There is a growing body of evidence that immunoregulation within the liver varies from all other peripheral organs in that the liver possesses unusual tolerogenic properties while still maintaining the ability to generate an effective immune response. However, the mechanisms that underlie the balance between intrahepatic tolerance and immunity are not fully understood.1 Recent research suggests that the liver's ability to induce tolerance may, in part, be related to activation of naïve T lymphocytes independent of lymphoid tissues.1–5 Normally, naïve T lymphocyte activation occurs in the secondary lymphoid organs. It involves T cell receptor (TCR) interaction with the appropriate major histocompatability complex (MHC) on the surface of an antigen-presenting cell (APC) along with the appropriate co-stimulatory signals. Dendritic cells (DCs) are considered “professional APCs,” but there are other cell types that can also be involved in antigen presentation. In fact, both hepatocytes and liver sinusoidal endothelial cells (LSECs) are able to act as APCs in vitro, activating antigen-specific CD8+ T cells.2, 6, 7 However, such intrahepatic activation does not lead to a fully active T cell phenotype.1 In fact, it has been demonstrated that this intrahepatic activation of T cells leads to apoptosis or differentiation into an anti-inflammatory phenotype (Fig. 1).2, 6, 8–10 The ability of LSECs to act as APCs seems plausible because LSECs readily contact circulating T cells within the sinusoids, as a result of sinusoids having a diameter similar to or even smaller than lymphocytes. The mechanism by which naïve T cells interact with hepatocytes is less readily apparent because an endothelial barrier prevents direct contact. Yet using various transgenic mice expressing cognate antigen only on hepatocytes results in activation of T cells, consistent with the view that hepatocytes are functional APCs in vivo.1, 4
Two potential mechanisms exist that could allow T cells to access hepatocytes. First, LSECs are morphologically distinct endothelial cells that are perforated by fenestrations of ≈100 nm in diameter. The fenestrations account for approximately 10% of the surface of the LSEC, so it is conceivable that these holes could be used by microvilli found on lymphocytes.11–13 Alternatively, the lack of interendothelial cell tight junctions led investigators to propose that gaps between endothelial cells could also function as access points for hepatocytes.13 Moreover, the fact that LSECs lack a basement membrane (common to other vascular beds) would further facilitate direct communication between lymphocytes and hepatocytes. Nevertheless the mechanism of lymphocyte-hepatocyte interactions remains unknown.
In this issue of HEPATOLOGY, Warren et al.13 sought to answer the question of how naïve T cells interact with hepatocytes in vivo given the presence of an endothelial barrier. In a series of electron microscopy studies, Warren et al. demonstrated that a number of intrahepatic lymphocytes within sinusoids were in direct contact with LSECs, and that lymphocyte cytoplasmic extensions pass through the fenestrations of LSECs to interact with hepatocyte microvilli in the space of Disse.13 In addition, Warren et al. demonstrated that there were no gaps between LSECs to allow lymphocyte-hepatocyte interactions to occur.13 The authors also demonstrated that although hepatocyte microvilli do protrude through the fenestrations into the sinusoidal lumen, they do not interact with circulating lymphocytes. An alternative explanation for this observation given by Warren et al. is that these particular lymphocytes could have been displaced during the fixation process.13 This leaves some ambiguity regarding the importance of this potential mechanism (discussed later).
In a second series of experiments, the authors endeavored to determine what mechanism naïve T lymphocytes use to interact with hepatocytes. The authors used the same model as Bertolino and co-workers,1, 4 where TCR transgenic CD8+ T cells specific for H-2Kb are transferred into transgenic mice expressing H-2Kb only on hepatocytes. Their findings demonstrate that naïve CD8+ T cells interact with hepatocytes by protruding microvilli through the fenestrations to contact hepatocytes (identical to but less frequent than intrahepatic lymphocytes). In addition, mice expressing H-2Kb on hepatocytes retain twice as many H-2Kb specific CD8+ T cells compared to control mice, suggesting that lymphocyte-hepatocyte interactions are sufficient to retain T cells within the liver.13
Primary T cell activation requires TCR and LFA-1 on the lymphocyte and MHC/peptide complexes and ICAM-1 on the APCs. Warren et al. demonstrate that LFA-1 is present on lymphocyte microvilli, and ICAM-1 and H-2Kb is expressed on the basolateral side of hepatocytes (polarized toward the sinusoidal lumen), where interactions with lymphocyte microvilli would occur.13 In addition, the authors demonstrate that the interaction of LFA-1 and ICAM-1/ICAM-2 is functionally important by showing that anti-ICAM-1/ICAM-2 blocking antibodies reduced the retention of H-2Kb lymphocytes in the liver of transgenic mice expressing H-2Kb on hepatocytes.13 It would be interesting to elucidate whether the functional blockade produced by the addition of antibodies to ICAM-1/ICAM-2 disrupted lymphocyte interactions with LSECs, thereby indirectly preventing interactions with hepatocytes, or whether they blocked direct contact between lymphocytes and hepatocytes. In other words, it would seem reasonable that an initial adhesive interaction with endothelium would be required before lymphocytes could contact the hepatocyte microvilli. Alternatively, the narrow diameter of sinusoids with low shear forces could allow for physical retention of T cells, giving them sufficient time to extend microvilli into the fenestrae.
What is the role of hepatocytes in the induction of primary activation of naïve T cells? Warren et al., suggest the interaction between naïve T cells and hepatocytes further solidifies the role of hepatocytes as APCs for high-affinity transgenic T cells and, in conjunction with previous studies, that this interaction may play a critical role in the development of tolerance within the liver (Fig. 1).1, 13 One might also question whether the duration, stability and number of lymphocyte-hepatocyte interactions are sufficient to induce T cell activation. Previous research in vitro suggests that T cell activation requires a long-lived immunological synapse,14–17 although it has been demonstrated that serial and short-lived contacts with DCs are sufficient to activate CD4+ cells in vitro.18, 19 In accordance with the second set of observations, Warren et al. propose that, in the presence of the appropriate antigen, repeated contacts between peptide/MHC complexes allow for a cumulative activation signal in the T cells.13 It seems plausible that this mechanism could lead to reduced effector function, because the strength and duration of antigenic and costimulatory signals can affect the differentiation process and the characteristics of the effector cells that develop.20, 21
There is always some concern about trying to interpret dynamic biological processes from “snap-shot” electron micrographs. Intravital microscopy approaches allowing visualization of these dynamic events greatly help to further resolve and support this work. For example, it is interesting that injection of naïve transgenic lymphocytes into mice lacking appropriate antigen leads to slowing down or transient stopping of the T cells, but very little lymphocyte retention or trapping within sinusoids.22 By contrast, injection of naïve TCR transgenic lymphocytes into mice with appropriate cognate antigen results in very profound retention of these cells.22 Clearly, a signal in the sinusoids must be present that allows free-flowing or transiently detained T cells to discriminate between presence and absence of cognate antigen in a very rapid fashion. One could argue that it is unlikely that the naïve T cell could extend microvilli into fenestrations while flowing through, slowing down or even transiently stopping in the sinusoids. This leaves either the sinusoidal endothelium perse or the hepatic microvilli that already protrude through the endothelial fenestrations into the sinusoidal lumen as the available intralumenal surfaces that could be rapidly sampled by a flowing or transiently stopped T cell. Because the cognate antigen was expressed only on hepatocytes, and not on LSECs, it is tempting to implicate the hepatocyte microvilli that protrude into the sinusoidal lumen as key players in this process. Although the authors saw no interactions between T cells and these hepatic microvilli, one wonders whether this type of transient interaction could be captured using electron microscopy approaches.
In fact, it is tempting to hypothesize that the hepatocyte microvilli function as a rapid screen for free-flowing lymphocytes, which could rapidly sample for cognate antigen, and then decide whether to adhere and send out microvilli to establish firm, long-lasting transendothelial lymphocyte-hepatocyte interactions for activation. The observation in the Warren et al. paper that there is a two-fold increase in lymphocyte retention if cognate antigen is present does not disagree with this model. Moreover, the identification of ICAM-1 and MHC on the basolateral membrane of the hepatocytes and presumably on the hepatic microvilli support these type of interactions between the hepatocyte microvilli and the T cells. Of course if these microvilli are selectively devoid of ICAM-1 and/or appropriate MHC, this would discount this possibility. At present, it remains unclear what the function of these hepatocyte microvilli are.
In summary, Warren et al. provide intriguing data describing the first in vivo physical interaction between naïve T cells and parenchymal cells. The notion that naïve T lymphocytes are capable of directly interacting with hepatocytes contradicts the dogma that naïve T cells cannot gain access to peripheral nonlymphoid tissues.23 The authors also further the notion that hepatocytes are capable of antigen presentation and may be involved in the development of intrahepatic tolerance. Like any good study, the work of Warren et al. answers important questions but also raises some new and intriguing areas for further exploration.