Programmed death-ligand 1
Programmed death-ligand 2
B1 B cells are the major source of natural antibody that is essential for innate immunity. The B1 repertoire is skewed toward production of phosphatidylcholine (PtC)-binding VH11 and VH12 immunoglobulin that plays a key role in immune defense against bacterial infection. Programmed death-ligand 2 (PD-L2) is a ligand for the immunosuppressive receptor programmed death-1 (PD-1). It has been reported that expression of PD-L2 is restricted to dendritic cells and macrophages in mice. Here we show that 50–70% of resting peritoneal B1 cells express PD-L2, which is not present or inducible on conventional B2 B cells or PD-L2– B1 cells. Although PD-L2+ and PD-L2– B1 cells are similar in proliferative responses and spontaneous immunoglobulin secretion, PD-L2+ B1 cells are highly enriched for expression of VH11 and VH12 genes and encompass the bulk of PtC-binding B1 cells. These findings extend the range of known PD-L2 expression to B cells and show that B1 cells identified by this marker express a specific repertoire associated with anti-bacterial immunity.
B1 B cells comprise a unique subset of B cells distinguished phenotypically, ontogenetically, and functionally from conventional (B2) B cells 1–3. B1 B cells are the major source of natural immunoglobulin in normal unimmunized mice, providing a vital front line defense against bacterial and viral infection 4–8. This function is associated with a skewed repertoire containing autoreactive specificities, one manifestation of which is expression of phosphatidylcholine (PtC)-binding VH11 and VH12 by a substantial portion of peritoneal B1 cells in unimmunized mice 9–11. PtC is the polar headgroup of common membrane phospholipids. Reconstitution of anti-PtC IgM in sIgM-deficient mice substantially reduced the bacterial load in a cecal ligation and puncture model 5. Thus, VH11/VH12-expressing B1 cells play an important role in immunity; considering that PtC-binding B1 cells recognize protease-treated autologous erythrocytes, they may play a role in hemolytic autoimmune dyscrasias as well 10, 12.
B cells, like T cells, express programmed death-1 (PD-1), engagement of which negatively regulates B cell activity in vitro13. Surface levels of PD-1 are up-regulated by BCR stimulation and down-regulated by cytokines generally grouped as “danger” signals 14. Because of the wide distribution of the ligand for PD-1, B7-H1 (PD-L1: programmed death-ligand 1), it has been suggested that PD-1 acts as a “fail-safe” means of down-modulating aberrantly activated autoreactive lymphocytes, and PD-1 deficiency in mice leads to autoimmunity 15, 16. The PD-1 pathway not only plays a pivotal role in preventing autoimmunity but also plays an important role in regulating viral and parasite infection 17. A second PD-1 ligand, B7-DC (PD-L2: programmed death-ligand 2), is much more restricted in its distribution, being inducibly expressed only on dendritic cells and macrophages, but appears to trigger the inhibitory function of PD-1 much like PD-L1 18–20. To study the potential role of the PD-1 pathway in B1 cell-mediated immunity and autoimmunity, we examined the expression of PD-1 ligands. Surprisingly, we found PD-L2 expression on a subset of B1 cells, which defines a population enriched for VH11/VH12-expressing and PtC-binding B cells.
Results and discussion
Peritoneal B1 cells uniquely express PD-L2
To evaluate PD-L1 and PD-L2 expression, cells from normal peritoneal washout fluids and spleens were stained with fluorescent antibodies to identify B1 cells, B2 cells and T cells and were counterstained to detect PD-L1 and PD-L2. Like B2 cells and T cells, B1 cells expressed a basal level of PD-L1, although to a much greater extent (Fig. 1A). Surprisingly, we found that a significant portion of peritoneal B1 cells (50–70%) expressed PD-L2, whereas B2 cells and T cells failed to do so (Fig. 1A). A smaller and less distinct population of splenic B1 cells also expressed PD-L2. To exclude the possibility of staining artifacts, we sorted PD-L2+ and PD-L2– peritoneal B1 cells and examined PD-L2 gene expression by real-time PCR. PD-L2+ B1 cells expressed much more PD-L2 RNA than did PD-L2– B1 cells, confirming specific expression of PD-L2 by a B1 subpopulation (Fig. 1B).
PD-L2 expression is not inducible on B1 or B2 cells
To confirm the unique specificity of PD-L2 expression by B1 cells, we examined splenic B2 cells for PD-L2 expression after stimulation for 2 days with the cytokines IL-4, IFN-γ, and GM-CSF, which have been reported to induce PD-L2 on macrophages and DC 19, 20 (Fig. 2A). PD-L2 was not induced by any of these cytokines, even in the presence of anti-IgM (Fig. 2A). Neither did B2 cells express PD-L2 after stimulation with CD40L, LPS, CpG, or PMA/ionomycin despite induction of B7.1, B7.2 and other activation markers (data not shown; 21).
To test whether PD-L2– B1 cells can be converted to PD-L2+ cells, we cultured sorted peritoneal B1 cells for 2 days with the cytokines and stimuli noted above. None of the stimuli induced PD-L2 expression on B1 cells that did not initially express PD-L2 (Fig. 2B), suggesting that PD-L2 expression is an intrinsic property of PD-L2+ B1 cells. Unexpectedly, rather than inducing PD-L2 expression on previously PD-L2– B1 cells, several stimuli, specifically CpG, LPS and CD40L, converted substantial numbers of initially PD-L2+ B1 cells to PD-L2– B1 cells (Fig. 2B).
Because inducible expression of PD-L2 on macrophages has been reported to be STAT6-dependent 19, we evaluated the STAT6 dependence of B1 cell PD-L2 expression by examining B cells obtained from STAT6-KO mice. Loss of STAT6 had little effect: STAT6-KO mice contained normal numbers of peritoneal B1 cells, and among those B1 cells, the fraction that was PD-L2+ (and the fraction that was PD-L1+) was similar to the corresponding fraction in WT control B1 cells (Fig. 2C, D).
Distinctive functional features are expressed similarly in PD-L2+ and PD-L2– B1 cells
To determine whether PD-L2+ and PD-L2– B1 cells differ with respect to distinctive aspects of B1 cell function, we evaluated proliferative responses and immunoglobulin secretion. PD-L2+ and PD-L2– B1 cells similarly incorporated thymidine in response to PMA alone and failed to respond to anti-Ig, as expected for unseparated B1 cells (data not shown; 22, 23) and in direct contrast to B2 cells (Fig. 3A; 23). PD-L2+ and PD-L2– B1 cells spontaneously secreted IgM similarly, as judged by ELISPOT assays, as expected for unseparated B1 cells (Fig. 3B; 24) and unlike B2 cells. Thus, PD-L2 expression does not demarcate B1 cells with these distinctive functional features.
PD-L2 positivity defines B1 cells that express VH11/VH12 and bind PtC
To determine whether PD-L2+ and PD-L2– B1 cells differ with respect to VH11/VH12 usage, we analyzed gene expression by real-time PCR. PD-L2+ B1 cells expressed substantially more VH11 and VH12 mRNA than did PD-L2– B1 cells, whereas these subsets did not differ in expression of several other (control) VH gene segments (Fig. 3C). To confirm these results, we then analyzed PtC binding by flow cytometry using fluorescent, PtC-bearing liposomes. Among peritoneal B1 cells, PtC-binding was contained predominantly in the PD-L2+ subset (Fig. 3D). Thus, PD-L2 expression identifies a B1 cell subset whose repertoire is heavily skewed toward VH11/VH12 usage and PtC binding, in contrast to B1 cells that lack PD-L2.
These results indicate that murine PD-L2 expression is not as highly restricted as previously believed and extends beyond dendritic cells and macrophages to include B1 cells. Although some B1 cells express PD-L1 alone, as B2 cells do, the majority of B1 cells express both PD-L2 and PD-L1, which would seem to represent redudant, dual expression of two PD-1 ligands. It is notable then that PD-L2 has a higher affinity for PD-1 than does PD-L1 25. In addition, blockade and/or elimination of PD-L2 or PD-L1 produce different effects on autoimmune dyscrasias 17. Thus, it may be speculated that PD-L1 and PD-L2 act individually and subserve distinct roles during co-expression on B1 cells. A further implication of this work is the unknown extent to which previous observations on animals in which PD-L2 and PD-L1 were manipulated may have involved B1 cells.
Many features of B1 cells are inducible in B2 cells, including CD44 expression, PMA responsiveness, and even expression of the B1 cell marker CD5 itself 2, 22, 26. This and other evidence has suggested that B1 cells are produced as a result of certain forms of BCR stimulation, in contradistinction to the idea that B1 cells represent a distinct lineage. The recent identification of a B1 cell progenitor lends weight to the latter concept 27. In line with this, PD-L2 expression represents a B1 cell characteristic that is not inducible in B2 cells and thus supports the notion that B1 cells are not an activated version of B2 cells.
Peritoneal B1 cells express a number of macrophage markers such as Mac-1, F4/80, and CD14. PD-L2 expression adds to this list. However, PD-L2 is not constitutively expressed on resident macrophages, in contrast to constitutive PD-L2 expression on B1 cells. Furthermore, STAT6 plays a key role in up-regulation of PD-L2 on macrophages, but absence of STAT6 had no effect on constitutive B1 cell PD-L2 expression. Thus, the rules regulating PD-L2 expression on B1 cells are very different from those regulating PD-L2 expression on macrophages.
The key finding of a correlation between VH11/VH12 expression and PtC binding on the one hand and PD-L2 expression on the other suggests an essential relationship. Recently it was reported that PD-L2 can mediate bidirectional signaling 28, and so it may be through receptor crosstalk that PD-L2 alters the threshold for BCR stimulation and expansion of this B1 subpopulation. Alternatively, PD-L2 expression may simply be a sign of specific heavy chain usage. Either way, the relationship between PD-L2 and repertoire expression is likely to have important implications for B cell development and therapeutic interventions directed toward B1 cell immunity and autoimmunity.
The work described herein demonstrates that murine PD-L2 expression is not as highly restricted as previously believed and extends beyond dendritic cells and macrophages to include B1 cells. Furthermore, PD-L2 expression divides peritoneal B1 cells into two distinct populations: PD-L2+ B1 cells are heavily skewed in terms of characteristic B1 cell VH usage and specificity, whereas PD-L2– B1 cells are not. Unlike dendritic cells and macrophages, PD-L2 expression does not appear to be inducible on PD-L2– B1 cells or B2 cells but is expressed constitutively on a major subpopulation of B1 cells and declines with some forms of B cell stimulation. Elucidation of the origin of these two distinct populations of B1 cells is likely to have important implications for understanding and manipulating B cell immunity.
Materials and methods
Male BALB/cByJ mice, C.129S2-Stat6tmlGru/J mice and control BALB/cJ mice were obtained from The Jackson Laboratory. Mice were handled in accordance with National Institutes of Health and Institutional guidelines.
Cell isolation and culture
Single-cell suspensions were prepared from pooled spleens and peritoneal washout fluids. Erythrocytes were depleted using red blood cell lysis buffer (Sigma Chemical Co.). Subpopulations of B1 cells were sorted according to B220, CD5, and PD-L2 surface staining using a MoFlo cytometer (Dako-Cytomation). Splenic B2 cells were sorted as B220+CD5– cells or were negatively purified using MACS (Miltenyi Biotec). Sorted cells were ⩾98% pure by post-sort analysis. All cells were cultured in RPMI 1640 containing 10% FCS.
Isolated cells were washed with PBS containing 2% fetal bovine serum (Sigma) followed by FcR blockade with 2.4G2 anti-CD16/CD32 antibody (BD Biosciences Pharmingen). Cell surface markers were then analyzed by staining with FITC-, CyChrome-, and PE-conjugated mAb specific for B220, CD5 (BD Pharmingen), and PD-L1 or PD-L2 (eBiosciences) for three-color staining. For four-color staining, biotin-labeled anti-PD-L2 and streptavidin-APC were used. PtC-bearing, fluorochrome-encapsulating liposomes 10 were the kind gift of Dr. S. H. Clarke (University of North Carolina, Chapel Hill, NC). Data were acquired using a FACSCalibur flow cytometer (BD Biosciences) with CellQuest software (BD Biosciences) and were analyzed using FlowJo software (TreeStar).
Total RNA was obtained from sorted PD-L2+ and PD-L2– B1 cells using the RNeasy mini kit (Qiagen). RT was performed using TaqMan Reverse Transcription Reagents (Applied Biosystems, Roche). The resulting cDNA was used as template for real-time PCR. TaqMan primer and probe sets for GAPDH, PD-L1, and PD-L2 (Applied Biosystems, Roche) were used according to the manufacturer's instructions. PCR settings were as follows: 50°C for 2 min, 95°C for 10 min, followed by 40–60 cycles of 95°C for 15 s and 60°C for 1 min. Relative expression was calculated as 2–ΔCt, where ΔCt refers to Ct (GOI) – Ct (GAPDH). SYBR Green real-time PCR assays were performed for all VH genes with the following primers (forward/reverse):
β2-microglobulin (CCCGCCTCACATTGAAATCC/GCGTATGTATCAGTCTCAGTGG); Cμ (CCACTACGGAGGCAAAAACAG/TGGAGTGAAGTTCGTGGCC); VH11 (GCAATAAACTACGCACCATCCA/TGTCCTCCGATCGCACATT); VH12 (CTTCTACAACCCATCCCTCCAG/TACATGGCTGTGTCCTCTGTGG); VH6v2 (GTCCTGCAAGGCTTCTGGC/ACCCAGTGCATCCAGTAGCTG); VH6v3 (CACCCTCCAGCACAGCCTAC/AGTCCTCAGATGTCAGGCTGC).
Thymidine incorporation assay
B cells were cultured in triplicate or quadruplicate in 96-well U-bottom plates (2 × 105 in 0.2 mL) and stimulated for 24, 48, or 72 h. [3H]-thymidine was added during the last 6 h, and cpm was measured. Results are presented as mean cpm ± SD.
FACS-sorted B2, PD-L2+, and PD-L2– B1 cells were distributed at various dilutions onto MultiScreen-IP Plates (Millipore) precoated with goat anti-mouse Ig (H+L) and were then incubated for 3 h at 37°C and 5% CO2. Plates were treated with alkaline phosphatase-conjugated goat anti-mouse IgM (Southern Biotechnology Associates) and developed with 5-bromo-4-chloro-3-indolyl phosphate/p-NBT chloride substrate (KPL). Immunoglobulin-secreting cells were enumerated using Phoretix Expression software (NonLinear Dynamics).
This work was supported by Public Health Service grants AI29690 and AI60896 (T. R.), an American Heart Association Scientist Development Grant (X. Z.), and a Boston University Medical Center Evans Foundation pilot grant (X. Z.).