Loss of a chromosomal region with synteny to human 13q14 occurs in mouse chronic lymphocytic leukemia that originates from early-generated B1 B cells

Abstract

A common feature of B cell chronic lymphocytic leukemia (CLL) is chromosomal loss of 13q14, containing the miR15a/16-1 locus controlling B cell proliferation. However, CLL etiology remains unclear. CLL is an adult leukemia with an incidence that increases with advancing age. A unique feature of CLL is biased B cell antigen receptor (BCR) usage, autoreactivity with polyreactivity, and CD5 expression, all suggest a role for the BCR in driving CLL pathogenesis. Among human CLLs, BCRs autoreactive with non-muscle myosin IIA (AMyIIA) are recurrent. Here we identify an unmutated AMyIIA BCR in mouse, with distinctive CDR3 segments capable of promoting leukemogenesis. B cells with this AMyIIA BCR are generated by BCR-dependent signaling during B-1 fetal/neonatal development with CD5 induction, but not in adults. These early-generated AMyIIA B1 B cells self-renew, increase during aging, and can progress to become monoclonal B cell lymphocytosis, followed by aggressive CLL in aged mice, often with loss of a chromosomal region containing the miR15a/16-1 locus of varying length, as in human CLL. Thus, the ability to generate this defined autoreactive BCR by B1 B cells is a key predisposing step in mice, promoting progression to chronic leukemia.

INTRODUCTION

A critical role for the BCR in development of CLL has been hypothesized, based on findings of biased immunoglobulin variable (V) region gene usage^{1, 2}. Approximately half of CLLs express unmutated BCRs, identifying cases with a more aggressive course compared to those bearing mutated BCRs^{3, 4}. These unmutated BCRs in CLL have been shown to be autoreactive and polyreactive, showing cross-reactivity to bacteria and/or viruses^{5, 6}. One clear example of autoreactivity by CLL is recognition of non-muscle myosin IIA by unmutated BCRs utilizing nearly identical V_H1-69/D3-16/J3 IgH paired with IgKV3-20 IgL⁷ found in ~1% of CLL patients⁸. In addition to binding intracellular non-muscle myosin IIA, this BCR also binds apoptotic cell determinants, where intracellular/nuclear components, including myosin IIA, are exposed outside the cell membrane as autoantigen-bearing blebs^{7, 9}. This suggests that B cells with this BCR provide the initial recognition of apoptotic cells^{9, 10}. These findings prompted the proposal that the initial step in CLL may be the generation of autoantigen-experienced B cells^{11, 12} bearing polyreactive unmutated BCRs.

In normal mice, generation of CD5⁺ B cells, termed B1a cells, occurs as the outcome of relatively strong BCR signaling induced by (self)-ligand exposure^13–15. Such BCR signal-dependent B1a cell generation is the predominant outcome of B-1 development that occurs in fetal/neonatal B lineage precursors expressing Lin28b and lacking miR Let-7, as the progeny of fetal hematopoietic stem cells. In contrast, adult bone marrow (BM) B lineage precursors do not express Lin28b and are Let-7⁺ resulting in a switch to B-2 development that predominantly yields CD5⁻ B cells ^16–18. After birth, the production of B1a cells declines; however, a fraction of B cells generated during fetal/neonatal B-1 development persists as a minor B cell subset that is maintained by self-renewal throughout life^{19, 20} as B1 B cells. Based on their autoreactivity and expression of CD5, B-1 derived B1 B cells have been suggested to have a propensity for leukemic progression. In order to test this idea, we first identified a recurrent BCR with non-muscle myosin IIA autoreactivity among CD5⁺ B cells that progressed to CLL, promoted by expression of the Eμ-hTCL1 transgene²¹. By establishing a set of BCR transgenic/knock-in mouse models, we demonstrate that B cell generation with this distinctive autoreactive BCR, having unique CDR3s, is restricted to B-1 development and poses a significant risk for progression to aggressive CLL/lymphoma. CLLs utilizing this BCR often show monoallelic loss of a region of mouse chromosome 14 that includes the miR15a/16-1 cluster, resembling human CLL.

MATERIALS AND METHODS

Mice

Eμ-hTCL1 Tg mice were backcrossed onto the C.B17 background. To establish the V_HQ52 VDJ knock-in line ON25, the V_HQ52 IgH-μ transgenic mouse line OK44, and the V_k9-96 kappa (IgL) transgenic line OW26, heavy and light chains were cloned from the V_HQ52/V_k9 hybridoma, 14-1H3. A detailed procedure to generate the zinc finger nuclease knock-in mouse line ON25 is described in Supplemental Information. In brief, as shown in Figure 2c, RNA coding for two pairs of Fok I heterodimeric ZFNs cutting the mouse Ig heavy chain locus in J_H1 and just downstream of J_H4 was injected into oocytes, together with a “donor” DNA segment containing the V_HQ52/D/J_H4 segment, with “arms” extending outside the ZFN target sites, facilitating homologous recombination into the J_H region. To generate the V_HQ52/D/J_H4-μ transgenic mouse line OK44, the rearrangement was cloned from hybridoma 14-1H3 DNA by long-PCR using a primer upstream of the V_H promoter region (identified from a database search) and a reverse primer downstream of the J_H4 segment. The promoter-V_HQ52/D/J_H4 segment was inserted into a Cμ vector previously used for generating heavy chain transgenic mice¹⁴. To generate the V_k9-96/J_k1-κ transgenic mouse line OW26, the kappa rearrangement was cloned from 14-1H3 DNA by long-PCR using a primer sequence upstream of the Vκ promoter region and an antisense primer downstream of the Cκ segment. In all cases, PCR-amplified coding segments were verified as unmutated. All knock-in and transgenic mouse lines (IgH μ^a) were crossed onto a μ^b C.B17 background. Animal experiments were conducted under a protocol approved by the FCCC Institutional Animal Care and Use Committee (IACUC).

B cell leukemia/lymphoma analysis, diagnosis, and purification

MBL and CLL stages were identified by analysis of PBL (total cell number and AMyIIA B cell frequency), performed every 2–3 months for each mouse. A predominance of AMyIIA BCR B cells (>70% of B cells) in PBL, without a significant increase in total PBL number was defined as MBL. For gene expression analysis, immunoglobulin sequencing, and western blotting, leukemia/lymphoma B cells (all IgM⁺) were purified as CD19⁺B220^loCD5^+/lo by flow cytometry. For AMyIIA CLL, B220^loCD5^+/lo24E1id^hi B cells were purified by sorting. Histologic diagnoses were made using formalin fixed paraffin embedded sections stained with H&E using established criteria²².

Flow cytometry analysis and anti-idiotype antibody

Multicolor flow cytometry analysis, sorting, and monoclonal antibody reagents have been described previously¹⁵, and also listed in Supplemental information. Rat anti-idiotype monoclonal IgG antibodies produced by hybridomas generated by immunizing rats with 14-1H3 IgM were made as described previously (Supplemental Information).

Immunoglobulin sequencing

Sequencing Igμ and Igκ variable regions of tumor B cell BCRs by μ or κ PCR with consensus V primers was carried out using standard procedures in most cases. When sequences were not recovered, 5′ RACE cloning was done using the SMARTer^™ RACE cDNA amplification Kit (Clontech). For single cell sequencing, individual cells were deposited using a FACSAriaII (Becton Dickinson), directly onto either AmpliGrid AG480F slides (Beckman Coulter) or onto 96 well plates prepared with SuperScript III RT Kit (Life Technologies). For details, see Supplemental Information.

Mouse V gene nomenclature

The V_H gene family name is based on Johnston et. al.²³ and V_H gene classification is based on identity with genes in Vbase2 (www.Vbase2.org). AMyIIA V_HQ52 has 100% identity with V222 from C57BL/6, closely related to the V_HOx-1 gene from BALB/c (V171) with some allelic differences ²⁴. V_k gene nomenclature is based on Thiebe et. al. ²⁵. V_k9-96 is ce9.

Comparative genomic hybridization (CGH) analysis

The Agilent DNA Microarray platform was employed for Array-CGH studies using the Agilent Mouse Genome CGH Microarray 244A chip with about 235,000 mouse sequences. Procedures were performed in the Fox Chase Cancer Center Genomics Facility. Data was analyzed using BioDiscovery Nexus Copy Number software.

Tissue section and cell line immunofluorescence, and flow cytometry analysis

Rag1^−/− mice were used for producing ethanol-fixed frozen tissue sections, and microscope imaging was done as described previously^{26, 27}. 1/100 diluted hybridoma IgM ascites (2–10 μg/ml) were incubated on tissue section slides for 90 minutes at 37°C, followed by AlexaFluor (AF)⁴⁸⁸ anti-IgM (331.12), together with DAPI (Thermo Scientific), with or without PE-anti-CD31 (eBioscience) (30 minutes at RT). Confocal microscopy was used for analysis of cell lines. MitoTracker Red CMXRos (Invitrogen) treatment was done in culture for 30 minutes before fixation and IgM staining. AF⁵⁴⁶ phalloidin was from Invitrogen.

Western Blotting

The OP9 stromal cell line (with and without siRNA treatment) was used for 14-1H3 IgM immunoblotting (1/100 IgM ascites) of 5% SDS-PAGE. Anti-myosin IIa, anti-myosin IIb, and HRP goat anti-rabbit IgG antibodies were all from Cell Signaling Technologies (CST). HRP-goat anti-mouse IgM was from Santa Cruz Biotech. Rabbit anti-β actin was from Bethyl Labs. For siRNA treatment, siRNA Myh9 (M-040013), siRNA Myh10 (M-062322), or non-targeting siRNA (D-001206), was mixed with DarmaFECT-1 (T-2001) to generate transfection media (all from Thermo Scientific, Dharmacon RNAi Technologies).

Quantitative RT-PCR assay

Gene expression was quantitated by real-time PCR, using TaqMan assays from Applied Biosystems, an ABI 7500 real-time thermal cycler, and ABI software (Life Technologies).

Heavy chain - surrogate light chain (SLC) association analysis

IgH-pMIG (MSCV-IRES-GFP) retroviral supernatants were generated as described in Supplemental Information. To examine SLC association, the Pro-B Abelson line N38 was transduced with IgH-pMIG retroviral supernatant. In GFP⁺ cells, IgH-μ transduction was examined by intracellular IgM staining, and extent of SLC assembly was examined by surface SLC staining, using the conformation-dependent anti-surrogate light chain antibody SL156²⁸. To test the capacity for pre-BCR-mediated proliferation, IgH-pMIG retroviral supernatant was added to Pro-B cell cultures of Rag1^−/−.BALB/c bone marrow on OP9 stromal cells together with IL-7. For details, see Supplemental Information.

B1a cell and bone marrow cell co-transfer

1–1.5 ×10⁶ purified B1a cells (B220^loCD5⁺) from spleen (or peritoneal cavity) of ≤2 month (10–60 day) Eμ-TCL1 Tg⁺ C.B17 mice (IgM^b allele) and 5 ×10⁶ total bone marrow cells from 2 month old TCL1⁻ BALB/c (IgM^a allele) were co-injected intravenously per C.B17 scid recipient (lightly irradiated, 3 Gy, 1 day prior), followed by PBL analysis, and CLL/lymphoma BCR sequence analysis.

RESULTS

Stereotyped V_HQ52/V_k9 BCR in mouse CLL with deletion of a 14q chromosomal region containing miR15a/16-1 locus binds non-muscle myosin IIA

In mice, transgenic expression of the human T-cell leukemia 1 oncogene in B lineage cells (Eμ-hTCL1) results in a high incidence of aggressive CD5⁺ CLL with a biased utilization of unmutated BCRs^{21, 29}. Among these BCRs on a C.B17 mouse background, we found that a V_HQ52 (V_HOx-1, V222)/V_k9 (V_k9-96) represented 9% of the total (14/155), and was a particularly restricted BCR with identical CDR3 and J segments (Figure 1a). Different from other TCL1⁺ (TC⁺) CLLs, this TC⁺ CLL often appeared as disseminated tumors in subcutaneous tissues (hypodermis), particularly in the cervical area including the salivary glands, resulting in a swollen neck (Figure 1b). Comparative genomic hybridization (CGH) analysis revealed that 11/14 cases showed variable length hemizygous deletions in 14q that included the miR15a/16-1 locus³⁰, syntenic to the 13q14 loss found in human CLL³¹ (Figure 1c).

Figure 1. Stereotyped V_HQ52/V_k9 BCR in 14q⁻ CLL binds non-muscle myosin IIA.

a. V_HQ52/V_k9 stereotyped BCR IgH and IgL CDR3 region amino acid residues and nucleic acid sequences. For IgH, two different D_H segment yield identical amino acid sequences. P-nucleotide in italic. V222 gene in Vbase2 (www.Vbase2.org). b. Subcutaneous (hypodermis) infiltration by V_HQ52/V_k9 BCR CLL with a CD19⁺B220^loCD5⁺ phenotype. Gross presentation and H&E staining and flow cytometry analysis of skin. Dotted squares are negative region. c. Summary of CGH data for chromosome 14 deletions in the 14qB-E4 region in V_HQ52/V_k9 BCR TC⁺ CLLs. The miR15a/16-1 locus is marked. d. Cryosection staining of Rag^−/− mouse heart. 14-1H3 IgM staining alone or merged with anti-CD31 and DAPI; 54-2F3 (V_HJ558 IgM) as control. cm = cardiac muscle; v = ventricle. Bars=100 μm e. Confocal image of NIH-3T3 cell line stained with 14-1H3 together with either phalloidin (1) or MitoTracker (2), and DAPI. Bars = 20 μm. f. 14-1H3 IgM (or control IgM) staining of OP9 stromal line, together with DAPI. Western blot of OP9 with anti-myosin IIA (plus anti-rabbit IgG), 14-1H3 IgM or control IgM (plus anti-IgM). g. Western blot of OP9 cell extracts with or without siRNA treatment. Scramb siRNA = scrambled sequence siRNA control.

To assess potential autoreactivity by this BCR, hybridomas were generated from CLL B cells. A V_HQ52/V_k9 IgM was produced from hybridoma 14-1H3, and an IgM from hybridoma 54-2F3 as a control. Immunofluorescence screening of cryosections from mouse heart, lung, spleen, and liver revealed strong binding by 14-1H3 IgM, specifically to blood clots in unperfused heart chambers (Figure 1d). Blood clots were CD31/PECAM-1⁺ with aggregated platelets. High reactivity to platelets raised the possibility of binding to non-muscle myosin IIA³². Confocal staining of the NIH3T3 mouse fibroblast cell line showed that 14-1H3 staining co-localized with phalloidin-stained F-actin filaments, the location of non-muscle myosin II, but not with mitochondria (Figure 1e).

Areas of actin filaments in the stromal cell line OP9 were also stained by 14-1H3 (Figure 1f). Western blot analysis of OP9 cell extracts showed that 14-1H3 binds to a protein of approximately 230 kDa, consistent with the size of non-muscle myosin II. This reactivity was abrogated by treatment with siRNA specific for the Myh9 gene that encodes myosin IIA, but not by siRNA for the Myh10 gene that encodes myosin IIB. This confirmed that the V_HQ52/V_k9 IgM, 14-1H3, binds non-muscle myosin IIA (MyIIA) (Figure 1g). MyIIA is known to be present in most non-muscle eukaryotic cells, but also in muscle cells³³. In addition to reactivity to platelets, fibroblasts, and stromal cells, 14-1H3 also showed binding to MyIIA in a mouse B lymphoma, a neuroblast cell line, and intestinal epithelium (Supplementary Figure S1). Thus, this stereotyped BCR is autoreactive for MyIIA (AMyIIA) present in a variety of cell types.

The AMyIIA BCR B cells generated in the V_HQ52/D/J knock-in mouse are B-1 derived

The AMyIIA IgH V sequence was consistent with a fetal/neonatal origin, based on lack of N-addition at the V-D-J junctions, except for a p (palindrome)-nucleotide (Figure 1a). TdT (terminal deoxynucleotidyltransferase), an enzyme that mediates N-addition, is low/absent in B lineage cells from mouse fetal liver³⁴. Furthermore, this V_HQ52 IgH failed to associate with surrogate light chain (SLC) (Figure 2a), leading to its inability to drive pre-BCR clonal expansion in bone marrow (BM) (Figure 2b). In adult B-2 development in BM, SLC assembles with IgH chain to generate a pre-BCR resulting in proliferation. However, for fetal B-1 development, strong IgH-SLC association is not required³⁵, suggesting a preference for B-1 development. Therefore, to construct a mouse model ensuring physiologic expression of this AMyIIA BCR, we replaced the endogenous J_H locus with this V_HQ52/D/J using a zinc finger nuclease (ZFN)-mediated approach (Figure 2c, details in Supplemental information). This knock-in procedure allowed us to develop a strain, ON25, containing the VDJ J_H insertion. B cells expressing this VDJ were identified by PBL staining with a V_HQ52 idiotype-specific antibody (Figure 2d).

As in human cord blood, CD5⁺ B cells are present in early in life in normal mice^{13, 36, 37}. One day after birth in ON25 mice (heterozygous knock-in), nearly all newly generated immature B cells in the liver expressed the V_HQ52id⁺ μ heavy chain, most with a κ light chain (Figure 2e top). AMyIIA BCR B cells already constituted 15–20% of the immature B cells stained by a V_HQ52/V_k9 idiotype-specific antibody, 24E1. Single cell Igκ sequence analyses confirmed that V_k9 light chain was identical to that utilized in the AMyIIA BCR, specifically in cells showing higher 24E1/κ binding ratios (24E1^hi) (Figure 2e, right top, red oval). CD5 induction had already occurred at this immature stage (AA4/CD93⁺) during IgD upregulation, and CD5⁺ immature AMyIIA B cells were present in both liver and spleen (Figure 2f, d1 upper panels). Maturation continued, and at 12 days after birth, AMyIIA B cells were the predominant subset among mature (AA4^lo/−) B220^lo B cells in both the spleen and peritoneal cavity (PerC), with further CD5 upregulation (Figure 2f, d12, lower panels). Thus, expression of V_HQ52/D/J IgH during fetal/neonatal B-1 development resulted primarily in the generation of AMyIIA mature B1a cells, as autoreactive B cells. CD5⁺ AMyIIA B cells also distributed to the intestine, including the colon, before B-2 derived B cells begin to accumulate (Figure 2g and Supplementary Figure S2). In contrast, by 2 months after birth, the frequency of V_HQ52id⁺ cells among newly-formed B cells in BM was reduced to ~40–50%, and production of AMyIIA immature B cells was significantly reduced (<2%) (Figure 2e, middle).

Figure 2. Predominant AMyIIA B1a cells from B-1 development in the ON25 knock-in line.

a. Inability of AMyIIA V_HQ52 IgH to associate with surrogate light chain (SLC). A pro-B cell line was transduced with IgH-pMIG retroviral supernatants (GFP⁺). Intracellular IgM level (top) and surface SLC level (bottom) of GFP⁺ cells. V_HJ558 (V627) IgH is a positive control. pMIG alone is in gray. b. BM preBCR-mediated proliferation. B cell numbers from pMIG alone (−) set to 1, and V_HJ558-μ to 10. n = 3 c. Schema for zinc finger nuclease (ZFN) generation of a V_HQ52/D/J_H4 knock-in mouse. d. 5 week old PBL with or without V_HQ52 V/D/J_H4 knock-in. e. Comparison of immature (AA4/CD93⁺) kappa⁺ B cells from day 1 liver with those from BM of 2 month old ON25 and wild type mice. Percentages of 13F11⁺ or 24E1^hi+ (red ellipse region) among kappa⁺ cells. f. Day 1 and 12 ON25 B cell analysis. Gated for CD5⁺ B (marked in red) for 24E1^hi+ cell percentages, IgM/IgD, and AA4 levels, in comparison with CD5⁻ B cells. g. Day 14 ON25 mouse intestine (Peyer’s patch in jejunum, and lymphoid cells in colon), spleen, and peritoneal cavity. Percentage of CD5⁺ B among total lymphoid cells (in figure), and 24E1^hi+ B among CD5⁺ B cell and out of total B cells. Representative of 4 mice.

However, 24E1^hi AMyIIA B cells continued to predominate in the mature B220^loCD5⁺ B1a pool in adult, with an IgM⁺IgD^med/lo phenotype in both spleen and PerC, accompanied by some B220^loCD5⁻ B cells (B1b) (Figure 3a). In spleen, 24E1^hi cells (some CD23⁺) were not present in mature follicular B (FO B) and marginal zone B (MZ B) subsets (Figure 3b). Few 24E1⁺ cells detected in MZ B and FO B cells were 24E1^med cells (Figure 3c); single cell Igκ sequence analysis revealed that 24E1^med cells utilized different V_k9-J_k junctional residues (Figure 3c). The mouse MZ B cell pool is known to contain autoreactive B cells^{26, 38}, as with B1 B cells. To assess whether the MZ B cell can include the stereotyped AMyIIA BCR, a V_HQ52/D/J-μ^a IgH transgenic mouse line, OK44, was generated as previously described¹⁴ (Supplementary Figure S3). Due to high-level transgenic IgH-μ expression, most newly-formed immature B cells in adult BM were V_HQ52id⁺ μ, and 30–50% of MZ B cells were 24E1⁺. However, these cells were 24E1^med, and all utilized a proline (P) at the light chain V-J junction, as in ON25 MZ B. Thus, we conclude that B cells with this AMyIIA BCR are predominantly generated during early B-1 development and reside in the B1 B cell pool in adult animals.

Figure 3. AMyIIA B cells are predominant in B1 B cells, but not in follicular B and marginal zone B cell subsets, in adult ON25 mice.

a. 2 month ON25 spleen and peritoneal cavity. On the left: total lymphoid cells and 24E1⁺ cell percentage in B1a (+B1b) cells. On the right: analysis of total B and 24E1⁺ B cells. The region marked in red for 24E1^hi+ cell percentages. b. AA4⁻ mature B cell subsets identified by CD21/CD23 staining and comparison of 24E1^hi+ versus 24E1 negative in adult ON25 spleen. 24E1^hi+ cells (red ellipse) and 24E1⁻ kappa⁺ cells (black box). CD21⁻CD23⁻ (some CD23⁺) mature 24E1^hi cells are predominantly CD5⁺ (as B1a, see Supplementary Figure S3). c. Amino acid differences at the V_k9/J_k junction in 24E1⁺ cells from mature B cell subsets. Percentages of 24E1⁺ cells in MZ B and FO B are marked by a vertical line, compared with B1 B staining pattern. Numbers of individual cells sequenced for each subset: B1 B, 20; MZ B, 36; FO B, 25.

AMyIIA BCR B cells can progress to aggressive CLL/lymphoma in aged animals

To assess the fate of B-1 derived AMyIIA B cells, ON25 (and OK44) mice were crossed with hTCL1 transgenic mice to generate TC⁻ and TC⁺ progeny, and data are summarized in Table 1. Analyses of ON25 mice at 6–8 months showed that the TC⁺ transgene promoted an increase in B220^loCD5⁺ B cells (B1a) in PBL with the AMyIIA BCR (Figure 4a and 4b, TC⁺). 100% (10/10) of TC⁺ ON25 mice developed CD5⁺ CLL by 8–10 months. In TC⁻ littermates, circulating B1a cells were few but were comprised primarily of 24E1⁺B220^lo AMyIIA B cells (Figure 4a, TC⁻), and gradually increased with age (Figure 4c). 34% of the TC⁻ mice developed a monoclonal B lymphocytosis (MBL) when older than 15 months, and 13% (4/32) developed CLL (Table 1), with or without splenomegaly, exemplified by the cases shown in Figure 4e. OK44 mice showed a similar outcome (Table 1). Thus, even without the TCL1 transgene, we found progression to CLL, promoted by the novel AMyIIA BCR.

Table 1.

AMyIIA B cell MBL and CLL/lymphoma incidence in ON25 knock-in and OK44 transgenic mouse lines.

AMyIIA B	TC+ 8–10 mo.	TC− > 15 mo.
VHQ52-JH4 Ki (“ON25”)
MBL	–	34% (11/32)
CLL/Lymphoma	100% (10/10)	13% (4/32)
VHQ52-JH4 μTg (“OK44”)
MBL	–	39% (13/33)
CLL/Lymphoma	100% (15/15)	21% (7/33)

Both V_HQ52-J_H4 ON25 and OK44 mouse lines showed 100% CLL/Lymphoma development promoted by TCL1 transgene (TC⁺) at 8–10 months of age. Without the TCL1 transgene (TC⁻), AMyIIA B cells could become MBL or progress to become CLL/lymphoma in older animals (>15 months).

Figure 4. Development of aggressive AMyIIA BCR CLL/lymphoma in aged mice.

a. Analyses of PBL from 6 to 8 month old TC^−/− or TC^+/− ON25 littermates. Center and lower panels show predominance of 24E1^hi+ cells in 24E1⁺B220^lo B cells. b. AMyIIA CLL in 6 to 8 month TC⁺ ON25 mice. Percentages of total B cells (left) and AMyIIA B cells among total B cells (right). c. Increased AMyIIA B cell frequency in aging TC⁻ ON25 mice. Average percentages indicated by lines. d. CGH analyses of AMyIIA CLLs with chromosome 14 deletions; data from TC⁺ (1–3, OK44; 4–5, ON25) and TC⁻ (6–8, OK44; 9, ON25 #85) mice. e. 1) TC⁻ ON25 mouse #85 before the tumor stage assessed by PBL and at the CLL stage. Gross findings at necropsy at 21 months; H&E staining of AMyIIA B cell infiltrated areas (marked). Flow cytometry of cells recovered from subcutaneous infiltrates. 2) TC⁻ ON25 mouse #24 before the tumor stage and at the CLL stage with splenomegaly and lymphadenopathy. Before the tumor stage, AMyIIA B cells were CD5^lo+ in PBL (marked by triangle), followed by further down regulation of both B220 and CD5 (red triangle). 24E1⁺ AMyIIA cell and total B cell percentages are shown.

The incidence of 14q deletion (hemizygous) occurred at the MBL stage (data not shown), and in CLL, 75% (6/8) in TC⁺ and 50% (4/8) in TC⁻, with subcutaneous infiltration, as originally described in Figure 1. This 14q deletion of the miR15a/16-1 locus of varying length often included the Rb1 gene at 14qD3 (Figure 4d) as found in the human aggressive CLL subtype³⁹. TC⁻ ON25 sample #85 (Figure 4e, group 1) is a case of 14q deletion (Figure 4d, TC⁻ 9, and Supplementary Figure S4). Following an increase in AMyIIA B cell frequency with a further reduction of B220 level in PBL, this ON25 mouse developed leukemia at 21 months. In addition to leukemic cells circulating through bone marrow, such cells were also predominant in mesenteric lymph nodes, peritoneal cavity (Supplementary Figure S5), and infiltrated the salivary glands and subcutaneous tissues. Mice with TC⁻ AMyIIA CLLs (and MBLs) also exhibited splenomegaly and lymphadenopathy, with or without subcutaneous infiltration, as exemplified by case #24 without 14q loss (Figure 4e, group 2). These TC⁻ leukemia/lymphoma cases often (60%, 20/33) exhibited eventual loss of CD5, without mutation of the BCR, as has been found in some cases of human CLL.

Sharp downregulation of B220 is due to further altered glycosylation of CD45, without loss of CD45 membrane expression, including CD45 isoform (Figure 5a, i). In addition to this B220^low/− phenotype as seen in human CLL and mantle cell lymphoma⁴⁰, retention of high-level CD38, increased CD43, and decreased CD62L, were also common in mouse AMyIIA B CLL/lymphomas (Figure 5a, ii). ZAP70 expression was also observed in mice. At the CLL stage, peritoneal cavity had increased numbers of B cells in both TC⁻ and TC⁺ mice, due to leukemic cell deposition from vessels (Figure 5b, left). Some of these leukemic AMyIIA B cells in the peritoneal cavity (p) showed ZAP70 mRNA levels, often higher than in spleen, as with the TC⁻ #85 case (Figure 5b, marked red). Thus, early generation of B cells with this AMyIIA BCR from B-1 development enhanced the risk for progression to aggressive CLL/lymphoma, and greatly increased the likelihood of loss of the miR15a/16-1 locus region, resembling human CLL. Addition of the Eμ-hTCL1 accelerated this CLL development process.

Figure 5. Mouse AMyIIA CLL resembling aggressive human CLL.

a. Comparison of surface phenotype of AMyIIA B cells (and non-AMyIIA FO B) at the young adult age (2 month) and at the tumor stage. b. Left panel: increased AMyIIA B cells in the peritoneal cavity (p) at the tumor stage in TC⁻ and TC⁺ ON25/OK44 mice. Right panel: ZAP70 expression determined by qPCR in AMyIIA B cells from spleen (s) and peritoneal cavity (p), before or after the tumor stage. Relative mRNA transcript levels with FO B set to 1.0.

Further restriction of HCDR3 in B1 B cells is required for progression to AMyIIA B-CLL

It remained unclear how frequent this BCR is present in the normal mouse fetal/neonatal derived B cell repertoire, without V_HQ52/D/J transgenesis. As shown in Figure 6a and 6b, CLLs with this stereotyped BCR developed following transfer of purified B1a cells from young (≤ 2 month) animals, such as 1 month TC⁺ mouse spleen, without BCR transgenesis. This confirmed that precursors for AMyIIA CLL are present among B1a cells in young mice. However, it was difficult to detect B1a with this BCR in young mice, indicating its low frequency. Thus, to pursue this issue, in particular to determine how restricted the heavy chain needs to be for CLL promotion, we developed a transgenic line expressing the AMyIIA V_k9 IgL, OW26, and evaluated IgH usage in 24E1⁺ cells by single cell sequence analysis (Figure 6c). One day after birth, 0.8–1.0% of V_k9 IgL⁺ B1a cells in liver were 24E1⁺ V_HQ52/D/J_H4, closely resembling the AMyIIA IgH chain; however, none contained the stereotyped HCDR3 (0/16), based on the limited number of V_HQ52/D/J_H4-V_k9 B cell analyzed. Two weeks after birth, as B1a cells increased (Figure 6c, right), V_HQ52/D/J_H4 rearrangements continued to show diverse CDR3 sequences (Supplementary Table S1); 30% of 24E1⁺ cells in PerC B1a expressed the V_HQ52/D/J_H4 heavy chain CDR3 with either LRR or LR, but not RLLR. However, the stereotyped BCR was detected among splenic 24E1⁺ B1a cells (1/22, 4.5%) at 2 weeks. At 6 months of age, B cells with this AMyIIA BCR increased in the B1a pool in spleen and PerC (data not shown), and also in circulating PBL (4/28, 10.5%) (Figure 6d, TC⁻). As found in ON25 mice, expression of the TC⁺ resulted in an increased pool of B1a cells in PBL, in which all 24E1⁺ B cells showed the stereotyped HCDR3 RLLR (Figure 6d, TC⁺), before the appearance of leukemia. These findings confirmed that rare AMyIIA stereotyped B cells were generated in neonatal/young mice in the B1a pool, that their frequency increased with age, becoming MBL, and that such cells were associated with an increased risk of progression to CLL. The TCL1 transgene promoted this leukemic process (Figure 6e).

Figure 6. Early generated B1a contain AMyIIA B cells with a restricted CDR3 that progress to AMyIIA CLL.

a. Stereotyped AMyIIA BCR CLL occurrence in aged recipient mice, transferred with B1a cells from spleen of young (1 month) TC⁺ C.B17 mice (IgM^b+) together with TC⁻ BALB/c BM cells (IgM^a+). Red arrow at 12 months indicates CLL of IgM^b origin. b. AMyIIA BCR CLL occurred in 2/20 CLL cases generated in transfer of young (≤ 2 month) B1a cells (from spleen or PerC). c. Incidence of AMyIIA B cells among B1a cells (but not in CD5⁻ B cells) in d14 V_k9-96/J_k1 IgL OW26 transgenic mice. Percentage of common IgH CDR3s expressed at the V_HQ52/D/J_H4 junction by 24E1⁺ cells (of 22 in spleen and 69 in PerC B1a cells). See Supplementary Table S1 for complete list of HCDR3s. Data are representative of 2 mice. d. 24E1⁺ B1a IgH sequence data from PBL of 6 month TC^−/− and TC^+/− OW26 littermates; RLLR CDR3: TC⁻ (4/28 =10.5%), TC⁺ (41/41=100%). Data are representative of 3 mice of each type. e. Summary: AMyIIA B cells with a restricted CDR3 in B1a generated in young mice become MBL and CLL/lymphoma during aging. Overexpression of TCL1 promotes this process.

Importantly, analysis of OW26 V_k9 Tg mice revealed that none of the CD5⁺ 24E1⁺ B1a cells with V_HQ52/D/J_H4 utilizing different CDR3s progressed to CLL, even in TC⁺ mice without BCR transgenesis (where this BCR was originally identified as a “stereotyped” V_H/V_L combination). Thus, not only was the variable region restricted, but there was a further key requirement for specific heavy and light chain CDR3s to generate AMyIIA CLLs from the B-1 derived B cell pool. This indicates multiple steps of BCR signaling are involved in increasing the risk of leukemia/lymphoma development (Figure 7). B1a B cell generation from fetal B-1 development is the first step involving selection by BCR signaling. Further BCR selection/restriction occurs during the self-renewal and proliferation of B1 B cells, leading to CLL with a stereotyped unmutated BCR.

Figure 7. Multistep BCR selection in generation of CLL/lymphoma that express unmutated BCRs, with or without loss of 14q chromosome region with synteny to human 13q14.

Multistep BCR selection in generation of CLL/lymphoma. BCR selection I: appropriate BCR ligand (including self-antigen)-mediated signaling allows maturation and results in CD5⁺ B cell (B1a) generation with biased BCR usage from fetal/neonatal B-1 development, as the progeny of mouse fetal hematopoietic stem cells. BCR selection II: their self-renewal/maintenance further selects BCRs (blue and red BCRs), and sensitivity to transformation requires a yet more restricted BCR (red BCR). Such CLLs, progressed to an aggressive stage, continue to express unmutated BCRs, with or without 14q chromosomal loss.

DISCUSSION

The origin of aggressive CLLs bearing unmutated BCRs has been long debated¹¹. In humans, mature CD5⁺ B cells in healthy adults show enrichment of stereotyped V gene rearrangements, and the highest concordance in gene expression to CLL compared with other B subsets, lending support to the idea that CD5⁺ B cells may be precursors for CLL⁴¹. Here, in mice, we demonstrate that the life-long maintained early-generated B1 B cell is one source of these CLLs. AMyIIA CLL incidence of B1 B origin resembles human aggressive CLL, accompanied by occurrence of deletion of the 14q locus, as found in human CLLs (13q14 deletion). CD5 induction on B cells is a common outcome of mouse perinatal B-1 development that utilizes a BCR ligand-mediated signal for maturation, often by autoreactivity, thus resulting in a pool of cells with biased BCR usage and a B220^loCD5⁺ phenotype as B1a. Importantly, our analysis of AMyIIA B cells revealed that not all B1a cells progress stochastically to CLL; rather, a further BCR restriction emerged from our analysis. Clearly, CDR3 structure plays an important role in this AMyIIA BCR. The arginine (R) residue in HCDR3, a charged amino acid, may be critical in providing crossreactivity⁴² together with light chain. Autonomous signaling capacity⁴³ may also be involved in selecting B1a cells with distinctive BCRs. Thus, BCR selection is not a single step process. Mouse AMyIIA IgM exhibits commensal bacteria crossreactivity (Supplementary Fig. S6). Microbiota provide crossreactive antigen(s) and also have an ability to trigger proliferation through TLR signal, suggesting that intestine may provide one such microenvironment. B1 B cells generated in neonatal liver initially distribute to intestinal mucosal tissues (Figure 2g), in addition to spleen and peritoneal cavity, and have the ability to circulate through lymphatic and blood vessels in the adult, including the intestine^{15, 44, 45}. Subcutaneous infiltration at leukemic stage is a unique feature found in this mouse AMyIIA CLL. Non-muscle myosin IIA is a central integrator to regulate cell migration and adhesion³³. Since our mouse AMyIIA models on a C.B17 mouse background showed high natural AMyIIA autoantibody production in serum (data not shown), continuous production of antibody and/or internal IgM interaction with the cytoskeleton may have contributed to this unique outcome. One possibility is the platelet aggregation by AMyIIA antibody, which may be responsible for blockage of leukemic B cell circulation from lymphatic vessels to the subclavian vein⁴⁶, leading to infiltration of the neck region promoted by deregulation of the miR15a/16-1 cluster. Targeted deletion in mouse of regions encompassing the miR15a/16-1 cluster promotes MBL/CLL, including cases with hemizygous deletion^{30, 47}, and around 80% of human CLL with 13q14 deletion is monoallelic. Thus, hemizygous deletion of 14q in mouse AMyIIA CLL appears to be sufficient to promote lymphoproliferation and infiltration.

Whether the AMyIIA CLL origin that we observed with mice also occurs in humans is an important issue for further investigation. Though there are no highly similar mouse and human V_H gene protein sequences, the human germline V_H most similar to mouse V_HQ52 AMyIIA BCR is V_H4-59 with 64% identity, and 83% similarity in framework segment FR1. Interestingly, V_H4-59 (V71-4) is expressed by B cells at the fetal/neonatal stage in humans, with increased usage in aged individuals^{48, 49}, and is found in human CLL². A well-known stereotyped AMyIIA CLL BCR in human is the V_H1-69/D3-16/J3 IgH paired with IgKV3-20 IgL^{7, 8} with polyreactivity^{50, 51}. This IgH rearrangement utilizes nearly identical CDR3s of greater length, also implying an important role for CDR3⁵². V_H1-69-expressing B lineage cells are present in human fetal liver⁵³, and V_H1-69 is one of the predominant V_H gene families found in unmutated CLLs^{8, 54} with crossreactivity to virus and also to intestinal commensal bacteria⁵⁵. V_H1-69-expressing B cells can be detected in healthy elderly individuals⁵⁶, but not at increased frequency, and this particular V_H1-69/D/J3 IgH rearrangement is not readily detectable in aged humans⁵⁷. Similarly, in mice, while V_HQ52 IgH is preferentially utilized early in life, the CLL-associated stereotype AMyIIA BCR is infrequent both in neonatal and aged mice. However, mouse BCR model analysis made it clear that CLLs with this BCR arise from B cells generated by early fetal/neonatal B-1 development, promoted by TCL1 transgenesis. Higher TCL1 expression is also correlated with aggressive B CLL in humans⁵⁸. Thus, human AMyIIA CLLs may also involve selection of particular CDR3s from an early-generated B cell pool, promoted both by genetic and exogenous influences.

Whether mouse-like fetal B-1 development occurs in humans remains an unanswered question. However, the differential expression patterns of Lin28b that distinguishes fetal and adult hematopoietic cells in mice has also been found in humans¹⁷, and CD5⁺ B cell generation, including at the immature stage, is the predominant outcome of development in the human fetus and in cord blood^{36, 37, 59}. Since the level of TdT is lower in human fetal development than adult, but is upregulated much earlier than in mice⁶⁰, N-addition in human V_H1-69/D3-16/J3 with a longer CDR3 does not exclude fetal/neonatal generation. Our mouse AMyIIA BCR CLL data raises the interesting possibility that distinctive fetal/neonatal B-1 development in humans may also generate a pool of self-reactive B cells that serve as precursors for unmutated BCR CLLs, where specific BCRs selected by this development, on certain genetic backgrounds, pose a high risk for leukemic progression.