Tug of war at the exit door

Abstract

The lipid sphingosine-1-phosphate has been identified as a key exit signal for lymph nodes. Pham et al show that its action can only be understood in the context of retention signals transduced by CCR7.

The molecular rules that control lymphocyte egress from lymph nodes are still poorly understood compared to the multistep paradigm that controls entry through high endothelial venules. The Cyster lab has been leading the charge in studying the role of the lipid sphingosine-1-phosphate (S1P) in this process exploiting T cells from S1P receptor 1 (S1P1) deficient mice and in this issue of Immunity Pham et al demonstrate that the effects of S1P1 deficiency and pharmacological agents that target S1P1 are dependent upon the level of the homeostatic chemokine receptor CCR7 (Pham et al, this issue). These data suggest a tug of war between S1P exit signals and CCR7 ligand retention signals for the attention of the T cell.

CCR7 is a chemokine receptor expressed on naive and central memory T cells, which is critical for efficient lymphocyte entry into secondary lymphoid tissues through high endothelial venules (Sallusto et al., 1999). CCR7 also contributes to the localization of T cells within T cell zones and make a quantitative contribution to the speed of T cell migration within secondary lymphoid tissues, but other Gi linked GPCR contribute to this process based on experiments with pertussis toxin treated T cells or mice (Okada and Cyster, 2007). Pham et al present findings that are consistent with CCR7 also playing a critical role in T cell retention within lymph nodes, working in opposition to S1P exit signals. They perform side by side comparison of WT and mutant T cell egress in vivo by adoptive transfer and steady state experiments with collection of thoracic duct lymph. T cells over-expressing CCR7 show a relative egress defect, while cells lacking CCR7 egress more rapidly. The most dramatic effect is that CCR7 deficiency partially restores egress in S1P1 deficient T cells and treating the T cells with pertusis toxin is even more effective in restoring egress in the absence of S1P1. Treatment of mice with the drug FTY720 largely phenocopies the S1P1 deficient phenotype by down-regulating S1P1 on the T cells (Mandala et al., 2002). Remarkably, CCR7 deficient or pertussis toxin treated T cells show restoration of egress in FTY720 treated mice. Thus, exit through cortical sinusoids, at least, appears to be a default T cell behavior in the absence of Gαi mediated signaling. How this works physically is not entirely clear since pertussis toxin treated T cells move very slowly in the parenchyma (Okada and Cyster, 2007). Direct imaging of this process by two photon laser scanning microscopy will be interesting to see in the future, but these experiments will need to have better markers for tissue context (stroma, lymphatics) as its clear that not all cortical and paracortical regions are equivalent.

Pham et al interpret these results in terms of a direct competition or a tug of war between CCR7 ligands, CCL19 and CCL21, and S1P that would be played out near the cortical sinusoids. S1P is at high levels in the blood and lymph and it is a chemotactic agent for T cells via S1P1 coupling to Gαi, which would thus also be inhibited by pertussis toxin. Stromal cells in the T cell zones produce CCL19 and CCL21. The superficial paracortex regions immediately around high endothelial venules are likely sites in which both S1P and CCL19/21 gradients may dove-tail. The concept of competing chemotactic signals has been explored previously for neutrophils and B cells responding to antigen. Hierarchies of competing stop and go signals have also been invoked in control of immunological synapse formation during antigen recognition (Dustin, 2004). Pham et al add new data to our understanding of chemotactic control following antigen recognition. They show that following antigen driven proliferation, effector T cells down-regulation CCR7 and increases S1P1 expression, perhaps favoring egress through cortical sinusoids. This process of egress to allow effector dissemination may be the key therapeutic target of FTY720 and related drugs.

The authors demonstrate these effects histologically in cortical sinusoids an exit compartment of lymph nodes that has been largely overlooked, at least since the 1970s. Most egress studies focus on the medullary cords, which are closer to the efferent lymphatics and are the place where T cells accumulate in FTY720 treated animals (Mandala et al., 2002), however the relative percentages of T cell egress mediated by the cortical sinusoids vs the medullary cords is not known. In the lymph node, the subcapsular sinus, cortical sinusoids and the medullary cords are lines by lymphatic endotelium. Since T cells are closer to the cortical sinusoids in the steady state it is possible that much of the homeostatic T cell egress takes place via the cortical sinusoids. The medullary cord parenchyma is occupied by plasma cells and macrophages and thus it is possible that lymphatic endothelium in this region are specialized for antibody transport and may regulate cellular traffic in response to innate immune signals from the macrophages to limit spread of infection.

While this paper adds to the impressive genetic data from the Cyster lab demonstrating that S1P1 on T cells is critical for egress from lymph nodes it still does not explain the discrepancy with data from Cahalan and colleagues in explanted lymph nodes that clearly demonstrates an important gate-keeper role for medullary sinusoid endothelial cells in controlling the egress of wild type T cells in an S1P1 dependent fashion (Wei et al., 2005). The only hypothesis that could bring these two stories together, short of experimental artifacts with knock-out T cells or explanted lymph nodes, is that the gate-keeper roles of cortical and medullary sinusoid edothelial cells are differently programmed as suggested above. Intravital microscopy and genetic studies to dynamically dissect the different role of these two lymphatic compartments will be important to fully understand the mechanics and biology of egress control in lymph nodes.

This is a complex problem in that there are multiple molecular components that act cooperatively in multiple compartments to control egress. The output signal- T cell in efferent lymph- is relatively limited in resolution. While direct imaging may provide some insights, the high level complexity due to cooperative dynamic events that span a wide range of length scales (molecular receptor-ligand interactions to cell populations in lymph nodes) makes it difficult to intuit underlying mechanisms. Such problems in immunology can benefit from synergistic in vitro, in vivo, and in silico studies (Chakraborty et al., 2003). For example, one could begin by asking how gradients in different ligands (e.g., S1P, CCL21) and temporal patterns of receptor expression (e.g., S1P1, CCR7) control egress. Models for random migration of T cells in lymphoid tissues (Beltman et al., 2007) would have to be augmented by incorporating chemotactic motion and combined with models (or measurements) that describe signaling stimulated by receptor-ligand binding. A computational model to examine the veracity of the mechanism suggested by the data reported by Pham et al. (this issue) could be set up as follows. T cells move in the paracortex via a default random process that is biased by chemokine gradients and receptor expression levels. Upon encountering an egress portal (cortical sinusoid), signals due to S1P1-S1P interactions could dominate over signals due to CCR7-CCL21 interactions with a certain probability, leading to egress. Theoretical and computational methods based on Langevin and Master equation approaches (van Kampen, N.G., 1992) could interrogate such a model to understand how the pertinent variables compete to modulate egress versus retention, and whether physiologically relevant values of quantities (such as receptor expression) can affect experimentally observed outcomes. In addition to data in Pham et al (this issue) and other studies on response as a function of ligand amount, such modeling efforts would also require dose-response curves as a function of receptor expression. This type of data (or sophisticated signaling models) is required to determine the probability with which one type of signal (e.g., stimulated by S1P1-S1P interactions) can dominate over competing ones. For the system examined by the Cyster lab, such data could be obtained from in vitro studies. Synergy between computational studies of increased sophistication, plus genetic and imaging experiments, may shed light on mechanistic principles that could then enable an understanding of which factors determine the compliance or resistance of gatekeeper lymphatic endothelium in different compartments.

Figure 1.

Three compartment model for a lymph node in which overplapping gradients of three chemokines CCL19/21 (green), CXCL13 (red) and CXCL12 (blue) define three parenchymal compartments, the deep paracortex, B cell follicles and medullary cords, respectively. These chemokines are captured on stromal cell surfaces, but are also may be detectable as gradients, particularly in boundary regions like those around the cortical sinusoid exit sites. The cortical and medullary sinusoids are bounded by lymphatic endothelial cells, which may have different properties. Pham et al provide evidence for functional S1P gradients that engage in a tug of war with CCL19/21 gradients around cortical sinusoids (purple), which are always permissive for T cell exit, whereas Wei et al (2005) provided evidence for control of exit at medullary sinisoids, in which S1P1 signaling closed exit doors.