The immunoglobulin μ chains of membrane-bound and secreted IgM molecules differ in their C-terminal segments

Cell, Vol. 21, 393-406,

September

1980,

Copyright

0 1980

by MIT

The lmmunoglobulin p Chains of Membrane-Bound and Secreted IgM Molecules Differ in Their C-Terminal. Segments M. Kehry,* S. Ewald,t R. Douglas,* C. Sibley,+ W. Raschke,§ D. Fambroughn and L. Hood* * Division of Biology California Institute of Technology Pasadena, California 91 125 t Department of Microbiology Montana State University Bozeman, Montana 59717 $ Department of Genetics University of Washington Seattle, Washington 98195 5 The Salk Institute for Biological Studies P.O. Box 85800 San Diego, California 92138 1 The Carnegie Institution of Washington Baltimore, Maryland 21210

Summary The B lymphocyte synthesizes two forms of IgM molecules during its development from a stem cell to a mature antibody-secreting plasma cell. The monomeric receptor IgM molecule is affixed to the plasma membrane and triggers the later stages of B cell differentiation, whereas the pentameric secreted IgM molecule is an effector of humoral immunity. The structural differences between membrane-bound and secreted IgM molecules are reflected in the differences between their heavy or mu chains. We have previously determined the complete amino acid sequence of a murine secreted mu (ps) chain. In this study, we have compared the structures of the secreted and membrane-bound mu (p,) heavy chains by peptide mapping, microsequence and carboxypeptidase analyses. These studies demonstrate that the p,,, and p., chains are very similar throughout their VH, C&l, C,2, C,3 and C,4 domains. The B,,, and ps chains differ in the amino acid sequence of their C-terminal segments. These studies in conjunction with those carried out on the p,,, and p= mRNAs and the C, gene suggest that the cl,,, and ps chains from a given B cell are identical except for their 41 and 20 residue C-terminal segments, respectively. The amino acid sequence of the 41 residue C membrane terminal segment predicted from the corresponding s,,, mRNA is in agreement with all the protein studies reported in this paper. Introduction In the maturation of B lymphocytes to antibody-secreting plasma cells, discrete stages of differentiation can be identified (Warner, 1974). The pre-B cell synthesizes only internal mu chains (Burrows, LeJeune and Kearney, 1979). In addition to other changes,

induction of light chain synthesis accompanies the formation of the small B lymphocyte containing membrane IgM molecules as integral membrane receptors (Vitetta, Baur and Uhr, 1971). Additional steps of differentiation and proliferation then occur to form a clone of IgM-secreting plasma cells. Thus the IgM molecule can exist as a membrane-bound receptor on the surface of B lymphocytes or is secreted by plasma cells as a hydrophilic serum antibody. The secreted IgM molecule is a pentamer composed of five monomeric IgM subunits plus a joining (J) chain [(p2L&J] (Della Corte and Parkhouse, 1973) whereas the membrane-bound IgM molecule is a monomer (pL2L2). We were interested in characterizing the structural features which distinguish these alternative forms of a single class of antibody molecules. Since the p heavy chain affixes the IgM molecule to the plasma membrane, the first step in the comparison of the membrane-bound (IJ~) and secreted (& chains was the determintion of the complete covalent structure of a pLs chain synthesized by the mouse plasmacytoma MOPC 104E (Kehry et al., 1979). This study revealed that the ps chain has five homology units or domains, one variable (V) and four constant(C) region domains (V,, C,l, C,2, C,3 and C,4), followed by a hydrophilic C-terminal segment of 20 amino acid residues which displays no homology to the remainder of the ,uschain. Several lines of evidence suggest that the P,,, and pL, chains differ in structure. IgM molecules are solubilized from the cell surface only by detergents (Melcher, Eidels and Uhr, 1975) and therefore are by definition integral membrane proteins. Detergent binding studies have shown that membrane IgM molecules and isolated (Jo chains bind small but significant amounts of detergent whereas secreted IgM molecules and ps chains bind negligible amounts of detergent (Melcher and Uhr, 1977; Vassalli et al., 1979; Parkhouse, Lifter and Choi, 1980). Thus pm chains possess a structural region capable of interacting with the hydrophobic plasma membrane. Molecular weight comparisons of pLmand ps chains by SDS-polyacrylamide gel electrophoresis and by ultracentrifugation have shown that the pm chain is larger than the EL, chain by approximately 1500 daltons (Melcher and Uhr, 1973; Bergman and Haimovich, 1978). Although the pm and ps chains are glycosylated, when they are synthesized in the presence of tunicamycin they are devoid of carbohydrate. Analyses of these nonglycosylated polypeptides indicate that the molecular weight difference is due to differences in protein structure (Vassalli et al., 1979; Singer, Singer and Williamson, 1980; Sibley et al., 1980; J. Haimovich, personal communication). Translation of pLmand ps mRNA derived from B lymphoma cells also results in the synthesis of a p,,, polypeptide which is larger in molecular weight than the p”s polypeptide (Singer, et al., 1980; J. Haimovich, personal communication). In addition,

Cell 394

there seem to be carbohydrate differences between p,,. and IJ~ chains (Bergman, Haimovich and Melchers, 1977; Bergman and Haimovich, 1978). At least some of the above differences are localized to the COOHterminal region of p,,, and ps chains. This localization has been demonstrated serologically by antibody competition experiments (Fu and Kunkel, 1974) and structurally by peptide mapping and electrophoresis of Fc fragments (Yuan, Uhr and Vitetta, 1980; Parkhouse et al., 1980) and by treatment of p,,, chains with carboxypeptidase (Williams, Kubo and Grey, 1978). Peptide mapping studies on p, and ps chains derived from a single cell line (Yuan et al., 1980) and on p,,, and ps chains from both a B cell lymphoma and the IgM-secreting hybrid of the B lymphoma with a plasmacytoma (Raschke, Mather and Koshland, 1979) have indicated that structural differences between ,u,,, and ps chains are small. However, there has been disagreement on whether the COOH-terminal amino acids of ps and pm chains are different or identical (Bergman and Haimovich, 1978; Mcllhinney, Richardson and Feinstein, 1978; Williams et al., 1978; Singer et al., 1980). Previous studies on kLsand pm mRNAs and the C, gene (Early et al., 1980; Rogers et al., 1980) suggest that the pm and pLschains are identical in their four C, domains but differ in their C-terminal segments. In this paper we present a structural analysis of three types of p chains derived from the mouse B cell lymphoma, WEHI 279 (W279). This cell line synthesizes two species of p chains, ,uLmand (Jo, in addition to an internal pool of p chains (pi) that contains precursors of both p,,, and ps chains (Sibley et al., 1980). Fusion of W279 cells with the myeloma cell line MPC 11 generates MPC 11 X W279.2 (MXW) hybrid cells that secrete pentameric IgM molecules composed of W279 I-L, chains. We demonstrate that these W279 p chains contain identical variable (V) regions and that these three species of W279 p chains, pi, ,uLmand ps, differ in molecular weight and charge. The delineation of their structural differences and similarities at the carbohydrate and protein levels is the subject of the present report.

of p chains (pi). In M XW cells, the pi pool consists only of precursors to ps chains, while pulse-chase experiments indicate that the /.Lipool of W279 cells contains precursors for both p,,. and pLschains (Sibley et al., 1980). In long-term (16 hr) “steady state” labelings, the pi pool in W279 cells is predominantly composed of pre-p., chains (shown below). These p chains can be distinguished by several chemical criteria. The pl of the pLschains is similar to that of pLmchains, and both are markedly more acidic than the pi chains (Figure 1, Table 1). In addition, pm chains are 4000 daltons larger in apparent molecular weight than p., chains (Figure 1, Table 1). This molecular weight difference is not due entirely to differences in the levels of glycosylation because in the presence of the glycosylation inhibitor, tunicamycin, W279 lymphoma cells synthesize two distinct p polypeptide chains differing in molecular weight by -1500 daltons (Table 1). The smaller of the two p polypeptides corresponds in molecular weight to precursor ps chains synthesized by M XW cells in the presence of tunicamycin (Table 1). Thus the B cell lymphoma synthesizes two p chains, p,,, and ps, which apparently differ in size by 15-20 amino acid residues.

basic

acidic

light Results Localization of Three Species of W279 c Chains Three species of p chains can be defined in W279 lymphoma and MxW hybridoma cells based on cellular location: pLm(membrane), pi (internal) and pL, (secreted). Two different cell-surface labeling techniques have demonstrated that pLmchains are present on the plasma membrane of W279 cells as 200,000 dalton IgM monomers (Sibley et al., 1980). In contrast, the MXW hybridoma cells have very low levels of cellsurface p chains and secrete large quantities of IgM pentamers (W. Raschke, unpublished observations). Both W279 and MXW cells contain an internal pool

“““““““““““““““““““”

Figure

1. Three

Species

of p Chains

Synthesized

by W279

Cells

Composite tracing of two-dimensional 10% polyacrylamide-SDS gel fluorographs of 3H-tyrosineand leucine-labeled W279 fi chains. The tracing represents the alignment of three different two-dimensional gels using stained marker Ml 04E p3 chains as an internal standard. Crosshatching of the p and light chain spots indicates the complete blackening of the X-ray film. (fi,) fi internal: (pm) b membrane: (p,) p secreted. The apparent molecular weight of the b chains in the electrophoretic system is approximately 78,000 daltons; light chains 22,000 daltons.

eraand pe Chains 395

The pi and pLsChains Appear To Be Very Similar in Primary Structure The cyanogen bromide fragments from the constant region of p* chains can be resolved individually by gel filtration (Kehry, 1980). The size and order of the cyanogen bromide fragments from MOPC 104E (Ml 04E) and W279 pL, chains are shown in the schematic drawing in Figure 2. In the M104E pLs chain, fragments CNl-2, 2, 3 and 4 contain variable region sequences, whereas fragments CN4, 5, 6, 7, 8, 8-9 and 9 contain C, sequences with the numbering extending from the NH2 to the COOH terminus (CNl to CN9, respectively). The 20 residue C-terminal segment is contained in fragments CN8, 9 and 8-9. Peptides with two numbers, such as fragment CN8-9, indicate an incomplete methionine cleavage (due to an adjacent serine residue) with the resulting peptide containing two methionine fragments. The Ml 04E ,u~ chain contains five C, region carbohydrate moieties. Fragments CN5 and CN6 contain complex carbohydrate moieties (boxed in the drawing of the ,u~ chain in Figure 2) whereas fragments CN7 and CN8 contain simple or high-mannose (circled) carbohydrate moieties. Table 1. Molecular Chains Cells and @ Chain Species

Weights

Molecular

and Isoelectric

Points

of W279

p

Weighta

- Tunicamycin

+ Tunicamycin

pl Range

w279 Pm

82

5.69-5.17

pre-pm

78

69

5.89-5.58

me-p,

76.5

67.3

6.22 and 6.10

7ab 76.5

67.3

MxW Ps

we-p,

5.83-5.33 6.22 and 6.10

a Molecular b Molecular

weight weight

in daltons x 1 Om3. range of 77,000-80,600

daltons.

W279 cells were labeled with 3H-tyrosine and leutine, and the labeled pi chains were preparatively isolated as the smaller molecular weight p peak on 10% polyacrylamide-SDS gels. These pi chains were cleaved with cyanogen bromide and the fragments were compared by gel filtration with EL,chains that had been secreted by 3H-tyrosineand leucine-labeled M x W cells and similarly prepared (data not shown): The V,, region cyanogen bromide peptides of these p, and ps chains have been identified by NHP-terminal amino acid sequence analyses (see below) and were identical in size. Because the W279 p chain has an unusual methionine distribution in the VH region, this size correspondence supports the hypothesis that the M x W hybridoma cells secrete ps chains with a W279 VH region. To compare pi and ps chains throughout their sequence, the W279 pi and M X W ps chains were labeled with either 3H-tyrosine, 3H-phenylalanine, 35S-cysteine or 35S-methionine and digested with trypsin and chymotrypsin. The resulting peptides were analyzed by high performance liquid chromatography (HPLC). This method of peptide separation involves the interaction of the amino acid residues with the derivatized packing, followed by elution with a continuous exponential gradient of acetone (McMillan et al., 1979). Thus differences in the charge and hydrophobicity of peptides may be detected. The /Lrand p., chains labeled with tyrosine, phenylalanine, cysteine or methionine exhibited very similar tryptic and chymotryptic patterns (Kehry, 1980). The major peaks in the pi and ps chains elute at identical positions. These chains are therefore very similar to one another, supporting the idea that the pi chains represent primarily an internal pool of pre-ps chains. We have incorporated various 3H-labeled amino acids into the pi and IL, chains for partial amino acid sequence analyses. In addition, we isolated sufficient quantities of W279 pLs chains from mice inoculated with the MxW hybridoma tumor so that we could determine directly the sequence of most of the W279

9

Figure 2. Location and Order ogen Bromide Fragments

of W279

Cyan-

Schematic drawing of the W279 p, chain showing the sites of cleavage by cyanogen bromide, the respective fragments (CNI-2’ to CN9), and the locations of the five C, region oligosaccharides (CHO) in the V,, and C, regions as compared to Ml 04E ps chains. The complex oligosaccharides are boxed and the high-mannose oligosaccharides are circled. The location of methionine residues in the W279 VH region is very different from that in the Ml 04E Va region. Fragments in the W279 p chains were ordered by comparing their NHAerminal sequences to those of Ml 04E ,u~ chains (Kehry et ai., 1979) and VW, sequences (Kabat et al., 1976). Numbering of the fragments is based on Ml 04E ps chain cyanogen bromide fragments. The p, fragments differing in molecular weight from corresponding M104E (1. chain cyanogen bromide fragments are denoted by a ’ (prime). The sequence of W279 (I, CN9 has not been determined. Its identity to Ml 04E CN9 is based on the alignment of a tyrosine-labeled cyanogen bromide fragment with the gel filtration pattern of M104E )&. Therefore, CN9 has not been included in the diagram. Crosshatched regions indicate the regions where a radiolabeled sequence has been determined for W279 p, chains (3H-phenylalanine, proline, valine, tyrosine and leucine for all p, fragments; 3H-lysine and alanine for CN7 only). (*) The radiolabeled sequence of M XW ps chains was also determmed for these cyanogen bromide fragments. ($) Includes the sequence of MxW GN8 and CN8-9. See Table 2.

Cell 396

variable region. The amino acid sequence data for each cyanogen bromide fragment of the pi chains and the V region and COOH-terminal cyanogen bromide fragments for the ps chains are compared with the amino acid sequence of the Ml 04E ,us chain in Table 2. These data also are summarized in a diagrammatic fashion in Figure 2. Several conclusions can be drawn. First, we have determined approximately 80% of the amino acid sequence of the VH region from the W279 ,LL~chains (Table 2). The W279 VI+ region has an unusual amino acid sequence that allows it to be distinguished from all other mouse VH regions (Kabat, Wu and Bilofsky, 1976). Second, the VH regions of the MXW ,u~ and W279 pi chains are identical at all the positions where they can be compared (Table 2). Because the W279 VH region is unusual in its sequence, we believe there is a high probability that the W279 pi and MXW ps VI+ regions will be identical throughout their sequences. This conclusion is also supported by the peptide map data discussed above. Fusion of the W279 lymphoma cell line with the mye-

loma cell line therefore induces the secretion of large quantities of pLschains containing the W279 VH region. Third, the C, region cyanogen bromide peptides from the W279 pi and MXW ps chains have identical NH*terminal sequences for all the positions where they can be compared (Table 2). In addition, the W279 pi C, region is identical to that of Ml 04E ps chains (Table 2, Figure 2). These partial sequence comparisons, together with peptide map comparisons, lead to the conclusion that the majority of the pi chains and the EL, chains are probably identical in amino acid sequence. Thus, within our limits of detection, the predominant pi chain species in W279 cells represents precursors to ps chains.

Table 2. Amino Terminal

Bromide

Sequence

Analyses

of W279

Mu Chain Cyanogen

The pi Chains Are Incompletely Glycosylated Complex carbohydrate moieties are constructed from a branched mannose and N-acetylglucosamine core structure by the addition of terminal N-acetylglucosamine, galactose, fucose and sialic acid residues (Kornfeld and Kornfeld, 1976). As illustrated in Figure

Fragments

a MOPC 104E sequence from Kehry et al., 1979. ’ The sequence of CN3a was determined on intact pL.chain: the sequence of CN3b was determined on CN3 derived from pa chains succinylated prior to cyanogen bromide cleavage. Location of CN3b in the V region was determined by alignment and homology with invariant VHI sequences (Kabat et al., 1976). ( 1 indicates residues with low yields. A key to the single letter amino acid code is provided by Dayhoff (1976). Fragments differing in molecular weight from 104E fiL, chain fragments are denoted by a ’ (prime). ’ Two exceptions to the identity of the W279 p, and 104E ps C region sequences have been noted in CN7. Both residues (Tyr at position 399 and Ala at position 417) were recovered in low yield, indicating they were not a part of the major CN7 sequence. We have not identified a possible contaminating sequence which would account for the two discrepancies. d (.....I Sequence not analyzed.

p, and ps Chains 397

2, the ps chains have three complex carbohydrate moieties (Kehry et al., 1979). Treatment of a detergent-solubilized lysate of W279 cells with galactose oxidase and NaB3H4 labels p,,. chains but not pi chains (Kehry, 1980). Therefore, we conclude that the ,rrLi chains are missing terminal galactose residues and presumably the other terminal residues such as sialic acid. In contrast, the I-L, chains have terminal galactose residues and presumably are fully glycosylated. The more basic pl of the pi chains compared with that of the p,,, and ps chains (Table 1) can then be explained by a lack of terminal sialic acid residues on the complex carbohydrate structures of the p, chains. Therefore, the final stages of glycosylation have not been completed on the internal precursor p chains. The pm Chain Lacks the High-Mannose Carbohydrate Moiety Present in the C-Terminal Segment of Pre-p, Chains and p. Chains W279 cells were labeled with 3H-mannose to label all the carbohydrate moieties. The labeled pi and pm chains were isolated by size separation on 10% polyacrylamide-SDS gels and cleaved with cyanogen bromide, and the fragments were separated by gel filtration. These gel filtration profiles were compared with the profile from ps chains that had been similarly labeled and prepared from M104E cells (Figure 3). When p chain cyanogen bromide fragments are separated by this gel filtration system, the excluded column peak consists of aggregated fragments and large uncleaved peptides. A small peak migrating on the gel filtration column between CN6 and CN7 has been found only in M104E ps chains and consists of a combination of Va and C, region partial cyanogen bromide cleavage products. Since the degree of glycosylation of pII,chains by M X W cells was found to be variable, we have used the M104E p$ chain as the standard ps chain for assessing the glycosylation of W279 pi and pm chains. The order and identity of cyanogen bromide fragments in the W279 ~1 chain have been discussed previously, and the sites of carbohydrate attachment in the mouse ,uLschain C, region are illustrated in the schematic drawing in Figure 2. Several points of interest arise when the glycosylated C, region cyanogen bromide fragments of W279 pi and Ml 04E ps chains are compared. First, the pi and ps chains have very similar 3H-mannose-labeled cyanogen bromide profiles (Figure 3A). These data again support the idea that the majority of W279 pi chains are precursors to pLschains. Both chains have identical numbers of 3Hmannose-labeled C, region peptides, and corresponding peptides in pi and pL, chains incorporate 3H-mannose in identical proportions. Second, the peptides with high-mannose carbohydrate structures, CN7, CN8 and CN8-9, have identical elution positions for the pi and Pi chains. Third, the peptides with complex carbohydrate moieties, CN5 and CN6, are smaller in

Figure 3. Comparison Peptides

of ‘H-Mannose-Labeled of p,, pS and pm Chains by Gel Filtration

Cyanogen

Bromide

W279 or Ml 04E cells were incubated with ‘H-mannose for 16 or 8 hr, respectively, in low glucose (1000 mg/ml) Dulbecco’s modified Eagle’s medium. The p,, pm and pS chains were preparatively isolated as described in Experimental Procedures. No 3H-mannose was incorporated into light chains. Cyanogen bromide cleavage in the presence of carrier Ml 04E ~1~chains and gel filtration on a column of ACA54 were performed as described in Experimental Procedures. The Rrs of the fractions from the different column runs were then normalized to the internal standard M104E pS chain cyanogen bromide fragments. Normalized fractions from the internal standards were then superimposable (not shown). The peak running with the excluded volume consists of aggregated material. Free salt is eluted around fraction 285, and column runs were eluted 10 fractions past this point. Each fraction volume is 5 ml. (A) W279 pa (-----); Ml 04E pS (-). See Figure 2 for a schematic drawing of the ordered p chain cyanogen bromide fragments. (B) W279 pLmc-----j; M104E pS (-).

molecular weight in the /I,r(denoted CN5’ and CN6’ in Figure 3A) than in the ps chains. Because the galactose oxidase studies suggest that the pi chains lack the terminal sugars of the complex carbohydrate moieties, the corresponding pi chain peptides should have a smaller molecular weight than their pLscounterparts, and this is what is observed. Because the CN5’, CN6’ and CN7 cyanogen bromide peptides are contiguous in the 1;1chain sequence (Figure 2, Table 2), the smaller size of the CN5’ and CN6’ peptides cannot be explained by shifts in the locations of the corresponding methionine residues unless one postulates that additional small C, region peptides are generated by cyanogen bromide cleavage of the pi chains. The cyanogen bromide fragmentation studies on 3Hamino acid-labeled pi chains also show these reproducible molecular weight differences for pi fragments CN.5’ and CN6’ and do not reveal any such additional

Cell 398

peptides (see Figure 6). The above studies suggest that the cyanogen bromide glycopeptides in the pLs chains and in the W279 pre-pS pool (the majority of pi chains) are identical, apart from differences in the levels of glycosylation for complex carbohydrate moieties. The cyanogen bromide glycopeptides of the W279 pm and M104E pS chains are compared in a similar manner in Figure 3B. Of the four carbohydrate-containing fragments found in p, chains, three are present in the ,u~ chains: fragments CN5, CN6 and CN7. These fragments are identical in molecular weight and in the proportions of 3H-mannose incorporated into their respective carbohydrate moieties. However, the position where the expected ,LL~CN8 and CN8-9 fragments should migrate is lacking mannose. The carbohydrate structure on the pS CN8 and CN8-9 fragments is located in the C-terminal segment, and therefore the ,LL~chain is lacking this high-mannose carbohydrate moiety (see Figures 2 and 3A). This conclusion is supported by comparing 3H-mannose-labeled W279 pi (pre& and p, chains after digestion with trypsin and chymotrypsin (Figure 4). The high performance, liquid chromatographs in Figure 4 show that the pm chains are lacking one 3H-mannose-labeled peptide (crosshatched) that is present in /L chains. As we noted earlier, the majority of W279 p, chains represent pre-p., chains which have the same number of 3H-mannose-labeled cyanogen bromide fragments (Figure 3A) and trypsin plus chymotrypsin peptides (data not shown) as the Ml 04E pS chains. The missing carbohydrate residue in the C-terminal segment of p,,, chains probably reflects the absence of the carbohydrate recognition sequence present in the C-terminal segment of p., chains. The p,,, and pLsChains Have Complex Carbohydrate Moieties Located in Identical Regions After treatment of intact W279 cells with galactose oxidase and NaB3H4, the labeled p,,, chains were iso-

lated, cleaved with cyanogen bromide and compared by gel filtration to Ml 04E pS chains which were similarly treated (Figure 5). Only the fragments CN5 and CN6 contain complex carbohydrate moieties (see Figures 2 and 3) and only fragments CN5 and CN6 contain radiolabeled galactose residues (Figure 5). These peptides migrate at identical positions for the W279 p,,. chains and the Ml 04E pLschains. We therefore conclude that complex carbohydrate moieties are located in identical regions in the pm and pS chains.

The pm Chain Differs from the cS Chain in its CTerminal Segment To localize any amino acid sequence differences between ,LL~and pL, chains, the 3H-amino acid-labeled cyanogen bromide fragments of W279 p, and /.Li(prepLs)chains were compared by gel filtration (Figure 6). In particular, 3H-tyrosine was included in the pi and p,,. chains because it labels the COOH-terminal cyanogen bromide fragment of pS chains, CN9 (see Figure 2; Kehry et al., 1979). The order and identity of these W279 cyanogen bromide fragments (shown in Figure 2) were deduced from a comparison of the NH,-terminal radiolabeled sequences of the pi cyanogen bromide fragments and the corresponding sequences of the myeloma M104E pS chain (Table 2). As expected, the CN5’ and CN6’ fragments from the partially glycosylated precursor pi chains are smaller in molecular weight than the CN5 and CN6 fragments from the completely glycosylated pm chains (Figure 6). The COOH-terminal cyanogen bromide fragment CN9, which is derived from a cleavage of a methionineserine bond in the C-terminal segment of pLschains, is absent in the pm chains, although twofold larger aliquots of the last 40 column fractions from the cyanogen bromide digest of pm chains were counted for radioactivity. The CN9 fragment is present in the /.Li chains, again supporting the idea that pi chains are composed mainly of the internal precursor EL,chains Figure 4. Glycopeptides Are Different

of pm and (L, Chains

3H-Mannose-labeled W279 p chains were combined with 0.25 mg pig IgG and cleaved with trypsin followed by exhaustive digestion with chymotrypsin as described in Experimental Procedures. The glycopeptides were separated by HPLC (McMillan et al., 1979). Fraction volume is 0.5 ml (0.5 min per fraction). (. .) Gradient of acetone used to elute peptides. 3H-Mannose-labeled W279 p, (-----): 3H-mannose-labeled W279 pm (~1. The major peptide difference is indicated by crosshatching of the peak. The flow-through peak from the column which is seen only in fi, is not included as a peptide difference since this peak is not reproducible.

Fractran

Number

p? and ps Chains 399

Figure 5. Comparison of NaB3H4-Galactose Oxidase-Labeled Cyanogen Bromide Peptides of pLmand ps by Gel Filtration

300: 3

- 2000

H-DDM

3H -DPM

Pm 200(

Ps

10

I20

140

160

180

FRACTION

zoo

220

240

260

Galactose residues of cell surface W279 b chains were labeled with NaB3H4 as described in Experimental Procedures. Pentameric M104E IgM was similarly labeled in solution. The pm and ps chains were isolated, cleaved with cyanogen bromide and separated as described. The column runs were normalized to internal standard Ml 04E fis cyanogen bromide fragments. The peak running with the excluded volume consists of aggregated material. Free salt is efuted around fraction 285. Fraction volume is 5 ml. W279 km (-); M104E 115(-----).

280

NUMBER

Figure 6. Comparison of ‘H-Amino beled Cyanogen Bromide Peptides y, Chains by Gel Filtration

Acid-Laof p, and

W279 cells were incubated with either 3Htyrosine and leucine (pm) or 3H-tyrosine, leutine, alanine and valine (p,). The p, and pm - 2000 chains were isolated, reduced, alkylated and 3H - DPM cyanogen bromide-cleaved, and the peptides were separated by gel filtration as described. Pm Aliquots of isolated p, and fiL, chains used for the cyanogen bromide cleavage were run on two-dimensional SDS gels to check the purity. The $,,, and fi., chains were always 90% free of cross-contamination. The individual column runs were aligned by normalization to internal standard Ml 04E ba chain cyanogen bromide fragments. The peak running with the excluded volume consists of aggregated mate100 120 140 160 180 200 220 240 260 -on rial. Free salt is eluted around fraction 285 FRACTION NUMBER (just after CN9). Fraction volume is 5 ml. “H-Amino acid cyanogen bromide fragments of W279 p,,. (-----); F, (-1. 0.2 ml aliquots were counted for 1-1,and 0.4 ml aliquots for pLmthrough fraction 250; 0.7 ml aliquots were counted for fractions 251-290. Cyanogen bromide fragments differing in molecular weight from Ml 04E p$ cyanogen bromide fragments are denoted by a ’ (prime). CN8 and CN8-9 in p, and CN8-9 in pm are not distinct peaks due to the migration of a Vn region incomplete methionine-serine cleavage product (CN3b-CNl-2’. detected by sequence determination) between CN5 and CN8-9. See Figure 2 for a schematic drawing of the order of the W279 p chain cyanogen bromide fragments.

before secretion. Identical results also have been obtained by comparing 3H-tyrosine-labeled cyanogen bromide fragments of pi and p,,- chains from a second B cell lymphoma, WEHI 231 (R. Douglas, unpublished observations). Since we reproducibly achieve 50% cleavage of the CNB-9 methionine-serine bond in p., chains by cyanogen bromide, the absence of the CN9 fragment from the p,,. chains suggests that the pm chains lack the methionine residue nine amino acids from the COOH terminus of the pL,chains. This supposition is substantiated by a comparison of peptides from W279 pL, and M XW pLs chains that were labeled with 35S-methionine, digested with trypsin and chymotrypsin and analyzed by HPLC (Figure 7). Some smaller variable peaks are seen in the early fractions from the HPLC column due to the shallowness of the exponential gradient of acetone in this region and to the difficulty in achieving complete digestion with chymotrypsin.

Nonetheless, the HPLC peptide maps in Figure 7 clearly show that the p,,- chains lack a major methionine peptide (crosshatched) that is present in the p., chains. A similar comparison of tryptic and chymotryptic peptides from pLm and p, chains labeled with 3H-phenylalanine (Figure 8) shows that k,,, chains have two phenylalanine peptides not present in p, chains. In summary, peptide map comparisons of ,u,,, and ps (p,) chains digested with chymotrypsin plus trypsin show that the pLm chains have additional phenylalanine peptides and are lacking both one carbohydrate moiety and one methionine residue present in ps chains. Comparison of gel filtration profiles of cyanogen bromide fragments from p,,. and pLschains localizes the carbohydrate and methionine differences to the C-terminal segment of the p chains. These results lead us to conclude that the p,,, and pLschains possess different COOH-terminal amino acid sequences.

Cell 400

Figure 7. Comparative %-Methionine-Labeled

Peptide Mapping pm and ~1~Chains

of

See legend to Figure 4. %-Methionine-labeled W279 pm (-----I; 35S-methionine-labeled M x W ps (-). Peptide differences are indicated by crosshatching of the peaks.

Fraction

Number Figure 8. Comparative 3H-Phenylalanine-Labeled

4000

MPCII

x W279.2pi,---m

w279

pm

Peptide Mapping of b and pLsChains

See legend to Figure 4. ‘H-Phenylalanine-labeled’W279 pm c-----j; ‘H-phenylalanine-labeled M X W p$ (-). Peptide differences indicated by crosshatching of the peaks.

4000

are

3000

3000

-100 I 2 a 13 2000 4 w-l

>-80 2000; k-60 m -40 1000

1000

-20

i 2 -? s: .u z 5 8

I ‘0

I 20

I 40

I 60

-n I 80

Fraction

I 100

I 120

I 140

I 160

1 18:

Number

The pm Chains Differ in Amino Acid Sequence from the ps Chains at Their COOH Termini The amino acids present at the COOH termini of W279 pm, pi and M x W ps chains were investigated by digestion of the radiolabeled p chains with carboxypeptidases A and B. The p,,,, pi and ps chains were labeled with either 3H-tyrosine, leucine, valine, phenylalanine or lysine and purified. The amounts of tritium in each amino acid pool were quantitated before combining the labeled p chains as ,u,,,. pi and ps pools. Digestion of W279 pLm,pi and M XW ps chains with carboxypeptidases A and B was followed by separation and quantitation of the released amino acids on an amino acid analyzer. The time course of release of the amino acids can be used to determine the COOH-terminal amino acids present in the different p chains. As shown in Figure 9, the rapid release of tyrosine from M XW pL,chains (Figure 9A) is expected since tyrosine

is the COOH-terminal amino acid in Pi chains (Kehry et al., 1979). Background release of valine and phenylalanine is also observed for M X W pLschains, a result which correlates with the nonspecific release of these amino acids from unlabeled M104E ps chains digested with carboxypeptidase (not shown). In addition, tyrosine is the predominant amino acid released from the W279 pi chain pool (Figure 9B). This confirms the previous peptide map and sequence determinations of pi chains showing that in W279 cells, the pi chains are predominantly incompletely glycosylated precursors to ps chains. Carboxypeptidase digestion of the W279 p,,, chains (Figure 9C) demonstrates that the COOH-terminal amino acids are different from those of their pL, counterparts. Significant amounts of phenylalanine and lesser amounts of valine, tyrosine, leucine and lysine are released. The simple interpretation of these data

f.~, and pS Chains 401

would be that the pLmchains have a COOH-terminal phenylalanine residue. This assignment is not consistent with the predicted COOH-terminal amino acid sequence derived from sequence analyses of pL, and ps mRNAs and the C, gene segment (Early et al., 1980; Rogers et al., 1980). We believe that these differences can be reconciled by the assumption that the pm chains have undergone partial proteolysis during isolation. It is important to stress once again that the carboxypeptidase data do establish that the COOH-terminal segments of the p, and pLschains differ in amino acid sequence. These observations are consistent with our previous peptide map and cyanogen bromide fragment comparisons which indicate that all the amino acid and carbohydrate differences between EL, and p, chains can be localized to the C-terminal segment.

A

. VAL &’ z

:

:

O

B,..

Discussion

0 0

20

40

60

TIME

80

100

120

(MINI

Figure 9. Release of Amino Acids from pm, p, and pS Chains Digestion with Carboxypeptidases A and B

upon

W279 p, and f~,,.chains and MxW pS chains labeled with 3H-tyrosine, valine, lysine, phenylalanine and leucine were digested with carboxypeptidases A and B for varying times as described rn Experimental Procedures. The p chains labeled individually with each amino acid were pooled for the digestion, so that the quantitation of residues released is based on radioactivity in the original undigested p chain, the proportion of the total sample removed and loaded on the amino acid analyzer for each time point, and the number of residues of each amino acid present in the W279 p chain [22 tyrosine, 45 valine, 31 lysine, 22 phenylalanine and 46 leucine residues (Kehry et al., 1979)]. Controls on the I-C,, fi, and pS chains were incubated for 2 hr and consisted of an undigested aliquot to which boiled carboxypeptidases A and i3 were added. No amino acids were released in any of the control incubations. (A) Release of COOH-terminal amino acids from MxW pS chains; these chains were not labeled with 3H-lysine or leucine. (B) Release of COOH-terminal amino acids from W279 r~, chains; these chains were not labeled with 3H-phenylalanme. (C) Release of COOH-terminal amino acids from W279 km chains.

W279 Cells Synthesize Two Distinct Species of p Chains, pL5and p,,, These p chains differ in electrophoretic mobility, localization in the cells and covalent structure. The pLm chains have been localized to the W279 cell plasma membrane by galactose oxidase and lactoperoxidase labeling (Sibley et al., 1980). The pLschain is secreted by M XW cells as a pentamer [(p2L&] in the presence of a J (joining) chain (Raschke et al., 1979; M. Koshland and E. Mather, personal communication). Pulselabeling studies (Sibley et al., 1980), galactose oxidase and NaB3H4 labeling, peptide maps and carboxypeptidase experiments indicate that the p, chain pool contains precursors for ,u,,, and p., chains and that these p chains lack the terminal sugars (that is, galactose, sialic acid and presumably fucose) in their complex carbohydrate moieties. Carboxypeptidase treatment and mannose incorporation suggest that in B cell lymphomas the flu, chain precursor appears to predominate in the p, pool. The significant amounts of intracellular ps chain precursors may constitute a pool that is readily mobilized for polymerization and secretion as pentameric IgM molecules containing J chains. Perhaps the induction of J chain synthesis is an important factor in the differentiation process that rapidly converts B cells to IgM-secreting plasma cells (Raschke et al., 1979). The W279 pm and p5 Chains Are Identical Except for Their C-Terminal Segments Gel filtration comparisons of cyanogen bromide fragments suggest that the ,u,,, and pL, chains are identical except for their C-terminal segments. The extensive similarity of the pm and ps chains is supported by the comparison of tryptic and chymotryptic peptides labeled with a variety of different amino acids (tyrosine, methionine, phenylalanine or cysteine). In addition,

Cell 402

the placement of high-mannose and complex carbohydrate moieties appears identical except for the highmannose carbohydr3e moiety that is present in the C-terminal segment of pLs chains and absent in p,,, chains. This observation implies that the corresponding Asn-X-Ser/Thr carbohydrate recognition sites are preserved throughout the four C, region domains of both pm and p* chains. The 41 Residue C-Terminal Segment Sequence Predicted from Nucleic Acid Studies Is in Complete Accord with Our Protein and Carbohydrate Findings on the p,,, Chain The amino acid sequences of the C-terminal segment of the ps chain (Kehry et al., 1979) and the predicted C-terminal segment of the p,,, chain (Early et al., 1980; Rogers et al., 1980) are given in Figure 10. The following conclusions are derived from a comparison of the two amino acid sequences. -Tryptic and chymotryptic peptide maps have shown that there are structural differences between p,,, and ps chains. Peptide maps of 3H-phenylalanine-labeled ps and CL,,,chains indicate that pm chains have two phenylalanine peptides not found in pL, chains. This is what is expected from the predicted p,,, C-terminal segment sequence, taking into account the fact that three of the five phenylalanine residues in the C-terminal segment will yield free phenylalanine with this type of enzymatic digestion (Figure 10). Peptide map comparisons of 3H-tyrosine-labeled pm and ps chains also show the existence of structural differences (Sibley et al., 1980). -There is no methionine residue in the predicted I*,,, C-terminal segment. Our comparative studies on cyanogen bromide fragments have localized a methionine difference between p,,, and pLschains to the Cterminal segment. Comparison of 35S-methionine-labeled tryptic and chymotryptic peptides also indicates that pm chains lack a methionine residue present in ,uLschains. -The recognition sequence for asparagine-linked carbohydrate moieties, Asn-X-Ser/Thr, is missing in the predicted sequence of the p,,, C-terminal segment. Studies of the mannose-labeled cyanogen bromide fragments of pL, and CL,,,chains show that the COOHterminal high-mannose carbohydrate moiety is absent cp4 I.Ls

c;,e:;;g

. . ..DKSTGKPTLYNVSLI@SDTGGT~ ’

Cp4 Pm

CHO 22

M segment

. . ..DKSTEGEVNAEEEGflENLWTTASTBIVLFLLSLFYSTTVTLF(KK

in EL,,,chains. Comparison of tryptic and chymotryptic glycopeptides also clearly indicates that p,,, chains lack a site of carbohydrate attachment which is present in ps chains. -The COOH-terminal sequences of the pm and pLs chains are distinct from one another. Our carboxypeptidase studies have demonstrated that the COOH-terminal residue of the pLschains is tyrosine, as expected (Kehry et al., 1979). Carboxypeptidase analysis of the ,LL~chains demonstrates that they are different from their p, counterparts and that they have an apparent COOH-terminal phenylalanine residue (Figure 9C). This observation disagrees with the COOH-terminal lysine residue predicted from the nucleic acid studies. Two points should be raised. First, COOH-terminal lysine residues present on yl (Honjo et al., 1979) and y2b (Tucker et al., 1979; Yamawaki-Kataoka et al., 1980) chains are removed by post-translational proteolysis. If the same is true of p,,, chains, then the COOH-terminal sequence would be lysine-valine (Figure 10). We have observed that digestion of pLmchains with carboxypeptidase releases valine and lysine with kinetics that are consistent with a lysine-valine sequence in the p,,, chains (Figure 90. The next two residues expected to be released would be phenylalanine and then leucine (Figure 10). Second, there is more phenylalanine and somewhat more tyrosine and leucine released than would be expected by the above analysis. Because the predicted C-terminal segment sequence of p,,, chains is rich in these amino acids, some COOH-terminal proteolysis could cause increased amounts of phenylalanine, tyrosine and leutine to be released during carboxypeptidase digestion. Clearly the unambiguous approach to this problem would be to determine the amino acid sequence of the C-terminal segment of the ,u~ chain. This technically difficult experiment is now in progress. -The p,,, C-terminal segment appears to be 15-20 amino acids longer than its p, counterpart when the p,,, and ps chains are synthesized in the presence of tunicamycin (Table 1). This observation is in excellent agreement with the sequences predicted from the nucleic acid data indicating that the C-terminal segment for ,LL,,,chains is 41 residues in length and the ps chains possess a 20 residue C-terminal segment. The COOH-terminal sequences predicted from the Figure 10. Amino Acid Sequences of the CTerminal Segments of pm and ps Chains The amino acid sequence of the C-terminal segment of ps chains (Kehry et al., 1979) and the predicted amino acid sequence of the Cterminal segment of pm chains (Early et al., 1980; Rogers et al., 1980) are compared. The junction between the C, 4 domain and the Cterminal segment is delineated. Boxed residues are described in the text. CHO indicates the site of carbohydrate attachment. A key to the single letter amino acid code is provided by Dayhoff (1976).

EL,,,and pS Chains 403

acid data have two additional interesting fea(Figure 10). First, the pLmC-terminal segment is missing a cysteine residue found in the ps chain one residue from the COOH terminus. This cysteine residue is essential for the formation of disulfide-linked pentameric secreted IgM molecules (Mestecky and Schrohenloher, 1974). However, membrane IgM moiecules lack the penultimate cysteine residue and are only capable of forming covalent monomers on the B cell surface, as has been shown by labeling studies with galactose oxidase (Sibley et al., 1980). Second, the ,um C-terminal segment contains a stretch of 26 uncharged amino acid residues that is comparable in hydrophobicity index (Segrest and Feldmann, 1974) to the membrane-associated regions in glycophorin (Tomita and Marchesi, 1975), phage Ml 3 coat protein (Wickner, 1976) and HLA heavy chains (H. Orr and J. Strominger, personal communication). The pLmchain is therefore an integral membrane protein possessing a hydrophobic C-terminal segment capable of binding detergents (Melcher and Uhr, 1977; Vassalli et al., 1979; Parkhouse et al., 1980). Thus the predicted amino acid sequence of the pm C-terminal segment is consistent with all these chemical and biological observations on p,,, chains. nucleic

tures

The Membrane-Bound IgM Molecule Represents a Structurally Well Characterized Eucaryotic Membrane Receptor Membrane receptors transduce signals between the external and internal environments of cells and as such constitute an extremely important class of biological effector molecules. Studies on the structure of most eucaryotic membrane receptors have been limited by the very small amounts of material generally available (Cuatrecassas et al., 1975; Gill, 1976; Andres, Jeng and Bradshaw, 1977; Carpenter and Cohen, 1979). Our studies on the IgM receptor molecule and its corresponding gene and mRNA (Early et al., 1980; Rogers et al., 1980) are important in two respects. First, we have characterized in detail the structure of the IgM receptor molecule. We have demonstrated that p,,- chains are very similar to IJ~ chains throughout the VH and C, region domains. The sites of carbohydrate attachment within the four C, domains appear identical in ,u,,, and !& chains. These chains, however, differ in the amino acid sequence of their Cterminal segments. Indeed, sequence analyses of the pLm and ps mRNAs indicate that the C-terminal segments of the p, and ps chains consist of 41 and 20 residues, respectively. The protein studies of pm chains are in agreement with the p,,, C-terminal segment sequence predicted from the mRNA sequences (Figure 10). These studies indicate that the pLmchain could be a transmembrane polypeptide which spans the plasma membrane with a 26 residue stretch of uncharged and hydrophobic residues near its COOH terminus (Figure 10) (see Rogers et al., 1980). Sec-

ond, studies on the pLmand pLsmRNAs as well as the C, gene suggest that the p,,. and pu, C-terminal segments are generated by RNA splicing from the RNA transcripts of a single C, gene (Early et al., 1980; Rogers et al., 1980). Thus RNA splicing allows IgM molecules and perhaps other immunoglobulins to be expressed as integral membrane receptors, or alternatively as soluble effector molecules of humoral immunity. It will be interesting to determine whether this mechanism of developmental regulation is used in other systems where protein molecules have to function in two or more very different environments. Experimental

Procedures

Cells The WEHI 279 (W279) cells (a gift from N. Warner) were grown in stationary suspension cultures at 37°C in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% heat-inactivated fetal calf serum (fcs), nonessential amino acids (Gibco) and 5 x 10m5 M fimercaptoethanol. We subsequently obtained a later clone of W279 cells (W279.1 /12) isolated by N. Warner and sent to us by V. Oi. These two cell lines correspond to slightly different stages in B lymphocyte development (Sibley et al., 1980). W279 cells were used for the experiments illustrated in Figure 3. some of the sequences in Table 2, and the cell fusion with MPG 11 to generate MPC 11 X W279 cells. The W279.1 /12 clone was used for all remaining experiments. Our previous characterization of these two cell lines (Sibley et al., 1980; N. Warner, personal communication) indicates that pm and p$ chains isolated from the two clones are equivalent, and we refer only to W279 cells in all descriptions. The MPC 11 x W279 (M XW) hybrid cells were generated by a fusion between MPC 11 and WEHI 279 cells (Raschke et al., 1979). MPC 11 X W279.2 cloned cells were stable for IgM production for periods of up to 6 months, and were grown in DMEM supplemented with 10% heat-inactivated fetal calf serum. MOPC 104E (Ml 04E) myeloma cells were passaged subcutaneously as a solid tumor in (BALB/c X DBA/2)F, mice. The MPC 11 x W279.2 (MxW) hybrid cells were grown as solid tumors in BALB/c mice. The cells were grown in ascites form for the production of milligram quantities of Ml 04E- or W279-secreted IgM or for isolation of Ml 04E cells to be radiolabeled. Labeling W279 and MXW cells were biosynthetically labeled by washing with fresh medium and resuspending at cell densities of 2 X 1 O6 per ml in labeling medium [DMEM containing 5% dialyzed heat-inactivated fcs, 0.3 mg/ml glutamine, 5 x 10e5 M P-mercaptoethanol and nonessential amino acids (Gibco)] lacking the amino acid or amino acids to be labeled, plus 50 gCi/ml of a ‘H-amino acid (New England Nuclear). Cells were incubated in a humidified atmosphere lacking CO* for 16 hr at 37°C. Ml 04E ascites fluid was diluted 1:2 with Hanks Balanced Salt Solution (HBSS) (Gibco), ceils were purified over Ficol-Hypaque, and the buffy coat was washed three times with HBSS, 5% heatinactivated fcs, and resuspended in labeling medium at 2 X lo6 M104E cells per ml. Labeling conditions were identical for all cell lines. Following the 16 hr labeling period, the cells were harvested by centrifugation at 150 x g for IO min and the medium was removed, made 1 mM in phenylmethylsulfonyl fluoride (PMSF), 1 X 1O-4 M P-mercaptoethanol and stored frozen at -70°C. Cells were washed once in cold DMEM, 20% fcs and resuspended in lysis buffer lo.01 M Tris-HCI (pH 7.4). 0.14 M NaCI, 0.5% Triton X-l 00, 1 mM PMSF, 1 X 1 O-” M P-mercaptoethanol] at a concentration of 2.5 X 10’ cells per ml, and incubated for 20 min at 4°C. The mixture was centrifuged at 1800 X g at 4°C for IO min to remove nuclei. Supernatant fractions were removed and stored frozen at -70°C. When tunicamycin was used, cells were preincubated in 2 @g/ml tunicamycin for 1 hr prior to the addition of ?Zi-methionine. Cells were labeled in the presence of tunicamycin for 1 hr.

Cell 404

Extrinsic labeling of galactose residues with NaB3H4 and galactose oxidase was performed by the method of Gahmberg and Hakomori (1973). Purified Ml 04E IgM molecules or W279 whole-cell lysates prepared as above were labeled, desalted on Sephadex G-25 equilibrated in 0.01 M Tris-HCI (pH 7.4), 0.14 M NaCI, 0.5% Triton X-100. and stored frozen at -7OOC. lmmunoprecipitation Lysates were preincubated with washed formalin-fixed Staphylococcus aureus (Kessler, 1976) (50 pl 10% suspension of S. aureus per 100 ~1 cell lysate) for 20 min at 4°C. After centrifugation at 1800 x g at 4°C for 5 min to remove the S. aureus, the supernatant was made 0.5% in SDS and incubated with rabbit anti Ml 04E k serum (rabbit anti p) (10 pl rabbit anti p per 100 gl cell lysate) for 1.5 hr on ice. Immune complexes were precipitated by addition of washed S. aureus (100 +I 10% suspension) for 20 min at 4OC. Precipitates were washed three times in washing buffer to.01 M Tris-HCI (pH 7.4), 0.14 M NaCI, 0.5% Triton X-l 00, 0.1% SDS) and the proteins were eluted in 50-l 00 pl sample buffer 150 mM Tris-HCI (pH 6.8), 2% /3-mercaptoethanol, 2% SDS] per 100 pl cell lysate by incubation in a boiling water bath for 2 min. For samples to be analyzed on 10% polyacrylamide-SDS gels, glycerol plus pyronin Y were added to a final glycerol concentration of 15%. Culture medium was precipitated according to the same protocol except that SDS was omitted from all the buffers and 20 (11rabbit anti p were used per 1 .O ml medium. Samples to be analyzed by isoelectric focusing were precipitated by 25% trichloroacetic acid (TCA) at 4°C for 1 hr, and the precipitates were washed at 4°C twice with 20% TCA, twice with ethanol:ether (1:2), once with ether, and air-dried. Gel Electrophoresis The p,, g,,, and p= chains were isolated from preparative 10% polyacrylamide-SDS gels (Laemmli. 1970) with a 2 cm stacking gel. Proteins were eluted from 1 mm slices by incubation of each 1 mm slice in 0.5 ml 0.5% SDS for 24 hr. Radioactivity was determined for an aliquot of each sample by liquid scintillation counting and the appropriate fractions were pooled. Two-dimensional gel analysis was performed according to the method of O’Farrell (1975) or Garrels (1979). In the O’Farrell procedure, isoelectric focusing gels were 5% polyacrylamide, 0.28% bisacrylamide, 9.2 M urea, 2% Triton X-l 00, 2% ampholines pH 3.510, 0.13% ampholines pH 5-7. 0.13% ampholines pH 7-9, 0.13% ampholines pH 4-6 and 0.07% ampholines pH 3.5-5 (LKB). TCA precipitates were dissolved in 50 ~1 sample buffer (9.5 M urea, 2% Triton X-l 00, 5% P-mercaptoethanol, 0.4% ampholines pH 3.5-i 0, 1.6% ampholines pH 5-7) at least 1 hr before loading. Gels were stained and fixed in 25% isopropanol, 20% sulfosalicyclic acid, 0.025% Coomassie Brilliant Blue; destained in 7.5% acetic acid, 5% methanol; and fluorographed according to the procedure of Bonner and Laskey (1974). Cyanogen Bromide Cleavage Unlabeled Ml 04E and MxW bs chains were purified as previously described (Kehry et al., 1979). To determine the sequence of M XW CN3b. the isolated ps chains were succinylated (Klapper and Klotz. 1972) prior to cyanogen bromide cleavage. This procedure blocked the NH2 terminus of the p. chains (for example, fragment CN3a). The b,, pm or ps chains pooled from appropriate slices of 10% polyacrylamide-SDS gels were reduced and alkylated as follows, Carrier pig IgG and Ml 04E IgM (3 mg each) were added. The pooled p chains were concentrated by lyophilization and brought to a final concentration of 9% SDS, 0.5 M iris-HCI (pH 8.5); boiled for 5 min; flushed with NP for 5 min; dithiothreitol was added to a concentration of 20 mM; boiled for 2 min; sealed under N2 and incubated at 37°C for 1.5 hr. Alkylation with 50 mM iodoacetamide (3x recrystallized) was performed for 1 hr at room temperature in the dark. Reagents were removed by desalting on Sephadex G-25 equilibrated with 0.05 M Tris-HCl (pH 7.4), 0.5% SDS. Proteins were TCA-precipitated as described above, and the air-dried precipitate was dissolved in 88% formic acid and combined with 30 mg Ml 04E fis chains in 70% formic

acid. Cyanogen bromide was added (50 mg/ml) and cleavage was performed for 20 hr at 4OC in the dark with constant stirring (Gross, 1967). Fragments were separated by gel filtration as previously described (Kehry et al., 1979) on a column of ACA54 (LKB) (3.5 x 140 cm) equilibrated with 3 M guanidine-HCI, 0.2 M ammonium bicarbonate, 0.02% NaN3. 5 ml fractions were collected. Aliquots were dissolved in Aquasol (New England Nuclear), and radioactivity was measured in a liquid scintillation counter (Beckman LS 9000). Peptide Mapping 0.25 mg of carrier pig IgG was added to each gel-purified p,, pm or ps sample intended for peptide mapping. Chains were completely reduced and alkylated as described above, TCA-precipitated and airdried. Samples were dissolved in 200 $0.2 M ammonium bicarbonate and digested with TPCK-trypsin (Worthington) at room temperature (total of 150 kg for 22 hr). followed by digestion with chymotrypsin (Worthington) at 37’C (total of 200 pg for 28 hr). The peptides were frozen and lyophilized to terminate the reaction, dissolved in 45 11 0.5 M phosphate buffer (pH 1.8):acetone. 2:1, and the peptides were separated by high performance liquid chromatography on a DuPont ODS C-l 8 column as described (McMillan et al., 1979). Fractions were collected at 0.5 min intervals (0.5 ml), dried in a vortex evaporator, redissolved in 0.25 ml 0.01% SDS and assayed for radioactivity in a liauid scintillation counter. Sequence Analysis Cyanogen bromide fragments from the ACA54 column were prepared for sequence determinations by pooling, dialyzing exhaustively against 5% formic acid in low molecular weight cutoff dialysis tubing (Spectra/Par 3) at 4°C and lyophilizing. Pools contained 1-3 mg carrier Ml 04E cyanogen bromide fragments. Automated sequence analyses on the milligram quantities of M x W ps chains and CN3 fragments were performed on a modified Beckman sequenator (Wittmann-Liebold, 1973: Wittmann-Liebold, Graffunder and Kohls, 1976; Hunkapiller and Hood, 1978). Samples were loaded onto carrier Polybrene (Aldrich), and phenylthiohydantoin (Pth) derivatives were identified by high performance liquid chromatography (HPLC) (Water Associates) as described (Hunkapiller and Hood, 1978; Johnson, Hunkapiller and Hood, 1979). Automated sequence analyses of radiolabeled cyanogen bromide fragments were performed on the Caltech sequenator (Hunkapiller and Hood, 1980). Samples were loaded in trifluoroacetic acid (Pierce) and 20% H20 with Polybrene (Aldrich) as carrier. The Pth derivatives were separated by HPLC (DuPont), and the peaks corresponding to the radiolabeled Pth-amino acids were fraction-collected, dried and counted for radioactivity in Liquifluor-toluene (McMillan et al., 1977). Carboxypeptidase A and B Digestions Carboxypeptidase A (Worthington) and diisopropylfluorophosphatetreated carboxypeptidase B (Sigma) were used at an enzyme-tosubstrate ratio of 1:40. Carboxypeptidase A was prepared as described (Ambler, 1972), and carboxypeptidase B in 0.1 M NaCl was thawed and diluted 1 :I 0 in 0.2 M N-ethylmorpholine acetate (pH 8.6) just prior to use. A known number of counts of isolated radiolabeled p chain were reduced and alkylated as for peptide maps and dissolved in 0.2 M N-ethylmorpholine acetate (pH 8.6) at a carrier pig IgG concentration of 2.5 mg/ml. A portion was removed for a substrate control, and the remainder was digested with an equimolar mixture of carboxypeptidases A and B at 37’C. Aliquots were removed at indicated times, frozen immediately on dry ice and lyophilized twice. For substrate controls, equivalent amounts of enzymes were boiled in 0.1 N acetic acid for 15 min. added to the substrate and incubated at 37OC for the same length of time as the longest reaction time point. Released amino acids were separated on a Durrum D-500 amino acid analyzer (92% of sample loaded). The ninhydrin coil temperature was turned to the minimum setting and 0.5 min fractions were collected from the time of injection and assayed for radioactivity in a liquid scintillation counter. Radiolabeled amino acid standards were analyzed by the same procedure. The amount of each amino acid released was quantitated (substrate controls contained no peaks) by

pm and ps Chains 405

comparison of the total radioactivity in each peak with the starting radioactivity and the number of residues of each amino acid in the W279 pLschain (Kehry et al., 1979). Antiserum Rabbit anti-mouse Ml 04E fi chain was produced by repeated immunization of an NZW rabbit with 300 pg partially reduced and alkylated Ml 04E p chain in complete Freunds adjuvant. Blood was clotted and centrifuged at 6000 x g. and the serum was divided into aliquots and stored frozen at -7O’C. Anti p serum was used without any further treatment.

We thank Dr. M. Hunkapiller for assistance and advice in using the Caltech sequenator, Dr. M. McMillan for teaching us peptide mapping and the analysis of radiolabeled Pth-amino acids, V. Farnsworth for setting up the Durrum D-500 fraction collection system for the carboxypeptidase experiments and Dr. Noel Warner for providing the WEHI 279 cell line. This work was supported by NIH grants to L. H. and C. S., and a National Cancer Institute grant to W. R. M. K. is the recipient of a Gordon Ross Medical Foundation Fellowship; S. E. was supported by an NIH fellowship: R. D. is a fellow of the Arthritis Foundation. C. S. is the recipient of a Faculty Research Award from the American Cancer Society. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisemenf” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. May 23, 1980

characteristic

IgM and IgD

Gahmberg. C. G. and Hakomori, S. (1973). External labeling of cellsurface galactose and galactosamine in glycolipid and glycoprotein of human erythrocytes. J. Biol. Chem. 248, 431 l-431 7. Garrels, J. I. (1979). Two-dimensional puter analysis of proteins synthesized Chem. 254, 7961-7977.

gel electrophoresis and comby clonal cell lines. J. Biol.

Gill, D. M. (1976). The Biochemistry 75, 1242-l

of subunits

arrangement 248.

Gross, E. (1967). The cyanogen zymology 7 7, 238-255.

Acknowledgments

Received

B lymphocytes: inability to detect certain antigens. J. Exp. Med. 740, 895-903.

bromide

reaction.

in cholera

toxin.

Methods

in En-

Honjo. T., Obata, M., Yamawaki-Kataoka, Y., Kataoka, T.. Kawakami, T., Takahashi, N. and Mano, Y. (1979). Cloning and complete nucleotide sequence of mouse immunoglobulin yl chain gene. Cell 78, 559-568. Hunkapiller, M. W. and Hood, L. E. (1978). Direct microsequence analysis of polypeptides using an improved sequenator, a nonprotein carrier (polybrene), and high pressure liquid chromatography. Biochemistry 7 7, 2124-2133. Hunkapiller, M. W. and Hood, L. E. (1980). New protein with increased sensitivity. Science 207, 523-525.

sequenator

Johnson, N., Hunkapiller. M. and Hood, L. (1979). Analysis of phenylthiohydantoin amino acids by high-performance liquid chromatography on DuPont Zorbax cyanopropylsilane columns. Anal. Biochem. 700, 335-338. Kabat, E. A., Wu, T. T. and Bilofsky, lmmunoglobulin Chains (Cambridge:

H. (1976). Variable Regions Bolt Beranek and Newman).

of

Kehry, M. R. (1980). Structure and function of murine immunoglobulin M from serum and cell membrane. Ph.D. thesis, California Institute of Technology, Pasadena, California.

References Ambler, R. P. (1972). Enzymatic hydrolysis Methods in Enzymology 25, 143-154.

with carboxypeptidases.

Andres, R. Y., Jeng, I. and Bradshaw, R. A. (1977). Nerve growth factor receptors: identification of distinct classes in plasma membranes and nuclei of embryonic dorsal root neurons. Proc. Nat. Acad. Sci. USA 74, 2785-2789. Bergman, Y. and Haimovich, J. W. (1978). B lymphocytes three species of p chains. Eur. J. Immunol. 8, 876-880. Bergman, Y.. Haimovich. J. and Melchers, ing tumor with biochemical characteristics Eur. J. Immunol. 7, 574-579.

F. (1977). An IgM-producof a small B lymphocyte.

Banner, W. M. and Laskey, R. A. (1974). A film detection tritium-labeled proteins and nucleic acids in polyacrylamide J. Biochem. 45, 83-88. Burrows, P., LeJeune, M. and Kearney, murine pre-B cells synthesize p heavy Nature 280, 838-840. Carpenter, G. and Cohen, S. (1979). Rev. Biochem. 48, 193-216.

contain

method for gels. Eur.

J. F. (1979). Evidence that chains but no light chains. Epidermal

growth

factor.

Ann.

Kehry, M., Sibley, C.. Fuhrman, J., Schilling, J. and Hood, L. E. (1979). Amino acid sequence of a mouse immunoglobulin p chain. Proc. Nat. Acad. Sci. USA 76, 2932-2936. Kessler, S. W. (1976). Staphylococcae protein 1482-l 490.

Cell membrane antigen A-antibody adsorbant.

Klapper, M. H. and Klotz, I. M. (1972). acid anhydrides. Methods in Enzymology Kornfeld, R. and Kornfeld, S. (1976). protein structure. Ann. Rev. Biochem.

isolation with the J. Immunol. 7 17,

Acylation with dicarboxylic 25, 531-536.

Comparative aspects 45, 217-237.

of glyco-

Laemmli. U. (1970). Cleavage of structural proteinsduring the assembly of the head of bacteriophage T4. Nature 277, 680-685. Mcllhinney, R. A. J., Richardson, N. E. and Feinstein, A. (1978). Evidence for a C-terminal tyrosine residue in human and mouse Blymphocyte membrane p chains. Nature 272, 555-557. McMillan, M., Cecka, J. M., Murphy, D. B., McDevitt, H. 0. and Hood, L. (1977). Structure of murine la antigens: partial NHz-terminal amino acid sequences of products of the I-E or I-C subregion. Proc. Nat. Acad. Sci. USA 74, 5135-5139.

Cuatrecassas. P., Hollenberg. M. D., Chang, K.-J. and Bennett, V. (1975). Hormone receptor complexes and their modulation of membrane function. Recent Prog. Hormone Res. 37, 37-94.

McMillan, M., Cecka. J. M., Hood, L., Murphy, 0. (1979). Peptide map analyses of murine subregion using HPLC. Nature 277, 663-665.

Dayhoff, M. 0. (1976). (Silver Spring, Maryland: pp. 189-l 90.

Melcher, U. and Uhr, J. W. (1973). An electrophoretic difference between surface and secreted IgM of murine splenocytes. J. Exp. Med. 738, 1282-l 287.

Atlas of Protein Sequence and Structure, 5 National Biomedical Research Foundation),

Della Carte. E. and Parkhouse, R. M. E. (1973). Biosynthesis of immunoglobulin A (IgA) and immunoglobulin M (IgM); requirement for J chain and a disulphide exchanging enzyme for polymerization. Biochem. J. 736, 597-606. Early, P., Rogers, J.. Davis, M., Calame, K., Bond, M., Wall, R. and Hood, L. (1980). Two mRNAs can be produced from a single immunoglobulin p gene by alternative RNA processing pathways. Cell 20, 313-319. Fu, S. M. and Kunkel, H. G. (1974). Membrane immunoglobulins of

D. B. and McDevitt, H. la antigens of the I-E

Melcher. U. and Uhr, J. W. (1977). Density differences between membrane and secreted immunoglobulins of murine splenocytes. Biochemistry 76, 145-152. Melcher, U., Eidels, L. and Uhr, J. W. (1975). Are immunoglobulins integral membrane proteins? Nature 258, 434-435. Mestecky, J. and Schrohenloher, J chain to human immunoglobulin

R. E. (1974). Site of attachment M. Nature 249, 650-652.

O’Farrell. P. H. (1975). High resolution two-dimensional resis of proteins. J. Biol. Chem. 250, 4007-4021.

electropho-

Of

Cell 406

Parkhouse, R. M. E., Lifter, J. and Choi, Y. S. (1980). characterisation of the Fab and Fc fragments from surface globulin. Nature 284, 280-281.

Chemical immuno-

Raschke, W. C., Mather, E. L. and Koshland. M. E. (1979). Assembly and secretion of pentameric IgM in a fusion between a nonsecreting B cell lymphoma and an IgG-secreting plasmacytoma. Proc. Nat. Acad. Sci. USA 76, 3469-3473. Rogers, J., Early, P., Carter, C., Calame, K., Bond, M.. Hood, L. and Wall, R. (1980). Two mRNAs with different 3’ ends encode membrane-bound and secreted forms of immunoglobulin p chain. Cell 20, 303-312. Segrest, J. P. and Feldmann, R. J. (1974). Membrane proteins: amino acid sequence and membrane penetration. J. Mol. Biol. 87, 853858 Sibley, C. H., Ewald, S. J., Kehry, M. R.. Douglas, R. H., Raschke. W. C. and Hood, L. E. (1980). Characterization of multiple immunoglobulin mu chains synthesized by two clones of a B-cell lymphoma. J. Immunol., in press. Singer, P. A., Singer, H. H. and Williamson, A. R. (1980). Different species of messenger RNA encode receptor and secretory IgM p chains differing at their carboxy termini. Nature 285, 294-300. Tomita. D. B. and Marchesi. V. T. (1975). Amino-acid oligosaccharide attachment sites of human erythrocyte Proc. Nat. Acad. Sci. USA 72, 2985-2968.

sequence and glycophorin.

Tucker, P. W., Marcu, K. B., Slightom, J. L. and Blattner, F. R. (1979). Structure of the constant and 3’ untranslated regions of the murine Y2b heavy chain messenger RNA. Science 206, 1299-i 303. Vassalli. P., Tedghi, R, Lisowska-Bernstein. B., Tartakoff, A. and Jaton. J.-C. (1979). Evidence for hydrophobic region within heavy chains of mouse B lymphocyte membrane-bound IgM. Proc. Nat. Acad. Sci. USA 76, 5515-5519. Vitetta, E. S., Baur, S. and Uhr, J. W. (1971). Cell surface immunoglobulin. II. Isolation and characterization of immunoglobulin from mouse splenic lymphocytes. J. Exp. Med. 734, 242-264. Warner, N. (1974). Membrane immunoglobulins and antigen tors on B and T lymphocytes. Adv. Immunol. 79, 67-216.

recep-

Wickner, W. (1976). Asymmetric orientation of phage Ml3 coat protein in Escherichia co/i cytoplasmic membranes and in synthetic lipid vesicles. Proc. Nat. Acad. Sci. USA 73, 1159-l 163. Williams, P. B.. Kubo, R. T. and Grey, H. M. (1978). p-chains from a non-secretor B cell line differ from secreted p-chains at the C-terminal end. J. Immunol. 721, 435-2439. Wittmann-Liebold, B. (1973). Amino acid sequence studies on ten ribosomal proteins of Escherichia co/i with an improved sequenator equipped with an automatic conversion device. Hoppe-Seyler’s 2. Physiol. Chem. 354, 1415-l 431. Wittmann-Liebold, B., Graffunder, H. and Kohls, H. (1976). A device coupled to a modified sequenator for the automated conversion of anilinothiazolinones into PTH amino acids. Anal. Biochem. 75, 621633. Yamawaki-Kataoka. Y., Kataoka, T., Takahashi. N., Obata. M. and Homo, T. (1980). Complete nucleotide sequence of immunoglobulin Y2b chain gene cloned from newborn mouse DNA. Nature 283, 786789. Yuan, D., Uhr, J. W. and Vitetta. E. S. (1980). A peptide between the p chains from cell-associated and secreted BCL, tumor. J. Immunol. 725, 40-46. Note Added

difference IgM of the

in Proof

Isolation and characterizarion of pm and p* mRNAs by F. W. Alt et al. (Cell 20, 293-300, 1980) has confirmed our findings. Labeling of B lymphocyte membranes with a reactive lipophilic reagent has shown that pm chains but not rrs chains contain a hydrophobic section that is buried in the lipid bilayer (M. J. Owen, J. C. A. Knott and M. J. Crumpton. Biochemistry 19, 3092-3099. 1980).