ligand binding by murine igm antibodies: intramolecular heterogeneity exists in certain, but not...

8
MoleculurImmunology, Vol. 25, No. 1, pp. 87-94, 1988 Printedin Great Britain. 0163-5890188 $3.00 + 0.00 PergamonJournalsLtd LIGAND BINDING BY MURINE IgM ANTIBODIES: INTRAMOLECULAR HETEROGENEITY EXISTS IN CERTAIN, BUT NOT ALL, CASES* DAVID PASCUAL and L. WILLIAM CLEM Department of Microbiology, University of Mississippi Medical Center, Jackson, MS 3921&4505. U.S.A. (First received 10 February 1987; accepted in revisedform 6 August 1987) Abstract-The ligand binding properties of eight hybridoma-derived murine anti-DNP IgM(K) antibodies were analysed by equilibrium dialysis. Four of these proteins exhibited the expected valances of - IO and relatively low affinities ($2.2 x IO5 M-‘). The remaining four proteins exhibited valences of considerably less than 10 (G8) and relatively high affinities (> 106Mm’). When these proteins were subjected to two cycles of lyophilization, those of the former group were observed to still exhibit - 10 sites per molecule with homogeneous affinities similar to those of the respective untreated molecules, However, molecules in the latter group (valences of < 8) were observed to exhibit only five to six binding sites subsequent to lyophilization with no changes in affinities. When the reductive subunits from each of the IgM(K)_proteins were subjected to trypsinization, two different patterns were observed in terms of the yields of Fabp fragments. Each of the proteins originally exhibiting - 10 binding sites yielded >90% of the expected Fabp fragments. In contrast each of the proteins exhibiting $8 binding sites yielded only - 50% of the expected Fabp fragments. Collectively these results indicate the existence of at least two different forms of murine IgM molecules, those with - 10 homogeneous, relatively stable sites and those with only approx. five stable sites. It is suggested that these intramolecular functional differences may be attributable to intramolecular conformational differences. INTRODUCTION Mammalian IgM antibodies are pentameric mol- ecules that should contain 10 functionally equivalent ligand-binding sites (reviewed in Metzger, 1970). Although the results of earlier studies had indicated that certain monoclonal IgM proteins exhibited 10 homogeneous sites, quite different results were fre- quently encountered with IgM antibodies isolated from immune sera. In fact an average of five high affinity and five low affinity sites was often the rule with “immune” IgM antibody populations obtained from a wide variety of species, including sharks (see Shankey and Clem, 1980a, b). Since the finding of both high and low affinity sites in “immune” IgM antibody populations could result from either inter- or intramolecular differences, our laboratory has attempted to resolve this issue by employing murine hybridoma-derived IgM molecules specific for the DNP moiety. Our initial efforts in this direction utilized the P3 myeloma line, which secretes the MOPC-21 protein, and resulted in the development of a hybridoma line secreting IgM antibodies with an average of only six homogeneous sites. The failure of this protein to exhibit a valence of 10 was attributable to intramolecular L chain heterogeneity, i.e. the *This work was supported by NIH grants lROI-AI-18342 and lROl-AI-19530. The sequencing facility at the University of Mississippi Medical Center was supported in part by NIH grants RR02745 and RR05386. presence of both MOPC-21 and anti-DNP L chains. It was noted, however, that the reductive subunits were homogeneous in terms of L chains and those (~60%) which contained active sites exhibited a valence of 2 (Giles et al., 1982). Since this type of intramolecular heterogeneity of primary structures and ligand binding was deemed an artifact of the hybridoma technology employed, additional fusions were done using non-Ig producing myeloma cells. The initial efforts with this approach resulted in the development of two hybridoma cell lines secreting IgM antibodies reactive with the DNP moiety. Al- though each of these proteins appeared to be struc- turally homogeneous, the results of binding studies revealed the somewhat unexpected finding of an average of only approx. five homogeneous binding sites (instead of 10) arranged such that each reductive p’z L, subunit exhibited a valence of about 1. Further- more, it was observed that the Fabp fragments containing the active binding sites from these mol- ecules were stable to trypsinization whereas the seem- ingly nonfunctional fragments were not (Giles et al., 1983). Consequently, the study reported here was undertaken in order to establish the generality of this apparent intramolecular heterogeneity for a greater number of IgM proteins, i.e. is it more apparent than real? The results obtained clearly indicate the exist- ence of at least two different forms of murine IgM antibodies to DNP, i.e. molecules with 10 homo- geneous, relatively stable sites and molecules with significantly fewer such sites. 87

Upload: david-pascual

Post on 19-Nov-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

Moleculur Immunology, Vol. 25, No. 1, pp. 87-94, 1988 Printed in Great Britain.

0163-5890188 $3.00 + 0.00 Pergamon Journals Ltd

LIGAND BINDING BY MURINE IgM ANTIBODIES: INTRAMOLECULAR HETEROGENEITY EXISTS IN

CERTAIN, BUT NOT ALL, CASES*

DAVID PASCUAL and L. WILLIAM CLEM

Department of Microbiology, University of Mississippi Medical Center, Jackson, MS 3921&4505. U.S.A.

(First received 10 February 1987; accepted in revisedform 6 August 1987)

Abstract-The ligand binding properties of eight hybridoma-derived murine anti-DNP IgM(K) antibodies were analysed by equilibrium dialysis. Four of these proteins exhibited the expected valances of - IO and relatively low affinities ($2.2 x IO5 M-‘). The remaining four proteins exhibited valences of considerably less than 10 (G8) and relatively high affinities (> 106Mm’). When these proteins were subjected to two cycles of lyophilization, those of the former group were observed to still exhibit - 10 sites per molecule with homogeneous affinities similar to those of the respective untreated molecules, However, molecules in the latter group (valences of < 8) were observed to exhibit only five to six binding sites subsequent to lyophilization with no changes in affinities. When the reductive subunits from each of the IgM(K)_proteins were subjected to trypsinization, two different patterns were observed in terms of the yields of Fabp fragments. Each of the proteins originally exhibiting - 10 binding sites yielded >90% of the expected Fabp fragments. In contrast each of the proteins exhibiting $8 binding sites yielded only - 50% of the expected Fabp fragments. Collectively these results indicate the existence of at least two different forms of murine IgM molecules, those with - 10 homogeneous, relatively stable sites and those with only approx. five stable sites. It is suggested that these intramolecular functional differences may be attributable to intramolecular conformational differences.

INTRODUCTION

Mammalian IgM antibodies are pentameric mol- ecules that should contain 10 functionally equivalent ligand-binding sites (reviewed in Metzger, 1970). Although the results of earlier studies had indicated that certain monoclonal IgM proteins exhibited 10 homogeneous sites, quite different results were fre- quently encountered with IgM antibodies isolated from immune sera. In fact an average of five high affinity and five low affinity sites was often the rule with “immune” IgM antibody populations obtained from a wide variety of species, including sharks (see Shankey and Clem, 1980a, b). Since the finding of both high and low affinity sites in “immune” IgM antibody populations could result from either inter- or intramolecular differences, our laboratory has attempted to resolve this issue by employing murine hybridoma-derived IgM molecules specific for the DNP moiety. Our initial efforts in this direction utilized the P3 myeloma line, which secretes the MOPC-21 protein, and resulted in the development of a hybridoma line secreting IgM antibodies with an average of only six homogeneous sites. The failure of this protein to exhibit a valence of 10 was attributable to intramolecular L chain heterogeneity, i.e. the

*This work was supported by NIH grants lROI-AI-18342 and lROl-AI-19530. The sequencing facility at the University of Mississippi Medical Center was supported in part by NIH grants RR02745 and RR05386.

presence of both MOPC-21 and anti-DNP L chains. It was noted, however, that the reductive subunits were homogeneous in terms of L chains and those (~60%) which contained active sites exhibited a valence of 2 (Giles et al., 1982). Since this type of intramolecular heterogeneity of primary structures and ligand binding was deemed an artifact of the hybridoma technology employed, additional fusions were done using non-Ig producing myeloma cells. The initial efforts with this approach resulted in the development of two hybridoma cell lines secreting IgM antibodies reactive with the DNP moiety. Al- though each of these proteins appeared to be struc- turally homogeneous, the results of binding studies revealed the somewhat unexpected finding of an average of only approx. five homogeneous binding sites (instead of 10) arranged such that each reductive p’z L, subunit exhibited a valence of about 1. Further- more, it was observed that the Fabp fragments containing the active binding sites from these mol- ecules were stable to trypsinization whereas the seem- ingly nonfunctional fragments were not (Giles et al., 1983). Consequently, the study reported here was undertaken in order to establish the generality of this apparent intramolecular heterogeneity for a greater number of IgM proteins, i.e. is it more apparent than real? The results obtained clearly indicate the exist-

ence of at least two different forms of murine IgM antibodies to DNP, i.e. molecules with 10 homo-

geneous, relatively stable sites and molecules with significantly fewer such sites.

87

Page 2: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

88 DAVID PASCUAL and L. WILLIAM CLEM

MATERIALS AND METHODS

The techniques employed for ~rfo~jng PEG- induced fusion (with the non-fg producing P3-X63-Ag 8.653 myeloma line), cloning in soft agar and propagating hybridoma cells in pristane-primed mice were as described in detail previously (Giles et ui., 1982, 1983). The seven new IgM(rc) secreting hybridomas studied here were developed from a single fusion using pooled spleen cells from three BALB/c mice primed 3 days previously with DNP- Ficoll. This fusion also resulted in the development of five additional cell lines secreting lgM(E,) anti-DNP proteins: these latter proteins have not yet been studied in detail.

Purification and characterization of IgM anti- bodies were accomplished as described in detail pre- viously (Giles et al., 1982). Equilibrium dialysis was also done as described previously (Giles et al., 1983) with the exception that AH-~-DNP-lysine (New

England Nuclear), instead of “H-DNP-c -

aminocaproate, was employed as the hapten. The preparation of reductive pL2-L2 chain subunits (actu- ally two covalent p-1, chain halfmers noncovalently bound together) of the IgM(K) antibodies was also performed as described previously; proteins BG-12, 1 B3, 1 FE, 1 F4 and 2E7 gave maximal yields of these subunits with 5 x lo-’ M 2-ME whereas proteins 1 ES and lD.5 only required 3 x lo-’ M 2-ME and protein 2F11 required 7.S x 10 ’ M 2-ME for similar yields.

Trypsinization of the 2~.-2L chain reductive sub- units from the various .fgM(K) molecules for various periods of time was performed as previously de- scribed (Klapper et al., 1971). Quantification of Fabp fragment yields on Coomassie blue-stained SDS- PAGE gels was accomplished by using a Zeineh soft laser scanning densitometer (LKB Instruments).

Lyophilization of affinity purified igM(rc) mole- cules was accomplished as follows: the proteins were dissolved at l-2 mg/ml in 0.01 M Tris-HCl, pH 7.4, and quickly frozen in polysulfonyl centrifuge tubes by swirling in liquid N,. Each tube was covered with punctured parafilm and freeze-dried on a Unitrap II (Virtis) for 18-24 hr. The lyophilized proteins were dissolved to the original vol in sterile deionized water and a second cycle of lyophilization was done as before. Such dissolved, twice lyophilized proteins were then studied immediately or after storage at .-40°C (for up to 1 month) with similar results.

Sequencing of isolated extensively reduced and alkylated p and L chains was done on an Applied Biosystems 470 sequencer using the 03RPTH pro- gram. PTH-derivatives were identified with an Ap- plied Biosystems dedicated PTH column (0.2 mm) on a Waters HPLC directly interfaced with the se- quencer. Titration of free sulfhydryl groups was done by mixing the proteins (at- lOmg/ml in 0.2 M

Tris-HCI, pH 8.6) with a 500-fold molar excess of “C-iodoacetamide (New England Nuclear) for 1 hr at 4°C followed by gel filtration of Sephadex G- 10 and ~intillation counting to ascertain the extent of pro- tein alkylation.

RESULTS

Each of the seven IgM(K) proteins reactive with DNP moiety were subjected to gel filtration (on a calibrated Agarose A5M column) and found to elute as symmetrical peaks in vols expected to contain -900,000 mol. wt material (not shown). In addition each unreduced protein migrated similarly (mol. wt -900,000) when subjected to SDS-PAGE as depicted in Fig. l(A): also depicted is the previously studied IgM(?c) anti-DNP protein BB-12 (SPZjO I-64 Cl-12; Giles et al., 1983). Upon extensive reduction each protein separated into heavy (p) and light chains upon SDS-PAGE as shown in Fig. l(B). Each of the I( chains from the different proteins migrated with negligible variability in mobility (mol. wt -70,000) on SDS-PAGE and represented a relatively constant amount (66-70%) of the total protein as determined by integration of soft laser densitometer scans of the

stained gels. It was noted that a minor band (mol. wt -53,000) below certain of the p chains on the SDS-PAGE gels corresponded to negligible amounts of protein, i.e. less than 1% of the total. Although the

Fig. 1. SDS-PAGE patterns of unreduced (A) and reduced (B) hybridoma-derived monoclonal IgM antibodies specific for the DNP moiety. The numbered lanes contained the following proteins: 1 = lD5; 2 = lF3; 3 = lF4; 4 = 2Fl I;

5= IB3; 6=1E5; 7=2E7: X=BG-12.

Page 3: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

Ligand binding by IgM antibodies from murine hybridomas 89

300

250 9 ‘a 200

5 150

100

50

0

A

~ 0 2 4 6 e 10

r

“v

$ n

f

Fig. 2. Scatchard plots of equilibrium dialysis data of ‘H-t-DNP-lysine binding to 1gM proteins lB3 (A) and 2FI 1 (B) before (0-O) and after (C---C) two cycles of lyophili~tion. The affinities and valences calculated from

these data are given in Table 1.

light chain SDS-PAGE migration patterns showed considerable variability in mobility (from mol. wts of _ 22,000 for protein lE5 to m 26,000 for protein lF3), soft laser densitometer scans indicated that the light chains accounted for 2630% of the total pro- tein in each case.

Each of the hybridoma-delved IgM monoclonal antibodies was studied for ligand binding by equi- librium dialysis with 3H-c-DNP-lysine. The results, depicted as Scatchard plots, for two of these proteins (IB3 and 2Fll) are presented in Fig. 2. On one hand it can be seen that protein 2Fll exhibited homo- geneous binding (K, = 4 x lo4 M-l, a = 1.0) for an average of 10 sites per molecule. On the other hand, protein 183 exhibited homogeneous binding (K, = 1.6 x lo6 M-‘, a = 0.98) for an average of only eight sites per molecule. The equilibrium dialysis results obtained with each of the light proteins (in- cluding the previously studied BG-12) are sum- marized in Table 1; the Sips heterogeneity indices (a) for each of these proteins was >0.96. It can be seen that, based upon binding criteria, they appear to fall into two distinct groups. One group. wherein the K, is c2.2 x 105M--’ and the valence is a9.4, is com- prised of proteins 2F11, 1D5, lF3 and 1F4. Another group, whereing the K, is > lo6 M-’ and the valence is < 8.2, is comprised of proteins lB3, lE5, 2E7, and

Table 1. At’Sties and valences of ‘H-c-DNP-lysine binding by monoclonal IgM antibodies

Cycles of lyophilization~

IgM(x ) protein __~. ._ 2Fll ID5 lF3 lF4

NolIe

Affinity Valence KM-‘) @I

4.0 x 104 IO.0 2.2 x 10s 10.0 1.1 x 105 9.6 1.4 x 105 9.4

TWO

Affinity Valence (&*M -‘1 (n,

‘Lx 104 t0.0 2.0 x 103 9.8 2.2 x 105 9.2 LOX 105 9.4

183 1.6 x lo6 8.0 1.4 x 106 6.0 lE.5 1.9 x 106 8.0 2.2 x I06 6.0 2E7 I.6 x 106 8.2 2.4 x 10’ 6.0 BG-I? 1.2 x 10’ 7.5 1.2 x 10’ 5.5

These values were calculated from equilibnmn dialysis data ob- tained with the various proteins both before and after two cycles of lyophilization.

BG-12. For the purposes of indentification these groups will subsequently be referred to as the “low affinity-high valance” and “high affinity--low valence” groups, respectively.

During the course of the above binding studies it was discovered that lyophilization of certain. but not all, of the purified IgM proteins resulted in losses of binding sites. Consequently each of the eight purified proteins being studied was subjected to one or two cycles of lyophilization and analysed by equilibrium dialysis. The Scatchard plots of two of these proteins (lB3 and 2Fll) after two cycles of lyophilization are depicted in Fig. 2 and the affinities and valences for each of the eight proteins are summarized in Table I. An important point of these results is that, although no significant changes in affinities for any of the proteins resulted from lyophilization, substantial losses in numbers of binding sites were apparent with some, but not all, of the IgM proteins. in particular it can be seen that the four proteins comprising the “high affinity-low valance” (183, lE5, 2E7 and BG12) group each appeared to lose -25% of their original sites resulting in stable valences of 5.5~~60 after two cycles of lyophilization; these proteins lost _ 15% of their original sites after one cycle of lyophilization (not shown). However, such losses of binding sites as a consequence of lyophilization were not apparent in the “low affinity--high valence” group of IgM proteins. In attempting to understand the nature of the lost sites, twice lyophilized proteins of both groups were subjected to SDS-PAGE: the rc- suits (not shown) indicated that the losses could not be explained by any changes in covalent structure as a consequence of lyophili~ation. Similarly another potential artifact, i.e. underestimation of valences due to the possibility of certain lyophilized IgM molecules adhering to the walls of the equilibrium dialysis cells. was ruled out or minimized by the finding that >95% of the various proteins were reproducibly recovered after dialysis. Thus, since an additional (third) cycle of lyophilization did not signnicantly further change the valences of any of these proteins (data not shown), it seems reasonable to consider that, whereas some murine IgM antibody molecules have only stable sites, others actually can have two somewhat different types of binding sites. i.e. stable vs labile to lyophilization. It should be noted that >90% of the gZLZ chain reductive subunits derived from all twice lyophilized proteins (irrespective of whether or not they exhibited an average of either five to six or y 10 stable sites) consistently bound to DNP affinity columns. Of additional importance in this respect was the finding that gel filtered (in 0.14 ;EJF NaC1, 0.01 h4 Tris-HCl, pH 7.4. on Agdrose A5M) pZLZ chain reductive subunits from both groups of proteins did not appear to aggregate as a con- sequence of lyophilization. Hence it seems reasonable to conclude that nearly all reductive subunits of protein from the “high affinity--low valence” group contained at least one stable site.

Page 4: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

90 DAVID PASCUAL and L. WILLIAM CLEM

A B

01 2 4 6120 12 4 612

H-L

H

Fab

Fig. 3. SDS-PAGE patterns of reductive subunits of monoclonal IgM proteins 2Fll (A) and 1 B3 (IS) subjected to trypsinization for various times. The numbers refer to digestion times. in hr.

The results of previous studies indicated the tryp- sinization of reductive subunits from protein BG-12

resulted in the recovery of only - 50% of the ex- pected Fat+ fragments. Furthermore, it was observed that each of the recovered Fabp fragments contained a binding site, suggesting that the inactive (non- binding) regions of the molecules uere differentially susceptible to trypsinoiysis (Giles et ul., 1983). Thus. trypsinization of reductive subunits from each of the proteins under consideration here was employed to ascertain the generality of this phenomenon for other IgM anti-DNP proteins. Of particular importance in this respect was the issue of whether or not “bona fide” 10 site proteins would actually yield 100% of the expected Fabp fragments. The results of

representative SDS-PAGE patterns for tryptic digests of proteins 1B3 and 2Fll are depicted in Fig. 3. In both cases it can be seen that the reductive subunits (actually -90,OOOmol. wt, [L-L chain haifmers in the presence of SDS) were almost entirely digested to Fabp fragments (defined as covalent 50,000-56,000 mol. wt material) within 4 hr. Further- more, and most importantly, when the stained gels were densitometrically analysed with a soft laser scanner a significant disparity in yields of Fabp fragment was apparent. Specifically, digests of pro- tein I B3 contained only - 54% of the expected Fabfi fragments whereas digests of protein 2F11 contained -98% of the expected yield, even after 12 hr of digestion. The results obtained with the tryp- sinization protocol for each of the proteins studied are presented in Table 2 and can be summarized as follows. Proteins of the originally discussed ‘*low

affinity-high valence” group (2F1 I, 1 D5, lF3 and lF4) reprodu~b~y yielded > 90% of the expected Fabfr fragments whereas those of the “high affinity-low valence” group (1 B, 1 ES. 2E7 and BG- 12) yielded < 60% of the expected fragments. Also included in Table 2 are the results for one protein from each of the aforementioned afinity-valence groups which indicate that two cycles of lyophiliz- ation prior to the preparation of reductive subunits did not alter the yields of FabFl fragments obtained by trypsinization.

In terms of firmly establishing the nature of the putative Fabp fragments obtained by trypsinization,

Table 2. Yields of Fabr fragments from the reductive subunits of IgM monoclonals subjected to trypsmiralion

WI Molecular wt F&I/l fragment” protein of Fabk fragments yields (%)

2F11 53,500 9x 2F1 Ih 53,500 94

ID.5 55,oQo 100 IF3 56.000 41

IF4 54.000 IO0

iB3 55.000 54 I B3h 55.000 55 IES 50,000 53 2E7 52,000 59 BG-12 52.000 55

“Vaiues aerrr obtained by scanning SDS-PAGE patterns ot ti-ypsinized IgM monoclonal subunits with a soft laser densitometer and comparing the staining intensities with those of unchgested material, takmg into account the relatwe masses of the original subunits tend the resultant Fabp fragment.

‘Subjected to two cycles of lyophilization before tryp slniratlon

Page 5: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

Ligand binding by IgM antibodies from murine hybridomas 91

A B

60.0 6.0

: * 0 40.0 4 k n f m ::

z 20.0

0.0 6.0 N 4.0

2.0

0.0 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0

r r

Fig. 4. Scatchard plots of equilibrium dialysis data of ‘H-t-DNP-lysine binding to reductive subunits (O--O) and Fabp fragments (G---O) from monoclonal IgM proteins 1B3 (A) and 2Fll (B). The affinities calculated from these data were: protein lB3 subunits = 1.8 x lo6 M-‘; Fubp fragments = 1.1 x IO6 M-‘; protein 2Fll subunits =

4.5 x IO4 M-l; Fabp fragments = 3.9 x IO4 M-‘.

as described above, several approaches were em- ployed. First, gel filtration (Sephadex G-200) of 46 hr tryptic digests of reductive subunits consis- tently yielded a 50,00&60,000 mol. wt component which, when further purified by hapten-elution from DNP affinity columns, exhibited the expected valence of - 1 and an affinity similar to that of the intact parent molecule. Two examples of these results are depicted as Scatchard plots in Fig. 4; the putative Fabp fragments of proteins lB3 and 2Fll each exhibited valences of -0.9-l .O and affinities of 1.1 x

IO6 Mm’ and 3.9 x lo4 Mm’, respectively. Second, when these affinity purified putative Fabp fragments were extensively reduced and subjected to SDS- PAGE, they were observed to dissociate into two components; one exhibited a mobility identical to the L chain of the parent molecule whereas the other appeared somewhat larger (mol. wt -30,000) and likely represented the Fd region (not shown). Immunoprecipitation (Ouchterlony) analysis indi- cated reactions of identity between the Fabn frag- ment and the intact IgM molecules when antiserum to mouse kappa chains was used. The anti-mouse p chain antiserum employed did not react visibly with the Fabp fragments (not shown).

a characteristic finding with this method of analysis, and (b) the yield ratios of the highest and next highest residues at position 2 were in the range of 10-20. When compared with a compilation of murine Ig variable region sequences (Kabat et al., 1987) the p chains from these two proteins each appeared to belong to heavy chain subgroup I(A). Similar com- parisons of the L chains of lB3 indicated that they belonged to kappa group II; the L chains of 2E7 appeared to belong to no defined group and thus should likely be assigned to the miscellaneous group. The p and L chains of proteins BG-12 (as previously reported, Giles et al., 1983) and lE5 were not amena- ble to sequence analysis, presumably due to blocked amino termini. An additional approach used in at- tempting to define structural differences between IgM proteins of the two affinity-valence groups involved ascertaining the number of free sulfhydryl groups present. The results obtained indicated that each of the proteins appeared to bind less than 0.005 moles of alkylating agent per mole of protein. Consequently, it seems highly unlikely that the observed functional differences could be attributable to differences in numbers of free sulfhydryls.

DISCUSSION

A priori it is possible, albeit unlikely, that the The original objectives of the work reported here failure of proteins in the “high affinity-low valence” were several. First, it was intended to develop a group to exhibit valences of 10 and to yield - 100% battery of structurally homogeneous murine pen- of the expected Fab fragments could be attributable tameric IgM molecules reactive with the DNP moiety to intramolecular differences in primary structure. and to establish their ligand-binding properties, es- Therefore extensively reduced and alkylated p and L pecially valences. Speculatively at least some of these chains from each protein of this group were isolated molecules would exhibit the expected (based upon (by gel filtration in the presence of 5 M dogma) 10 homogeneous binding sites, whereas guanidine-HCl) and subjected to limited amino ter- others would exhibit fewer such sites, a situation minal sequence analysis. The results obtained for two reminiscent of that previously reported for a limited of these proteins (2E7 and lB3) are presented in number of such mouse proteins (Giles et al., 1983). Table 3 and warrant several comments. It seems clear Second, it was intended that the reductive subunits that the p and L chains from each of these proteins from each of the monoclonal antibodies be subjected were homogeneous by this criterion. Of particular to tryptic digestion in order to ascertain the yields of importance in this regard were the observations that Fabp fragments. Of particular importance in this (a) the % yields at position 1 were uniformly >75%, regard was the need for comparing IgM molecules

Table 3. Amino terminal sequences of heavy and light chains of hybridoma-derived IgM antibodies exhibiting fewer than IO stable

binding sites for the DNP ligand

Protein

2E7 183

Position H chains L chains H chains L chains

1 GlU Asp GIU Asp 2 Val Ile Val Val 3 Gin Gln Gin Val 4 Leu Met Leu Met 5 Gln Asn Gln Thr 6 GIU Gin Glu Gin 7 Ser Ser Scr Thr 8 Gly Pro Gly Pro 9 Pro Scr Pro Leu

10 Ser Ser Ser 11 Leu LclI Leu I2 Val Val Pro

Page 6: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

92 DAVID PASCUAL and L. WILLIAM CLEM

with 10 homogeneous sites with others exhibiting fewer sites. As will be discussed below, it would appear that each of these objectives was fulfilled and the results can be interpreted as supporting the notion that intramolecular heterogeneity of ligand-binding sites may be a real phenomenon in certain IgM antibody molecules.

The results obtained strongly indicate that each of the hybridoma-derived IgM(w) antibodies studied represented bona fide - 900,000 mol. wt pentameric molecules composed of equimolar p chains (mol. wi - 70,000) and L chains (mol. wt ~22,00~26,000) arranged in pizLZ chain subunits (mol. wt - 180,000 wdS determined by gel filtration). Thus it seems highly unlikely that any observed variations in numbers of binding sites between the different IgM anti-DNP molecules could be attributed to too few (or too many) jr and/or L chains. In this regard it was noted that the IgM proteins studied here did appear to contain minor components (mol. wt - 53,000) which, at least superficially, resembled the truncated p chains described by others (Marks and Bosma, 1985). Since these minor components were variously present in at1 the IgM molecules studied here but did not represent more than 1% of the total protein, it seems highly unlikely that they would be responsible for any major differences in valences observed between the various IgM molecules.

The results of the equilibrium dialysis studies have revealed several potentially important points regard- ing ligand-binding by hybridoma-derived IgM anti- bodies to the DNP moiety. First, it seems clear that certain of these antibodies exhibit an average of about 10 homogeneous binding sites as would be expected for a molecule composed of t 0 p and 10 L chains. In contrast it is equally clear that certain other IgM molecules, also containing 10 p and t 0 L chains, can exhibit significantly fewer sites. Second. in terms of affinity considerations, it appears that although each of the IgM proteins studied exhibited hom- ogeneity of DNP binding (Sips heterogeneity indices of >0.96). the range of affinities for the various IgM proteins was considerable, i.e. from -4.0 x lo4 M _I to -1.2 x IO’M ‘. The fact that the higher affinity proteins studied exhibited the lower valences and the lower affinity proteins exhibited valences of nearly IO is intriguing. Unfortunately this observation is inex- plicable at present although intuitively it seems un- likely that it is just coincidental.

The finding of an average of seven to eight binding sites for some of the IgM molecules was unexpected, particularly in light of the fact that one of the proteins (BG-12) studied here and found to exhibit about seven and a half sites had previously (- 2 yr) been considered to have approx. five sites (Giles et al., 1983). Unfortunately the basis for this discrepancy cannot be explained at present other than to say approaches involving repeated subcloning of the BG- 12 secreting hybridoma line, the use of tissue culture supernatants rather than ascitic fluids as sources of

BG-12 antibody and the usage of a variety of thin- layer chromatography schemes for purifying the “H-c-DNP-lysine hapten failed to reveal any tangible evidence that could account for the apparent change in binding site numbers (D. Pascuat. MS. thesis, University of Mississippi Medical Center). Another potential source of error (or variability) in assessing antibody valences could involve the requi- site determination of antibody concns. For example, it the protein concns in certain of the IgM antibody solutions studied by eq~lilibrium dialysis were over- estimated by -250/b, the valences of those prepara- tions might artifactually appear to be -7.5 rather than 10. For the work reported here an extinction

coefficient (EIROnm iz, ) of 12 was assumed for each protein; this value is consistent with those reported for other IgM proteins (Miller and Metzger, 1965; Mukkur. 1972; Oriol and Rousset, 1974) and for protein BG-I2 (Giles et ul.. 1983). In addition, com- parisons by relative calorimetric methods (Lowry and Rio Rad, not shown) of mouse IgM protein solutions of identical optical densities (0.D.s) (280 nm) did not reveal differences which could explain the large differences in valences between the groups of IgM proteins studied here. Hence, it seems highly likely that the variability in valences (7.5--K? vs 9.4-10.0) observed here between the two groups of proteins is real, i.e. not an artifact; it is not clear whether or not the smaller variations in valences within groups can be attributed to minor differences in extinction coeflicients. It does, however, seem appropriate here

to speculate that the results of the lyophilization studies may have some bearing on this. as welt as other issues. In fact these data strongly argue for the notion that certain mouse IgM antibodies (repre-

sented here by the proteins in the “high affinity-low valence” group) actually can contain three somewhat different types of combining site regions. The first of these would be the nonbinding functionally inactive sites (presumably involving certain ,u-L chain pairs) demonstrable in freshly isolated material, i.e. inferred to exist based upon the finding of less that t 0 sites per molecule. The second would represent those two to three labile sites per molecule which are lost as a consequence of lyophilization. The third would repre- sent the remaining five to six stable sites that are resistant to this treatment. Consequently, it might be appropriate to suggest that all freshly synthesized and secreted IgM antibodies may actually contain t 0 functionally similar binding sites but, for some un- known reason(s), approximately half of these sites on molecules from certain cell lines are relatively labile, i.e. more susceptible to the “denaturing” effects of seemingly rather mild environmental factors. Obvi- ously, proof of this hypothesis requires further study. In future considerations of this issue it may be important to emphasize that >90% of the lyophi- lized p2L2 reductive subunits which contain an aver- age of about one site also bound to DNP affinity columns. This finding strongly supports the notion

Page 7: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

Ligand binding by IgM antibodies from murine hybridomas 93

that nearly every p2L2 subunit of this type must contain at least one stable and one unstable site, i.e. clearly indicating an intramolecular functional het- erogeneity situation for these particular IgM pro- teins.

The results of the trypsinization experiments with reductive subunits revealed that murine IgM mole- cules exhibit two distinctly different digestion pat- terns in terms of the yields of Fabp fragments, i.e. -50% vs - 100% of the expected values. Further- more there was a seemingly conclusive correlation between Fabp fragment yields and the number of ligand-binding sites eshibited by the parent molecule. Specifically those molecules exhibiting - 10 stable sites yielded - 100% of the expected fragments whereas those with approx. five to six sites after lyophilization (or significantly less than 10 sites in the native state) yielded - 50% of the expected fragments with the remainder being degraded to small peptides. The consistent finding that the Fabp fragments recov- ered from each of the IgM preparations exhibited valences of - 1, and consequently accounted for nearly all stable binding sites present in the parent molecule, strongly argues that the labile (and non- functional) sites present on certain of the proteins are quite susceptible to trypsinization. If this interpre- tation is correct, then clearly another manifestation of intramolecular heterogeneity for some, but not all, IgM antibodies is apparent.

In terms of rationalizing the above results indi- cating several levels of apparent intramolecular heterogeneity in certain murine IgM antibodies, there are several issues that must be considered. To begin, it was possible, albeit unlikely in our opinion, that the observed intramolecular functional heterogeneity was actually a reflection of varibility at the primary structural (amino acid sequence) level of either the p or L chains. Clearly the limited sequence and sulfhy- dry1 titration studies performed here argue against this potential problem although it is possible that intramolecular structural heterogeneity exists in other regions, i.e. hypervariable residues. An indirect argu- ment against this latter notion would be based upon the observed homogeneity of ligand-binding for those native molecules exhibiting both stable and labile DNP binding sites. It seems highly unlikely that each of four different IgM molecules containing a mixture of p chains (or L chains) which might differ at the primary structural level would exhibit identical affinities (manifest as Sips heterogeneity indices of - 1.0) for the stable and labile sites present on that particular molecule. Theoretically such a circum- stance is possible, but its occurrence in four out of four cases seems highly improbable. Thus, pro- ceeding on the assumption that the IgM molecules studied here were homogeneous at the primary struc- tural level, it seems appropriate to search for an alternative explanation for the observed intra- molecular functional heterogeneity. One possibility, as suggested previously (Giles et al., 1983). would be

to invoke the idea of intramolecular conformational differences. Although it seems that each of the above observations (regarding binding site and Fabp frag- ment stabilities) are compatible with the idea of intramolecular conformational differences, obtaining direct proof of this issue may be difficult. Clearly the proper strategy would involve searching for evi- dence of conformational differences between “good” (stable) and “bad” (labile) Fabp fragments from molecules exhibiting the putative intramolecular bet.- erogeneity. Unfortunately the lability to tryp- sinization of the latter fragments from such molecules precludes this approach at present. Optimistically ongoing work involving the usage of other pro- teolytic protocols (Lin and Putnam, 1978) will result in the availability of these necessary fragments.

In terms of the origins of these putative intra- molecular conformational differences, it could be postulated that they might arise as a consequence of different modes of intracellular assembly of newly synthesized polypeptides into intact IgM molecules. Our initial and ongoing approaches here have in- volved pulse-chase protocols to establish the pat- tern(s) of covalent p-L chain intracellular assembly. The results indicate rather clearly the existence of two different assembly modes (Deuter et ul., in prepara- tion). On one hand, proteins 2Fll and IF3 (each exhibiting 10 stables sites) were assembled covalently via the

pathway, previously considered characteristic for IgM antibodies (Baumal and Scharff, 1973; Park- house, 1971). On the other hand, proteins BG-12 and 1E5 (each exhibiting stable and labile sites) were assembled in a different fashion involving a covalent p,L chain intermediate. Thus, if intramolecular con- formational differences actually exist in these latter proteins, it certainly seems possible that such differences in symmetry could somehow be generated by the later addition of the second L chain to the /I:L intermediate resulting in two somewhat different con- formations for the two Fabp regions within the subunit. In the former case, where two existing /(-I, intermediates form pZLZ subunits one might expect to see true intramolecular symmetry, i.e. 10 functionally equivalent sites per pentameric molecule.

A final point to be considered would be the ques- tion of the biological significance of two different forms of IgM molecules in the mouse. Obviously there is as yet no rational answer to this question. It does seem, however, that future efforts at under- standing the significance of the intramolecular phe- nomenon must not only consider both the membrane and secreted forms of IgM but also the issue of whether or not intramolecular symmetry at the idiotype level is found among certain IgM molecules. Clearly the work reported here raises more questions than it answers: foremost among these is the question

Page 8: Ligand binding by murine IgM antibodies: Intramolecular heterogeneity exists in certain, but not all, cases

94 DAVID PASCUAL and L. WILLIAM CLEM

of why all combining sites in a given IgM molecule are not identical, as would be expected from current dogma.

REFERENCES

Baumal K. and Scharff M. (1973) Synthesis, assembly, and secretion of mouse immunoglobulin. Transpl. Rev. 14, 163-183.

Giles B., Klapper D. G. and Clem L. W. (1982) Intrasubunit homogeneity in heterogeneous IgM antibodies to the DNP moiety derived from a murine hybridoma cell line. Molec. Immun. 19, 747-754.

Giles B., Klapper D. G. and Ciem L. W. (1983) Intra- molecular heterogeneity of ligand binding of two IgM antibodies derived from murine hybridomas. Molec. Immun. 20, 737-744.

Kabat E. A., Wu T., Reid-Miller M., Perry H. and Gottens- man K. (Editors) (1987) In Sequences of Proteins of Immunological Interest, 4th edn: U.S. Department of Health and Human Services. Bethesda. MD.

Klapper D. G., Clem L. W. and Small P. A. (1971) Proteolytic fragmentation of elasmobranch immune: globulins. Biochemistry 10, 645-652.

Lin L.-C. and Putnam F. W. (1978) Cold pepsin digestion: a novel method to pruduce the Fv fragment from human immunoglobuhn M. Proc. natn. Acad. Sci. U.S.A. 75, 2649-2653.

Marks R. and Bosma M. J. (1985) Truncated ~01’) chains in murine IgM: evidence that p’ chains lack variable regions. J. exp. Med. 162, 186221877.

Metzger H. (1970) Structure and function of yM macro- globulins. Adv. Immun. 12, 57-116.

Miller F. and Metzger H. (1965) Characterization of a human macroglobulin-I. The molecular weight of its subunit. J. biol them. 240, 3235-3333.

Mukkur T. K. S. (1972) Valence and association constant of bovine colostral immunoglobulin M antibody (IgM). Immunochemistry 9, 1049-1055.

Oriol R. and Rousset M. (1974) The IgM antibody site-II. Comparison of the two populations of IgM antibody sites and analysis of the behavior of the low affinity population at equilibrium with a dinitrophenylated hapten. J. Immun. 112, 2235-2240.

Parkhouse R. M. E. (1971) Immunoglobulin M bio- synthesis: production of intermediates-and excess light chain in mouse mveloma MOPC 104E. Biochem J. 123. 635-641. -

Shankey V. and Clem L. W. (1980a) Phylogeny of immuno- globulin structure and function-VIII. Intermolecular heterogeneity of shark 19s IgM antibodies to pneumoccal polysaccharide. Molec. Immun. 17, 366-375.

Shankey V. and Clem L. W. (19806) Phylogeny of immuno- globulin structure and function-IX. Intramolecular heterogeneity of shark 19s IgM antibodies to the dinitro- phenyl hapten. J. Immun. 125, 269G2698.