Immune role of hepatic TLR-4 revealed by orthotopic mouse liver transplantation


  • Beena John,

    Corresponding author
    1. The David H. Smith Center for Vaccine Biology and Immunology, The Aab Institute for Biomedical Research, University of Rochester, Rochester, NY
    • David H. Smith Center for Vaccine Biology and Immunology, University of Rochester Medical Center, 601 Elmwood Avenue, Box 609, Rochester, NY 14642, USA
    Search for more papers by this author
    • fax: 585-273-2452

  • Ingo Klein,

    1. The David H. Smith Center for Vaccine Biology and Immunology, The Aab Institute for Biomedical Research, University of Rochester, Rochester, NY
    2. Department of Surgery, University Hospital Wuerzburg, Julius-Maximilians University, Wuerzburg, Germany
    Search for more papers by this author
  • I. Nicholas Crispe

    1. The David H. Smith Center for Vaccine Biology and Immunology, The Aab Institute for Biomedical Research, University of Rochester, Rochester, NY
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.


Activated CD8+ T cells migrate to the liver at the end of an immune response and go through apoptosis there, but this mechanism is impaired in mice lacking Toll-like receptor-4. This allowed us to test the importance of liver trapping in an ongoing immune response. In the absence of Toll-like receptor-4, reduced liver accumulation was associated with an increase in the circulating CD8+ T cell pool, more long-lived memory T cells and increased CD8+ T cell memory responses. Using experimental orthotopic liver transplantation, we showed that the effect of Toll-like receptor-4 on the formation of the CD8+ T cell memory resides in the liver. Conclusion: These studies reveal a new function for the liver, which is to regulate the magnitude of T cell memory responses through a Toll-like receptor-4–dependent mechanism. (HEPATOLOGY 2007;45:178–186.)

The ability to respond to a pathogen more vigorously upon second exposure is a central feature of adaptive immunity. Primed CD8+ T cells go through massive clonal expansion and systemic dissemination,1, 2 followed by extensive T cell apoptosis.3 A small population of effector CD8+ T cells survives the contraction phase, and is the source of memory cells.2 Many factors influence the size of the memory CD8+ T cell pool,4–6 but the migratory pattern of effector T cells has not been assigned a key role.

Primed T cells migrate to multiple tissues,7–9 and among these the liver is a preferential site for the accumulation of activated CD8+ T cells.10 Such trapping is associated with T cell apoptosis in diverse situations, including mouse models driven by antigenic peptide, in SIV infection, and during influenza.11–14 The activated CD8+ T cells are trapped in the liver primarily through ICAM-1 and VCAM-1,15 and undergo apoptosis due to TNF-α.16 Thus, the liver acts as a sink for activated CD8+ T cells.

Constant exposure of the liver to lipopolysaccharide (LPS) from the intestinal bacteria contributes to the unique hepatic microenvironment.17, 18 LPS tolerance prevents continuous liver inflammation,19, 20 but constitutive exposure to low levels of LPS may increase the expression of adhesion molecules.19, 21 We recently showed that the LPS receptor, Toll-like receptor-4 (TLR-4), promotes the trapping of activated CD8+ T cells in the liver.22 Without TLR-4, the liver trapped fewer activated CD8+ T cells, leading to an increase of these cells in the circulation. Thus, the liver acts as a negative regulator during the acute CD8+ T cell response.

The blood-borne primed T cell pool gives rise to tissue memory populations.23 In this report, we exploited the TLR-4–dependent intrahepatic trapping of CD8+ T cells to test whether the previously documented effect on short-term T cell distribution in vivo modulates the formation of T cell memory. Our data show that lack of TLR-4 is associated with increased CD8+ T cell memory responses in multiple tissues. Using experimental orthotopic liver transplantation, we show that TLR-4 expressed in the liver is important for this effect, and thus reveal a novel function of the liver as an immunomodulatory organ. This could prove to be an important tool in vaccination and adoptive immunotherapy, where better memory responses might be achieved by limiting the migration of activated CD8+ T cells to the liver.


APC, antigen-presenting cell; CFSE, carboxyfluorescein diacetate succinimidyl ester; DC, dendritic cell; LPS, lipopolysaccharide; TLR-4, Toll-like receptor-4.

Materials and Methods


C57BL/10SnJ mice (with normal TLR-4, referred to as WT) and C57BL/10ScN mice (which lack the TLR-4 gene, referred to as TLR-4–deficient mice) were from the Jackson Laboratory (Bar Harbor, ME, USA), and OT-1 T cell receptor transgenic mice were maintained in-house on the CD45.1 background. Mice were housed under specific pathogen-free conditions and used at 6-8 weeks of age. All experiments were approved by the Institutional Animal Care and Use Committee.

Adoptive Transfer Experiments.

Purified CD8+ T cells were isolated from pooled peripheral lymph node and spleen cells of CD45.1 OT-1 mice using antibody-mediated depletion of B cells, monocytes, and dendritic cells (DCs) with magnetic beads. OT-1 T cells (5 × 106, of which >90% were CD8+) were injected i.v. into recipient mice. In some experiments, the T cells were labeled using 5 μM carboxyfluorescein diacetate, succinimidyl ester (CFSE) for 10 minutes at 37°C, followed by 2 washes in Hank's balanced salt solution. T cells were activated in vivo using spleen-derived DCs,24 injected i.p. 24 hours later. These DCs were pulsed with SIINFEKL peptide (1 μM), and the cultures were analyzed by fluorescence-activated cell sorting (FACS) to determine the frequency of CD11c+ cells. CD11c+ cells (5 × 105) were injected i.p. into either WT or TLR-4–deficient hosts. Mice were challenged 6 weeks later with 3 daily injections of 25 nmol of SIINFEKL peptide in sterile saline.

Cell Harvest.

Lymph node, spleen, liver lymphocytes, peripheral blood cells, and bone marrow cells were harvested from mice during the primary response, 6 weeks after priming, or 24 hours after the last dose of the peptide challenge. Liver lymphocytes were isolated by mechanical homogenization of the liver, treatment with collagenase (0.05% wt/vol) and DNase (0.002% wt/vol) for 45 minutes at 37°C, low-speed (30g) centrifugation to deplete hepatocytes, then separation using a 22% Optiprep gradient (Accurate Chemical, Westbury, NY, USA).

In Vivo Cytotoxic Assay.

Mice were injected i.v. with a 50:50 mixture of 107 CFSE-high and 107 CFSE-low C57BL/10 spleen cells, labeled with 2 μM and 0.2 μM CFSE, respectively.22 The CFSE-high cells were additionally pulsed with 1 μM SIINFEKL peptide for 1 hour (specific targets), while the CFSE-low cells were incubated in saline (control targets). Lymphoid cells were harvested from multiple tissues after 5 hours. The percent specific lysis was calculated as: [1 − (%specific targets/%control targets)] × 100.

Liver Transplantation.

Orthotopic liver transplantation was performed following the procedure of Steger et al.25 In brief, the recipient's liver was removed under isoflurane anesthesia, and the donor liver installed with reconstruction of the superior and inferior vena cava, the portal vein, and the bile duct. Full technical details were as described in previous25, 26 publications. This technique does not reconnect the hepatic artery, but such liver transplants are histologically normal with no ischemic features.25

Staining and Flow Cytometry.

Cells were stained for surface markers using standard techniques. Caspase-3 activity was detected using a CaspaTag caspase-3 (DEVD) activity kit (Intergen, Purchase, NY, USA) following manufacturer instructions. Intracellular cytokines were detected by culture of lymphoid cells in media containing 50 U/mL of recombinant IL-2 (Endogen) and 1 μM GolgiPlug (BD Biosciences), in the presence or absence of antigen (1 μM SIINFEKL). After 6 hours, cells were washed and stained for surface markers, then fixed, permeabilized, and subject to intracellular staining using the Cytofix/Cytoperm kit (BD Pharmingen). Stained cells were analyzed using a BD Biosciences FACSCalibur instrument and Cellquest software.

Statistical Analysis.

Statistical significance was tested using either the Student's t test (unpaired, 2-tailed) or using a 2×3 factorial or 2×4 factorial ANOVA for independent variables (Vassar Stats; In all the cases, P < 0.05 was considered significant.


Memory CD8+ T Cells in TLR-4–Deficient Mice.

TLR-4 regulates the trafficking of activated CD8+ T cells to the liver.22 On this basis, we formulated the hypothesis that the reduced trapping of activated CD8+ T cells in the liver of TLR-4–deficient mice would make more cells available to enter the peripheral pool of primed CD8+ T cells. To test this, adoptively-transferred OT-1 T cells were primed in vivo using WT SIINFEKL-pulsed DCs in either WT or TLR-4–deficient host mice. Using this priming protocol, T cell activation was identical between the WT and the TLR-4–deficient mice.22 Peripheral blood samples were analyzed to detect OT-1 T cells in the circulation, and at 3 and 5 days after in vivo priming, the OT-1 T cells were more abundant in the blood of TLR-4–deficient mice (Fig. 1A). This is consistent with the lack of trapping of these cells in the liver.22 To test the alternate possibility that the greater abundance of OT-1 cells was due to differential apoptosis, lymph node, spleen, and liver lymphocytes from WT versus TLR-4–deficient host mice were analyzed by staining for caspase-3 activity. The frequency of apoptotic OT-1 T cells was low (around 5%) in all tissues at day 3, higher in all tissues, and highest in the liver (around 12%) at day 5, but there was no difference in the caspase-3 activity in OT-1 T cells undergoing activation in WT versus TLR-4–deficient mice (Fig. 1B).

Figure 1.

CD8+ memory precursors in TLR-4–deficient mice. (A) OT1 T cells (CD45.1+CD8+) in the peripheral blood of either WT (closed symbols) or TLR-4–deficient mice (open symbols) at various time points after primary immunization with peptide-pulsed APCs. (B) (top) Caspase 3+ cells among OT1 cells in the spleen, lymph nodes, and liver of WT and TLR-4–deficient mice on the peak of the response (day 5). (B) (bottom) The average percentage of Caspase3+ OT1 cells from the different organs of WT and TLR-4–deficient mice (n = 6 per group) on days 3 and 5. (C) The percentage of the OT1 cells in the spleen, liver, bone marrow, and lymph nodes of the WT (black bars) or TLR-4–deficient mice (open bars) 6 weeks after primary immunization with peptide. The differences between the WT and TLR-4−/− mice were significant (P = 0.025) additively in the 4 tissues, as tested by a 2×4 factorial ANOVA (n > 12 for all groups).

To determine whether the increase in the circulating pool of OT-1 T cells resulted in increased formation of memory T cells, we determined the frequency of OT-1 T cells in the spleen, liver, bone marrow, and lymph nodes of such mice 6 weeks after in vivo priming. In all these tissues, there was an approximately 1.5-fold increase in OT-1 T cells (CD45.1+) in the TLR-4–deficient mice compared to the WT mice (Fig. 1C). This increase was significant (P = 0.025) using a 2×4 factorial ANOVA, for all the 4 tissues (liver, spleen, lymph nodes, and bone marrow) additively.

Quality of Memory CD8+ T Cells in TLR-4–Deficient Mice.

Memory OT-1 T cells isolated from both WT and TLR-4–deficient mice expressed a normal pattern of cell surface markers (Fig. 2) and were functional in terms of their capacity to synthesize interferon-gamma (IFN-γ). This was revealed by a 5-hour in vitro culture of these cells with SIINFEKL peptide, in the presence of GolgiPlug. In the absence of antigen, cytoplasmic staining revealed IFN-γ in 1%-5% of the OT-1 cells (Fig. 2, right panels, under both WT and TLR-4−/− headings). In the presence of the SIINFEKL peptide, between 48% and 77% of the OT-1 T cells contained detectable IFN-γ. There was no consistent difference between T cells primed in WT versus TLR-4–deficient hosts in the capacity to synthesize IFN-γ.

Figure 2.

The CD8+ memory T cell precursors primed in WT and TLR-4–deficient host are functionally and phenotypically identical. The expression of the activation markers CD62L, CD44, and CD127 on the OT1 cells in WT (left panel) or TLR-4–deficient mice (right panel), 6 weeks after primary immunization with peptide. Also shown is the production of IFNγ by the OT1 cells after 6 hours of restimulation with or without SIINFEKL peptide in culture (n > 9 in each group).

Memory Responses in TLR-4–Deficient Mice.

When OT-1 T cells were primed in TLR-4–deficient mice, memory T cells were qualitatively normal but more abundant. Thus, it could be predicted that the memory T cell response would be enhanced in these mice. To test this, we evaluated the numbers of OT-1 T cells in liver, lymph nodes, and spleens of DC-primed mice before and after a challenge with SIINFEKL peptide in saline. When DC-primed OT-1 T cells were challenged in WT versus TLR-4–deficient mice, the percentage (Fig. 3, left panels) and absolute number (Fig. 3, right panels) of OT-1 T cells was increased in the TLR-4–deficient mice. The increase in cell numbers was approximately 2.5-fold and was statistically significant (P < 0.05) in all 3 tissues.

Figure 3.

T cells primed in TLR-4–deficient mice show better recall responses 6 weeks after immunization. Percentage (A) and numbers (B) of OT1 T cells in the liver, lymph nodes, and spleens of WT or TLR-4–deficient mice 6 weeks after primary immunization (1°) with SIINFEKL peptide-pulsed APCs. In the secondary challenge (2°), the mice either received phosphate-buffered saline (PBS) or SIINFEKL peptide, and all the responses were measured on day 3 after secondary challenge. n = 11 in each group. The significance values were obtained using the Student's t test (unpaired, 2-tailed).

We tested whether these effector cells were functional. In all 3 tissues, the OT-1 T cells maintained CD44 expression, lost CD62L from a proportion of the cells, and lost CD127 from all of the cells (Fig. 4A); these changes were similar in OT-1 T cells from WT and TLR-4–deficient hosts. Cultured without antigen in vitro, these cells synthesized IFN-γ, with 13%-29% positive; however, no difference was found between the OT-1 cells taken from mice with or without TLR-4. When such OT-1 T cells were restimulated in vitro with SIINFEKL peptide, the frequency of IFN-γ–synthesizing cells was increased, to 49%-59% in the examples shown. Again, there was no difference between OT-1 T cells taken from WT and from TLR-4–deficient hosts.

Figure 4.

Functional secondary effectors generated in TLR-4–deficient mice. (A) expression of activation markers CD62L, CD44, and CD127 on OT1 cells from WT (left panels) or TLR-4–deficient (right panels) mice that were immunized with peptide-pulsed DCs and challenged with either PBS or peptide 6 weeks later. (B) Target cell lysis (in vivo cytotoxicity) in the spleen, lymph nodes, and livers of WT and TLR-4–deficient mice 6 weeks after primary immunization with peptide-pulsed APCs. The mice were either given a secondary challenge with antigenic peptide in vivo (SIINFEKL) or left unchallenged (PBS). n = 6 in each group.

The prototypic function of effector CD8+ T cells is cytotoxicity, so we tested this activity using an in vivo cytotoxic assay. Six weeks after priming, both WT and TLR-4–deficient host mice contained functional “effector memory” cytotoxic T lymphocytes against SIINFEKL-loaded targets (Fig. 4B, marked “PBS”). Thus, around 30%-40% of the specific targets were eliminated in these mice. After peptide challenge, the specific in vivo killing was increased to around 90%, but there was no difference between WT and TLR-4–deficient host mice.

Taken together, the experiments shown in Figs. 3 and 4 show that in TLR-4–deficient mice, there were more CD8+ memory T cells, but these cells were qualitatively normal in terms of their phenotype and effector function.

Increased Secondary Clonal Expansion in TLR-4–Deficient Mice.

Our data show that higher secondary responses in the TLR-4–deficient mice were associated with the higher percentage of memory cells in these mice. However, the increase in the frequency of the OT1 memory precursors was approximately 1.5-fold (Fig. 1C), while the increase in the magnitude of the secondary response was greater, about 2.5-fold (Fig. 3). This raised the possibility that TLR-4 was also regulating the magnitude of secondary expansion. To test this, we isolated CD8+ T cells from WT or TLR-4–deficient mice 6 weeks after we primed with peptide-pulsed antigen-presenting cells (APCs), and injected 0.5 × 106 CFSE-labeled CD45.1+ CD8+ T cells into new recipients together with SIINFEKL peptide.

Figure 5 shows individual examples of WT or TLR-4–deficient mice that had received an adoptive transfer of OT-1 T cells primed and maintained in either a WT or a TLR-4–deficient host. On day 0 (immediately after adoptive transfer), the OT-1 T cells were present at 0.10%-0.15% in the peripheral blood, and all of these cells were CFSE-high. After 3 days, the T cells underwent cell division, revealed by dilution of their CFSE staining. In the examples shown, the OT-1 T cells expanded to 0.52%-0.61% of lymphoid cells in the WT host mice, regardless of whether the OT-1 T cells were taken from a WT or a TLR-4–deficient donor (Fig. 5A). In contrast, the OT-1 T cells primed in a WT donor and transferred into a TLR-4–deficient host expanded to 1.18%; these differences were reproducible (Fig. 5B). Thus, OT-1 T cells primed in a WT primary host gave rise to a larger secondary expansion in a TLR-4–deficient secondary host. These data show that the lack of TLR-4 results in increased secondary clonal expansion in response to a peptide challenge. Thus, TLR-4 modulates the magnitude of memory CD8+ T cell responses by a compound effect that influences both memory cell numbers and secondary expansion.

Figure 5.

Secondary clonal expansion is controlled by host TLR-4 expression. Memory cells generated in the WT or TLR-4–deficient mice, transferred in equal numbers into WT mice, expanded to the same extent. However, they expand more when transferred into a TLR-4–deficient host. (A) percentage of the OT1 memory cells generated in either WT or TLR-4−/− mice that were re-transferred into either WT or TLR-4–deficient mice. The responses seen are before (day 0) and 3 days after challenge with SIINFEKL peptide in saline (day 3). The dilution of CFSE by the OT1 cells (CD45.1+) in the peripheral blood is also shown in (A). (B) Shown are the average percentages of OT1 memory cells that have expanded in the spleen, lymph nodes, peripheral blood, and liver of WT or TLR-4–deficient hosts 3 days after challenge with peptide (n = 6 in each group). The significance values were obtained using the Student's t test.

TLR-4 Expressed in the Liver Regulates CD8+ Cell Responses.

TLR-4 controls the acute trapping of CD8+ T cells in the liver,22 and also the abundance of memory CD8+ T cells and the magnitude of secondary responses (Figs. 1–5). To test whether these effects were linked to TLR-4 acting in the liver, we determined the effect on CD8+ T cell memory of the replacement of a WT with a TLR-4–deficient liver. Orthotopic liver transplants were performed to create control animals in which a WT liver was transplanted into a WT host, and experimental mice in which WT hosts received a TLR-4–deficient liver. After 4 weeks, the mice received an adoptive transfer of OT-1 T cells and were primed with WT SIINFEKL-pulsed DCs. Six weeks later, these mice were challenged with 3 i.p. injections of SIINFEKL peptide in saline, and the clonal expansion of the OT-1 T cells was documented by flow cytometry.

In WT mice that received a normal B6 liver (WT → WT), secondary clonal expansion was evident as an increased percentage of OT1 cells in the spleen, lymph nodes, and liver in the SIINFEKL-challenged mice (Fig. 6A, left side). In WT recipients carrying a TLR-4–deficient liver (TLR-4 → WT), the clonal expansion was greater (Fig. 6A, right side). Thus, both the percentage (Fig. 6A) and absolute number (Fig. 6B) of OT1 T cells were increased in multiple tissues of the mice that received a TLR-4–deficient liver.

Figure 6.

WT mice transplanted with TLR-4–deficient livers show the same phenotype as that seen in intact TLR-4–deficient mice. Shown are the (A) percentages and (B) cell numbers of OT1 TCR transgenic cells in the liver, lymph nodes, and spleens of WT mice that were transplanted with WT livers (WT → WT) or WT mice that were transplanted with TLR-4−/− livers (TLR-4 → WT). The primary immunization was with peptide-pulsed APCs, and mice were rechallenged with either PBS or SIINFEKL peptide 6 weeks after primary immunization. The data shown are an average of 6 mice per group. The significance values were obtained by a 2×3 factorial ANOVA (VassarStats).

This experiment shows that TLR-4 in the liver has a profound effect on the magnitude of systemic immune responses. In particular, hepatic TLR-4 negatively regulates the magnitude of a CD8+ T cell memory response. However, this experimental design cannot distinguish which population of liver cells is most important in sensing LPS via the TLR-4. The transplanted liver in our experiments would be predicted to contain a mixture of recipient-derived (TLR-4 intact) and donor-derived (TLR-4–deficient) Kupffer cells,27 in the context of its TLR-4–deficient parenchyma.


The role of the liver as a unique tolerance-inducing organ was first revealed through early transplantation experiments,28 and subsequently through its effects on the induction of oral and portal vein tolerance.29 The liver is also capable of sustaining effective immune responses to pathogens,30, 31 which suggests that a complex interplay of factors shifts the balance toward either intrahepatic tolerance or immunity.32, 33 We and others have shown that the liver also plays an important role in clearing activated CD8+ T cells during a systemic CD8+ T cell response,13 but the consequences for immune memory are not understood.

We have shown previously that TLR-4 signaling promotes the localization of circulating activated CD8+ T cells to the liver, both in a short-term homing experiment and during an in situ immune response.22 The mechanism by which TLR-4 mediates this effect remains unclear. Because ICAM-1 is important for the retention of activated CD8+ T cells in the liver,15 we measured ICAM-1 RNA in livers of WT and TLR-4–deficient mice by reverse transcription-PCR, but found no difference; nor did a blocking antibody against ICAM-1 differentially affect trapping of activated CD8+ T cells in WT versus TLR-4 mutant mice (data not shown). We conclude that TLR-4 does not modulate ICAM-1 in the liver.

As a consequence of reduced intrahepatic CD8+ T cell trapping in the TLR-4–deficient mice,22 we detected more activated CD8+ T cells in the peripheral blood at the early time points (Fig. 1). Because the frequency of apoptotic cells among the activated OT1 cells at the peak of liver accumulation was not different between the WT and TLR-4–deficient mice, the effect of TLR-4 in the liver was primarily on trapping and not on the apoptosis of intrahepatic CD8+ T cells. However, because fewer cells were trapped in the livers of the TLR-4–deficient mice, there were also fewer dying cells in the liver, and this can explain the greater percentage of cells seen in the peripheral circulation. Although the difference in each of the peripheral tissues tested was small, additively there was a significantly higher percentage of total memory precursors in the TLR-4–deficient mice than in the WT mice. While the higher secondary responses seen in TLR-4–deficient mice could be partially explained by the higher T cell memory precursor frequency, the absence of TLR-4 during secondary expansion also contributed to this increase. The 2.5-fold difference in memory responses seen in our experimental model would be unlikely to make a difference to protective memory; however, the experiments reveal a biological mechanism that may be more important after weak priming against “stealth” pathogens.

Models for CD8+ T cell memory generation evolve constantly. The current understanding involves distinct phases of clonal expansion, clonal contraction, and the memory phase in which T cell numbers are stabilized and homeostatically maintained.6 Recent studies indicate differences in the rate of apoptosis of activated cells that migrate to the lymphoid versus nonlymphoid compartments;30 however, experiments involving adoptive transfer of memory cells from various nonlymphoid tissues34 and the use of parabiotic mice23 make it clear that such tissue memory cells are capable of recirculation.

On the basis of data presented here, we propose a modification to current models of CD8+ T cell memory generation. Activated T cells traffic through various tissue compartments, including the liver, which preferentially sequesters activated CD8+ T cells and not simply T cells already undergoing apoptosis.12 Sequestration starts as soon as activated CD8+ T cells leave priming sites and begin to circulate in the blood. In the liver, a proportion of the trapped CD8+ T cells are subjected to apoptosis. Our model is that at each passage through the liver, activated CD8+ T cells that have not entered either lymphoid or nonlymphoid tissues will be depleted. Thus, the liver acts as a “sink” for activated T cells that do not rapidly localize either to sites of infection or to sites where they can mature into long-lived memory cells. This interpretation fits the available data better than the earlier “graveyard” model, in which the liver was thought to sequester T cells already committed to apoptosis. Thus, the liver controls the size of the memory CD8+ T cell pool by modulating the contraction phase of the effector CD8+ T cell response. The liver carries out this function through a TLR-4–dependent mechanism.