Purification and chemical characterization of papain-solubilized histocompatibility-2 antigens from mouse liver.

A large scale purification of histocompatibility-2 (H-2) antigens from mouse liver is described. The antigens were solubilized by a limited papain digestion of a crude preparation of liver membranes (strain A/J) and purified by ion exchange chromatography, gel filtration, affinity chromatography, and isoelectric focusing. The overall degree of purification of H-2Kk was 1,300-fold and that of H-2Dd was 1,500-fold; approximately 8 mg of purified H-2a antigens were obtained from 1 kg of liver. The purification was followed by a sensitive radioimmunoassay in which H-2a-containing fractions were used to inhibit the binding of 125I-labeled H-2a to appropriate antisera. H-2Dd and H-2Kk co-purified through all the steps but the concentration of H-2Kk was 2- to 3-fold higher than that of H-2Dd in the liver homogenate as well as in the purified H-2 preparation. beta 2-microglobulin was initially present in a 3- to 10-fold excess over H-2 in the liver homogenate, but the purified H-2 preparation contained approximately 2 mol of alloantigenic heavy chain/mol of beta 2-microglobulin. Isoelectric focusing and disc-gel electrophoresis showed a charge heterogeneity of H-2, with a mean isoelectric point of pH 4.9. Electrophoresis on sodium dodecyl sulfate gels showed one band. Denaturing conditions were required to remove beta 2-microglobulin and small amounts of impurities from H-2. The amino acid sequence of the first 27 residues of the isolated heavy chains was determined.

A large scale purification of histocompatibility-2 (H-2) antigens from mouse liver is described. The antigens were solubilized by a limited papain digestion of a crude preparation of liver membranes (strain A/J) and purified by ion exchange chromatography, gel filtration, affinity chromatography, and isoelectric focusing. The overall degree of purification of H-2Kk was 1,300fold and that of H-2Dd was 1,500-fold; approximately 8 mg of purified H-2" antigens were obtained from 1 kg of liver. The purification was followed by a sensitive radioimmunoassay in which H-2"-containing fractions were used to inhibit the binding of '2SI-labeled H-2" to appropriate antisera. H-2Dd and H-2Kk co-purified through all the steps but the concentration of H-2Kk was 2-to 3-fold higher than that of H-2Dd in the liver homogenate as well as in the purified H-2 preparation. j12-microglobulin was initially present in a 3-to lo-fold excess over H-2 in the liver homogenate, but the purified H-2 preparation contained approximately 2 mol of alloantigenic heavy chain/m01 of j&-microglobulin. Isoelectric focusing and disc-gel electrophoresis showed a charge heterogeneity of H-2, with a mean isoelectric point of pH 4.9. Electrophoresis on sodium dodecyl sulfate gels showed one band. Denaturing conditions were required to remove /12-microglobulin and small amounts of impurities from H-2. The amino acid sequence of the first 27 residues of the isolated heavy chains was determined.
The major histocompatibility complex of the mouse, the histocompatibility-2 (H-2) complex (1,2), is composed of five regions defined by recombination.
Two of these are the socalled K and D regions that code for H-2K' and H-2D antigens. These two species of glycoproteins are integral components of the plasma membrane (3)(4)(5) and apparently play a major role as targets of graft rejection. It is unknown, however, whether this is part of their normal function. H-2 antigens are very polymorphic with numerous serologically defined antigenie specificities associated with alleles of different haplotypes. Structural studies have shown (6-9) that the antigens are composed of a glycoprotein with a molecular weight of approximately 40,000 associated with a smaller protein, /32microglobulin, with a molecular weight of approximately * 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 18 U.S.C. Section 1734 solely to indicate this fact. $ Present address; Base1 Institute for Immunology, 487 Grenzacherstrasse,  Basel, Switzerland. ' The following abbreviations are used: H-2, histocompatibility-2 antigens; SDS, sodium dodecyl sulfate; IgG, immunoglobulin G; dansyl, 5-dimethylaminonaphthalene-1-sulfonyl. 12,000. All of the antigenic determinants are associated with the larger of the two components (10).
Several investigators have shown that immunochemically and immunologically reactive H-2 antigens can be solubilized from cell membranes of lymphoid cells either by detergents (11) or by a limited proteolytic digestion (12)(13)(14), but the amount of material that has been obtained from this kind of tissue has been very limited. In contrast, the HLA antigens of human origin have been successfully purified in large quantities from tissue culture cells (15,16). Hence, only limited studies on the primary structure have been carried out on H-2 antigens labeled in situ with radioactive amino acids and carbohydrates (17)(18)(19)(20), but quite extensive investigations of HLA have been reported (21-24).
We have previously described the structural characterization of purified, papain-solubilized, H-2" antigens from mouse liver (25). In this paper we provide the detailed description of the purification and radioimmunoassay of H-2". In addition, further immunochemical and structural studies of the purified H-2" are reported.

Partial
Purification of H-2 from Spleen H-2, partially purified from spleen of A/J mice, was used as the initial radiolabeled antigen for the radioimmunoassay of H-2". The purification of this material has been described earlier (9,10). Some of the properties of the partially purified H-2 from spleen are shown in Fig. 1 In the blank samples 1 to 2% of the total radioactivity was normally found in the precipitate.
The H-2 content of the material of interest was expressed by its degree of inhibition of the binding of '*"I-H-2" to the antiserum: where the fraction: bound radioactivity/total radioactivity is calculated as follows: S, in the absence of inhibitor; T, in the presence of was mixed with unlabeled, partially purified H-2" and anti-H-2Kk alloantiserum as described for the radioimmunoassay. After precipitation of the antigen. antibody complexes the radioactivity was measured.
inhibitor; B, in the absence of inhibitor but with normal mouse serum instead of antiserum.
A very reproducible quantitation of H-2 was obtained by measuring the inhibition, as described above, by serial dilutions of the samples of interest.
The appropriate dilution factor of a particular sample was determined in preliminary inhibition experiments, and a stock dilution of the sample was made. Then, 5, 10, 20, 30, and 40 ~1 of the stock solution were used as inhibiting material in the radioimmunoassay. One inhibitory unit is defined as that amount of protein that inhibits the binding of "'I-H-2 to its antibody by 50% under the conditions described above. The assay of /3z-microglobulin was done in an analogous fashion, but with rabbit anti-mouse @-microglobulin antiserum as the first and goat anti-rabbit IgG antiserum as the second antiserum.
The specificity of the radioimmunoassay used to monitor the purification of H-2 was demonstrated by inhibition experiments. Table I   Proteins were also reacted with dansyl-chloride (36) in the presence of 2% SDS and precipitated with acetone.

RESULTS
Purification ofH-2"-The preparation of crude liver membranes and their subsequent solubilization by limited papain digestion gave a recovery of more than 50% of the initial H-2 antigenic material (Table II).
A major consideration during the succeeding purification of H-2 was to avoid methods that separated proteins according to their charge since earlier studies by others (3,10,16,37) have shown that H-2, as well as HLA antigens under such conditions, behave very heterogenepusly.
Hence, the solubilized H-2-containing material was passed through a DEAEcellulose ion exchange column at a pH and ionic strength which allowed the H-2 and the bulk of the protein to remain unabsorbed to the column material. The purpose of this step was to remove highly negatively charged material such as nucleic acids; that this was accomplished is evident from the ratio of the absorptions at 280 and 260 nm, 0.7 before and 1.4 after the DEAE-cellulose column. The next step separated H-2 from the bulk of the proteins by gel filtration (Fig. 2). The H-2 activity was eluted from the column in a sharp peak. Other investigators (3) have reported H-2 activity associated with proteins of quite different sixes; such a pattern was not observed in the present study.
The following step, affinity chromatography on concanavalin A-Sepharose, was the most effective one ( Fig. 3 and Table II). A 13-fold purification and a 60% yield was obtained. The lectin from lens culinaris (lentils) has been used by other investigators (38,39) for the purification of HLA antigens and by our laboratory (9) for the purification of H-2 from spleen. However, during a preliminary study of the properties of this lectin and those of concanavalin A it was found that the latter gave a more complete separation of H-2 from impurities, with less H-2 appearing in the nonadsorbed fraction. Various column chromatographic methods were tried after the concanavalin A step but all gave relatively large losses of activity. Isoelectric focusing proved to be the method of choice for the last step of the purification.
In order to obtain a distinct separation of H-2 from impurities, it was necessary to use a pH gradient containing 4% ampholine. At lower concentrations, most of the proteins remained close together. Fig. 4 shows that the bulk of the protein as well as the activity was present in the electrofocusing column at pH values between 4.6 and 5.2, which gives a mean value of 4.9 for the isoelectric point of H-2. The absence of a sharp, well defined peak of H-2 is in agreement with an earlier report by Hess and Davies (37) who observed a heterogeneous distribution of activity after ion exchange chromatography, disc-gel electrophoresis and isotachophoresis.
Turner et al. (16) also noticed a similar heterogeneity during the purification of HLA antigens. Only the distribution of H-2Kk is shown in Fig. 5. However, the distribution of H-2Dd was identical, and even after a repeated electrofocusing in a very narrow pH gradient of the fractions containing most of the H-2 activity no separation of H-2Dd and H-2Kk could be obtained.
H-2Kk and H-2Dd co-purified through the various steps (Table II). The final purification of H-2Kk was approximately 1,300-fold with a yield of 8% of the initial activity, and H-2Dd was purified approximately 1,500-fold with a 10% yield. The apparent ratio of H-2Kk/H-2Dd was 3:l (Table II).  Purity-Polyacrylamide gel electrophoresis of the electrofocused H-2 in the presence of SDS (Fig. 5A) showed a nearly homogeneous preparation with the expected (6-9) two-component structure.
The large component had a molecular weight of approximately 37,800 and the small component had a molecular weight of approximately 12,000. Disc-gel electrophoresis in the presence of 6 M urea resolved three to four distinct protein bands (Fig. 5B). In one study (not shown) the proteins were eluted from a gel (without urea) and assayed for H-2Dd and H-2Kk activity; both activities were associated with each band. &Microglobulin, having an isoelectric point of approximately 7 (lo), has a lower electrophoretic mobility than the main H-2 bands. However, the concentration of this protein in the H-2 preparation was so low that it was not apparent after disc-gel electrophoresis in the presence of urea (Fig. 5B).
The electrofocused H-2 was also analyzed by immunochemical means. The immunoprecipitation curves for the precipitation of lz51-H-2" with alloantisera directed against H-2Dd and H-2Kk show that more of the radioactivity was present as H-2Kk than as H-2Dd (Fig. 6). A mixture of both antisera was able to form a complex with an additive amount of the radioactivity.
Similar precipitation experiments were carried out with a rabbit antiserum prepared against the purified liver H-2 (Fig. 7). These experiments suggest that over 70% of the material in the liver preparation and 65% of the spleen material could form a complex with the antiserum.
&-Microglobulin-&-Microglobulin was present in a 3-to lo-fold excess over the alloantigenic chains of H-2 in the liver homogenate of several different preparations (Table II). Seventy per cent of the excess was removed when the membranes were isolated. A further loss of &microglobulin activity occurred when the preparation was concentrated and subjected to gel filtration.
However, for the last three steps of purification, the ratio of heavy chain activity/&microglobulin activity was 3:l.
Precipitations of electrofocused H-2 with a rabbit antiserum directed against ,&microglobulin brought down less of the radiolabeled liver H-2 and less of the spleen H-2 than did precipitation with a alloantisera against H-2Dd and H-2Kk (Fig. 6). This observation again suggests that the amount of Pz-microglobulin associated with purified H-2 is less than 1:l. For one preparation of carboxymethylated H-2 (see below), both the heavy chains and &microglobulin were subjected to amino acid analysis. A ratio of 1.5 mol of heavy chain/l mol of &microglobulin was found. Carboxymethylated H-2 and Separation of Heavy Chains-For chemical analyses of the alloantigenic heavy chains, electrofocused H-2 was fully reduced and alkylated and subjected to gel filtration (25). This procedure separated the heavy chains from /32-microglobulin and small amounts of  impurities.
The carboxymethylated heavy chains were obtained in yields of 60 to 70%. Fig. 50 shows their homogeneous appearance by SDS-gel electrophoresis.
Electrophoresis in the presence of urea resolved two proteins of slightly different charge (Fig. 5C). These were identified as H-2Kk (upper band) and H-2Dd (lower band) by immune complex formation and autoradiography (25). Amino Acid Composition-The amino acid compositions of the fully reduced and carboxymethylated heavy chains of three different preparations of H-2" are shown in Table III.  The compositions agree well, which indicates that the method of purification gives reproducible results. A total half-cystine content of 5 residues/m01 (of 290 residues) was obtained. Analysis of partially carboxymethylated heavy chain showed that 1 of the 5 residues was present as free cysteine ( Gly-Ser-His-Ser-Leu-Arg-Tyr-Phe-(ti)-Thr-Ala-Val-( ) ( ) Pro-Gly-Leu-Gly-Gl~-  Parentheses indicate unidentified residues or uncertain assignments.degradation (19 nmol) was performed in the presence of 1% SDS (32). The initial yield increased to 35% but the repetitive yield was only 71%. Glycine, serine, and leucine were identified at cycles 1,2, and 5, respectively. In the third degradation the preparation (31 nmol) was reacted with 4-SPITC (32, 33) in the presence of 0.5% SDS. The sequence of the first 27 residues was established (repetitive yield, 91%) with unidentified residues at positions 13, 14, 22, and 24 (Table V and Fig. 8). An automated degradation of electrofused H-2" (2 mg) confirmed this sequence. Two NH&erminal sequences were obtained. One was that of the heavy chain (Fig. 8) and the other was that of /32-microglobulin (43). Although j&microglobulin also has valine, proline, glycine, and proline at positions 9, 15, 18, and 20, respectively, the yields at these steps established their presence in the heavy chains as well (data not shown). The native H-2" was much more soluble than the carboxymethylated chains and less than one-half the amount of material was required to obtain the same data.
The NHz-terminal sequences of H-2" agree with the published sequences of intrinsically radiolabeled H-2Kk and H-2Dd (17-20,44) with the following exceptions. Glycine, and no methionine, was found at position 1 and serine, instead of proline, was found at position 2. Histidine, in addition to valine, was detected at position 9 in only one degradation. The yield of valine at cycle 9 (Table V) indicated that H-2Kk, present in a 3-fold excess over H-2Dd, has valine at this position. Similarly, the yield of valine-21 indicates that one of the polypeptide chains has a different residue at this position. The failure to identify any PTH-derivatives at positions 13 and 14 as well as positions 22 and 24 could be due either to heterogeneity or to the fact that these are residues which are difficult to identify quantitatively, e.g. serine and arginine. Some lysine was detected at position 19 (Table V), but glutamine/glutamic acid was detected in much higher yield.

DISCUSSION
In this report, a procedure for the purification of papainsolubilized H-2 antigens that yields up to 10 mg of highly purified material has been described. The two most important factors in the purification scheme were the choice of liver as starting material and the introduction of a quantitative radioimmunoassay of H-2. The yield of purified H-2 was 8.1 pg/ liver uersus 1.3 pg/spleen (45). This 6-fold difference in yield appears to be due mainly to the difference in size of the two organs, since the specific activities of H-2 in the two membranes are comparable (data not shown). The choice of liver over spleen as a source for the antigen then has important practical consequences. In order to obtain a yield of H-2 from spleen equivalent to that from one preparation of 1000 livers, the expenditure of 5000 more mice and many more man-hours would be required.
Earlier studies in our laboratory (9) and by others (3,37) have used the antibody-mediated cytotoxicity assay to monitor the purification of H-2. This assay is time-consuming; in addition, quantitative values are difficult to obtain. The use of a radioimmunoassay in the present study has overcome these diffkulties and has made it possible to measure the relative amounts of H-2Kk, H-2Dd, and &microglobulin at all stages of the purification (Table II). This technique indicated that the two transplantation antigens, H-2Kk and H-2Dd, were not present in equal amounts in H-2 preparations purified from liver or from spleen. The 2-to 3-fold lower amounts of H-2Dd detected could not be attributed to a preferential loss during the purification since the recovery and increase in specific activity of H-2Dd closely followed that of H-2Kk. The relative amounts of the two proteins eluted from urea gels (Fig. 5C and Ref. 25) support the immunochemical observation.
SDS-gel electrophoresis and disc-gel electrophoresis of isoelectric-focused H-2 in the presence of urea revealed only minor impurities in the preparation. Chemical analyses of the electrofocused H-2 and of the carboxymethylated chain confiied the presence of only minor impurities. We conclude that the purity of the electrofocused material was approximately 90%.
The molecular weight of the papain fragments of the heavy chains was 37,800 as estimated by SDS electrophoresis and 37,000 if calculated from the sum of the molecular weights of the polypeptide chain and the carbohydrate moiety (25). These values are in agreement with those determined by Kvist et al. (45) wherein three different methods gave an average molecular weight of 37,900. Thus, although the heavy chains of detergent-solubilized H-2 and HLA both have molecular weights of approximately 45,000 (11, 24), papain cleavage yields heavy chain fragments of different molecular weights, namely 38,000 (H-2) versus 34,000 (HLA).
From the amino acid composition analyses and the limited sequence data, it appears that the primary structures of H-2 and HLA antigens are very similar (70% homology in the first 27 residues for H-2" and HLA-B7 and 85% homology for H-2" and H-2Kb, Fig. 8) as are those of the H-2D and H-2K molecules (25). Comparative tryptic peptide analysis of allelic H-2K molecules and K and D region products of the same haplotype yielded only 30 to 45% identical peptides (46). However, this technique can overestimate the variability in the primary structure; for example, an interchange of 45 out of 300 amino acid residues in two related proteins could affect the properties of a large percentage of the tryptic peptides even though there is 85% sequence homology.
Comparison of our sequence data with published sequences obtained by using radiolabeling techniques demonstrates good agreement between the two. Two differences are, however, apparent. We identified glycine at position 1 of H-2", whereas methionine was reported for radiolabeled H-2K and H-2D molecules (17)(18)(19)(20). Recently, H-2Kk and H-2Kq were reanalyzed (44). No methionine could be detected at position 1 of either antigen; glycine was tentatively assigned to this position for both molecules. Glycine has been reported at position 1 for all HLA antigens examined to date. It is possible that this position is invariant in both species. Serine was identified at position 2 of H-2" but proline has been reported for other H-