Propagation and Control of T Cell Responses by Heparan Sulfate-Bound IL-2¹ Free

Early studies of T cell lymphocytes in culture revealed that proliferation and survival of lymphocytes in vitro are supported by soluble factors (1). Throughout the 1960s, various growth factors were reported to enhance proliferation of lymphocytes in short-term cultures (2). In 1976, Morgan et al. (3) made the critical observation that conditioned medium from PHA-stimulated lymphocytes selected for and maintained the growth of human T cells. This growth-promoting activity was ultimately ascribed to a soluble protein, T cell growth factor (4), which today is known as IL-2 (5).

Although IL-2 was identified and isolated based on its mitogenic properties, IL-2 is now known to play a critical role in determining the extent of the proliferative response and in termination of the response via cell death (6). Thus, by priming T cells for activation-induced cell death, IL-2 limits the number of effector T cells in an immune response and regulates the number of memory T cells that remain. Although other cytokines promote proliferation of T cells, IL-2 is the only cytokine that primes cells for activation-induced cell death under physiologically relevant conditions (6, 7, 8). In light of these crucial functions, mechanisms that regulate the availability of IL-2 are of significant import to immune responses.

Given the experiments that led to the discovery of IL-2 and its properties in culture, it has been logical to consider IL-2 only as a soluble cytokine. However, our laboratory (9) and others (10, 11) recently demonstrated that IL-2 binds to the extracellular matrix, suggesting the possibility that bound IL-2 might contribute to immune responsiveness. We tested this concept in model systems in which the functional effects of bound vs free IL-2 could be compared. Our studies, reported here, suggest that contrary to the common concepts concerning IL-2, it is the bound and not the soluble form of IL-2 that drives and controls T cell responses.

Rag2KO mice were purchased from Taconic Farms (Albany, NY). All other mice were bred in the animal facility at the University of Washington (Seattle, WA) under specific pathogen-free conditions. DO11.10 mice, bearing a transgenic TCR for OVA peptide, were originally obtained from Dr. M. Jenkins (University of Minnesota, Minneapolis, MN) and bred with BALB/c IL2^+/− males (The Jackson Laboratory, Bar Harbor, ME) to eventually yield DO11.10/IL-2^−/− mice. TCR-SFE mice on a BALB/c background (expressing a transgenic TCR recognizing hemagglutinin peptide) were a kind gift from D. Lo (Digital Gene Technologies, La Jolla, CA). These mice were crossed with BALB/c IL2^+/− males to eventually obtain TCR-SFE/IL2^−/− mice. This cross was performed to improve survival of IL-2^−/− mice on a BALB/c background, which otherwise die in utero or shortly after birth.

FITC-labeled or biotinylated KJ1-26 (mouse IgG), a mAb recognizing the OVA-specific TCR, was generously provided by Dr. M. Jenkins. PE-labeled anti-CD4 (GK1.5), neutralizing rat anti-mouse IL-2 (S4B6), allophycocyanin-labeled CD25, FITC-labeled CD69, FITC-labeled goat anti-rat IgG, and hamster anti-mouse anti-CD3 (clone 311C) were obtained from BD PharMingen (San Diego, CA). Anti-heparan sulfate Abs (10E4) were purchased from Seikagaku (Tokyo, Japan). Human IL-2 was purchased from R&D Systems (Minneapolis, MN). The T51P mutant of human IL-2 was a kind gift from Dr. T. Ciardelli and Dr. D. Lauffenberger.

Cells isolated from lymph nodes and spleen were stained with FITC-labeled KJ1-26 followed by CD4-PE. Stained cells were then analyzed for the frequency of double-positive cells using a FACScan (BD Biosciences, San Jose, CA).

Five million KJ1-26⁺CD4⁺IL-2^+/+ or KJ1-26⁺CD4⁺IL-2^−/− T cells were injected via tail vein into BALB/c wild-type (WT)GL72³ or BALB/c TCR-SFE IL-2^−/− recipients. Twenty-four hours later, the mice were injected with 2 mg of OVA i.p.. Lymphocytes and splenocytes were harvested on days 1, 4, and 7 postinjection and prepared for flow cytometry as described above. To assess the impact of the T51P mutant of IL-2 on immune responses, Rag2KO mice were injected, via tail vein with 1 μg of murine IL-2, human IL-2, the T51P mutant of IL-2, or PBS. Twenty-four hours later, the mice received 5 × 10⁶ KJ1-26⁺CD4⁺IL-2 ^−/− T cells. Approximately 2 h later, the mice were injected with 2 mg of OVA i.p.. Lymphocytes and splenocytes were harvested on select days as above.

Heparan sulfate glycosaminoglycan was extracted from human spleen (12, 13) and captured onto microtiter plates preincubated with anti-heparan sulfate Abs (2 μg/ml), yielding a final concentration of immobilized heparan sulfate glycosaminoglycan of ∼10 μg/well. Serial dilutions of human IL-2, the T51P mutant of human IL-2, or PBS were incubated overnight at 4°C in the presence or absence of immobilized heparan sulfate glycosaminoglycan. Free IL-2 was subsequently removed by washing. The proliferative response of the IL-2-dependent cell line CTLL-2 (American Type Culture Collection, Manassas, VA), plated at a concentration of 5000 cells/well, was assessed 24 h later based on incorporation of tritiated thymidine.

Tissues were snap frozen in liquid nitrogen and stored at −80°C until use. Five-micrometer tissue sections were then fixed in acetone and stained with rat anti-mouse IL-2 Abs (clone S4B6) and preabsorbed goat anti-rat FITC-labeled secondary Abs.

Splenocytes from WT BALB/c or DO11.10 IL-2^−/−mice, at a concentration of 5 × 10⁶/ml, were incubated with 5 μg/ml neutralizing anti-mouse IL-2 Abs (clone S4B6), 1 μg/ml anti-CD3, and 25 U/ml of human IL-2. Proliferative responses were measured 48 h later via incorporation of tritiated thymidine.

To investigate the potential importance of bound IL-2, we compared immune responses initiated in the presence of matrix-bound IL-2 to responses initiated in the presence of free IL-2. As depicted in Fig. 1, we used a murine model (14) in which T cells expressing a transgenic TCR specific for OVA peptide 323–399 were infused into either WT BALB/c mice, in which heparan sulfate-bound IL-2 is detectable, or IL-2-deficient mice, which lack heparan sulfate-bound IL-2 (9). Presumably, the IL-2 present in WT animals is retained from previous immune responses. Although the half-life of bound IL-2 is not known, IL-2 given i.v. into nude mice is detectable for at least 72 h postinfusion (9). IL-2 is found in the perifollicular regions of the spleen and colocalizes with perlecan, a heparan sulfate proteoglycan (Fig. 2 and Ref.9).

All BALB/c IL-2-deficient mice used in our experimental model were crossed with an irrelevant transgenic TCR to improve their overall health and viability. Immune responses were generated by the systemic injection of OVA and tracked through the use of an anti-idiotypic Ab recognizing the OVA-specific TCR (KJ1-26). As shown in Fig. 3, the response of IL-2 WT, TCR-transgenic T cells in WT recipients, i.e., controls, peaked 4 days after injection of OVA and then declined, approaching baseline by day 7. The response of IL-2-deficient, TCR-transgenic T cells in WT mice, with bound IL-2 available, also peaked on day 4 and reached a similar intensity as controls (see groups KO→WT and WT→WT). In contrast, responses on day 4 of IL-2 WT, TCR-transgenic T cells in IL-2-deficient mice (WT→KO) initiated in the presence of free IL-2 were approximately one-third of the responses generated in the presence of bound IL-2. The peak response of IL-2-deficient T cells in IL-2-deficient animals (KO→KO) was also less than that of either WT or IL-2-deficient T cells in WT animals. This result suggests that bound IL-2 is more effective than free IL-2 in generating immune responses.

In addition to promoting T cell proliferation, IL-2 is thought to contribute to the regulation or extinction of immune responses (6, 7, 8). For example, when OVA is administered as described above, the frequency of OVA-specific T cells typically reaches a maximum 3–4 days postinjection and then decreases dramatically during the ensuing 3 days due primarily to T cell death (14, 15, 16). To explore whether bound IL-2 might contribute to such control of T cell responses, we monitored the frequency and total number of Ag-specific T cells on subsequent days in mice treated as described above. As shown in Fig. 3, the frequency and total number of either WT or IL-2-deficient, TCR-transgenic T cells in WT hosts bearing bound IL-2 decreased precipitously by day 7, consistent with the expected response. In contrast, frequencies and total numbers of IL-2 WT or IL-2-deficient T cells responding in IL-2-deficient hosts continued to increase. These results suggest that in the absence of bound IL-2: 1) both T cell proliferation (days 1–4) and deletion (days 5–7) are impaired or 2) the kinetics of the immune response are significantly modified. Consistent with the former hypothesis, preliminary data from our laboratory show that the number of apoptotic cells is decreased in the spleens of IL-2-deficient vs WT hosts treated as described above. We have also previously shown that bound IL-2 promotes activation-induced cell death in vivo using a Rag2 KO model (9).

The response of IL-2-deficient T cells in WT mice (KO→WT group) was not due to production of IL-2 by host WT T cells, since IL-2-producing, transgenic T cells responded poorly in IL-2-deficient recipients (WT→KO group). The former result is consistent with the observations of Khoruts et al. (17), who found no mRNA for IL-2 in the spleens of naive BALB/c mice given OVA. Conversely, the poor response of WT T cells in KO hosts was unlikely to be due to competition for IL-2 from KO T cells as 1) the expression of the high-affinity receptor (αβγ) for IL-2 is impaired in KO cells in the absence of IL-2 (Ref.18 and Fig. 4), 2) the number of endogenous T cells in naive hosts recognizing and responding to OVA is extremely low (14), and 3) the responses of KO and WT T cells to IL-2 are identical (Fig. 5).

The diminished response of WT T cells in KO hosts may be related to several factors. Although the IL-2 produced by T cells in the WT→KO group will bind heparan sulfate, the WT→KO and KO→WT groups differ in that KO T cells responding in WT hosts are exposed to retained IL-2 at the initiation of the immune response. T cells in the WT hosts may need to undergo several cell divisions before producing IL-2 (19). Bound IL-2 may differ in potency, by virtue of retention alone or due to differences in presentation, from free IL-2. Finally, because only a fraction of the T cells in the WT→KO group are producing IL-2, the amount of IL-2 eventually retained is likely much less than that accumulated in WT hosts in which all T cells can produce IL-2.

To examine in greater detail responses of the transgenic T cells in the presence or absence of retained IL-2, we measured the expression of CD69 and CD25 on the transferred cells. Expression of CD69 was chosen as a reflection of T cell activation, and CD25 was chosen as a reflection of exposure to IL-2. As seen in Fig. 4 a, the frequency of CD69⁺ TCR-transgenic T cells on day 1 postinjection of OVA was highest in the WT→WT, KO→WT, and WT→KO groups, although there was a reproducible tendency for a decreased frequency of CD69⁺-transgenic T cells in the WT→KO group. The frequency of CD69⁺-transgenic T cells in the KO→KO group was diminished in comparison to the other groups. These results suggest that T cell activation is greatest in the presence of host-derived (tissue-bound) IL-2, slightly decreased in the presence of free IL-2 alone (donor derived), and lowest in the absence of IL-2. Interestingly, as time progressed, the frequency of CD69⁺ TCR-transgenic WT or KO T cells responding in WT hosts gradually decreased, whereas the frequency of CD69⁺ TCR-transgenic WT or KO T cells responding in KO hosts gradually increased. These results suggest that the regulation of T cells activated in the absence of host-derived (tissue-bound IL-2), either by death of activated T cells or by down-regulation of activation markers, is impaired.

Because expression of CD25 is largely dependent on IL-2 (18), we assessed expression of CD25 on the transferred T cells as a measure of exposure to IL-2. On day 1 postinjection of OVA, the frequency of CD25⁺ TCR-transgenic T cells was highest in the WT→WT and KO→WT groups and lowest in the KO→KO group (Fig. 4 b). In four separate experiments, frequencies of transgenic T cells in the WT →KO group were intermediate between the KO→WT and KO→KO groups or similar to the KO→KO group. The frequencies of CD25⁺ TCR-transgenic T cells on days 4 and 7 were very low in all groups. These results suggest that transgenic T cells in both the KO→WT and WT →KO groups are exposed to IL-2, but that the exposure of IL-2-deficient T cells to tissue-bound IL-2 in WT hosts is either greater or more efficient than the exposure of IL-2 WT T cells to autocrine or paracrine sources of IL-2 in KO hosts.

To ensure that the disparity in responses of IL-2-deficient and WT T cells did not reflect differential responsiveness to IL-2, we compared the responses of IL-2-deficient and IL-2 WT T cells to human IL-2 in vitro, while blocking the effect of endogenous IL-2 with a neutralizing Ab. As seen in Fig. 5, IL-2-deficient and WT T cells responded similarly to stimulation with human IL-2 and anti-CD3.

The findings in Fig. 3 suggest that tissue deposits of IL-2 drive the immune response to OVA. Because IL-2 has been reported to bind collagen and mannose as well as heparan sulfate (9, 10, 11, 20), we asked whether IL-2 must bind heparan sulfate in vivo to manifest this function. To address this question, we compared T cell responses generated in the presence of WT IL-2, with responses generated using a mutant IL-2 that does not bind heparan sulfate (11). The mutant IL-2, bearing a proline for threonine substitution at aa 51, has equivalent bioactivity to that of WT IL-2 in vitro (21, 22). As seen in Fig. 6,a, used in solution in the absence of heparan sulfate, the WT and mutant IL-2s elicited similar proliferative responses, as expected. When responses to IL-2 bound by heparan sulfate were assessed, only the WT IL-2 elicited a significant response, demonstrating that the T51P mutant of IL-2 does not bind heparan sulfate (Fig. 6 b).

To determine whether IL-2 must be retained by heparan sulfate to drive immune responses in vivo, we “reconstituted” the matrices of IL-2-deficient mice with IL-2 by administering the cytokine systemically, as previously described (9). Twenty-four hours after the injection of IL-2, no free IL-2 was detectable in the blood, consistent with its brief half-life (23). However, IL-2 deposits were detected in the perifollicular regions of the spleen (Fig. 2). To test the impact of heparan sulfate-bound IL-2 in this model, IL-2-deficient TCR-transgenic T cells were transferred into BALB/c Rag-2-deficient mice, which have no detectable perifollicular IL-2 (data not shown). The matrices of these mice were reconstituted as above with either WT or mutant IL-2 24 h before the infusion of T cells, and immune responses were subsequently initiated by the systemic administration of OVA. As seen in Fig. 7, IL-2-deficient TCR-transgenic T cells responded vigorously in mice whose matrices were reconstituted with either human or murine IL-2, but responded poorly in mice whose matrices were reconstituted with mutant IL-2 or PBS. These results suggest that the IL-2 driving T cell responses in vivo is associated with heparan sulfate.

IL-2 is thought to control the activation, proliferation, and elimination of Ag-specific T cells. In this study, we show that the functions of IL-2 in vivo are largely mediated by tissue-bound cytokine and specifically by heparan sulfate-bound IL-2. If bound rather than free IL-2 predominantly regulates T cell responses, then such responses may be subject to variables not generally considered in models of T cell immunity. Given the heterogeneity of heparan sulfate (24, 25), binding of IL-2 in a tissue may be conditioned by the chemical properties of heparan sulfate in that tissue. In fact, we have found substantial differences in the location and amount of IL-2 retained in various organs (Ref.9 and our unpublished observations). Although IL-2 binds readily to heparan sulfate extracted from spleen and kidney, it is possible that IL-2 might bind less or more avidly to heparan sulfate from tumors or virus-infected tissues. Moreover, our results suggest the possibility that organisms (or tumors) might evade immune responses through synthesis of heparan sulfate with modifications (26) that decrease the affinity of heparan sulfate for IL-2. This concept is supported by the finding that undersulfated heparan sulfate from Engelbreth Holmes Swarm sarcoma cells does not bind IL-2 (data not shown). As another variable, the availability of bound IL-2 might be conditioned by glycosylation of the cytokine. Mammalian cells release both glycosylated and unglycosylated forms of IL-2, although little is understood about the regulation of this modification (27). Preliminary data from our laboratory indicate that the binding of glycosylated IL-2 to heparan sulfate is ∼60% that of unglycosylated IL-2 (data not shown).

Our findings may have the most import for understanding the fate of T cells in peripheral tissues. IL-2 is particularly retained in the spleen, liver, and kidney, organs that clear native proteins and foreign Ags from the blood and gut. It is possible that IL-2 bound in these organs provides a first response to organisms that invade the blood and a first line of defense against the development of autoimmunity to the products of normal cells. Because naive T cells must divide several times before producing IL-2 (19), bound IL-2 may serve to quickly initiate immune responses in these tissues. Since substantial numbers of CD4⁺ and CD8⁺ T cells are retained in nonlymphoid organs following immune responses (15, 28), heparan sulfate-bound IL-2 may provide a survival factor for these cells. Finally, because soluble Fas ligand may also be bound by the extracellular matrix (29), colocalization of both IL-2 and Fas ligand by the extracellular matrix may represent a potent means to limit the number of memory T cells remaining in tissues following immune responses or to eliminate T cells inappropriately stimulated by self-Ag.

We thank C. Nath, L. Chen, and N. Wander for technical assistance.