CD8 T cells activated in distinct lymphoid organs differentially express adhesion proteins and co-express multiple chemokine receptors

Abstract

Previous work from this laboratory showed that generation of memory CD8 T cells by different immunization routes correlates with control of tumors growing in distinct sites. We hypothesized that effector CD8 T cell expression of adhesion proteins and chemokine receptors would be influenced by activation in different secondary lymphoid organs. In this report, CD8 T cells were activated by immunization with bone marrow derived dendritic cells via intraperitoneal, intravenous, or subcutaneous routes. Three distinct populations of activated CD8 T cells arise in mesenteric, axillary/brachial, and mediastinal lymph nodes and spleen based on differential expression of α4β7 integrin, E-selectin ligand, and α4β1 integrin, respectively. In contrast, three subsets of CD8 T cells defined by differential expression of P-selectin ligand and chemokine receptors were induced irrespective of activation site. The majority of activated CD8 T cells expressed CXCR3, with one subset additionally expressing P-selectin ligand, and another subset additionally expressing CCR3, CCR4, CCR5, CCR6, and CCR9. In the mesenteric lymph node, a fourth subset expressed CCR9 and CXCR3 in the absence of CCR5. Similar homing receptor profiles were induced in the same sites after localized vaccinia immunization. Homing receptor expression on CD8 T cells activated in vitro was distinct, revealing influences of both dendritic cells and the lymphoid microenvironment. Collectively, these results identify previously undescribed populations of activated CD8 T cells based on adhesion protein expression and co-expression of chemokine receptors that arise after activation in distinct secondary lymphoid organs.

INTRODUCTION

The process of T cell activation in secondary lymphoid organs (SLO)³ is accompanied by the up-regulation of several different adhesion proteins and chemokine receptors associated with recruitment to sites of inflammation in peripheral tissues (1). T cell migration to gut associated tissue and peripheral skin tissue occurs via distinct homing receptors. The α4β7 integrin and CCR9 are associated with specific homing of CD4 (2, 3) and CD8 (4, 5) T cells to the gut, and the ligands for P-selectin and E-selectin (PSL and ESL, respectively) have been associated with CD4 (6, 7) and CD8 (8, 9) T cell homing to the skin. CD4 T cell homing to the skin is partially dependent on CCR4 (4, 10, 11), but the role of this receptor in mediating CD8 T cell homing to the skin is unknown. In contrast, the α4β1 integrin and the chemokine receptors CXCR3, CCR3, CCR5, and CCR6 have been associated with homing of CD4 (12, 13) and CD8 (14-20) T cells to multiple inflamed sites. While expression of several chemokine receptors has been associated with entry into peripheral tissues, the extent to which they are expressed coordinately or independently of one another on the same CD4 T cells has not been comprehensively evaluated (21-23), and has not been previously analyzed on CD8 T cells activated in vivo.

Several studies have sought to understand how the site of T cell priming influences their expression of homing receptors. Initial studies analyzing CD4 T cells demonstrated that T cells activated in mucosal and skin sites differentially express α4β7 and PSL (24). Some of these have analyzed the induction of homing receptors on CD8 T cells activated in vitro by resting dendritic cells (DC) isolated from different SLO. DC from skin draining lymph node (LN) were found to induce ESL and PSL, but not α4β7 or CCR9 (11, 25, 26). Conversely, Peyer’s patch or mesenteric (mes) LN DC induced α4β7 and CCR9 with minimal induction of ESL and PSL. Based on chemotaxis assays, expression of CCR2, CCR4, and CCR5 by these two populations of activated CD8 T cells appeared similar (11, 25). A possible limitation is that these studies employed resting DC, whose ability to induce homing receptors might differ from that of mature DC. These in vitro data were complemented by in vivo studies. CD4 T cells activated in mes LN and cutaneous peripheral LN expressed α4β7 and PSL in a mutually exclusive fashion (24). In addition, CD4 T cells activated in both compartments migrated to a CXCR3 ligand, while only cells activated in mes LN migrated to a CCR9 ligand. Similarly, CD8 T cell effectors generated by s.c. immunization with bone marrow derived DC (BMDC) expressed ESL and CCR4 but not α4β7, while those generated by i.p. BMDC immunization expressed α4β7 with minimal ESL and lower levels of CCR4 (27, 28). Blood-borne CD8 T cells generated by i.v. BMDC immunization expressed CCR4 on a subset of cells, and failed to express either ESL or α4β7. Similarly, CD8 T cells activated in mes LN by i.p. injection of vaccinia virus acquired α4β7, but not ESL or PSL, while those activated in inguinal LN by s.c. viral injection expressed ESL and PSL, but not α4β7 (16). These studies collectively support the concept that CD8 T cells activated in gut and skin draining SLO differentially express α4β7 compared to ESL and PSL. However, the adhesion proteins induced in other SLO remain incompletely described. In contrast, the induction of most chemokine receptors on activated CD8 T cells remains poorly described both in vitro and in vivo. Finally, the influence of activating CD8 T cells in the same SLO using different immunogens has not been directly compared.

A previous report from this laboratory demonstrated that BMDC immunization generated tumor-specific CD8 T cell memory that provided protection against the outgrowth of subsequently injected melanoma cells (29). Interestingly, the route of DC immunization dictated the ability to control tumors in different locations: s.c., but not i.v., immunization protected against s.c. tumor challenge, while i.v. immunization conferred significantly greater protection against lung metastases. We subsequently determined that i.v. immunization resulted in BMDC accumulation and activation of CD8 T cells only in the mediastinal (med) LN and spleen (29-32), while s.c. immunization in the scapular fold lead to BMDC accumulation and activation of CD8 T cells only in the axillary/brachial (ax/brach) LN (29, 30, 32). Conversely, i.p. immunization only activated CD8 T cells in the mes and med LN (31, 32). Expression of high levels of α4β7 was confined to those cells activated in mes LN, while α4β1 was expressed on T cells activated in both mes and med LN and spleen (31). However, neither the expression of chemokine receptors and selectin ligands on CD8 T cells activated in the mes LN, med LN and spleen or the expression of homing receptors induced in the ax/brach LN after s.c. immunization were examined.

In the present work, we have conducted a more comprehensive evaluation of the homing receptor profiles induced on CD8 T cells activated in different compartments, and compared both BMDC and vaccinia virus immunogens. Our results extend earlier observations of differential adhesion molecule expression to define a new population induced in med LN and spleen. They also show that, in contrast to adhesion molecules, expression of both PSL and a variety of chemokine receptors is similar in all sites of activation. Furthermore, PSL and multiple chemokine receptors are commonly co-expressed on a subset of activated cells in each of these sites. The latter results reemphasize the importance of chemokine and selectin expression by inflamed peripheral tissue as the principal determinant of T cell homing.

MATERIALS AND METHODS

Mice, viruses, and viral infection

C57BL/6, OT-I RAG1^-/-, and C57BL/6 Thy-1.1 mice were obtained from Charles River, Taconic, and The Jackson Laboratories, respectively. OT-I Thy1.1 mice were first generation crosses of OT-I RAG1^-/- and C57BL/6 Thy-1.1 mice. All animals were maintained in pathogen-free facilities. Recombinant vaccinia virus expressing ovalbumin (vaccinia-ova) was a kind gift from Dr. J. Yewdell (National Institute of Allergy and Infectious Diseases). All protocols were approved by the Institutional Animal Care and Use Committee.

Adoptive transfer of OT-I cells

Single cell suspensions from spleen and pooled inguinal, ax/brach, cervical, mes, and med LN of OT-I Thy1.1 mice were labeled with 5μM CFSE (Molecular Probes) in phosphate buffered saline/0.1% bovine serum albumin for 15min at 37°C prior to injection. One to 2 days before immunization, the indicated number of OT-I cells were injected i.v. into the dorsal tail vein of sex-matched recipients.

Immunogens and immunizations

CD40L-activated BMDC were generated as previously described (33). Prior to injection, BMDC were pulsed with 10μM ova₂₅₇ (corresponding to residues 257-264 of chicken ovalbumin) for 1h at 37°C. Mice were immunized with 10⁵ BMDC or 10⁵ PFU vaccinia-ova in 100-200μl into the dorsal tail vein (i.v.), the peritoneal cavity (i.p.), or the scapular fold (s.c.).

FTY720 treatment

Mice were injected i.p. with 1mg/kg FTY720 (a generous gift of Dr. V. Brinkmann [Novartis Pharma AG]) in 200μl 48h after adoptive transfer of OT-I cells (24h after BMDC immunization) and then every 24h until harvest.

In vitro activation of CD8 T cells by BMDC

OT-I cells were labeled with 1μM CFSE in phosphate buffered saline/0.1% bovine serum albumin for 15min at 37°C. Cells (2.5×10⁴) were plated in 96 well round bottom plates together with 2.5×10⁴ CD40L-activated BMDC and cultured for 5 days before staining for FACS analysis.

Flow cytometric analysis of surface markers

Single cell suspensions were incubated with anti-CD16/32 (eBioscience) or 5% normal rat serum to block Fc receptors. PerCP anti-CD90.1, APC-Alexa-750 anti-CD8α, Alexa-647 anti-α4, PE anti-α4 integrin, PE anti-α4β7, biotin anti-α4β7, PE-Cy7 anti-CD62L, biotin anti-CCR5, and biotin anti-CCR9 were from eBiosciences. PE-Cy7 anti-CD43 (clone 1B11) and alexa-647 anti-CCR3 was from Biolegend. PE anti-CCR5 was from BD Pharmingen. PE anti-CXCR3, PE anti-CCR6, E-selectin fusion protein and P-selectin fusion protein were from R&D Systems. Fusion proteins were detected using a PE anti-human IgG (Jackson Immunoresearch). Anti-mouse CCR4 (Caprologics) was detected using PE anti-goat IgG (Jackson Immunoresearch). Viability assay for in vitro activated cells was performed using LIVE/DEAD stain (Invitrogen). Samples were analyzed on FACSCalibur and FACSCanto instruments (Becton Dickinson) using Flow Jo software (Treestar).

RESULTS

Expression of Adhesion Proteins is Restricted to Distinct SLO

To examine the differentiation of CD8 T cells in different SLO, we adoptively transferred 10³ OT-I cells and immunized the recipients with BMDC that had been pulsed with the ova₂₅₇ peptide. FTY720 was administered daily beginning one day after BMDC immunization as this drug causes retention of effector cells that would otherwise have left the LN (32), providing a complete assessment of cells activated in the priming LN. The induction of adhesion proteins was evaluated four days after BMDC immunization on CD8 T cells that were activated in different SLO and had undergone at least three cell divisions. These cells were CD62L^lo (Fig. 1) suggesting that these cells were fully differentiated (34). Consistent with previously published data (31), a substantial fraction (45.5 ± 26.9% [n=6]) of divided CD8 T cells activated in the mes LN by i.p. immunization with BMDC expressed the α4β7 integrin (Fig. 1). Because the expression of α4β1 is detected by differential staining with anti-α4 and anti-α4β7, it was not possible to ascertain whether α4β7⁺ cells activated in the mes LN were also α4β1. However, a smaller fraction of activated CD8 T cells in the mes LN are α4⁺, α4β7^neg, and thus are α4β1⁺ (Fig. 1). Few cells bound the E-selectin fusion protein (12.2 ± 10.6% [n=8]) or expressed the altered glycoform of CD43 that is recognized by the antibody 1B11 (13.2 ± 9.1% [n=5]) and can act as an ESL (35) (Fig. 1). However, a significant fraction of activated CD8 T cells expressed PSL (Fig. 1 and Table I). Therefore, activation of CD8 T cells in the mes LN results in a dominant population expressing α4β7 and smaller populations expressing α4β1 and PSL.

FIGURE 1.

Adhesion protein expression on CD8 T cells is restricted to distinct sites of activation. 10³ Thy 1.1⁺ OT-I cells were adoptively transferred into Thy 1.2⁺ C57BL/6 mice 1 day prior to immunization with 10⁵ BMDC via the indicated route. 1 day after immunization, mice were treated with FTY720 daily until SLO were harvested 5 days after immunization. CD8⁺Thy1.1⁺ lymphocytes that had undergone 3 or more divisions (as determined by CFSE dilution) were gated and analyzed for the indicated proteins. The priming SLO is indicated above the plots with the route of DC immunization to target that SLO indicated in parenthesis. Adoptive transfer only plots were from mice that received 10⁶ CFSE labeled Thy 1.1⁺ OT-I cells, and then were treated with FTY720 and harvested on a schedule comparable to that of BMDC immunized mice. Data are representative of at least 5 experiments, with exact numbers given in the text.

Table I. Homing receptor induction on CD8 T cells activated in distinct lymphoid organs.

Homing Receptor	Secondary Lymphoid Organ
	Mes LN (i.p.)	Med LN (i.v.)	Spleen (i.v.)	Ax/Brach LN (s.c.)	AT onlya
PSL (n=10)	38 ± 11%	48 ± 14%	44 ± 15%	58 ± 28%	12 ± 5% (n=6)
CXCR3 (n=5)	82 ± 7%	86 ± 11%	67 ± 22%	92 ± 7%	16 ± 7% (n=4)
CCR3 (n=2)	32 ± 1%	32 ± 4%	32 ± 2%	31 ± 9%	n.d.
CCR4 (n=6)	30 ± 14%	36 ± 14%	30 ± 13%	42 ± 26%	8 ± 10% (n=4)
CCR5 (n=21)	23 ± 11%	32 ± 14%	18 ± 8%	26 ± 14%	6 ± 4% (n=7)
CCR6 (n=15)	33 ± 10%	41 ± 15%	35 ± 13%	38 ± 17%	4 ± 3% (n=5)
CCR9 (n=17)	49 ± 23%d	28 ± 19%	34 ± 20%	29 ± 19%	10 ± 5%b, 17 ± 6%c (n=7)

Mice were treated as described in Figure 3. Values in parentheses indicate the number of mice evaluated for each receptor. n.d. not done.

^aAT only is the average of mes, med, and ax/brach LN, and spleen for all homing receptors except CCR9

^baverage of med LN, ax/brach LN, and spleen

^caverage of mes LN.

^dP<0.05 compared to other lymphoid organs

We next evaluated CD8 T cells activated in either the spleen or med LN after i.v. BMDC immunization. Although these SLO drain blood and lung, respectively, most CD8 T cells activated in both compartments expressed α4β1 (87.6 ± 11.7% [n=17] in spleen and 79.7 ± 22.8% [n=17] in med LN). A highly variable subset (21.2 ± 21.4% [n=8]) of cells activated in med LN expressed α4β7 (Fig. 1). However, these cells have a lower mean fluorescence intensity (MFI) of α4β7 than those activated in mes LN and show diminished ability to enter the gut (31). Similar to observations in mes LN, relatively small percentages of cells activated in these two SLO expressed 1B11 (12.8% ± 9.3% [n=8] in spleen and 22.8 ± 16.0% [n=8] in med LN) and ESL (8.3 ± 7.3% [n=12] in spleen and 18.2 ± 10.9% [n=12] in med LN) (Fig. 1), but substantial fractions expressed PSL (Fig. 1 and Table I). While about half of the PSL⁺ activated CD8 T cells co-expressed α4β7, only a third of the α4β7⁺ cells co-expressed PSL (Fig. 2). This suggests that differentiation to express PSL or α4β7 were not mutually exclusive outcomes of T cell activation in mes LN, and reemphasized α4β7 expression as the dominant phenotype. These results identify the dominant adhesion receptor profile for CD8 T cells activated in both spleen and med LN as α4β1⁺ α4β7^lo ESL^lo, with about half of these cells expressing PSL.

FIGURE 2.

The expression of selectin ligands and integrins on CD8 T cells are not mutually exclusive. Mice received 10⁶ OT-I cells and were immunized with 10⁵ BMDCs i.p. or s.c.. The mes LN or ax/brach LN was harvested 4 days after DC immunization. CD8⁺Thy1.1⁺ lymphocytes that had undergone 3 or more divisions (as determined by CFSE dilution) were gated and analyzed for the indicated proteins.

A final site of CD8 T cell activation examined was the ax/brach LN targeted by s.c. BMDC immunization. Most CD8 T cells activated in these LN expressed 1B11 (73.7 ± 19.0% [n=10]) and ESL (46.9 ± 21.7% [n=14]) (Fig. 1) with about two-thirds of the 1B11⁺ cells expressing ESL, consistent with observations in other model systems (35). CD8 T cells activated in ax/brach LN did not express α4β7 (6.0 ± 4.2% [n=11]) (Fig. 1). As observed in other SLO, a subset of these activated CD8 T cells expressed PSL (Fig. 1 and Table I). While the percentage of cells expressing PSL in ax/brach LN was greater than that in the mes LN (Table 1), this difference was barely below the level of statistical significance (p=0.053). Finally, a subset (31.3 ± 25.4% [n=25]) of T cells activated in ax/brach LN expressed α4β1 (Fig. 1). While somewhat less than half of the ESL⁺ activated CD8 T cells co-expressed α4β1, about 3/4 of the α4β1⁺ cells co-expressed ESL (Fig. 2). This suggests that differentiation to express ESL or α4β1 were not mutually exclusive outcomes of T cell activation in skin-draining LN, and reemphasized ESL expression as the dominant phenotype. CD8 T cells activated in inguinal LN after s.c. immunization in the flank showed similar patterns of expression of all of these adhesion proteins (data not shown). Therefore, CD8 T cells activated in skin draining LN results in a dominant population that expresses ESL and smaller populations co-expressing α4β1 and/or PSL.

Overall, the same major phenotypes were evident in the presence or absence of FTY720 and over a range of adoptively transferred numbers (10³ to 10⁶, data not shown). Collectively, these data establish that CD8 T cell activation in mes LN, med LN/spleen and skin-draining LN leads to three dominant populations that differentially express α4β7, α4β1 and ESL/1B11, respectively.

Similar chemokine receptor profiles are induced on CD8 T cells irrespective of priming site

Based on the differential expression of adhesion proteins on CD8 T cells activated in different SLO, we wanted to determine whether there was a similar selectivity in the induction of chemokine receptors. However, we observed that most chemokine receptors were expressed similarly regardless of activation sight. The majority of activated cells in all SLO expressed CXCR3 (Fig. 3 and Table 1). In contrast, CCR3, CCR4, CCR5, and CCR6 were all induced on much smaller subsets, although comparably in each SLO (Fig. 3, 4 and Table I). We failed to detect any induction of either CXCR4 or CXCR6 on activated CD8 T cells in all SLO (data not shown). Interestingly, we also observed CCR9 induction in all SLO, although the percentage of cells expressing CCR9 was clearly elevated in mes LN (Table I). These expression patterns were consistent over a range of adoptively transferred OT-I cells (data not shown). These results indicate that, in contrast to the adhesion proteins, induction of these chemokine receptors was not restricted to particular SLO.

FIGURE 3.

Activated CD8 T cells express the same profile of chemokine receptors irrespective of the site of activation. Mice received 10⁶ OT-I cells and immunized as described in Figure 1. Lymphocytes were gated for CD8⁺Thy1.1⁺ cells and analyzed for CFSE and the indicated chemokine receptor.

FIGURE 4.

Co-expression of chemokine receptors and adhesion proteins on CD8 T cells activated in distinct SLO. Mice were treated as described in Figure 3. CD8⁺Thy1.1⁺ lymphocytes that had undergone 3 or more divisions (as determined by CFSE dilution) were gated and analyzed for the indicated proteins.

Definition of CD8 T cell subsets based on co-expression of chemokine receptors and PSL

The expression of CCR3, CCR4, CCR5, and CCR6 of activated CD8 T cells was consistent with the possibility that all of these receptors were co-expressed. Therefore, we evaluated the extent to which cells expressing CCR5 also expressed CCR3, CCR4, and CCR6. Regardless of the LN examined, all activated CD8 T cells that expressed CCR5 also expressed these other chemokine receptors (Fig. 4). The co-expression of these chemokine receptors suggested that these cells might represent more differentiated effector cells. Consistent with this possibility, activated CD8 T cells expressing CCR5 also expressed PSL (Fig. 4). However, a proportion of PSL⁺ cells did not express CCR5. These results establish the existence of three subpopulations of activated cells in each SLO based on chemokine receptor expression: one expressing only CXCR3 and one additionally co-expressing PSL and one additionally co-expressing CCR3, CCR4, CCR5, and CCR6.

Based on the reported involvement of CCR9 in the selective migration of activated CD8 T cells to the small intestine (36), we were interested in understanding whether it was co-expressed with other chemokine receptors. In all LN examined, we found that the chemokine co-expressing population defined using CCR5 also co-expressed CCR9 (Fig. 4). However, in mes LN, we found an additional population that was CCR9⁺, but CCR5^neg. Furthermore, these cells expressed CXCR3. The presence of this additional population accounts for the elevated level of CCR9⁺ cells in mes LN compared to other SLO (Table I).

Induction of homing receptor expression by BMDC in vitro

To examine whether the induction of multiple chemokine receptors in all SLO might be a consequence of the intrinsic programming capability of BMDC, we analyzed their expression on CD8 T cells activated with BMDC in vitro. A small fraction of T cells cultured in the absence of BMDC expressed low levels of CXCR3, CCR4, CCR5, CCR6, and CCR9 (Fig. 5 and data not shown). The percentages of cells expressing CXCR3 and CCR9 were modestly increased in the presence of BMDC. In contrast, neither CCR4, CCR5, nor CCR6 was induced on CD8 T cells activated in vitro by BMDC. These results suggest that the in vivo expression of all these chemokine receptors on CD8 T cells is not a result of the intrinsic programming capablility of BMDC, but depends on additional factors in the lymphoid microenvironment.

FIGURE 5.

Differential expression of adhesion proteins and chemokine receptors by CD8 T cells activated in vitro by BMDC. CFSE-labeled OT-I cells (2.5×10⁴) were incubated either alone or with 2.5×10⁴ SIINFEKL pulsed BMDC for 4 days in vitro. Live lymphocytes as determined by LIVE/DEAD cell viability stain were gated for CD8⁺Thy1.1⁺ cells and analyzed for CFSE dilution and the indicated protein.

We were also interested in how the expression patterns of adhesion proteins in different SLO in vivo were related to the intrinsic programming capability of BMDC. Consistent with previous results (26, 31), α4 was induced on CD8 T cells activated by BMDC in vitro, while α4β7 was not (Fig. 5). We expanded these results to include 1B11, which was expressed at a high level on most divided cells. Despite this, none of the activated cells expressed ESL (Fig. 5). PSL was induced on T cells cultured in vitro in the absence of DC stimulation making an examination of the induction of PSL unfeasible (data not shown). The lack of ESL and α4β7 induction in vitro contrasts with their selective induction in vivo, suggesting that induction is due to microenvironmental factors in specific LN that directly or indirectly augment the programming capabilities of BMDC. Conversely, the absence of α4 expression on most CD8 T cells activated in the ax/brach LN, and the lack of 1B11 expression on cells activated in the mes LN, med LN, and spleen, demonstrates the existence of microenvironmental factors that suppress the intrinsic ability of BMDC to induce these proteins on activated CD8 T cells.

Homing receptor profiles induced by vaccinia immunization

The in vitro results above demonstrate that the expression of homing receptors in vivo after BMDC immunization reflected a combination of the programming capabilities of BMDC together with microenvironmental influences in the SLO. To compare the programming capabilities of BMDC with those of endogenous DC, we immunized OT-I adoptively transferred mice via i.v., i.p. or s.c. routes with vaccinia-ova. As with BMDC immunization, mice were also treated with FTY720 and analyzed 4 days later. As was seen in mice immunized with BMDC, substantial fractions of cells activated in the mes LN, med LN, and ax/brach LN by vaccinia-ova expressed α4β7, α4β1, and ESL, respectively (Fig. 6). Similarly, in each SLO a subset of activated CD8 T cells expressed PSL (Fig. 6). CCR4, CCR5, CCR6, and CCR9 were expressed on subsets of cells activated in all SLO that were comparable in size to those elicited by BMDC, with the subset of cells expressing CCR5 entirely included within the PSL⁺ subset (Fig. 6 and data not shown). However, in contrast to CD8 T cells activated by BMDC, a higher percentage of those activated by vaccinia-ova in ax/brach LN expressed α4β1 (Fig. 6). In addition, a population of cells activated in mes LN and med LN expressed 1B11, although few of these also expressed ESL compared to cells activated in the ax/brach LN. These results demonstrate that the patterns of homing receptors induced by BMDC and by endogenous DCs targeted by a virus are very similar.

FIGURE 6.

Vaccinia infection induces similar homing receptor profiles based on site of CD8 T cell activation. 10⁶ Thy 1.1⁺ CFSE labeled OT-I cells were adoptively transferred into Thy 1.2⁺ C57BL/6 mice 1 day prior to immunization with Vaccinia-OVA via the indicated route. Lymphocytes from the indicated SLO were gated for CD8⁺Thy1.1⁺ cells and analyzed for CFSE and the indicated proteins or analyzed cells that had undergone 3 or more divisions for the indicated proteins.

DISCUSSION

The migration of activated CD8 T cells requires the expression of adhesion proteins and chemokine receptors to enable entry into peripheral tissues and tumors. In the present study, we examined the expression of homing receptors on CD8 T cells after in vivo activation in different lymphoid compartments by two different immunogens. Our experiments establish that activation of CD8 T cells by BMDC in mes LN, med LN/spleen, and skin draining LN results in three dominant populations based on differential expression of α4β7, α4β1 and ESL. In contrast, regardless of SLO most activated CD8 T cells expressed CXCR3, with a subset additionally expressing PSL, and a subset additionally expressing CCR3, CCR4, CCR5, CCR6, and CCR9. The expression of adhesion proteins and chemokine receptors on activated CD8 T cells following vaccinia immunization was similar to that observed after BMDC immunization, indicating similar site-dependent and site-independent homing receptor profiles are induced by both endogenous and exogenous DC. Collectively, these results identify previously undescribed populations of activated CD8 T cells based on adhesion protein expression and co-expression of chemokine receptors that arise after activation in distinct SLO.

Our observation that high levels of α4β7 are selectively expressed on CD8 T cells activated in mes LN is consistent with previous in vivo data analyzing CD8 effectors generated by i.p. immunization with BMDC (27, 28) as well as the phenotype of T cells activated in vitro by DC from Peyer’s patches (25) or mes LN (4). Similarly, our observation that ESL is selectively expressed on CD8 T cells activated in skin draining LN is consistent with the phenotype of CD8 T cells generated by s.c. immunization with BMDC (27, 28) and of CD8 T cells activated in vitro by DC from skin draining LN (25-27). However, we also identified a distinct differentiation pathway of CD8 T cells activated in the med LN and spleen after i.v. BMDC or vaccinia immunization, which results in uniform expression of α4β1, but limited expression of α4β7 or ESL. Previous work demonstrated cells arising from i.v. BMDC immunization failed to accumulate in the gut (31), were unable to mediate skin contact hypersensitivity (27), and could not control s.c. tumor outgrowth (29). However, i.v. BMDC immunization elicited CD8 T cells that were able to protect mice against outgrowth of tumors in their lungs (29). Our results suggest that these functional properties are based on a homing receptor phenotype that is programmed in med LN and spleen, which is distinct from that programmed in either skin draining LN or mes LN.

ESL and α4β7 are not induced on CD8 T cells activated in vitro. Thus, their induction in only certain LN in vivo exemplifies the positive influences the lymphoid microenvironment can exert on homing receptor programming. On the other hand, an apparent suppressive influence of the lymphoid microenvironment on the homing receptor profiles induced by BMDC was evident in the reduced expression of α4β1 on CD8 T cells activated in skin-draining LN, and the reduced expression of 1B11 on cells activated in mes and med LN and spleen. In both cases, this in vivo constraint was relieved when vaccinia was used as an immunogen instead of BMDC. In skin-draining LN, the extent of cell division and the expression of chemokine receptors and ESL on α4β1⁺ and α4β1^neg cells was comparable, suggesting that this was not a reflection of differences in T cell activation efficiency. However, α4β1 was preferentially expressed on cells that were also ESL⁺, suggesting that it represented a modification of the dominant programming in skin-draining LN, rather than a distinct program. In a similar way, although vaccinia immunization induced 1B11 non-selectively in all LN, only those cells activated in skin-draining LN went on to express an ESL, and 1B11 expression in other LN was not associated with a lack of expression of α4 integrins.

The ability of vaccinia to circumvent the constrained induction of both α4β1 and 1B11 by BMDC demonstrates that different immunogens can alter homing receptor expression. This may reflect alterations in the microenvironment of the LN due to virus associated inflammation, or viral induction of a distinct DC maturation state. It remains controversial whether cells within the SLO itself (28) or the tissue draining the SLO (37) are responsible for homing receptor programming, and how these might be affected by viral infection. Previous studies have identified retinoic acid (38) and vitamin D3 (39) as factors that act within the LN to positively regulate α4β7 and CCR9 and CCR10 expression on activated T cells. However, while cytokines have also been shown to regulate the expression of homing receptors in vitro, their influence on homing receptor expression during T cell activation in LN remains poorly understood (40).

Our demonstration that PSL is expressed on significant fractions of CD8 T cells activated in every SLO by both BMDC and vaccinia virus is at odds with previous work suggesting it is not induced on T cells activated in vitro by DC from Peyer’s patches (25) or in mes LN by i.p. with vaccinia virus (16) or intralymphatic injection of BMDC (28). This discrepancy may reflect the fact that these studies analyzed T cells at an earlier time point. In addition, we found the percentage of cells expressing PSL in mes LN was generally less than that in ax/brach LN, although this difference was barely below the level of statistical significance. However, our observation is consistent with the idea that PSL is less tissue selective in T cell homing than ESL. In support of this, P-selectin itself is expressed by a much broader array of inflammed peripheral tissues than E-selectin (41). Collectively, our results support the importance of different lymphoid microenvironments in dictating the expression of α4β7 and ESL on activated CD8 T cells, while expression of PSL is not limited to distinct SLO.

An unexpected finding from these studies was the induction of CCR9 on activated CD8 T cells in all SLO in response to multiple immunogens. This contrasts with previous reports demonstrating that CCR9 is expressed by CD8 T cells activated in vivo in mes LN and not peripheral LN (28, 36) and after activation in vitro by DC from gut draining SLO, but not skin draining LN (4, 25, 26). However, our results are in keeping with the demonstration that CD8 T cells activated in vitro by BMDC express CCR9 (37) and that s.c. BMDC immunization elicits CCR9⁺ memory CD8 T cells (26). These descrepancies may reflect differences in the activation state of the DC. In keeping with this possibility, the use of LPS dramatically increases CCR9 expression compared to ovalbumin alone indicating the activation state of the DC can have an impact (42). How the activation state of DC compares between BMDC activated by CD40L, vaccinia infection, and LPS remains unclear as the quality of activation in the peripheral LN in these experiments was not addressed. Nevertheless, our data demonstrate the expression of CCR9 by CD8 T cells activated in multiple sites.

Another surprising finding from this study was the extent to which multiple chemokine receptors were co-expressed on a subset of CD8 T cells activated in vivo. Previous work has associated the independent expression of CXCR3, CCR4, and CXCR5 with functionally distinct CD4 T cell subsets (22), but also documented the coordinate expression of CCR4, CCR6, and CCR10 (21). While previous studies established that CD8 T cells express CCR4, CCR6, CCR9 and CXCR3 after BMDC immunization, the co-expression of these receptors on the same CD8 T cell was not evaluated (11, 27, 43). Our demonstration of the co-expression of 6 different chemokine receptors on a subset of activated CD8 T cells, with minimal evidence of cells expressing single chemokine receptors other than CXCR3 and CCR9, suggests that the control of CD8 T migration will be largely dependent on the selective expression of chemokines by target tissues, rather than selective imprinting. Collectively our results suggest a model in which the site of CD8 T cell activation influences migratory potential through the selective induction of adhesion proteins, while chemokine receptor expression is less selective and potentially dependent on DC activation state.

The impact of adhesion proteins and chemokine receptors on migration of the CD8 T cells to peripheral non-lymphoid tissue remains to be explored. Of particular interest is the necessity of these proteins for the infiltration of CD8 T cells into tumors. An important component of future studies is the role of these homing receptors in CD8 T cell infiltration of tumors in different sites of the body. Understanding the homing receptor profile required for efficient CD8 T cell infiltration will indicate the best route of DC immunization to target particular SLO in relation to the tumor site. Applying this knowledge to the development of vaccines is an important consideration to ensure efficient targeting of the immune response to the desired peripheral tissue compartment.