Targeted regulation of a lymphocyte-endothelial-interleukin-6 axis by thermal stress

Abstract

Immune protection from microbial invaders or malignant progression is dependent on the ability of lymphocytes to efficiently traffic across morphologically and biochemically distinct vascular sites throughout the body. Lymphocyte trafficking to target tissues is orchestrated by adhesion molecules and chemokines that stabilize dynamic interactions between circulating lymphocytes and endothelial cells lining blood vessels. While the molecular mechanisms that regulate the efficient migration of lymphocytes across specialized high endothelial venules (HEVs) in secondary lymphoid organs have been extensively characterized, there is a paucity of information available regarding the mechanisms that dictate the rate of lymphocyte entry into tumor tissues. This article summarizes recent evidence that inflammatory cues associated with fever-range thermal stress promote lymphocyte extravasation across HEVs of lymphoid organs through a highly regulated lymphocyte–endothelial–interleukin-6 (IL-6) biological axis. The potential for using thermally-based strategies to improve lymphocyte delivery to the tumor microenvironment during T cell-based immunotherapy will also be discussed.

• Immunotherapy
• heat shock and immune response (i.e., immunotherapy, innate immune responses, inflammation, and cytokines)
• lymphocyte trafficking
• fever
• vascular endothelium
• hyperthermia

Advantages and limitations of T-cell based cancer immunotherapy

Significant progress has been achieved in the development of T cell-based immunotherapeutic modalities for the treatment of cancer [1–6]. These strategies are based on the notion that the adaptive immune system can be harnessed to mount a specific attack on tumor cell targets. Target-specific immune approaches are attractive as they avoid the serious side-effects of other conventional treatments such as chemotherapy and radiation that have relatively non-specific mechanisms of action. CD8 T cell subsets are potent effectors of adaptive immunity by virtue of their ability to initiate cell death following receptor-mediated engagement by antigens expressed on the surface of target cells. T cells mediate tumor destruction by the release of pore-forming lytic enzymes (granzyme, perforin) as well as initiation of apoptotic pathways following ligation of death-receptors on tumor cells (TRAIL receptor, Fas receptor) [2], [7]. T cell cytotoxicity requires direct contact with target cells, thereby limiting damage to bystander cells. An inherent feature of the immune response, unlike conventional cancer therapies, is that it can elicit durable long-term protection from recurring disease.

An additional advantage of T cell-based anti-tumor immunity is that this treatment strategy does not depend on knowledge of tumor location. T cells, through the expression of trafficking molecules, are equipped with the potential to ‘search out and destroy’ widely disseminated tumor cell targets. However, the prevailing paradigm is that the ratio of T cells to tumor targets within tissues must be high in order to cause tumor regression. This assumption is based on in vitro studies where effective killing requires relatively high effector-to-target ratios (e.g., ranging from 20 : 1 to 1 : 1) that are difficult to achieve during immunotherapy. Thus, CD8 T cell-based anti-tumor responses are likely to be most effective when the tumor burden is limited such as in micrometastases or in a setting of minimal residual disease after tumor debulking by surgery, chemotherapy, or radiation.

The generation of CD8 T cell-mediated anti-tumor immunity involves two integrated arms of the immune system. In the first arm, tumor-derived antigens are captured in peripheral organs by resident antigen-presenting cells (i.e., dendritic cells) and carried via the afferent lymphatics to tumor-draining lymph nodes (Figure 1) [2–4], [8], [9]. The capture and delivery of tumor antigens to draining lymph nodes follows the same sequence of events as for microbial-derived antigens. Antigen-loaded dendritic cells localize in close proximity to specialized blood vessels, termed high endothelial venules (HEVs), which are the major portals of entry of blood-borne naïve and central memory CD8 T cells into lymphoid organs [10], [11]. Thus, newly arrived T cells have the opportunity to sample a vast array of antigens presented by dendritic cells upon their entry into lymph nodes.

Figure 1. Tumor immunity depends on the generation of tumor-reactive T cells in the lymph node compartment and subsequent trafficking to the tumor microenvironment. Antigens derived from microbial pathogens or tumor cells in peripheral tissues are carried by dendritic cells to draining lymph nodes via the afferent lymphatic network. Naïve T cells enter lymph nodes across high endothelial venules (HEVs) and proliferate in response to antigen presentation by dendritic cells. Following exit of effector T cells via the efferent lymphatics, these cells circulate throughout the body in search of cognate target cells. The inset illustrates how fever-range thermal stress improves immune surveillance by increasing naïve T cell trafficking across HEVs. Thermal stress modifies the intravascular landscape in HEVs by increasing the density of trafficking molecules (CCL21 and ICAM-1) that support lymphocyte extravasation. Vessels within tumor sites are hypothesized to be non-permissive to effector T lymphocyte trafficking. Thus, effector T cells are physically segregated from tumor cell targets, providing one potential tumor escape mechanism. The role of thermal stress in regulating endothelial adhesion in tumor vessels remains to be determined.

HEVs lined by cuboidal endothelial cells are distinguished from ubiquitous flat-walled vascular beds found throughout the body by their ability to support efficient extravasation of lymphocytes into the underlying parenchymal tissue [10], [12], [13]. HEVs are found exclusively in lymphoid organs (i.e., lymph nodes and Peyer patches) located in close proximity to the skin or the respiratory, gastrointestinal or genitourinary tracts. These organs are strategically positioned to provide the first line of defense against invading pathogens. Lymph nodes are also a major site of immune detection of endogenous antigens derived from transformed tumor cells. It is estimated in mice that lymphocytes continuously cross the total HEVs within a single lymph node at a rate of ∼1.25 × 10⁵ cells per hour [14]. As a result, the lymphocyte content of an individual murine lymph node comprised of ∼10⁶ cells turns over ∼3 times a day. This high throughput screening process is critical for maintaining immune homeostasis and ensures that rare antigen-specific T cells present at a frequency of only 1 in 10⁵ − 10⁶ T cells [15], [16] will encounter dendritic cells carrying cognate antigens from the periphery.

Antigen recognition leads to expansion and differentiation of CD8 T cells with potent lytic potential within the interior of lymph nodes (Figure 1) [2], [4], [8]. Tumors can interfere with the development of tumor-reactive CD8 T cells by multiple mechanisms including preventing dendritic cell maturation, promoting the development of regulatory T cells (T_reg), and limiting the rate of entry of naïve and central memory CD8 T cells across HEVs in tumor-draining lymph nodes [3], [8] [17–20]. These limitations can be overcome clinically by hyper-vaccination with purified tumor antigens or by adoptive T cell transfer regimens in which large numbers of patient-derived tumor-reactive T cells are infused back into cancer patients [1–6]. Despite these advances, clinical trials of T cell-based immunotherapy have been disappointing, thus far, with objective response rates achieved in only a low proportion of patients [1–4], [21].

The second arm of the anti-tumor immune response depends on the delivery of blood-borne CD8 T cells to peripheral sites of primary and metastatic tumor lesions. ‘Armed to kill’ CD8 T cells exit the lymph nodes via the efferent lymphatics and ultimately return to the bloodstream where they patrol the body in search of antigen depots within tissues (Figure 1). CD8 T cell entry at all tissue sites is dependent on a highly ordered sequence of lymphocyte-endothelial adhesion events mediated by matching pairs of ‘Velcro-like’ receptors on T cells and venular walls as well as by interactions between chemokine receptors on T cells with chemoattractant molecules displayed on vascular endothelial cells [10], [11], [22], [23]. While the success of T cell-based cancer immunotherapy clearly depends on the ability of blood-borne T cells to gain access to malignant tissues in order to spearhead tumor-target destruction, remarkably little is known about the molecular mechanisms that control this dynamic process.

A landmark paper by Galon et al. has reinvigorated interest in this topic [24]. This retrospective study in colorectal cancer patients demonstrated that the extent of infiltration in tumor lesions by CD3 T cells as well as by CD8 effector T cell subsets is a better prognostic indicator of patient outcome than conventional TNM (tumor, node, metastasis) staging. Thus, stage III patients with high tumoral infiltration by CD3 and CD8 T cells have a greater probability of disease-free survival and overall survival than stage I patients with low CD3 or CD8 T cell infiltration. Similar observations have been reported in melanoma, colon, lung, and ovarian cancer [23], [25], [26]. Moreover, CD8 T cells have been identified in close proximity to tumor cells undergoing apoptosis in melanoma patients manifesting spontaneous tumor regression [27], [28]. Taken together, these observations suggest that T cell infiltration is causally linked to tumor control and improved patient outcome. Consequently, the mechanisms controlling T cell trafficking to the tumor microenvironment are likely to be a significant determinant of patient responses during T cell-based immunotherapy.

A frequently overlooked consideration is that the extent of CD8 T cell infiltration depicted in a histologic ‘snapshot’ reflects a balance between the rate of entry across a vascular barrier and subsequent events that affect the fate of T cells in situ including survival, proliferation, retention, and egress from tumor tissues [23], [26], [29]. Despite the obvious importance of the initial entry step, surprisingly little is known about the molecular events controlling adhesive interactions between CD8 T cells and the vascular endothelium lining tumor vessel walls. Studies correlating the extent of vascular expression of specific adhesion molecules or chemokines with intratumoral accumulation of T cells over a period of days, while informative, do not provide precise information about entry mechanisms, per se, since many of these same molecules also regulate T cell survival, proliferation, and retention within tissues.

Limited leukocyte interactions with tumor vessels under steady-state conditions

Tumor microvessels are differentiated from vessels of normal organs with respect to their structure, organization, and ability to support leukocyte interactions [30], [31]. Normal vessels are typified by a hierarchical organization in which blood flow is unidirectional through arterioles, capillaries, venules, and subsequently collecting veins. Leukocyte trafficking occurs preferentially across postcapillary venules which respond to acute inflammatory cues because of expression of histamine receptors that trigger vasodilation and increased vascular permeability as well as expressions of adhesion molecules that support leukocyte-endothelial interactions [30], [32–37]. Tumor vessels lack the sequential hierarchy of normal vessels such that arterioles, capillaries, and venules typically cannot be discriminated within tumor tissues [30], [37], [38].

The postcapillary venules of secondary lymphoid organs (lymph nodes, Peyer patches) have additional refinements with respect to structure and function which allow them to support efficient homeostatic trafficking of naïve and central memory lymphocytes. Within the venular tree, successive increases in diameter and branch points provide landmarks for the discrimination of high-order level V through level III venules (with level V being most proximal to capillaries) which comprise the specialized HEVs [14]. Immediately downstream are flat-walled, low-order level II–I venules which are not supportive of lymphocyte interactions under non-inflammatory conditions. Postcapillary HEVs (high-order level V–III venules) of lymphoid organs constitutively express a high density of adhesion molecules and chemokines that support the multistep adhesion cascade required for lymphocyte extravasation that includes: (1) primary tethering and rolling of lymphocytes along vessel walls, (2) activation of chemokine receptors on lymphocytes by chemokines presented on the glycocalyx of vascular endothelium, (3) firm arrest of lymphocytes on vessels, and (4) migration of lymphocytes across the vascular endothelial barrier (Figure 1) [10], [11], [22], [39]. The efficiency of these interactions has been demonstrated by intravital microscopy. For example, in high-order level V venules of inguinal lymph nodes, ∼50% of blood-borne lymphocytes undergo tethering and rolling, and ∼40% of these cells transition to firm arrest [40]. This is in sharp contrast to low-order level I or level II venules where ≤10% of lymphocytes initiate tethering and rolling while ≤5% of these cells undergo firm arrest [40]. Lymphocyte-endothelial interactions in HEVs occur over seconds (tethering/rolling and chemokine activation) and minutes (firm arrest), while it takes approximately 20–30 minutes for lymphocytes to fully complete the extravasation process resulting in entry into the underlying parenchyma [13], [40].

This four-step adhesion cascade is required for trafficking of leukocytes throughout the body although the precise adhesion molecules and chemokines can differ under various inflammatory conditions and at different anatomical sites [36]. Postcapillary venules within non-lymphoid organs normally support only low levels of trafficking of leukocytes under homeostatic conditions but are rapidly transformed to active recruitment sites in response to acute inflammatory signals provided by local cytokines (e.g., tumor necrosis factor (TNF), interleukin-1β (IL-1β), IL-6, interferon-γ (IFN-γ), and lymphotoxin) released by resident mast cells and macrophages in response to infectious agents or inflammatory stimuli [26], [36], [37]. Lymphocyte trafficking can also occur across modified ‘HEV-like’ capillaries and arterioles during chronic inflammation [37], [41].

The highly disordered organization of flat-walled tumor vessels is exemplified by irregular diameters, aberrant branching patterns, abnormal blood flow rates, and anastomotic structures [30], [31], [35], [37], [38], all of which could influence leukocyte trafficking. Moreover, intratumoral vessels in murine tumor models (e.g., RIP-Tag5 pancreatic tumors, EMT6 mammary tumors, CT26 colon tumors, B16 melanoma) poorly express primary adhesion molecules (e.g., E-selectin), chemokines (e.g., CXCL9, CXCL10) or prototypical arrest molecules such as intercellular adhesion molecule-1 (ICAM-1) or vascular adhesion molecule-1 (VCAM-1) [23], [25], [26], [42–48]. These observations in murine tumors correspond with histologic studies showing limited expression of adhesion molecules and chemokines in the intratumoral region of human tumors [23], [25], [26], [42], [45], [49–51].

Intravital microscopy has been used to evaluate the interaction of circulating leukocytes with tumor vessels [26], [31], [52–55]. These studies are primarily based on the analysis of leukocytes labeled in situ with fluorescent tracking dyes (e.g., rhodamine-6-G). Thus, neutrophils are likely to be the major leukocyte population under observation since they comprise the predominant circulating leukocyte pool. These studies provide insight into the intrinsic adhesive properties of tumor microvasculature, although intravital microscopy has yet to be applied for the direct analysis of interactions between CD8 effector T cells and tumor microvessels. In contrast to the efficient interactions of lymphocyte observed in HEVs of lymphoid organs or inflamed vessels in normal organs [10], [11], [14], [25], [35], [36], [40], [56], leukocytes interact poorly with tumor vessels, at a level comparable to or less than what is detected in normal vessels under homeostatic conditions [52–55], [57], [58]. Angiogenic factors present at high concentrations in the tumor microenvironment such as basic fibroblast growth factor (bFGF) and vascular endothelial growth factor (VEGF) C and D have been implicated in repressing adhesion in tumor vessels by in vitro and intravital studies [54], [58–61].

Short-term (1 hour) homing studies further suggest that CD8 T cell entry across tumor vessels is limited. This time point is relevant to the 20–30 minute time frame required for lymphocyte entry into tissues across vascular endothelial barriers. In this regard, tracking of CD8 T cells during the first hour after adoptive transfer into mice bearing MCA-205 tumor nodules in the lung showed that CD8 T cells interacted with capillaries but not venules in normal lung tissue and interacting cells were rarely found in the intratumoral region [62]. Moreover, in murine studies examining the fate of tumor-specific T cells within the first 24 hours after adoptive transfer, it has been shown that CD8 T cells fail to accumulate in large numbers in tumor tissues while these cells are readily detected in other organs such as the spleen [29], [48], [62–64].

While direct experimentation is required to elucidate CD8 T cell extravasation mechanisms in tumor vessels, the overall picture to emerge is that the lymphocyte-endothelial interface in tumor tissues is an important determinant controlling access of CD8 effector T cells to tumor targets. Restrictions imposed by the endothelial barrier provide a plausible explanation for observations in melanoma patients that tumor progression occurs despite high levels of tumor-reactive T cells in the peripheral blood after repeated vaccination or adoptive T cell therapy [3], [21], [65–67]. Tumor growth is similarly unimpeded in a B16 melanoma mouse model even though >90% of the CD8 T cell population expresses a T cell receptor transgene capable of recognizing the gp100 antigen (pmel) on melanoma cells [64], [68].

Identification of the lymphocyte-HEV axis as a thermally sensitive alert system that heightens immune surveillance

There is an accumulating body of evidence indicating that fever-range thermal stress actively promotes egress of blood-borne lymphocytes across HEVs of secondary lymphoid organs where the probability of encountering cognate antigens or pathogens is enhanced (Figure 1) [13], [23], [26], [42], [69]. The notion that thermal stress improves immune surveillance by augmenting lymphocyte trafficking has primarily been studied within the context of the relevance to physiological fever. These findings have provided insight into the beneficial mechanism of action of the thermal component of fever that has generally been relegated to a bystander function during infection and inflammation. Studies of the dynamics of lymphocyte trafficking across HEVs also have broader relevance to cancer immunotherapy since lymph nodes are the major site of priming of CD8 T cells by tumor antigens as well as a common destination of early metastatic spread. Moreover, there is strong speculation that optimal antitumor immunity depends on the ability of a subset of tumor-reactive T cells to gain access to secondary lymphoid organs during adoptive T cell transfer immunotherapy [1], [70–72].

The molecular mechanisms underlying thermal control of lymphocyte trafficking are highly integrated, involving all four steps of the adhesion cascade. In addition, these mechanisms represent independent, but complementary responses in both lymphocytes and high endothelial cells that line HEVs. Temperatures that mimic febrile episodes (38–40°C) were shown to act directly on T and B lymphocytes to enhance the avidity and/or affinity of two lymphocyte homing molecules, L-selectin and α₄β₇ integrin [69], [73–78]. These molecules are obligatory for the initial tethering and rolling of lymphocytes along the luminal surface of HEVs in lymph nodes (Figure 1) or Peyer patches, a requisite first step in the adhesion cascade that ultimately directs lymphocytes into the underlying parenchymal tissue. Fever-range thermal stress augments the binding activity of L-selectin and α₄β₇ integrin without altering the surface density of these molecules on lymphocytes [74–76], [78]. The effects of thermal stress are highly specific such that no change in the binding activity of leukocyte function-associated molecule-1 (LFA-1) for its endothelial ligand, ICAM-1, has been detected [74].

Studies focusing on L-selectin further determined that thermal stress does not alter the intrinsic lectin activity of the N-terminal extracellular domain of L-selectin [74]. Nor does it alter the topographical distribution of L-selectin on microvillous processes where it must reside in order to initiate transient tethering and rolling interactions [74]. Instead, thermal stress causes L-selectin to become stably associated with the detergent-insoluble cytoskeletal matrix prior to engagement by its cognate HEV ligand, peripheral lymph node addressin (PNAd) [75], [78]. The C-terminal 11 amino acids of the cytoplasmic tail of L-selectin are necessary for the thermal response [75]. Notably, this region contains a binding site for the cytoskeletal linker protein α-actinin [79]. On the basis of these studies, it has been proposed that stable associations with the cytoskeletal scaffold induced by fever-range thermal stress enhance L-selectin tensile strength and the ability to withstand hemodynamic shear within HEVs [75], [78].

Fever-range thermal stress has also been found to improve immune surveillance by augmenting the binding function of HEVs [13], [26], [40], [77]. These studies employed the administration of fever-range whole body hyperthermia (WBH) in order to elevate core body temperatures to the range of physiological fever (39–40°C for 3–6 hours) [76], [80–84]. Systemic thermal therapy using fever-range WBH causes a transient decrease in the number of circulating lymphocytes in mice as well as in advanced cancer patients [77], [81], [83]. In mice it was shown that the loss of lymphocytes from the peripheral blood pool is accompanied by increased trafficking of lymphocytes across HEVs of lymphoid organs [13], [23], [40], [77]. This process is illustrated in Figure 2 where increased accumulation of lymphocytes labeled with a fluorescent tracking dye (TRITC, red) is detected within the walls of HEVs (demarked by staining for the HEV-specific adhesion molecule, PNAd; green) as well as in the parenchyma of lymph nodes following WBH-treatment of recipient mice when compared with normothermic (NT) controls [13], [23], [40]. An important aspect of the experimental design in these studies is that the evaluation of the adhesive properties of HEVs is performed after the cessation of heat treatment [13], [40]. This allows for the segregation of the endothelial response to thermal stress from the effects on lymphocyte adhesion (as detailed above) or hemodynamic parameters such as vasodilation, vascular permeability, and blood flow that could influence trafficking.

Figure 2. Fever-range thermal stress increases lymphocyte recruitment across HEVs. Photomicrographs show typical images of TRITC-labeled lymphocytes (red) associated with PNAd-positive HEVs (green) or infiltrated into the stroma of PLN tissue. NT, normothermic control; WBH, whole body hyperthermia. Bar = 50 µm.

Fever-range thermal therapy was further found to improve homeostatic trafficking of lymphocytes across HEVs [40]. In this regard, thermal stress increases the absolute number of blood-borne lymphocytes that migrate across HEVs without preferentially affecting the composition of cells that traffic to peripheral lymph nodes. Thus, naïve and central memory T lymphocytes and B lymphocytes represent the major population of cells trafficking under both normothermic and hyperthermic conditions while neutrophils, monocytes, and activated effector/memory T cells are essentially excluded from crossing HEVs. This is a key difference from the response in inflamed nodes driven by the administration of potent inflammatory cytokines (e.g., TNF) or antigenic challenge which ‘open the HEV gateways’ for cells normally excluded from trafficking at this vascular site including NK cells, monocytes, and effector CD8 T cells [11], [85–91]. Based on these findings, it is tempting to speculate that heightened immune surveillance by fever in draining lymph nodes as well as in distal secondary lymphoid organs would provide a substantial advantage to the host by protecting against widespread dissemination of rapidly multiplying infectious agents.

Several approaches have been taken to pinpoint the adhesion molecules or chemokines that mediate the thermal response in HEVs. Intravital microscopy revealed that fever-range temperatures do not affect the ability of HEVs to support primary tethering and rolling interactions of lymphocytes [40]. Instead, thermal stress causes a profound increase in the transition of lymphocytes from rolling interactions to firm arrest in high-order level III–V postcapillary venules. The remarkable site-specificity of the response is evidenced by findings that adhesion is not altered in low-order level I–II venules, i.e., the vascular segments immediately proximal to the HEVs. Consistent with these observations, thermal stress does not change the expression of trafficking molecules that mediate primary adhesion (tethering and rolling) such as PNAd in lymph node HEVs (Figure 3) or mucosal addressin cell adhesion molecule-1 (MAdCAM-1) in Peyer patch HEVs [13], [26], [40], [63], [77]. In contrast, fever-range thermal stress increases the intravascular display of two gatekeeper homing molecules, ICAM-1 (Figure 3) and the CCL21 chemokine, exclusively in HEVs [40], [63], [69]. Moreover, enhanced endothelial expression of ICAM-1 and CCL21 is causally linked to improved lymphocyte trafficking across HEVs. While elevated ICAM-1 expression is generally considered a signature of inflammation, the findings for CCL21 are particularly unexpected since this chemokine was previously considered to be a homeostatic chemokine that is not susceptible to inducible expression on blood vessel walls.

Figure 3. Fever-range thermal stress increases ICAM-1 display on HEVs of peripheral lymph nodes. Intravascular staining of ICAM-1 (red) and PNAd counterstaining (green). Bar = 50 µm.

The site-specific nature of the fever-range thermal response in vascular endothelium would be predicted to be a benefit during physiologic fever since it would maintain the focus of the immune response in lymphoid organs, where lymphocytes have the greatest opportunity to encounter antigens. The failure of flat-walled non-HEVs to augment adhesion in response to fever-range thermal stress in vivo [40] is paralleled by results of in vitro studies which examined the effects of fever-range hyperthermia (39–40°C for 6 hours) on endothelial expression of adhesion molecules and the ability to support leukocyte adhesion [92], [93]. Fever-range thermal stress has no effect on baseline endothelial expression of adhesion molecules (ICAM-1, E-selectin, VCAM-1) by endothelial cells in vitro, nor does it improve the ability of cultured endothelial cells to support adhesion of lymphocytes or neutrophils [92], [93]. Febrile temperatures further do not affect TNF-induced upregulation of ICAM-1, E-selectin, or VCAM-1 [93]. Flat-walled endothelial cells are not entirely refractory to fever-range thermal stress however, since heat-treated endothelial cells release proadhesive factors in vitro that can act in trans to stimulate L-selectin or α4β7 integrin adhesion in lymphocytes [78], [93].

Contradictory results have been observed in endothelial cells in vitro in response to high temperature (≥41°C), short-duration (>60 minutes) heat shock. Differences among the various studies may reflect the types of endothelial cells and the temperature conditions under investigation. Several studies have reported a moderate induction of ICAM-1 in endothelial cells in response to high temperature heat shock alone [93], [94]. Other studies have reported that heat shock does not influence induction of adhesion molecules (ICAM-1, VCAM-1) in cultured endothelial cells either in the absence or presence of inflammatory stimuli such as TNF [93], [95–97]. Finally, several studies have shown that high temperature heat shock actually inhibits cytokine-induced upregulation of ICAM-1, E-selectin, and VCAM-1 by preventing NFκB activation [98], [99]. While heat shock temperatures are not achieved in vivo during physiologic febrile responses, these temperatures could be relevant during local-regional clinical hyperthermia regimens.

Conserved role of IL-6 trans-signaling in mediating thermal control of lymphocyte-endothelial adhesion

An unexpected unifying aspect of the fever-range thermal response is that the same proinflammatory cytokine, IL-6, is required for increased adhesion in both lymphocytes and HEVs. Thus, IL-6 is obligatory for thermally-induced increases in L-selectin binding activity in lymphocytes as well as enhanced intravascular display of ICAM-1 in HEVs [40], [69], [73], [78]. Other proinflammatory cytokines that are more often recognized for the regulation of lymphocyte-endothelial adhesion such as TNF, IL-1β, IFN-γ are not necessary for the response to fever-range thermal stress [40], [78]. Thermal regulation of lymphocyte-HEV adhesion further involves a tightly regulated IL-6 trans-signaling mechanism whereby the membrane-anchored gp130 signal transduction molecule is engaged by a complex comprised of IL-6 and a soluble form of the IL-6 receptor α (sIL-6Rα) binding subunit (Figure 4) [40], [73], [78]. The dual requirement for both IL-6 and sIL-6Rα provides an additional level of control which is typical of potent proinflammatory cytokines, where multiple checks and balances prevent adverse bystander effects during inflammation [69], [100], [101]. The majority of cell types throughout the body are refractory to IL-6 despite ubiquitous expression of membrane-associated gp130 because of their limited expression of membrane IL-6Rα. These cells are rapidly converted to IL-6–responsive cells during acute inflammation as a result of IL-6 trans-signaling due to an increase in local availability of sIL-6Rα [100], [101].

Figure 4. Fever-range thermal stress regulates lymphocyte and HEV adhesion through a common IL-6 trans-signaling mechanism. IL-6 trans-signaling is mediated by ligation of the membrane-anchored gp130 signal transducing subunit by a heterodimeric complex comprised of IL-6 and a soluble form of the IL-6 receptor α (sIL-6Rα) binding subunit. Downstream activation of ERK1/2 mediates thermal enhancement of L-selectin binding activity in lymphocytes. The signal transduction pathways underlying thermal induction of ICAM-1 density in HEVs are currently unknown.

Perhaps one of the most intriguing findings relates to evidence for the evolutionary conservation of the IL-6 trans-signaling mechanism governing lymphocyte adhesion during thermal stress. Elevated temperatures in the range of physiological fever stimulate L-selectin adhesion via an endogenous IL-6 trans-signaling mechanism in leukocytes of species representing four major taxa of jawed vertebrates (mammals, avians, amphibians, teleosts) [39], [73–78]. The evolutionary conservation of a cytokine-dependent, thermally-regulated adhesion response in species that evolved from a node of common ancestry ∼450 million years ago suggests that this process is beneficial since conserved functions point to preservation of organismal integrity. MEK1-ERK1/2, but not p38 MAPK or JNK, has been identified in the IL-6/sIL-6Rα/gp130 signaling pathway upstream of activation of L-selectin adhesion by fever-range thermal stress (Figure 4) [78]. The canonical IL-6 signaling molecule, STAT3, is also activated by a gp130-dependent mechanism during lymphocyte stimulation by fever-range temperature [78] although it remains to be determined if a STAT3 pathway acts cooperatively with MEK1-ERK1/2 to promote L-selectin binding activity. The downstream signal transduction pathway(s) responsible for mediating ICAM-1 induction in HEVs during thermal stress is under investigation.

Several lines of evidence suggest that fever-range thermal regulation of lymphocyte trafficking involves local rather than systemic action of IL-6. In the case of lymphocytes, the IL-6 trans-signaling response necessary for thermal enhancement of L-selectin adhesion in vivo is fully recapitulated during heat treatment of lymphocytes in vitro [69], [73], [78] In this regard, IL-6 and sIL-6Rα present in the conditioned medium of heat-treated leukocytes are capable of acting in trans to stimulate L-selectin adhesion in non-heat treated lymphocytes [78]. Thus, an intact hormonal-neuronal-cytokine axis is not required to induce thermal responses in lymphocytes. Further support for the notion that IL-6/sIL-6Rα acts locally is provided by evidence that thermal stimulation of ICAM-1–dependent firm arrest of lymphocytes is restricted to high-order level III–V venules whereas adhesion in endothelial cells lining low-order venules (levels I–II), immediately adjacent to HEVs, is unaltered [40]. Conversely, systemic administration of a recombinant fusion protein comprised of IL-6 linked to sIL-6Rα indiscriminately induces ICAM-1 on both HEVs and non-HEVs via a trans-signaling mechanism [40].

Perspectives

Insight into the potential protective mechanism of action of fever is provided by emerging evidence that a thermally sensitive lymphocyte–HEV–IL-6 axis positively regulates lymphocyte trafficking (Figure 1). These observations may also have clinical relevance during fever-range thermal therapy in cancer, since lymphocyte trafficking into tumor-draining lymph nodes would be predicted to be improved. Enhanced migration of naïve CD8 T cells into tumor-draining nodes could result in the generation of greater numbers of tumor-reactive CD8 T cells that are licensed to kill tumor cell targets in peripheral organs. Systemic fever-range thermal therapy is well tolerated in cancer patients [81], [84], suggesting that this modality could be used in an adjuvant setting together with T cell-based immunotherapies.

It remains to be determined if the site-restricted specificity of the vascular thermal response reflects intrinsic properties of specialized high endothelial cells that line HEVs or extrinsic influences of the surrounding cytokine-rich microenvironment within lymphoid organs. A related question is whether tumor microvessels will respond to clinical thermal therapy in a manner analogous to HEVs, or whether they will be non-responsive like non-HEVs. There are early indications that the tumor microenvironment could be converted to a site that actively recruits immune cells based on findings that systemic thermal therapy increases the intratumoral accumulation of leukocytes including natural killer cells, granulocytes, and lymphocytes [25], [80], [102]. The tumor microenvironment is known to be rich in inflammatory cytokines and chemokines such as TNF, IL-1β, IL-6 and CCL2 [103–106]. Moreover, studies in murine tumor models indicate that tumor cells are capable of releasing sIL-6Rα into the local tissue environment [101], [107–109]. The inflammatory milieu associated with these factors drives tumor progression by promoting angiogenesis, tumor cell proliferation and survival. Thus, it is feasible that thermal therapy can target the tumor vasculature by exploiting this pre-existing inflammatory environment, thereby opening the gateways for trafficking of cytolytic effector lymphocytes (Figure 1).

Acknowledgements

We thank E.A. Repasky, H. Baumann, and J. Subjeck for valuable discussions during the development of this work and M.M. Appenheimer, J.B. Muhitch, W.C. Wang, and T.D. Vardam for critical review of the manuscript. Supported by the US National Institutes of Health (CA79765 and CA094045) and the Department of Defense (W81XWH-04-1-0354).