A novel mechanism of liver allograft rejection facilitated by antibodies to liver sinusoidal endothelial cells


  • Presented at the American Transplantation Congress May 2004, Boston, MA.


Liver sinusoidal endothelial cells (LSECs) may be implicated in the induction of liver allograft rejections. We studied the clinical consequences of LSEC-reactive antibodies and their functional capacity in modulating T-cell responses during acute rejections. Pre- and posttransplant sera and T lymphocytes from 95 liver transplant patients were used in this study. LSECs were isolated from normal healthy liver. Binding of antibodies to LSECs was detected using flow cytometry. To study whether LSEC antibodies facilitated cell-mediated immunity, a mixed cell culture (MCC) assay was used. Cytokines in the supernatants of MCC were measured by enzyme-linked immunosorbent assay. Liver biopsy sections were stained to detect the deposition of immunoglobulins in LSECs during rejections. The 2-year patient survival was 86.3%. A significantly higher number of patients with rejections had LSEC antibodies (35/50; 70%) than those without rejections (8/45; 18%) (P < .0001). Purified fractions of LSEC antibodies induced the expression of the costimulatory molecule CD86 on LSECs. A significantly higher number of patients with LSEC antibodies and rejections had an increased proliferation of T cells and markedly decreased levels of transforming growth factor β (TGF-β) in the MCC than those without antibodies and rejections (P < .0001, P < .0001, respectively). Deposition of antibodies in LSECs during rejection episodes was observed in the biopsies of patients with LSEC antibodies but not in those without LSEC antibodies. In conclusion, antibodies to LSECs may facilitate acute liver allograft rejections by down-regulating the immune modulating cytokine TGF-β and thus up-regulating alloreactive T-cell proliferation. (HEPATOLOGY 2004;40:1211–1221.)

The tolerance-inducing capacity of the liver is a well-recognized fact. In a variety of species, liver allografts are accepted spontaneously across major histocompatibility complex (MHC)-incompatible strain combinations.1–3 Acceptance of a liver graft induces specific tolerance to subsequent transplants of other tissues that would otherwise be rejected.4–7 Mechanisms proposed to account for the tolerogenic properties of the liver may be due to a combination of several biological properties that distinguish the liver from other organs.8 One of the principle findings is that liver sinusoidal endothelial cells (LSECs) are likely to be important in tolerance induction.9, 10

LSECs are strategically located in the sinusoids to enable the extraction of blood-borne material.11 Like Kupffer cells, they are also ideally located to interact with T lymphocytes in the sinusoids. Rejection of liver allografts and inflammatory liver disease are believed to be mediated by resident antigen-presenting cells (APCs) to CD4+ T cells.12 In rodents, LSECs function as APCs to both CD4+ and CD8+ T cells in the liver.8, 13–15 In organ transplantation, break in tolerance is manifested clinically as allograft rejection. Vascular endothelial cells are one of the important targets recognized by immune cells during acute rejections.16–20 Sinusoidal endothelialitis, characterized by adhesion of immune cells to LSECs, is a histopathological feature of acute liver rejection.21 These observations suggest a pivotal role for LSECs in acute and chronic liver allograft rejections. Various pathological conditions of the liver affect the reticuloendothelial function of LSECs, and considerable attention has been focused on the changes in the endocytic function of LSECs in liver disease.22 So far there has been no functional analysis of LSECs in modulating T-cell responses during acute rejections of liver allografts.

We were interested in investigating factors that may adversely affect the role of LSECs in liver allograft immunity. We hypothesized that one factor might be the presence of antibodies to LSECs, which upon binding may alter LSEC function, leading to liver allograft rejection. Here, we studied the clinical consequences of antibodies binding to LSECs present in the sera of liver disease patients before and after liver allotransplantation. We also studied the functional capacity of LSEC-reactive antibodies in modulating T-cell responses.


MHC, major histocompatibility complex; LSECs, liver sinusoidal endothelial cells; APCs, antigen-presenting cells; PBMCs, peripheral blood mononuclear cells; HAECs, human aortic endothelial cells; HLA, human leukocyte antigen; TNF, tumor necrosis factor; IFN-γ, interferon γ; MCC, mixed cell culture; cpm, counts per minute; IL, interleukin; TGF-β, transforming growth factor β; Ig, immunoglobulin; IQR, interquartile range; I/R, ischemia-reperfusion.

Patients and Methods

Patient Population.

Ninety-five consecutive primary liver transplants performed at our center were included in this study. Median age was 51 years (interquartile range, 44 to 58), and 40% of the patients were women. The indications for liver transplantation were mainly end-stage chronic liver diseases (Table 1). Recipient/donor selection was based on ABO blood group compatibility and size matching. Follow-up time for every patient was 2 years. A total of 314 serum samples were tested. Pretransplant serum samples were taken immediately prior to liver transplantation. In general, posttransplant serum samples were collected during the first 4 weeks after transplantation and at 3 months thereafter, and also during acute rejection episodes and in the case of hospitalization. Serum from 20 healthy individuals were included as controls. Peripheral blood mononuclear cells (PBMCs) were obtained from all 95 patients before transplantation.

Table 1. Diagnosis of Patients and Presence of LSEC Antibodies in Different Patient Groups
IndicationsNo. of PatientsNo. of LSEC Antibody-Positive Patients (%)
Hepatitis C cirrhosis227 (31.8%)
Primary sclerosing cholangitis (PSC)188 (44.4%)
Familial amyloidotic polyneuropathy (FAP)143 (21.4%)
Primary biliary cirrhosis (PBC)87 (87.5%)
Autoimmune hepatitis (AIH)76 (85.7%)
Hepatitis B cirrhosis72 (28.6%)
Alcoholic cirrhosis76 (85.7%)
Other124 (33.3%)
Total9543 (45.3%)
Healthy controls202 (10%)

Immunosuppressive protocol for most of the patients (n = 71) was based on tacrolimus/FK506 (Prograf; Fujisawa, Munich, Germany) and steroids, with or without mycophenolate mofetil (Cellcept; Roche, Stockholm, Sweden). For the other 24 patients, cyclosporine A–based (Neoral; Novartis, Basel, Switzerland) triple immunosuppression with steroid and either mycophenolate mofetil or basiliximab (Simulect; Novartis Pharma, East Hanover, NJ) was used.

Acute rejection episodes (88% occured within 1 month after transplantation in the present study) were diagnosed by clinical signs and histopathological confirmation. In some rejection episodes, biopsy was contraindicated or not available due to logistic reasons, but all of these patients received standard rejection treatment based on clinical and biochemical data and responded to the treatment. The clinical and biochemical manifestations include systemic illness with fever and malaise in association with changes in bile characteristics (decreased output, pale color), increases in serum bilirubin, alkaline phosphatase, gamma glutamyl-transpeptidase, and concomitant rise in aminotransferases.

The standard treatment for acute rejection episodes in this study was a 1,000-mg bolus dose of intravenous methylprednisolone (Solu-Medrol; Pharmacia & Upjohn Co., Stockholm, Sweden), followed by a full recycling of prednisolone (i.e., tapering daily dose from 200 mg to 20 mg). In the case of steroid-resistant rejection, treatment was switched to tacrolimus/FK506 (Fujisawa) for cyclosporine patients or treatment for 7 to 14 days with OKT3 (Ortho-Biotech, Inc., Raritan, NJ). Permission for the present study was granted from the local ethics committee.

Isolation and Characterization of LSECs.

Human LSECs were freshly isolated from the liver of one healthy liver donor and characterized as described earlier.23 LSECs were routinely cultured in an endothelial cell selective medium, MCDB 131 (GIBCO BRL, Gaithersburg, MD), containing 10% heat-inactivated human AB serum. The medium was further supplemented with endothelial cell growth medium (EGM-2) SingleQuots obtained from Bio Whittaker (Clonetics; Walkersville, MD). Human aortic endothelial cells (HAECs) were purchased from Clonetics and grown in medium recommended by the suppliers. HAECs served as control cells. Single-color fluorescence was used to phenotypically characterize LSECs and HAECs.23 Antibodies were from Becton Dickinson (San Jose, CA), unless otherwise specified. The primary antibodies used for staining were anti-CD141 (thrombomodulin), anti-CD142 (tissue factor), anti-CD144 (vascular endothelial-cadherin), anti-CD105 (endoglin), anti-CD58, anti–HLA-DR, anti-AcLDL (Molecular Probes, Eugene, OR), anti–Ulex europaeus (Sigma, Munich, Germany), anti-CD106 (vascular cell adhesion molecule 1; Biogenesis, Poole, UK), anti-CD62E (E-selectin, Biogenesis), anti-CD31 (platelet-endothelial cell adhesion molecule), anti-FVIIIRAg (Dakopatts, Glostrup, Denmark), anti-CD80, anti-CD86 (R & D Systems, Abingdon, UK), and anti–human leukocyte antigen (HLA) class I (Dakopatts).

All experiments were performed using primary LSECs in passages 1 to 4, where one set of cell samples remained untreated while another set was stimulated with recombinant tumor necrosis factor α (TNF-α) and interferon γ (IFN-γ; 20 ng/mL and 200 ng/mL, respectively; R & D Systems), which were added to the culture medium overnight prior to harvesting of cells for analysis.

Determination of the Presence of LSEC Antibodies in Sera of Liver Transplant Patients.

For the flow cytometric assay, unstimulated and stimulated LSEC/HAEC were used, and the procedure was carried out as described earlier.24 The cells were analyzed on a Becton Dickinson flow cytometer (FACSorter; Becton Dickinson), as described earlier.23 A shift in the mean fluorescence of 20 channels in the test sample compared to negative control was considered positive.24 Appropriate positive and negative controls were included. All sera giving a positive reaction were further diluted (1:50, 1:100, 1:1000) in phosphate-buffered saline (PBS) to determine the titer of the antibodies.

Specificities of LSEC Antibodies.

We described previously a new and quick method for the isolation of HLA class I and class II antigens using paramagnetic microbeads.25 Microbeads coated with pooled HLA class I or II antigens were used for the removal of HLA antibodies from patient sera. The absorbed sera were retested for binding to LSECs. To further study whether binding of antibodies to LSECs modulated T-cell responses, a mixed cell culture (MCC) assay was used.

Mixed Cell Culture Assay.

Serum and CD4+ T cells obtained from each patient were used in this assay. Previously, we showed that LSECs express MHC class II upon activation with IFN-γ for 3 days.23 Therefore, for this assay, IFN-γ–activated LSEC were used. CD4+ T cells were magnetically isolated from PBMCs of patients using a commercially available kit (Dynal, Oslo, Norway). The procedure followed was that described by the manufacturer. For the MCC assay, 1 × 105 irradiated (35 Gy) IFN-γ–stimulated LSECs or HAECs were added to round-bottomed microtiter plates (Nunc-ImmunoPlate, MaxiSorp, Nunc, Denmark) in triplicate and allowed to adhere to the plastic. Sera from the patients with or without LSEC antibodies were added to each cell type and incubated for 1 hour. The plates were washed 5 times with PBS, and 1 × 105 CD4+ T cells from the same patients were added to the wells. The rest of the procedure was carried out according to standard protocol. Controls included MCCs with irradiated autologous cells (negative control) or allogeneic PBMCs (positive control). Results were expressed as mean counts per minute (cpm). A response was considered positive when it was more than 50% of the response of the autologous cell combination. Parallel control MCC experiments with endothelial cells and T cells—but no serum—from patients were also performed.

Cytokine Production in Supernatants of the Mixed Cell Cultures.

We investigated whether LSEC antibodies could alter the cytokine profile in the MCC supernatants of patients with LSEC antibodies compared to those without antibodies. The assay was performed as described in “Mixed Cell Culture Assay.”23 The cell culture supernatants were collected after 72 hours, sterile filtered and kept frozen at −70°C until assayed. The cytokines TNF-α, IFN-γ, interleukin (IL)1β, IL-2, IL-4, IL-10, and transforming growth factor β (TGF-β) were measured by standard sandwich enzyme-linked immunosorbent assay techniques using Quantikine sandwich enzyme immunoassay from R & D systems (Abingdon, UK). Assays were performed according to the manufacturer's instructions.

Determination of Costimulatory Molecule Expression on LSECs Induced by LSEC Antibodies.

To determine whether LSEC antibodies induced expression of CD86, purified immunoglobulin G (IgG) fractions were isolated from the sera of 5 patients with rejections and LSEC antibodies using goat anti-human IgG agarose beads (Sigma), according to standard procedure. IgG fractions were also purified from 5 patients without rejections and no LSEC antibodies. F(ab′)2 fractions of the IgG immunoglobulins was prepared as described earlier.23 LSEC (5 × 105) were incubated with IgG F(ab′)2 fragments (3 mg/mL) for 18 hours in 6-well culture plates. Immunocytochemistry was performed by enzyme staining. The cells were incubated with anti-CD80 or anti-CD86 antibodies (1:50) for 1 hour at room temperature, washed 3 times with PBS, and stained with secondary goat anti-mouse antibodies (1:500). The immunoperoxidase procedure was carried out using Vectastain Elite ABC kit (ImmunKemi, Stockholm, Sweden) as described by the manufacturer.

The purified IgG F(ab′)2 fragments were also used to confirm that the results obtained using the MCC assay and cytokine production in the MCC supernatants were indeed due to LSEC antibodies.

Blocking of the Lymphocyte Proliferative Response With Anti-CD86 Antibodies.

To determine whether the expression of CD86 induced by LSEC antibodies is important for the proliferative response, we blocked the response (n = 5) by addition of anti-CD86–blocking monoclonal antibodies (BD Pharmingen, San Diego, CA). For this purpose, the MCC assay was performed with the addition of 15 μg/mL (concentration determined in initial experiments) of anti-CD86 antibodies on day 2 of the MCC assay. An isotype control IgG antibody (15 μg/mL; BD Pharmingen) was also used in the assay.

Liver Biopsy Staining.

Liver biopsies were obtained from 10 patients with liver allografts (5 patients with rejections and LSEC antibodies before and after transplantation; 5 without rejections and with no detectable antibodies) and from 3 normal liver samples from transplant donors. Liver specimens were snap frozen in liquid nitrogen and stored at −70°C. Five-micrometer cryostat sections were fixed in acetone at room temperature, and immunohistochemistry was performed by enzyme staining. The sections were stained with F(ab′)2 fragments of goat anti-human IgG or IgM antibodies (Jackson ImmunoResearch, West Grove, PA). The immunoperoxidase procedure was carried out using Vectastain Elite ABC kit (ImmunKemi), as described by the manufacturers. The Vector NovaRed kit was used as color developer. Sections were counterstained with Mayer's hematoxylin (Sigma, Stockholm, Sweden).

Statistical Analysis.

Categorical outcome measures were analyzed by either chi-square test or Fisher exact test, as appropriate. Continuous outcome measures are presented in terms of medians and interquartile ranges (IQR), and differences between groups were assessed by a 2-sample t test or Mann-Whitney test depending on whether the data was normally distributed. Log-rank test was used to compare patient survival rate between different groups. All statistical analyses were done using Statistica version 6.0 (StatSoft Scandanavia AB, Uppsala, Sweden). Differences were considered significant if P was less than .05.


Patient Results.

Two-year patient survival rate was 86.3%. Fifty patients (52.6%) developed acute rejection episodes after liver transplantation. Thirty-eight (76%) of 50 of the rejection episodes were confirmed by biopsy result, and 7 of them were steroid-resistant rejections. There was no difference in acute rejection episodes between FK506-based (50.7%) and cyclosporine A–based (58.3%) immunosuppression regimens (P not significant).

Characterization of LSECs.

We previously isolated and characterized LSECs in our laboratory from a healthy normal liver.23 For the present study, we freshly isolated LSECs from another healthy normal liver and confirmed the unique phenotype of the isolated LSECs using electron microscopy and flow cytometry. The LSECs isolated in this study lacked expression of the molecules CD31, FVIIIRAg, and CD62E and had no basement membrane, but showed presence of fenestrae under electron microscopy (Fig. 1). The expressions of other adhesion markers, including the T-cell costimulatory markers CD80 and CD86, on LSECs and HAECs are shown in Table 2.

Figure 1.

Electron microscopic analysis of isolated LSECs. (A) Scanning electron microscope photomicrograph showing liver endothelial cells cultured in Transwell tissue culture inserts displaying fenestrations (arrows) in the cytoplasm. Scale bar, 1 μm. (B) Transmission electron microscopy image showing an endothelial cell (C) attached to the Transwell membrane (M) without any matrix (*); no formation of tight junctions between cells is seen. Scale bar, 200 nm.

Table 2. Expression of Adhesion Molecules on Human LSECs and HAECs
AntibodiesUnstimulated LSECsStimulated* LSECsUnstimulated HAECsStimulated* HAECs
  • Abbreviation: ICAM-1, intercellular adhesion molecule 1.

  • *

    Cells activated with TNF-α and IFN-γ for 18 hours.

Ulex europaeus++++
HLA class I++++

Significantly Higher Numbers of Patients With Rejections Have LSEC Antibodies.

LSEC antibodies were found in 2 (10%) of 20 healthy individuals. In total, 43 (45.3%) of 95 patients had LSEC antibodies either before and/or after liver transplant (Fig. 2A). LSEC antibody reactivity could be detected in all patient groups, but a higher frequency of patients with primary biliary cirrhosis, autoimmune hepatitis, and alcoholic cirrhosis had LSEC antibodies (7/8, 6/7, and 6/7, respectively) than other patients (Table 1).

Figure 2.

Presence and specificity of anti-sinusoidal endothelial cell antibodies in patients with liver transplants (tx). (A) Presence of LSEC antibodies in sera of patients with liver transplants was found before and/or after transplantation. Significantly higher numbers of patients with rejections had LSEC antibodies compared with patients without rejections (P < .0001). (B) A high fraction of patients had non-HLA antibodies. However, no difference in correlation between HLA and non-HLA antibodies with acute rejections was observed.

A significantly higher number of patients with rejections had LSEC antibodies: 35 (70%) of 50 compared with 8 (17.8%) of 45 without rejections (P < .0001; Fig. 2A). In 25 (71%) of 35 patients with rejection and LSEC antibodies, these antibodies were already detected before onset of clinically defined rejection episodes. LSEC antibodies bound to both unstimulated and cytokine-stimulated LSECs. Interestingly, a higher percentage of patients with LSEC antibodies had steroid-resistant rejections than those without LSEC antibodies (P < .05; Table 3). In general, patients with LSEC antibodies had a mixture of both IgM and IgG classes. A significantly higher titer of LSEC antibodies was detected in patients with rejections (1:500-1:1000) than in those without rejections (1:5- 1:100) (P < .001).

Table 3. Correlation Between LSEC Antibodies and Clinical Results
LSEC ReactivityNo. of PatientsMedian Age, yrFemale (%)Acute Rejections (%)Steroid-Resistant Rejections (%)Two-Year Survival (%)
  1. Abbreviation: ns, not significant.

LSEC antibodies (+)4351.021 (48.8)35 (81.4)6 (14.0)38 (88.4)
LSEC antibodies (−)5250.017 (32.7)15 (28.8)1 (1.9)44 (84.6)
P value nsns<.0001.03ns

We found that a lower number of patients (10/95; 10.5%) had antibodies against HAECs compared to LSECs. The presence of HAEC antibodies did not correlate with acute rejections.

Specificities of LSEC Antibodies.

Absorption assays with HLA class I and II antigen-coated magnetic beads indicated that a high fraction of LSEC reactivity was due to non-HLA antibodies (Fig. 2B). In the rejection group, HLA antibodies were detected in only 4 patients in the pretransplant period and 9 in the posttransplant period. It is important to remember that the HLA reactivity detected was not donor specific but against the HLA type expressed by the LSECs. No significant differences between patients with HLA antibodies and non-HLA antibodies were observed with regard to acute rejection episodes, steroid-resistant rejections, and 2-year patient survival rate (P not significant).

Patients With LSEC Antibodies and Rejections Had Significantly Increased CD4+ T-cell Proliferation in the MCC Assay.

The baseline T-cell proliferation in the MCC assay with LSEC and CD4+ T cells—but without addition of serum (antibodies)—is shown in Table 4. Addition of serum to the MCC showed that 29 (83%) of 35 patients with LSEC antibodies and rejections had significantly increased proliferation of T cells compared with 5 (13%) of 37 without antibodies and rejections (P < .0001). The median T-cell proliferation was 9,001 cpm (IQR, 7,609-9,612 cpm) in patients with LSEC antibodies and rejections compared with 422 cpm (IQR, 379-496 cpm) in patients without LSEC antibodies and rejections (P < .0001) (Fig. 3). However, the T-cell proliferation induced in the LSEC antibody and rejection-positive group was less efficient compared with allogeneic PBMCs (median, 21,268 cpm; IQR, 17,251-24,250 cpm). Interestingly, no proliferation (median, 974 cpm; IQR, 917-1,000 cpm) was induced in T cells by HAEC pretreated with antibodies (in patients with HAEC antibodies) compared with patients without HAEC antibodies (median, 846 cpm; IQR, 822-896 cpm; n = 10 in each group).

Table 4. CD4+ T-cell Proliferation and TGF-β Levels in Different Patient Groups
Patient GroupsNo. of PatientsT-cell Proliferation (counts/min)TGF-β (pg/mL)
Without Addition of Patient SerumWith Addition of Patient SerumWithout Addition of Patient SerumWith Addition of Patient Serum
  • NOTE. Values are shown as median (interquartile range).

  • *

    P < .0001.

  • P > .05.

LSEC antibodies (+)/Rejection (+)352,232 (2,100–2,897)*9,001 (7,609–9,612)*1,320 (1,105–1,371)*430 (405–452)*
LSEC antibodies (+)/Rejection (−)8515 (501–642)807 (749–940)1,904 (1,756–1,962)1,680 (1,623–1,701)
LSEC antibodies (−)/Rejection (+)153,000 (2,520–3,863)3,219 (3,000–4,100)1,110 (1,029–1,527)1,015 (963–1,180)
LSEC antibodies (−)/Rejection (−)37401 (315–456)422 (379–496)2,099 (1,921–2,220)2,009 (1,928–2,113)
Figure 3.

CD4+ T-cell proliferation in different patient groups. The median T-cell proliferation in patients with LSEC antibodies (Ab) and rejections (Rej.; 9,001 cpm; IQR, 7,609-9,612 cpm) was significantly higher than the 422 cpm (IQR, 379-496 cpm) in patients without LSEC antibodies and rejections (P < .0001).

LSEC Antibodies Induce Expression of CD86 on LSECs.

Because we found an increase in the proliferation of T cells of patients with LSEC antibodies, we investigated whether these antibodies could induce expression of the costimulatory molecules CD80 and CD86 needed for T-cell stimulation. Unstimulated and cytokine-stimulated LSECs did not express CD80 or CD86. However, treatment of cytokine-stimulated (IFN-γ for 3 days)—but not unstimulated—LSECs with purified IgG F(ab′)2 fragments from patients with LSEC antibodies for 18 hours induced expression of CD86 but not CD80 (Fig. 4), while IgG F(ab′)2 from patients without LSEC antibodies did not induce the expression of either CD80 or CD86. LSEC antibodies did not induce CD80 or CD86 expression on HAECs.

Figure 4.

Induction of CD86 expression on LSECs by LSEC antibodies. Unstimulated and cytokine-stimulated LSECs did not express CD80 or CD86. However, treatment of LSECs with purified IgG F(ab)2 fragments from patients with LSEC antibodies (Ab) for 18 hours induced expression of (A) CD86 but not (B) CD80. (Original magnification, ×60.) Flow cytometric analysis showed (C) the lack of expression of CD86 and MHC class II on nonactivated LSECs. (D) However, 3-day stimulation of LSECs with IFN-γ (200 ng/mL) induced expression of MHC class II (dashed line) but not CD86 (dotted line). (E) Treatment of IFN-γ–activated LSECs with IgG F(ab)2 fractions from patients with LSEC antibodies for 18 hours induced expression of CD86 (dotted line). Staining with only secondary antibodies served as negative control (gray line).

We found that addition of anti-CD86 antibodies to MCC giving a proliferative response significantly blocked the response (Fig. 5, P = .009), indicating that CD86 expression induced by LSECs is important for the proliferative response.

Figure 5.

Blocking of T-cell proliferation by anti-CD86 antibodies. T cells from liver patients with antibodies to LSECs gave a proliferative response in a mixed cell culture. However, the proliferative response was significantly reduced when blocked with anti-CD86 monoclonal antibodies (15 μg/mL, P = .009). No effect of blocking with the isotype control was observed.

Significantly Decreased Levels of TGF-β in the MCC Supernatants of Patients With LSEC Antibodies and Rejections.

Of the cytokines tested, the only cytokine found at significantly high levels was TGF-β in the supernatants of the mixed cell cultures (Table 5). The TGF-β levels were markedly decreased in MCC supernatants of patients with LSEC antibodies and rejections (median, 430 pg/mL; IQR, 405-452 pg/mL) compared with those without antibodies and rejections (median, 2009 pg/mL; IQR, 1928-2113 pg/mL) (P < .0001; Fig. 6). Moreover, within the LSEC antibody and rejection positive group, a significant difference in T-cell proliferation and TGF-β levels was obtained in MCC combinations with serum compared to those without serum (P < .0001, Table 4). The same was not observed in the patient group without rejections and antibodies (P not significant; Table 4). These findings were confirmed when purified fractions of LSEC antibodies—IgG F(ab)2 fragments—instead of whole serum were used in the MCC assay (Table 6).

Table 5. Cytokine Levels in Mixed Cell Cultures of LSECs With CD4+ T Lymphocytes and Sera From Patients With Liver Transplants
GroupsIL-1β (pg/mL)IL-2 (pg/mL)IL-4 (pg/mL)IL-10 (pg/mL)TNF-α (pg/mL)IFN-γ (pg/mL)TGF-β (pg/mL)
  1. NOTE. Values are shown as median (interquartile range).

LSEC antibodies (+)/Rejection (+) (n = 35)<30191 (189–199)<30<3050 (50–55)<30430 (405–452)
LSEC antibodies (+)/Rejection (−) (n = 8)<30<30<30<30<30<301,680 (1,623–1,701)
LSEC antibodies (−)/Rejection (+) (n = 15)<30152 (150–155)<30<3080 (25–91)<301,015 (963–1,180)
LSEC antibodies (−)/Rejection (−) (n = 37)<30<30<3053 (51–57)<30<302,009 (1,928–2,113)
Figure 6.

TGF-β levels in different patient groups. The TGF-β levels were markedly decreased in MCC supernatants of patients with LSEC antibodies (Ab) and rejections (Rej.; median, 430 pg/mL; IQR, 405-452 pg/mL) compared with those without antibodies and rejections (median, 2009 pg/mL; IQR, 1928-2113 pg/mL) (P < .0001).

Table 6. CD4+ T-cell Proliferation and TGF-β Levels Obtained in Mixed Cell Cultures Using Purified Fractions of LSEC Antibodies
 Addition of IgG F(ab)2 Fragments From Patients Without Rejections and LSEC Antibodies (n = 5)Addition of IgG F(ab)2 Fragments From Patients With Rejections and LSEC Antibodies (n = 5)
  1. NOTE. Values are shown as median (interquartile range).

T-cell proliferation (cpm)550 (100–620)*10,256 (9,850–10,612)*
TGF-β (pg/mL)2,100 (1,990–2,220)*415 (358–420)*

Deposition of Immunoglobulins in the Livers of Patients With LSEC Antibodies During Rejection Episodes That Bind to Sinusoidal Endothelial Cells.

Intense binding of IgG and IgM to LSECs (red-brown staining) was observed in livers of patients with LSEC antibodies and rejections (Fig. 7A-7D). Binding of antibodies to LSECs was not observed in patients with rejections but without LSEC antibodies (Fig. 4E-4F). No immunoglobulin deposition was observed in LSECs of normal livers (Fig. 7G).

Figure 7.

Deposition of immunoglobulins on LSEC during rejections. Intense binding of IgG and IgM to LSEC (red-brown staining) was observed in livers of patients with LSEC antibodies and rejections. (A-D) Representative pictures are shown from 2 different patients with LSEC antibodies during a rejection episode. (E, F) Binding of antibodies to LSECs was not observed in patients without LSEC antibodies but with rejections. (G) No staining of LSECs in normal livers for IgG and IgM was observed. (Original magnification, ×40.)


Our study demonstrates that the presence and binding of antibodies to LSECs is significantly associated with acute rejections of liver allografts. Furthermore, the LSEC antibodies induced (1) expression of the T-cell costimulatory molecule CD86 but not CD80 on LSEC, and (2) increased T-cell proliferation and decreased TGF-β levels in mixed cell cultures. In general, patients with rejections—compared to those without rejections—had higher T-cell proliferation in the MCC even in the absence of serum. However, addition of serum significantly and specifically increased T-cell proliferation and decreased TGF-β in the MCC of patients with LSEC antibodies and rejections compared to those without antibodies and rejections. The deposition of antibodies in LSECs detected in biopsies of patients with rejections further implicates the role of LSEC antibodies in facilitating rejections. Thus, our results suggest that LSEC antibodies may facilitate acute liver allograft rejections by down-regulating the growth arresting cytokine TGF-β26–29, thus up-regulating alloreactive T-cell proliferation. Although the mechanism by which TGF-β induces lymphocyte suppression is unknown, it is reported that high levels of TGF-β antagonize the effects of IL-1 on the induction of lymphocyte proliferation,29 inhibit IFN-γ production in the cultures of peripheral blood lymphocytes stimulated with IL-12,30 down-regulate expression of adhesion molecules, monocyte chemoattractant protein, and TNF receptors on endothelial cells,31 and induce antigen-specific unresponsiveness in naive T cells.32 Thus by inhibiting important cytokines and adhesion molecules involved in proliferation and activation of T cells, TGF-β may contribute to inhibition of T-helper 1 (Th1)–like responses in lymphocytes. Therefore, decrease in TGF-β levels by anti-LSEC antibodies might favor Th1–like responses, resulting in proliferation of alloreactive T lymphocytes leading to rejections.

Specificity studies indicated that a high fraction of the patients had antibodies to non-HLA compared to HLA antigens. However, no difference in correlation between the presence of HLA and non-HLA antibodies with regard to rejections was observed. Nevertheless, history of the immunization status of the patients in our study indicated that most patients with LSEC antibodies prior to transplantation were not immunized. Therefore, non-HLA autoantibodies may play an important role in the rejections of liver allografts. HLA antibodies were mainly detected in the posttransplant period. Whether the pathophysiological role of the autoantibodies in the pretransplant period is different from the alloinduced HLA/non-HLA antibodies in the posttransplant period is currently not known. We have previously shown that anti-LSEC autoantibodies transform LSEC into a vascular type and may therefore play an important role in the development of hepatocellular failure and portal hypertension in primary biliary cirrhosis and autoimmune hepatitis patients.23 Whether capillarization of LSEC occurs during liver allograft rejections and, furthermore, whether capillarized LSECs support proliferation of lymphocytes is currently being determined at our laboratory. Ongoing studies at our center using a panel of LSECs from different individuals indicate that the non-HLA antigen(s) recognized by LSEC antibodies may not be polymorphic (preliminary results). Furthermore, our unpublished observations show that LSEC antibodies are not directly cytotoxic to LSECs, as determined in a microcytotoxic assay.

In vitro experiments have shown that endothelial cells can function as APC to CD4+ T cells,15 and murine LSECs may act as APC in cell-mediated immune responses in the liver.12, 14 Recently, we have shown that purified isolates of human LSECs do not constitutively express HLA class II, but expression can be induced by IFN-γ.23 In addition, nonactivated or cytokine-activated human LSECs do not express the 2 costimulatory molecules CD80 and CD86. However, in this study we demonstrate that LSEC antibodies induce the expression of CD86 on LSECs. Thus, LSECs in patients with LSEC antibodies during rejections may develop a phenotype that facilitates antigen presentation. However, the antigen-presenting capacity of human LSECs still remains to be elucidated. Although T-cell proliferation by human LSECs was less efficient than proliferation by alloreactive APCs, this property was exhibited by LSECs but not by HAECs. Use of the purified IgG F(ab)2 fractions of the LSEC antibodies confirmed that it was the LSEC antibodies in the serum that induced expression of CD86 and T-cell proliferation.

Previous studies have demonstrated various sinusoidal abnormalities during rejection.21, 33, 34 Ischemia/reperfusion (I/R) injury to the liver is frequently induced after liver transplantation and surgery, leading to local or systemic organ dysfunction. Dysfunction in the LSEC of human liver allografts during transplantation has been shown to be an early event that is critical in I/R injury and probably plays a key role in primary liver dysfunction after transplantation.35 In fact, functional abnormalities of LSECs in rats with acute liver rejections have been reported.22 LSECs play a key role in hepatic clearance. It has been shown that LSECs possess endocytic function for various types of substances, such as circulating IgG immune complexes and hyaluronic acid.11 Thus, damage to LSECs in the early posttransplant period by various mechanisms, including I/R, may prevent efficient clearance of LSEC antibodies, resulting in antibody-facilitated cell-mediated immunity and rejections.

Detection of LSEC antibodies in the pre- and/or postliver transplant period would identify patients who are at a high risk for acute rejections. Importantly, in a significantly high fraction of patients with rejections and LSEC antibodies, these antibodies were already detected before transplantation. Detection of these antibodies in the pretransplant period will be important in tailoring individual immunosuppressive regimens. Thus, processes such as I/R, known to damage important targets, such as endothelial cells, may be factors that further increase the risk for rejections in patients with LSEC antibodies. This suggests that clinical transplantation protocols may benefit from reevaluation to improve cold preservation-warm reperfusion conditions as well as to include immunosuppressive agents that suppress antibody-mediated responses. Furthermore, marginal livers that are more vulnerable to I/R and other injuries are widely used nowadays to expand the donor pool.36 It may also be advisable not to allocate marginal livers to liver patients already known to have LSEC antibodies.

In conclusion, we demonstrate a novel mechanism of liver allograft rejection facilitated by antibodies to LSECs in patients with liver transplants. The LSEC antibodies may facilitate acute rejection by down-regulating TGF-β and up-regulating T-cell proliferation via induction of CD86 expression.