Exposure to Systemic Immunosuppressive Ultraviolet Radiation Alters T Cell Recirculation through Sphingosine-1-Phosphate

The UV radiation contained in sunlight is a powerful and broad-spectrum immune suppressant. UV suppression of the local cutaneous immune response is a known risk factor for the development of UV-induced skin cancers (1) and is used to treat a variety of dermatoses (2). Exposure to UV is also associated with protection from autoimmune diseases like multiple sclerosis (MS) (3–5), and systemic suppression of immunity at distant nonskin sites is hypothesized to be one mechanism by which UV affords this protection. In animal models, UV exposure of the skin can delay the onset of experimental autoimmune encephalomyelitis (EAE) and reduce EAE severity and overall disease burden in this mouse model of MS (6–11). This protection from autoimmunity is partly dependent on B cells (6), a reduction of CCL5 in the CNS leading to fewer infiltrating inflammatory cells, and a systemic increase in IL-10 (11). Although the mechanisms underpinning local UV-induced immune suppression are well known, precisely how UV suppresses systemic immunity is unclear. This is important if we are to harness the beneficial effects of sunlight exposure while avoiding the detrimental.

The T cell immune response is a key component of antitumor immunity and is primarily responsible for the autoimmune destruction seen in diseases like MS (12). Interference with the activation and/or migration of T cells is therefore an effective immune-suppressive strategy. In mice, UV radiation inhibits T cell activation at least in part by directly altering the Ag presentation capability of UV-exposed epidermal Langerhans cells (13). Although this explains the ability of UV to suppress local skin immunity, it does not explain its ability to suppress systemic immune responses (14, 15). Indeed, low, suberythemal doses of UVB systemically suppress the development of normal T cell responses to Ag, inhibiting the activation and expansion of effector CD4⁺ and CD8⁺ T cells in the skin-draining lymph nodes (16, 17). In addition, UV is a potent inducer of a subtype of regulatory T cells (Tregs) known as UV-Tregs (18). These Tregs can persist in the skin for up to 2 wk post-UV exposure and suppress the cutaneous immune responses nonspecifically in an IL-10–dependent manner (19–21). In addition to UV-Tregs, UV suppression of local cell-mediated immunity involves the production of PGE₂ (22), pyrimidine dimers (23), cis-urocanic acid (24), and reactive oxygen species (25). However, none of these local UV mechanisms are involved in suppressing systemic cell-mediated immune responses (16, 17). In the context of EAE, UV-mediated protection is independent of vitamin D (8) and urocanic acid (26), meaning that another mechanism is responsible for UV suppression of systemic immune responses.

In this study, we have tested the hypothesis that systemic UV-induced immune suppression is caused by an alteration to normal lymphocyte recirculation. UV-induced PGE₂ blocks the egress of immune cells from lymph nodes (27), resulting in a preferential accumulation of circulating immune cells in peripheral lymph nodes (28). Indeed, one way in which UVB systemically suppresses the contact hypersensitivity response is by altering the trafficking of effector T cells, reducing the number of effector memory CD8⁺ T cells infiltrating into the challenged ear skin (16). This indicates a defect in T cell homing to inflamed sites.

T cell recirculation is dependent on the lipid chemoattractant sphingosine-1-phosphate (S1P), which can be produced by keratinocytes post–UV irradiation (29). S1P, which binds to the receptors S1P_1-5, is higher in concentration in the blood than lymphoid tissues (30). The low S1P concentration within the lymphoid tissue, maintained by high S1P lyase activity, generates a gradient that attracts S1P receptor–positive lymphocytes out of the lymphoid tissues and into the blood. Lymphocytes express low amounts of S1P_2-5 but high levels of S1P₁, which are required for their egress from lymphoid tissues into the blood (31, 32). Engagement of S1P₁ with an agonist leads to the activation and subsequent internalization of the receptor (30, 33), disrupting lymphocyte egress out of lymphoid tissues. Interference with normal S1P-mediated circulation of lymphocytes can have a profound effect on immunity. Fingolimod (FTY720), an agonist of S1P receptors 1, 3, 4, and 5 and, more recently, siponimod and ozanimod, which more selectively agonizes S1P receptors 1 and 5, has been approved for the treatment of MS. These drugs work primarily by altering lymphocyte circulation, decreasing peripheral naive and central memory T cell numbers (34, 35). In mice, the FTY720-mediated decrease in circulating T cells is accompanied by an increase in cells within the lymph nodes (36). In this study, we show that UV exposure preferentially blocks the recirculation of naive and central memory T cells from the skin-draining lymph nodes by downregulating S1P₁ expression. This downregulation and alteration to recirculation was dependent on the activation of sphingosine kinase 1 (SphK1) and the consequent enhanced production of S1P in lymph nodes. Inhibition of SphK activity using SKI-II prevented the UV-induced increase in S1P and blocked the accumulation of T cells within the lymph nodes. Together, these results confirm the S1P pathway as a novel mechanism involved in UV-mediated systemic manipulation of cell-mediated immunity.

All animal experiments were approved by The University of Sydney Animal Ethics Committee (#6020 and #1352). The shaved backs of 7- to 10-wk-old female C57BL/6 mice (Australian BioResources, Moss Vale, NSW, Australia) were exposed to 25.5 J/cm² solar-simulated UV (delivered as seven daily doses of 3.64 J/cm², which we have shown to protect from EAE) (6) or a single, immune-suppressive 8 J/cm² dose (37) of solar-simulated UV generated by a 1000 W xenon arc solar simulator (92.3% UVA plus 7.7% UVB) (Oriel Instruments, Stratford, CT). UV irradiations always occurred in the mornings at the same time of day. Control mice were littermates of irradiated mice. Controls were shaved, sham-irradiated, and cohoused with irradiated mice.

Shaved C57BL/6 mice were topically treated on the back with either 50 μg SphK inhibitor SKI-II (Cayman Chemical, Ann Arbor, MI) or the equivalent 2.5 μl DMSO (1.25% v/v, vehicle control) in 200 μl surfactant (2:1:1 of 100% ethanol/propylene glycol/water) (38). Twenty-four hours later, the mice were irradiated (or not) with 8 J/cm² UV. The mice were sacrificed 24 h post-UV, and skin-draining lymph nodes were collected.

Splenocytes and lymph node cells (2.4–3.1 × 10⁶) from CD45.1⁺ B6.SJL-Ptprca Pepcb/BoyJ mice (Australian Resource Centre, Perth, WA, Australia) were i.v. transferred into control or irradiated CD45.2⁺ congenic C57BL/6 mice 6 h after the last dose of UV. The skin-draining (inguinal) lymph nodes were collected 48 h after transfer and CD45.1⁺ cells analyzed by flow cytometry.

Lymph nodes were dissociated using 25-gauge needles and passed through a cell strainer. Migration of lymphocytes to murine CCL19 (PeproTech, East Windsor, NJ) was performed in transwell plates (96-well, 5-µm pores, polycarbonate membrane; Corning, Corning, NY) according to the manufacturer’s protocols. A total of 2.5 × 10⁵ cells resuspended in media (RPMI 1640 GlutaMAX with 10% charcoal-stripped FCS) were loaded into the top wells with or without 100 nM CCL19 in media in the bottom wells. The experiment was three biological replicates (n = 3 mice/group) performed in triplicate and incubated in a 37°C incubator. After 1 h, cell counts and flow cytometry analysis were performed on the cells in the bottom wells. Background migration levels were calculated as an average of the no-chemokine controls and subtracted. To obtain enough cells, triplicates were then pooled and stained for flow cytometry analysis.

Lymph nodes were dissociated using 25-gauge needles and passed through a cell strainer. Peripheral blood collected via cardiac puncture into sodium citrate–coated tubes was added to 12 ml RBC lysis buffer (16.8 mM Tris with 139 mM ammonium chloride) and incubated at 22°C. After 10 min, the samples were spun down at 200 × g for 10 min, washed with FACS buffer (PBS with 0.5% BSA and 0.4% EDTA), and centrifuged again at 500 × g for 5 min.

A total of 10⁶ cells was stained with Fixable Viability Dye eFluor 455UV (Thermo Fisher Scientific, Waltham, MA) and anti-Fcγ receptor Abs (CD16/32, final concentration of 1.25 µg/ml; clone 2.4G2; BioLegend, San Diego, CA). Cells were then stained with CD45.1-allophycocyanin (1 µg/ml; clone A20; BD Biosciences, Franklin Lakes, NJ), CD45.2-FITC (2.5 µg/ml; clone 104; BD Biosciences), CD3ε-PE-CF594 (1 µg/ml; clone 145-2C11; BD Biosciences), CD4-PerCP (1 µg/ml; clone RM4-5; BioLegend), CD8α-FITC (2.5 µg/ml; clone 5H10-1; BioLegend), CD62L-BV421 (1 µg/ml; clone MEL-14; BD Biosciences), CD44-BV786 (0.2 µg/ml; clone IM7; BD Biosciences), CD69-PE-Cy7 (1 µg/ml; clone H1.2F3; Thermo Fisher Scientific), and S1P₁-allophycocyanin (1 in 100; clone 713412; R&D Systems, Minneapolis, MN) for 30 min at 4°C. The cells were washed twice and analyzed on a BD LSRII flow cytometer (BD Biosciences).

Four hours after the last dose of UV radiation, the skin-draining lymph nodes were collected and RNA extracted using TRIzol (Life Technologies). A total of 3.1 µg RNA was synthesized into cDNA (SuperScript III; Life Technologies) and analyzed using TaqMan OpenArray Mouse Inflammation Panel on a QuantStudio 12K Flex Real-Time PCR System (Life Technologies). In addition, expression of Actb (housekeeping), Cxcr3, Ccr7, Cxcl9, Bmal1, Clock, Rev-verbα, Cry1, and Cry2 were analyzed by RT-PCR on a Roche LightCycler 480 using the following primers: Actb forward, 5′-AGATCAAGATCATTGCTCCTCCT-3′; Actb reverse, 5′-ACGCAGCTCAGTAACAGTCC-3′; Cxcr3 forward, 5′-GCCATGTACCTTGAGGTTAGTGA-3′; Cxcr3 reverse, 5′-TGACTCAGTAGCACAGCAGC-3′; Ccr7 forward, 5′-CGCAACTTTGAGCGGAACAA-3′; Ccr7 reverse, 5′-GCAATGTTGAGCTGCTTGCT-3′; Cxcl9 forward, 5′-TGGAGTTCGAGGAACCCTAGT-3′; Cxcl9 reverse, 5′-AGGCAGGTTTGATCTCCGTT-3′; Bmal1 forward, 5′-ACATAGGACACCTCGCAGAA-3′; Bmal1 reverse, 5′-AACCATCGACTTCGTAGCGT-3′; Clock forward, 5′-CCTATCCTACCTTCGCCACACA-3′; Clock reverse, 5′-TCCCGTGGAGCAACCTAGAT-3′; Rev-verbα forward, 5′-GCCCCCTTGTACAGAATCGAA-3′; Rev-verbα reverse, 5′-GGCCAGAGGCTCATCTTGGA-3′; Cry1 forward, 5′-TTGCCTGTTTCCTGACTCGT-3′; Cry1 reverse, 5′-GACAGCCACATCCAACTTCC-3′; and Cry2 forward, 5′-TCGGCTCAACATTGAACGAA-3′; Cry2 reverse, 5′-GGGCCACTGGATAGTGCTCT-3′. A total of 6 µl master mix containing SYBR Green (Thermo Fisher Scientific) and primer mix at a ratio of 5:1 was added to 4 µl cDNA. The PCR program consisted of 55 cycles each containing 20 s of denaturing and annealing at 95°C and extension for 40 s at 60°C. Experimental replicates (experiment conducted in triplicates) were averaged and the relative expression to housekeeping calculated using the 2^−ΔΔCT method.

Lymph node lipids were extracted using a modified Bligh and Dyer protocol (39). Sample and 1.25 fM C17 S1P internal standard (d17:1 S1P; Avanti Polar Lipids, Alabaster, AL) were added to 2 ml chloroform (Sigma-Aldrich, St. Louis, MO) and 2 ml methanol (Sigma-Aldrich) and vortexed for 2 min. Another 1 ml chloroform was added and vortexed for 30 s. To separate the lipid-containing layer, 1 ml ultrapure water was added, and the tube was vortexed for 30 s. The lipid-containing chloroform layer was collected and dried under nitrogen flow. For S1P analysis, the lipid precipitate was resuspended in 25 µl 100% methanol and transferred to autosampler vials.

Lipid extraction of blood followed the optimized protocol of Frej et al. (40). Blood collected in sodium citrate was centrifuged at 300 × g for 15 min to generate platelet-rich plasma. The plasma layer was collected and spun at 20,000 × g for 20 min to obtain platelet-poor plasma. A total of 10 µl platelet-poor plasma was diluted with 55 µl TBS (50 nM Tris-HCl and 0.15 M NaCl [pH 7.5]). A total of 200 µl protein-precipitation solution (methanol containing C17 S1P at a final concentration of 50 nM) was added and vortexed for 30 s. The samples were centrifuged at 17,000 × g for 2 min to precipitate the proteins. A total of 150 µl supernatant was collected and transferred to autosampler vials. All samples were immediately analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS).

A total of 2 µl lipid extract was injected into a reverse-phase liquid chromatography column (Phenomenex Synergi 4 µm Polar-RP 80 Å, 50 × 1 mm) coupled to an Agilent Technologies 1100 HPLC system. The HPLC operated with a flow rate of 100 µL/min and a gradient beginning with 50% solvent A (97:2:1 of water/methanol/formic acid) and 50% solvent B (68:29:2:1 of methanol/acetone/water/formic acid). At 2 min, the gradient became 100% solvent B, returned to 50% solvent B at 12 min, and held at 50% solvent B until 45 min. Each sample was injected twice for technical replicates and the average taken.

The ions were detected using a SCIEX QTRAP 5500 triple-quadrupole mass spectrometer in negative ion mode with multiple-reaction monitoring. S1P (d18:1) and C17 S1P (d17:1) were detected scanning for the following transitions: mass-to-charge ratio (m/z) −378.1 to −79.0 (S1P) and m/z −364.1 to −79.0 (C17 S1P). All samples were analyzed using MultiQuant Software (SCIEX) with areas under the peak measured for each molecule. The area of S1P in each sample was normalized to the area of the internal standard, C17 S1P. The amount of S1P in each sample was quantified using an S1P standard curve covering the range 7.8 nM to 1000 nM (each concentration containing 1.25 fM C17 S1P).

Lipids were extracted from the skin-draining lymph nodes of control unirradiated and UV-irradiated mice using a modified version of the above protocol. Lymph nodes were added to 500 µl each of chloroform and methanol without any internal standards and vortexed for 2 min. A total of 250 µl chloroform was added, followed by 30 s of vortexing; 250 µl distilled water was added, vortexed for 30 s, and left until the layers separated. The lipid-containing chloroform layer was isolated and evaporated to dryness under nitrogen flow.

Naive skin-draining lymph nodes were collected and processed into single-cell suspension under sterile conditions within 3 h. Cells were pooled together for hemocytometer cell counting and then seeded into U-bottom 96-well plates at 2 × 10⁵ cells/well. The lipid preparations were reconstituted in complete RPMI (RPMI 1640 with 10% charcoal-stripped FCS, 0.05 µM 2-ME, and 1% penicillin-streptomycin) and added to the cells at a ratio of one lymph node amount of lipids to one lymph node number of cells. Because an average of 2 × 10⁶ cells was obtained from each lymph node, one tenth of the lipid preparation from each lymph node was added to each well. The cells were incubated at 37°C for 24 h and then stained for flow cytometry analysis.

Skin-draining lymph nodes were placed in 400 μl lysis buffer (20 mM HEPES [pH 7.4], 10 mM KCl, EDTA-free complete protease inhibitor mixture [Roche], 1 mM DTT, and 3 mM β-glycerophosphate) before homogenizing via bead-beating at medium intensity for 1 min. Tissue lysates were cleared by centrifugation at 1000 × g for 10 min. The supernatant was collected and total protein concentrations determined using the Bradford assay (Bio-Rad Laboratories, Hercules, CA).

SphK1 and SphK2 activity in lymph node lysates were assayed by measuring the conversion of d17:1 (C17) sphingosine (C17) to d17:1 (C17) S1P using reaction conditions modified from Couttas et al. (41). Different buffer conditions that favor the activity of each isoform were used to independently assay their activity. SphK1 assay buffer contained 50 mM HEPES (pH 7.4), 15 mM MgCl₂, 3% CHAPS, 2 mM ATP, 5 mM NaF, 10 μM fumonisin B1, 1 mM 4-deoxypyridoxine, 2 mM Na₂VO₄, 0.1% BSA, and 10 μM SphK2 inhibitor K145 (Sigma-Aldrich). SphK2 buffer contained 50 mM HEPES (pH 7.4), 15 mM MgCl₂, 0.5 M KCl, 2 mM ATP, 5 mM NaF, 10 μM fumonisin B1, 1 mM 4-deoxypyridoxine, 2 mM Na₂VO₄, 0.1% BSA, and 0.1 μM SphK1 inhibitor 1a (a kind gift from Prof. Kevin Lynch, University of Virginia) (42). S1P phosphatase activity was measured as the conversion of C17 S1P to C17 sphingosine (41). The buffer used contained 50 mM HEPES (pH 7.4), 2 mM ATP, 20% glycerol, 1 mM EDTA, 150 mM NaCl, and 0.1% BSA.

A total of 5 μg lymph node protein lysate was added to the above buffers with either 500 pmol C17 sphingosine (for SphK) or C17 S1P (for S1P phosphatase). Reactions ran for 30 min at 37°C with agitation on a thermomixer and were stopped by adding 200 μl methanol containing 100 pM d18:0 dihydrosphingosine (dhSph) and d18:0 dhSph 1-phosphate as internal standards (Avanti Polar Lipids). Reaction mixtures were centrifuged to clear insoluble debris (10,000 × g, 10 min), and supernatants were transferred to autosampler vials for quantification using LC-MS/MS.

The following LC-MS/MS method was used to calculate Sphk and S1P phosphatase activity and S1P levels within skin-draining lymph nodes following topical cutaneous application of Sphk inhibitor. S1P and sphingosine were quantified using a Vanquish UHPLC system (Thermo Fisher Scientific) coupled to a TSQ Altis Triple Quadrupole mass spectrometer (Thermo Fisher Scientific), operating in positive ion mode. Lipid extracts were resolved on a Phenomenex Synergi 50 × 1-mm Polar-RP column using a 7-min binary gradient with solvent A (water with 0.2% formic acid and 2 mM ammonium formate) and solvent B (methanol with 1% formic acid and 1 mM ammonium formate), at a flow rate of 300 µl/min. The gradient was held at 50% solvent B for 1 min, increased to 95% B from 1 to 3 min, held at 95% B to 3.5 min, returned to 50% B at 4 min, and held at 50% solvent B until 7 min.

The ions were analyzed with selective-reaction monitoring, scanning for peaks at transitions m/z 286.3 to 250.2 (C17 sphingosine), m/z 366.3 to 250.2 (C17 S1P), m/z 302.3 to 266.3 (dhSph), and m/z 382.3 to 266.3 (dhSph 1-phosphate). C17 S1P and C17 sphingosine peak areas were expressed as a ratio to their internal standards and subsequently converted to picomoles in the sample using an external standard curve. Enzymatic activity was calculated as picomoles of substrate per minute per microgram of protein.

S1P lyase activity was measured using the fluorogenic substrate 2S-ammonio-3R-hydroxy-5-((2-oxo-2H-chromen-7-yl)oxy)pentyl hydrogen phosphate (Cayman Chemical). Once cleaved by S1P lyase, fluorescent umbelliferone is released. A total of 30 μg lymph node protein lysate was assayed under reaction conditions modified from previous publications (43, 44) consisting of 0.5 M K₃PO₄, 25 μM Na₂VO₄, 250 μM pyridoxal-5′-phosphate, and 125 μM S1P Lyase Fluorogenic Substrate. Reactions ran for 6 h before the fluorescence of each well was read using a plate reader (370-nm excitation and 460-nm emission).

Statistical analysis was conducted using Prism v9.1.2 (GraphPad Software, San Diego, CA). Shapiro-Wilks normality test was performed (α = 0.05; p < 0.05) for all experiments. If the data were normally distributed, a Student t test (comparing two groups) or an ordinary one-way ANOVA with a Holm-Šídák multiple-comparisons test (comparing more than two groups) was used. If the data were not normally distributed, a nonparametric analysis using Mann–Whitney U test (comparing two groups) or Kruskal–Wallis test with Dunn multiple comparisons (comparing more than two groups) was used.

We have previously shown that seven consecutive daily doses of UV for a total of 25.5 J/cm² prior to induction of EAE can successfully protect mice from an autoimmune attack on the CNS (6). To determine whether changes in lymphocyte trafficking are involved in systemic immune suppression, we exposed mice to our EAE-protecting regimen. Exposure of the skin to UV radiation significantly increased the total number of cells within the skin-draining (inguinal) lymph nodes by 1.6-fold (Fig. 1A). This was associated with a corresponding decrease in total cell numbers within non–skin-draining lymph nodes: mesenteric (gut-draining) and lumbar (CNS-draining). Similar effects on skin-draining lymph node cellularity were observed even when a single 8 J/cm² dose of UV, which we have previously shown to be immune-suppressive (37), was given to mice. UV-exposed mice had significantly higher cell numbers within the skin-draining lymph nodes (1.35-fold), and this correlated with an overall increase in lymph node weight (Fig. 1B). Hence, exposure to a systemically immune-suppressive dose of solar-simulated UV is associated with a redirection of lymphocytes toward the skin-draining lymph nodes and away from other sites.

A prior study showed that the change in leukocyte distribution following UV exposure is not due to a direct effect on the circulating immune cells themselves, but rather an extrinsic downstream change in the lymph nodes draining the UV-exposed skin (28). To confirm this with our EAE-protective UV regimen, we adoptively transferred CD45.1⁺ cells from unirradiated donors into CD45.2⁺ recipients 6 h after the last dose of UV. At 48 h after transfer, CD45.1⁺ cells within the skin-draining lymph nodes were analyzed by flow cytometry. Although the overall number of CD45.1⁺ cells in the skin-draining lymph nodes was not significantly altered by UV (2.0 ± 0.3 × 10⁵ versus 2.1 ± 0.4 × 10⁵ for no-UV and UV recipients, respectively), there was a significant increase in the proportion of CD45.1⁺CD44^–CD62L⁺ naive T cells and a concomitant decrease in the proportion of CD44⁺CD62L⁺ effector memory T cells (Fig. 1C). Naive and memory T cells follow CCR7-CCL19/21 and CXCR3-CXCL9/10/11 pathways, respectively (45, 46). We therefore examined the changes in expression of these chemokines and receptors by quantitative RT-PCR in skin-draining lymph nodes; however, there were no significant UV-induced changes in Cxcr3, Ccr7, Cxcl9, Cxcl10, or Ccl19 (Fig. 1D). To rule out a direct effect of UV on the intrinsic migratory properties of T cells, we used a transwell migration assay. We found no difference in the capacity of cells isolated from the skin-draining lymph nodes of UV-irradiated and control mice to migrate toward CCL19 (Fig. 1E). Virtually all lymph node cells that had migrated to CCL19, regardless of whether they were from UV-irradiated or control mice, were CD3⁺CD19^– T cells (Fig. 1F). Circulating lymphocyte numbers can also be affected by changes in the circadian rhythm (47), which can impact the cutaneous response to UV-induced erythema (48) and carcinogenesis (49). However, no UV-induced changes to Bmal1, Clock, Rev-verbα, Cry1, and Cry2 circadian rhythm genes were observed (Fig. 1G). Thus, changes within the UV-irradiated host are responsible for alterations in T cell trafficking, and this is independent of direct effects on T cell migration or changes in chemokine or circadian rhythm pathways.

We next determined which immune cells were responsible for the increase in lymph node cellularity observed post–UV exposure (Fig. 1B). Flow cytometry analysis of the skin-draining lymph node cells of irradiated mice revealed no changes in the proportion of CD3^–CD19⁺ B cells, CD3^–CD19^– cells, or any subset of CD3⁺CD19^– T cells (Fig. 2A). However, an increase in the number of B cells and CD3^–CD19^– cells was observed (Fig. 2B), which is consistent with what we have previously described using this dose of UV (37, 50). In addition, a 1.3-fold increase in lymph node T cells was observed in UV-exposed mice. This was contributed to by a significant increase in naive and central memory CD4⁺, as well as naive and central memory CD8⁺ T cells (Fig. 2C). Analysis of the peripheral blood of these same mice showed a corresponding decrease in the number of circulating T cells with no changes in B cells or other immune cells (Fig. 2D). The overall decrease in peripheral blood T cells was attributed to a significant reduction in the number of naive and central memory CD4⁺ (1.78-fold and 2.3-fold, respectively) and naive and central memory CD8⁺ T cells (2.06-fold and 2.03-fold, respectively) (Fig. 2E). No significant changes were observed with effector memory T cells in either the lymph nodes or blood. Together, this suggests that UV selectively affects the recirculation and lymph node accumulation of naive and central memory helper and cytotoxic T cells.

The selective targeting of naive and central memory T cells by UV parallels the effect of fingolimod treatment in people with MS (34). In light of the fact that chemokine and circadian rhythm pathways did not appear to be affected by UV (Fig. 1D–(G), this suggested that UV may be causing changes to T cell recirculation by targeting the S1P pathway, specifically the S1P₁ receptor and possibly S1P itself.

Exposing mice to an immune-suppressive dose of UV increased the number, but not proportion (Supplemental Fig. 1B), of S1P₁-negative cells in draining lymph nodes (mean increase of 1.4 × 10⁶ cells) (Fig. 3A). Of these S1P₁-negative cells, the majority were T cells (mean increase of 0.96 × 10⁶ cells) (Fig. 3B), accounting for 94% of the increase in total T cells in the lymph nodes. This indicated that UV may be affecting T lymphocyte recirculation by downregulating S1P₁ expression. Indeed, mirroring our previous findings (Fig. 2B), UV-exposed mice had a significant increase in the number of S1P₁-negative naive CD4⁺ T cells, naive CD8⁺ T cells, and central memory CD8⁺ T cells in their skin-draining lymph nodes (Fig. 3C). An increase in S1P₁-negative CD4⁺ central memory T cells, though not significant, was also seen. Again, no effect was observed with effector memory T cells. This indicates that UV-mediated changes in lymphocyte recirculation on naive and central memory T cells may be associated with decreases in surface S1P₁ expression.

One possible explanation for the reduction in surface S1P₁ on lymph node T cells is via UV modulation of the early lymphocyte activation marker CD69, which can bind S1P₁ to promote receptor internalization and degradation (51, 52). However, flow cytometry analysis comparing CD69 expression on S1P₁-negative T cells of control and UV-irradiated mice showed no significant differences (Fig. 3D). Thus, UV-induced downregulation of S1P₁ on T cells in the skin-draining lymph nodes is not associated with changes to surface CD69. S1P₁ expression can also be downregulated by ligand-mediated activation. Indeed, LC-MS/MS quantification of S1P showed a significant increase in the skin-draining lymph nodes 24 h after exposure to UV (1.24-fold; (Fig. 3E). No-UV–induced changes to S1P were observed in non–skin draining (mesenteric) lymph nodes or plasma, indicating a local UV-induced change in the S1P gradient mediated by a specific increase in S1P within the skin-draining lymph nodes.

To determine whether the increased S1P in the skin-draining lymph nodes was biologically significant, we cultured primary lymph node cells with lipids extracted from skin-draining lymph nodes of control or UV-irradiated mice. Twenty-four hours later, we assessed S1P₁ expression by flow cytometry. CD4⁺ T cells cultured with lipids from UV-irradiated mice showed a significant reduction in S1P₁ expression (Fig. 3F). A similar trend was observed with CD8⁺ T cells but this did not reach statistical significance. These results show that the modest increase in S1P within skin-draining lymph nodes of UV-irradiated mice is sufficient to downregulate S1P₁ expression on T cells.

S1P is produced following the phosphorylation of sphingosine by Sphk1 and Sphk2. Degradation of S1P can be achieved in two ways: 1) dephosphorylation of S1P by S1P phosphatases back into sphingosine, or 2) irreversible cleavage of S1P by S1P lyase into hexadecanal and phosphoethanolamine (53). The UV-induced increase in lymph node S1P could be due to increased SphK activity or a reduction in the activity of S1P phosphatases and lyase. To assess the activity of these S1P-associated enzymes, protein lysates were obtained from control or UV-irradiated skin-draining lymph nodes. UV-irradiated mice had a significant increase in Sphk1, but not Sphk2, activity measured as conversion of C17 sphingosine to C17 S1P (Fig. 4A). To determine whether a UV-triggered decrease in S1P degradation might also be involved in increased S1P in lymph nodes, the activity of S1P phosphatase and S1P lyase were assessed. S1P phosphatase activity was assayed as conversion of C17 S1P to C17 sphingosine, whereas S1P lyase activity was assayed with a fluorogenic substrate. No differences in S1P phosphatase or lyase activity were seen between control and UV-irradiated mice (Fig. 4B). Together, these data indicate that the UV-induced increase in skin-draining lymph node S1P was due to an increase in Sphk1 activity.

If this was the mechanism responsible, inhibition of SphK activity should reverse the UV-mediated increase in both S1P and T cells within skin-draining lymph nodes. To confirm this, we topically applied SKI-II (38), which inhibits both Sphk1 and Sphk2 (or vehicle) onto the shaved back skin of mice 24 h prior to UV exposure. Local topical cutaneous application of SKI-II avoided any systemic effects of SphK inhibition. As expected and consistent with our previous findings (Fig. 3E), there was a significant increase in lymph node S1P within the UV-irradiated vehicle-treated mice in comparison with the unirradiated vehicle-treated mice (Fig. 4C). There was no UV-induced increase in S1P in mice treated with the SphK inhibitor (Fig. 4C). SKI-II alone did not affect S1P levels in the skin-draining lymph nodes (2.29 ± 0.33 pmol/mg for no-UV vehicle compared with 2.47 ± 0.34 pmol/mg for no-UV SKI-II). This confirms that the activation of SphK by UV is responsible for the increase in S1P within the skin-draining lymph nodes. To confirm that UV-induced S1P was responsible for the accumulation of T cells in skin-draining lymph nodes, another group of mice were topically treated with SKI-II. Twenty-four hours after UV exposure, flow cytometry analysis was performed on cells isolated from the inguinal lymph nodes. Consistent with our previous findings (Fig. 3C), there was a significant increase in both S1P1^–CD4⁺ and S1P1^–CD8⁺ naive T cells (Fig. 4D). However, this UV-induced accumulation of naive T cells did not occur in mice pretreated with SKI-II (Fig. 4D). Thus, UV activation of SphK in the skin leads to increases in S1P at distant sites that result in the accumulation of T cells in skin-draining lymph nodes.

The mechanisms driving local and systemic UV immune suppression are different. Although much is known about how UV suppresses local cutaneous immune responses, the mechanisms by which UV suppresses distant immunity are only just being revealed. Our study has shown for the first time, to our knowledge, that UV exposure of the skin systemically alters T cell recirculation, preferentially trapping naive and, to a lesser extent, central memory T cells in the UV-exposed skin-draining lymph nodes. This cellular sequestration away from peripheral blood was the result of a UV-triggered increase in SphK activity, leading to elevated levels of S1P in regional lymph nodes. This increase in S1P in turn caused a downregulation of S1P₁ expression on T cells leading to their sequestration.

Pharmacological interference with normal lymphocyte trafficking is a highly effective way to induce therapeutic immune suppression. Natalizumab for example, is an mAb that prevents leukocyte trans-migration by blocking α₄β₁ integrin leukocyte adhesion to endothelial VCAM-1 (54). Another example is fingolimod, a functional S1P receptor agonist that induces the internalization and downregulation of S1P₁, the functional result of which is a reduction in circulating leukocytes (35), in particular naive and central memory T cells (34), and an accumulation of T cells within lymph nodes (36). Interference with normal lymphocyte trafficking following administration of natalizumab and fingolimod causes profound systemic immune suppression that is effective in the treatment of autoimmune diseases, particularly MS (55, 56). We have shown in this study that exposure to UV, which is a known immune suppressant and associated with protection from MS (4), also interferes with normal lymphocyte trafficking.

The ability of UV to induce peripheral T cell subset changes in mice is consistent with observations in humans following exposure to therapeutic doses of artificial UV. Patients with clinically isolated syndrome had increased proportions of circulating FrIII Tregs following narrowband-UVB phototherapy (57). Similarly, patients with polymorphic light eruption had increased percentages of peripheral blood Tregs following photohardening (58). The increase in proportion of peripheral Tregs and restoration of their suppressive function is associated with the therapeutic benefit of UV in polymorphic light eruption. In this study, we demonstrated that in mice exposed to a single immune-suppressive dose of UV, circulating naive T cell numbers are significantly reduced, accounting for 78% of the fall in peripheral T cell numbers. UV also reduced central memory T cells in the peripheral blood of mice although, as expected, the absolute number of these cells was very low, accounting for only 7% of the overall decrease in circulating T cells. Whether UV also affects naive and central memory T cell subsets in the circulation and skin-draining lymph nodes of humans has yet to be investigated.

UV exposure stimulates mechanisms known to promote lymph node “shutdown,” such as PGE₂ production (59, 60), the activation of the complement pathway (61, 62), and increased expression of high endothelial venules (27, 28). However, we showed that the accumulation of naive and central memory T cells in the skin-draining lymph nodes was almost completely (97%) accounted for by an accumulation of S1P₁-negative naive and central memory T cells. The downregulation of S1P₁ was not associated with changes to CD69, but rather a significant increase in S1P. An increase in S1P within lymph node tissue would be expected to maintain low S1P₁ expression on lymphocytes (30), leading to an inability to sense S1P gradients. Consistent with this hypothesis would be an accumulation of S1P₁-negative cells in the skin-draining lymph node, which is what we observed. This was confirmed by our adoptive transfer of CD45.1⁺ cells into CD45.2⁺ irradiated mice, in which there was a specific increase in the percentage of naive CD45.1⁺ T cells in lymph nodes that drained UV-exposed skin. We showed that UV did not affect the expression of chemokines or their receptors nor alter circadian rhythm genes. Chemotaxis assays confirmed that UV did not directly impact the intrinsic ability of T cells to migrate to lymph node chemokine gradients. Thus, UV affects the host lymph node environment rather than the lymphocytes themselves.

The ability of UV to alter T cell recirculation, promote the expansion of UV-activated Tregs (18), and inhibit T cell proliferation upon stimulation (16) provides a potent formula for the comprehensive suppression of cell-mediated immunity. The trapping of naive T cells in the skin-draining lymph nodes maximizes the chance of interactions with other cellular mediators of UV-induced immune suppression, such as dendritic cells (63), mast cells (37), B cells (50, 64), and UV-Tregs (19). It is known that if an Ag is presented during the first 24–72 h of exposure to UV, T cell activation is suppressed (16) in part by UV-Treg and B cell production of IL-10 (19, 20, 65). We have now shown that even if activation of naive T cells occurs, UV not only limits clonal expansion (16), but it also interferes with normal lymphocyte recirculation by reducing the likelihood of cells exiting the lymph nodes.

Immune modulatory B cells activated in the skin-draining lymph nodes (50, 64) mediate UV-immune suppression (6) and skin carcinogenesis (66). In this study, we found that although total B cell numbers increased in skin-draining lymph nodes following UV exposure, there was no concomitant decrease in circulating B cells. This observation of stable peripheral B cell numbers is consistent with what others have shown in the context of fingolimod treatment in EAE (67). One possible explanation for this difference between T and B cells is that circulating B cell numbers are tightly regulated and maintained by the egress of B cells from the bone marrow (68, 69), an immunological organ that is known to be indirectly affected by UV (70).

UV radiation of the skin has previously been reported to alter cutaneous lipid profiles by decreasing free triglyceride levels and stimulating the production of immune-suppressive lipids such as platelet-activating factor (71–73). In addition, cutaneous UV exposure induces the production of immune-modulating extracellular vesicles (74) and sphingolipids by keratinocytes (75). In this study, we show for the first time, to our knowledge, that UV radiation can induce immunomodulatory lipids within the skin-draining lymph nodes, enhancing S1P levels by 20% and the consequent downregulation of S1P₁ expression on T cells. UV was found to activate SphK1 to increase production of S1P in the skin-draining lymph nodes. Topical treatment with an SphK inhibitor on the skin prior to UV irradiation inhibited both the UV-mediated increase in lymph node S1P and the accumulation of naive T cells in these distant organs. These results confirm that local stimulation of SphK at the skin, or possibly draining lymph node, is responsible for the immunological effects we have observed. Alternatively, topical inhibition may have prevented UV promotion of S1P synthesis in keratinocytes (75), although it is not yet known whether skin-derived S1P can reach the draining lymph nodes.

In conclusion, alterations in the S1P pathway are a novel mechanism involved in UV-mediated systemic immunomodulation. By increasing the local S1P concentration in lymph nodes apposed to the skin and downregulating S1P₁ expression on T cells, UV-induced lipids limit T cell recirculation and thus their ability to perform effector functions. Considering that this is the same mechanism of action for powerful immune suppressants like fingolimod, it will be important to determine whether this UV-triggered event plays any role in UV suppression of distant autoreactive immune responses. Targeting these processes may lead to better control over the impacts of UV exposure on human health.

We thank the Animal, Sydney Mass Spectrometry, and Sydney Cytometry core facilities for subsidized access and the support staff in these core facilities for the assistance.

The authors have no financial conflicts of interest.