Spatiotemporal dynamics of effector CD8⁺ T cell responses within the liver

Review on the dynamic imaging of hepatic effector CD8⁺ T cell migration and function.

Abstract

CD8⁺ T cells play a critical role in controlling hepatotropic viral infections, such as those caused by hepatitis B and hepatitis C viruses. The capacity of these cells to protect against such pathogens is mediated by antigen-experienced effector cells and relies on their ability to home to the liver, recognize pathogen-derived antigens, and deploy effector functions. Here, we review how dynamic imaging of hepatic effector CD8⁺ T cell migration and function in mouse models of hepatitis B virus pathogenesis has recently revealed a unique and novel mode of adaptive immune surveillance. Circulating effector CD8⁺ T cells initially arrest within liver sinusoids by docking onto adherent platelets and then actively crawl along the liver vasculature, probing hepatocytes for the presence of antigens by extending protrusions through the fenestrated sinusoidal endothelial cells. Hepatocellular antigen recognition and effector functions occur while CD8⁺ T cells are still confined to the intravascular space and are inhibited by the pathologic processes that characterize liver fibrosis. A detailed understanding of the spatiotemporal dynamics of effector CD8⁺ T cells within the liver is important for the rational design of targeted immunotherapeutic approaches for chronic liver infections.

Introduction

Efficient priming of antiviral T cell responses is thought to occur within secondary lymphoid organs (primarily the spleen and lymph nodes). This seems to be the case even during infection with viruses, such as HBV, whose tropism is restricted to the parenchymal cell of the liver, the hepatocyte. Despite the absence of data during natural infection, recent evidence obtained in mouse models indeed indicates that when primed by HBV-expressing hepatocytes, naïve, HBV-specific CD8⁺ T cells receive inhibitory signals that result in the expansion of defective cells that do not express IFN-γ or the cytolytic molecule granzyme B [1].

Functional effector T cells that have expanded in secondary lymphoid organs, instead, are released into the circulation and can eventually migrate to infection sites to perform their protective role. In most instances, this is made possible as activation-dependent signals program T cells to express a variety of homing molecules that are required to enter specific nonlymphoid tissues [2–4]. For instance, the chemokine receptor CCR6 plays a key role in effector T cell migration to the CNS [5]; the integrin α1β1 and the chemokine receptors CXCR3 and CCR5 regulate migration of effector T cells to the lungs [6]; effector T cells migrating to the skin up-regulate E-selectin ligands and chemokine receptors CCR4 and CCR10 [7]; and effector T cells migrating to mucosal tissues express the integrin α4β7, which binds to MAdCAM-1 [8, 9]. Whereas the nature of these and other tissue-specific T cell homing signals has been elucidated in the last few years [4], the in vivo requirements regulating T cell trafficking within the liver have lagged behind until very recently.

A number of observations pertaining to leukocyte subsets other than T cells suggest that the liver may not follow the classic multistep leukocyte migration paradigm involving rolling, adhesion, and extravasation from postcapillary venules [2, 10]. For example, leukocyte adhesion to the endothelium of hepatic blood vessels in response to a chemotactic stimulus occurs primarily within liver sinusoids (not in postcapillary venules) and in the absence of any notable rolling. This is consistent with selectins not being required for leukocyte recruitment into the inflamed liver vasculature [11]. Notably, LSECs lack tight junctions between cells, as well as a fully formed basal membrane, and contain numerous fenestrae of up to 200 nm in diameter [12]. This is in stark contrast to most microvascular beds in other tissues and organs, where a continuous endothelial cell layer and a basement membrane physically separate parenchymal cells from circulating leukocytes [12]. Thus, liver sinusoids provide the opportunity for direct interaction of circulating cells with underlying hepatocytes [13, 14].

Effector CD8⁺ T cell TRAFFICKING WITHIN THE LIVER

Recent data indicate that effector CD8⁺ T cells circulating through the liver initially arrest within sinusoids (not postcapillary venules), and they do so independently of selectins, Gαi-coupled chemokine receptors, β2- and α4-integrins, PECAM-1, and vascular adhesion protein 1 (all previously thought to be variably important for leukocyte trafficking in other organs) [15]. In keeping with the notion that the initial sinusoidal arrest occurs regardless of the location of antigen-producing hepatocytes, cotransfer of MHC-matched and MHC-mismatched antigen-specific effector CD8⁺ T cells into recipients that did or did not express cognate antigen demonstrated that hepatocellular antigen recognition is also not required for the arrest and initial accumulation of effector CD8⁺ T cells in the liver [15].

Several studies in mouse models of HBV immunopathogenesis showed that platelets play a critical role in the hepatic recruitment of effector CD8⁺ T cells [15–17]. Indeed, the first of those studies demonstrated that platelet depletion is associated with a significant reduction in the hepatic accumulation of effector CD8⁺ T cells 1–2 d post-transfer and with a proportional reduction in liver disease [16]. Both phenotypes are restored upon reconstitution with platelets, unless platelets are treated with the activation inhibitor PGE₁ before transfusion [16]. The notion that platelet activation is required for the hepatic accumulation of effector CD8⁺ T cells is demonstrated further by the reduction in the liver-homing potential of these cells when mice are experimentally treated with a low dose of combined aspirin and clopidogrel, 2 antiplatelet drugs widely used in humans [17].

With the use of an ex vivo model-reactive surface simulating the hepatic blood flow conditions, we found that effector CD8⁺ T cells tightly interact with platelets [16]. The concept of physical platelet–T cell interaction leading to hepatic accumulation of the latter cells has been substantiated more recently in vivo by intravital microscopy experiments; in those studies, it was observed that circulating effector CD8⁺ T cells preferentially arrest within liver sinusoids by docking onto platelets that have previously adhered to sinusoidal hyaluronan via CD44 [15]. Platelets have been shown recently to firmly adhere to bacterially infected Kupffer cells (the liver-resident intravascular macrophages), preventing the systemic spread of bacteria [18]. In contrast to the abovementioned study, we found no evidence for a preferential formation of platelet aggregates on Kupffer cells in mouse models of HBV pathogenesis, where Kupffer cells are neither infected with bacteria nor targeted by CD8⁺ T cells. Consistent with these results, Kupffer cell depletion did not affect hepatic effector CD8⁺ T cell accumulation in these mouse models [19].

Experiments that use anti-glycoprotein-Ibα antibodies to deplete >98% circulating platelets (reducing normal platelet counts from ∼10⁶ platelets/μl to <2 × 10⁴ platelets/μl) decreased the hepatic accumulation of effector CD8⁺ T cells by only 2-fold [15]. These results suggest that either the number of circulating platelets greatly exceeds the number required to arrest effector CD8⁺ T cells efficiently in the liver or that some effector CD8⁺ T cell home to the liver in a platelet-independent fashion. The failure to decrease hepatic effector CD8⁺ T cells homing additionally by treating platelet-depleted animals with CD44-blocking antibodies seems to support the latter hypothesis [15].

The aforementioned results unequivocally identify the molecules involved in platelet interaction with LSECs; what are the mechanisms supporting effector CD8⁺ T cells docking onto platelet aggregates? Platelets possess numerous surface adhesion molecules, among which only P-selectin has a known ligand (i.e., PSGL-1) on effector CD8⁺ T cells [20]. Passive PSGL-1 neutralization or reconstitution of platelet-depleted mice with P-selectin-deficient platelets, however, did not affect the hepatic accumulation of effector CD8⁺ T cells [15]. Other platelet molecules that had been implicated previously in the cross-talk between platelets and the adaptive immune system, such as CD40 ligand [21] or serotonin [22], were also shown to be dispensable for this process. It is possible that other constitutive or inducible platelet molecules contribute directly or indirectly (via the formation of molecular bridges with ligands expressed by T cells) to effector CD8⁺ T cells docking onto platelet aggregates. An appealing alternative hypothesis involves modification in local blood flow dynamics by intrasinusoidal platelet aggregates so that effector CD8⁺ T cells can slow down and engage platelets and/or liver sinusoidal cells via noncovalent interactions.

The interference with the liver homing potential of effector CD8⁺ T cells might be exploited therapeutically in the context of chronic HBV infection. This pathologic condition is characterized by a dysfunctional CD8⁺ T cell response that fails to eliminate HBV from the liver but maintains continuous cycles of low-level hepatocellular injury, promoting the development of liver fibrosis/cirrhosis and ultimately, hepatocellular carcinoma [23, 24]. Instead of trying to restore the CD8⁺ T cell functionality to the levels observed in patients undergoing self-limited acute infection, one could envision a therapeutic approach that further reduces the capacity of CD8⁺ T cells to induce liver damage and therefore, prevents or delays fibrosis/cirrhosis and hepatocellular carcinoma. In keeping with this and building on the observation that platelets are instrumental to intrahepatic effector CD8⁺ T cell homing, a recent study showed that the antiplatelet drugs aspirin and clopidogrel are able to blunt the hepatic accumulation of pathogenic effector CD8⁺ T cells in mouse models of sustained liver injury, thus preventing or delaying the development of hepatocellular carcinoma and significantly improving overall survival [25]. The identification of specific molecules controlling the capacity of platelets to promote liver CD8⁺ T cell homing might lead to the design of more targeted immune-modulating strategies for the treatment of chronic HBV infections.

HEPATOCELLULAR ANTIGEN RECOGNITION BY INTRAVASCULAR EFFECTOR CD8⁺ T CELLS

Additional intravital microscopy studies revealed that, after the initial platelet-dependent arrest, effector CD8⁺ T cells actively crawl along liver sinusoids at an average speed of ∼10 μm/min [15] (which is 500- to 1000-fold slower than sinusoidal flow [26]). This multidirectional intrasinusoidal crawling is similar to what was observed for hepatic CXCR6⁺ NKT cells [27]. What regulates intrasinusoidal effector CD8⁺ T cells crawling (e.g., chemokine or integrin cues, noradrenergic neurotransmitters from sympathetic nerves [28], or physical structures [29–31]) is currently under investigation in our laboratory. Importantly, this intrasinusoidal crawling behavior represents a form of immune surveillance, as it occurs independently of antigen, and it ceases following hepatocellular antigen recognition [15]. Indeed, hepatocellular recognition of HBV antigens leading to cytokine production (Fig. 1) and hepatocyte killing (Fig. 2) occur in a diapedesis-independent manner, that is, when effector CD8⁺ T cells are still confined to the intravascular space [15]. These processes are mediated via the extension of cellular protrusions through sinusoidal endothelial cell fenestrae by effector CD8⁺ T cells producing contact sites with the hepatocyte membrane that are consistent with the formation of an immunologic synapse [15]. This antigen-probing activity by intravascular effector CD8⁺ T cells would require the formation and retraction of cellular protrusions, events that might seem at odds with average CD8⁺ T cell migration rates of ∼10 μm/min. The high variability among effector CD8⁺ T cell velocities (between 1 and up to 30 μm/min [15]) is compatible with the hypothesis that slower cells are more active in extending and retracting transendothelial protrusions than faster-moving ones. The notion that circulating leukocytes can interact with hepatocytes through endothelial fenestrations has been proposed in previous studies [13, 14, 27]. Our results reveal that this process has functional significance, as reducing sinusoidal fenestrae or creating a collagen barrier between sinusoidal fenestrae and hepatocellular membranes—2 pathologic conditions that typify liver fibrosis and are referred to as sinusoidal defenestration and capillarization, respectively—inhibits hepatocellular antigen recognition by effector CD8⁺ T cells [15]. These results suggest that the process of liver fibrosis might reduce CD8⁺ T cell immune surveillance toward infected or transformed hepatocytes, thus potentially favoring the development and progression of hepatocellular carcinoma.

Figure 1. Effector CD8⁺ T cells recognize hepatocellular antigens and produce antiviral cytokines in a diapedesis-independent manner.

Confocal micrograph showing intravascular HBV-specific effector CD8⁺ T cells (Cor93 T_E; red; two examples are denoted by arrowheads) that produce IFN-γ (yellow) upon recognition of hepatocellular antigen within the liver of HBV replication-competent transgenic mice. Note that nearby intravascular MHC-mismatched, HBV-specific effector CD8⁺ T cells (Env28 T_E; green) do not produce IFN-γ. Sinusoids are in gray. Original scale bar, 10 μm.

Figure 2. Effector CD8⁺ T cells kill HBV-expressing hepatocytes in a diapedesis-independent manner.

Confocal micrograph showing an intravascular, HBV-specific effector CD8⁺ T cell (Cor93 T_E; red) juxtaposed to an apoptotic, HBV-expressing hepatocyte (Caspase 3; brown). Sinusoids are in gray. Original scale bar, 10 μm.

Surprisingly, effector CD8⁺ T cell extravasation from the liver microcirculation follows, rather than precedes, hepatocellular antigen recognition and effector function [15]. What is the fate of extravasated CD8⁺ T cells? One possibility is that they invade hepatocytes, enter endosomal/lysosomal compartments, and are degraded—a process of suicidal emperipolesis that was described recently for naïve CD8⁺ T cells undergoing primary activation in the liver [32]. We are attracted by the hypothesis that effector CD8⁺ T cell extravasation into the liver parenchyma (whether through emperipolesis or other yet-undefined mechanisms) might actually represent a way to limit antigen recognition and therefore, to regulate excessive liver damage caused by T cells. This concept is supported by the notion that hepatocellular MHC-I expression is polarized and localized predominantly to the portion of the basolateral membrane facing the sinusoidal lumen [14] so that MHC-I-peptide complexes might be less accessible to CD8⁺ T cells residing in extravascular spaces.

It should be pointed out that the abovementioned imaging studies uncovered events governing the initial phases of effector CD8⁺ T cells-mediated acute liver pathology in mice. Liver disease in these models, as well as in HBV-infected patients, progresses within the following days with the formation of CD8⁺ T cell clusters around antigen-expressing hepatocytes, CD8⁺ T cell proliferation [33], and the hepatic influx of large numbers of pathogenic antigen-nonspecific inflammatory cells that largely contribute to organ damage [34, 35]. It will be important to extend the imaging studies to these later time points, as well as in settings of chronic hepatitis and hepatocellular carcinoma, to enhance our understanding of the pathogenic versus protective role of virus-specific CD8⁺ T cells during HBV infection.

CONCLUSIVE REMARKS

Recent intravital microscopy studies have unraveled the spatiotemporal constrains, whereby effector CD8⁺ T cells home to the liver, recognize hepatocellular antigens, and perform effector functions. It is now clear that effector CD8⁺ T cells arrest within liver sinusoids by preferentially docking onto platelet aggregates, crawl along liver sinusoids probing hepatocytes for the presence of antigen by extending protrusions through sinusoidal fenestrae, and perform effector function in a diapedesis-independent manner (see a schematic representation in Fig. 3). Several questions remain unanswered, and they include the mechanism whereby effector CD8⁺ T cells dock onto platelet aggregates, what molecules effector CD8⁺ T cells use to crawl intrasinusoidally, and what the fate is of extravascular effector CD8⁺ T cells. The answer to these questions will be important for the rational design of targeted immunotherapeutic approaches for chronic liver infections.

Figure 3. Schematic representation of hepatic effector CD8⁺ T cell behavior leading to hepatocellular antigen recognition.

First, effector CD8⁺ T cells arrest within liver sinusoids by docking onto platelet aggregates; second, they crawl along liver sinusoids, probing hepatocytes for the presence of antigen by extending cellular protrusions through sinusoidal fenestrae; third, they perform effector function in a diapedesis-independent manner (i.e., while still confined to the intravascular space).

Glossary

HBV

hepatitis B virus

LSEC

liver sinusoidal endothelial cell

MAdCAM-1

mucosal vascular addressin cell adhesion molecule 1

PECAM-1

platelet endothelial cell adhesion molecule

PGE1

prostaglandin E1

PSGL-1

P-selectin glycoprotein ligand 1

DISCLOSURES

The authors declare no competing financial interests.