Circulating Activated Platelets Reconstitute Lymphocyte Homing and Immunity in L-selectin–Deficient Mice

Function blocking anti–murine L-selectin mAb Mel-14, anti–murine LFA-1 (CD11a) mAb TIB 213, and anti– human P-selectin mAb WAPS 12.2 were provided by Dr. E.C. Butcher (Stanford University, Stanford, CA). Anti–human CD41 mAb 7E3 was a gift of Dr. B. Coller (Mt. Sinai Medical Center, New York). FITC-conjugated mAbs to murine TCRα/β, Mac-1 (CD11b), and anti-B220, and PE-conjugated nonblocking anti– human P-selectin (mAb S12) were purchased from PharMingen (San Diego, CA) and Becton-Dickinson (San Jose, CA), respectively. Anti–human CD31 mAb Hec 7 and unlabeled anti–human P-selectin mAb S12 were gifts of Drs. W.A. Muller (The Rockefeller University, New York) and R.P. McEver (University of Oklahoma Health Sciences Center, Oklahoma City, OK), respectively. Nonbinding isotype-matched antibodies were used as negative controls.

Mice were anesthetized and catheterized, and the left subiliac LN was microsurgically prepared as previously described (5). Human platelets were isolated from blood of healthy donors following established procedures (11). For microscopic visualization of endogenous white blood cell (WBC) interactions with vascular endothelium, a bolus injection of saline (10 ml/kg body weight) containing 1 mg/ml of the nuclear dye Rhodamine 6G (Molecular Probes, Inc., Eugene, OR) was given intravenously. Interacting and freely flowing blood-borne WBCs were videotaped during their passage through LN HEV under fluorescent stroboscopic epi-illumination and observation through a ×40 Zeiss objective (Achroplan, numerical aperture 0.75 ∞•, water; Carl Zeiss, Inc., Thornwood, NY). The rolling fraction, defined as the percentage of interacting WBCs in the total number of fluorescent cells passing through a venule during 3 min, was determined by off-line analysis of video recordings. The sticking fraction, defined as the percentage of rolling WBCs that became stationary for a minimum of 30 s, was determined during the same period. After an initial observation period to establish baseline interactions, some animals were injected with 3 boluses (200 μl each at 2 min intervals) of resting or TRAP (thrombin receptor activating peptide)-activated human platelets (2 × 10⁸/ml), and WBC behavior was recorded in the same vascular bed. Some aliquots of activated platelets were pretreated with mAb WAPS12.2 (20 μg/ml) before injection. In some experiments, platelets were fluorescently labeled with 2′7′,-bis-(2-carboxyethyl)-5(and 6) carboxyfluorescein (Molecular Probes, Inc.) to distinguish WBCs and injected platelets as previously described (10). To test the role of L-selectin and LFA-1 in HEV, some animals were treated with a 300 μl bolus of saline containing 100 μg of mAb Mel-14 or mAb TIB 213, and WBC adhesion before and after injection of activated platelets was analyzed.

Murine PBMCs were isolated by density gradient centrifugation using lympholyte-M (Accurate Chemical and Science Corp., Westbury, New York), washed in PBS with 1% heat-inactivated human serum albumin (HSA) and 5 mM EDTA, pH 7.4 (PBE) and subsequently resuspended in HBSS containing 10 mM Hepes, 2 mM CaCl₂, and 0.2% HSA, pH 7.4 (assay media).

Purified PBMCs (2 × 10⁶/ml) in PBE were preincubated at 4°C for 30 min with FITC-conjugated antibodies that recognize murine Mac-1, TCR-α/β, or B220. TRAP-activated human platelets (10⁸/ml) were labeled with nonblocking PE-conjugated mAb S12 to human P-selectin. Stained cells were washed twice and resuspended in assay media. 50 μl each of platelets and PBMCs were subsequently coincubated in microtiter wells at 4°C for 1 h. In some experiments, labeled platelets were pretreated for 30 min with 20 μg/ml P-selectin mAb WAPS 12.2 (which blocks the lectin domain) or nonblocking control mAb S12 (which recognizes the SCR domain) and subsequently mixed with the PBMC suspension without washing. The two mAbs do not interfere with each other's binding to P-selectin (Diacovo, T.Y., and U.H. von Andrian, unpublished data). Platelet adhesion to PBMCs was quantitated by measuring red and green fluorescence on a FACScan^® flow cytometer (Becton Dickinson, San Jose, CA). Gating on events identified by light scatter as lymphocytes or monocytes was performed to exclude single platelets. For each experiment, a sample containing either labeled platelets mixed with unlabeled PBMCs or labeled PBMCs alone was used to set threshold markers on the red and green fluorescence scale, respectively. MAb- labeled PBMCs showing the platelet marker were interpreted as cells that bound platelets. Results are presented as percentage of platelet-bound PBMCs expressing Mac-1, TCR-α/β, or B220 in the total number of cells in the same subset.

A cutaneous delayed type hypersensitivity immune response was induced by application of 0.5% 2,4-dinitrofluorobenzene (DNFB) in acetone/olive oil (4:1) on the shaved abdomina of mice as previously described (7). 18 h after sensitization, mice were anesthetized, the right external jugular vein was cannulated, and a total volume of 3 ml of TRAP-activated human platelets (10⁸/ml) were administered intravenously over 6 h as repeated small boluses. Activation status of platelets was confirmed by flow cytometry of P-selectin expression. For antibody inhibition studies, platelets and mice were pretreated with function blocking antibodies (50 μg/ml and 100 μg/mouse, respectively) to human P-selectin (mAb WAPS 12.2) or human CD31 (mAb Hec 7). Mice were rechallenged with 0.25% of DNFB on the left pinna (10 μl) on day 5, and ear swelling responses were measured 24 h after elicitation. Subsequently, the animals were killed and lymphocytes pooled from cervical, subiliac, and axillary nodes were counted on a hemocytometer.

For analysis of antigen-specific T cells in the PLN of mice, nodes were harvested 5 d after DNFB sensitization of mice with and without infusion of platelets and antibodies as described above. Isolated lymphocytes (2 × 10⁴–1.6 × 10⁵ cells) were cultured for 36 h in triplicate in the presence or absence of DNBS, the water soluble form of DNFB (50 μg/ml), and then pulsed with 1 μCi/well of [³H]thymidine for 18 h before harvesting cells as previously described (7).

Where appropriate, data are presented as the arithmetic mean ± 1 SD. A statistical analysis of the ear swelling response was performed using the Kruskal-Wallis test for one-way analysis of variance followed by multiple comparisons on ranks using the Bonferroni correction of P. Results were considered statistically significant when P <0.05.

The behavior of rhodamine 6G–labeled circulating WBCs in PLN–HEV of L-selectin– deficient mice was evaluated using intravital microscopy (5). In agreement with previous studies using a neutralizing antibody to L-selectin (10), the rolling fraction of WBCs in mice lacking this homing receptor (12.1 ± 5.3%, mean ± SD) was considerably reduced in comparison to that in wild-type littermates (72.1 ± 12.1%; Figs. 1 and 2 a). Injection of activated, but not resting, human platelets enhanced leukocyte rolling in the HEV of mutant mice (rolling fraction of 48.9 ± 9.6%), approaching a level similar to that seen in wild-type mice. The pivotal role of platelet P-selectin was confirmed by the ability of mAb WAPS 12.2, directed against human, but not murine, P-selectin, to abolish this effect.

Although selectin binding to PNAd is highly efficient at initiating rolling interactions in flow, in vitro studies indicate that engagement of additional adhesion receptors is required for lymphocyte arrest in HEV (12). Indeed, recent in vivo experiments have shown that L-selectin–dependent lymphocyte rolling is followed by an activation–dependent arrest mediated by LFA-1 (6). We hypothesized that in order for L-selectin–deficient lymphocytes to accumulate in tissues where PNAd is expressed, a similar sequence of events must occur to stabilize dynamic interactions initiated by platelet P-selectin. Infusion of activated platelets in L-selectin–deficient mice not only restored rolling, but >50% of interacting WBCs became stably adherent to HEV, and a large fraction of stuck cells eventually transmigrated (Fig. 2, b and c). Since both WBCs and platelets express several integrin receptors that are capable of strengthening adhesive interactions with vascular endothelial cells, mutant mice or platelets were treated with function-blocking mAb to LFA-1 (TIB 213) or to CD41 (7E3) before platelet injection. MAb TIB 213, but not mAb 7E3, reduced leukocyte sticking by ∼90%; neither had any effect on rolling (Figs. 1 and 2,d and data not shown). Since human platelets do not express β2-integrins and TIB 213 binds only murine, but not human LFA-1 (reference 13 and our unpublished data), these results demonstrate that the sticking of WBC–platelet aggregates was mediated by LFA-1 on WBCs. Therefore, the cellular bridge between lymphocytes and HEV that was initially created by platelets allowed for a critical period of exposure of lymphocytes to chemoattractant signals on HEV that was sufficient to activate and engage LFA-1. Moreover, in intravital microscopy experiments where WBC and platelets were labeled with different fluorescent dyes, we observed that platelets frequently returned to the circulation after the WBCs they had delivered became firmly adherent on the surface of an HEV (Fig. 3). This suggests that platelets may be able to participate repeatedly in the recruitment of leukocytes.

Although in situ labeling of circulating WBCs with rhodamine 6G allows adequate visualization of nucleated cells in HEV, it does not enable us to differentiate between the various leukocyte populations that may participate in antigen-specific immune responses. To identify the specific subsets of murine lymphocytes capable of interacting with human platelets, PBMCs from wild-type or L-selectin–deficient mice were preincubated with fluoresceinated mAbs detecting specific antigens for either α/β T cells or B cells. Cells were then incubated with activated human platelets prelabeled with a nonblocking PE-conjugated P-selectin antibody, and platelet adhesion was quantified using two-color flow cytometry (Fig. 4,a). For reference, platelet binding to monocytes was also analyzed. Approximately half of all B and T cells from both wild-type and L-selectin–deficient mice formed aggregates with activated platelets (Fig. 4 b). Over 98% of monocytes from either genotype also interacted with platelets. The critical role of P-selectin in mediating these interactions was determined by the ability of mAb WAPS 12.2, but not by the nonblocking P-selectin mAb S12, to significantly reduce the percentage of all PBMC subsets bound by platelets. These data suggest that a substantial fraction of circulating B and T lymphocytes can associate with activated platelets in the blood stream.

Our results indicate that lymphocytes can use the platelet delivery mechanism to enter the LN interstitium where they may encounter their cognate antigen. Since this intravascular adhesion cascade does not mandate a contribution by L-selectin, we tested whether activated platelets in the circulation of L-selectin–deficient mice can reconstitute the defect in CHS that has been previously described in these animals (7, 9). To address this issue, the abdominal skin of mutant mice was exposed to DNFB. Topically applied antigens are known to be taken up primarily by dermal Langerhans' cells which then migrate to draining PLN where they present antigen to homed lymphocytes. Our analysis has indicated that a peak in the number of Langerhans' cells is achieved 18–24 h after sensitization (Catalina, M.D., and M.H. Siegelman, unpublished data). To ensure optimal lymphocyte delivery to PLN, mutant mice were repeatedly injected during this 6 h time interval with a total number of 3 × 10⁸ activated platelets.

When ears of sham-treated mutant mice were rechallenged with DNFB 5 d after sensitization, there was no detectable CHS response (Fig. 5 a). Moreover, activated platelet treatment of sensitized wild-type mice did not enhance the CHS response, which, in fact, was slightly lower in this group (P <0.05 versus sensitized wild-type without platelets). In contrast, a dramatic CHS immune reaction was elicited in sensitized mutant animals that had received activated platelets. The extent of ear swelling elicited with DNFB in platelet-treated L-selectin–deficient mice was not different from that elicited in sensitized wild-type mice irrespective of whether or not the latter had received activated platelets. Pretreatment with P-selectin function-blocking mAb WAPS 12.2, but not control mAb Hec 7 to the human platelet antigen CD31, abolished the CHS response in L-selectin–deficient mice. Since WAPS 12.2 neutralizes human, but not murine, P-selectin, its inhibitory effect was most likely due to blocking of platelet-mediated lymphocyte delivery to PLN during the sensitization phase, and not a result of the inhibition of endothelial P-selectin– mediated trafficking of effector cells during the elicitation phase.

To address whether the effect of activated platelets on the CHS response was in the afferent phase of the response, the presence of antigen-specific T cells within draining PLN of sensitized mice was determined by the ability of DNBS to induce in vitro lymphocyte proliferation (Fig. 5,b). LN T cells isolated from sensitized platelet-treated mutant mice proliferated in response to the antigen comparably to those of wild-type animals, whereas T cells from sensitized untreated L-selectin–deficient animals did not proliferate at all. Furthermore, DNFB-reactive T cells resided overwhelmingly in the PLN of both L-selectin–deficient and wild-type mice after contact sensitization, as no detectable proliferation was found in either mesenteric LNs or spleens of these animals (reference 7 and data not shown). The magnitude of the CHS response in platelet-treated mutant animals directly correlated with the number of resident lymphocytes in PLN, suggesting that the number of antigen-specific T cell clones generated during the sensitization phase determined the extent of the cutaneous reaction (Fig. 5 c).

The observation that circulating, activated platelets can promote not only adhesion of lymphocytes to HEV but also reconstitute cell-mediated immunity in L-selectin– deficient mice is important in several ways. It emphasizes the mandatory requirement for lymphocyte localization to PLN in a T cell–dependent immune response to a cutaneous antigen challenge. Moreover, it shows that L-selectin per se is not the exclusive mechanism for lymphocyte homing to PLN, at least in situations that lead to platelet activation in the blood stream. Platelets, through surface P-selectin, can provide a significant compensatory mechanism for initiating lymphocyte–HEV interactions without perturbing subsequent adhesion steps or later events such as antigen recognition or lymphocyte migration and differentiation, which are requisite for activation and clonal expansion of immunocompetent T cells.

Our intravital microscopy observations of PLN–HEV in L-selectin–deficient mice also suggest a second constitutive adhesion pathway that can mediate rolling and perhaps lymphocyte recruitment to HEV. Although impaired in their ability to interact with HEV, a minority of circulating L-selectin–deficient WBCs (∼12%) were still capable of rolling, even when no activated platelets were injected. This low level interaction was also observed in wild-type mice treated with an L-selectin function-blocking antibody (10). Interestingly, treatment of L-selectin–deficient animals with a P-selectin function-blocking antibody reduced these residual lymphocyte–HEV interactions by ∼75% (Diacovo, T.G., and U.H. von Andrian, unpublished data). However, it remains to be determined whether this L-selectin–independent component of rolling is mediated by endogenous P-selectin expressed on platelets, on endothelial cells, or on both.

Recent intravital microscopy studies of PLN–HEV in wild-type mice have shown that stationary adhesion of lymphocytes requires that L-selectin–mediated rolling is followed by a G protein–coupled transmembrane signal that results in the upregulation and subsequent engagement of LFA-1 (6). Consistent with these observations, our present results indicate that lymphocytes that become tethered to HEV by platelets must also activate LFA-1 before they can arrest. Although the precise mechanism(s) that control(s) the physiologic modulation of LFA-1 activity remain(s) to be defined, previous in vitro studies have implicated a role for L-selectin (14–17). In fact, cross-linking of L-selectin on naive, but not memory, T cells with either antibodies or the HEV-derived L-selectin ligand GlyCAM-1, has been reported to promote binding of LFA-1 to one of its counterreceptors, ICAM (intercellular adhesion molecule)-1 (17). Signal transduction through PSGL-1 has been suggested to mediate similar interactions in nonlymphoid leukocytes (18–20). However, platelet-mediated cross-linking of PSGL-1 on T cells did not appear to directly induce integrin activation; platelets bound to purified human peripheral blood T cells failed to convert rolling interactions to firm adherence on PNAd coimmobilized with ICAM-1 in flow assays (Diacovo, T.G., and U.H. von Andrian, unpublished data). Similarly, surface-adherent platelet monolayers supported P-selectin–dependent T cell rolling, but not firm adhesion, which has been shown to occur during platelet interactions with neutrophils (21, 22). Thus, the predominant contribution of platelets to lymphocyte trafficking appears to be the facilitation of lymphocyte margination. Although our data do not exclude more subtle platelet-mediated effects that could potentially affect integrin function on lymphocytes, it appears likely that additional direct communication events between the passively tethered lymphocyte and the vascular wall must occur to initiate the necessary downstream events that are required for lymphocyte accumulation in lymphoid or chronically inflamed tissues.

Our results demonstrate that activated platelets can have a profound effect on the formation of immunologic memory because they can significantly impact on the trafficking pattern of lymphocytes. In particular, activated platelets are likely to affect the migratory behavior of lymphocyte subsets that express little or no L-selectin. This would include populations such as activated and/or memory T cells, which play a vital role in the initiation and maintenance of inflammatory conditions. It is interesting in this context that Th1, but not Th2, cells were recently found to acquire the ability to interact with P-selectin, and this differential ability was shown to be responsible for the preferential recruitment of Th1 cells into inflamed skin and joints (23). However, it appears that P-selectin binding activity is probably not restricted to this specific subset of T cells, as our studies indicate that a much larger proportion (40–50%) of both T and B cells can bind activated platelets, at least in the murine system.

In humans, platelets may not only participate in hemostasis but may similarly modulate immune function by virtue of their ability to interact with both vascular endothelium and circulating leukocytes. This dual function of platelets may contribute to vascular pathology associated with chronic inflammation. For instance, it is conceivable that platelets may play a role in exacerbations of inflammatory bowel disease, a pathological condition associated with an increase in activated platelets in the systemic circulation (24), and endothelial cell expression of PNAd in inflamed tissues (25). Circulating, activated platelets may not only occlude the intestinal microvasculature resulting in tissue infarction, but also alter the homing pattern of T cells that play a pivotal role in the pathogenesis of this and other chronic inflammatory diseases (26–28).

Clinical studies suggest that intravascular platelet activation may be a frequent occurrence in other settings as well. For instance, elevated numbers of P-selectin–expressing platelets have been detected in patients with diabetes mellitus, and in healthy subjects at increased risk for developing insulin-dependent diabetes mellitus (29, 30). In these settings as well as in inflammatory bowel disease, activated platelets may be encountered in the blood stream for extended periods of time. However, our experiments indicate that even a single episode during which large numbers of activated platelets are present in the systemic circulation can affect lymphocyte trafficking. The total number of platelets that were infused into L-selectin–deficient mice during a 6-h period corresponded to ∼40–50% of the circulating platelet pool in each animal. This dose of activated platelets may be clinically relevant since similar levels may be achieved as a result of platelet transfusions to individuals with thrombocytopenia. At the time of transfusion, 15–50% of platelets stored under blood bank conditions have been reported to express P-selectin, depending on the method of isolation and concentration in the transfusate (31, 32). Acute activation of circulating platelets can also occur as a consequence of surgical procedures requiring cardiopulmonary bypass or hemodialysis with bioincompatible filter membranes (33, 34). Thus, our studies simulate an extremely common clinical condition that has not been examined so far with regards to its consequences for a patient's immune system.

We wish to thank Drs. E.C. Butcher, B. Coller, R.P. McEver, and W. Muller for supplying antibodies.

This work was supported by National Institutes of Health grants K08 HL-03361 (to T.G. Diacovo); HL-54936, HL-48675, and HL-56949 (to U.H. von Andrian); and CA-56746 and HL-56746 (to M.H. Siegelman); and a grant from the Arthritis Foundation (to M.H. Siegelman). M.H. Siegelman is an Established Investigator of the American Heart Association.