Hepatic Stellate Cells Directly Inhibit B Cells via Programmed Death–Ligand 1 Free

The liver is considered an immunoprivileged organ (1). In rodents, almost all liver allografts are spontaneously accepted without rejection (2–4); in humans, patients who have received liver allograft transplantation generally require lower doses of immunosuppressive drugs to prevent rejection than patients who have received other transplanted organs, such as kidney or heart (5). Also, ∼20–40% of patients with liver transplants can gradually be weaned off immunosuppressive drugs without rejecting their new liver (6, 7). In addition, the liver must maintain overall homeostasis while physiologically it is under chronic exposure to many foreign Ags (e.g., food and microbial Ags) and stimulants (e.g., LPSs) (8, 9). Interestingly, although whole-liver transplantation can be tolerated, transplanted hepatocytes are rejected (10), suggesting that nonparenchymal cells in the liver are critical for maintaining the immunoprivileged status of the liver.

Hepatic stellate cells (HSCs) account for about one third of such nonparenchymal cells (11). Upon activation, HSCs secrete a number of local growth factors (12, 13) and matrix metalloproteinases associated with liver repair and fibrosis, as well as systemic acute-phase proteins (14, 15). Although the role of HSCs in liver injury and fibrosis has been investigated extensively, the potential role of HSCs in liver immunoregulation and their underlying mechanisms remains understudied. We reported previously that HSCs directly inhibit T cells through programmed death–ligand 1 (PD-L1) on the HSC cell surface (16) and that HSCs also indirectly suppress the adaptive immune system by inducing the propagation of myeloid-derived suppressor cells from hematopoietic stem cells (17). However, whether HSCs have any direct effect on B cells, and if this is the case, what the underlying mechanisms might be, remain unknown. B cells reside in the liver together with T cells under physiological conditions (18–20). In addition to producing Abs, in response to pathogens and associated with autoimmune disorders, activated B cells are the major source of inflammatory cytokines, including IL-6, in the lymphoid organs (21) and serve as APCs to promote T cell responses (22). It is established that B cells play important roles in the traditional humoral immunity–mediated diseases, as well as in many diseases conventionally believed to be mediated by T cells (23). Using isolated mouse primary HSCs, we found in vitro and in vivo evidence suggesting that HSCs directly inhibit B cells, in which the HSC-expressed PD-L1 is integrally involved.

C57BL/6 mice were purchased from The Jackson Laboratory (Bar Harbor, ME), and PD-L1–knockout (KO) mice (on C57BL/6 background) were kindly provided by Lieping Chen (Yale University, New Haven, CT) (24). All mice were housed in Cleveland Clinic’s Biological Resources Unit in accordance with guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care International. The animal experimental protocols were approved by the Institutional Animal Care and Use Committee of the Cleveland Clinic. Mice were used at 8–16 wk of age in all experiments.

HSCs were isolated from mouse liver and cultured in RPMI 1640 medium supplemented with 20% FBS (Life Technologies, Grand Island, NY) in 5% CO₂ in air at 37°C for 14–21 d, following protocols well established in the laboratory, as previously described (16, 17, 25, 26). Purity of the isolated HSCs was generally >95%, as assessed using α-smooth muscle actin as a marker (Supplemental Fig. 1), followed by flow cytometry analysis. All of the flow cytometry experiments were performed using a BD FACSCalibur flow cytometer and FlowJo version 7 software.

B cells (>98% pure) were purified by negative selection (STEMCELL Technologies, Vancouver, BC, Canada) from splenocytes (Supplemental Fig. 2). The purified B cells were activated by incubation with 10 μg/ml anti-IgM IgGs (Jackson ImmunoResearch Laboratories, West Grove, PA) or 10 μg/ml anti-CD40 IgGs (BioLegend, San Diego, CA), together with 100 U/ml IL-4 (PeproTech, Rocky Hill, NJ), and then cocultured with different numbers of HSCs. After 24 h of incubation, B cells were assessed for the expression of activation markers CD69 and CD86 by flow cytometry after staining with 1 μg/ml PE–anti-mouse CD69 or FITC–anti-mouse CD86 mAbs (BioLegend).

The proliferation of activated B cells was assessed by CFSE dilution assay and/or BrdU incorporation assay. For CFSE-based proliferation assays, purified B cells were incubated with CFSE at 37°C for 10 min and then activated by incubation with 10 μg/ml anti-IgM IgGs or anti-CD40 IgGs, together with 100 U/ml IL-4. After 72 h, proliferation of the activated B cells was assessed by flow cytometric analysis of the CFSE dilution on B cells. For BrdU incorporation–based proliferation assays, BrdU was added to the HSC/B cell cocultures 1 d before the assay and then suspended B cells were washed gently and collected to measure their proliferation (BrdU incorporation) using a BrdU ELISA Kit (Roche Applied Science, Indianapolis, IN), following the manufacturer’s protocols. At the same time, culture supernatants were collected to measure levels of IL-6, IgG, and/or IgM by ELISA, following the manufacturer’s protocols.

HSCs were cultured at the bottom of the 24-well Transwell culture system (BD Biosciences, San Jose, CA) in 500 μl media; anti-CD40/IL-4–activated and CFSE-labeled B cells were cultured in the inserts, which were separated from the bottom cells by a membrane with 0.1 μM pore size. After 72 h of culture, B cells were analyzed for proliferation by flow cytometry, and supernatants were collected to measure the levels of IL-6 produced by the activated B cells.

Mice were anesthetized, and a transverse upper abdominal incision was used to expose the spleen. The splenic artery was visually identified and separated from the mesenteric adipose tissues. After closing off the proximal artery using a microvascular clamp clip, the artery was punctured by a sterile 32-Ga needle. Using the needle tip as a canal, a tip-modified 10-0 suture “guidewire” was inserted into the artery. Then, using a wire catheter–exchange technique, the modified catheter was placed into the lumen. After this step, 0.2 × 10⁶ wild-type (WT) or PD-L1–KO HSCs in 50 μl sterile PBS were injected into the splenic artery. After injection, the proximal side of the injected artery was ligated, and 1 ml warm 0.9% saline was injected into the abdominal cavity to replenish fluid losses and prevent dehydration. The abdomen and skin were closed in layers with running 4/0 silk sutures or wound clips. Sham-operated mice that had not received an injection of HSCs were included as controls. To demonstrate the distribution of the injected HSCs in the spleen, the same numbers of HSCs labeled with Vybrant Dil Cell-Labeling Solution (Life Technologies, Carlsbad, CA) were injected into the mouse; after sacrifice, the spleen was collected to make cryosections for examination under a fluorescence microscope (Leica Microsystems, Buffalo Grove, IL).

Each HSC-injected or sham-operated mouse was immunized by i.p. injection of 10 μg 4-hydroxy-3-nitrophenylacetyl (NP)-Ficoll (Biosearch Technologies, Petaluma, CA). Serum samples were collected by tail bleeding, and NP-specific IgG and IgM levels in the sera were measured by ELISA using plates coated with NP-BSA (Biosearch Technologies), following previously described protocols.

Serum samples were collected from immunized mice at days 3, 9, and 14, and titers of NP-specific IgM and IgG were measured by ELISA, using methods that were described previously. In brief, serum samples were diluted 1:500 in PBS and added to wells of a 96-well plate coated with 5 μg/ml NP-BSA (Biosearch Technologies). After 2 h of incubation, HRP-conjugated rabbit anti-mouse IgG (1:4000) or anti-mouse IgM (1:4000) was added and incubated for an additional hour. The titers of NP-specific IgGs and IgMs in the sera were assessed by measuring OD₄₅₀ after development using tetramethylbenzidine (Thermo Scientific, Rockford, IL).

All data were analyzed using GraphPad Prism software (GraphPad, La Jolla, CA). To combine results from different proliferation or B cell–activation (CD69 and CD86 upregulation) experiments, the following equation was used to normalize the data before combination: relative proliferation (or activated B cells) = [(A − B)/(C − B)] × 100%, where A represents the mean experimental proliferation or activated B cell percentages, B represents the basal cell proliferation or activation (resting B cells without stimuli), and C represents the maximum proliferation or activation (activated B cells alone, 100%). To determine whether groups were statistically different, results were compared using the Student t test. A p value <0.05 was considered significant.

To determine whether HSCs have an effect on B cells, we first isolated HSCs from WT C57BL/6 mice and cocultured different numbers of cells with 0.5 × 10⁶ purified B cells from WT C57BL/6 mice in the presence of anti-CD40/IL-4 (to activate the B cells). After 24 h of incubation, we assessed B cell–activation status by analyzing the activation markers CD69 and CD86. These experiments showed that, consistent with previous reports (27, 28), expression of both CD69 and CD86 was significantly upregulated on B cells after activation. However, upregulation of both CD69 and CD86 (especially CD86) was significantly suppressed in proportion to the numbers of HSCs in the cocultures, suggesting that HSCs inhibited B cell activation (Fig. 1A–D).

We next tested whether HSCs inhibited the proliferation of, and Ig production from, activated B cells, as well as their IL-6 secretion. We cultured different numbers of HSCs with CFSE-labeled B cells in the presence of anti-IgM IgG/IL-4 or anti-CD40 IgG/IL-4 (to activate B cells) and then assessed the proliferation of the activated B cells by measuring CFSE dilutions on B cells using flow cytometric analysis and measuring BrdU incorporation in these B cells using a BrdU ELISA. We found that HSCs inhibited the proliferation of activated B cells in a dose-dependent manner, as assessed by the CFSE-based (Fig. 2A–D) and BrdU-based (Fig. 2E, 2F) B cell–proliferation assays. We also measured levels of IL-6, IgM, and IgG in the culture supernatants by ELISA. Because the anti-IgM Abs that we used to activate B cells interfere with sequential measurements of IgM produced by the activated B cells, and because this B cell–activation method leads to little IgG production, we measured levels of IgG/IgM only in cocultures with B cells activated by anti-CD40 IgG/IL-4. These ELISA experiments revealed that, in addition to diminished proliferation, activated B cells showed reduced production of IL-6, IgM, and IgG in proportion to the numbers of HSCs in the cocultures (Fig. 2G–J). These and the studies described above, taken together, show that HSCs inhibit B cell activation, proliferation, and IL-6/Ig production, demonstrating that HSCs have a previously unknown role in the direct inhibition of B cells.

To explore the mechanism by which HSCs directly inhibit B cells, we carried out Transwell experiments in which HSCs were cultured in wells at the bottom, with anti-IgM/IL-4–activated B cells cultured in top inserts that allow only soluble factors to be exchanged, without direct cell–cell contact. We then measured B cell proliferation and levels of IL-6 produced by the activated B cells in different wells at 72 h to assess the efficacy of B cell inhibition. These assays found that, without direct contact of HSCs with B cells, the inhibitory effect of the HSCs was significantly reduced, as indicated by more proliferation of the activated B cells (Fig. 3A, 3B) and higher levels of IL-6 produced (Fig. 3C) in Transwells than in wells with direct cell–cell contact. These results suggest that HSC cell surface molecules are important for the direct B cell inhibitory activity of HSCs.

We demonstrated that activated HSCs have elevated levels of PD-L1 on their surface and that these molecules are important for HSCs to suppress T cells via interactions with its ligand, programmed cell death protein 1 (PD-1), on activated T cells (14). It also was reported that PD-1 is present on activated B cells in both mice (29) and humans (30). In light of this previous work, and the above results that HSC cell surface molecules are crucial for HSCs to directly inhibit B cells, we first tested a putative role for PD-L1 on HSCs to directly inhibit B cell activity by blocking PD-L1 on HSCs using a PD-L1 function neutralizing Ab (eBioscience, San Diego, CA). We cocultured HSCs with different numbers of anti-CD40/IL-4–activated or anti-IgM/IL-4–activated B cells in the presence of a PD-L1–blocking mAb or control IgGs and then assessed the efficacy of HSCs in inhibiting the activated B cells by measuring the proliferation of the activated B cells and levels of IL-6 produced by the activated B cells in the culture supernatants. From these experiments, we found that, under both B cell–activation protocols, blocking PD-L1 reduced the efficacy of HSCs in inhibiting both the proliferation of (Fig. 4A–D) and IL-6 production from (Fig. 4E, 4F) the activated B cells; this result suggests that PD-L1 is required for HSCs to directly inhibit B cells.

We next tested the effect of blocking PD-1 in the HSC-mediated B cell–inhibiting experiments. We incubated CFSE-labeled- and anti-CD40/IL-4–activated B cells with WT HSCs in the presence of 10 μg/ml PD-1–blocking IgGs (BioLegend) or control IgGs for 72 h and then analyzed the proliferation of activated B cells by measuring CFSE dilution using flow cytometry. These experiments found that blocking PD-1 significantly improved the proliferation of the activated B cells in the presence of HSCs (Fig. 4G, 4H). These results, taken together, demonstrated that blocking PD-1/PD-L1 interactions by an anti–PD-L1 or an anti–PD-1 mAb diminished the newly discovered B cell–inhibitory activity of HSCs.

In addition to the above-described PD-L1– and PD-1–blocking experiments, we explored the importance of HSCs in B cell inhibition using a genetic approach by comparing HSCs isolated from WT mice and PD-L1–KO mice (both on the C57BL/6 background) in the same B cell–activation marker–upregulation and IL-6/Ig–production assays. These experiments showed that, in accordance with the Ab-blocking experiments, compared with the same numbers of WT HSCs, PD-L1–KO HSCs showed reduced efficacy in inhibiting the upregulation of both CD69 and CD86 on activated B cells (Fig. 5A, 5B) and in reducing the production of IL-6 and/or Igs from the activated B cells (Fig. 5C–F), further confirming that PD-L1 on HSCs is required for the efficient inhibition of B cells.

To determine whether HSCs directly inhibit B cell responses in vivo, we used a model in which mice were immunized by i.p. injection with the T cell–independent Ag NP-Ficoll (31). We first tried to deliver HSCs into the mice by i.v. injection into the tail vein after NP-Ficoll immunization and then to evaluate the effect of HSCs in suppressing the in vivo production of NP-specific IgG/IgM. However, most mice died of pulmonary embolism immediately after tail vein i.v. injection of HSCs (data not shown), presumably as a result of the large size of HSCs. To solve this problem, in view of previous reports that, after i.p. immunization of NP-Ficoll, most of the reactive, IgM/IgG-producing B cells are located within the marginal zone of the spleen (32, 33), we developed a protocol to locally deliver HSCs into the spleen by intrasplenic artery injection. In brief, we anesthetized the mice, performed laparotomy to identify the splenic artery (Fig. 6A, 6B), and injected 0.2 × 10⁶ red Vybrant Dil-labeled HSCs in 50 μl of sterile PBS into the spleen through the splenic artery using a wire-guided catheter-exchange technique. To demonstrate that the HSCs had been delivered successfully into the spleen, at sacrifice, we took out the injected spleen to prepare cryosections and examined them for the presence of HSCs (red cells) under a fluorescence microscope. The red HSCs were distributed throughout the spleen in the red pulp after intrasplenic artery injection (Fig. 6C–H), showing that HSCs can be delivered locally into the spleen by this approach without inducing lethal pulmonary embolism in mice.

After establishing the protocol for local delivery of HSCs into the spleen through intrasplenic artery injection, we tested the efficacy of HSCs in inhibiting B cells in vivo and the role of PD-L1 in this process. We injected each mouse with 0.2 × 10⁶ WT or PD-L1–KO HSCs using our established intrasplenic artery injection method; sham-operated mice without HSC injection were included as controls. At 24 h after the injection, we immunized the mice with NP-Ficoll by i.p. injection and collected blood from the tail vein on days 3, 9, and 14. We measured serum levels of NP-specific IgM and IgG by ELISA. These experiments showed that, on day 3, NP-specific IgMs were barely detectable (data not shown). On day 9 after local delivery of HSCs, immunized mice with injected WT HSCs (but not PD-L1–KO HSCs) showed significantly reduced titers of NP-specific IgMs compared with NP-immunized mice with sham operations (Fig. 7A). By day 14, there was no difference in NP-specific IgM titers among the three groups (Fig. 7B). For IgG measurements, again, there were no detectable NP-specific IgGs on day 3 (data not shown). On day 9, NP-specific IgGs started to be measurable in the sera; compared with the NP-immunized sham mice, NP-immunized mice injected with WT HSCs (but not PD-L1–KO HSCs) showed significantly reduced titers of NP-specific IgGs (Fig. 7C). The differences in NP-specific IgG titers among the groups of mice were even more significant on day 14 (Fig. 7D), demonstrating that HSCs directly inhibit B cells in vivo and that PD-L1 on HSCs are required in this process.

To examine whether HSCs concurrently inhibit B cells as well as T cells, we developed an intrasplenic artery injection approach to deliver HSCs locally into the spleen to test the effects of HSCs on B cells in vivo. We found that HSCs directly inhibit B cells in vitro and in vivo and that PD-L1 expressed by HSCs is integrally involved in these processes. These new results, together with our previous report showing that HSCs inhibit T cells through PD-L1, suggest that PD-L1 is needed for HSCs to concurrently inhibit T and B cells in the adaptive immune system, which could be an important mechanism by which HSCs help to maintain liver homeostasis.

Although T cells were thought to be the major mechanism underlying allograft rejection, an increasing number of studies recently suggested that B cells also play important roles in allograft rejection (34), as well as in many other pathological situations (e.g., multiple sclerosis and autoimmune uveitis) in which T cells were thought to be the major player in the pathogenesis (35). B cells can induce tissue damage by secreting pathological Abs or by producing cytotoxic cytokines, including IL-6. They can also shape the splenic architecture (36) and serve as APCs to stimulate T cells (37), mechanisms that amplify T cell–mediated tissue damage. Under normal conditions, B cells reside in the liver together with T cells (19, 20). Upon stimulation with LPSs, these hepatic B cells produce massive amounts of inflammatory cytokines, including IL-6, IFN-γ, and TNF-α, but little IL-10, which could initiate liver inflammation in situ (18). In addition, B cell infiltration was found in transplanted renal and liver allografts during acute rejection; consequently, depletion of B cells with rituximab ameliorates renal graft rejection (38). This B cell–depletion therapy is emerging as a new and effective approach for treating many autoimmune diseases that were originally thought to be mediated by T cells (39). Our discovery that HSCs directly inhibit B cells adds a new mechanism to explain how HSCs regulate immune reactions in the liver, which could contribute to the importance of HSCs in maintaining the immune tolerance status of the liver.

In Transwell experiments, when HSCs were cultured separately from activated B cells, their ability to inhibit B cells was significantly impaired, suggesting that direct cell–cell contact is required for HSCs to efficiently inhibit B cells. This finding led us to investigate the role of PD-L1, expressed on the surface of HSCs, in this process. However, in the presence of large numbers of HSCs (at a 1:10 ratio), even without direct contact of HSCs with B cells in the Transwell culture system, B cells were still inhibited by the HSCs to some degree, suggesting that some soluble factors produced by the HSCs still play a role in the HSC-mediated, direct B cell inhibition. These soluble factors remain to be fully characterized.

PD-L1 is a member of the B7 family, serving as a major ligand of PD-1, which is a strong, inducible, negative immune-modulatory receptor present on activated T and B cells. PD-L1 is expressed on the cells of many tissues, including many tumor cells, retinal pigment epithelial cells, and HSCs, and PD-L1 expression is also significantly upregulated in response to inflammation. Interactions between PD-L1 on these peripheral tissue cells and PD-1 on activated T cells potently suppress T cell responses, which, in turn, enables tumors to escape immunosurveillance and help organs like the eye or liver to maintain their immunoprivileged status. Reagents targeting PD-L1 or PD-1 are under intensive study in clinical trials for cancer treatment, with positive preliminary results. Pembrolizumab and nivolumab, humanized anti–PD-1 mAbs, were recently approved for treating unresponsive melanoma (40) and non-small cell lung cancer (41), further confirming that the PD-L1/PD-1 interaction is a potent T cell–inhibitory mechanism.

Although the role of the PD-L1/PD-1 pathway in inhibiting T cells has been investigated extensively, reports on the significance of this pathway in regulating B cell responses are sporadic. Since early reports in the 1990s showing that PD-1 is present on activated B cells (29) and that PD-1–KO mice develop elevated levels of autoantibodies (42), there was only one follow-up report in 2013 showing that PD-1 is important in regulating B cell function (30). Nearly all investigative efforts focused on the role of PD-1 on T cells in mediating T cell suppression. The impact of PD-1 on B cells and its interaction with PD-L1 on other cells have been neglected. Using flow cytometric analysis, we found that, consistent with previous reports (29), PD-1 is present on activated B cells (data not shown); our results, described above, indicate that HSCs directly suppress B cells through PD-L1, which interacts with PD-1 on B cells to inhibit the proliferation of and Ab/cytokine production from activated B cells. These results provide further evidence that the PD-L1/PD-1 pathway is an important pathway regulating B cell function.

Systemic i.v. injection of cells is the most common approach for cell-based therapy studies in animals and humans. It is established that, immediately after i.v. injection, most cells are trapped in the lung; some of them migrate out and into other organs. Although tail vein i.v. injection has been used successfully in mice to systemically deliver many cell types [e.g., dendritic cells (43), myeloid-derived suppressor cells (44), and mesenchymal stem cells (45)], we found that tail vein i.v. injection of primary HSCs led to lethal pulmonary embolisms, potentially due to the large size of HSCs, which made our in vivo studies to test the effect of HSCs in inhibiting B cell responses impossible. Because it was demonstrated that most of the reactive B cells are located in the marginal zone of the spleen after i.p. NP-Ficoll immunization, we hypothesized that we could locally deliver HSCs into the spleen to suppress the NP-Ficoll–responding B cells without causing pulmonary embolisms. Although direct intrasplenic injection was used to deliver cells, including hepatocytes (46) and tumor cells (47), into the spleen in experimental studies, this approach was not useful for our purposes, because it would likely cause splenic injury and uneven distribution of the injected cells in the spleen. Given the sinusoidal structure of the spleen, we speculated that intrasplenic artery injection of the HSCs would be a better approach to locally delivering HSCs, thus enabling them to directly interact with marginal zone B cells. Indeed, we found that intrasplenic artery injection was practical.

HSCs are one of the major components of nonparenchymal cells in the liver (11), and the significance of HSCs in liver fibrosis has been the focus of intensive studies. Emerging evidence also suggests that HSCs are an important group of resident cells that regulate immune responses in the liver. We hypothesized that HSCs are essential for the liver to maintain its homeostasis. In support of this concept, we (16) and other investigators (48) found that HSCs are strongly immunosuppressive. We further demonstrated that HSCs inhibit T cells through PD-L1 (16) and that HSCs induce the propagation of myeloid-derived suppressor cells, which concurrently inhibit T and B cells (44). The results presented in this article establish that, in addition to the previous discoveries, HSCs directly inhibit B cells and that the PD-L1 expressed on the surface of HSCs is integrally involved in the underlying mechanism.

In summary, we examined the potential role of HSCs in regulating B cells, the second major component of the adaptive immune system. We found that HSCs directly inhibited B cells through PD-L1/PD-1 interactions in vitro. By establishing a protocol to locally deliver HSCs into the spleen through intrasplenic artery injection, we further confirmed the direct B cell–inhibiting activity of HSCs and the importance of the PD-L1/PD-1 interaction in vivo. These results established a new approach to locally deliver HSCs and other large cells into the spleen without causing pulmonary embolism and demonstrated a novel mechanism by which HSCs regulate immune responses in the liver.

The authors have no financial conflicts of interest.