernatants obtained by incubation of corresponding amounts of HS-2 and HS-3 antibody fractions with immobilized nonimmune rabbit IgG were 15,492 f 636 cpm and 3268 f 503 cpm (columns I and 3, respectively). 0 4

'x A B

I

0

15

+*

* C

ence of a single affinity binding site in IgG from HS-2 ( K d 29.3 nM, BmaX 46 pmol/mg IgG) and two distinct binding sites 7.5 0.025 0 with distinct affinities in IgG from HS-3 (Kd9.3 and 210 nM; x (u BmeX 0.7 and 39 pmol/mg IgG). 0 4 L, -3 I In conformity with previous observation (9, lo), HS-2 and 2 I \ HS-3 IgG prepared by ammonium sulfate precipitation and e o I 0 protein G-Sepharose chromatography were free of detectable I contaminant proteins. Nonreducing SDS-polyacrylamide gel 25 35 E 15 electrophoresis of overloaded IgG revealed a singlesilverRetention time, min stained band of mass 150-160 kDa (Fig. 3A). Reduction with FIG. 5. [Tyr'0-'251]VIP hydrolysis by Fab from HS-3. Fab mercaptoethanol resulted in appearance of two protein bands of mass 61 and 25 kDa, immunoblots of which were stained was prepared as in Fig. 3 and rechromatographed on a gel filtration column (Superose-12; fractionation range: 1-300 kDa). VIP hydrolywith anti-H and anti-L antibodies,respectively (not shown). tic activity ( 0 )was assayed in duplicate50-pl aliquots of the effluent. To enrich VIP-specific hydrolytic antibodies, total IgG frac- -is the profile. A, B, and C are the elution positions of IgG tions were fractionatedon immobilized VIP. The specific (150 kDa), BSA (67 kDa), and soybean trypsin inhibitor (20 kDa) activities (counts/min hydrolysis/pg protein) of the antibod- (Sigma).

; -

'

Hydrolyzed Bonds Peptide

by Antibodies

16131

Under nonreducing conditions, this fraction exhibited a single band with mass 56 kDa (Fig. 3A). Two closely spaced bands with mass 26 and 28 kDa stainable with anti-L and anti-H antibodies, respectively, were observed under reducing conditions (not shown), consistent with formation of L-chains and Fd-chains by reduction of S-S bonds in Fab. Upon rechromatography of the Fab fraction on the gel filtration column, coincident peaks of hydrolytic activity and Azm eluting between BSA (67 kDa) and bovine trypsin inhibitor (20 kDa) were observed (Fig. 5 ) . The specific hydrolytic activities of three fractions with the greatest hydrolytic activity (retention times 24.5,25.5, and 26.5 min) were comparable (454, 538, and 554 cpm/pg protein, respectively). We concluded thattheVIP hydrolytic activity is preserved in the Fab fragments of the antibodies. Scissile Bonds-To locate the scissile bonds, unlabeled VIP was treated with antibody preparations from HS-3 or HS-2 and the resultant VIP fragments separated by reverse phase HPLC andidentified by N-terminal sequencing. The antibody preparations used in these experiments were: (i) the total IgG fraction from HS-3 purified by protein G-Sepharose chromatography and (ii) the VIP-antibody fraction from HS-2 enriched by affinity chromatography of IgG on immobilized VIP. In control incubations,VIP was treated with nonimmune IgG or assay diluent. Treatment of VIP with IgG from HS-3 I I resulted in appearance of several AZl4 absorbing peptides 20 30 1’0 separated by reverse phase HPLC that were absent in the RETENTION TIME, min control HPLC analysis of VIP treated with nonimmune IgG (Fig. 6A). To betterresolve these peptides, the fractions from the initial chromatogram were divided into two pools (retenB tion times 3-12 min and 13-17 min) and rechromatographed using lower acetonitrile concentrations (Fig. 6B). The fractions from the nonimmune IgG control were treated in the same way. Ten peptide fractions absent in the control chromatograms were sequenced. In every case, the peptides identified were fragments of VIP. Some of the HPLC fractions contained two peptide fragments each. Two amino acids were present per sequencing cycle in these cases, which were assigned to individual peptides to fit with the parent sequence of VIP. The scissile bonds were identified based on detection of peptide fragmentsthat: (i)contained an N-terminalresidue ”. 0 1030 20 40 50 other than the N-terminal His of full-length VIP, (ii) possessed a clearly identifiable amino acid sequence five or more residues in length that was identical to internalsubsequences of VIP, and (iii) sequenced at the level of 10 pmol or more. Seven fragments were identified that met the criteria for localization of the scissile bonds (Table I).The peptide bonds 10 on the N-terminal side of these fragments were identified as 9 b the scissile bonds. In the case of the HS-2 affinity-purified antibodies, nine peptide peaks absent in the control reaction .02 mixtures were identified (Fig. 7). Five VIP fragments satisfied the criteria for localization of scissile bonds (Table I). Several fractions in Fig. 6B (fractions 1, 2, 5, 7, 8, and 10) and Fig. 7 0 0 10 20 30 40 (fractions I, 3, and 5-8) contained peptides beginning with the N-terminal pentapeptide sequence of VIP (His-Ser-AspRetention Time, min Ala-Val). With one exception, the C-terminal residues of these FIG. 6. Separation of VIP fragments generated by HS-3 peptides could not be assigned with certainty, due to progresIgC. A , 50 pg of unlabeled VIP was treated with588 pg of IgG purified sively decreasing yields of the phenylthiohydantoin-derivaby protein G-Sepharose chromatography fromHS-3 (lower panel)or tives in successive sequencing cycles. The exception was a nonimmune IgG (upper panel)and the reaction mixture extracted on a C-18cartridgeandseparated by reverse phase HPLC with a peptide present in fraction 3 of Fig. 7, which was identified as His-Ser-Asp-Ala-Val-Phe-Thr-Asp-Asn-Tyr-Thr-Arggradients of solvent B in solvent A (solvent B: 80% acetonitrile in 0.1% trifluoroacetic acid; solvent A: 0.1% trifluoroacetic acid). B, to Leu-Arg (VIP-(l-14)), because it sequenced at >lo0 pmol/ resolve the peptide fragments further, fractions with retention times residue through its C-terminalArg. On the basis of these data,

*

Ill

3-12 min and 13-17 were pooled and rechromatographed (upper and lower panels, respectively) using lower concentrations of acetonitrile for elution (upper panel:15-35% solvent B, 50 min; lower panel: 2040% solvent, 40 rnin). The profiles shown are those observed for the

HS-3 IgG reaction mixture. Peptide fractions numbered 1-10 were absent in control chromatographic analyses of the nonimmune IgG reaction mixture performed under identical conditions (not shown).

16132

Peptide Bonds Hydrolyzed by Antibodies TABLEI Peotide bonds cleaved bv V I P antibodies human subjects HS-2 and HS-3 Fraction numbeP (source of antibody)

3 (HS-3) 3 (HS-3) 4 (HS-3) 6 (HS-3) 9 (HS-3)

9 (HS-3) 10 (HS-3) 2 (HS-2) 2 (HS-2) 3 (HS-2) 3 (HS-2)

D(41), N W ) , Y(17), "(331, R(16), LU4L W6) Y(707), L(911), N(505), S(298),I(279), L(370), N(218) K(360), Y(276), L(399), N(260), S(136), I(134),L(179), N(111) V(92), K(67), K(59), Y(45), L(64), N(48), S(22), I(32), L(33),N(41) H(204). S(306). D(332), A(332), V(403), F(344). T(244). D(297). N(306).

4 (HS-2) 9 (HS-2) Obtained from the HPLC separations shown in Fig. 6B (HS-3) and Fig. 7 (HS-2). *Complete amino acid sequences are shown for the peptides listed. Values in parentheses are amounts (in picomoles) of the amino acids. With @-lactalbuminas standard, the initial yield for sequencing was 53%, and the repetitive yield was >go%. a

vIP(1-28)

n

i

.lo .08

+

.06

d

CQ .04

.02

I

0

20

40

60

80

Retention Time, min FIG. 7. Separation of VIP fragments generated by affinity purified HS-2 antibodies. 50 pg of unlabeled VIP was treated with 88 p g of antibodies purified by affinity chromatography on a VIPSepharose column and the reaction mixture separated by HPLC as in Fig. 7 (15-50% solvent B, 80 min) in 0.1% trifluoroacetic acid. Peptide fractions numbered 1-9 were absent in control incubations, in which the anti-VIP was replaced by assay diluent or nonimmune IgG (not shown).

we concluded that antibodies from HS-2 and HS-3 cleave VIP at six and seven bonds, respectively (Fig. 8). The six bonds cleaved by both antibody preparations are clustered in the segment of VIP spanningresidues 14-22. The seventh scissile bond, located at residues 7-8, was cleaved only by HS-3 antibodies. The amino acid residues at theseven scissile bonds differ in the charge, size, and hydrophobicity. Thus, the P1

site can be occupied by Thr, Arg, Gln, Met, Ala, or Lys and the P1' site byAsp,Lys, Met, Ala,Val andTyr. If the recoveries of VIP fragments during extraction,chromatography and sequencing are assumed to be comparable, cleavage of VIP by HS-2 antibodies was most pronounced at theLysZ1TyrZ2bond (Table I) and by HS-3 antibodies, at the Ala1'Val" bond. Substrate Specificity-To test antibody specificity two peptides unrelated to VIP, lZ5I-1abeledinsulin and lZ51-labeled ANP were treated with IgG from HS-3 (Fig. 9) or HS-2 (not shown) and hydrolysis of the peptides monitored by reverse phase HPLC.The elution profiles of the peptides treated with antibodies and assay diluent were essentially equivalent, suggesting that the antibodies did not hydrolyze insulin and ANP. Under experimental conditions identical to those used for insulin and ANP, the antibodies hydrolyzed 39.0 and 53.7% of [Tyr10-1261]VIP, respectively. Essentially similar results were obtained with HS-2 IgG. Antibody Turnover-The hydrolytic activity of the affinitypurified antibodies from HS-2 was saturable with increasing VIP concentration, and the data could be fitted to the Michaelis-Menten equation (Fig. 10). Half-maximal velocity was observed at 1.85 f 0.33 p M VIP. Based on the observed V,,, value (46.3 +_ 4.1 pmol/3 h) and theamount of antibody used in this experiment (1.1pmol), we estimate an average turnover of 45 molecules of VIP/IgG molecule over the course of the hydrolysis assay (3 h). It is noteworthy that the antibodies we used are polyclonal; it is not possible, therefore, to accurately predict the kinetic properties of individual antibody catalysts that may be present in this preparation.

1

1 1.11

1.1

HS-1 HS-2

t

t t t

t t

HS-3

FIG. 8. Peptide bonds cleaved by VIP antibodies (arrows)from three human subjects (HS-1, HS-2, and HS-3). Data for HS-1 antibodies are from Ref. 9. The scissile bonds were located by identification of fragments generated by HS-2 and HS-3 antibodies by amino acid sequencing.

Peptide Bonds Hydrolyzed by Antibodies

A

16133

B

4 0 ~

p:

40 T

50

0

15

30

45

Retention Time, min

T

60

Retention Time, min

FIG.9. Substrate specificity of HS-3 IgG. Reverse phase HPLC profiles of '"I-ANP (toppanel) and control insulin (top panel) and control [Tyr10-1251]VIP (bottompanel) treated with 100 pg of HS-3 IgG ( A ) and 1251-labeled [Tyr'o-1251]VIP (bottompanel) treated with 37.5 pg of HS-3 IgG. In each case, radioactivity in fractions corresponding to 1 min each was determined. Solid lines indicate reaction mixtures treated with IgG, and broken lines indicate control reaction mixtures treated with assay diluent. '251-labeledinsulin was chromatographed as in Ref. 17. labeled ANP reaction mixtures were separated with a gradient of 10-50% solvent B in solvent A (50 min) and [Tyr'0-'261]VIPreaction mixtures with gradients of 10-50% (60 min) ( A ) or 30-60% (45 min) ( B ) solvent B in solvent A (see Fig. 6 for solvent composition). The retention times of intact 1251-labeledANP and 1251-labeled insulin (A-chain) were 39.3 and 56.0 min, respectively, and those of intact [Tyr10-'251]VIP were 53.4 min ( A ) or 20.8 min ( B ) . (9) andin the present study, unlabeled VIP is also cleaved by the antibodies. It is highly unlikely, therefore, that cleavage of [Tyr'0-'251]VIP by antibodies is due to a radiolytic event, as suggested recently (19). Moreover, breakdown of[Tyr'O'251]VIP in the absence of antibody was undetectable, suggesting that radiolysis is not a significant factor. The segment of VIP spanning residues 14-22 is most sensitive to antibody-mediated hydrolysis, and six of the seven scissile bonds are located in this region. The seventh scissile bond links residues 7 and 8. Uncatalyzed hydrolysis of peptide bonds usually requires exposure to harsh conditions like low O Y 0 1 2 3 pH and elevated temperatures. Asn-X bonds can undergo slow spontaneous cleavage at physiological pH and temperaVIP [PMI FIG.10. Saturable hydrolysis of VIP by HS-2 antibodies ture via intramolecular nucleophilic attack on the peptide (168 ng; purified by VIP-Sepharose chromatography). The bond carbonyl by the amide side chain; tl/* values for this data are fitted to the Michaelis-Menten equation. Values are means deamidative cleavage process are on the order of months (13). of triplicate determinations. Theoretically, Gln-X bonds could undergo analogous deamidative cleavage. The Glnl6-Met'' bond belongs to thisclass of DISCUSSION peptide bonds but the other six bonds cleaved by antibodies We have described previously an autoantibody from a hu- do not. In viewof the diversity of residues at the scissile man subject that catalyzed the hydrolysis of VIP at theGln16- bonds, the potential destabilization of peptide bonds by the Met17bond (9,12). Thepresent study shows cleavage of VIP side chains of certain amino acids is not sufficient to explain at several peptide bonds by VIP-cleaving antibodies present antibody-mediated cleavage of VIP. Certain acyl transfer reactions that are energetically less demanding than peptide in two additionalhuman subjects. The evidence that the hydrolytic activities isolated from these subjects are antibody- bond hydrolysis are shown to be catalyzed by anti-hapten mediated includes: overloaded hydrolytic IgG preparations antibodies produced by immunization against transition state were free of detectable contaminants visualized by SDS- analogs (20, 21). Presumably, tight binding of antibodies to facilitates thesereactions. The free energy electrophoresis under nonreducing and reducing conditions; the transition state highly purified Fab fragments of the antibodies retained the of antigen binding by antibodies can be as large as 12-18 hydrolytic activity; adsorption on immobilized antibody to kcal/mol (corresponding to KO 109-10'3 M-'). Since peptide human light chains resulted in virtually complete loss of the bonds are generally stable, the antibody-antigen binding enhydrolytic activity; andaffinity chromatography on VIP- ergy may not be sufficient, in itself, to effect the hydrolysis Sepharose resulted in increased specific activity of the anti- of VIP. This conclusion is supported by observations that IgG body preparations. We used [Tyr'0-'251]VIPto determine some preparations from six human subjects exhibited binding of of the properties of these antibodies. As described previously VIP but failed to hydrolyze the peptide. Although the mech-

16134

Peptide Bonds Hydrolyzedby Antibodies

anism of the hydrolysis of VIP remains to be defined, the antibodies probably act like conventional proteases, in that chemically reactive amino acids in the catalytic site of the antibody (e.g. His, Ser, Asp) may participate directly in peptide bond hydrolysis. Since the scissile bonds in VIP link amino acids with distinct size, charge, and hydrophobicity, the hydrolytic repertoire of the antibodies is considerable. The extent to which substrate structural requirements contribute in the observed cleavage patterns remain to be determined. For instance, the cleavage reactions may occur consecutively if fragments produced by the initial cleavage of full-length VIP serve as substrates for the antibodies. Alternatively, the fragments may be unrecognizable by antibodies, in which case, cleavage at different peptide bonds must take place simultaneously, using full-length VIP as substrate. Insulin and ANP, peptides unrelated in sequence to VIP, were not cleaved bythe antibodies. Moreover, the VIPhydrolytic activity of the antibodies was observed in the presence of a 300-fold excess (by weight) of BSA. These considerations suggest that VIP hydrolysis by the antibodies is sequencespecific. The antibodies we have examined are polyclonal and the number of component hydrolytic antibody species is not known. In a related study (26), light chains purified from HS3 antibodies were observed to hydrolyze VIP. Subunit rearrangement via disulfide bond exchange reactions could generate antibody oligomers with altered hydrolytic specificity. It is not possible to predict at this time, therefore, the relationship between epitopes recognized by the antibodies and the peptide bonds that are cleaved. For example, cleavage at different peptide bonds could be due to different antibodies, each with a unique bond specificity. The second possibility is that a single antibody may cleaveat different types of peptide bonds, if conformational flexibility in the antigen or in the antibody active site permits productive hydrolytic contact at more than one type of peptide bond. Antibody active site residues contact large epitopes on antigens composed of as many as 15-22 amino acid residues (22-24). We have shown that residues 22-28 of VIP contribute in recognition by a catalytic antibody that cleaves VIP between residues 16 and 17 (Gln16-Met17) (12).The specificity of a hydrolytic antibody for a particular polypeptide could arise, therefore, from recognition of a large peptide epitope that includes amino acids distant from the bonds hydrolyzed. An antibody that is polyspecific or nonspecific with the respect to the type of peptide bonds it can hydrolyze may fail to hydrolyze potential polypeptide substrates that are incapable of sufficiently strong binding. The binding step alone may be sufficient, therefore, to confer to a hydrolytic antibody the ability to discriminate between different polypeptide substrates. VIP is a neuropeptide with a broad range of biological activities. Fragments of VIP generally display little or no biological activity compared with theintact peptide (25). Cleavage of VIP by antibodies at the sites identified in the present study is likely to result, therefore, in fragments with greatly reduced activity. The antibodies are specific for VIP,

the hydrolytic activity of these antibodies is not limited to a particular type of bond, and the antibodies have now been identified in three different human beings. These properties are consistentwith a biological role for catalytic autoantibodies to VIP as effector agents of the immunological system. Although the VIPautoantibodies arethe only known peptide bond cleaving antibodies at this time, the formation of such antibodies could extend to other self- and nonself-protein antigens. Acknowledgment-Technical assistance from D. Volle is acknowledged. REFERENCES 1. Said, S. I. (ed) (1982) Vasoactive Intestinal Peptide, Raven Press, New York 2. Said, S. I. (1984) Peptides 5, 144-150 3. Brenneman, D. E., Nicol, T., Warren, D. & Bowers, L. M. (1990) J.Neurosci. Res. 2 5 , 386-394 4. Hegerstrand, A., Jonzon, G., Daalsgaard, C.-J, & Nilsson, J. (1989) Proc. Natl. Acad. Sci. U. S. A. 86,5993-5996 5. Prysor-Jones, R.A., Silverlight, J. J. & Jenkins, J. S. (1989) J. Endocrinol. 1 2 0 , 171-177 6. Bloom, S. R., Barnes, A. J., Adrian, T. E. & Polak, J. M. (1979) Lancet i:14, 14-15 7. Paul, S. & Said, S. I. (1988) Life Sci. 4 3 , 1079-1084 8. Paul, S., Said, S. I., Thompson, A., Volle, D. J., Agrawal, D. K., Foda, H. & De la Rocha, S. (1989) J. Neuroimmunol. 23,133142 9. Paul, S., Volle, D. J., Beach, C. M., Johnson, D. R., Powell, M. J. & Massey, R. J. (1989) Science 244, 1158-1162 10. Paul, S., Volle, D. J., & Mei, S. (1990) J. Zmmunol. 145, 11961199 11. Paul, S. (1989) Cold Spring Harbor Symp. Quant. Biol. 5 4 , 283286 12. Paul, S., Volle, D. J., Powell, M. J. & Massey, R. J. (1990) J.Biol. Chem. 265, 11910-11913 13. Geiger, T. & Clarke, S. (1987) J. Bwl. Chem. 262, 785-794 14. Fink, A. L. (1987) in Enzyme Mechanisms (Page, M. I., and Williams, A., eds) pp. 159-177, Royal Society of Chemistry, London 15. Barrett, A. J. (1986) in Proteinase Inhibitors (Barrett, A. J., and Salvesen, G., eds pp. 3-22, Elsevier Science Publishers B. V., Amsterdam 16. Ellis, K. J . & Morrison, J. F. (1982) Methods Enzymol. 87,405 17. Hamel, F. G., Peavy, D. E., Ryan, M. P. & Duckworth, W. C. (1986) Endocrinology 118,328-333 18. Pharmacia (1986) Affinity Chromatography:Principles and Methods pp. 12-18, Pharmacia, Uppsala, Sweden 19. Shokat, K.M. & Schultz, P. G. (1990) Annu. Reu. Zmmunol. 8, 335-363 20. Tramontano A., Janda K. D. & Lerner R. A (1986) Science 2 3 4 , 1566-1573 21. Blackburn, G. M., Kang, A. S., Kingsbury, G..A. & Burton, D. R. (1989) Biochem. J. 2 6 2 , 381-390 22. Amit, A.G., Mariuzza, R. A., Phillips, S. E. V. & Poljak, R. J. (1985) Nature 313, 156-158 23. Colman, P. M., Laver, W.G., Varghese, J. N., Baker, A. T., Tulloch, P. A., Air, G. M.& Webster, R. G. (1987) Nature 3 2 6 , 358-363 24. Davies, D. R., Padlan, E. A. & Sheriff, S. (1990) Annu. Reu. Biochem. 59,439-473 25. Rosselin, G. I. (1986) Peptides 7,(Suppl. 1) 89-100 26. Mei, S., Mody, B., Eklund, S. H., and Paul, S. (1991) J. Biol. Chem. 2 6 6 , in press