Analyses of Recombinant Stereotypic IGHV3-21–Encoded Antibodies Expressed in Chronic Lymphocytic Leukemia

Abstract

Chronic lymphocytic leukemia (CLL) cells that use IgH encoded by IGHV3–21 and that have a particular stereotypic third CDR (HCDR3), DANGMDV (motif-1), almost invariably express Ig L chains (IgL) encoded by IGLV3–21, whereas CLL that use IGHV3–21–encoded IgH with another stereotypic HCDR3, DPSFYSSSWTLFDY (motif-2), invariably express κ-IgL encoded by IGKV3–20. This nonstochastic pairing could reflect steric factors that preclude these IgH from pairing with other IgL or selection for an Ig with a particular Ag-binding activity. We generated rIg with IGHV3–21–encoded IgH with HCDR3 motif-1 or −2 and IgL encoded by IGKV3–20 or IGLV3–21. Each IgH paired equally well with matched or mismatched κ- or λ-IgL to form functional Ig, which we screened for binding to an array of different Ags. Ig with IGLV3–21–encoded λ-IgL could bind with an affinity of ~2 × 10⁻⁶ M to protein L, a cell-wall protein of Peptostreptococcus magnus, independent of the IgH, indicating that protein L is a superantigen for IGLV3–21–encoded λ-IgL. We also detected Ig binding to cofilin, a highly conserved actin-binding protein. However, cofilin binding was independent of native pairing of IgH and IgL and was not specific for Ig with IgH encoded by IGHV3–21. We conclude that steric factors or the binding activity for protein L or cofilin cannot account for the nonstochastic pairing of IgH and IgL observed for the stereotypic Ig made by CLL cells that express IGHV3–21. The Journal of Immunology, 2011, 186: 000–000.

The Ig expressed in chronic lymphocytic leukemia (CLL) cells have restricted diversity. The IgH used by CLL cells of nonrelated patients frequently use certain IgH V region genes (IGHV), diversity, and junctional gene segments, often with restricted reading frames (1–5). Moreover, the leukemia cells of different patients that use the same IGHV often can be found to have the same or similar H chain CDR3 (HCDR3) sequence. Such stereotypic sequences are found in 28% of CLL patients with leukemia cells that express unmutated IGHV and 12% of CLL patients with leukemia cells that express mutated IGHV (6).

In addition, there are examples in which certain IgH preferentially are expressed with certain Ig L chains (IgL) by CLL cells (3, 7). Tobin and colleagues (8), for example, reported that CLL cells that make IgH encoded by one IGHV, namely, IGHV3–21, frequently express λ-IgL encoded by IGLV3–21. The IGHV3–21–encoded IgH paired with such l-IgL almost invariably have a stereotypic HCDR3, characterized by the amino acid sequence D(A/G)NGMDV, which we designate here as motif-1. On the other hand, we found that some CLL cells also can make another IgH encoded by IGHV3–21 with a different stereotypic HCDR3, characterized by the amino acid sequence DPSFYSSSWTLFDY, which we designate as motif-2. The CLL cells that express this IgH invariably were found to express a K-IgL encoded by IGKV3–20 (9). The nonstochastic pairing of particular IgL with IgH with certain HCDR3 could reflect selection for Ig with binding properties for an Ag(s) that potentially plays a role in leukemogenesis.

Alternatively, the nonstochastic pairing of certain IgH and IgL could be because of steric factors that preclude such IgH or IgL from pairing with other Ig polypeptides to form an intact Ig molecule. Previous studies provided evidence for how such biased assembly of certain IgH with certain IgL could affect B cell maturation in at least two developmental checkpoints: pre-B to immature B cell transition and immature to mature B cell differentiation. As an example of the former, investigators observed that not all IgH could associate with surrogate λ L chain (10). Moreover, pre-B cells that express only IgH encoded by IGHV11–2 may not develop normally into immature B cells because of the poor association of such IgH with surrogate λ L chain (10, 11). Furthermore, the repertoire of mature B cells might be limited through a failure for certain IgH to pair with any but a few IgL (12–18). The most notable example of this is provided from studies on the repertoire of mouse B cells specific for phosphatidyl choline (PtC). PtC-specific B cells express predominantly one of two IgH/IgL combinations with variable regions encoded by V_H12/Vκ4/5H and V_H11/Vκ9, respectively. Moreover, the HCDR3 of such Abs is made up of 10 aa with an invariant glycine in the fourth position and a tyrosine encoded by J_H1 in the fifth position, a motif designated 10/G4 (19). Although the apparently biased association of IgH encoded by VH12 with IgL encoded by VK4/5H could be interpreted as demonstrating selection for Ig with binding to PtC, subsequent studies revealed a biased use of Vk4/5H by VH12-expressing B cells, even B cells making Ig that did not bind PtC (20). Moreover, the IgH encoded by VH12 were found physically unable to associate with different IgL (20). As such, the biased pairing of VH12-encoded IgH with IgL encoded by VK4/5H appeared secondary to selection of VH12-expressing B cells for making functional Ig, rather than for making Ig with a binding activity for a particular Ag, such as PtC. Similarly, there are other examples in which certain IgH and IgL are biochemically incompatible to form dimers required for assembly of functional Ig (12–17).

Conceivably, the respective association of IGHV3–21–encoded IgH with motif-1 or motif-2 HCDR3 for IgL encoded by IGLV3–21 or IGKV3–20 in CLL could reflect similar biochemical constraints that preclude nonnative combinations of IgH and IgL from assembling into intact Ig. To assess the significance and biological basis for selective pairing of IgH and IgL in CLL, we examined the Ig used by CLL cells that express IGHV3–21.

Materials and Methods

Patients and leukemia cells

Blood was collected from consenting patients who satisfied diagnostic and immunophenotypic criteria for CLL and who presented for evaluation at the referral centers of the CLL Research Consortium. Institutional review board approval and informed consent were obtained for the procurement of the samples in all cases, in accordance with the Declaration of Helsinki.

PBMCs were isolated by density gradient centrifugation using Ficoll-Hypaque 1077 (Sigma, St. Louis, MO), washed twice, and analyzed directly or suspended in FCS containing 10% DMSO (Sigma) for storage in liquid nitrogen. All samples contained >90% CLL B cells as assessed by flow cytometric analyses, and the isotype of the expressed IgL was determined by flow cytometry as described previously (5).

Sequence analyses of IGHV, IGKV, and IGLV

The IGHV gene expressed by the CLL cells was determined by reverse transcription-PCR. The cDNA from each sample was amplified using IGHV, IGKV, or IGLV family-specific primers for the sense strand of the gene of interest and antisense IGHM, IGKC, or IGLC consensus primers, respectively (21). Nucleic acid sequence analyses were conducted as previously described (9).

Recombinant Ab production and purification

All nested PCR reactions with gene-specific primers or primer mixes were performed with 3.5 μl purified first PCR product as already described (22). Digested PCR product purification and ligation into human Igγ1, Igκ, and Igλ expression vectors were performed as previously described (22). Irrespective of their original isotype, all rAbs are generated as IgG1 molecules to facilitate the detection and comparability in reactivity assays.

Escherichia coli XL1-blue supercompetent cells (Stratagene, San Diego, CA) were transformed at 42°C with 2 μl ligation product. Colonies were screened by PCR using same primers used for nested PCR. PCR products were sequenced to confirm the identity with the original PCR products. Plasmid DNA was isolated from bacteria cultures grown for 16 h at 37°C in Luria-Bertani medium (MP Biochemicals, Solon, OH) containing 100 μg/ ml ampicillin (Sigma) using QIAprep columns (Qiagen, Valencia, CA).

293A human embryonic kidney (HEK) cells were cultured in DMEM supplemented with 10% ultra-low IgG FBS (Invitrogen, Carlsbad, CA) and cotransfected with 12.5 μg IgH and 13.7 μg IgL encoding plasmid DNA by calcium phosphate precipitation. Twelve hours after transfection, the cells were washed with serum-free DMEM and thereafter cultured in DMEM supplemented with 1% Nutridoma SP (Roche, Indianapolis, IN). Supernatants were collected after 8 d of culture (23). Cell debris was removed by centrifugation at 800 × g for 10 min, and rAbs were purified from culture supernatants with protein G beads (GE Healthcare, Piscataway, NJ) as described previously (22). We examined the purified rIgG by SDS gel electrophoresis under nonreducing and reducing conditions. For reduction, 5% DTT was added to the denaturing sample buffer and all samples were incubated for 10 min at 70°C. One microgram of Ig protein was added to each lane of a 4–15% Tris-HCl polyacrylamide gel (BioRad, Hercules, CA) for electrophoresis. Protein bands were visualized by silver staining.

Evaluation of whether steric factors affect the IgH and IgL pairing

293A cells were cotransfected by calcium phosphate precipitation with IGLV3–21 and IGKV3–20 encoding plasmid DNA and limiting amounts (33 and 10%, respectively) of IgG expression plasmids driving expression of IgH with HCDR3 motif-1 or motif-2. Supernatants were collected after 8 d of culture, and cell debris was removed by centrifugation at 800 × g for 10 min. rAbs were purified from culture supernatants with protein G beads (GE Healthcare). Total amount of intact IgH/κ-IgL and IgH/λ-IgL after protein G purification was determined by sandwich ELISA, as previously described (22).

Ag microarray

We examined binding activity with a panel of 45 self-Ags or environmental Ags (Supplemental Table I): small nuclear ribonucleoprotein 68/70 (snRNP 68/70), tetanus toxoid, malondialdehyde BSA (MDA-BSA), malondialdehyde-modified low-density lipoprotein (MDA-LDL), Sjögren syndrome Ag B (SSB), small nuclear ribonucleoprotein BB (snRNP BB), small ribonucleoproteins (SmRNP), smith Ag (Sm), small nuclear ribonucleoprotein A (snRNP A), histone, Sjögren syndrome Ag A of 60 kDa (60), aggrecan, proteoglycan, centromere protein A (CENP-A), Hep-2 extract, glomerular extract, Ku, centromere protein B (CENP-B), alanyl-tRNA synthetase (PL-12), oxidized LDL (OxLDL), human transglutaminase (HtTG), Sjögren syndrome Ag A of 52 kDa (SSA 52), C1q, ssDNA, dsDNA, nucleolar extract, chromatin, laminin (mouse), pneumococcal polysaccharide 23 (pneumococcal PS 23), pneumococcal multiserotype vaccine (Pneumovax), pneumococcal polysaccharide 12 (pneumococcal PS 12), pneumococcal polysaccharide 14 (pneumococcal PS 14), phosphorylcholine conjugated to BSA (PC-BSA), LPS, anti-Saccharomyces cerevisiae Ab (ASCA), capsular polysaccharide (C-PS), BSA, thyroglobulin, fibrinogen type I-S, ribosomal-P, ATPase-associated Ag (M2), entactin, annexin V, collagen IV ms, and protein L. Microarrays were printed on nitrocellulosecoated FAST slides (Whatman, Florham Park, NJ) with a QSoft QArray Mini, using QSoft microarray software (Genetix USA, Boston, MA), with six replicates of each Ag. After blocking with Whatman blocking buffer, the slides were incubated with rIgG (5 μg/ml), or control IgG, in PBS, pH 7.4. The slides were processed and data determined as described previously (24).

Immunoblot for Peptostreptococcus magnus protein L

After electrophoresis of purified recombinant protein L (Pierce, Rockford, IL), the size-separated protein was transferred from the gel onto a nitrocellulose membrane (Invitrogen) by dry electrotransfer using iBlot dryblotting system (Invitrogen). Nitrocellulose membranes were blocked with 5% milk TBS, 0.1% Tween 20 overnight at 4°C. Incubation of native and mismatched Abs was performed using the Cassette Miniblot System (Immunetics, Cambridge, MA) with 28 channels. Native and mismatched Abs were adjusted to 30 μg/ml and incubated for 3 h at room temperature. We performed immunoblot using 30 μg/ml of each rAb to detect Abs with low binding affinity for Ag. After incubation, native and nonnative pairs of IgH/IgL were washed with TBS, 0.1% Tween 20. Secondary HRP-conjugated anti-human IgG (Southern Biotechnology, Birmingham, AL) was incubated for 45 min and then washed as before. The blots were prepared for ECL and subsequent autoradiography.

ELISA for Abs to Peptostreptococcus magnus protein L

We determined the concentration of rIgG in culture supernatants by ELISA using human monoclonal IgG1 λ (Antigensite, San Diego, CA) or human monoclonal IgG1 κ (Southern Biotechnology) to generate a standard curve. Protein L binding activity was measured by ELISA using Reacti-Bind protein L-coated plates (Pierce).

Competitive ELISA with Peptostreptococcus magnus protein L

We cotransfected 293A cells with equal amount (6 mg) of IgG expression plasmids driving expression of IgH with HCDR3 motif-2 and IGKV3–20 encoding plasmid DNA, and increasing amounts of IGLV3–21 encoding plasmid DNA (2, 4, 6, 8, 16, and 24 μg, respectively) using calcium phosphate precipitation. Supernatants were collected after 8 d of culture, and cell debris was removed by centrifugation at 800 × g for 10 min. rIgG were purified from cell culture supernatants with protein G beads (GE Healthcare). We examined the purified IgG by SDS gel electrophoresis under nonreducing conditions. Protein bands were visualized by silver staining.

We determined the concentration of IgG in tissue culture supernatants by ELISA using human monoclonal IgG1 λ (Antigensite) or human monoclonal IgG1 κ (Southern Biotechnology) to generate a standard curve. Protein L binding activity was measured by ELISA using Reacti-Bind protein L-coated plates (Pierce) and secondary anti-IgG Ab HRP conjugated (Southern Biotechnology).

Ab affinity measurements

We experimentally determined the equilibrium dissociation constant (K_d) of the following rIgG: mutated IGHV3–21 with HCDR3 motif-1 paired with its native λ-IgL encoded by IGLV3–21 designated as H_AL_A, unmutated IGHV3–21 with HCDR3 motif-1 paired with its native λ-IgL encoded by IGLV3–21 designated as H_BL_B, unmutated IGHV3–21 with HCDR3 motif-2 paired with its native κ-IgL encoded by IGKV3–20 designated as H_CL_C. The K_d measurements were performed on a KinExA 3200 instrument (Sapidyne Instruments, Boise, ID) using polystyrene beads as a solid phase to immobilize the ligand, protein L. For each experiment, a series of solutions with fixed concentrations of rIgG and various concentrations of ligand were equilibrated. The solutions were then briefly exposed to the solid phase, and free IgG was captured and measured by detection with goat anti-human IgG conjugated with Cy5 (Jackson ImmunoResearch, West Grove, PA).

Immunoblot analyses for Abs reactive with cofilin

Lysates of human monoblastic leukemia (U937) cells, expressing cofilin (nonmuscle type, cofilin 1), in radioimmunoprecipitation assay buffer were size separated on two-dimensional wells SDS-PAGE. The size-separated proteins were transferred from the gel onto a nitrocellulose membrane (Invitrogen) by dry electrotransfer using iBlot dry blotting system (Invitrogen). We performed immunoblot to test rIgG composed of native and nonnative pairs of IgH and IgL, using the Cassette Miniblot System (Immunetics, Cambridge, MA) with 28 channels. Native and mismatched Abs were adjusted to 30 μg/ml and incubated for 3 h at room temperature. We performed immunoblot using 30 μg/ml of each recombinant Ab to detect Abs with low binding affinity for Ag. After incubation, native and nonnative pairs of IgH/IgL were washed with TBS, 0.1% Tween 20. Secondary HRP-conjugated anti-human IgG (Southern Biotechnology) was incubated for 45 min and then washed as before. The blots were prepared for ECL and subsequent autoradiography. Rabbit polyclonal anti-cofilin 1 (ECM Biosciences, Versailles, KY) was used as positive control.

Results

rIgG with native pairs of IgH and IgL

CLL that make IgH encoded by IGHV3–21 with a HCDR3 harboring the amino acid sequence D(A/G)NGMDV (motif-1) generally also express a λ-IgL encoded by IGLV3–21 (8). We generated the IgH cDNA of two representative cases, one encoded by IGHV3–21 harboring a modest number of somatic mutations, which we designated as H_A, and another encoded by IGHV3–21 without somatic mutations, but with variant motif-1 HCDR3 (DQNGMDV), which we designated as H_B (Table I). This variant motif-1 HCDR3 is found in the IgH produced by subset of CLL cases that express IGHV3–21 and IGLV3–21 (9). The D segment encoding motif-1 or its variant cannot be assigned because of the short length of these HCDR3. We also generated the cDNA for the IGLV3–21–encoded λ-IgL from each of these two cases, which we designated as L_A and L_B, respectively. In contrast, CLL that make IgH encoded by IGHV3–21 with a HCDR3 harboring the amino acid sequence DPSFYSSSWTLFDY (motif-2) instead express Ig κ L chains encoded by IGKV3–20 (9). From a representative CLL case, we generated the cDNA encoding the IgH and κ-IgL, which we designated as H_C and L_C, respectively (Table I). We also generated the Ig cDNA from another CLL case that used an unrelated IgH encoded by IGHV4–39 and a λ-IgL encoded by IGLV1–40, which we designated as H_D and L_D, respectively. Finally, we made the Ig cDNA from yet another unrelated CLL case (CLL69 C/C) (3, 7), which expressed IgH and IgL encoded by IGHV1–69 and IGLV3–09, respectively. From these cDNA, we generated expression vectors encoding human IgG H chains and λ- or κ-IgL, which we cotransfected into 293A cells to produce recombinant human IgGλ or IgGκ, the characteristics of which are provided in Table I.

Table I.

Characteristics of rIgG

rIgG ID (Reference)	IGHV Gene	IGHD Gene	IGHJ Gene	% Germline Identity	HCDR3	IGKV/IGLV Gene	% Germline Identity	LCDR3
HALA (9)	IGHV3–21		JH6	95.4	DGNGMDV	IGLV3–21	97	QVWDSGSDHPWV
HBLB (9)	IGHV3–21		JH6	99	DQNGMDV	IGLV3–21	99.3	QVWDAGSDHPWV
HCLC (9)	IGHV3–21	D6–13	JH4	100	DPSFYSSSWTLFDY	IGKV3–20	100	QQYGSSPLT
HDLD	IGHV4–39	D2–2	JH6	100	HRLGYCSSTSCYYYYYGMDV	IGLV1–40	100	QSYDSSLSVV
CLL69 C/C (7)	IGHV1–69	D3–3	JH6	99.3	AGGYDFWSGYYSNYYYYGMDV	IGLV3–09	100	QVWDSSTEV

Amino acid sequences of the Ig HCDR3 and L chain LCDR3 of the Ig expressed by unrelated patients who have CLL cells that expressed IGHV3–21 with either HCDR3 motif-1 or -2, IGHV4–39, or IGHV1–69.

IGHD, IgH diversity; IGHJ, IgH junctional.

The IgG produced by transfected 293A cells was evaluated by SDS gel electrophoresis under nonreducing (Fig. 1A) or reducing (Fig. 1B) conditions. As anticipated, we noted that the native pairs of IgH and IgL assembled into intact IgG of the expected molecular size.

FIGURE 1.

SDS gel electrophoresis of purified rIgG under nonreducing (A, C) or reducing conditions (B, D). The kiloDaltons of size-separated proteins are indicated at the left margin on each panel. In the first two lanes of each gel are the lanes with m.w. markers (M; BioRad) or 1 μg control human IgG1κ Ab (control IgG; Southern Biotechnology). In each of the other lanes, 1 μg IgG composed of native pairs of IgH and IgL (A, B) or nonnative pairs of IgH and IgL (C, D) were loaded to each lane, as indicated at the top of each lane. Protein bands were visualized by silver staining. Sizes of IgG (150 kDa), IgH (50 kDa), and IgL (25 kDa) are indicated on the left of each panel.

Capacity of nonnative pairs of IgH and IgL to assemble into functional IgG

We examined whether the IgH could assemble with different IgL to form intact Ig that ordinarily are not expressed by CLL cells. For this, we cotransfected 293A cells with expression vectors encoding IgH H_A, H_C, or H_D with vectors encoding IgL L_C, L_A, or L_D, respectively. We found that cotransfection of vectors encoding nonnative pairs of IgH and IgL generated proteins of the expected molecular size for intact IgG, as assessed via PAGE under nonreducing (Fig. 1C) or reducing conditions (Fig. 1D).

We also cotransfected 293A cells with fixed and limiting amounts of expression vector directing synthesis of IgH H_A or H_C together with fixed amounts of plasmids directing synthesis of IgL L_A or L_C. We determined the total amounts of intact Ig that were generated in a fixed amount of time by transfected 293A cells by ELISA. We did not observe significant differences between the amounts of intact Ig produced by 293A cells transfected with vectors encoding H_A/L_C or H_A/L_A (1.8 ± 0.3 versus 1.6 ± 0.4 μg, respectively, p = 0.4). Similarly, we did not observe significant differences between the amounts of intact Ig produced by 293A cells transfected with vectors encoding H_C and L_C or H_C and L_A (1.9 ± 0.4 versus 1.6 ± 0.3 μg, respectively, p = 0.2). These results indicate that L_C could assemble with H_A and L_A could assemble with H_C, even in 293A cells transfected with limiting amounts of expression vector encoding H_A or H_C, respectively.

Ag microarray, protein L ELISA, and immunoblot analyses

We examined the capacity of the rIgG to bind one or more Ags present in an array of tissue proteins, plasma proteins, enzymes, mitochondrial proteins, and microbial proteins (Supplemental Table I). We found that IgG composed of H_AL_A, H_BL_B, or H_CL_C, but not H_UL_D, could bind protein L (Figs. 2A, 2B), but not any of the other Ags spotted on the array (Supplemental Fig. 1, Supplemental Table II). Protein L is a multidomain cell-wall protein isolated from Peptostreptococcus magnus, a Gram-positive commensal bacterium that comprises a large portion of the human bacterial gut flora. The capacity for IgG composed of H_CL_C to bind protein L was not surprising, as prior studies indicated that this microbial protein could bind to human Ig κ L chains encoded by IGKV genes of the VκI, VκIII, or VκIV subgroups (25, 26). However, human Ig with λ L chains had not been observed to bind protein L. Because all IgG with IgH encoded by IGHV3–21 could bind protein L, including IgG with λ-IgL, such as H_AL_A or H_BL_B, we speculated that protein L could be a target Ag driving selection of the stereotypic Ig used by CLL cells that express IGHV3–21.

FIGURE 2.

Native rAbs binding tests. A, Protein L ELISA. Ordinate depicts OD measurements at 450 nm of the ELISA for native rIgG using plates coated with protein L. The abscissa provides the concentration of IgG used in the ELISA in micrograms per milliliter (μg/mf). ELISA data using IgG composed of H_AL_A (diamonds), H_BL_B (circles), H_CL_C (gray squares), and H_DL_D (triangles) are represented by the symbols, as indicated in the fegend on the upper right margin. B, Protein L immunoblot of rIgG. Recombinant protein L was size separated on SDS-PAGE and transferred to nitrocellulose membrane (Invitrogen). This was probed with various concentrations of different rIgG, at 30, 10, and 1 μg, as represented by the triangles at the top of the gel. The different IgG used are indicated above each triangle.

To evaluate for this, we examined rIgG composed of nonnative combinations of IgH and IgL. We found IgG with the λ-IgL L_A could bind protein L, even when the IgG had the IgH H_C or H_D (Fig. 3A, 3B). Similar to the binding activity of IgG composed of H_AL_A, the binding activity of IgG composed of H_CL_A or H_DL_A for protein L appeared less than that of IgG that had the κ-IgL L_C (Fig. 3A). In contrast, IgG with the λ-IgL L_D failed to bind protein L, regardless of its IgH (Fig. 3A).

FIGURE 3.

Nonnative rAbs binding tests. A, Protein L ELISA. Ordinate depicts OD measurements at 450 nm of the ELISA for nonnative rIgG using plates coated with protein L. The abscissa provides the concentration of IgG used in the ELISA in micrograms per milliliter (μg/ml). ELISA data using nonnative rAbs are represented by the symbols indicated in the legend on the upper right margin. B, Protein L immunoblot of rIgG. Recombinant protein L was size separated as in Fig. 2B and probed with various concentrations of different rIgG at 30, 10, and 1 μg, as depicted in Fig. 2B. The different IgG used are indicated above each triangle. C, Competitive ELISA using native rIgG (H_AL_A, H_BL_B, and H_DL_D) or IgG made by 293A cells transfected with fixed amounts of vector encoding IgH HC and IgL L_C together with increasing amounts of vector encoding IgL L_B (e.g., 2, 4, 6, 8, 16, or 24 μg). The number after L_B indicates the micrograms of vector encoding IgL L_B used.

We tested whether the IgL L_B encoded by IGLV3–21 could compete with the IgL L_C encoded by IGKV3–20 to assemble with IgH H_C to form an IgG with functional binding activity for protein L. For this, we transfected 293A cells with increasing amounts of plasmid encoding IgL L_B together with fixed amounts of plasmids encoding IgH H_C and IgL L_C. At the lowest amount of L_B-encoding plasmid used (2 μg), the IgG generated by the transfected 293A cells still bound protein L, albeit with a lower apparent binding activity than IgG composed of H_CL_C (Fig. 3C). Cotransfection of 293A cells with ≥6 μg plasmid encoding the IgL L_B resulted in production of IgG that had binding activity for protein L similar to that of IgG composed of H_AL_A or H_BL_B (Fig. 3C). These results are consistent with the notion that the λ-IgL L_B could compete with K-IgL L_C in vivo to form an intact IgG with binding activity for protein L.

Ab binding affinity measurements

We determined the binding affinity of the rIgG for protein L. We found that IgG composed of H_CL_C had a K_d for protein L of 1.5 ± 0.5 × 10⁻⁷ M (Fig. 4A), which agreed with the results of prior studies using fluorescence measurements (27). In contrast, IgG composed of H_AL_A or H_BL_B each had a K_d for protein L of 2.1 ± 0.5 × 10⁻⁶ or 1.7 ± 0.5 × 10⁻⁶ M, respectively (Fig. 4A).

FIGURE 4.

Binding affinity measurements to protein L. Binding affinity measurements for (A) native rIgG (H_AL_A, H_BL_B, or H_CL_C) and for (B) nonnative rIgG (H_AL_C, H_DL_C, H_CL_A, or H_DL_A). In A and B, the ordinate depicts the percentage of free rIgG, and the abscissa provides the molar (M) concentration of protein L used to determine the K_d of the native or nonnative rIgG. K_d values for each rIgG are indicated in each panel.

We determined the binding affinity for protein L of rIgG composed of nonnative pairs of IgH and IgL (e.g., H_AL_C, H_DL_C, H_CL_A, or H_DL_A). We found that IgG composed of H_AL_C or H_DL_C each had a K_d for protein L comparable with that of IgG composed of H_CL_C (1.7 ± 0.5 × 10⁻⁷ and 1.6 ± 0.5 × 10⁻⁷ M, respectively; Fig. 4B). In contrast, IgG composed of H_CL_A or H_DL_A each had K_d for protein L that was comparable with that of IgG composed of H_AL_A (2.0 ± 0.5 × 10⁻⁶ and 2.3 ± 0.5 × 10⁻⁶ M, respectively; Fig. 4B). We conclude that rIgG composed of either native or nonnative pairs of IgH and IgL can bind protein L with similar binding affinities, which appeared dependent on the type of IgL used.

Ab binding activity for cofilin

We also tested whether the IgG listed in Table I could bind cofilin. Cofilin is an actin- and phosphatidylinositol 4,5-bisphosphate-binding protein that is well conserved and ubiquitous (28). We found that the IgG with native pairs of IgH and IgL (e.g., H_AL_A, H_BL_B, H_CL_C, or H_DL_D) could bind cofilin by immunoblot analysis on lysates of U937 (Fig. 5A). In contrast, the IgG-derived CLL69 C/C (7) did not have detectable binding activity (Fig. 5A). As noted for the Ab binding activity for protein L, the weak binding activity for cofilin of the tested IgG did not appear dependent on having a native pair of IgH and IgL, as IgG composed of nonnative pairs (e.g., H_DL_A, H_AL_D, or H_DL_C) also bound cofilin by immunoblot analyses (Fig. 5B). From these studies, we conclude that the binding activity for cofilin of these IgG could not account for the nonstochastic coexpression of these IgH and IgL by CLL cells.

FIGURE 5.

Ab binding activity for cofilin. Lysates of U937 cells that express nonmuscle type, cofilin 1, were size separated on SDS-PAGE and transferred to nitrocellulose membrane (Invitrogen). The membranes were probed with 30 or 10 μg of different rIgG, H_AL_A, H_CL_C, H_DL_D, or CLL69C/C (A) or H_DL_A, H_AL_D, or H_DL_C (B). The triangles at the top of the gels represent the relative concentration of IgG used. The different IgG are as indicated above each triangle. The last lane of each gel was probed with rabbit anti-cofilin 1 Abs (ECM Biosciences), which served as a positive control (+ α-cofilin).

Discussion

We investigated whether steric factors were responsible for the nonstochastic coexpression of stereotypic IGHV3–21–encoded IgH with certain IgL in CLL. Prior studies identified examples in which some IgH preferentially assembled with certain IgL. Moreover, in some cases, the IgH was physically unable to assemble with any but a few IgL to form an intact Ig (20). As has been evaluated in studies on other types of Abs (29–33), we considered whether similar constraints in IgH/IgL assembly could account for the frequent coexpression of certain IgL with stereotypic IGHV3–21–encoded IgH in CLL. However, our studies failed to identify biochemical constraints in the capacity of stereotypic, IGHV3–21–encoded IgH to assemble with any IgL to form IgG with IgH and IgL that typically are not coexpressed by CLL B cells.

The finding that IGHV3–21–encoded IgH with HCDR3 motif-1 is not precluded from binding different κ-IgL has implications for models proposing prior Ig receptor editing in cases of CLL that use λ-IgL encoded by IGLV3–21 (9). Analyses of the rearranged and nonexpressed κ-IgL locus of such cases revealed evidence for productive rearrangements and somatic mutations in the rearranged κ-IgL V region genes that presumably occurred before secondary rearrangements with kappa deleting element that aborted expression of the κ-IgL. Receptor editing has been found to occur for normal B cells that express autoreactive Ig, allowing such B cells to express Ig with different binding properties that do not lead to clonal deletion or autoimmunity (34). Conceivably, the de novo expression of λ-IgL encoded by IGLV3–21 after such receptor editing generated Ig that had binding properties for some self-Ags or environmental Ag conducive to clonal expansion and/or survival of preleukemia and/or nascent leukemia B cells. In this regard, the nonstochastic association of stereotypic IGHV3–21–encoded IgH with particular IgL appears to be due to selection for B cells that have Ig with the binding activity required for their potential malignant transformation into CLL.

We examined for Ags that could bind the stereotypic IGHV3–21–encoded Ig observed in CLL. We observed that IgGλ composed of H_AL_A or H_BL_B could bind protein L, a cell-wall protein isolated from a commensal Gram-positive bacterium, Peptostreptococcus magnus. Prior studies found protein L could bind human Ig with κ-IgL encoded by IGKV genes of the I, III, and IV subgroups, but not human Ig with l-IgL (25–27). However, another study found that Ig with an IgH such as H_A and an IgL such as L_A (Table I) could react against Gram-positive bacteria (e.g., Streptococcus pyogenes, Enterococcus faecium, and Enterococcus faecalis) (35). Conceivably, these Ig react with proteins similar to that of protein L, which might also be expressed by these bacteria. In any case, because these microorganisms commonly colonize the human mouth, skin, or gastrointestinal tracts, there is prevalent and continuous exposure to the Ags they produce, such as protein L. For this reason, these microbial products are considered good candidate Ags for driving constitutive stimulation and/or proliferation of Ag-specific B cells, which, over time, might undergo neoplastic transformation (35–37).

However, we found that IgG with -IgL encoded by IGLV3–21 bound protein L regardless of the IgH used. This is similar to what we observed for the Ig expressed in CLL that had κ-IgL encoded IGKV3–20. As such, protein L is a superantigen for B cells that express Ig with l-IgL encoded by IGLV3–21 or κ-IgL encoded by IGKVof the VκI, VκIII, or VκIV subgroups (25–27).

Superantigens are substances that bind site(s) in the Ig V region that are outside the conventional Ag-binding region (38). The binding sites for superantigens typically are not influenced by the HCDR3 or the combinatorial diversity afforded by the pairing of IgH and IgL in the intact Ig molecule. Consequently, a superantigen can bind to the Ig made by many different clones of B cells, which collectively have undergone disparate Ig gene rearrangements and that express different Ig chains together with the Ig chain(s) that possesses the superantigen-binding site(s). The fate of B cells that express such Ig can be influenced through interaction with its respective superantigen, which can cause oligoclonal B cell expansion or deletion, depending in part on the relative binding affinity of Ig for the superantigen (38–41). Moreover, superantigens may drive oligoclonal B cell proliferation that might factor in the development of autoimmune disease or B cell leukemogenesis (42–44).

However, selection for superantigen binding cannot account for the nonstochastic coexpression of stereotypic IgH with certain IgL in CLL unless the irrelevant Ig chain somehow influences the relative binding activity of the intact Ig for its superantigen. We considered this possibility and hence studied the binding affinities for protein L of the stereotypic IGHV3–21 Ig expressed in CLL. We found that the binding affinity for protein L by IgG with an IgH encoded by IGHV3–21 and an IgL encoded by IGKV3–20 was similar to that noted in prior studies on IgGκ made by normal, nonleukemia B cells (27). Furthermore, we found that IgG composed of the stereotypic IGHV3–21–encoded IgH paired with a λ-IgL encoded by IGLV3–21 had binding affinity for protein L (K_d, ≈2 × 10⁻⁶ M) that was ~ 10-fold lower than that noted for such IgGk, including Ig with the stereotypic κ-IgL encoded by IGKV3–20 that was expressed in CLL. Moreover, the binding affinity for protein L of IgGI with λ-IgL encoded by IGLV3–21 was not affected by use of different IgH. As such, we conclude that stimulation by the protein L superantigen cannot account for the non-stochastic coexpression of stereotypic IGHV3–21–encoded IgH with certain IgL in CLL.

A prior study showed that the stereotypic IGHV3–21–encoded Ig expressed in CLL, such as the Ig represented by H_AL_A in this study (Table I), also could bind cofilin (36, 37). Cofilin is a family of three highly conserved and ubiquitous actin-binding proteins that disassemble actin filaments, can facilitate cellular entry of Gram-positive bacteria into nonphagocytic mammalian cells, and potentially play a role in neurodegenerative diseases associated with aging (45–47). Because it is a ubiquitous self-Ag that might have altered expression or distribution in conditions associated with aging, cofilin is an attractive candidate Ag for driving B cell proliferation and potential neoplastic transformation leading to CLL. In this study, we confirmed that the stereotypic IGHV3–21–encoded Ig expressed by CLL cells could bind cofilin by immunoblot analyses. Moreover, we found that IgG with IgH and IgL respectively encoded by IGHV4–39 and IGLV1–40 also could bind to this self-Ag. However, as with our studies on the interactions of such Ig with protein L, we found the detected binding for cofilin was not restricted to Ig having the native pair of IgH and IgL that typically are expressed together in CLL. As such, there are no data to support the notion that cofilin can select for B cells that use the stereotypic Ig expressed in CLL.

This study provides an important caveat to the conclusions reached from studies examining the binding activity of the Abs expressed in CLL for self-Ags or environmental Ags. Indeed, the observed binding activities of Ig for such Ags might be fortuitous and may not account for the apparent selection of CLL B cells that express such Abs. Future studies on the binding properties of the Ig used by CLL B cells should take into account whether the binding for Ag is affected by the combinatorial diversity resulting from the pairing of disparate IgH and IgL, which also appears restricted in the repertoire of Abs expressed in this disease.

Abbreviations used in this article

CLL

chronic lymphocytic leukemia

HCDR3

H chain CDR3

IGHV

IgH V region gene

IgL

Ig L chains

K_d

equilibrium dissociation constant

PtC

phosphatidyl choline