The electrophoretically 'slow' and 'fast' forms of the alpha 2-macroglobulin molecule. - PDF Download Free (2023)

Biochem. J. (1979) 181, 401-418 Printed in Great Britain

401

The Electrophoretically 'Slow' and 'Fast' Forms of the ao2-Macroglobulin Molecule By Alan J. BARRETT, Molly A. BROWN and Christine A. SAYERS Molecular Pathology Department, Strangeways Research Laboratory, Worts' Causeway, Cambridge CB1 4RN, U.K. (Received 21 December 1978)

a2-Macroglobulin (a2M) was isolated from human plasma by a four-step procedure: poly(ethylene glycol) fractionation, gel chromatography, euglobulin precipitation and immunoadsorption. No contaminants were detected in the final preparations by electrophoresis or immunoprecipitation. The protein ran as a single slow band in gel electrophoresis, and was designated 'S-a2M'. S-a2M bound about 2mol of trypsin/mol. Treatment of S-a2M with a proteinase or ammonium salts produced a form of the molecule more mobile in electrophoresis, and lacking proteinase-binding activity (F-a2M). The electrophoretic mobility of the F-x2M resulting from reaction with NH4' salts was identical with that of proteinase complexes. We attribute the change in electrophoretic mobility of the a2M to a conformational change, but there was no evidence of a change in pI or Stokes radius. Electrophoresis of S-a2M in the presence of sodium dodecyl sulphate gave results consistent with the view that the a2M molecule is a tetramer of identical subunits, assembled as a non-covalent pair of disulphide-linked dimers. Some of the subunits seemed to be 'nicked' into two-thirds-length and one-third-length chains, however. This was not apparent with F-a2M produced by ammonium salts. F-a2M produced by trypsin showed two new bands attributable to cleavage of the subunit polypeptide chain near the middle. Immunoassays of F-a2M gave 'rockets' 12-29 % lower than those with S-a2M. The nature of the interactions between subunits in S-a2M and F-a2M was investigated by treating each form with glutaraldehyde before electrophoresis in the presence of sodium dodecyl sulphate. A much greater degree of cross-linking was observed with the F-a2M, indicating that the subunits interact most closely in this form of the molecule. Exposure of S-a2M to 3M-urea or pH3 resulted in dissociation to the disulphide-bonded half-molecules; these did not show the proteinase-binding activity characteristic of the intact a2M. F-a2M was less easily dissociated than was S-a2M. S-a2M was readily dissociated to the quarter-subunits by mild reduction, with the formation of 3-4 new thiol groups per subunit. Intact reactive a2M could then be regenerated in high yield by reoxidation of the subunits. F-a2M formed by reaction with a proteinase or ammonium salts was not dissociated under the same conditions, although the interchain disulphide bonds were reduced. If the thiol groups of the quarter-subunits of S-a2M were blocked by carboxymethylation, oxidative reassociation did not occur. Nevertheless treatment of these subunits with methylammonium salts or a proteinase caused the reassembly of half-molecules and intact (F-) tetramers. It is emphasized that F-a2M does not have the properties of a denatured form of the protein. The suggestion is made that one attractive interpretation of the difference in properties between the S- and F-forms is that the S-form is a planar tetramer, whereas the F-form is pseudotetrahedral. The results are considered in relation to the 'trap' hypothesis for the reaction of a2M with proteinases. a2-Macroglobulin (ac2M) is a plasma glycoprotein of about 725000 mol.wt. (for reviews see Laurell & Jeppsson, 1975; Harpel, 1976; Starkey & Barrett, Abbreviations used: a2M, a2-macroglobulin; F-a2M, the electrophoretically fast form of a2M; S-a2M, the electrophoretically slow form of a2M; IgA, immunoglobulin A; IgM, immunoglobulin M; SDS, sodium dodecyl sulphate.

Vol. 181

1977). a2M has the unique property of binding and inhibiting the great majority of endopeptidases, regardless of their specificity or catalytic mechanism, and there is good reason to think that the protein is important in the physiological control of extracellular proteolytic activity. In pointing out the broad reactivity of a2M with proteinases, Barrett & Starkey (1973) put forward the 'trap' hypothesis for

402

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

the reaction, It was argued that most proteinases have the capacity to attack a particularly vulnerable region in the macroglobulin molecule, and that this triggers a rapid conformational change in the macroglobulin molecule that results in the trapping of the enzyme molecule within it. The result is a characteristic steric inhibition of the reactivity of the bound proteinase with large substrate or inhibitor molecules, and also with antibodies (Geokas et al., 1977). Subsequently two new pieces of evidence were obtained in support of the 'trap' hypothesis. It was shown that reaction of a2M with proteinases does indeed involve proteolytic cleavage (Harpel, 1973), and that this occurs at essentially the same point in the molecule despite widely differing proteinase specificities (Barrett et al., 1974). It was also demonstrated that the electron-microscopic appearance of a2M is consistent with a conformational change being associated with the reaction with proteinases (Barrett et al., 1974). Research on the interaction of a2M with proteinases has received a stimulus from the suggestion that a2M of people homozygous for the hereditary disease cystic fibrosis reacts abnormally (Shapira et al., 1977; Wilson & Fudenberg, 1976). Clearly it is of interest to understand in detail the structure of the a2M molecule, and to know how it is changed in course of reaction with a proteinase. The a2M molecule is a tetramer of identical subunits linked in pairs by disulphide bonds, and then assembled non-covalently (Hall & Roberts, 1978) (Fig. 1). Here we report that the native tetramers can exist in two forms that differ in electrophoretic mobility ('slow' or S-, and 'fast' or F-) as well as other properties and that the apparently irreversible transition of the S-form into the F-form is associated with the binding of a proteinase. The results are considered in relation to the structure of a2M and its mechanism of action. Materials and Methods

Araterials Sources of chemicals and other materials were as follows: Brij 35, iodoacetic acid (special purity), methylammonium chloride, ethylammonium chloride, 1,2-diaminoethane, ethanolamine and poly(ethylene glycol) 6000, from BDH Chemicals, Poole, Dorset BH12 4NN, U.K.; 1,3-diaminopropane from Aldrich Chemical Co., Gillingham, Dorset SP8 4JL, U.K.; albumin (bovine serum, crystallized), carbonic anhydrase (bovine erythrocyte), catalase (bovine liver, 2 x crystallized), a-chymotrypsin (bovine, 3 x crystallized), cytochrome c (horse heart, type Ila), ferritin (horse spleen, type I), papain (2 x crystallized), phenyl-

Fig. 1. Diagrammatic representation of the a2M molecule The identical quarter-subunits of the a2M molecule are linked in pairs by disulphide bonds, and the resultant half-molecules are associated non-covalently (at the sites marked 'r'). The shapes indicated for the subunits and the whole molecule are not intended to resemble their actual shapes, which are far more complex.

methanesulphonyl fluoride, phosphorylase a (rabbit muscle, 2 x crystallized), thermolysin, transferrin (human), neuraminidase (Clostridium perfringens, type V), trypsin (bovine, 2 x crystallized), trypsin inhibitor (soya-bean, type I-s), thyroglobulin (pig, type II), from Sigma (London) Chemical Co., Poole, Dorset BH17 17NH, U.K. Trypsin preparations were standardized by active-site titration (Chase & Shaw, 1967), but weights of trypsin given in the text refer to the commercial solid (52 % active enzyme). Ultrogel AcA 22 and AcA 34, Ampholines and the preparative isoelectric-focusing apparatus were from LKB Instruments, South Croydon, Surrey CR2 8YD, U.K., and benzoyl-DL-arginine 4-nitroanilide hydrochloride and benzyloxycarbonyl-L-arginine 2-(4-methoxy)naphthylamide hydrochloride from Bachem Feinchemikalein, Liestal CH-4410, Switzerland. Myosin from rabbit skeletal muscle was kindly given by Dr. A. Weeds, M.R.C. Laboratory of Molecular Biology, Hills Road, Cambridge, U.K. Immune globulin preparations (rabbit) against human immunoglobulins, IgA, IgM, fibrinogen, plasminogen and haptoglobin were from Mercia Brocades, Weybridge, Surrey KT14 6RA, U.K. Horse anti-(human serum) antiserum was from Wellcome Reagents, Beckenham, Kent BR3 3BS, U.K. Antiserum (rabbit) against human a2-pregnoglobulin (pregnancy-associated a-globulin) was a 1979

a2-MACROGLOBULIN FORMS gift from Dr. A. R. Bradwell, Immuno Diagnostic Research Laboratory, Department of Experimental Pathology, Birmingham B15 2TJ, U.K. Rabbit anti-(human a2-macroglobulin) sera were a gift from Dr. N. Heimburger, Behringwerke A.G., Marburg, Germany. Standard human plasma and serum were from Behring Diagnostics, Hoechst Pharmaceuticals, Hounslow, Middx. TW4 6JH, U.K.

Electrophoresis The systems of gel electrophoresis are distinguished as 'rate' electrophoresis of native proteins or SDS-protein complexes and as 'pore-limit' electrophoresis of native proteins run to pseudoequilibrium in gradients of increasing polyacrylamide concentration. Rate electrophoresis of native proteins. The buffer system was the discontinuous 2-amino-2-methylpropane-1-diol (Ammediol)/glycine/chloride system of Wyckoff et al. (1977), in which resolution occurs at pH 9.58. Polyacrylamide slab gels (5% or 7% total acrylamide concentration, 2.6 % of this as methylenebisacrylamide) were prepared as follows. Pairs of glass plates (81 mm square, 1.2mm thick) were spaced with plastic strips (4mm wide, 2.8mm thick) and taped together at two opposite edges. Sets of ten such cassettes were placed with the taped sides vertical in a Perspex [poly(methyl methacrylate)] box. The box was filled to 1 cm depth with aq. 25% (v/v) ethanol. The polyacrylamide monomer solution was then pumped into the bottom of the box, so that it gradually rose to fill the cassettes, overlayered with the aqueous ethanol. When the gel had polymerized, the slabs were removed from the box, sealed in polythene bags and stored at 40C until required. For use, each slab was provided with an upper gel (3.2 % total acrylamides, 6.25 % as methylenebisacrylamide) in which six sample slots were formed by use of an acrylic comb (Pharmacia, code no. 91 482). Samples containing about 1 mg of protein/ml were mixed with an equal volume of double-strength upper-gel buffer containing 40% (v/v) glycerol and 0.01 % Bromophenol Blue. A 20,u1 portion was applied to each slot. Electrophoresis was in a Pharmacia GE-4 tank or a similar apparatus constructed in the laboratory. Samples were run into the gel at a current of 2OmA (-100V), before the current was increased to 4OmA (-200V) for the remainder of the run. The run was stopped when the Bromophenol Blue marker approached the lower edge of the gel (after about 60min). Gels were stained with 0.1 % Coomassie Brilliant Blue G in methanol/acetic acid/water (5 :1 :4, by vol.) during 1 h at 55°C, before destaining in methanol/ Vol. 181

403 formic acid/water (30:1:69, by vol.) and eventual storage in I % (v/v) formic acid. Rate electrophoresis of SDS-protein complexes. The buffer system was as described for native proteins, except that the upper reservoir buffer contained 0.1 % SDS, and the gels were of 7 % total acrylamide concentration. Samples known to contain proteinases, e.g. trypsin-ox2M complex, were treated to inactivate the proteinase (with 1 mM-phenylmethanesulphonyl fluoride, in this instance) before exposure to SDS. Samples (501 containing about 50,ug of protein) to be run reduced were mixed with an equal volume of sample buffer containing doublestrength upper-gel buffer with 0.01 % Bromophenol Blue, 40% (v/v) glycerol, 2% (w/v) SDS and I % (v/v) 2-mercaptoethanol, and immediately heated at 100°C for 5min. Alternatively, some were incubated at 37°C for 45min, as indicated in the text. Portions (20,ul) were placed in the sample slots of the upper gel. It was found that boiling in SDS at pH8.37 (the pH of the upper-gel buffer) itself caused some cleavage of disulphide bonds. Samples to be run with disulphide bonds intact were therefore boiled with SDS (2 %, w/v) and 2 mM-sodium iodoacetate at pH6-7 for 5 min before the upper-gel buffer, glycerol and Bromophenol Blue were introduced. Electrophoresis, staining and storage of the gels were as described above for the non-SDS-containing gels. The standard proteins used for molecularweight calibration of gels, and the molecular weights assumed for them, were as follows: rabbit muscle myosin heavy chain (mol.wt. 200000), Escherichia coli f,-galactosidase (mol.wt. 130000), rabbit muscle phosphorylase a (mol.wt. 95000), human transferrin (mol.wt. 78000), bovine plasma albumin (mol.wt. 68000), sheep immunoglobulin G heavy chain (mol.wt. 50000) and bovine erythrocyte carbonic anhydrase (mol.wt. 29000). Pore-limit electrophoresis of native proteins. Slab gels containing concave gradients of polyacrylamide concentration (4-26 Y.) for pore-limit electrophoresis (Margolis, 1973) were prepared as described above, except that the acrylamide monomer gradient was produced by two Pharmacia P-3 pumps, used as described by Lakshman & Lieberman (1954); the two pumps were run at 162 and 180 ml/h respectively. The buffer used in gel, reservoirs and samples was 50mM-Tris/7mM-disodium EDTA/2 mM-boric acid, pH9.1 (Manwell, 1977). Gels were stored as described above, and when required were fitted with polypropylene sample applicators (Pharmacia, code no. 91 093). Samples (50,ul, containing 50g of protein) were mixed with an equal volume of double-strength buffer containing 0.01 % Bromophenol Blue and 40 % (v/v) glycerol, and 10#1 portions were applied to the slots in the applicator. Electrophoresis was at 125 V,

404

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

in the Pharmacia GE-4 apparatus with buffer circulation and water-cooling, overnight. Staining and storage of gels were as described above. Proteins used in an attempt to calibrate the porelimit gels with regard to molecular weight were bovine thyroglobulin (mol.wt. 670000), horse spleen ferritin (mol.wt. 440000) and bovine liver catalase (mol.wt. 230000). For some experiments gels were formed with a formic acid (50mM)/Tris (25 mM) buffer, pH 3.1, by use of an H202/thiourea catalyst system (Hurley, 1977), and samples were run in this buffer with Methyl Green as tracking dye. Densitometry of electrophoresis gels. Coomassie Blue-stained gels were scanned at 592 nm by use of the model 2520 accessory of a Gilford Instruments model 250 spectrophotometer, the absorbance being recorded and integrated with a Smiths Servoscribe RE 542.20 recorder. Preparative isoelectric focusing The LKB Instruments type 8101 column was used as described by the manufacturers, with a 9 :1 (v/v) mixture of Ampholines pH4-6 and pH 3-10.

Radiolabelling ofproteinases a-Chymotrypsin was labelled by treatment with [14C]acetic anhydride, in aqueous solution, essentially as described by Barrett & Heath (1977) for haemoglobin, but without the precipitation step. Trypsin was labelled by the method of Bolton & Hunter (1973), N-hydroxy-3-(p-hydroxyphenyl)propionylsuccinimide (Pierce and Warriner, Chester CHI 4EF, U.K.) being treated with [1251]NaI in the presence of chloramine-T, and allowed to react with the protein. Protein determination

Protein concentrations were determined essentially as described by Lowry et al. (1951), with bovine serum albumin as standard. The concentration of solutions of highly purified a2M was determined by spectrophotometry, on the basis that A 8 1% = 9.1 (Dunn & Spiro, 1967). Immunochemical methods All immunoplates were stained with Coomassie Brilliant Blue G (Barrett, 1974), and some also with Sudan Black for detection of fl-lipoprotein. Double immunodiffusion and immunoelectrophoresis in agarose gels were performed as described by Barrett (1974). 'Rocket' immunoassay (Laurell, 1972) was used to determine the antigenic activity of a2M; the gel contained 0.4% (v/v) rabbit anti-(human a2M)

serum, and standard human serum was used for standardization. Quantitative radial immunodiffusion (Mancini et al., 1965) was used for determination of immunoglobulins, haptoglobin and fibrinogen. Plates contained 0.2 % (v/v) of rabbit immune globulin preparations, and were standardized with standard serum or plasma. Immunoadsorbents were prepared by coupling the rabbit immunoglobulins against human immunoglobulins, immunoglobulin A, a2-pregnoglobulin, haptoglobin and fibrinogen to Sepharose 4B that had been activated with CNBr essentially as described by Cuatrecasas & Anfinsen (1971). Each gram of damp activated gel was allowed to react overnight at 20°C with 7.5-50,ul of the commercial immune globulin solutions in 50mM-sodium phosphate buffer, pH6.5. The adsorbents were then treated with 1 % (w/v) glycine in the same buffer for 1 h, and washed with 100mM-sodium formate buffer, pH3.0, followed by 100mM-sodium phosphate, pH 7.0, containing 0.1 % NaN3. Assay ofproteinase-inhibiting activity of a2-M Activity was determined as the inhibition of proteolytic activity of trypsin against a particulate substrate, hide powder azure (Calbiochem, London W1H, U.K.). In each of the duplicate assay tubes, a 0.1ml sample containing 10-50pg of a2M/ml was incubated for 10min at 37°C with 0.1 ml of trypsin (2.0jg/ml in 0.1OM-Tris/HCl buffer, pH8.1, containing 20mMCaCI2 and 0.1 % Brij 35) in a total volume of 0.5 ml made up with the same buffer/detergent solution. Then 0.5 ml of water was added to the tube, followed by 0.8 ml of substrate suspension, i.e. 12.5 mg of hide powder azure/ml in 0.6 M-sucrose/0. 1 % Brij 35/0.03 % toluene. The tube was incubated in a roller rack (2 rev./min) at 37°C for 20 min, and 1 ml of water was added. After centrifugation (1200g for 5min), the A595 of the supernatant was determined. A standard curve of 0-0.2,ug of trypsin/tube was run in parallel, and results were calculated as the amount of trypsin inhibited per unit weight of a2M, the trypsin itself being standardized by active-site titration (Chase & Shaw, 1967). A linear relationship between amount of a2M and degree of inhibition was obtained in the range 0-80% inhibition, and this relationship was used to calculate the inhibitory capacity of the a2M. Inhibitory activity against thermolysin was measured in the same way, except that the buffer was 0.4M-Tris/HCI, pH7.5, containing 40mM-CaC12.

Preparation of a2-M in the S-form a2M was purified by precipitation with poly(ethylene glycol), gel chromatography, euglobulin preci1979

2-MACROGLOBULIN FORMS pitation and immunoadsorption. In detail, the procedure was as follows. Fresh human blood unsuitable for transfusion was provided by a local blood-transfusion service. In each bag 425 ml of blood had been mixed with 75 ml of ACD anticoagulant solution containing 0.60g of citric acid monohydrate, 1.65g of trisodium citrate dihydrate and 1.68g of anhydrous glucose. The blood had been stored at 4°C for 2-5 days after collection. Plasma of haptoglobin type 1-1 was used, having been identified by rate electrophoresis after saturation with haemoglobin (Halliday, 1976). The plasma was separated by centrifugation (100OOg for 15min) in plastic bottles, and stored at -20°C until required. Plasma (250ml) was mixed with 0.28vol. of aq. 25% (w/v) poly(ethylene glycol) 6000, pH6.5, and left at least 30min at 20°C. The precipitate was removed by centrifugation and discarded, and a further 0.72vol. (on the basis of the original volume) of the 25 % (w/v) poly(ethylene glycol) added to the supernatant. After 30min the -precipitate [formed at 5.5-12.5 % (w/v) poly(ethylene glycol) (Giroux, 1975)] was centrifuged as before, dissolved in 0.1 M-sodium citrate buffer, pH 6.5, to an A'c' of about 50, and run on a bed (5cm x 80cm, 1570cm3) of a 2: 1 (v/v) mixture of Ultrogel AcA 34 and AcA 22, in the same buffer, at a flow rate of 20ml/h. A peak of a2M was detected (by gel electrophoresis or immunodiffusion) in fractions from the gel chromatography column early in the elution pattern, but not at the excluded volume. Fractions showing little contamination in polyacrylamide-gel electrophoresis were combined, concentrated by ultrafiltration over a Sartorius type 12133 membrane under pressure of N2 (in a C-50 ultrafiltration cell; Chemlab Instruments Ltd., Hornchurch, Essex, U.K.) to about lOmg/ml and dialysed against 1 % (w/v) glycine (two l00vol. portions) during 24-48 h. A precipitate (largely composed of immunoglobulins) was removed by centrifugation (26000g for 15min) and discarded. The final step of purification was immunoadsorption. The partially purified a2M (about 100mg in lOml of 1% glycine) was passed through a bed (20mm x 24cm) containing a mixture of adsorbents for total immunoglobulins (30ml), IgA (20ml), a2-pregnoglobulin (5ml), haptoglobin (20ml) and fibrinogen (20ml). The unadsorbed protein was washed through with 100mM-sodium phosphate buffer, pH 7.0, and reconcentrated to A280 about 10 by ultrafiltration. The immunoadsorbent column was regenerated for re-use by washing with 0.1 Msodium formate buffer, pH 3.0, containing 0.5 MNaCl. The immunoadsorption process was repeated as necessary until no immunoprecipitation reaction could be obtained for any of the original contaminants. Sometimes small (20ml) columns of adsorVol. 181

405 bents for individual contaminants, IgA or a2-pregnoglobulin, were also used. Finally, the solution of a2M was twice dialysed for 24h against l00vol. of 20mM-sodium citrate buffer, pH6.5, and stored at 4°C, sometimes after freeze-drying. Results and Discussion We have found that native a2M can exist in two forms readily distinguishable by mobility in gel electrophoresis and many other properties. The form found in plasma, and capable of binding proteinases, is the electrophoretically 'slow' S-form. S-a2M can be converted into F-a2M, the electrophoretically 'fast' form, but we have not seen any evidence of the reverse change. An attempt has been made in Fig. 2 to summarize our main findings and conclusions on the structures of S-a2M and F-a2M.

Purification of S-a2M The purification procedure was developed with the aim of preparing a2M free of detectable contamination, and also completely in the S-form. In the past, thorough purification has normally been achieved by procedures that have yielded a mixture of the S- and F-forms (e.g. Hall & Roberts, 1978). In our procedure (see the Materials and Methods section) a typical yield of a2M from 250ml of anticoagulated plasma was 200mg (40%) before immunoadsorption, and 160 mg (32 %) after. Even before immunoadsorption, the purified a2M showed only a single precipitin arc in immunoelectrophoresis against anti-(human serum proteins) serum, and a single line against this antiserum in double immunodiffusion. Nevertheless assays by single radial immunodiffusion at this stage typically revealed contamination by haptoglobins (0.41.4%), fibrinogen (0.4-1.2%), IgA (3.8-8.1 %), IgM (0.52-2.2 %) (these values being ranges for three preparations) and a2-pregnoglobulin. After removal of these proteins by immunoadsorption, even sensitive tests by double diffusion against specific antisera revealed no plasminogen, haptoglobin, immunoglobulins, fibrinogen or a2-pregnoglobulin. On one occasion a2M was purified from plasma of haptoglobin type 2-1; 14% haptoglobin was then present before immunoadsorption, but this was successfully removed to yield x2M at the normal standard of purity. The method that we have developed avoids the disadvantages of working from serum (reviewed by Starkey & Barrett, 1977) and yields extremely highly purified a2M in satisfactory yield with a minimum of steps. Fractionation with poly(ethylene glycol) is a mild procedure that removes the ,B-lipoprotein more conveniently than does preparative ultracentrifugation, and without the conversion of S-form into

406

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

CU

E

._

-

C.,

Cd .

0.',

CUC

.= 0 0d0 o

Qs CO

EE 4)

C

2

°4

F

.Co 4._ s.o

k. >, 0-

-.

'0

o

~

._

1979

407

a2-MACROGLOBULIN FORMS F-form risked by the use of (NH4)2SO4 (see below). Gel chromatography has always been one of the most valuable steps in the purification of a2M; we obtained much better results with columns containing the mixture of Ultrogel AcA 22 with AcA 34 than with others containing Bio-Gel A1.5m, Sepharose 6B or Sephadex G-200. After removal of euglobulins insoluble in 1 % glycine, pH 6.5, the a2M contained only 6 % of contaminants detectable by our methods, and these were removed by immunoadsorption. Attempts at final purification by preparative isoelectric focusing were unrewarding. The freshly isolated a2M ran as a single zone of low mobility in pore-limit electrophoresis (Fig. 3, track a) and in rate electrophoresis. There was no protein with the mobility of the 'fast' or F-form a2M (see below), nor were there zones attributable to plasma proteins other than a2M. During storage at pH5.0-6.5 at -20°C or 4°C in various buffers, the S-a2M gradually changed into the F-form, with the appearance of transitional forms of intermediate mobility. In the intermediate stages, the protein appeared either as a doublet similar to those described previously for preparations of individual rat macroglobulins (Gauthier & Mouray, 1976) and human a2M (Zais & Roberts, 1977) or as a group of four or five bands similar to those studied by Saunders et al. (1971), according to loading. The change was accompanied by a partial or

(a)

-

-

(c)

(b)

4~w

(d)

complete loss of inhibitory activity. The rate of this deterioration varied between preparations, but S-a2M could normally be stored for several weeks in sodium cittate buffer, pH6.5, or in 20% sucrose, with little change. S-L12M that had been dialysed into 20mM-sodium citrate buffer, pH 6.5, and freeze-dried was reasonably stable during storage for several months at 40C. Formation of F-a2M

S-ar2M can be converted into the electrophoretically more mobile F-a2M by two distinct kinds of treatment: reaction with proteinases, or exposure to salts of ammonia or some closely related bases. We have studied these effects and examined the products. Effect of proteinases. Solutions of S-a2M (about 1 mg/ml) in sodium phosphate or Tris/HC1 buffers at pH values in the range 6-8 were treated with trypsin, papain or thermolysin (50,ug of commercial enzyme/mg of x2M, representing about 2 mol of proteinase/mol) during 5min at 20°C, after which each enzyme was inactivated (with di-isopropyl phosphorofluoridate, iodoacetate or EDTA respectively, each at a concentration of I mM). It was found that the treatment with each proteinase slightly increased the mobility of the a2M both in rate and pore-limit electrophoresis (Fig. 3, track b). In each system the mobilities of the 'fast'

(e)

(f )

0 S-(12 M

4i

-OWN&

slo

-

F-a2M

Half-molecule (dimer) 6

-Quarter-subunit

Fig. 3. Pore-limit gel electrophoresis of a2-M The samples, run in a gradient of increasing polyacrylamide concentration as described in the Materials and Methods section, were (a) S-a2M, (b) F-a2M formed by saturation with trypsin, (c) F-12M formed by treatment with methylammonium chloride, (d) subunit (dimeric) formed by exposure of S-a2M to pH 2.0, (e) subunit (monomeric) formed by treatment of S-a2M with 1 mM-dithiothreitol and (f) reduced tetramer formed by treating F-a2M (trypsin-treated) with 1 mM-dithiothreitol. Vol. 181

408

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

(F-) forms produced by reaction with the various proteinases were indistinguishable; a single fast zone was obtained even with some aged preparations that gave double zones before reaction. In a further experiment the a2M was titrated with increasing amounts of trypsin (5-40,ug/mg). Rate and pore-limit electrophoresis showed the formation of mixtures of S-a2M and F-a2M, with a progressive increase in the proportion of the F-form, and no detectable intermediate component. Effects of trypsin and chymotrypsin on electrophoretic mobility have previously been described (Saunders et al., 1971; Roberts et al., 1974; Hall & Roberts, 1978), and we suspect that the effect will be seen with any proteinase with which ac2M has the capacity to bind. We suggest that the change is a result of the conformational change in a2M postulated to occur in the reaction with proteinases. Effect of NH41 salts. Solutions of a2M in 50mMTris/HCl buffer, pH8.2, were made 100mM with respect to (NH4)2SO4 or 200mM with respect to methylammonium chloride for 30min at 20°C before being run in rate or pore-limit electrophoresis. Each of the ammonium salts caused an increase in mobility in both systems identical with that seen with the proteinases (Fig. 3, track c). The effect on S-a2M was restricted to the salts of low-molecular-weight ammonium analogues: it was obtained with (NH4)2S04, methylammonium chloride and ethylammonium chloride, but not under these conditions with hydrochlorides (at 200mM) of trimethylamine, triethylamine, 1,2-diaminoethane, 1,3-diaminopropane, 1,4-diaminobutane, glycine, f,-alanine, 6-aminohexanoate, hydroxylamine, ethanolamine, triethanolamine, Tris, spermidine or a variety of other organic bases. Hydrazine produced the increase in electrophoretic mobility, and also released some free half-molecules. Neither (NH4)2SO4 nor methylammonium chloride (at 200mM concentration of the anions during several hours) caused the change in mobility at pH 6.0 (50mM-sodium citrate buffer). The effects of concentration and time on the modification of a2M by (NH4)2SO4 and methylammonium chloride at pH 8.0 were studied. Mixtures containing the anions at 10, 50 or 200mM concentration were incubated at 20°C. After 30min and 6h, samples were applied to electrophoresis gels and immediately run. The degree of conversion from the S-form into the F-form in each sample was estimated visually. The effects of the two salts were indistinguishable. At 30min the three concentrations gave about 0, 90 and 100 % conversion respectively, whereas the values after 6 h were estimated to be 70, 100 and 100%. To determine whether the salts of amines structurally very closely related to ammonia and methylamine have the capacity to bring about the transition

of the S-form into the F-form at all, although much more slowly, prolonged incubations were made with 1,3-diaminopropane and ethanolamine (each 300mM) in Tris/HCI buffer, pH8.2, at 4°C. With each amine, complete conversion of S-a2M into the F-form required 5-6 days. Since the conversion is complete after 30min (and possibly much less) with (NH4)2SO4 or methylammonium chloride, it can be said that the effect of the higher-molecular-weight amines was at least several hundred-fold slower (although at a lower temperature). A sample of a2M that had been completely converted into the F-form by 30min exposure to 200mM(NH4)2S04 was dialysed against two changes of l00vol. of 50mM-Tris/HCl, pH8.1 during 48h. It was then re-examined by electrophoresis." No reversion into the S-form was detectable. We do not know what kind of reaction occurs between a2M and salts of ammonia or methylamine. Ratnoff et al. (1954) experimented with hydrazine because it was already known to inactivate complement components C3 and C4, and found that a serum inhibitor of plasmin was inactivated by this reagent and also by ammonia, methylamine and ethylamine, but not by several other amines. Norman & Hill (1958) selectively inactivated this 'immediate' inhibitor of plasmin with 0.2M-methylamine in assays of 'slow' inhibitor, and the inactivation of a2M as an inhibitor of proteinases by methylamine has been reported explicitly by Steinbuch et al. (1968), Nagasawa et al. (1970), Saunders et al. (1971) and Ganea et al. (1974). The risk of inactivation of a2M by ammonia during fractionation with (NH4)2SO4 was pointed out by Mehl et al. (1964) and re-emphasized by Heimberger et al. (1971), who mentioned that the inactivation occurs at pH8 but not pH5. Fractionation with (NH4)2SO4 seems to have been used successfully by Hamberg et al. (1973) at pH 6. We have confirmed that salts of ammonia, methylamine, ethylamine and hydrazine rapidly convert a2M from the S-form into the F-form at pH8. Closely related compounds such as diaminoethane and ethanolamine are vastly less active, and none of the bases has much effect at pH 6. The pronounced pH-dependence of the inactivation of a2M by salts of ammonia or methylamine may reflect a requirement for the unprotonated form of a crucial group in the protein, or of the nitrogen base, for the interaction, but there is no apparent relationship between pK and effectiveness of the various amines. The effective organic bases are all primary amines, but other primary amines of larger molecular size have very little action on a2M. It can be postulated that, at pH 8, each subunit of a2M contains a 'pocket' capable of accommodating NH4+ or NH3 and binding it. Apparently the pocket can accommodate the extra bulk of a methyl or an ethyl group, but very

1979

409

a2-MACROGLOBULIN FORMS little more. It is unclear whether the nitrogen (Nbases) are effective in their protonated forms. If the cations react with a2M, hydrogen-bonding is presumably involved, as inorganic ions of similar size are inactive (Barker, 1968). The alternative possibility that the free bases act as nucleophiles in a covalent reaction with some highly reactive group on the protein seems to be supported by a recent brief report (Swenson & Howard, 1979). The requirement for a high concentration of Nbase to produce the transition of S-form into F-form implies that S-x2M has a low affinity for these compounds. Since dialysis does not reverse the transition, it seems improbable that the situation is a simple equilibrium in which F-a2M is the S-a2MN-base complex. A system like that in Scheme 1 seems more probable. It is proposed that, whereas S-a2M is very slowly transformed into F-a2M in normal storage, the S-a2M-N-base complex undergoes this transition very rapidly. We have no definite evidence that the reversal steps occur, however.

Properties ofS-a2M and F-a2M In this section various properties of S-a2M and F-a2M are considered, with a view to understanding more about the structure of the molecule in its two forms, and the relation of this to the biological activity of a2M. The evidence relating to reductive dissociation of a2M forms a separate section. Proteinase-inhibiting activity of S-a2M. The trypsin-inhibiting activity of S-a2M was determined as described in the Materials and Methods section. The results were then calculated in terms of mol of trypsin inhibited/mol of a2M on the basis that a2M has a molecular weight of 725000 (see the introduction). Several experiments were made with each of three preparations of S-x2M, with the following results: preparation M57, 1.6, 1.6, 1.5; M58, 1.7, 1.6; M59, 2.1, 2.3, 1.8. These results point to an integer ratio of 2: 1, which contrasts with our earlier finding of binding ratios only in the order of 1:1 (Barrett & Starkey, 1973). The a2M isolated in the earlier study from serum showed both 'fast' and 'slow' components in electrophoresis (A. J. Barrett, unpublished work) and so cannot have been fully active (see below). Also, our previous results depended on the assumption that all trypsin molecules bound by a2M are resistant

S-x2M -------------------E Very slow

N-basc

S-OC2M4N-base).

N-base

.

Fa

Scheme

Vol. 181

F-a2M

1.

F-a2M-(N-base),

to inhibition by soya-bean trypsin inhibitor. We have recently found that a certain proportion of the bound enzyme molecules do react with the soya-bean trypsin inhibitor, however (P. M. Starkey, C. A. Sayers & A. J. Barrett, unpublished work). Lack of proteinase-inhibiting activity in F-a2M. Samples of a2M that had been converted into the F-form by treatment with (NH4)2SO4 or methylammonium chloride at pH8.0 were tested for the capacity to inhibit the action of trypsin and thermolysin on dyed hide powder (see the Materials and Methods section). No inhibitory action (less than 2 % of untreated control) was detected. No recovery of inhibitory activity occurred when the F-a2M was dialysed against two changes of 100vol. of 50mMTris/HCl buffer, pH8.1. Samples of the (NH4)2SO4- or methylammonium chloride-treated x2M were also treated with trypsin (50,ug/mg, for 5min) and run in SDS/polyacrylamide-gel electrophoresis with reduction. All of the quarter-subunits were seen to have been cleaved, with the formation of fragments indistinguishable in mobility from those seen in experiments with trypsin-treated S-x2M (Fig. 4, track b). a2M treated with 200 mM-diaminopropane hydrochloride for 3 days (which had resulted in a partial transition from the S-form into the F-form) showed only 20 % of the trypsin-inhibiting activity of a control. After 5 days the electrophoretic change was complete, and no activity remained. It thus seems that the specificity of cations for the inactivation of a2M is the same as that for the modification of electrophoretic mobility, in consistency with the view that the two effects are aspects of the same phenomenon. If, as we suggest, the F-form of x2M produced by the amines is a 'closed' form of the molecule similar to the proteinase complex, it is not surprising that it lacks proteinase-binding activity. The finding that this F-a2M was cleaved by trypsin in the same way as is S-ac2M suggests to us that the transition of S-form into F-form involves primarily a change of tertiary structure in the subunits, whereas the pattern of sensitivity to trypsin is dictated by secondary (as well as primary) structure. It is also important that the pattern of proteolysis of F-a2M clearly distinguishes the molecule from denatured x2M, which is cleaved quite differently from S-a2M, presumably because the secondary structure has been disturbed (Harpel, 1973; Barrett et al., 1974). Effect of removal of sialic acid on electrophoretic mobilities. S-a2M and the electrophoretically fast x2M-trypsin complex were treated with clostridial neuraminidase (10,g/ml) in 100mM-sodium acetate buffer, pH 5.5, containing 5 mM-CaCl2 at 37°C during 16h. No attempt was made to quantify the liberation of sialic acid, but the electrophoretic mobility of both forms was markedly decreased, presumably owing to

410

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

the loss of sialic acid residues and lowering of net negative charge. Thus the difference in electrophoretic mobilities did not tend to be obscured, as would have been expected if sialic acid contributed more to the mobility of one form than the other. Gel chromatography of o2M forms. It seems likely that the greater mobility of F-ac2M in gel electrophoresis reflects a more compact molecular structure. An attempt was made to detect this by gel chromatography. Mixtures of S-a2M (untreated) and F-a2M (methylammonium chloride-treated), in equal parts, were run on a column of Ultrogel AcA 34 in 100mMsodium citrate buffer, pH 6.5. Electrophoretic analysis of the effluent fractions failed to reveal any separation of the two forms in several such experiments. Further experiments were made with mixtures of S-a2M with a2M complexed with ["4C]acetyl-chymotrypsin. Again no separation of the S-a2M from proteinase complexes (identified as radioactivity) was obtained. These results were taken to show that there is no large difference in Stokes radius between S- and F-forms of a2M. Nevertheless the resolution of gel electrophoresis is undoubtedly greater than that of gel chromatography, and the possibility of a significant size difference is not excluded. Isoelectric focusing of a2M forms. Mixtures of equal quantities of control (S-)ac2M with F-x2M produced by reaction with trypsin (50,ug/mg; subsequently inactivated with phenylmethanesulphonyl fluoride) or with methylammonium chloride (as above) were subjected to isoelectric focusing in the LKB 8101 column. No separation of S- and F-forms was detected by electrophoresis of the fractions, suggesting that the forms do not differ appreciably in isoelectric point. Similarly, when complexes formed with ['4C]acetyl-chymotrypsin or 125I-labelled trypsin were mixed with S-a2M and focused, peaks of both radioactivity and protein (A280) were coincident, in fractions of pH5.2. We were thus unable to confirm the report by Ohlsson & Skude (1976) of a large change in pl of a2M (from 5.0 to 6.0) associated with the complexing of a

proteinase. Antigenic activity of S-a2M and F-(x2M. Triplicate

samples of S-a2M were treated with 200mM-methylammonium chloride (to produce F-a2M) or 200mMNaCl (control) for 30min at pH8.1 at 22°C. Each sample was then run in triplicate in 'rocket' immunoelectrophoresis (see the Materials and Methods section) into agarose gel containing specific anti(human a2M) serum. The complete experiment was done with each of four separate antisera, and the results are given in Table 1. It is apparent that the heights of 'rockets' for F-a2M were 12.3-29.3 % lower than those for S-a2M, the exact value depending on the antiserum. The difference in height of the 'rockets' indicates-that somewhat different sets of antigenic determinants are recognized on the two forms of ox2M by the antisera. In separate double-immunodiffusion experiments, only precipitating reactions of complete identity were obtained, so that the difference is likely to involve few antigenic determinants. The commercial standard serum (Behringwerke batch no. 1001 R) used to standardize our immunoassays for serum proteins was found to contain a2M only in the F-form. Clearly this factor would needto be taken into account, as well as the form of a2M being determined and the discriminating tendency of the individual antiserum, if one were to make valid assays of a2M by 'rocket' immunoelectrophoresis. SDS/polyacrylamide-gel electrophoresis of S-cc2M. Samples of a2M prepared without reduction (see the Materials and Methods section) ran in 7 % gels as a single major band with molecular weight greater than that of myosin heavy chain or reduced thyroglobulin, and thus possibly in the region of 400000mol.wt. (Fig. 4, track a). This was consistent with the view that the minimal covalent subunit of a2M is the half-molecule. Samples prepared with reduction showed a major band (I) of mol.wt. 182000, undoubtedly the single quarter-subunit of the molecule of about 725000 mol.wt. [in confirmation of the conclusions of Harpel (1973) and Hall & Roberts (1978)]. When the sample preparation involved heating at 100°C, the gels showed, in addition to band I, a less intense band of 139 000 mol.wt. (band II) and two further bands of 68000 and 64000 mol.wt. (bands IlIa and IlIb

Table 1. Comparison of heights of 'rockets' in Laurell electrophoresis of S-a2M and F-OC2M against four antisera The heights are expressed as percentages of the control (S-X2M) mobility, and the statistical limits are standard errors (n = 9). The antisera nos. 4602, 4603 and 2743 were from Behringwerke, whereas no. RP-5 was raised in our laboratory. Mobility (% of control)

Antiserum

Sample S-a2M F-a2M Difference

no.

...

4602 100+1.2 83.6+ 0.8 16.4

4603 100+0.7 70.7+0.7 29.3

2743 100+0.6 87.7 + 0.9 12.3

RP-5 100+1.2 74.4+ 1.0 25.6

1979

411

a2-MACROGLOBULIN FORMS (a)

(b)

(c)

(d) -

__ _..

Half-molecule (dimer)

Quarter-subunit (Component 1) Component 11 Tryptic fragments

Component lila Component Illb

Fig. 4. SDS/polyacrylamide-gel electrophoresis of a2M Samples were prepared at 100°C and run in 7% gels as described in the Materials and Methods section; (a) S-a2M, run without reduction, and (b) S-a2M, (c) a2M treated with 50,ug of trypsin/mg and (d) F-a2M (methylammonium chloride-treated), all run with reduction.

respectively) (Fig. 4, track b). In one experiment quantitative densitometry showed that the amount of stain in the band II and band III represented 24 and 16% respectively of the total. When native quarter-subunits were liberated by mild reduction (see below) and chromatographed on Ultrogel AcA 34, SDS/polyacrylamide-gel electrophoresis of the single protein peak again showed bands I, II, lIla and ITlb. We have no doubt that our minor bands II, IIIa and IlIb (previously referred to as fi, Yi, and Y2; Barrett & Salvesen, 1978) correspond to bands II and III described by Harpel (1978), which were shown by peptide 'mapping' to be fragments of the quarter-subunit. In consistency with the latter report, we found (not shown) that the minor bands were much weaker when samples were prepared by incubation at 37°C rather than 100°C. It seems that components II and III arise from cleavage of a peptide bond about one-third of the way from one end of some of the chains (in contrast with the cleavage near the mid-point that triggers the trapping of a proteinase by a2M), and we considered whether the bond cleavage occurs in vivo, during storage in vitro or during the treatment with SDS before electrophoresis. Comparison of some older preparations with new ones gave no indication that the quantity of components II and III increases during storage. It was also found that precipitation of samples with 10 % (w/v) trichloroacetic acid before boiling with SDS, or the use of various proteinase inhibitors, did not decrease the quantity of components II and III, suggesting that enzymic Vol. 181

activity was not involved at that stage. Like Harpel (1978), we consider that it is unlikely that proteolytic cleavage of the chains occurs at any stage in vitro, but we are not able to decide whether the cleavage occurs by limited proteolysis in vivo or chemical cleavage in the SDS at 100°C. It may be that some subunits are subject to this limited proteolysis in vivo, and that since this does not trigger the transformation of S-form into F-form, the molecules remain in the circulation. It would then be assumed that the association of the fragments is so tight in S-a2M that high temperature is required to separate them in SDS. Alternatively it might be that the bond cleavage is due to chemical cleavage of an especially labile bond at high temperature. This point is considered further in connection with the results of SDS/ polyacrylamide-gel electrophoresis of F-a2M below. SDS/polyacrylamide-gel electrophoresis of F-ac2M. SDS/polyacrylamide-gel electrophoresis of a2M that had been converted into the F-form by treatment with proteinases or ammonium salts was done with and without reduction of the samples. Samples of a2M that had been treated with 2 molar equivalents of trypsin, pancreatic elastase or thermolysin, and run without reduction, showed the presence of half-molecules; no lower-molecularweight material was detected, but sometimes traces of intact tetramer remained. It was concluded that the fragments of subunits produced in the reaction with proteinases are held together in the halfmolecules by disulphide bonds. When run with reduction, the samples showed the expected frag-

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

412

mentation of the quarter-subunits (Barrett et al., 1974), but the electrophoretic resolution was now sufficient to separate the individual fragments (mol.wts. 111000 and 98000, for tryptic digestion) (Fig. 4, track c). a2M that had been converted into the F-form by treatment with methylammonium chloride was run in SDS/polyacrylamide-gel electrophoresis with reduction. A single zone corresponding to band I (the intact polypeptide chain) was seen, but the minor bands II and III that had been seen with S-a2M (see above) were absent, even when samples were prepared at 100°C (Fig. 4, track d). Still harsher conditions of sample preparation were imposed, by inclusion of 5M-urea in the sample, by autoclaving the sample (115°C for 5min) or by succinoylation of the F-ox2M with a 5-fold weight excess of succinic anhydride (Klotz, 1967) before the SDS treatment, but still the bands II and III did not appear. In contrast, the demonstration of the fragments of 111000 and 98000 mol.wt. was not interfered with when trypsin-treated a2M was further treated with methylammonium chloride before SDS/polyacrylamide-gel electrophoresis. The failure of the F-a2M produced by methylammonium chloride to show bands II and III could be explained either in terms of the two-thirds/one-third cleavage of some quarter-subunits occurring in vivo, or in the preparation of samples for electrophoresis. In the first case one would conclude that the new conformation induced in the subunits by the amine was so rigid that the fragments were not released even under harsh conditions in SDS; in the second, one

...

(a)

(b)

(c)

_.

(d)

would have to assume that the reaction with the amine prevented the chemical attack in hot alkaline SDS by a direct modification of the scissile region, or through a conformational effect, or again prevented separation of the fragments. Inactive F-a2M formed in prolonged storage of S-a2M at 4°C tended not to show fragments smaller than mol.wt. 182000, like methylamine-treated rather than trypsin-treated a2M. Subunit interactions of S-a2M and F-a2M as indicated by SDS/polyacrylamide-gel electrophoresis after treatment with glutaraldehyde. Dilute solutions (0.50 mg/ml) of S-a2M and F-a2M (methylammonium chloride- or trypsin-treated) were exposed to lOmM-glutaraldehyde in 50mM-sodium borate buffer, pH8.5, for 30min at 22°C, and run reduced and unreduced in SDS/polyacrylamide gels. The S-a2M showed the formation of covalent dimers (half-molecules) in both unreduced and reduced samples (Fig. 5, tracks a and b). F-x2M produced with methylammonium chloride or trypsin yielded only tetramers (Fig. 5, tracks c and d). Reduced and alkylated quarter-subunits of cx2M did not show any tendency to form dimers or tetramers under the same conditions of treatment with glutaraldehyde (Fig. 5, track e). Since the separated subunits were not linked under the conditions used, it seemed clear that the crosslinking detected with the intact molecules was attributable to the way in which the subunits associate in the tetramers. In the S-a2M, cross-linking occurred between subunits normally linked by disulphide bonds, but not between those associated

(e)

(f) Tetramer

-

* Dimer

7-Monomer

-Components 11 and III

i:

Fig. 5. Effect ofglutaraldehyde cross-linking on S-c2M and F-a2M seek in SDS/polyacrylamide-gel electrophoresis The samples run in a 5 % gel were (a) S-OC2M treated with glutaraldehyde (rduced), (b) as (a), but unreduced, (c) F-a2M (trypsin-treated) treated with glutaraldehyde (reduced), (d) as (c), but with F-a2M (methylammonium chloride-treated), (e) alkylated (carboxymethylated) quarter-subunits treated with glutaraldehyde (reduced), and (f) S-OC2M untreated (reduced). S-a2M was mainly cross-linked in dimeric half-molecules that separated without reduction. Both forms of F-a2M were completely cross-linked as tetramers. No oligomers were produced in the control with isolated subunits, and the untreated S-x2M sample was included merely for reference. Further experimental details are given in the text.

1979

o2-MACROGLOBULIN FORMS non-covalently. In contrast, the contacts between subunits in F-a2M permitted stabilization of tetramers by linking of each subunit with at least two others. In view of other indications in the present paper, we consider it very likely that in fact there are larger areas of close contact between subunits in F-a2M than in S-a2M. S-a2M that had been treated with 5 mM-glutaraldehyde for 30min had no detectable trypsin-binding activity, nor was electrophoretic mobility increased after exposure of the treated protein to the enzyme. Separation of half-molecules of a2M in denaturing conditions. Previously it has been reported that exposure of a2M to low pH, urea, sodium thiocyanate or SDS without reduction causes the protein to dissociate into the minimal covalent subunit halfmolecules (Jones et al., 1972; Harpel, 1973; McConnell & Loeb, 1974; Starkey & Barrett, 1977). Samples of S-a2M, and of F-a2M produced with trypsin or methylammonium chloride as usual, were kept in 5 mM-sodium phosphate buffer, pH 7.0, containing 4.0M-urea, for 4h at 22°C, and then run in pore-limit electrophoresis. It was seen (Fig. 6) that the S-a2M had dissociated almost completely, whereas the F-a2M samples still contained mainly tetramer. Results with KI or NaSCN as dissociating agent, in place of urea but under the same conditions, were very similar. Pore-limit electrophoresis of S-a2M and F-a2M at pH 3.1 showed complete dissociation of the S-a2M, much less dissociation of the methylammonium chloride-produced F-x2M (a)

(Ib)

Ic)

_lll

]-Tetramer

3m Dimer

0_

Fig. 6. Dissociationi of half-molecules of a2M in urea The samples run on a pore-limit electrophoresis gel as described in the were (a) S-a2M, (b)

Materials and Methods section F-12M (trypsin-treated) and (c) F-a2M (methylammonium chloride-treated), each incubated in 4M-urea for 4h at 22°C before the run. Dissociation of the tetramer was much more complete for S-X2M than for either form of F-a2M. Vol. 181

413 tetramer, and less still for the trypsin-produced F-a2M molecule. The above results again point to the strength of subunit interactions in F-a2M. Krebs et al. (1978) similarly concluded from their work with the ultracentrifuge that a2M complexed with trypsin is more resistant to dissociation at low pH than is unmodified a2M. Lack of inhibitory activity of half-molecules. S-a2M almost completely converted into half molecules during 40h in 3M-urea was dialysed into 0.15 M-NaCl/10 mM-sodium phosphate/0.9 % glycine, pH7.9, as described by McConnell & Loeb (1974), and tested for inhibitory activity against trypsin and thermolysin. No activity was detectable. This finding contrasted with that of McConnell & Loeb (1974) for plasma kallikrein. Treatment of the half-molecules with 50,ug of trypsin/mg at pH 8 caused fragmentation different from, and more extensive than, that seen with 'fast' or 'slow' tetramers, or native quarter-subunits (see above), in that degradation was apparent on SDS/polyacrylamide gels even without reduction, and the pattern of bands produced with reduction was quite different from that described above for the native protein. Our conclusion was that halfmolecules dissociated in urea behave as if denatured. Jones et al. (1972) similarly concluded that there was extensive unfolding of half-molecules dissociated in urea, on the basis of ultracentrifugation experiments.

Effect of mild reduction on forms of a2M Reductive dissociation of S-a2M. S-a2M, at a concentration of about 1 mg/ml in 50mM-Tris/HCl buffer, pH 8.1, was treated with 1 mM-dithiothreitol for 30min at 22°C. The solution was then made 3 mM with respect to sodium iodoacetate, and samples were run in electrophoresis. Electrophoresis in 5 % polyacrylamide gels showed the complete disappearance of the normal x2M band after dithiothreitol, and the formation of a new component that ran as a more-diffuse faster zone. In pore-limit electrophoresis the new component ran as a sharp zone at much higher gel concentration (slightly behind catalase, 230000 mol.wt.) (Fig. 3, track e). In SDS/polyacrylamide-gel electrophoresis the single zone of mol.wt. 182000 attributed to the quartersubunit chains was seen even when samples were prepared without reduction in the SDS/polyacrylamide-gel electrophoresis. When the experiment was repeated with exposure to dithiothreitol in 50mM-sodium citrate buffer, pH6.0, rather than the pH8 buffer, electrophoresis showed dithiothreitol to have had very little effect. It was concluded that I mM-dithiothreitol, pH 8, reduces the intersubunit disulphide bonds in a2M, and that the quarter-subunits then dissociate. The lack of effect at pH 6 was attributable to the low

414

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

concentration of the reactive thiol anion of dithiothreitol at this value, and is typical of thiol-disulphide reactions. Jones et al. (1972) have previously detected dissociation of a2M after exposure to 10mM-Nacetylcysteine, pH 8.0, by ultracentrifugation. The alkylated quarter-subunits could be stored in solution (1mg/mI) in the Tris/HCI buffer, pH8.1, at 4°C for several weeks without appreciable precipitation. Thiol content of separated quarter-subunits. S-a2M (2mg/ml in 50mM-Tris/HCI buffer, pH8.0) was treated with 1 mM-sodium iodoacetate for 20min to block any free thiol groups before treatment (30 min) with dithiothreitol (2mm, 1.5mm excess over iodoacetate) to cause dissociation, as described above. The reduced protein was freed from dithiothreitol by passage through a column (13 cmxl cm) of Sephadex G-25 in 50mM-sodium acetate buffer, pH4.7, containing 2mM-disodium EDTA, and bubbled with N2 before use. Fractions comprising the protein peak from the Sephadex column were combined, and the material was examined by pore-limit electrophoresis and assayed for protein (Lowry et al., 1951) and thiol groups (ElIman, 1959). Pore-limit electrophoresis showed that reductive dissociation had not been quite complete, as some ('slow') a2M molecules remained with the quartersubunits. Nevertheless 3-4 thiol groups were detected for each unit of 182000 mol.wt. in two separate experiments. It was concluded that treatment with dithiothreitol sufficient to cause complete dissociation of a2M into quarter-subunits is associated with the reduction of about four disulphide bonds per half-molecule. We cannot say, however, whether all of these are inter-subunit bonds or whether some may be internal. Reoxidation of quarter-subunits. In further experiments reduction of S-a2M with 1 mM-dithiothreitol at pH 8 was followed, not by alkylation of thiol groups, but by addition of 2-hydroxyethyl disulphide (to 2 mM), to allow re-formation of disulphide bonds. Pore-limit electrophoresis of samples taken after 16h at 4°C and alkylated with iodoacetate showed the disappearance of the quarter-subunits, and the reappearance of S-form tetramer with a faint band of the intermediate mobility characteristic of halfmolecules. As was expected, no reassembly of tetramers occurred when alkylated quarter-subunits were treated with hydroxyethyl disulphide. Alkylated and non-alkylated quarter-subunits and re-formed tetramers were tested for inhibitory activity against trypsin and thermolysin (with the hide powder substrate). The quarter-subunits showed no measurable inhibitory activity, whereas the reoxidized samples showed 50-90% of the initial inhibitory activity of the a2M (four experiments). The recovery of activity after reoxidation of the

quarter-subunits indicated that the dissociation of a2M by mild reduction liberated quarter-subunits in their native state. In an earlier study of reductive dissociation of a2M, Jones et al. (1972) noted that the subunits retained the optical-rotary-dispersion characteristics of the parent molecule, and therefore probably retained their native structure. Effect of mild reduction on F-a2M. F-a2M that had been formed by saturation with trypsin or treatment with 200 mM-NH4+ or -methylammonium ions at pH 8 was treated with 1 mM-dithiothreitol at pH 7.1 for 30min, as described for the S-form. Electrophoresis in the rate and pore-limit systems showed no dissociation (after trypsin) (Fig. 3, track f) or very little (after methylamine). Nevertheless SDS/polyacrylamide-gel electrophoresis without reduction showed. that the disulphide bonds linking quartersubunits had been cleaved by the dithiothreitol, just as they were with the S-form. The a2M that had been treated with trypsin showed 'intact' quarter-subunits when samples were not reduced in SDS, but the expected smaller fragments after reduction. We concluded that the conformation of the subunits in the 'fast' form of ac2M is such that there are important non-covalent interactions between them, and the disulphide bonds, though not protected from reduction, are no longer essential for the integrity of the tetrameric molecule. Disulphide bonds within subunits (i.e. intra-chain bonds) link the peptides produced by tryptic cleavage, and at least some of these are resistant to reduction by 1 mM-dithiothreitol. Effect of proteinases or ammonium salts on free quarter-subunits. A preparation of free subunits was obtained by treatment of S-a2M with 1 mM-dithiothreitol, at pH 8.0, as described above. Part of the preparation was alkylated with 3mM-sodium iodoacetate for 30min. Samples were treated with trypsin (40,ug/mg of protein) or thermolysin (40,ug/ mg of protein) or papain (6ug/mg of protein) as described for intact a2M above, and others were made 200mM with respect to methylammonium chloride and left at 20°C. Samples were applied to pore-limit electrophoresis gels at various times. When unblocked quarter-subunits were treated either with the proteinases or with methylammonium chloride for 30min, pore-limit electrophoresis of the products showed a strong band of F-form tetramers and a weaker band of intermediate mobility attributable to half-molecules. When papain was used to bring about the reassembly of unblocked quarters, staining for enzymic activity showed papain both in the F-a2M zone and (more weakly) in the zone corresponding to half-molecules. Similarly the alkylated subunits re-formed F-form tetramers when treated with trypsin or thermolysin, and also in the presence of methylammonium 1979

a2-MACROGLOBULIN FORMS

chloride, although 3 h incubation was required for substantial conversion by the amine. SDS/polyacrylamide-gel electrophoresis with reduction of F-a2M formed from quarter-subunits by treatment with trypsin showed fragments indistinguishable from those produced by proteolytic cleavage of intact a2M. SDS/polyacrylamide-gel electrophoresis without reduction showed that the tetramers were not disulphide-bonded. It seemed probable from these results that isolated quarter-subunits of S-cx2M are affected by exposure to (NH4)2S04 or methylammonium chloride, or by limited proteolysis, just as they are in intact S-form tetramers. They are converted into a state in which they have the capacity to form strong non-covalent interactions with other subunits, and the assembly of dimers and tetramers is favoured. When a proteinase is used, some enzyme molecules may be trapped as F-form tetramers re-form, and binding by halfmolecules also may occur.

Conclusions The aim of the present study was to obtain a clearer idea of the molecular structure of a2M, and of the mechanism of interaction with proteinases. If it is true that the molecule acts literally as a trap one would like to know the mechanics of this. We agree with previous reports (e.g. Hall & Roberts, 1978) that the subunits of a2M seem to be identical. They give a single band in pore-limit electrophoresis after reductive dissociation, and are antigenically homogeneous (Jones et al., 1972; McConnell & Loeb, 1974; Starkey & Barrett, 1977). It remains to be established whether the two-thirds/one-third cleavage of the polypeptide chain that is seen in electrophoresis in the presence of SDS is a natural phenomenon or an artifact of the method. It is also unclear why this fragmentation is not seen with F-a2M produced by ammonium

compounds. Properties of subunits liberated by dissociation of S-a2M in urea or at pH3, or in SDS without reduction, plainly point to the existence of identical half-molecules, again in agreement with Harpel (1973), McConnell & Loeb (1974) and Hall & Roberts (1978). In S-x2M it seems that the disulphide bonds stabilizing the half-molecules are a major element in the structure of the tetramer. These are rather easily reduced, with the result that the quarter-subunits separate. Active S-a2M can be regenerated in high yield by reoxidation; however. The areas of contact between half-molecules in S-a2M may not be extensive, since glutaraldehyde cross-links did not tend to be formed. S-a2M was converted into F-a2M by two different kinds of treatment, namely reaction with a proVol. 181

415 teinase or exposure to an ammonium salt. The F-a2M molecules formed are not identical, because reaction with a proteinase involves peptide-bond cleavage in some or all of the subunits. Nevertheless the properties of F-a2M molecules formed in the different ways were very similar: their electrophoretic mobilities were indistinguishable, they lacked proteinase-inhibiting activity and there was a characteristic tightening of subunit interactions. Thus reduction of inter-chain disulphide bonds did not cause dissociation of the F-tetramers, and separation of half-molecules occurred less readily in the presence of chaotropic agents and at low pH. The tightening of subunit interactions was tather more clear-cut in the proteinase-generated a2M, but both types readily formed cross-linked tetramers with glutaraldehyde, and we consider that it is a reasonable working hypothesis that F-a2M produced by the action of ammonium salts has a very similar structure to that of a2M-proteinase complexes. This implies that the conformational change in the subunits that causes them to interact as they do in the F-tetramer can occur without any proteolytic cleavage. It seems unlikely that the proteinase molecules themselves are directly responsible for holding the subunits of F-a2M together, as was suggested by Krebs et al. (1978). Rather, it seems that each individual quarter-subunit as existing in S-x2M can respond to proteolysis or an environmental change (e.g. a high concentration of ammonium salts) by a conformation transition that creates strong noncovalent binding sites for other F-subunits. The new interactions can occur whether or not the S-subunits are already linked in pairs by disulphide bonds in the S-tetramer, but once they have occurred the disulphide bonds are no longer required to hold an F-tetramer together. We do not yet know what is the minimum number of subunits that must be proteolytically cleaved to trigger the binding of a proteinase molecule and the complete transition of S-a2M into F-a2M. It is, however, most improbable that all four subunits must be cleaved, so presumably some subunits normally adopt the F-conformation without peptidebond cleavage. Perhaps the transition from S-form into F-form of a subunit in S-a2M causes a change in the environment of the adjacent uncleaved subunits such that they tend also to undergo the same transition. The ammonium salts may mimic this environmental influence. Since the a2M molecule is a 'dimer of dimers', each form of dimer behaving as a 'closed structure' in the terms of Cornish-Bowden & Koshland (1971), it seems certain that all of the subunit interactions are 'isologous' (Monod et al., 1965), as they are indicated in Fig. 1. Closed tetramers can exist either as planar ('square') arrangements or as pseudo-tetrahedra. The data available for a2M do not give any definite

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

416

(a)

(b)

Fig. 7. Planar and tetrahedral arrangements possible for the tetramer If the 'a'-faces of the subunits in the planar form of the tetramer (a) are imagined to carry potential sites of interaction (marked 't'), the transition to the pseudo-tetrahedral form (b) would involve two subunits coming up out of the plane of the paper to make the new interactions, and the other two going down to do likewise. The transition would seem to produce molecules that should pass more readily through sieving gels, would have more contacts between subunits and would enclose space. Again, the shapes indicated for the subunits and the whole molecule are not intended to resemble their actual shapes, which are far more complex. Moreover, there is an important oversimplification, because it is most improbable that the subunits are the same shape in the S- and F-forms of the tetramer.

information in favour of either arrangement, but it is tempting to speculate that the S-form may be planar, as it is represented in Fig. 1, whereas the F-form may be tetrahedral. In the planar form each subunit would interact with only two of the others, which would be consistent with the ease of dissociation of S-a2M. It could then be suggested that the new conformation of each F-subunit allows binding to three others, to form a pseudo-tetrahedron. In Fig. 7 this would be represented by the two subunits of which we see the a-faces coming up out of the plane of the paper to form new interactions at the sites 't', while the other two go down, to bind similarly. The idea that the transition of S-form into F-form is a change from a planar to a tetrahedral arrangement of subunits is attractive for several reasons. Firstly, the tetrahedral molecule could well be more compact than the planar form. Electron microscopy has already shown that the F-a2M molecule is more compact than that of S-12M (Barrett et al., 1974), and this presumably explains why F-a2M migrates more rapidly through gel media than does S-a2M, although the molecular weights are the same. Secondly, the larger number of interaction sites between subunits could help to explain the greater resistance of the F-tetramer to dissociation. Finally, and most speculatively, it may be significant that the tetrahedron, being a three-dimensional array, can enclose space (an essential requirement for a physical trap), whereas the planar arrangement of subunits might not do so. A crucial requirement of the 'trap' hypothesis for the action of a2M was the conformational change representing closure of the trap (Barrett & Starkey, 1973). Such a change has already been indicated by

electron microscopy and some less direct lines of evidence, but the existence of the S- and F-forms of a2M, and their interconversion by proteinases, represents valuable confirmation. Since the hypothesis was put forward the interaction of a2M with many more proteinases has been described, so that the idea that the reaction is rather general for endopeptidases is supported, although it is also clear that some very specific proteinases do not react at an appreciable rate. The original proposal that only 1 mol of proteinase is bound/mol of a2M now seems to have been incorrect; instead, it seems that each molecule of a2M can act as a double trap. Native quarter-subunits do not bind proteinases, and we were not able to isolate active half-molecules, but we believe that native half-molecules might well have activity. Nevertheless we have seen no electrophoretic evidence for the formation of molecules that are 'half-fast', in titration experiments, so the two traps may not act independently. A consequence of the great structural stability of F-a2M, seen in the individual subunits by the resistance to separation of components II and III, and in the whole molecule by the tight association of subunits, may well be that the molecule is rather resistant to proteolysis 'from within', and is thus an effective trap. There are, however, some characteristics of the a2M-proteinase reaction that were not predictable from the 'trap' hypothesis. These include the finding that some molecules of trypsin bound by a2M are available to inhibition by soya-bean trypsin inhibitor (P. M. Starkey, C. A. Sayers & A. J. Barrett, unpublished work) and the apparently covalent binding of plasmin to a2M (Harpel, 1978). The hypothesis can 1979

a2-MACROGLOBULIN FORMS be elaborated to explain such findings, however, and may remain the simplest hypothesis consistent with all of the data. The rapid clearance of x2M-proteinase complexes from the circulation (Ohlsson, 1971) is probably due to the recognition of the characteristic structure of F-a2M by the reticuloendothelial cells. If so, F-ac2M produced by the action of NH4+ salts should also be cleared rapidly. Note Added in Proof (Received 3 May 1979) Recent experiments in which S- and F-ac2M have been subjected to isoelectric focusing in thin-layer polyacrylamide gel have given results qualitatively similar to those of Ohlsson & Skude (1976), in contrast with the lack of separation in sucrose gradients. It seems that one system or the other does not give a correct impression of the respective isoelectric points of the two forms. We thank our colleagues, especially Dr. Phyllis M. Starkey and Dr. C. Graham Knight, for much valuable discussion. Dr. J. Darnborough, Blood Transfusion Centre, Long Road, Cambridge, U.K., kindly made available the suipply of blood. We are also grateful to Dr. M. Steinbuch, Centre National de Transfusion Sanguine, Paris, France, and Dr. R. C. Roberts, Marshfield Medical Foundation, WI, U.S.A., for discussions and the communication of unpublished results.

References Barker, A. V. (1968) Biochim. Biophys. Acta 168, 447-455 Barrett, A. J. (1974) Biochimn. Biophys. Acta 371, 52-62 Barrett, A. J. & Heath, M. F. (1977) in Lysosomes: a Laboratory Handbook, 2nd edn. (Dingle, J. T., ed.), pp. 19-145, North-Holland, Amsterdam Barrett, A. J. & Salvesen, G. S. (1978) in The Physiological Inhibitors of Blood Coagulation and Fibrinolysis (Collen, D., Wiman, B. & Verstraete, M., eds.), pp. 247-254, Elsevier/North-Holland, Amsterdam Barrett, A. J. & Starkey, P. M. (1973) Biochem. J. 133, 709-724 Barrett, A. J., Starkey, P. M. & Munn, E. A. (1974) in Proteinase Inhibitors (Fritz, H., Tschesche, H., Greene, L. J. & Truscheit, E., eds.), pp. 574-580, Springer-Verlag, Berlin Bolton, A. E. & Hunter, W. M. (1973) Biochem. J. 133, 529-539 Chase, T. & Shaw, E. (1967) Biochem. Biophys. Res. Commun. 29, 508-514 Cornish-Bowden, A. J. & Koshland, D. E. (1971) J. Biol. Chem. 246, 3092-3102 Cuatrecasas, P. & Anfinsen, C. B. (1971) Methods Enzymol. 22, 345-378 Dunn, J. T. & Spiro, R. G. (1967) J. Biol. Chem. 242, 5549-5555 Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70-77 Ganea, D., Dumitrescu, A. & Szegli, G. (1974) Arch. Roum. Pathol. Exp. .MIicrobiol. 33, 325-333

Vol. 181

417 Gauthier, F. & Mouray, H. (1976) Biochem. J. 159, 661-665 Geokas, M. C., Brodrick, J. W., Johnson, J. H. & Largman, C. (1977)J. Biol. Chem. 252, 61-67 Giroux, E. L. (1975) Biochem. Med. 12, 258-266 Hall, P. K. & Roberts, R. C. (1978) Biochem. J. 171, 27-38 Halliday, R. B. (1976) Med. Lab. Sci. 33, 23-56 Hamberg, U., Stelwagen, P. & Ervast, H.-S. (1973) Fur. J. Biochenm. 40, 439-451 Harpel, P. C. (1973) J. Exp. Med. 138, 508-521 Harpel, P. C. (1976) Methods Enzymol. 45, 639-652 Harpel, P. C. (1978) in The Physiological Inhibitors of Blood Coagulation and Fibrinolysis (Collen, D., Wiman, B. & Verstraete, M., eds.), pp. 231-238, Elsevier/ North-Holland, Amsterdam Heimburger, N., Haupt, H. & Schwick, H. G. (1971) in Proc. Int. Res. Conf. Proteinase Inhibitor-s 1st (Fritz, H. & Tschesche, H., eds.), pp. 1-22, Walter de Gruyter, Berlin and New York Hurley, C. K. (1977) Anal. Biochem. 80, 624-626 Jones, J. M., Creeth, J. M. & Kekwick, R. A. (1972) Biochem. J. 127, 187-197 Klotz, 1. M. (1967) Methods Enzymol. 11, 576-580 Krebs, G., Guinand, S. & Breda, C. (1978) C. R. Hebd. Seances Acad. Sci. 286, 1219-1222 Lakshman, T. K. & Lieberman, S. (1954) Arch. Biochem. Biophys. 53, 258-281 Laurell, C.-B. (1972) Scand. J. Clin. Lab. Invest. 29, Suippl. 124, 21-37 Laurell, C.-B. & Jeppsson, J.-O. (1975) in The Plasma Proteins (Putnam, F. W., ed.), 2nd edn., vol. 1, pp. 229-264, Academic Press, New York Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. (1951) J. Biol. Chem. 193, 265-275 Mancini, G., Carbonara, A. 0. & Heremans, J. F. (1965) ImmunochemistrY 2, 235-254 Manwell, C. (1977) Biochem. J. 165, 487-495 Margolis, J. (1973) Lab. Pract., 107-109 Markus, G. (1965) Proc. Natl. Acad. Sci. U.S.A. 54, 253-258 McConnell, D. J. & Loeb, J. N. (1974) Proc. Soc. Exp. Biol. Med. 147, 891-896 Mehl, J. W., Conne!l, W. 0. & Degroot, J. (1964) Science 145, 821-822 Monod, J., Wyman, J. & Changeux, J. P. (1965) J. Mol. Biol. 12, 88-118 Nagasawa, S., Hoon, B. H., Sugihara, H. & Suzuki, T. (1970) J. Biochem. (Tokyo) 67, 821-832 Norman, P. S. & Hill, B. M. (1958) J. Exp. Med. 108, 639-649 Ohlsson, K. (1971) Scand. J. Clin. Lab. Invest. 28, 13-19 Ohlsson, K. & Skude, G. (1976) Clin. Chim. Acta 66, 1-7 Ratnoff, 0. D., Lepow, I. H. & Pillemer, L. (1954) Bull. Johns Hopkins Hosp. 94, 169-179 Roberts, R. C., Riesen, W. A. & Hall, P. K. (1974) in Bayer Symp. 5: Proteinase Inhibitors (Fritz, H., Tschesche, H., Greene, L. J. & Truscheit, E., eds.), pp. 63-71, Springer-Verlag, Berlin Saunders, R., Dyce, B. J., Vannier, W. E. & Haverback, B. J. (1971) J. Clin. Invest. 50, 2376-2383 Shapira, E., Martin, C. L. & Nadler, H. L. (1977) J. Biol. Chem. 252, 7923-7929

418

A. J. BARRETT, M. A. BROWN AND C. A. SAYERS

Starkey, P. M. & Barrett, A. J. (1977) in Proteinases in Mamnwalian Cells and Tissues (Barrett, A. J., ed.), pp. 663-696, North-Holland, Amsterdam Steinbuch, M., Pejaudier, L., Quentin, M. & Martin, V. (1968) Biochim. Biophys. Acta 154, 228-231 Swenson, R. P. & Howard, J. B. (1979) Fed. Proc. Fed. Ani. Soc. Exp. Biol. 38, 837

Wilson, G. B. & Fudenberg, H. (1976) Pediat. Res. 10, 87-96 Wyckoff, M., Rodbard, D. & Chrambach, A. (1977) Anal. Biochem. 78, 459482 Zais, D. P. & Roberts, R. C. (1977) Clin. Cheni. 23, 590-592

1979