Regulatory T Cells Use Programmed Death 1 Ligands to Directly Suppress Autoreactive B Cells in Vivo

Home , Memory B cell, Peripheral tolerance

Regulatory T cells use programmed death 1 ligands to directly suppress autoreactive B cells in vivo

Janine Gotota, Catherine Gottschalka, Sonny Leopolda, Percy A. Knollea, Hideo Yagitab, Christian Kurtsa,1,2, and Isis Ludwig-Portugalla,1,2

aInstitutes of Molecular Medicine and Experimental Immunology (IMMEI), Rheinische Friedrich-Wilhelms-Universität, 53105 Bonn, Germany; and bDepartment of Immunology, Juntendo University School of Medicine, Bunkyo-ku, Tokyo 113-8421, Japan

Edited by David Tarlinton, The Walter and Eliza Hall Institute of Medical Research, Parkville, VIC 3050, Australia, and accepted by the Editorial Board May 4, 2012 (received for review January 20, 2012)

The mechanisms by which regulatory T cells (Tregs) suppress auto- Programmed death-1 (PD-1, CD279) is an activation-induced antibody production are unclear. Here we have addressed this member of the extended CD28/CTLA-4 family that suppresses T question using transgenic mice expressing model antigens in the cells (18–21). It has been associated with exhausted memory kidney. We report that Tregs were essential and sufficient to sup- T cells in chronic viral infection (22, 23) and with cytotoxic T-cell press autoreactive B cells in an antigen-specific manner and to cross-tolerance (24). PD-1 has two known ligands, PDL-1 (B7-H1, prevent them from producing autoantibodies. Most of this sup- CD274) and PDL-2 (B7-DC, CD273) (25, 26), and both of them pression was mediated through the inhibitory cell-surface-mole- are sufficient to mediate T-cell suppression (27, 28). Tregs have cule programmed death-1 (PD-1). Suppression required PD-1 been shown to express PDL-1, but such expression was dispens- expression on autoreactive B cells and expression of the two PD- able for T-cell suppression in vitro (29). Also B cells specific for 1 ligands on Tregs. PD-1 ligation inhibited activation of autoreactive foreign antigens express PD-1 ligands and interact with follicular B cells, suppressed their proliferation, and induced their apoptosis. helper T cells known to express high levels of PD-1, resulting in Intermediate PD-1+ cells, such as T helper cells, were dispensable increased germinal center B-cell survival and plasma cell differ- fi for suppression. These ndings demonstrate in vivo that Tregs use entiation (30). On the other hand, PD-1 knockout mice develop PD-1 ligands to directly suppress autoreactive B cells, and they high levels of auto-Ab (31), which is hard to reconcile with identify a previously undescribed peripheral B-cell tolerance mech- a positive effect on antibody production. Thus, the role of PD-1 on IMMUNOLOGY anism against tissue autoantigens. B cells is unclear. To address these open questions, we have used mice expressing peripheral tolerance | autoimmunity | autoantibodies | inhibitory receptors OVA and HEL in kidney glomerular podocytes (32), which allow detailed studies in an organ where auto-Ab–mediated diseases are utoantibodies (auto-Ab) cause various autoimmune dis- prevalent (33). We demonstrate direct suppression of B cells by Aeases, such as systemic lupus erythematosus and certain Tregs and identify PD-1 signaling as the underlying mechanism. forms of glomerulonephritis. Depletion of autoreactive B cells ameliorates many, but not all, autoimmune diseases. However, Results fi this approach causes severe immunosuppression due to the Tregs Speci cally Suppress Auto-Ab Production Against Glomerular fi general loss of B cells. Specific control of autoreactive B cells is Auto-Ag. To study how Tregs speci cally suppress auto-Ab pro- required for improved therapies. duction against peripheral tissue antigens, we used transgenic mice Regulatory T cells (T ) are powerful suppressors of auto- expressing a membrane-bound fusion protein of OVA and HEL regs under the control of the nephrin promoter in kidney podocytes reactive T cells with high therapeutic potential (1–3). Tregs also suppress auto-Ab production (4, 5). We recently showed in vivo (NOH mice) (32). We used a vaccination scheme of applying OVA in aluminium hydroxide (Alum) three times (experimental scheme that they do so in an antigen-specific (Ag-specific) manner (6, 7). in Fig. 1A), which induced robust anti-OVA titers after 3 wk in These studies used rat insulin promoter HEL/OVA (ROH) mice nontransgenic wild-type (WT) control mice (6, 7). When we vac- expressing ovalbumin (OVA) and hen egg lysozyme (HEL) in cinated NOH mice with this scheme, OVA-specific IgG antibody pancreatic islet β-cells. Autoreactive OVA- and HEL-specificB B fi titers were sevenfold lower compared with WT mice (Fig. 1 ), cells, but not B cells speci c for a foreign antigen, failed to indicating immune tolerance. Consistent with our previous studies proliferate in response to in vivo autoantigen (auto-Ag) challenge on pancreatic auto-Ag (6, 7), treatment with the antibody PC61 1 d and instead underwent apoptosis in a strictly T -dependent + reg before vaccination depleted about 90% of the FoxP3 Tregs (Fig. fashion. Tregs can affect B cells indirectly by suppressing the S1) and restored anti-OVA antibody production in NOH mice to T-helper (Th) cells required for antibody production (8, 9). This 85% of that in WT controls (Fig. 1B). Antibody production against did not rule out that Tregs might also suppress B cells directly. the foreign antigen β-galactosidase (β-Gal) was unchanged in + Cell culture systems have revealed that CD25 Tregs can kill NOH and WT mice, and PC61 treatment had no effect either (Fig. – cocultured B cells (10 12). A recent in vivo study showed that 1C), demonstrating auto-Ag–specific suppression by Tregs. These Tregs enter germinal centers and suppress B cells in this site (13, 14). The question whether this occurred directly or indirectly remained open (15). This question is difficult to address in vivo Author contributions: J.G., C.K., and I.L.-P. designed research; J.G., C.G., S.L., and I.L.-P. performed research; P.A.K. and H.Y. contributed new reagents/analytic tools; J.G., C.G., because it requires an experimental system where Tregs can sup- P.A.K.,H.Y.,C.K.,andI.L.-P.analyzeddata;andJ.G.,C.K.,andI.L.-P.wrotethepaper. press B cells but not Th cells. fl Another open question concerns the molecular mechanisms by The authors declare no con ict of interest. This article is a PNAS Direct Submission. D.T. is a guest editor invited by the which Tregs suppress. In principle, Tregs may suppress other T cells Editorial Board. by (i) secreting inhibitory mediators; (ii) deprivation of survival 1C.K. and I.L.P. contributed equally to this study. factors; (iii) killing target cells by granzyme/perforin; and (iv) 2To whom correspondence may be addressed. E-mail: [email protected] or isis.ludwig- modulation of DCs by ligating inhibitory T-cell receptors (16, 17). [email protected]. The exact contribution of these mechanisms in relevant in vivo This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. situations and the mechanisms by which Tregs suppress are unclear. 1073/pnas.1201131109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1201131109 PNAS Early Edition | 1of6 Downloaded by guest on September 30, 2021 Fig. 1. Blocking PD-1 restores Ag-speciﬁc auto- Ab titers in vivo. (A) Experimental scheme for B

and C:miceweredepletedofTregs with PC61 Ab on days −4, −1, 6, and 13 (white arrows) and immunized with 10 μgOVAand10μg β-Gal in Alum on days 0, 7, and 14 (black arrows). (B and C) IgG titers against OVA (B)andβ-Gal (C)in NOH (black bars) or nontransgenic wild-type (WT) control mice (white bars) after depletion

of Tregs on day 21. (D) Experimental scheme for E and F: mice were depleted of Tregs with PC61 Ab on days −4and−1 and immunized with OVA/Alum on day 0. (E and F)Percentage(E) and mean ﬂuorescence intensity (MFI) (F)ofPD- 1+ OVA-speciﬁc B cells on day 3. (G) Experi- mental scheme for H and I: mice were immu- nizedondays0,7,and14andinjectedwithPD- 1–blocking RMP1-14 or isotype control Abs on thesamedays.(H) Anti-OVA serum IgG titers on day 21 in WT mice (white bar), NOH mice (black bar), NOH mice treated with RMP1-14 (light gray bar), or with isotype control (dark gray bar). (I) Anti-NP titers on day 14 in NOH (black bars) or WT (white bars) mice immunized with either OVA-NP or BSA-NP in Alum. *P < 0.05; **P < 0.01; ***P < 0.001 (ANOVA and Bonfer- roni). Data are representative of two experiments using three to four mice in each group.

findings indicate that Tregs prevent autoreactive B cells from by central tolerance mechanisms. To address this possibility, we producing anti-glomerular auto-Ab. immunized these mice with OVA-nitrophenol (NP) or BSA-NP as a control and determined the response of NP-specific B cells. PD-1 Mediates Peripheral B-Cell Tolerance Against Glomerular Auto- Such B cells were not autoreactive yet still were controlled by Ag. To identify candidate molecules for B-cell suppression, we OVA-specific Th cells or Tregs. NP-specific antibody serum titers isolated OVA-specific B cells (representative FACS plot in Fig. in NOH mice were 2.8-fold lower than in WT control mice after S2) from immunized NOH or WT mice and determined ex- immunization with OVA-NP, but were unchanged when mice pression of molecules previously implicated in peripheral im- were immunized with NP-BSA (Fig. 1I). When PD-1 was mune tolerance, such as PD-1 or Fas (11, 18, 34). OVA blocked, anti-NP titers were restored to 67% of those in WT vaccination increased PD-1 mRNA expression in OVA-specific controls, which was not significantly different (Fig. 1I), verifying B cells of NOH mice stronger (3-fold) than in WT mice (1.7- that PD-1 did not operate during B-cell development. Thus, PD- fold) (Fig. S3A), whereas no such preferential up-regulation was 1 mediates peripheral B-cell tolerance. seen for Fas (Fig. S3B). Flow cytometry confirmed selective PD- 1 protein up-regulation by OVA-specific B cells from the spleens Autoreactive B Cells Require PD-1 for Being Suppressed in NOH Mice. of OVA and β-Gal–immunized NOH mice, but not in non- To study whether B cells required PD-1 to be suppressed in vivo, OVA–specificorinβ-Gal–specific B cells (Fig. S4 A–C). Both we wanted to create a situation where only B cells were PD-1– the proportion of PD-1+ B cells (Fig. 1E and Fig. S3C) and the deficient by transferring PD-1–deficient autoreactive B cells into amount of PD-1 per B cell (Fig. 1F and Fig. S3D) were in- PD-1–competent NOH or WT recipient mice. To this end, we creased. PD-1 was up-regulated per B cell by only 25% (Fig. used transgenic B cells specific for HEL (IgHEL cells) that can S3D); nonetheless, this small increase was quantitatively com- be distinguished from wild-type B cells by their expression of parable to functionally relevant increases seen in other models IgM of the subtype a (IgMa). We had previously shown that (35). In contrast to PD-1, the PD-1 ligands remained unchanged these cells were deleted 3 d after transfer into ROH mice on OVA-specific B cells after immunization (Fig. S4 D and E). expressing HEL in pancreatic islets (7). Consistent with these Treg depletion prevented PD-1 up-regulation on OVA-specificB findings, IgHEL cells were deleted also after transfer into NOH cells but had no influence on β-Gal–specific B cells (Fig. 1 E and mice (but not into WT recipients), unless Tregs were depleted F, black bars; Fig. S4 F and G), supporting a connection between (experimental scheme in Fig. 2A, results in Fig. 2B). However, Treg-mediated B-cell suppression and PD-1. when we used PD-1–deficient IgHEL cells generated by crossing − − − − To test this hypothesis, we injected the PD-1 blocking antibody IgHEL mice to PD-1 / mice (IgHEL×PD-1 / cells, experi- RMP1-14 (18) into NOH mice immunized with OVA or β-Gal mental scheme in Fig. 2C), their numbers were similar in NOH (experimental scheme in Fig. 1G). This indeed restored anti- and WT recipient mouse groups (Fig. 2D), verifying that B-cell OVA IgG antibody production in NOH mice to 75% of that in suppression depended on their expression of PD-1. WT controls, whereas an isotype antibody control showed no Previously, we showed that reduction of IgHEL cell numbers such effect (Fig. 1H). Antibody levels against the foreign antigen occurred both by inhibiting their proliferation and by inducing β-Gal were unchanged (Fig. S5). The restoration of suppression apoptosis (7). When we examined proliferation by intracellular −/− was reminiscent of that seen after Treg depletion, albeit slightly Ki67 staining, IgHEL×PD-1 cells proliferated equally well in less effective (Fig. 1B), suggesting that B-cell suppression by Tregs NOH and WT mice, whereas WT IgHEL cells failed to divide in at least partially occurred via PD-1. NOH mice (Fig. 2E). Also, induction of apoptosis measured by In NOH mice, OVA-specific B cells are autoreactive and thus determining activated caspase 3 levels was prevented in − − might have been functionally altered during their development IgHEL×PD-1 / cells, in contrast to PD-1–competent IgHEL

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1201131109 Gotot et al. Downloaded by guest on September 30, 2021 – not in recipients of NOH CD25 T cells or of Tregs from WT mice (Fig. 3B). This demonstrated that Ag-speciﬁcTregs are sufﬁcient for in vivo suppression of B cells. −/− When we transferred NOH Tregs into IgHEL×PD-1 mice (experimental scheme in Fig. 3A), auto-Ab production was no longer suppressed (Fig. 3C) compared with PD-1–competent IgHEL recipients. Titers resembled those in IgHEL recipients of NOH Th cells or of WT Tregs (Fig. 3C). Thus, adoptively transferred Tregs required PD-1 expression by host cells, consistent with the necessity of PD-1 in B cells as shown above (Fig. 2).

PD-1 Ligand Expression by Tregs Is Required for B-Cell Suppression. We next wanted to adoptively transfer PD-1 ligand-deficient Tregs into IgHEL mice to clarify whether these ligands were necessary for B-cell suppression. To this end, we crossed NOH mice with −/− PDL-1 mice and transferred their Tregs into immunized IgHEL mice. PDL-1–deficient Tregs still prevented autoantibody production (Fig. 4B), suggesting that PDL-2 also was involved in suppression. This was consistent with a previous report showing that both PD-1 ligands are sufficient on their own to signal through PD-1 (26) and implied that we needed to incapacitate both of them −/− in our system. To do so, we transferred Tregs from NOH×PDL-1 mice and blocked PDL-2 with TY25 antibodies (36) (experimental scheme in Fig. 4A), which was faster than generating NOH×PDL- Fig. 2. B cells need to express PD-1 to be suppressed by Tregs in NOH mice. (A) Experimental scheme: 5 × 106 IgHEL were transferred into untreated or

Treg-depleted WT (white bars in B) or NOH mice (black bars in B), which were

immunized with HEL in Alum on the next day. (B) Absolute numbers of IMMUNOLOGY surviving B220+ IgMa+ HEL+ IgHEL cells 3 d after adoptive transfer. (C)Ex- − − perimental scheme for D–F: Either 5 × 106 IgHEL or IgHEL×PD-1 / cells were transferred into WT (white bars in D–F) or NOH mice (black bars in D–F), which were immunized with HEL in Alum on the next day. (D) Absolute numbers of surviving B220+ IgMa+ HEL+ IgHEL cells 3 d after adoptive transfer. (E) Proportion of Ki67+ proliferating B220+ IgMa+ HEL+ IgHEL B cells. (F) Representative bar graph of cleaved caspase 3 expression on IgHEL B cells. *P < 0.05 (ANOVA and Bonferroni). Data are representative of two experiments with three mice in each group.

cells (Fig. 2F). There were about 50% more apoptotic WT IgHEL cells after 3 d in NOH mice, and although this small increase was reproducible, it did not reach statistical signiﬁcance (Fig. 2F). Taken together, these ﬁndings supported the view that Tregs suppressed B cells in vivo only when the latter expressed PD-1 and that inhibition of B-cell proliferation and also apoptosis induction were involved.

+ Ag-SpecificTregs Are Sufficient for Suppression of PD-1 B Cells in Vivo. Next we wondered whether Tregs might provide the PD-1 ligands that triggered PD-1 on B cells. We first demonstrated that Tregs expressed PDL-1/2 and up-regulated these ligands after immunization with autoantigen (Fig. S6 A–C). T-cell levels of PDL-2 were very small, consistent with previous studies (21), but up- regulated on Tregs from PDL-1–deficient mice (Fig. S6C), supporting the notion that PDL-2 becomes functionally relevant when PDL-1 is lacking (30). Expression of mRNA for other inhibitory molecules described as suppressive mediators of Tregs like FasL, granzyme B, or perforin remained unchanged between groups or showed a reduction compared with Tregs from WT mice (Fig. S7). Studying the functional relevance of such PD-1 ligand expression was theoretically possible by adoptively transferring Tregs fi fi + lacking such ligands. This, however, first required establishing Fig. 3. Ag-speci cTregs are suf cient for suppression of PD-1 B cells in vivo. (A) Experimental scheme: 1 × 106 T or Th cells were transferred into IgHEL that Tregs are able to suppress B cells after adoptive transfer regs (B) or IgHELxPD-1−/− (C) mice on day −1, which were then immunized on because our previous studies had shown only that Tregs are es- days 0, 7, and 14 with HEL in Alum i.p. (B) HEL-specific IgMa serum Ab titers sential, not that they are sufficient, for such suppression (6, 7). To − − on day 21. (C) HEL-specific serum IgMa titers in IgHEL or IgHEL×PD-1 / mice this end, we isolated Tregs from NOH or WT mice and injected after adoptive transfer of Tregs or Th cells. The Ab titers of unimmunized them into IgHEL mice before immunization (experimental control mice are shown as background. *P < 0.05; **P < 0.01; ***P < 0.001 scheme in Fig. 3A). On day 21, HEL-specific IgMa titers were (ANOVA and Bonferroni). Data are representative of two experiments with reduced in those IgHEL mice that had received NOH Tregs,but four mice in each group.

Gotot et al. PNAS Early Edition | 3of6 Downloaded by guest on September 30, 2021 − − IgHEL×PD-1 / cells were 10-fold higher than in recipients of WT cells (Fig. 5F, right pair of bars), indicating that PD-1 on B cells alone was essential and sufficient to suppress auto-Ab formation almost completely. Also after transfer into PD-1–competent NOH mice, PD-1+ B cells were prevented from producing auto-Ab (Fig. 5F, left pair of bars). In these recipients, IgHEL cells produced even somewhat more auto-Ab than after − − transfer into NOH×PD-1 / mice (Fig. 5F), excluding a contribution of PD-1 on other host cells to autoantibody inhibition in our system because in this case less auto-Ab should have been produced. When we examined the numbers of autoreactive B cells surviving 14 d after transfer, results were similar (Fig. 5G). In PD-1–deficient NOH recipient mice, PD-1 deficiency in IgHEL cells and Treg depletion increased B-cell survival to a similar extent, and the combination was not synergistic (Fig. 5H), supporting the conclusion that PD-1 on B cells, and only on B cells, was the target molecule of Tregs. These findings showed

Fig. 4. PDL-1/2 expression by Tregs is required for B-cell suppression. (A)Ex- 6 −/− perimental scheme: 1 × 10 Tregs from spleens of NOH or NOH×PDL-1 mice were transferred into IgHEL mice that were immunized with HEL/Alum i.p. and injected with PDL-2–blocking TY25 Ab on days 0, 7, and 14. The Ab titers of unimmunized control mice were shown as background. (B) HEL-speciﬁc serum IgMa titers on day 21. (C) Proportion of cells producing anti-HEL IgMa Ab determined by ELISpot. (D) Proportion of apoptotic IgHEL cells among splenocytes on day 21. *P < 0.05; **P < 0.01 (ANOVA and Bonferroni). Data are representative of two experiments in groups of four mice each.

− − − − 1 / .PDL-2 / mice. This restored autoantibody production to levels in control mice (Fig. 4B), indicating that B cells were suppressed by PDL-1 on Tregs and by PDL-2 on unidentiﬁed cells. These PDL-2+ cells cannot be host cells because in that case autoantibody production should have been restored in mice treated with TY25 antibody but not injected with Tregs, yet this was not so (Fig. 4B). And if host cells indirectly used PDL-2 to render the transferred Tregs suppressive, then TY25 antibody should have disabled PDL-1–competent Tregs. Thus, our results imply that Tregs used both PD-1 ligands to suppress B cells in our system. Mech- anistic analysis revealed that the lack of both ligands augmented the number of antibody-forming cells (Fig. 4C) and reduced B-cell apoptosis (Fig. 4D), consistent with the effects of PD-1 signaling on B cells (Fig. 2 E and F).

B-Cell Suppression by SpecificTregs Does Not Require Intermediate Th Cells. The above findings established that Tregs need to express PD-1 ligands and that B cells need to express PD-1 in our system, consistent with direct cross talk between these cells. However, there is a theoretical scenario where intermediate Th cells are + still involved: Tregs might suppress PD-1 Th cells using PD-1 ligands, which then up-regulate PDL-1 and/or PDL-2 to suppress PD-1+ B cells. We experimentally addressed this possibility first by analyzing PD-1 expression on Th cells and Tregs. PD-1 was unchanged on these cells after immunization or Treg depletion (Fig. S8 A–C). To test for functional relevance, we generated −/− NOH×PD-1 mice, in which the lack of PD-1 on all cells, in- 6 Fig. 5. Tregs directly suppress B cells. (A) Experimental scheme: Either 5 × 10 − − cluding Th cells, precluded the scenario described above. When IgHEL (white bars in B–D, F,andG) or IgHEL×PD-1 / (black bars in B–D, F,and − − we adoptively transferred IgHEL cells expressing or lacking PD-1 G) B cells were transferred into PD-1–competent NOH or into NOH×PD-1 / − − into NOH×PD-1 / mice (experimental scheme in Fig. 5A), after mice, which were immunized with HEL/Alum on the next day. (B) Absolute 3 d we observed lower numbers, reduced proliferation, and in- numbers of IgHEL cells in the spleen 3 d after immunization. (C) Proportions creased apoptosis of PD-1–competent, but not of PD-1–deficient of proliferating IgHEL cells. (D) Proportion of apoptotic IgHEL cells at 18 h B–D + after immunization. (E–G) Same experiment as in A except that mice were IgHEL cells (Fig. 5 ). Thus, PD-1 IgHEL cells were sup- fi pressed even when all other cells in the system, including Th cells, immunized twice in a weekly interval. (F) HEL-speci c IgMa serum Ab titers – fi B–D on day 14. (G) Numbers of antibody-forming cells in the spleens measured by were PD-1 de cient (Fig. 5 , right pair of bars). ELISpot. (H) Numbers of HEL-specific antibody-forming cells in the spleens of −/− Finally, we examined auto-Ab titers and numbers of auto- Treg-depleted NOH–PD-1 mice measured by ELISpot. *P < 0.05; **P < 0.01; reactive B cells in this system after 14 d (experimental scheme in ***P < 0.001 (ANOVA and Bonferroni). Data are representative of two −/− Fig. 5E). Anti-HEL IgMa titers in NOH×PD-1 recipients of experiments in groups of three to five mice.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1201131109 Gotot et al. Downloaded by guest on September 30, 2021 that PD-1 expression by cells other than autoreactive B cells, Our findings identified PD-1 ligands as suppressive effector including Th cells, was dispensable for suppression, implying that molecules of Tregs. It has been previously reported that PDL-1 Tregs directly suppressed B cells in vivo. affects the development of Tregs (27), raising the question whether the PDL-1–deficient Tregs that we adoptively transferred might carry Discussion developmental defects. However, this cannot explain our observa- Tregs potently inhibit autoreactive T and B cells, but the un- tions because PDL-1–deficient Tregs were still able to suppress, derlying molecular mechanisms remain unclear, especially those unless also PDL-2 was blocked. Although PDL-2 possesses high that suppress B cells. Here we show that Tregs control autoreactive affinity to PD-1 (21), its levels were very low on PDL-1–competent fi B cells speci c for tissue auto-Ag through the suppressive surface Tregs. However, it was higher on PDL-1–deficient Tregs,whichmay molecule PD-1. Others have previously shown that follicular Th indicate that Tregs use PDL-2 only when they lack PDL-1. cells express high amounts of PD-1 (37) and that PD-1 inhibits Th An open question in our system concerns the intrasplenic site cells (38), suggesting that indirect B-cell suppression, by curbing where Tregs and autoreactive B cells meet. Tregs and B cells make help for auto-Ab production, may play a role. Cell culture studies contacts at the T-cell–B-cell border and within germinal centers – + by others had hinted at the possibility of direct suppression (10 (9). CXCR5 Tregs were very recently shown to enter germinal 12), but providing in vivo evidence required an experimental centers to suppress B cells, affinity maturation of antibodies, and system by which either the direct or the indirect route of sup- plasma cell differentiation (13, 14). Further studies are necessary pression could be incapacitated. This became possible by our to identify the site of suppression in our model. identification of PD-1 as the main suppressor of auto-Ab forma- PD-1–blocking antibodies are being discussed for clinical ap- tion. Performing a series of adoptive transfer experiments, we plication in cancer (43), persistent viral hepatitis (44), and AIDS found that B cells, and only B cells, needed to express PD-1 and (45) as a means to invigorate suppressed or exhausted T-cell that Tregs needed to express PD-1 ligands, implying direct com- responses. Also antibody responses were found to be increased munication between Tregs and B cells. On the basis of our findings after PD-1 blockade, which our results suggest to be due to in- i ii ( ) that direct suppression entirely depended on PD-1, ( ) that hibition of Treg function. Our findings also suggest that auto-Ab PD-1 reduced auto-Ab production by 65–75%, and (iii) that Tregs production might occur as a side effect of such blockade. A reduced such production by 85–90%, it can be calculated that 65– current preclinical study showed little evidence of auto-Ab 85% of the Treg effect in our system resulted from PD-1–mediated generation (45), which may be explained by the short duration of direct suppression. The PD-1–independent part of the Treg effect antibody treatment, by maintained PD-1–independent suppres- may occur by indirect suppression, for example, by inhibiting Th sion of autoreactive Th cells, or by the absence of adjuvants such IMMUNOLOGY fi cells. Thus, our ndings are not inconsistent with previous studies as Alum in our study. Finally, our study suggests that Tregs that showing indirect suppression in other systems (8, 39). were caused to express PD-1 ligands might allow treatment of PD-1 expressed by follicular Th cells has previously been shown antibody-mediated autoimmune disease. Given that Tregs spe- to improve antibody formation against foreign antigens by pro- cifically suppressed auto-Ab formation, this approach should be moting the survival of B cells expressing PD-1 ligands (30). This associated with fewer side effects than therapies targeting all B positive effect was difficult to reconcile with the high auto-Ab cells, such as depletion with CD20-specific antibodies. titers and the lupus symptoms observed in PD-1 knockout mice (21, 31). In our study, Tregs expressing PD-1 ligands acted nega- Materials and Methods + tively on autoreactive PD-1 B cells, explaining the phenotype of Mice and Reagents. All mice were bred and maintained under specific PD-1–deficient mice (21, 38). This is no contradiction because pathogen-free conditions at the central animal facility of the University Clinic of Bonn (House of Experimental Therapy) and used at 8–14 wk of age Tregs will operate only in responses against self-antigens and pre- vail over the positive effect on follicular Th cells. Thus, the according to German animal experimentation laws. All studies were ap- proved by an external review board (Bezirksregierung Köln, Cologne). All mechanism described here will not operate in antibody generation fi + against foreign antigens. reagents, if not otherwise speci ed, were from Sigma-Aldrich. CD25 cells were depleted by injecting 300 μg of PC61 antibody (purified from hybrid- Also, in humans, there is evidence that PD-1 suppresses an- oma supernatant) i.p. For in vivo blockage of PD-1 or PDL-2, 250 μg of RMP1- tibody-mediated autoimmunity. For example, a single-nucleotide 14 (18) or 250 μg of TY25 antibodies (34), respectively, were injected i.p. at polymorphism of the PDCD1 gene that incapacitated PD-1 ex- weekly intervals. Mice were immunized i.p. with 10 μg antigen in aluminum pression was linked to the presence of rheumatoid factors and hydroxide at a 1:1 ratio in 200 μL total volume at weekly intervals. rheumatoid arthritis symptoms and to lupus erythematosus with nephritis (40). Our findings suggest that B cells lacking PD-1 Isolation and Transfer of Primary B-Cell and Tregs. B cells were isolated from functionality in these individuals might have been unable to re- IgHEL (MD4) mice, and Tregs or Th cells were isolated from NOH or C57BL/6 mice. Single-cell suspensions were treated with erythrocyte lysis buffer (146 ceive suppressive signals from Tregs. mM NH4Cl, 10 mM NaHCO3 and 2 mM EDTA) to remove red blood cells. B B-cell apoptosis induced by Tregs has been previously shown fi in vitro and occurred by granzyme B/perforin-mediated cell lysis cells were further puri ed by magnetic cell separation using the negative B-cell isolation kit from Miltenyi (purity ∼97%). T were purified using the (10, 12) or by Fas (11). Our findings provide in vivo evidence for regs Treg isolation kit from Miltenyi (purity ∼90%). Treg-induced B-cell apoptosis by PD-1. Engagement of PD-1 on B cells has been shown to inhibit B-cell receptor (BCR) signaling ELISA and ELISpot. Antigen-specific serum IgG, IgM, and IgMa titers were by recruiting SHP-2 to its phosphotyrosine and dephosphor- measured by ELISA, and antibody-forming cells (AFC) numbers by ELISpot as ylating key signal transducers of BCR signaling (41), which may described (6). deprive B cells of survival signals. It remains to be clarified whether this molecular mechanism applies in our system. Real-Time PCR. RNA was isolated using TRIzol. Contaminating DNA was elimi- Throughout this study, the accrual of apoptotic B cells was less nated with DNase. One microgram of RNA was converted to cDNA using High prominent than the increase of autoantibody titers or of viable Capacity Reverse Transcription kit. The SYBR Green-PCR was done on an Abi autoreactive B cells. This may be due to the rapid in vivo clear- Prism 7900HT System. One microliter of cDNA was used with the following ance of apoptotic cells in healthy organisms (42), which prevented settings: 40 cycles of 15 s of denaturation at 95 °C and 1 min of primer annealing and elongation at 60 °C. Samples were run in triplicates, and GAPDH served as accumulation of large numbers of apoptotic B cells. However, fi internal control for normalization. Primers were designed by Primer Express even if apoptosis induction occurred slowly, it may still be suf - Software and purchased from Invitrogen. Sequences were the following: PD- cient for peripheral tolerance induction because self-antigens are 1—forw. 5′AAGCTTATGTGGGTCCGGC′3 and rev. 5′GGATCCTCAAAGAGGCC′ normally always present, allowing continual incapacitation of 3; Fas—forw. 5′TCTGGTGCTTGCTGGCTCAC′3 and rev. 5′CCATAGGCGATTTC- autoreactive B cells. TGGGAC′3; PDL-1—forw. 5′TGCTTCTCAATGTGACC′3 and rev. 5′GGAACAAC-

Gotot et al. PNAS Early Edition | 5of6 Downloaded by guest on September 30, 2021 AGGATGGAT′3; PDL-2—forw. 5′TGACCCTCTGAGTTGGATGGA′3 and rev. 5′ determine absolute cell numbers, 1 × 105 CaliBRITE APC beads were added GCCGGGATGAAAGCATGA′3; FasL—forw.5′CGGTGGTATTTTTCATGGTTCTGG before flow cytometry as an internal reference. For intracellular staining, ′3 and rev. 5′CTTGTGGTTTAGGGGCTGGTTGTT′3; granzyme B—forw. 5′GGG- cells were stained for surface makers, fixed with 2% (vol/vol) para- AAGATGAAGATCCTCCTGC′3 and rev. 5′TGATCTCCCCTGCCTTTGTC′3; and formaldehyde, permeabilized with 0.25% Triton X in PBS, and stained with — ′ ′ ′ perforin forw.5 CCCTAGGCCAGAGGCAAAC 3 and rev. 5 AAAATTGGCT- Foxp3- or Ki67-specific antibodies. Results were analyzed with FlowJo soft- ′ ACCTTGGAGTGG 3. ware (TreeStar).

Flow Cytometry. Cells were stained in PBS 0.1% BSA 0.1% sodium azide for 20 Statistical Analysis. Results are expressed as mean ± SEM; *P < 0.05; **P < fl min on ice using uorochrome-labeled monoclonal antibodies from eBio- 0.01; ***P < 0.001. Comparisons were made using ANOVA test with Bon- sciences if not otherwise specified: CD4 (GK 1.5), CD25 (PC61), DO11.10 TCR ferroni posttest using Prism 4 software (Graphpad Software). (KJ1-26), Foxp3 (FJK-16s), B220 (RA3-6B2), IgMa (DS-1; BD Bioscience), PD-1 (J43), PDL-1 (MIH5; BD Bioscience), PDL-2 (TY25), Fas (15A7), FasL (MFL3), and ACKNOWLEDGMENTS. We thank Liping Chen and Linda Diehl for PD-1 and Ki67 (SP6; Thermo Scientific). Apoptosis was determined by a caspase 3/7 fi PDL-1 knockout mice, Tim Sparwasser for DEREG mice, and Achmet Imam FLICA kit (Immunochemistry Technologies). To identify Ag-speci c B cells, Chasan for help with primer design. The authors acknowledge support by fl soluble HEL was conjugated to Alexa647 uorochrome with a commercial kit the Central Animal Facilities and the Flow Cytometry Core Facility at the (Invitrogen) and used at 2.5 μg/mL. Dead cells were excluded by Hoechst Institutes of Molecular Medicine and Experimental Immunology. I.L.-P. and 33258 or 7-AAD and analyzed on a BDCanto II (Becton Dickinson). To C.K. were supported by Deutsche Forschungsgemeinschaft Grant Lu1387/2-1.

1. Frey O, et al. (2005) The role of regulatory T cells in antigen-induced arthritis: Ag- 25. Freeman GJ, et al. (2000) Engagement of the PD-1 immunoinhibitory receptor by gravation of arthritis after depletion and amelioration after transfer of CD4+CD25+ T a novel B7 family member leads to negative regulation of lymphocyte activation. J cells. Arthritis Res Ther 7:R291–R301. Exp Med 192:1027–1034. 2. Reddy J, et al. (2004) Myelin proteolipid protein-specific CD4+CD25+ regulatory cells 26. Latchman Y, et al. (2001) PD-L2 is a second ligand for PD-1 and inhibits T cell acti- mediate genetic resistance to experimental autoimmune encephalomyelitis. Proc Natl vation. Nat Immunol 2:261–268. Acad Sci USA 101:15434–15439. 27. Francisco LM, Sage PT, Sharpe AH (2010) The PD-1 pathway in tolerance and auto- 3. Salomon B, et al. (2000) B7/CD28 costimulation is essential for the homeostasis of the immunity. Immunol Rev 236:219–242. CD4+CD25+ immunoregulatory T cells that control autoimmune diabetes. Immunity 28. Greenwald RJ, Latchman YE, Sharpe AH (2002) Negative co-receptors on lymphocytes. 12:431–440. Curr Opin Immunol 14:391–396. 4. Bystry RS, Aluvihare V, Welch KA, Kallikourdis M, Betz AG (2001) B cells and pro- 29. Baecher-Allan C, Brown JA, Freeman GJ, Hafler DA (2003) CD4+CD25+ regulatory cells fessional APCs recruit regulatory T cells via CCL4. Nat Immunol 2:1126–1132. from human peripheral blood express very high levels of CD25 ex vivo. Novartis 5. Seo SJ, et al. (2002) The impact of T helper and T regulatory cells on the regulation of Found Symp, 252: 67–88; discussion 88–91, 106–114. anti-double-stranded DNA B cells. Immunity 16:535–546. 30. Good-Jacobson KL, et al. (2010) PD-1 regulates germinal center B cell survival and the 6. Ludwig-Portugall I, Hamilton-Williams EE, Gotot J, Kurts C (2009) CD25+ T(reg) spe- formation and affinity of long-lived plasma cells. Nat Immunol 11:535–542. cifically suppress auto-Ab generation against pancreatic tissue autoantigens. Eur J 31. Nishimura H, Minato N, Nakano T, Honjo T (1998) Immunological studies on PD-1 Immunol 39:225–233. deficient mice: Implication of PD-1 as a negative regulator for B cell responses. Int 7. Ludwig-Portugall I, Hamilton-Williams EE, Gottschalk C, Kurts C (2008) Cutting edge: Immunol 10:1563–1572. CD25+ regulatory T cells prevent expansion and induce apoptosis of B cells specificfor 32. Heymann F, et al. (2009) Kidney dendritic cell activation is required for progression of tissue autoantigens. J Immunol 181:4447–4451. renal disease in a mouse model of glomerular injury. J Clin Invest 119:1286–1297. 8. Fields ML, et al. (2005) CD4+ CD25+ regulatory T cells inhibit the maturation but not 33. Kurts C, Heymann F, Lukacs-Kornek V, Boor P, Floege J (2007) Role of T cells and the initiation of an autoantibody response. Journal Immunol 175(7):4255–4264. dendritic cells in glomerular immunopathology. Semin Immunopathol 29:317–335. 9. Lim HW, Hillsamer P, Banham AH, Kim CH (2005) Cutting edge: Direct suppression of 34. Yamazaki T, et al. (2002) Expression of programmed death 1 ligands by murine T cells B cells by CD4+ CD25+ regulatory T cells. J Immunol 175:4180–4183. and APC. J Immunol 169:5538–5545. 10. Iikuni N, Lourenço EV, Hahn BH, La Cava A (2009) Cutting edge: Regulatory T cells 35. Kao C, et al. (2011) Transcription factor T-bet represses expression of the inhibitory directly suppress B cells in systemic lupus erythematosus. J Immunol 183:1518–1522. receptor PD-1 and sustains virus-specific CD8+ T cell responses during chronic in- 11. Janssens W, et al. (2003) CD4+CD25+ T cells lyse antigen-presenting B cells by Fas-Fas fection. Nat Immunol 12:663–671. ligand interaction in an epitope-specific manner. J Immunol 171:4604–4612. 36. Yamazaki T, et al. (2005) Blockade of B7-H1 on macrophages suppresses CD4+ T cell 12. Zhao DM, Thornton AM, DiPaolo RJ, Shevach EM (2006) Activated CD4+CD25+ T cells proliferation by augmenting IFN-gamma-induced nitric oxide production. J Immunol selectively kill B lymphocytes. Blood 107:3925–3932. 175:1586–1592. 13. Chung Y, et al. (2011) Follicular regulatory T cells expressing Foxp3 and Bcl-6 suppress 37. Chtanova T, et al. (2004) T follicular helper cells express a distinctive transcriptional germinal center reactions. Nat Med 17:983–988. profile, reflecting their role as non-Th1/Th2 effector cells that provide help for B cells. 14. Linterman MA, et al. (2011) Foxp3+ follicular regulatory T cells control the germinal J Immunol 173(1):68–78. center response. Nat Med 17:975–982. 38. Wong M, La Cava A, Singh RP, Hahn BH (2010) Blockade of programmed death-1 in 15. Campbell DJ, Koch MA (2011) Treg cells: Patrolling a dangerous neighborhood. Nat young (New Zealand black × New Zealand white)F1 mice promotes the activity of Med 17:929–930. suppressive CD8+ T cells that protect from lupus-like disease. J Immunol 185: 16. Sakaguchi S, Wing K, Onishi Y, Prieto-Martin P, Yamaguchi T (2009) Regulatory T cells: 6563–6571. How do they suppress immune responses? Int Immunol 21:1105–1111. 39. Guay HM, Larkin, J, III, Picca CC, Panarey L, Caton AJ (2007) Spontaneous autoreactive 17. Shevach EM (2009) Mechanisms of foxp3+ T regulatory cell-mediated suppression. memory B cell formation driven by a high frequency of autoreactive CD4+ T cells. J Immunity 30:636–645. Immunol 178:4793–4802. 18. Agata Y, et al. (1996) Expression of the PD-1 antigen on the surface of stimulated 40. Prokunina L, et al. (2004) The systemic lupus erythematosus-associated PDCD1 poly- mouse T and B lymphocytes. Int Immunol 8:765–772. morphism PD1.3A in lupus nephritis. Arthritis Rheum 50:327–328. 19. Ishida Y, Agata Y, Shibahara K, Honjo T (1992) Induced expression of PD-1, a novel 41. Okazaki T, Maeda A, Nishimura H, Kurosaki T, Honjo T (2001) PD-1 immunoreceptor member of the immunoglobulin gene superfamily, upon programmed cell death. inhibits B cell receptor-mediated signaling by recruiting src homology 2-domain- EMBO J 11:3887–3895. containing tyrosine phosphatase 2 to phosphotyrosine. Proc Natl Acad Sci USA 98: 20. Sharpe AH, Wherry EJ, Ahmed R, Freeman GJ (2007) The function of programmed cell 13866–13871. death 1 and its ligands in regulating autoimmunity and infection. Nat Immunol 8: 42. Muñoz LE, Lauber K, Schiller M, Manfredi AA, Herrmann M (2010) The role of de- 239–245. fective clearance of apoptotic cells in systemic autoimmunity. Nat Rev Rheumatol 6: 21. Keir ME, Butte MJ, Freeman GJ, Sharpe AH (2008) PD-1 and its ligands in tolerance 280–289. and immunity. Annu Rev Immunol 26:677–704. 43. Stagg J, et al. (2011) Anti-ErbB-2 mAb therapy requires type I and II interferons and 22. Blackburn SD, et al. (2009) Coregulation of CD8+ T cell exhaustion by multiple in- synergizes with anti-PD-1 or anti-CD137 mAb therapy. Proc Natl Acad Sci USA 108: hibitory receptors during chronic viral infection. Nat Immunol 10(1):29–37. 7142–7147. 23. Mueller SN, et al. (2010) PD-L1 has distinct functions in hematopoietic and non- 44. Radziewicz H, Hanson HL, Ahmed R, Grakoui A (2008) Unraveling the role of PD-1/PD- hematopoietic cells in regulating T cell responses during chronic infection in mice. J L interactions in persistent hepatotropic infections: Potential for therapeutic appli- Clin Invest 120:2508–2515. cation? Gastroenterology 134:2168–2171. 24. Diehl L, et al. (2008) Tolerogenic maturation of liver sinusoidal endothelial cells 45. Velu V, et al. (2009) Enhancing SIV-specific immunity in vivo by PD-1 blockade. Nature promotes B7-homolog 1-dependent CD8+ T cell tolerance. Hepatology 47:296–305. 458:206–210.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1201131109 Gotot et al. Downloaded by guest on September 30, 2021