Mechanistic Insights into Immunomodulation by Hepatic Stellate Cells in Mice: A Critical Role of Interferon-γ Signaling

Abstract

The liver is considered to be an immune privileged organ which favors the induction of tolerance. The underlying mechanisms remain not completely understood. Interestingly, liver transplants are spontaneously accepted in a number of animal models, but hepatocyte transplants are acutely rejected, suggesting that liver non-parenchymal cells (NPC) may effectively protect the parenchymal cells from immune attack. We have shown the profound T cell inhibitory activity of hepatic stellate cells (HSC). Thus, co-transplantation with HSC effectively protects islet allografts from rejection in mice. In this study, using T cell receptor (TCR) transgenic and gene knockout approaches, we provided definitive evidence that HSC protected co-transplanted islet allografts via exerting comprehensive inhibitory effects on T cells, including apoptotic death in graft infiltrating antigen-specific effector T cells, and marked expansion of CD4⁺ Forkhead box protein (Foxp)3⁺ T regulatory (Treg) cells. All these effects required an intact interferon (IFN)-γ signaling in HSC, demonstrated by using HSC isolated from IFN-γ receptor1 knockout mice. B7-H1 expression on HSC, a product molecule of IFN-γ signaling, was responsible for induction of T cells apoptosis, but had no effect on expansion of Treg cells, suggesting that a yet to be determined effector molecule (s) produced by IFN-γ signaling is involved in this process.

Conclusion

Upon inflammatory stimulation, the specific organ stromal cells, such as HSC in the liver, demonstrate potent immune regulatory activity. Understanding of the mechanisms involved may lead to development of novel strategies for clinical applications in transplantation and autoimmune diseases.

Introduction

Hepatic tolerance has been recognized by spontaneous acceptance of liver transplants in a number of animal models (1) and by induction of tolerance to antigens delivered via portal vein (2). Compared with other organ transplants, human liver transplants manifest absence of hyperacute rejection and low incidence of chronic rejection (3). Certain percentage of liver transplant patients have been weaned from immunosuppression without graft rejection (2). Interestingly, liver allografts are accepted, whereas hepatocyte transplants are acutely rejected (4), suggesting that liver non-parenchymal cells (NPC) play a role in protecting parenchymal cells (hepatocytes) from immune injury. We have examined a variety of mouse liver NPC, and found that hepatic stellate cells (HSC), abundant liver stromal cells, well known for storing retinoids and participating in fibrogenesis, have potent immune regulatory activity. HSC can effectively protect islet allografts from rejection when they are co-transplanted (5,6).

IFN-γ is an important proinflammatory cytokine mainly produced by Th1 T cells and NK cells, mediating both innate and adaptive immune responses. Recent accumulating evidence suggests that IFN-γ is also critical for tolerance induction (7–12). Thus, IFN-γ stimulation is required for liver transplant tolerance, as liver allografts transplanted into wild type (WT) mice achieve long-term survival, whereas no allografts survived beyond 14 days in IFN-γ^−/− recipients or IFN-γ receptor (R)^−/− allografts in WT recipients (13). The underlying mechanism(s) are not completely understood (8).

In this study, we demonstrated that islet allografts achieved long-term survival when HSC were co-transplanted, which required IFN-γ stimulation to HSC, since HSC isolated from IFN-γ receptor (R)1 knockout (KO) mice were unable to provide the same degree of protection. Co-transplanted HSC inhibited T cell responses via elimination of graft infiltrating effector T cells and expansion of CD4⁺ Forkhead box protein (Foxp)3⁺ T regulatory (Treg) cells. Both effects required IFN-γ stimulation. B7-H1, a downstream product of IFN-γ signaling, contributes to deletion of effector T cells, but has no effect on expansion of Treg cells, suggesting other yet-to be identified IFN-γ signaling product participates in this process.

Materials and Methods

Mice

Male C57BL/6 (B6; H-2^b), C3H (H-2^k), BALB/c (H-2^d) and IFN-γR1 KO (B6.129S7-Ifngr1^tm1Agt/J) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). B7-H-1 KO mice were kindly provided by Dr. Lieping Chen (Johns Hopkins University Medical School, Baltimore, MD). 2C (H-2^b) and DES (H-2^k) TCR transgenic mice, expressed L^d and H-2K^b specific TCR transgene on CD8⁺ T cells, respectively. All mice were used under NIH guidelines.

Preparation of HSC

HSC were isolated from the mouse liver NPC as previously described (5). The liver was perfused via the portal vein with collagenase IV. The smashed cells were filtered through a nylon mesh. The HSC were purified by Percoll density gradient centrifugation, and cultured in complete medium supplemented with 20% fetal bovine serum for 7 to 14 days, unless otherwise indicated. The purity of HSC ranged from 90% to 95% determined by desmin immunostaining.

Islet Transplantation

Diabetes was induced in recipients with a single intraperitoneal injection of streptozotocin (STZ) (180 mg/kg, Sigma-Aldrich, St. Louis, MO). Only the mice with blood glucose exceeding 350 mg/dL were used in experiments. Islets were isolated from donor pancreata by collagenase V digestion, and separated on a Ficoll gradient (Type 400, Sigma-Aldrich). The islets were purified by hand picking. 300 islets alone or mixed with 3×10⁵ HSC were aspirated into polyethylene tubing (PE-50), pelleted by gent centrifugation, and placed under renal capsule. Transplantation was considered successful when blood glucose returned to and remained normal (≤150 mg/dL) for 4 days after transplantation. The first day of two consecutive readings of blood glucose greater than ≥350 mg/dL was defined as the date of graft failure. No immunosuppressive agents were administered throughout the experiment.

Glucose tolerance test

Following fasting overnight, mice were intraperitoneally injected with 50% dextrose solution at 2g/kg. Blood glucose and insulin were measured before and after glucose injection at indicated time by glucosemeter and ELISA kit (Linco Research, St. Charles, MO), respectively.

Flow cytometric analysis

Anti-CD4, -CD8, -CD25, -CD44, -CD62L, and -IFN-γ mAbs and annexin V detection kit were purchased from BD PharMingen (San Diego, CA). The DES TCR determinant was stained with anti-DES mAb. The appropriate isotype-matched irrelevant Abs were used as controls in all experiments. Flow cytometric analysis was performed on a FACSCalibur flow cytometer (BD Biosciences). For carboxyfluorescein diacetate succinimidyl ester (CFSE) labeling, cells (10⁷/mL) wer incubated with 2.5 mol/L CFSE (Molecular Probes, Eugene, OR) for 8 minutes. For intracellular IFN-γ stining, cells were pre-fixed using 2% paraformadehyde, permeabilized by 0.1% saponin in 0.1 % bovine serum albumin solution. Foxp3 was stained using fixation and permeabilization buffers in a FoxP3 kit (eBioscience, San Diego, CA).

Caspase activity

30 μg proteins isolated from cells were incubated with caspase 3-specific fluorescent substrate Ac-DEVD-AFC (20 μM) for 30 minutes, and measured by a fluorescence spectrometer (Tecan, Hayward, CA) at 400 nm/505 nm.

Immunohistochemistry

The cryostat sections were stained using anti-CD4, -CD8 (BD PharMingen) or -Foxp3 mAb (FJK-16s, eBioscience). The color was developed using avidin-biotin-alkaline phosphatase complex (ABC) as the substrate. Insulin was identified using anti-insulin mAb (Santa Cruz Biotechnology, Santa Cruz, CA), and developed color using 3-amino-9-ethylcarbazole. The isotype-matched irrelevant Abs were used as controls. The slides were counterstained with Harris’ hematoxylin. The immunofluorescence protocols were also used for CD4, CD8 and Foxp3 staining. Collagen was stained overnight by Sirius Red (saturated picric acid with 0.1%Sirius Red F3BA, Sigma-Aldrich).

Quantitative real-time PCR

Total RNA was extracted with TRIzol Reagent. cDNA was synthesized with SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA). The primers were GACTGGCTTCATGTCCATTCAC/GTT GGG TTG TCA CAG CAT GGT for granzyme B and TGCCACTCGGTCAGAATG C/CGGCGGAGGGTAGTCACAT for perforin. The mRNAs were quantified using Applied Biosystems 7500 Fast PCR System in duplicate. The expression levels were normalized to 18S mRNA.

Statistical Analysis

Graft survival between groups of transplanted animals was compared using the log-rank test. The statistical significance of parametric data was determined using the Student t test. p< 0.05 was considered statistically significant.

Results

Expression of B7-H1 on HSC in response to INF-γ stimulation

HSC isolated from B6 mice were incubated with various concentrations of mouse IFN-γ (0.1–200 U/mL) for 24 hours. Expression of B7-H1 analyzed by flow cytometry was enhanced after treatment with IFN-γ at as low as 0.1 U/ml, reaching a maximum at 10 U/ml in a dose-dependent manner (Fig. 1A). To study time course, HSC were exposed to IFN-γ (100 U/ml) for 0.5–48 hours. B7-H1 expression began to increase after 0.5 hour exposure and continued thereafter in a time-dependent manner, with maximal expression at 24–48 hours (Fig. 1B). IFN-γ specifically interacts with IFN-γ receptor (R) containing IFN-γR₁ binding chain and internal IFN-γR₂ transducing chain (12). Expression of B7-H1 on HSC isolated from IFN-γR1 KO mice showed no response to IFN-γ (Fig. 1C), indicating that B7-H1 expression depends on IFN-γ signaling. This was supported by the fact that B7-H1 response to IFN-γ-stimulation was almost entirely absent on HSC from Stat1^−/− mice (Fig. 1D), since Stat1 is a key transcription mediator for IFN-γ signaling (14). Expression of B7-H1 on IFN-γR1^−/− HSC was slightly increased by co-culture with activated T cells (Fig. 1C), suggesting that activated T cells may produce other factor (s), participating in upregulation of B7-H1. Different from activated HSC, expression of B7-H1 on quiescent HSC (culture for 2 days, instead of 7–14 days) was only marginally upregulated in response to IFN-γ (supplement Fig. 1).

Figure 1. Expression of B7-H1 on HSC in response to IFN-γ stimulation.

(A) HSC isolated from B6 (H-2^b) mice were exposed to graded concentrations of IFN-γ (0.1–200 U/ml) for 24 hours in vitro, and stained with anti-B7-H1 mAb, or (B) HSC were incubated with IFN-γ (100 U/ml) for varying times (3–48 hours), and analyzed by flow cytometry, displayed as histograms. For panels A and B, the filled areas represent isotype controls. (C) B7-H1 expression on HSC is dependent on IFN-γ stimulation. HSC isolated from wild type (WT) or IFN-γR1 KO mice (both on B6 background) were exposed to IFN-γ (100 U/ml), or cultured with activated B6 T cells (pre-cultured with irradiated BALB/c (H-2^d) DC at DC:T ratio of 1:20 for 2 days) at a HSC:T ratio of 1:20 for 48 hours, or (D) HSC isolated from WT or Stat1 KO mice (B6 background) were exposed to IFN-γ (100 U/ml) for 48 hours. Cells were stained using anti-B7-H1 mAb and analyzed by flow cytometry, displayed as histograms. For panels C and D, the non-filled areas represent isotype controls. The data are representative of two separate experiments.

HSC deficient in IFN-γR1 or B7-H1 expression lose the capacity to protect co-transplanted islet allografts

BALB/c islet allografts were mixed with B6 HSC and co-transplanted under the renal capsule of the STZ-induced diabetic B6 recipients. Islet allografts transplanted alone, serving as controls, were rejected within 15 days, while co-transplantation with WT HSC effectively protected islet allografts from rejection (Fig. 2A, p<0.05, islet vs. islet+WT HSC) with 60% grafts achieving long-term survival (>90 days). Removal of the kidney bearing the islet grafts were performed at various times (from day 90–170) in all animals maintaining long-term euglycemia, and resulted in a prompt recurrence of hyperglycemia (data not shown), indicating the euglycemia was maintained by the functioning subcapsular islet transplants. Fig. 2B shows photomicrographs of the revascularized islet grafts obtained 142 days after co-transplantation with HSC. Insulin production is demonstrated by immunohistochemical staining with anti-insulin mAb. Glucose tolerance tests in three recipients with long-term surviving islet allografts showed normal response to glucose challenge, indicating that the islet allografts co-transplanted with HSC maintain long-term function (supplement Fig. 2). The collagen deposition was assessed in islet allografts alone or co-transplanted HSC by Sirius Red staining. The presence of HSC did not enhance collagen deposition at early stage (day 7) post transplant, compared to islet alone grafts, while Sirius Red material markedly increased in long-term surviving islet allografts, indicating a deposition of large amounts of collagens (supplement Fig. 3).

Figure 2. IFN-γ signaling and B7-H1 expression are required for HSC to protect islet allografts.

300 freshly isolated islets from BALB/c (H-2^d) mice were mixed with 5 × 10⁵ HSC from WT, B7-H1 KO or IFN-γR1 KO mice [all on B6 (H-2^b) background], and transplanted under renal capsule of STZ-induced diabetic B6 recipients. Transplantation of islet allografts alone served as controls. No immunosuppressive reagents were administered. Survival of the islet allografts was determined by examination of the blood glucose levels. (A) Co-transplantation with HSC effectively protected islet allografts from rejection (p<0.05, WT HSC vs. islet alone), which was dependent on expression of B7-H1 or IFN-γR1 (p<0.05, WT HSC vs. B7-H1 KO or IFN-γR KO HSC). (B) The islet allografts were harvested from a mice receiving islet and HSC co-transplantation for 142 days; left panel: vigorous revascularization was seen in islet allografts (white area); right panel: the graft sections were stained with anti-insulin mAb (red).

To address the role of IFN-γ signaling and its product B7-H1 in immune regulatory activity of HSC, islet allografts were co-transplanted with HSC isolated from IFN-γR1^−/− or B7-H1^−/− (both on B6 background). HSC deficient in either B7-H1^−/− or IFN-γR1^−/− largely lost their protective effect on islet allografts (Fig. 2A, both p<0.05, vs. WT HSC group), indicating that the protection of islet allografts by co-transplanted HSC requires IFN-γ stimulation, and that its downstream product, B7-H1, is a crucial effector molecule.

HSC co-transplantation attenuates CD8⁺ T cells in islet allografts

In order to study the effect of HSC on T cell responses, we harvested the islet grafts, spleen, draining (kidney hilium) and irrelevant LN from the recipients of islet allograft alone or co-transplantation with at various post operative days (POD). Flow analysis of T cells isolated from these tissues showed that HSC co-transplantation had little effect on the percentages of CD4⁺ and CD8⁺ cells in all tested compartments on POD 7. However, by POD 14, the proportion of CD4⁺ T cells in HSC-co-transplanted islet grafts significantly increased, while the percentage of CD8⁺ T cells conversely decreased in HSC co-transplanted groups. These changes were further enhanced in co-transplanted recipients with long-term surviving islet allografts (Fig. 3A). However, the actual number of CD4⁺ and CD8⁺ T cells within the allografts showed that HSC co-transplantation led to significant reduction of CD8⁺ T cell numbers in islet grafts on POD 7 and continuously decreasing thereafter. CD4⁺ T cells in HSC co-transplanted grafts were reduced on POD 7, compared to islet only group, but they were not further significantly decreased on POD 14 and long-term (Fig. 3B). These data suggest that HSC co-transplantation is associated with marked reduction of CD8⁺ T infiltrates in islet allografts, but has little effect on the systemic CD8⁺ T cell pool.

Figure 3. Co-transplantation of HSC leads to reduction of CD8⁺ T cells in islet allografts.

BALB/c (H-2^d) islets mixed with 5 × 10⁵ HSC from B6 (H-2^b) mice were transplanted under renal capsule of diabetic B6 recipients. Transplantation of BALB/c islet allografts alone served as control. Animals were sacrificed on day 7, 14 and long-term (LT, >POD 90). Lymphocytes isolated from the islet grafts, spleen, irrelevant (I-) and draining (D-) LN were triple stained for CD3, CD4, CD8 and analyzed by flow cytometry (n=3 in each group). (A) Percentage of CD4⁺ and CD8⁺ cells in CD3⁺ population, expressed as mean ± 1SD. (B) Absolute number of CD4⁺ and CD8⁺ cells in islet grafts were calculated based on flow analysis data, expressed as cell numbers/graft. (C and D) Co-transplantation of HSC does not inhibit activation of CD8⁺ T cells. (C) Lymphocytes isolated from the grafts on POD 7 were stained with anti-CD62L, -CD44 or -IFN-γ mAbs, and analyzed by flow cytometry gated on CD8⁺ cell populations (C); RNA was isolated from CD8⁺ T cells purified from the isolated lymphocytes. mRNA expression of granzyme B and perforin was examined by qPCR (D).

To determine whether the HSC-induced reduction of CD8⁺ T cells was a result of impaired activation, expression of activation-related molecules CD44, CD62L, IFN-γ, granzyme B and perforin in CD8⁺ T cells was examined by flow cytometry (protein) or qPCR (mRNA). HSC co-transplantation did not suppress expression of CD44, CD62L, IFN-γ, granzyme B and perforin in either islet grafts or draining LN (Fig. 3C and D), suggesting that the reduction of CD8⁺ T cells is unlikely a consequence of inhibiting activation.

HSC co-transplantation promotes apoptosis in specific CD8⁺ T cells

We utilized TCR transgenic marker to track the fate of allospecific CD8⁺ T cells. 1.5 × 10⁷ CFSE-labeled DES TCR transgenic T cells (H-2^k background, CD8⁺ T cells specifically recognize H-2^b antigen) were adoptively transferred on POD 4 into C3H (H-2^k) recipients after transplantation of allogeneic (B6, H-2^b) islets with or without C3H HSC. Lymphocytes were isolated, 1 to 3 days after adoptive transfer, from islet allografts, spleen, draining and irrelevant LN, and double stained with DES clonotypic and anti-annexin V mAbs for flow analysis. Very few dividing DES⁺ cells were detected in spleen and irrelevant LN (data not shown), while dividing DES⁺ cells were clearly identified in draining LN and islet allografts. Compared to islet allograft alone, co-transplantation with HSC demonstrated little effect on DES⁺ cells in draining LN, but the presence of HSC was associated with marked reduction (~50%) of DES⁺ cells in the allograft in both percentage and absolute number (Fig. 4A, upper and right panels). The reduction was seen predominantly in dividing DES⁺ cell population (Fig. 4A, upper left panel), indicating that HSC induced deletion of activated specific CD8⁺ T cells. In contrast to islet alone group, in which ~90% of dividing DES⁺ cells were annexin V^low, approximately 45% dividing DES⁺ cells were annexin V^high in the islet grafts co-transplanted with HSC (Fig. 4A, lower panel), suggesting that HSC induced reduction of specific CD8⁺ T cells is a consequence of apoptotic death.

Figure 4. HSC induces specific CD8⁺ T cell apoptosis in islet grafts.

Islets from B6 (H-2^b) donors were mixed with HSC from C3H (H-2^k) mice, and transplanted into diabetic C3H recipients. Transplantation of islet allografts alone served as control. On POD 4, recipients were intravenously injected with 1.5 × 10⁷ CFSE-labeled T cells from DES TCR transgenic mice (H-2^k, CD8⁺ T cells specifically recognizing H-2K^b antigen). Two days later, lymphocytes were isolated from islet allografts and draining LN, and double stained with anti-DES and -annexin V mAbs for flow analysis. (A) CFSE dilution analysis. The number represents the percentage of DES⁺ cells in each group. Dividing DES⁺ cells were identified in both draining LN and islet grafts. Right panel: The absolute number of DES⁺ cells was calculated based on flow analyses. A marked reduction of DES⁺ cells was demonstrated in islet allografts that were co-transplanted with HSC, but not in draining LN. Lower panel: expression of annexin V was analyzed on dividing and non-dividing DES⁺ cells in the islet allografts, respectively. HSC-induced reduction of DES⁺ cells is associated with enhanced apoptotic activity in dividing cell population. (B) Co-transplanted HSC exert local effect. Islets allografts and HSC were co-transplanted in left kidney, the same number of islet allografts without HSC were transplanted in right kidney. CFSE labeled DES T cells were intravenously given on POD 4. Lymphocytes were isolated from islet allografts and draining LN two days later, and stained with anti-DES mAb for flow analysis and CFSE dilution analysis (left panels). The absolute number of DES⁺ cells was calculated based on flow analysis (right panel). (C) HSC induce apoptosis in activated specific CD8⁺ T cells in vitro. 2 × 10⁶ CFSE-labeled T cells isolated from DES TCR transgenic mice were cultured with irradiated B6 spleen cells and at a ratio of 1:1 for 4 days. Irradiated HSC from C3H mice (H-2^k) were added at the beginning of the culture at HSC:T ratio of 1:40, 1:20 or 1:10. Cells were double stained with anti-DES and -annexin V mAbs for flow analysis gated on DES⁺ cell populations. Upper left panels show that addition of HSC was associated with reduction of dividing DES⁺ T cells in a dose dependent manner. The right panel shows the corresponding absolute number of dividing and non-dividing DES⁺ cells in each group, based on flow analysis, expressed as mean ± 1SD. Lower panels: analysis of annexin V expression in dividing cell populations. The data are representative of three experiments.

To determine whether the HSC exerted a systemic effect, DES T cells were adoptively transferred into the recipients that were simultaneously transplanted islet allografts in both side kidneys, while HSC were co-transplanted only in left kidney. Flow analysis showed that dividing DES⁺ cells were markedly less in islet allografts in the kidney where HSC were co-transplanted (Fig. 4B), indicating that the effect of co-transplanted HSC is locally mediated.

To obtain direct evidence of HSC inducing apoptosis in specific CD8⁺ T cells, graded numbers of HSC from C3H mice were added into a culture of CFSE labeled DES T cells whose activation was elicited by irradiated allogeneic (B6) spleen cells. The addition of HSC was associated with marked reduction of dividing DES T cells (both percentage and absolute numbers) with increase in expression of annexin V in a dose dependent manner, indicating that HSC induce apoptosis in activated CD8⁺ T cells (Fig. 4C). This was confirmed by the results of caspase activity analysis. The presence of HSC enhanced T cell caspase 3 activity in a dose dependent manner (supplement Fig. 4).

IFN-γ/B7-H1 axis is essential for HSC-induced deletion of specific CD8⁺ T cells

To examine whether IFN-γ signaling and expression of B7-H1 were required for HSC-mediated apoptosis of CD8⁺ T cells in the islet allografts, co-transplanted HSC were isolated from IFN-γR1^−/− or B7-H1^−/− mice. HSC from WT mice served as controls (n=3 in each group). The lymphocytes were isolated from the islet allografts on POD 7 and 14, and stained with anti-CD8 mAb for flow cytometric analysis. Comparison of CD8⁺ T cells number in the islet grafts revealed that IFN-γR1^−/− or B7-H1^−/− HSC induced significantly less reduction of CD8⁺ T cells on either POD 7 or 14 (Fig. 5A, both p<0.05, compared to WT group), and the numbers were similar to levels seen in the islet allografts alone (POD 7), suggesting that an intact IFN-γ signaling and expression of B7-H1 are essential for HSC-induced deletion of CD8⁺ T cells. Since all islet allografts transplanted alone were rejected before POD 14, corresponding values on POD 14 could not be obtained for comparison (Fig. 5A).

Figure 5. Elimination of graft CD8⁺ T cells is dependent on expression of IFN-γ receptor and B7-H1 on co-transplanted HSC.

Islets from BALB/c (H-2^d) donors were mixed with HSC from B7-H1^−/− or IFN-γR1^−/− (both on B6 (H-2^b) background) mice (HSC from WT B6 mice were used for comparison), and transplanted into diabetic B6 recipients. Transplantation of BALB/c islets alone served as control (n=3 in each group). (A) CD8⁺ T cells in islet allografts. Animals were sacrificed on POD 7 and 14, and the islet allografts were sectioned and staining with anti-CD8 mAb for immunohistochemical analyses. A total of 10 high power fields (hpf) were randomly selected in each graft for CD8⁺ cells enumeration. The data are expressed as mean cell number/graft± 1SD. Islet allografts transplanted alone were rejected before POD 14, therefore the related data were absent. (B and C) Use of TCR transgenic T cell adoptive transfer. On POD 3, recipients were intravenously given 1.5 × 10⁷ CFSE-labeled T cells from 2C TCR transgenic mice (H-2^b). 3 days later, lymphocytes were isolated from islet allografts, and stained with anti-CD8 mAb for flow analysis. The CFSE⁺CD8⁺ cells were identified as L^d specific CD8 T cells. Upper panels: CFSE dilution analysis; the numbers are the percentage of CFSE⁺CD8⁺ cells in whole cell population. Lower panels: CFSE dilution analysis gated on CFSE⁺CD8⁺ cells, expressed as histograms. The numbers are the percentage of dividing cells in specific CD8 T cells.

The HSC-induced deletion of CD8⁺ T cells in islet grafts was reexamined using TCR transgenic mice. BALB/c islets were co-transplanted with HSC from B7-H1^−/− or IFN-γR1^−/− mice (both B6 background) into B6 diabetic recipients. 2C mice (B6 background, recognizing MHC L^d antigen) were used as source of allospecific CD8⁺ T cells for adoptive transfer. Since flow analysis showed that 90% of CD3⁺ cells in 2C mice were CD8⁺, among which 99% were positive for clonotypic 1B2 mAb staining (data not shown), we simply defined CD3⁺CD8⁺CFSE⁺ as allospecific CD8⁺ T cells. As expected, co-transplanted WT HSC induced marked reduction of allospecific CD8⁺ T cells in islet allografts, which mainly occurred in dividing cell population. The HSC associated reduction of allospecific CD8⁺ T cells was partially inhibited when co-transplanted HSC were deficient in B7-H1 (Fig. 5B), suggesting that B7-H1 is required for elimination of activated allospecific CD8⁺ T cells. When HSC were deficient in IFN-γR1, the HSC associated reduction of specific CD8⁺ T cells was almost totally inhibited compared to the islet only group (Fig. 5C), indicating that elimination of activated specific CD8⁺ T cells is dependent on IFN-γ signaling in HSC.

HSC co-transplantation markedly expands CD4⁺FoxP3⁺ Treg cells

CD4⁺FoxP3⁺ Treg cells are essential for the induction and maintenance of peripheral tolerance (11). To examine the effect of HSC co-transplantation on Treg cell activity, lymphocytes from islet allografts, spleen, blood, draining and irrelevant LN were isolated on POD 7, 14 and long-term (>POD 90) for identification of CD4⁺FoxP3⁺ cells by either flow cytometry or immunohistochemistry. Normal mice (no transplantation) and the mice received islet allografts alone served for comparison. As shown in Fig. 6A, on day 7 after islet transplantation alone, the incidence of Foxp3⁺ cells in CD4⁺ population was increased in all tested compartments, compared to normal mice, reflecting a collateral expansion of Treg cells during the allo-response, as previously described (12). HSC co-transplantation markedly enhanced the proportion of FoxP3⁺ cells in all tested compartments with highest in the islet grafts (>40%), peaking at POD 7 and 14. In the recipients bearing long-term surviving islet allografts, Foxp3⁺ cells substantially declined in the allograft, spleen and peripheral blood, but maintained high in LN (Fig. 6A), consistent with the characteristics of induced (i) Treg cells, which are thought to be generated in LN (15,16). These change patterns were confirmed by calculating the absolute number of CD4⁺FoxP3⁺ cells in the allografts (Fig. 6B, left panel) and draining LN (data not shown). Compared to islet alone, CD4⁺FoxP3⁺ cells in HSC co-transplanted grafts were significantly elevated on POD 7, and returned back to the control levels by POD 14. However, in the long-term surviving allografts, CD4⁺FoxP3⁺ cells actually declined to the levels lower than controls (Fig. 6B, left panel). This was validated using in situ immunohistochemical double staining of islet allografts with anti-CD4 and anti-Foxp3 mAbs (Fig. 6B, right panels), supporting the conclusions generated based on flow analysis. In contrast to what seen in the grafts, the absolute number of CD4⁺FoxP3⁺ cells in the draining LN remained high in graft long-term survivors based on flow analysis (Fig. 6C, p<0.05, compared to any other group). These data indicate that, in long-graft-survival recipients, many Treg cells remained in LN, the immunological significance of which need to be further elucidated.

Figure 6. Co-transplantation with HSC markedly expands Foxp3⁺ Treg cells, which is dependent on IFN-γ signaling in HSC.

BALB/c (H-2^d) islets mixed with B6 (H-2^b) HSC were transplanted under renal capsule of diabetic B6 recipients. Transplantation of islet allografts alone and without transplantation (none) served as controls. Animals were sacrificed on POD 7, 14 or >90 (long-term, LT). Lymphocytes isolated from the islet grafts, blood, spleen, irrelevant (I-) and draining (D-) LN were triple stained for CD4, CD25 and Foxp3 and analyzed by flow cytometry (n=3 in each group). (A) HSC co-transplantation markedly expands CD4⁺Foxp3⁺ cells. Expression of CD25 and Foxp3 was analyzed gated on CD4⁺ cells, and the numbers are percentages. (B) The absolute numbers of CD4⁺Foxp3⁺ cells in islet allografts were counted by flow analysis, as well as immunohistochemitry. Total of 10 high power fields (hpf) were randomly selected in each graft. The right lower panels show double staining of CD4 (red) and Foxp3 (green). (C) The numbers of CD4⁺Foxp3⁺ cells in the draining LN were counted by flow analyses. CD4⁺FoxP3⁺ cells in the draining LN remained high in graft long-term survivors, p<0.05, compared to any other group. (D) Adoptive transfer of CD4⁺CD25⁺ cells prolongs survival of islet allografts. Lymphocytes were isolated from B6 recipients of BALB/c islets and B6 HSC for 2 weeks, and CD4⁺CD25⁺ T cells were enriched using MACS regulatory T cell isolation kit (Miltenyi Biotec) with purity >90%. 5×10⁵ CD4⁺CD25⁺ cells were i.v. injected into diabetic B6 mice one day before receiving BALB/c islet allografts. Injection of same number of bulk spleen cells from normal B6 mice served as control. Upper panel: Blood was withdrawn at the indicated time points post transplantation. The incidence of Foxp3⁺ cells were analyzed by flow cytometry, expressed as percentage in CD4⁺ cell population. Lower panel: impact of adoptive transfer of CD4⁺CD25⁺ cells on survival of islet allografts. (E) Expansion of CD4⁺Foxp3⁺ T cells is dependent on expression of IFN-γ receptor on HSC. Islets from BALB/c mice mixed with 5 × 10⁵ HSC from WT, IFN-γR1^−/− or B7-H1^−/− (all on B6 background) mice were transplanted to diabetic B6 recipients. Animals were sacrificed on POD 7 and 14. Lymphocytes isolated from the islet grafts, spleen, irrelevant (I-) and draining (D-) LN were triple stained for CD4, CD25 and Foxp3, and analyzed by flow cytometry. The numbers are percentages in CD4⁺ cell population.

The in vivo function of the generated CD4⁺FoxP3⁺ T cells was examined by transfer of magnetic bead purified CD4⁺CD25⁺T cells from long-graft-surviving animals (the purity ranged from 93 to 95%; 90% of these CD4⁺CD25⁺ cells were Foxp3⁺) into diabetic mice one day before islet transplantation. The transfer of isolated CD4⁺CD25⁺ cells markedly enhanced CD4⁺Foxp3⁺ cell levels in peripheral blood similar to the recipients of islet and HSC co-transplantation (Fig. 6D, upper panel), and significantly prolonged survival of islet allografts compared to the controls (Fig. 6D, lower panel, p<0.05), indicating that these expanded CD4⁺FoxP3⁺ cells are Treg cells.

IFN-γ signaling is required for HSC to induce Foxp3⁺ Treg cells, but not B7-H1

To elucidate the molecular mechanism of HSC-induced CD4⁺FoxP3⁺ cell expansion, HSC from IFN-γR^−/− or B7-H1^−/− mice were co-transplanted with islet allografts. HSC from WT mice were used as controls. Lymphocytes from islet allografts, spleen, draining and irrelevant LN were isolated on POD 7, 14 and long-term (>POD 90) for identification of CD4⁺FoxP3⁺ cells by flow cytometry. Compared to WT HSC controls, IFN-γR^−/− HSC induced markedly less CD4⁺Foxp3⁺ cells in all tested compartments, suggesting that induction of Treg cells is dependent on IFN-γ signaling in the co-transplanted HSC. In addition, animals co-transplanted with B7-H1^−/− HSC, generated similar levels of CD4⁺Foxp3⁺ cells in all tested areas, compared to WT HSC controls POD 14 (Fig. 6E), indicating that B7-H1 expressed on co-transplanted HSC is not responsible for induction of Treg cells. These data demonstrate that IFN-γ signaling in co-transplanted HSC is required for inducing Treg cells, but is unlikely mediated by B7-H1. A yet to be identified downstream product(s) of IFN-γ signaling in HSC may participate in the induction of CD4⁺Foxp3⁺ Treg cells.

Discussion

We have demonstrated that co-transplantation with HSC led to long-term survival of islet allografts without the requirement for immunosuppressive therapy (17). Further elucidation of the mechanisms of this finding may substantially improve clinical cellular transplantation. Thus, the current limitation of requiring chronic administration of immunosuppression associated with severe complications may be overcome. Only by reducing the need for such agents, will cellular transplantation be widely accepted (18). The present study showed that co-transplanted HSC exerted extensive immunomodulatory effects, including elimination of activated specific CD8⁺ T cells and marked expansion of Treg cells. Both effects were dependent on IFN-γ stimulation. Although the mechanisms are largely unknown, it has been revealed that IFN-γ is required for the depletion of alloreactive T cells and induction of tolerance (10). The underlying mechanisms remain unclear (8,19). We showed that IFN-γ induced apoptosis of specific activated T cells was mediated by B7-H1 on HSC. This is in agreement with previous reports that B7-H1 expressed on cancer cells leads to increased apoptosis of activated T cells (20), and B7-H1 maintains long-term acceptance of the corneal allografts by inducing apoptosis of effector T cells (21).

Treg cells play a key role in controlling immune responses, but the involved mechanisms are not completely understood (22). The data in this study demonstrated that transplantation of islet allografts alone slightly increased the incidence of CD4⁺Foxp3⁺ T cells, reflecting a collateral expansion of Treg cells in allo-response. Co-transplanted HSC dramatically increased Foxp3⁺ Treg cells to about 40% in all tested compartments. They were markedly attenuated in long-term surviving islet allografts, correlated with reduction of immune effector cells at the local site, while Treg cells remained high in LN, implicating that the Treg cells are generated and maintained in LN, consistent with previous reports (23). It is however, unlikely that these Treg cells are directly induced by co-transplanted HSC since there is no evidence that co-transplanted HSC are able to migrate to the draining LN. In addition, in a recent study, we have showed that co-culture of HSC with CD4⁺ T cells in vitro only marginally increased CD4⁺Foxp3⁺ cells, and that neutralization of TGF-β by addition of blocking Ab does not affect the marginal increase in Treg cells, suggesting that the TGF-β produced by activated HSC is not crucial in induction of Treg cells in this experimental model (24). We postulate that co-transplanted HSC may influence the development of the antigen-presenting cells (APC) when they process the alloantigens in the islet grafts. These modulated APC migrate to the draining LN and induce Treg cells (25). It would be intriguing to determine whether these modulated APC produce TGF-β and other Treg inducing factors, and their role in induction of Treg cells. Adoptive transfer of these Treg cells prolonged survival of transplanted islets, but it was not effective as the protection of islets by HSC co-transplantation. In addition, the protective effect of HSC appeared to be local. These data suggest that HSC-induced deletion of the activated effector T cells is critical in protection of co-transplanted islet allografts. This is in agreement with the recent study demonstrating that Treg cells play a role in tolerance induced by costimulation blockade, but deletion of reactive T cells is a major mechanism (26). Induction of Treg cells by co-transplanted HSC was dependent on IFN-γ stimulation, it is however, not mediated by B7-H1, suggesting that a yet to be determined product(s) of IFN-γ signaling in HSC is responsible for induction of Treg cells. Quiescent HSC (culture for 2 days) poorly responded to IFN-γ stimulation for B7-H1 expression (supplement Fig. 1), and co-transplantation of quiescent HSC with islet allografts did not markedly expand Treg cells (supplement Fig. 5). This may partially explain the failure of quiescent HSC to prolong islet allograft survival (17). The data of the present study suggest that the collagen encapsulation seems not affect the islet graft function at least up to 142 days post transplant (supplement Fig. 2). We will follow up its effect on longer-term outcome of the graft function.

It is not surprising that the liver, an immune privileged organ (27), contains cells with immunomodulatory properties. The liver encounters various antigens from the GI tract, such as food-derived antigens and bacterial products, which would impose a state of constant immune activation in the liver unless it is controlled by regulatory activities. We have previously shown that the liver induces apoptosis of activated T cells during allo-responses (28). The liver appears to have inherent mechanisms to modulate immune response. Liver may not be the only tissue/organ containing the immunosuppressive tissue cells. Thus, co-transplantation with Sertoli cells in the testis, another immune privileged site, could also protect islet allografts from rejection, but with ~10 times less potent than HSC as to comparing cell numbers required for effectiveness, suggesting involvement of different mechanisms (29). This phenomenon has been observed in non-immune privileged organs. Resident mesenchymal stem cells show immunomodulatory properties in a variety of tissues (30), suggesting that tissue based immunomodulation is a widespread property of many tissues, but to differing degrees (31). This study provides a model as to how the immune response is regulated in peripheral tissues, which is triggered by inflammatory response, and mediated by specific tissue immunomodulatory cells, such as HSC, to exert differential effects on regulatory and effector T cells.

Abbreviations

APC

atigen-presenting cells

B6

C57BL/6

Foxp3

Forkhead box protein 3

HSC

hepatic stellate cell

IFN

Interferon

KO

knockout

LN

lymph node

NPC

nonparenchymal cells

POD

post operation days

STZ

streptozotocin

TCR

T cell receptor

Treg

T regulatory (cells)

WT

wild type