Based primarily on studies in the mouse, it has been postulated that at least two separate lineages of B cells exist (reviewed in reference 1). The first lineage, characterized by the cell surface expression of the Ly-1 antigen, appears to be primarily responsible for the secretion of antibodies that display extensive autoreactivity. The second lineage, characterized by the lack of Ly-1 expression, is primarily responsible for the production of "conventional", non-self-reactive antibodies (2-5). This distinction has been most convincingly demonstrated in inbred strains of mice with a genetic predisposition for the development of various autoimmune syndromes (6, 7).The human counterpart ofthe Ly-1 antigen, recently termed CD5, was originally defined using the anti-Leu-1 mAb (8). In this (8) and subsequent (9-12) studies, it was demonstrated that the expanded monoclonal population of B cells that develops in patients with chronic lymphocytic leukemia (CLL)' expresses the CD5 molecule. Thus, CLL appears to represent a clonal overexpansion of the putative autoantibody-producing B lymphocyte . In support ofthis notion are the clinical observations that the sera of patients with CLL occasionally contain autoantibodies and that -20% of such patients develop autoimmune phenomena such as hemolytic anemia (13).On the basis of these findings, the following prospective studies were performed to determine, at the clonal level, if the C135-expressing B cells in patients with CLL and other CD5' B lymphoproliferative disorders secrete mAbs reactive with autoantigens . These in vitro studies demonstrate that leukemic cells from >50% of patients with overexpansions of CD5' cells synthesize and can be made to secrete Ig that is autoreactive . Considering that these secreted mAbs were screened for reactivity with only a limited panel of autoantigens, it is conceivable that an even higher

SummaryChronic lymphocytic leukemia (CLL) is characterized by the clonal expansion of CD5-expressing B lymphocytes. Most studies have found that these leukemic CD5 + B cells, like their normal counterparts, use immunoglobulin (Ig) variable (V) region genes that exhibit minimal, if any, somatic diversity. These and other observations have suggested that CD5 § B cells may be incapable of generating Ig V gene diversity, and therefore may not be able to develop higher affinity binding sites that could be selected by antigen. However, most of the studies of CLL and normal CD5 + B cells have focused on IgM-producing cells. Since somatic mutations are most often seen in B cells that have undergone an isotype class switch, we analyzed the Ig heavy (H) and light (L) chain variable region genes of seven IgG + CD5 + CLL B cells to determine if somatic diversification and antigen selection had occurred. The data derived provide evidence for skewed use, somatic diversification, and antigenic selection of the Ig V region genes. Nonrandom use of both H and L chain V region genes was manifested by an overrepresentation of VH4 and V~I family genes and the underrepresentation of the JH4 gene segment. Furthermore, VH4 gene use was restricted to only two family members (4.21 and 4.18). In four of the seven cases, the V. and VL genes displayed />5% difference from the most homologous known germline counterparts. Polymerase chain reaction and Southern blot analyses performed in two of these patients demonstrated that their unique V. CDR2 and adjacent sequences were not present in their germline DNA. In addition, a significant level of diversity was seen in the rearranged DJ8 segments and at the VL-JL junctions of every patient that occurred both at the time of recombination and subsequently. The localization of replacement changes to complementarity determining regions of some patients suggested that antigen selection had occurred. Furthermore, the mutations identified in the V~ and VL genes of each individual patient were strikingly similar, both in number and location. Collectively, the data indicate that a subset of CD5 § CLL B cells can display Ig V region gene mutations. In addition, they are consistent with the notions that in some cases antigen selection of these mutations may have occurred, and that antigen stimulation may be a promoting factor in the evolution of certain CLL clones. most invariably expresses surface membrane CD5 (for a review see reference 1). Over the years, investigators have studied these cells to understand both the normal physiology of this subset of human B cells as well as the abnormal features of the B lymphocytes expanded in this disease. Studies to date suggest that the B cells that are clonally expanded in CLL use a biased set of Ig V genes (for a review see reference 2) to code for low affinity, polyreactive autoantibodies (3-5), predominantly of the IgM class. Although numerous studies have noted that these Ig V genes rarely undergo somatic mu- 1507J. Exp.

Several questions exist regarding CD5+ B cells. These include the ability of these cells, as compared to CD5- B cells, to undergo an Ig isotype class switch, the subclasses utilized, and the effects that switching may have on antigen binding. To address these issues, ten patients with chronic lymphocytic leukemia (CLL) whose CD5+ leukemic B cell clones produced IgG were studied. Monoclonal IgG was collected from PMA-stimulated CLL cells and from heterohybridomas constructed with these cells, and then analyzed for IgG subclass utilization, autoreactivity, and DNA idiotype expression. The monoclonal B cells from 80% of the CLL patients produced IgG1 and those from 20% produced IgG3. None produced IgG2. In contrast to the known autoreactivity of IgM-producing CD5+ CLL cells (> 50% autoreactive), none of these IgG antibodies reacted significantly with the autoantigens tested. However, three did react significantly with autoantigen after artificially increasing antibody valency by crosslinking. Whereas five of the IgG molecules expressed a cross reactive idiotypic (CRI) marker characteristic of non-mutated kappa anti-DNA antibodies, three expressed a CRI displayed primarily on mutated IgG anti-DNA antibodies. Thus, some CD5+ human B cells can undergo an isotype class switch that for these CLL cells is biased against IgG2 and in favor of the IgG1 and IgG3. In their native state the IgG molecules secreted by these isotype-switched CD5+ cells have diminished autoreactivity, as compared to IgM-producing CLL cells. Since some of the IgG antibodies could be made auto- and poly-reactive by increasing antigen-binding valency, while others expressed idiotypic markers of mutated antibodies, certain of these CD5+ B cells probably utilize non-mutated Ig V genes coding for polyreactive antibodies, whereas others may use genes that have undergone somatic mutation and that code for more restricted specificities. Therefore, both valency and VH gene mutation may account for the diminished autoreactivity of these CD5+ B cell-derived IgG antibodies.

A large battery of anti-CD23 mAb were compared for their epitope specificities and for their abilities to alter both IgE binding to cell-associated CD23 and IgE production in vitro in response to three sets of stimulants. The nine mAb tested can be divided into four families which define four antigenic epitopes (A-D) of CD23. Of these four families, two bind antigenic sites, (A and D) that appear to lie outside the IgE ligand binding site and two bind sites (B and C) that appear to be located within or close to this site, as determined by the abilities of appropriate mAb to alter IgE binding to CD23. The effects that these mAb had on IgE secretion by normal peripheral blood mononuclear cells (PBMNC) varied depending on the stimulant employed to induce IgE production. Interactions with epitope A, which was found to lie outside the ligand binding site and to be made more accessible by binding of mAb to other epitopes, had different effects on IgE production than interactions with the other epitopes. Indeed, mAb binding to this epitope lead to as much as a 10 fold enhancement in IgE biosynthesis induced by IL-4 alone or by IL-4 + hydrocortisone whereas interactions at the other sites resulted in almost complete inhibition of IgE production. In addition, mAb reactive with epitopes B and C had minimal effects on IgE production induced by IL-4 + anti-CD40 mAb whereas interactions at epitope A consistently enhanced IgE production. Finally, no apparent direct correlation was found between the ability of individual anti-CD23 mAb to alter IgE binding to cell-associated CD23 and their ability to modulate IgE production by PBMNC. These studies suggest that IgE binding to cell-associated CD23 does not have a major role in the de novo synthesis of IgE that involves CD23 interactions. In addition, the different effects that binding to epitope A vs B or C have on IgE synthesis suggest that molecular interactions between distinct portions of the CD23 molecule and other cell surface molecules expressed on the same B cell or adjacent communicating cells may lead to divergent cellular effects on IgE production. Finally these studies imply that only epitope A is involved in the generation of an IgE response through the CD40 pathway.