Regulation of PD-1, PD-L1, and PD-L2 expression during normal and autoimmune responses

Authors

  • Spencer C. Liang,

    1. Division of Medical Sciences, Harvard Medical School, Boston, USA
    2. Immunology Research Division, Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, USA
    Search for more papers by this author
  • Yvette E. Latchman,

    1. Immunology Research Division, Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, USA
    Search for more papers by this author
  • Janet E. Buhlmann,

    1. Immunology Research Division, Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, USA
    Search for more papers by this author
  • Michal F. Tomczak,

    1. Immunology Research Division, Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, USA
    Search for more papers by this author
  • Bruce H. Horwitz,

    1. Immunology Research Division, Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, USA
    Search for more papers by this author
  • Gordon J. Freeman,

    1. Department of Medical Oncology, Dana-Farber Cancer Institute, Department of Medicine, Harvard Medical School, Boston, USA
    Search for more papers by this author
  • Arlene H. Sharpe

    Corresponding author
    1. Immunology Research Division, Department of Pathology, Brigham and Women's Hospital and Harvard Medical School, Boston, USA
    • Eugene Braunwald Research Center, 221 Longwood Avenue, Boston, MA 02115, USA
    Search for more papers by this author

Abstract

Newer members of the B7-CD28 superfamily include the receptor PD-1 and its two ligands, PD-L1 and PD-L2. Here, we characterize the expression of PD-1, PD-L1, and PD-L2 in tissues of naive miceand in target organs from two models of autoimmunity, the pancreas from non-obese diabetic (NOD) mice and brain from mice with experimental autoimmune encephalomyelitis (EAE). In naive mice, proteiexpression of PD-1, PD-L1, and PD-L2 was detected in the thymus, while PD-1 and PD-L1 were detected in the spleen. PD-L1, but not PD-L2, was also detected at low levels on cardiac endothelium, pancreatic islets, and syncyciotrophoblasts in the placenta. In pre-diabetic NOD mice, PD-1 and PD-L1 were expressed on infiltrating cells in the pancreatic islets. Furthermore, PD-L1 was markedly up-regulated on islet cells. In brains from mice with EAE, PD-1, PD-L1, and PD-L2 were expressed on infiltrating inflammatory cells, and PD-L1 was up-regulated on endothelium within EAE brain. The distinct expression patterns of PD-L1 and PD-L2 led us to compare their transcriptional regulation in STAT4–/–, STAT6–/–, or NF-κB p50–/–p65+/– dendritic cells (DC).PD-L2, but not PD-L1, expression was dramatically reduced in p50–/–p65+/– DC. Thus, PD-L1 and PD-L2 exhibit distinct expression patterns and are differentially regulated on the transcriptional level.

Abbreviations:
AEC:

Aminoethyl carbazole

CHO:

Chinese hamster ovary

FDC:

Follicular dendritic cell

MFI:

Mean fluorescence intensity

PNA:

Peanut agglutinin

1 Introduction

The recently described PD-1:PD-ligand pathway within the CD28-B7 superfamily consists of one receptor, PD-1, and two ligands, PD-L1 (B7-H1) and PD-L2 (B7-DC) 13. PD-1, originally isolated from apoptotic T cell lines 4, is a 55-kDa transmembrane protein with one extracellular IgV-like domain and a 97-amino acid cytoplasmic tail containing one immunotyrosine inhibitory motif (ITIM) and one immunotyrosine switch motif (ITSM) 5. While the expression of CD28, CTLA-4, and ICOS is limited to T cells, PD-1 can be expressed on activated T cells, B cells, and myeloid cells 1. The function of PD-1 in vivo has been revealed through studies of PD-1-deficient (PD-1–/–) mice. C57BL/6 PD-1–/– mice developed a lupus-like glomerulonephritis and arthritis starting at 6 months of age, while knockout mice on the BALB/c background developed a dilated cardiomyopathy as early as 5 weeks of age 6. These findings suggest an inhibitory role for PD-1 in vivo and a potential role in regulating tolerance and autoimmunity.

There are two ligands known to bind PD-1; PD-L1 was identified through homology searches with known B7 molecules 7, 8, and PD-L2 was identified both by its homology to PD-L1 and by analysis of genes differentially expressed in libraries generated from dendritic cells (DC) and activated macrophages 9, 10. The functions of these ligands are still under investigation. Similar to other B7 family members, both ligands contain one extracellular IgV and one IgC domain. While the expression of B7-1 (CD80) and B7-2 (CD86) is limited primarily to lymphoid cells, PD-L1 and PD-L2 mRNA have been detected in a variety of tissues, including heart, lung, and placenta.

Little is known about the tissue-specific expression of PD-L1 or PD-L2 at the protein level in vivo or the molecular mechanisms that regulate their expression. Here, we report the expression of PD-1, PD-L1, and PD-L2 at the protein level in vivo in both hematopoietic and non-hematopoietic tissues of naive mice and in two models of autoimmunity. We demonstrate that PD-L1 and PD-L2 have distinct expression patterns in vivo and that these ligands are up-regulated in autoimmune settings. We also demonstrate that STAT6 and NF-κB, but not STAT4, control PD-L2 expression on DC. In contrast, PD-L1 expression is only slightly altered in NF-κB-deficient but not in STAT4- or STAT6-deficient DC. The differential expression of PD-L1 and PD-L2 suggests distinct functions for these PD-1 ligands.

2 Results

2.1 Expression of PD-1, PD-L1, and PD-L2 in lymphoid organs

To examine the protein expression of PD-1 and its ligands, we generated antibodies to murine PD-1 (clone 29F.1A12) and PD-L1 (clone 10F.5C5). Clone 29F.1A12 bound to PD-1-transfected Chinese hamster ovary (CHO) cells but not untransfected CHO cells (Fig. 1 and data not shown). Anti-murine PD-L1 clone 10F.5C5 bound PD-L1-transfected CHO cells but not untransfected or PD-L2-transfected cells (Fig. 1A and data not shown). We also analyzed the ability of these new mAb to block PD-1 interactions with PD-L1 or PD-L2 in an in vitro blocking assay. PD-1-transfected CHO cells were first incubated with the anti-PD-1 (29F.1A12) mAb. Biotinylated PD-L1 or PD-L2 immunoglobulin (Ig) fusion protein was then added to the PD-1-transfected CHO cells to determine the ability of the mAb to inhibit binding of PD-L1 and PD-L2 Ig. Clone 29F.1A12 anti-PD-1 mAb substantially blocked PD-L1 Ig and PD-L2 Ig binding at high concentrations but only blocked 35% of PD-L1 Ig or PD-L2 Ig binding at 10 μg/ml (Fig. 1B). A similar blocking assay was performed to characterize the anti-PD-L1 (10F.5C5) mAb. PD-L1-transfected CHO cells were incubated with the 10F.5C5 mAb, followed by addition of the biotinylated PD-1 Ig fusion protein. Clone 10F.5C5 anti-PD-L1 mAb was able to completely block binding of the PD-1 fusion protein at concentrations as low as 2.5 μg/ ml (Fig. 1C).

To examine the expression of PD-1, PD-L1, and PD-L2 in lymphoid organs, we performed confocal fluorescent microscopy on spleen and thymus sections from naive BALB/c mice. PD-1 expression was detected primarily in the T cell zone of the spleen and co-localized with CD3 expression (Fig. 2i–iv). This population was also CD45RBlo as determined by flow cytometry, suggesting that the T cells are antigen-experienced or regulatory T cells (data not shown). Expression of PD-L1, but not PD-L2, was detected in the spleen (Fig. 2v – x and data not shown). PD-L1 expression was most intense in the marginal zone surrounding the white pulp, but weak PD-L1 expression was found diffusely within the white pulp, co-localizing with expression of both CD3 (Fig. 2viii, long arrow) and IgM (Fig. 2viii, short arrow). PD-L1 also co-localized with CD11c, suggesting that PD-L1 is expressed on a DC population in the naive mouse (Fig. 2ix–xi). In summary, PD-1 was expressed on the CD3+ population within the spleen, whereas PD-L1 was expressed on DC, T cells, and B cells, with the strongest expression in the marginal zone. PD-L2 expression was not detected within the spleen of naive BALB/c mice.

In the thymus, PD-1 was found on thymocytes primarily in the medulla but also in the cortex (Fig. 2xii–xiv). The expression patterns of PD-L1 and PD-L2 in the thymus were distinct. PD-L1 was expressed in both the cortex and the medulla, with a staining pattern suggestive of cortical epithelium, medullary epithelium, and DC (Fig. 2xv). PD-L2 expression in the thymus was confined to the medulla on interdigitating cells, suggestive of DC (Fig. 2xviii). When sections were stained for CD11c and the PD ligands, both PD-L1 and PD-L2 partially co-localized with CD11c staining (Fig. 2xvii, xx). Our data agree with previous reports demonstrating expression of PD-L1 mRNA in the thymus 11.

Figure 1.

Specificity and blocking ability of anti-murine PD-1 (29F.1A12) and anti-murine PD-L1 (10F.5C5). (A) Anti-murine PD-1 (29F.1A12) (line) and anti-murine PD-L1 (10F.5C5) (line) were tested for specificity against PD-1-transfected CHO cells and PD-L1-transfected CHO cells. Control rat isotype is shown as the filled histogram. (B) An in vitro blocking assay was performed to determine the ability of anti-murine PD-1 (29F.1A12) to inhibit PD-1 interactions with PD-L1 Ig (solid line) and PD-L2 Ig (dashed line). (C) A similar in vitro blocking experiment was done with anti-murine PD-L1 (10F.5C5). Percent inhibition of binding is calculated based on the MFI of each sample compared to the MFI of the isotype control.

Figure 2.

Expression of PD-1, PD-L1, and PD-L2 in spleen and thymus. In images i-viii, spleen sections from naive BALB/c mice were triple-stained for PD-1 (i) and PD-L1 (v) in green, CD3 (ii, vi) in red, and IgM (iii, vii) in blue and analyzed by confocal microscopy. For each section, the three panels were merged (iv, viii). Colocalization of PD-1 or PD-L1 (green) with CD3 (red) resulted in a yellow color, while colocalization of PD-1 or PD-L1 (green) with IgM (blue) resulted in cyan. For images iv and viii, long arrows denote colocalization between PD-1 or PD-L1 with CD3, while short arrows denote colocalization between PD-L1 and IgM. In images ix–xi, splenic sections were co-stained for PD-L1 (ix, green) and CD11c (x, red), and the merge of both images is shown in panel xi. Similarly, BALB/c thymus sections were stained for PD-1 (xii), PD-L1 (xv), and PD-L2 (xviii) in green and co-stained for CD3 (xiii) or CD11c (xvi, xix). In panels xiii, xvi, and xix, M=medullary region, C=cortical region of the thymus. Merge panels are shown in xiv, xvii, and xx, with arrows denoting colocalization. Magnification ×200.

2.2 Expression of PD-1, PD-L1, and PD-L2 in germinal centers

Previous studies have demonstrated that PD-1 is expressed on activated B cells in vitro12. This led us to examine the expression of PD-1 and its ligands in germinal centers. Mice were immunized with ovalbumin emulsified in CFA, and draining lymph nodes were isolated 10 days later. Sections were double stained for PD-1, PD-L1, or PD-L2 (blue) and for peanut agglutinin (PNA, red) to delineate germinal centers (Fig. 3). Within germinal centers, PD-1 expression was detected on small round cells, suggestive of germinal center B cells. In contrast, neither PD-L1 nor PD-L2 was detected on any cell within the germinal center. The presence of follicular DC in these sections was confirmed by staining consecutive sections with anti-follicular DC (FDC) mAb (data not shown). Expression of PD-L1 and PD-L2 was not detected on murine FDC.

Figure 3.

Expression of PD-1, PD-L1, and PD-L2 in germinal centers. Mice were immunized in the footpad with 100 μg ovalbumin emulsified in CFA. On day 10, popliteal lymph nodes were frozen, sectioned, and stained with control IgG (A), anti-PD-1 (B), anti-PD-L1 (C), or anti-PD-L2 (D). mAb were detected by biotinylated anti-rat IgG, followed by streptavidin-AP, and developed with Fast Blue (blue). Sections were co-stained with PNA-HRP and developed with AEC (red) to identify germinal centers. Magnification ×100.

2.3 PD-1, PD-L1, and PD-L2 expression in non-hematopoietic organs

PD-L1 and PD-L2 mRNA was previously found to be expressed in non-hematopoietic organs such as heart, lung, and kidney. For this reason, we used immunohistochemistry to examine protein expression of the PD-1 ligands in the organs of a naive mouse. PD-L1 was expressed in heart (Fig. 4B), pancreas (Fig. 4E), small intestines (Fig. 4H), and placenta (Fig. 4K) but not in testes, kidney, or brain of naive mice (data not shown). In heart tissue (Fig. 4B), PD-L1 expression was seen on endothelium. PD-L1 expression was detected at low levels on endothelium and islet cells in the pancreas (Fig. 4E). We confirmed the staining of PD-L1 on endothelium by staining sections with the endothelium marker CD31 (data not shown). Mononuclear cells within the lamina propria of the small intestine also expressed PD-L1. In the placenta, we detected PD-L1 on the syncytiotrophoblasts. We also detected PD-L1 on lung macrophages, as previously reported (data not shown) 13. In marked contrast, PD-L2 staining was not detected in brain, heart, small intestine, placenta, lung, pancreas, kidney, or testes of naive BALB/c mice (Fig. 4 and data not shown).

Figure 4.

Expression of PD-L1 and PD-L2 in naive BALB/c organs. Sections of heart (A–C), pancreas (D–F), small intestines (G–I), and placenta (J–L) from naive BALB/c mice were stained with control IgG, anti-PD-L1, and anti-PD-L2 followed by biotinylated anti-rat IgG. After subsequent incubation with streptavidin-HRP, sections were developed with AEC and counterstained with hematoxylin. Magnification ×400.

2.4 PD-1, PD-L1, and PD-L2 expression is up-regulated in NOD and EAE mice

Because PD-L1 is expressed in non-hematopoietic organs and PD-1 has been implicated in regulating autoimmunity, we further examined the expression of PD-L1 and PD-L2 in the target organs of two tissue-specific models of autoimmunity. The NOD murine model of diabetes exhibits many similarities to human type 1 diabetes 14. Mice develop insulitis by 3 weeks of age, with large numbers of DC, macrophages, and T cells infiltrating the islet 15. Disease is also dependent on inflammatory cytokines as well as genetic factors 16, 17. We examined the expression of PD-1 and its ligands in pancreatic islets of pre-diabetic NOD mice at 3 and 9 weeks of age (Fig. 5A–H). In 3-week-old NOD mice, no PD-1, PD-L1, or PD-L2 staining was observed, despite an initial infiltration of mononuclear cells. At 9 weeks of age, expression of PD-1 and PD-L1 was detected on infiltrating mononuclear cells in the islet (Fig. 5F, G). Strikingly, PD-L1 was dramatically up-regulated on islet cells (Fig. 5G). PD-L2 expression was not detected on cell infiltrates or on islet cells (Fig. 5H). No expression of PD-1, PD-L1, or PD-L2 was detected in the exocrine pancreas. Thus, PD-L1, but not PD-L2, is expressed in the pancreas of NOD mice.

Experimental autoimmune encephalomyelitis (EAE) is a largely Th1-mediated disease in which IL-12 and IL-23 play crucial roles in development and maintenance of disease 18. Inflammatory foci containing T cells, B cells, and macrophages are seen within the brains of mice with EAE 19. To induce disease, C57BL/6 mice were immunized with MOG33–55 in CFA and pertussis toxin. Disease symptoms developed by day 10, and mice were killed on day 14 for immunohistochemical analysis. PD-1, PD-L1, and PD-L2 were expressed on mononuclear cells within the meninges (Fig. 5J–L). PD-L1 expression was also up-regulated on the endothelium surrounding the cell infiltrates (Fig. 5K). PD-L2 expression was detected on small round cells, suggestive of macrophages or B cells, in the brain. In summary, PD-L1 and PD-L2 are differentially expressed in the brains of mice with clinical EAE.

Figure 5.

Expression of PD-1, PD-L1, and PD-L2 in pancreas of NOD mice and brain of EAE mice. Pancreas from 3-week-old (A–D) and 9-week-old (E–H) NOD mice were stained with isotype IgG (A, E), anti-PD-1 (B, F), anti-PD-L1 (C, G), and anti-PD-L2 (D, H). For panels I–L, brain tissues from mice exhibiting clinical symptoms 14 days after EAE induction were stained for control isotype IgG (I), PD-1 (J), PD-L1 (K), and PD-L2 (L). Magnification ×400.

2.5 Expression of PD-L2, but not PD-L1, is controlled by NF-κB and STAT6

Our immunohistochemical data revealed that PD-L1 and PD-L2 have distinct expression patterns on murine cells and tissues, suggesting that these ligands are regulated by different signaling molecules or transcription factors. For this reason, we sought to identify molecules that control the expression of PD-L1 and PD-L2. We focused our study on DC, as they express both PD-L1 and PD-L2 20. Previous data had suggested that Th1 cytokines are an important stimulus for PD-L1 expression, while IL-4 and GM-CSF are important stimuli for PD-L2 expression 10, 21. We therefore examined DC from STAT4–/– and STAT6–/– mice for altered expression of these ligands (Fig. 6). STAT4 is critical for IL-12 signaling, and STAT4–/– mice cannot mount an effective Th1 response 22. Conversely, STAT6 is involved in the IL-4 signaling pathway, and mice deficient in STAT6 are exhibit immune responses skewed towards Th1 23. DC from these mice were expanded in vivo with Flt-3L Ig treatment for 10 days, followed by purification and overnight culture in media alone or with exogenous IL-12 or IL-4. Wild-type DC cultured overnight in media alone expressed PD-L1 and PD-L2, but addition of IL-12 or IL-4 led to further up-regulation of the PD-1 ligands. As shown in Fig. 6A and B, STAT4–/– and STAT6–/– DC cultured in media alone did not exhibit altered basal levels of PD-L1 or PD-L2. PD-L1 and PD-L2 expression was also unaltered in STAT4–/– DC cultured with exogenous IL-12. STAT6–/– DC cultured with exogenous IL-4 had lower PD-L2 expression, but similar PD-L1 expression, when compared to DC from wild-type mice. In summary, STAT6 partially regulates IL-4-mediated up-regulation of PD-L2.

The NF-κB family consists of five proteins containing the Rel homology domain that exist as dimeric complexes in the cytoplasm. In resting cells, these complexes are bound to proteins of the IκB family. Upon stimulation, IκB is phosphorylated and degraded, allowing NF-κB to translocate into the nucleus 24. NF-κB has been shown to play a role in the transcription and regulation of other B7 family members 25, 26. We therefore hypothesized that NF-κB may also play a role in regulating PD-L1 and PD-L2 expression. Because p50–/–p65–/– mice are embryonic lethal, we chose to examine DC from NF-κB p50–/–p65+/– mice for altered PD-L1 and PD-L2 expression. These mice have recently been described to develop severe colitis when infected with Helicobacter hepaticus27. As shown in Fig. 7, PD-L1 expression was not altered in p50–/–p65+/– DC cultured in media alone or with exogenous LPS. Expression of PD-L1 was slightly decreased in IL-4-treated and IFN-γ-treated p50–/–p65+/– DC and slightly increased in GM-CSF-treated p50–/–p65+/– DC as compared to wild-type DC. Overall, however, PD-L1 expression was only modestly altered in p50–/–p65+/– DC. In contrast, PD-L2 expression was dramatically decreased in p50–/–p65+/– DC compared to wild-type DC cultured in media alone, LPS, IL-4, or IFN-γ. While IL-4 treatment up-regulated PD-L2 expression in the p50–/–p65+/– DC, it could not fully restore PD-L2 expression to wild-type levels. Addition of GM-CSF to the cultures was able to restore PD-L2 expression to near wild-type expression, although a difference in mean fluorescence intensity (MFI) still existed (wild type, 206.25±25.3 vs. p50–/–p65+/–, 134.1±21.9; n=4). In summary, NF-κB plays a significant role in regulating the expression of PD-L2 and a lesser role in regulating PD-L1 expression.

Figure 6.

Expression of PD-L1 and PD-L2 in STAT4–/– and STAT6–/– DC. DC from STAT4–/– (A) and STAT6–/– (B) mice were expanded in vivo with Flt-3L Ig. DC were purified and cultured overnight in media alone or with exogenous IL-12 (10 ng/ml) or IL-4 (1,000 U/ml). DC were triple-stained with anti-CD11c-APC, anti-PD-L1 (9G2)-FITC, and anti-PD-L2 (TY25)-PE. Histograms are gated on CD11c+ cells, and expression of PD-L1 and PD-L2 is shown for wild-type (solid), STAT4–/– (dashed), and STAT6–/– (dashed) DC, along with the appropriate isotype control (shaded). Data shown is representative of three experiments.

Figure 7.

Altered expression of PD-L2 in p50–/–p65+/– DC. DC from NF-κB p50–/–p65+/– mice were expanded with Flt-3L Ig. Purified DC were cultured overnight in media or with exogenous LPS (0.1 μg/ml), IL-4 (1,000 U/ml), GM-CSF (1.5 ng/ml), or IFN-γ (1,000 U/ml). Cells were then triple-stained with anti-CD11c-APC, anti-PD-L1-FITC (9G2), and anti-PD-L2-PE (TY25) and relevant control isotype IgG and analyzed by flow cytometry. Histograms are gated on CD11c+ cells, and expression of PD-L1 and PD-L2 is shown for wild-type (solid) and p50–/–p65+/– (dashed) DC, along with the appropriate isotype control (shaded). Data shown is representative of four experiments.

3 Discussion

In this report, we define the expression of PD-1, PD-L1, and PD-L2 protein in vivo. In naive mice, PD-L1 protein expression was detected on cardiac endothelium, pancreatic islet cells, mononuclear cells within the lamina propria, alveolar macrophages, and syncyciotrophoblasts within the placenta. In humans, PD-L1 also was detected on endothelial cells 28, myocardium 29, and syncyciotrophoblasts 29, 30. Thus, in the tissues examined, expression of PD-L1 is similar in human and mouse. In contrast, PD-L2 protein was not detected in any of the non-lymphoid tissues examined. Previously published data suggested expression of PD-L2 RNA in tissues such as lung, brain, and kidney 9. A possible explanation for this discrepancy is that the RNA used in the Northern blots was contaminated with RNA from the draining lymph nodes for each tissue. In contrast to the mouse, PD-L2 protein was detected in a variety of human tissues, including cardiac endothelium, placental endothelium, and myocardium 29. Overall, the expression of PD-L2 protein appears to be more restricted in mice than in humans, suggesting that PD-L2 may have different functions in the two species.

Examination of lymphoid organs revealed expression of PD-1 in the T cell zone of the spleen. PD-L1 was expressed diffusely throughout the white pulp, with strong expression in the marginal zone, while PD-L2 expression was not detected in the spleen. It is possible that PD-L2 is expressed at a low level or on a small population not detected in our samples. In the thymus, PD-1-positive cells were detected primarily in the medulla, which consists mostly of single-positive thymocytes, although some thymocytes in the cortex also expressed PD-1. Previously published data showed the majority of PD-1 expression on two populations, the CD44CD25TCRβ+ cells, found primarily in the cortex, and γ δ thymocytes 31. It is possible that, in our studies, the PD-1+ thymocytes detected in the medulla had recently transitioned there from the cortex. PD-L1 in the thymus was expressed on DC and on cells with a morphology suggestive of cortical and medullary epithelial cells. PD-L2 was found on thymic DC. Cortical epithelial cells were previously shown to play a role in the transition from double-positive to single-positive thymocytes 32, while medullary epithelial cells and DC are thought to be involved in mediation of anergy or deletion 33, 34. PD-1 was shown to play a role in facilitating CD8+ T cell positive selection but not negative selection 35. However, the role of the PD-1 pathway in CD4+ T cell-positive and -negative selection is unclear, and the expression of PD-1 ligands in the cortex medulla raises the possibility that this pathway modulates selection of CD4+ cells. It also remains to be seen if the recently reported second receptor for PD-L1 plays a role in T cell development 13.

We also examined germinal centers for the expression of PD-1, PD-L1, and PD-L2. PD-1+ cells were found in the germinal centers of mice immunized with ovalbumin emulsified in CFA. Similarly, expression of PD-1 in the germinal centers of human tonsils was shown previously 36. However, no expression of PD-L1 or PD-L2 was detected on murine germinal center B cellsor FDC, while PD-L1 and PD-L2 expression has been previously reported on human FDC 29. It is possible that these ligands are expressed at low levels on FDC or that different conditions can lead to their up-regulation.

We further demonstrate that PD-L1 and PD-L2 are expressed in the target organs of autoimmune responses. In the NOD model of diabetes, PD-1 and PD-L1 were detected on lymphocytes infiltrating the pancreatic islets. PD-L1 was also dramatically up-regulated on pancreatic islet cells in 9-week-old mice. Previous studies in the NOD model have shown that B7-1 and B7-2 are expressed on infiltrating cells, but not on the islet cells themselves 37, 38. Treatment of an islet cell line, NIT-1, with inflammatory agents such as IFN-γ or TNF-α also failed to up-regulate B7-1 37. The expression of PD-L1 on pancreatic islet cells and PD-1 on infiltrating cells suggests that this pathway may act to modulate pancreatic immune responses in NOD mice.

In the EAE model, PD-1, PD-L1, and PD-L2 were expressed on cells infiltrating the brain. In addition, PD-L1 was up-regulated on the endothelium and brain tissue surrounding the infiltrates. The presence of PD-L2 in EAE infiltrates but not NOD infiltrates may be due to different cell types infiltrating the brain versus the pancreas or to a different cytokine milieu. The expression of B7 family members on infiltrating cells in EAE is not unprecedented; previous reports have documented the expression of B7-1 and B7-2 on inflammatory cells in the spinal cords of mice with EAE 39, 40. However, this is the first report demonstrating that PD-L1 protein is actually up-regulated in the target organs of tissue-specific autoimmune disease. The expression of PD-L1 but not PD-L2 on non-hematopoietic tissues suggests that PD-L1 may be the more dominant of the two ligands in modulating tissue-directed inflammatory responses.

Because of the contrasting expression patterns of PD-L1 and PD-L2, we investigated the molecular pathways that might regulate expression of the ligands. We examined DC from STAT4–/–, STAT6–/–, and NF-κB p50–/–p65+/– mice for altered expression of the PD-1 ligands. PD-L1 expression was not altered in STAT4–/– or STAT6–/– DC, but was slightly reduced in NF-κB p50–/–p65+/– DC cultured with IL-4 or IFN-γ. In contrast, PD-L2 expression was decreased in STAT6–/– DC incubated with IL-4 but notmedia alone. While our data agree with a recent report demonstrating that PD-L2 up-regulation in response to IL-4 in macrophages is dependent on STAT6 41, we also show that PD-L2expression is not solely dependent on STAT6 signaling. In NF-κB p50–/–p65+/– DC cultured in media, basal levels of PD-L2 expression were dramatically reduced when compared to wild-type DC. Addition of exogenous IL-4, LPS, IFN-γ, or GM-CSF led to up-regulation of PD-L2 on p50–/–p65+/– DC, but could not restore expression to wild-type levels. The ability of exogenous cytokines to up-regulate PD-L2 on p50–/–p65+/– and STAT6–/– DC suggests that other signaling pathways also play a role controlling PD-L2 expression. The dramatic decrease of PD-L2 expression in p50–/–p65+/– mice is of interest, as these mice develop severe colitis when exposed to H. hepaticus 27. Decreased expression of PD-L2 on DC may enable inappropriate or prolonged activation of T cells through decreased PD-1 signaling. The NF-κB pathway has been previously shown to play a role in the regulation of other B7 family members. Using a retroviral transduction system, expression of B7-1 and B7-2 was shown to decrease when the NF-κB repressor, IκBα, was introduced into DC 26. In addition, analysis of the promoter region for human B7-1 has revealed NF-κB binding sites 42, and another B7 homologue, B7h (ICOSL, B7-RP1), was originally identified through a screen searching for genes induced by NF-κB 25.

The PD ligands represent a unique pair of ligands in the B7 family because the expression of PD-L1 and PD-L2 is so distinct. This distinct expression suggests that the two may have largely non-overlapping functions in regulating immune responses and autoimmunity.

4 Materials and methods

4.1 Generation of mAb specific for PD-1 and PD-L1

mAb to mouse PD-1 (29F.1A12) and PD-L1(10F.5C5) were generated by immunizing rats with plasmid DNA containing PD-1 or PD-L1. Boost immunizations were done with PD-1 and PD-L1 Ig fusion proteins. Rat spleens were fused with SP2/0 myeloma cells and screened for reactivity using PD-1- and PD-L1-transfected cells.

4.2 Blocking assay

PD-1-transfected CHO cells were incubated with 29F.1A12 for 30 min on ice. Biotinylated murine PD-L1 Ig or PD-L2 Ig was then added and the cells incubated for another 30 min on ice. Cells werewashed twice with 1% BSA/PBS, and binding of PD-L1 Ig or PD-L2 Ig was detected with streptavidin-PE. A similar protocol was followed using PD-L1-transfected CHO cells and biotinylated human PD-1 Igto characterize the 10F.5C5 mAb.

4.3 Antibodies and reagents

PE-conjugated anti-PD-L2, (TY25), rat IgG isotype, and purified anti-CD16 (2.4G2) were purchased from eBioscience (San Diego, CA). Allophycocyanin (APC)-conjugated anti-mouse CD11c (HL3) and purified FDC-M1 were purchased from BD Biosciences (San Jose, CA). Alexa 488-conjugated streptavidin, goat anti-mouse IgM Alexa 647, and goat anti-hamster IgG Alexa 568 were purchased from Molecular Probes (Eugene, OR). Anti-mouse CD11c (N418) was purified from cultured supernatants. Anti-mouse PD-L1 (9G2) was prepared as previously described 43. Purified anti-CD3 (2C11) was prepared by Bioexpress (West Lebanon, NH). Human PD-1 Ig, murine PD-L1 Ig, and murine PD-L2 Ig were gifts from Wyeth Research (Cambridge, MA). Recombinant mouse IFN-γ and IL-4 were purchased from BD Biosciences, and mouse GM-CSF was purchased from R&D Systems (Minneapolis, MN). Recombinant mouse IL-12 was a gift from Wyeth Research. LPS was purchased from Sigma (St Louis, MO). Flt-3L Ig was prepared by Bioexpress as previously described 44.

4.4 Immunohistochemistry and immunofluorescence

Tissues were snap-frozen with liquid nitrogen in OCT (Sakura Inc., Torrence, CA) and 5 μm sections were prepared. For immunohistochemistry, samples were fixed briefly in cold acetone and treated with 0.1% H2O2 in PBS to eliminate endogenous peroxidase activity. After blocking with 10% rabbit serum and biotin-avidin block (Vector Laboratories, Burlingame, CA), primary antibodies to PD-1 (29F.1A12), PD-L1 (10F.5C5), and PD-L2 (TY25) diluted in 2%BSA/PBS were applied to the sections. Biotinylated rabbit anti-rat IgG was applied as a second layer followed by streptavidin-horseradish peroxidase (HRP) (Endogen, Woburn, MA). Sections were developed with aminoethyl carbazole (AEC, Sigma), counterstained with Mayer's hematoxylin (Sigma), and mounted with Crystal Mount (Biomeda, Foster City, CA). For double staining, sections were stained for PD-1, PD-L1, and PD-L2 followed by biotinylated rabbit anti-rat IgG as described above. Sections were then incubated with streptavidin-alkaline phosphatase (AP, PharMingen) and PNA-HRP and developed with Fast Blue (Sigma) and AEC. For confocal microscopy, samples were frozen, cut, and fixed as above. Sections were blocked with 10% rabbit serum and 5% goat serum, followed by biotin-avidin block. A similar protocol was used to detect PD-1, PD-L1, and PD-L2, with the exception that streptavidin-Alexa 488 was used asthe final layer. For three-color staining, sections were co-stained with anti-mouse IgM Alexa 647 and with anti-CD3 (2C11) detected by anti-hamster IgG Alexa 568. For detection of DC, sections werestained with anti-CD11c (N418) followed with goat anti-hamster IgG Alexa 568. Sections were mounted with Gel Mount (Vector Laboratories), coverslipped, and analyzed with the Radiance system confocal microscopy system (Bio-Rad, Hercules, CA).

4.5 Cell culture and cell lines

CHO cells were generated and maintained as previously described 9. All other cell culture was performed in RPMI 1640 (Invitrogen, Carlsbad, CA) supplemented with 10% FBS (Sigma), 2 mM L-glutamine (Invitrogen), 10 mM Hepes (Invitrogen), antibiotic-mycotic (Invitrogen), and 50 μM 2-mercaptoethanol (Sigma).

4.6 DC preparation

Mice were injected intraperitoneally with 20 μg Flt-3L every other day for 10 days. DC were purified using anti-CD11c magnetic microbeads (Miltenyi Biotec, Auburn, CA) and cultured overnight with 1,000 U/ml IL-4, 1,000 U/ml IFN-γ, 0.1 μg/ml LPS, 1.5 ng/ml GM-CSF, or 10 ng/ml IL-12.

4.7 Flow cytometry

Cells (2×105 to 1×106) cells were incubated with the appropriate antibodies diluted in 1% BSA and 0.02% sodium azide in PBS on ice for at least 20 min. Unconjugatedanti-CD16/32 was added to prevent nonspecific Fc binding. Cells were washed twice in 1%BSA/PBS and analyzed on a FACSCalibur (BD Biosciences).

4.8 Mice

Mice were housed in American Association for the Accreditation and Laboratory Animal Care-accredited institutions at Brigham and Women's Hospital (Boston, MA). BALB/c and C57BL/6 were obtained from Taconic Farms (Germantown, NY). STAT4–/– and STAT6–/– mice and NOD/Ltj were obtained from Jackson Laboratories (Bar Harbor, ME). Blood glucose was measured using Ascensia Elite XL glucometer (Bayer, Elkhart, IN) on a weekly basis. Mice were considered diabetic when two consecutive readings greater than 300 mg/100 ml were measured. NF-κB-deficient p50–/–p65+/– mice were backcrossed six generations onto the 129 background and housed in a Helicobacter-free environment.

4.9 Immunizations

To induce germinal center formation, mice were immunized with 100 μg ovalbumin (Sigma) emulsified in CFA (Sigma) in the hind footpad. At 10 days after immunization, popliteal lymph nodes were isolated and snap-frozen. For induction of EAE, C57BL/6 mice were immunized subcutaneously with 100 μg MOG33–55 (Biopolymer Facility, Brigham and Women's Hospital, Boston, MA) emulsified in CFA (Difco Laboratories, Detroit, MI) supplemented with 400 μg Mycobacterium tuberculosis H37RA (Difco Laboratories) in the two flanks. Pertussis toxin (200 ng; List Biological Laboratories, Campbell, CA) was administered intravenously on the day of immunization and 2 days later.

Acknowledgements

We are grateful to Y. Wu for her help with EAE as well as B. Chang and J. Burgess for their technical support. This research was funded by National Institute of Health grants (AI40614 and AI38310) and a National Multiple Sclerosis Society grant (NMSS RG2779) to A.H.S. and National Institutes of Health grants (CA84500 abd AI39671) to G.J.F.

Footnotes

  1. 1

    WILEY-VCH

  2. 2

    WILEY-VCH

  3. 3

    WILEY-VCH

  4. 4

    WILEY-VCH

  5. 5

    WILEY-VCH

  6. 6

    WILEY-VCH

  7. 7

    WILEY-VCH

Ancillary