All mice were housed at the Central Animal Laboratory at the University of Turku. The animal experiments were approved by the Ethical Committee for Animal Experimentation (under license number 5587/04.10.07/2014) in Finland, and they were performed according to the 3R-principle and in adherence with the Finnish Act on Animal Experimentation (497/2013).

The B6.129P2-Lyve1^{tm1.1(EGFP/cre)Cys}/J (Lyve1-CRE), B6.129S6(FVB)-S1pr1^tm2.1Rlp/J (S1pr1^f/f), B6.Cg-Gt(ROSA)26Sor^{tm14(CAG−tdTomato)Hze}/J (R26-tdTomato), C57BL6-Tg (CAG-EGFP) 10sb/J (beta-actin-eGFP), and B6.Cg-Tg(CAG-DsRed*MST)1Nagy/J (beta-actin-DsRed) mice were purchased from Jackson Laboratory. The Lyve1-CRE mice were bred with S1pr1^f/f or R26-tdTomato mice to generate Lyve1-CRE/S1pr1^f/f mice for functional studies and to generate Lyve1-CRE/R26-tdTomato mice for imaging experiments.

For immunohistochemical analysis of the thymus and lymph nodes (LNs), 6-μm thick frozen sections were fixed with acetone at −20°C and incubated with Alexa647-conjugated ERTR7 (Santa Cruz), Alexa488-conjugated anti-Lyve1 (223322; R&D Systems), and rabbit polyclonal anti-S1PR1 antibodies (H60; Santa Cruz), followed by incubation with Alexa546-conjugated goat anti-rabbit IgG antibody (Life Technologies). The images were captured using a confocal microscope (LSM 780; Zeiss). To detect Qa-2⁺ cells and tdTomato⁺ cells, the thymuses were fixed with 4% paraformaldehyde and then subsequently embedded in 4% low-melting agarose (Lonza) in PBS and cut into 200-μm thick sections using a vibratome (VT1200S; Leica). The sections were incubated overnight at 4°C in PBS containing 0.1% BSA and 1% Triton X-100, and Hoechst 33342 (Life Technologies), Alexa647-conjugated anti-Qa-2 antibody (695H1-9-9; BioLegend), DyLight550-conjugated anti-CD31 antibody (MEC13.3; Novus Biologicals), or APC-conjugated anti-CD31 antibody (MEC13.3; BioLegend). The images were captured using a spinning disk confocal microscope (Intelligent Imaging Innovations) and processed using Imaris software (Bitplane). Quantitation of the cortical and medullary area of the thymus was performed using ImageJ software (National Institute of Health).

Thymocytes and lymphocytes were obtained from the thymus, LNs, and spleen by mechanically dissociating the tissues. Peripheral blood lymphocytes were isolated using Ficoll-Paque PLUS (GE Healthcare). Stromal cells in the thymus and LNs were isolated as described previously, with minor modifications (29). Briefly, the tissues were digested with 0.8 mg/ml dispase (Gibco), 0.2 mg/ml Collagenase P (Sigma), and 0.1 mg/ml DNase (Roche). The resulting single-cell suspensions were enriched for non-hematopoietic stromal cells by using CD45 microbeads (Miltenyi Biotech).

The lymphocytes and stromal-enriched cells were resuspended in PBS containing 1 mM EDTA, 2% fetal calf serum, and 0.1% sodium azide and incubated with the following antibodies: APC-Cy7-conjugated anti-CD4 (GK1.5), PerCP-Cy5.5-conjugated anti-CD8 (53-6.7), Pacific Blue-conjugated anti-B220 (RA3-6B2), BV510-conjugated anti-CD62L (MEL-14), FITC-conjugated anti-CD44 (IM7), PE-Cy7-conjugated anti-CD69 (H1.2F3), Alexa647-conjugated anti-Qa-2 (695H1-9-9), Pacific Blue-conjugated anti-HSA (M1/69), anti-6C10 (SM6C10) (30), V450-conjugated anti-CD45 (30-F11), PE-Cy7-conjugated anti-Gp38 (8.1.1), or APC-conjugated anti-CD31 (MEC13.3), all of which were purchased from either BioLegend or BD Biosciences except anti-6C10. To detect 6C10, lymphocytes were further incubated with phycoerythrin-conjugated goat anti-rat IgM antibody (Southern Biotech).

S1PR1 expressed on thymocytes and stromal cells was detected, as described previously (31). Briefly, single-cell suspensions were incubated with rat anti-S1PR1 monoclonal antibody (40 μg/ml, 713412; R&D Systems) followed by incubation with phycoerythrin-conjugated goat anti-rat IgG antibody (Southern Biotech).

All flow cytometric data were acquired on an LSR Fortessa (BD Biosciences) and analyzed using FlowJo software (FlowJo, LLC).

CD62L⁺ CD4⁺ SP cells were isolated from the thymus using a mouse CD4⁺ T cell isolation kit (StemCell Tech), followed by the positive selection of CD62⁺ cells (Miltenyi Biotech). To label the cells, they were incubated with 5 μM CFSE or 5 μM CMTMR for 30 min at 37°C in RPMI containing 2% FCS. The cells (1 × 10⁶ cells in 10 μl PBS) were then injected into the thymus of recipient mice, as described previously, with some modifications (32). Young female mice (less than 4 weeks old) were used as the recipient mice to control thymus size, and the single-cell suspension was injected directly through the skin into the thoracic cavity immediately above the sternum using a 10-μl Hamilton syringe equipped with a 33-gauge needle. Two days after the injection, the recipient mice were sacrificed, and the donor-derived cells in the thymus, blood, spleen, and LNs were quantified by flow cytometry.

Lymphocytes were collected from LNs of beta-actin-eGFP mice. The cells (1 × 10⁷ cells in 200 μl PBS) were then intravenously injected into the recipient mice. Twelve hours after the injection, the recipient mice were sacrificed, and the donor-derived cells in the inguinal LNs were quantified by flow cytometry.

CD62L⁺CD4⁺ cells were isolated from the thymus, as described above, prior to RNA purification (RNeasy Plus Micro, QIAGEN). cDNA was generated using Superscript VILO cDNA synthesis kit (Life Technologies) and analyzed by a quantitative TaqMan RT-PCR method with the primers and probes designed by Universal ProbeLibarary Assay Design Center (Roche). S1pr1 primers; forward: CGGTGTAGACCCAGAGTCCT, reverse: AGCTTTTCCTTGGCTGGAG, Actb primers; forward: CTAAGGCCAACCGTGAAAAG, reverse: ACCAGAGGCATACAGGGACA. The expression values were normalized using Actb expression as endogenous controls.

Differences between groups were evaluated with Student’s t-test for single comparisons or one-way ANOVA, followed by post hoc Tukey tests for multiple comparisons. The statistical analyses were performed using Prism software (GraphPad). A P-value <0.05 was considered to be statistically significant. Data are presented as the mean ± SD unless otherwise indicated.

While T cell egress from lymphoid tissues has been shown to require S1PR1 expression on T cells (8, 9), a recent study indicates that S1PR1 activation occurs mainly in the vascular structures rather than on the T cells in lymphoid tissues (11). As shown in Figure 1, S1PR1 was readily detected immunohistologically in the vascular structures including the blood vessels and lymphatics of the thymus and LNs (Figure 1A; Figure S1 in Supplementary Material), and FACS studies confirmed the distinct S1PR1 expression in the blood endothelial cells and lymphatic endothelial cells of these tissues (Figures 1B,C). To address the functional significance of the lymphatic endothelial cell-expressed S1PR1 in lymphocyte egress from lymphoid tissues, we developed Lyve1-CRE/S1pr1^f/f mice, in which S1pr1 was deleted selectively in Lyve1 lineage cells due to Cre-mediated excision of the loxP-flanked S1Pr1 allele. The S1pr1^f/f mice were used as controls. As expected, Lyve1-CRE/S1pr1^f/f mice exhibited a strongly downregulated expression of S1PR1 on the lymphatic endothelial cells of the thymus and LNs (Figure 1C); however, their lymphatics were morphologically indistinguishable from those in S1pr1^f/f mice (data not shown). Lyve1-CRE/S1pr1^f/f mice also exhibited a markedly reduced S1PR1 expression on blood vessel endothelial cells (Figure 1C), suggesting that the Lyve1 promoter is active in both lymphatic and blood vessel endothelial cells in the thymus.

Analysis of the lymphoid tissues and peripheral blood of Lyve1-CRE/S1pr1^f/f mice revealed that the number of both CD4⁺ T and CD8⁺ T cells in the LNs, spleen, and peripheral blood was substantially reduced compared with those in S1pr1^f/f control mice (Figure 2A). In contrast, there was no significant difference in the number of B cells between the two groups of mice. Further analysis showed that naïve (CD62L^high and CD44^low) T cells, but not memory-type (CD62L^high and CD44^high) or effector-type (CD62L^low and CD44^high), T cells were markedly reduced in the LN of Lyve1-CRE/S1pr1^f/f mice compared to the control mice (Figure 2B). Analysis of the lymphocytes in the primary lymphoid organs showed that the numbers of CD4⁺ and CD8⁺ SP cells in the thymus of Lyve1-CRE/S1pr1^f/f mice were significantly increased compared to those in S1pr1^f/f mice, whereas the number of B cells in the bone marrow and CD4⁺CD8⁺ double positive (DP) cells in the thymus was comparable in the two types of mice (Figure 2A).

Because S1PR1 expression was disrupted not only in lymphatic endothelial cells but also in blood vessel endothelial cells in Lyve1-CRE/S1pr1^f/f mice (Figure 1B), we investigated whether the marked reduction in naïve T and B cells in the peripheral LNs in these mice was due to an impaired ability of the high endothelial venules to mediate lymphocyte recirculation. To this end, we performed a lymphocyte adoptive transfer assay by intravenously injecting LN lymphocytes from beta-actin-eGFP mice into either Lyve1-CRE/S1pr1^f/f mice or S1pr1^f/f mice. We then examined the migration of the eGFP⁺ cells into the recipients’ LNs. As shown in Figure 2C, adoptively transferred CD4⁺ T cells, CD8⁺ T cells, and B cells were found in comparable numbers in Lyve1-CRE/S1pr1^f/f and S1pr1^f/f LNs, indicating that S1pr1 deletion in the Lyve1-expressing cells did not compromise the ability of high endothelial venules to mediate lymphocyte trafficking from blood to lymph. These results indicated that S1PR1 deletion in Lyve1-expressing cells reduced the number of circulating T and B cells without affecting high endothelial venule-mediated lymphocyte recirculation.

Because naïve T cells were strongly reduced in the LNs of Lyve1-CRE/S1pr1^f/f mice, we further examined the T cell subsets in the thymus of these mice. As described above, although the Lyve1-CRE/S1pr1^f/f and S1pr1^f/f thymuses contained similar numbers of CD4⁺CD8⁺ DP cells, there were modest but reproducible increases in the proportion of CD4⁺ and CD8⁺ SP subsets in the Lyve1-CRE/S1pr1^f/f thymus (Figures 2A and 3A). Among the SP subsets, the CD62L^highCD69^high cells, which barely detectable in the WT thymus, were easily detected in the Lyve1-CRE/S1pr1^f/f thymus, while the relatively immature (CD62L^lowCD69^high) and mature (CD62L^highCD69^low) SP subsets were proportionately reduced in Lyve1-CRE/S1pr1^f/f compared with the S1pr1^f/f thymus (Figures 3B,C). The increased CD62L^highCD69^high CD4⁺ (Figure 3D) and CD8⁺ SP subsets (data not shown) expressed high levels of Qa-2, a cell surface marker that is normally expressed strongly on CD69^low recent thymic emigrants (33) (Figure 3D), although CD62L^highCD69^high subsets in S1pr1^f/f thymus expressed low levels of Qa-2 (Figure 3D), raising the possibility that these cells were defective in thymic egress. Consistent with this possibility, the Lyve1-CRE/S1pr1^f/f mice exhibited markedly enlarged thymic medullas (Figures 3E,F). Thus, our results indicated that S1PR1 ablation in Lyve1-expressing cells leads to a thymic accumulation of SP T cells, possibly due to their impaired ability to exit from the thymus.

We next sought to determine why S1PR1 ablation in Lyve1-expressing cells resulted in the accumulation of semi-mature CD4⁺ and CD8⁺ SP T cell subsets in the thymic medulla. As shown in Figures 3D and 4A, further analysis indicated that these CD62L^highCD69^high cells were phenotypically very similar to the egress-competent cells of the thymus in that they were Qa-2^high6C10⁻ HSA^low (33, 34); however, the S1PR1 expression was uniformly low or absent in these cells (Figure 4B). Quantitative PCR analysis confirmed that S1pr1 expression was strongly attenuated in mature SP thymocytes in Lyve1-CRE/S1pr1^f/f mice (Figure 4C). This was an unexpected finding, because (1) SP thymocytes most closely resembling recent thymic emigrants (HSA^lowCD62L^highQa-2^high) normally express very high levels of S1PR1 (9, 35) and (2) S1PR1 was deleted selectively in the cells that expressed the lymphatic endothelial cell-specific marker, Lyve1, in these mice.

The lack of S1PR1 expression in the SP T cell subsets in the Lyve1-CRE/S1pr1^f/f mice suggested that these cells may belong to the Lyve1 cell lineage, consistent with a recent lineage tracing study, indicating that a small subset of hematopoietic stem cells arise from the Lyve1 lineage cells (23). To determine if the SP T cells belong to the Lyve1 cell lineage, we crossed Lyve1-CRE mice with the Rosa26-floxed stop-tdTomato (R26-tdTomato) mice to generate Lyve1-CRE reporter mice. In these mice, the Lyve1 promoter-induced expression of Cre recombinase in Lyve1-expressing cells results in excision of a floxed stop cassette preceding the tdTomato sequence at the Rosa26 locus, leading to intracellular tdTomato fluorescence. As shown in Figure 5A, flow cytometric analysis of thymocytes from these mice showed strong tdTomato expression in approximately half of the CD4⁺CD8⁺ DP cells and the CD4⁺ and CD8⁺ SP T cells, suggesting that a considerable proportion of these cells are derived from the Lyve1 lineage. Consistent with this possibility, histochemical analysis of thymus revealed that tdTomato was expressed in both the thymic cortex and the medulla, with the medulla exhibiting much stronger fluorescence than the cortex (Figure 5B). Both tdTomato-marked and -unmarked mature thymocytes exhibited comparable expression level of CD62L, CD69, and S1PR1 (Figures 5C,D). The peripheral LNs and spleen also contained large numbers of tdTomato-marked and -unmarked T and B cells (Figure 5E), indicating that Lyve1 and non-Lyve1 lineage cells emigrate from the primary lymphoid tissues to repopulate the peripheral lymphoid tissues. Notably, however, we were unable to detect Lyve1 expression on the cell surface of the tdTomato-marked cells either immunohistochemically (Figure 5B) or by flow cytometry (data not shown). These results collectively indicate that Lyve1 is expressed in a large proportion of lymphocytes during ontogeny and that, in the case of the T cell lineage, both Lyve1 and non-Lyve1 lineage lymphocytes seed and emigrate from the thymus to replenish the recirculating T cell pool in the periphery.

To determine whether the increased SP T cell subsets in the Lyve1-CRE/S1pr1^f/f thymus were egress defective, we performed intrathymic adoptive transfer assays (36, 37) using fluorescently labeled CD62L⁺CD4⁺ SP thymocytes from Lyve1-CRE/S1pr1^f/f and S1pr1^f/f mice. As shown in Figure 6, when a 1:1 mixture of S1pr1^f/f CD62L⁺CD4⁺ SP T cells and Lyve1-CRE/S1pr1^f/f CD62L⁺CD4⁺ SP T cells labeled with CFSE and CMTMR, respectively, were cotransferred into the thymus of wild-type (WT) recipients, the Lyve1-CRE/S1pr1^f/f CD4⁺ SP T cells were retained in the thymus more abundantly than the S1pr1^f/f CD4⁺ SP T cells, and they migrated much less readily to the recipient’s LNs than did the S1pr1^f/f CD4⁺ SP T cells (Figures 6A,B). Similar results were obtained when the fluorescent labels were swapped (data not shown), indicating that the fluorescent labels had no effect on lymphocyte migration. In contrast, when DsRed-expressing WT CD62L⁺CD4⁺ SP T cells were adoptively transferred into the thymus of Lyve1-CRE/S1pr1^f/f or S1pr1^f/f mice, their migration to the LNs of the two different type of mice was comparable (Figure 6C), indicating that S1PR1 expression of the thymic blood or lymphatic vasculature is dispensable for thymocyte egress. These results collectively demonstrated that Lyve1-CRE/S1pr1^f/f CD4⁺ SP T cells exhibit defective thymus egress and that the T cell intrinsic, but not stromal cell-associated S1PR1 plays a critical role in T cell emigration from the thymus. Our results also suggested that subsets of both T and B cells arise from the Lyve1 linage.