Inhibition of sphingosine 1-phosphate lyase activates human keratinocyte differentiation and attenuates psoriasis in mice

Sphingosine 1-phosphate (S1P) lyase is an intracellular enzyme that catalyzes the irreversible degradation of S1P and has been suggested as a therapeutic target for the treatment of psoriasis vulgaris. Since S1P induces differentiation of keratinocytes, we examined whether modulation of S1P lyase and altered intracellular S1P levels regulate proliferation and differentiation of human neonatal epidermal keratinocytes (HEKn). To identify the physiological functions of S1P lyase in skin, we inhibited S1P lyase in HEKn with an S1P lyase-specific inhibitor (SLI) and with SGPL1-specifc small interference RNA (siSGPL1). In HEKn, pharmacological treatment with the SLI caused G1 arrest by upregulation of p21 and p27 and induced keratin 1, an early differentiation marker. Similarly, genetic suppression by siSGPL1 arrested the cell cycle at the G1 phase and activated differentiation. In addition, enzyme suppression by siSGPL1 upregulated keratin 1 and differentiation markers including involucrin and loricrin. When hyperproliferation of HEKn cells was induced by interleukin-17 (IL-17) and IL-22, pharmacologic inhibition of S1P lyase by SLI decreased proliferation and activated differentiation of HEKn cells simultaneously. In addition, SLI administration ameliorated imiquimod-induced psoriatic symptoms including erythema, scaling, and epidermal thickness in vivo . We thus demonstrated that S1P lyase inhibition reduces cell proliferation and induces keratinocyte differentiation and that inhibition may attenuate psoriasiform changes. Collectively, these findings suggest that S1P lyase is a modulating factor for proliferation and differentiation, and support its potential as a therapeutic target for psoriasis in human keratinocytes. examined whether differentiation and of by inhibition of S1P lyase and altered S1P levels. We found that keratinocytes with suppressed S1P lyase had elevated S1P levels and induced G1 growth arrest with the suppression of cyclin D3, and CDK2, 4, 6. This is mediated mainly by the induction of p21, p27 and is finally followed by keratinocyte differentiation. Collectively, these findings suggest that S1P lyase could be a modulating of proliferation in human keratinocytes. These data indicate that SLI induced morphological changes similar to S1P-treated cells, but not to the levels of fully differentiated cells as observed by Ca 2+ treatment. We also measured cell proliferation rate after treatment with IL-17 and IL-22, to induce hyperproliferation and mimic the pathological condition of psoriasis. involved in psoriasis therapy is not clear at this moment and deserves further study. Taken together, this study showed that the differentiation of keratinocytes is induced by the inhibition of S1P lyase and elevated cellular S1P. We found that cell cycle was arrested at G1 phase and led to the differentiation of keratinocytes. Cytokine-mediated hyperproliferation of keratinocytes was inhibited by S1P lyase inhibition. Consequently, our study suggests that inhibition of S1P lyase could be an important regulator of psoriasis.


Introduction
Sphingosine 1-phosphate (S1P) is a bioactive sphingolipid metabolite that regulates many aspects of cell growth and differentiation. Sphingosine kinases (Sphk) catalyze the formation of S1P from sphingosine and ATP (1-4). Since S1P is an important regulator of proliferation and differentiation of human keratinocytes, S1P has been considered as an inducer of differentiation after inhibition of cell proliferation (5)(6)(7). As a signaling molecule that initiates differentiation of keratinocytes, S1P binding to 6 system was applied according to the Psoriasis Area Severity Index (PASI) (25). Scores of 0 (none), 1 (mild), 2 (moderate), 3 (severe), and 4 (very severe) were given for each of the four symptoms: erythema (redness), scaling, and skin thickness on a scale of 0-4 were checked daily. Scoring of psoriatic effects on the back skin was performed by independent observers blinded to treatment.
Ear thickness of mice was checked using a micrometer (Mitutoyo, Kanagawa, Japan) every 2 days. All experimental procedures were approved by Gachon University Institutional Animal Care and Use Committee (IACUC).
To induce hyperproliferation of keratinocytes, cells were treated with 100 ng/ml of IL-17 or IL-22 (R&D systems, MN, USA), which promote the proliferation of keratinocytes. Prior to stimulation with interleukins, cells were starved for 24 h in growth supplement-free medium.

RNA preparation and real-time PCR
Total RNA from cells was extracted using easy-spin total RNA extraction kit (Intron Biotechnology, Sungnam, Korea) in accordance with the manufacturer's protocol. cDNA was synthesized using the iScript™ cDNA Synthesis Kit (BioRad, CA, USA). Real-time PCR analysis was performed in StepOnePlus equipment (Applied Biosystems, MA, USA) using 2X SYBR Green Master Mix (Takara, Shiga, Japan). Expression levels of mRNA were measured as a ratio with β-actin, which was the normalization control. Primer sequences used in this study are stated in Supplemental Table S1.

Sphingolipid analysis by LC/MS/MS
For quantitative analysis of sphingoid bases in keratinocytes, cells were harvested and lysed in phosphatebuffered saline (PBS). C17:0 ceramide as an internal standard was added to each cell extract (1 mg of protein), and sphingolipids were extracted by MeOH/CHCl3 (1:2, v/v) including 0.01% butylated hydroxytoluene. To saponify phospholipids, KOH was added and incubated at 37°C for 2 h. Lipid extracts were neutralized by the addition of acetic acid and organic phase was separated and evaporated under N2.

Cell cycle analysis by flow cytometry
Cells were treated with SLI or transfected with siRNA and were washed with PBS. The cells were then centrifuged and fixed in 70% EtOH in PBS at 4°C. After washing in PBS, the cells were stained with propidium iodide (PI) containing RNase at room temperature for 45 min. Cell cycle distribution of keratinocytes was analyzed using the flow cytometer, Cytomics FC 500 (Beckman Coulter, CA, USA).
Hundred thousand events were counted during data collection. The percentage of cells in G1, S, G2/M phase was determined using CXP analysis software (Beckman Coulter, CA, USA).

siRNA-mediated inhibition of SGPL1 expression
Silencer Select Pre-designed siRNA against human SGPL1 and scrambled siRNA were purchased from Ambion (Life Technologies, CA, USA). Cells were cultured for 24 h before transfection and when 50% confluence was reached in growth supplement-free medium, cells were transfected with either scrambled siRNA or siRNA against SGPL1, overnight. Lipofectamine3000 reagent (Invitrogen, CA, USA) was used for all siRNA transfections. Then, cells were harvested to analyze the expression of mRNA and protein.

Statistical analysis
Differences between groups were tested by two-tailed t test, one-way or two-way ANOVA test within multiple groups, followed Dunnett's or Tukey's multi-comparison tests. Statistically significant differences were indicated as *p < 0.05, **p < 0.01, ***p < 0.001. by guest, on March 9, 2020 www.jlr.org

Results
Human epidermal keratinocyte differentiation is associated with increased S1P and downregulated S1P lyase. S1P is associated with inhibition of cell proliferation and induction of differentiation in human keratinocytes (5). To evaluate the expression of the genes involved in sphingolipid biosynthetic pathway in human epidermal keratinocytes, their protein levels were measured during proliferation and differentiation of keratinocytes. Since Ca 2+ is known as an important inducer of keratinocyte differentiation, HEKn cells were incubated for 5 days in medium either with low Ca 2+ (60 µM) to maintain normal cell growth condition or high Ca 2+ (2 mM) concentrations to induce differentiation (24). To measure protein expression of the enzymes, western blotting was performed. The protein levels of alkaline ceramidase 1 (Acer1) and sphingosine kinase 1 (Sphk1), enzymes that catalyze the production of S1P concomitantly, were increased under low Ca 2+ conditions as cells proliferated. The expression of S1P lyase was also increased in a time-dependent manner ( Figure 1A). In contrast, the high Ca 2+ condition downregulated Acer1 but upregulated Sphk1. No change was found in S1P lyase ( Figure 1B). To examine whether altered expression of S1P metabolizing enzymes under high Ca 2+ condition changes cellular S1P levels, we measured the S1P levels by LC/MS/MS. The levels of cellular S1P and sphingosine were increased in a time-dependent manner ( Figure 1C). These results support that high calcium condition induced increased of S1P level in keratinocytes during differentiation. To examine whether S1P induces differentiation of HEKn, cells were treated with extracellular S1P at various concentrations. We found that the protein levels of keratin1, the marker of early differentiation, and involucrin, the marker of late differentiation, were increased in the cells by extracellular S1P treatment ( Figure 1D). These results suggest that S1P metabolism is directly associated with keratinocyte differentiation.
While the level of S1P was increased by 2.5 fold, sphingosine (SO) level was not changed. In addition, the degree of alteration in sphinganine (SA) and sphinganine-1-phosphate (Sa1P) levels was higher than those of sphingosine or S1P ( Figure 2A). Based on the finding that SLI increased the levels of S1P in the cells, we questioned whether the proliferation of HEKn cells is altered by S1P lyase inhibition. Notably, treatment with SLI reduced the growth rate of HEKn cells in a time-dependent manner ( Figure 2B). This proportion was not changed regardless of SLI treatment except for 24 h time point that showed slightly decreased G2 proportion. These changes might be due to the induction of G1 phase growth arrest. To analyze the expression of regulators that are involved in the G1 phase, we measured the mRNA levels of CDKN1A encoding p21 and PCNA, a cell cycle regulator in DNA replication. We found that CDKN1A was upregulated but PCNA was downregulated. In contrast, the expression of CDKN1B encoding p27 was not altered ( Figure 2E). Activation of p21 and p27 signaling is a crucial mechanism to prevent G1 phase from entering S phase. While SLI elevated the protein levels of p21 and p27, the protein levels of cyclin D1, cyclin D3, CDK2, CDK4, and CDK6 were decreased after 24 h or 48 h post-treatment with SLI ( Figure 2F). To confirm that S1P lyase inhibition by SLI is associated with maintaining the effects of p21 and p27 by inhibiting degradation or blocking synthesis, cells were treated with SLI and by guest, on March 9, 2020 www.jlr.org cycloheximide for various times. In the presence of SLI, p21 and p27 stayed longer than those in the presence of only cycloheximide. Therefore, SLI treatment prevented the clearance of both p21 and p27 in HEKn cells ( Figure 2G). These results suggest that pharmacological inhibition of S1P lyase induces G1 arrest via induction and increased half-life of p21 and p27, and contributes to the differentiation of keratinocytes.

SLI upregulates the expression of differentiation marker proteins and alters cell morphology with reduced cell proliferation
Cell cycle arrest is required to initiate keratinocyte differentiation (28, 29). Since inhibition of S1P lyase by SLI led to G1 arrest via p21 and p27, we investigated whether the growth arrest determines induction of keratinocytes. To verify the changes at the protein level, we measured the protein levels of differentiation markers ( Figure 3A). The protein levels of keratin1 and involucrin increased, although the level of keratin10, an early differentiation marker of keratinocytes, was reduced. These results demonstrated that not all differentiation marker proteins were increased by altered S1P. However, the protein levels of loricrin, a terminal differentiation marker was decreased only at the 48 h time point.
Taken together, S1P lyase inhibition by SLI caused activation of early and late keratinocyte differentiation.
Next, we found that the cell morphology was altered when cells were treated with SLI. To compare with fully differentiated HEKn, the cells were treated with SLI, S1P, and Ca 2+ and cell morphology was observed ( Figure 3B). While cells incubated in a medium without any treatment showed proliferative cell morphology, cells treated with either S1P or SLI showed an extended cell shape and did not indicate proliferation. However, S1P-treated keratinocytes were restored with morphology like proliferative cells after 72 h of incubation, whereas cells incubated with SLI were differentiated at 48 h. These data indicate that SLI induced morphological changes similar to S1P-treated cells, but not to the levels of fully differentiated cells as observed by Ca 2+ treatment. We also measured cell proliferation rate after treatment with IL-17 and IL-22, to induce hyperproliferation and mimic the pathological condition of psoriasis.

Downloaded from
Under this condition, we examined whether inhibition of S1P lyase reduces the growth rate of keratinocytes and induces differentiation. Compared to cells of the control group, the sustained treatment with IL-17 or IL-22 promoted proliferation (Supplemental Figure S2). Interestingly, co-treatment with both SLI and interleukins reduced the growth rate when compared to the cells treated only with each cytokines ( Figure 3C). In addition, p21 and p27 were equally increased in HEKn cells treated with both SLI and interleukins when compared to the cells with no SLI treatment. Reduction of cyclin D1 expression was found in SLI-treated cells in the presence of interleukins. Both keratin1 and involucrin were increased when cells were treated with SLI regardless of the presence of interleukins ( Figure 3D).
These results indicate that proliferation of keratinocytes induced by interleukins is suppressed and differentiation is induced by S1P lyase inhibition.

S1P lyase suppression by siRNA induces G1 arrest and keratinocyte differentiation
Inhibition of S1P lyase by SLI induced the differentiation of keratinocytes. To verify whether induction of differentiation in human keratinocytes by S1P lyase suppression is not compound-specific, we genetically suppressed S1P lyase expression by SGPL1-specific siRNA (siSGPL1). siSGPL1 suppressed the expression of SGPL1 at both mRNA and protein levels (Supplemental Figure S3A, B). To examine whether genetic suppression of SGPL1 affects the sphingolipid profile, we measured cellular levels of sphingoid bases in siSGPL1-treated HEKn cells by LC/MS/MS ( Figure 4A). Similar to the results observed in the pharmacological study, which showed that SLI treatment did not alter sphingosine levels, in siSGPL1-transfected cells S1P, sphinganine, and Sa1P levels were elevated when compared to those of siCON controls. We investigated whether S1P lyase suppression induces growth arrest by induction of p21 or p27 as observed in SLI treated cells. Cells transfected with siRNA were analyzed for cell cycle distributions using flow cytometry ( Figure 4B, C). Compared to siCON, the siSGPL1transfected cells in the G0/1 phase were increased to 61.73 ± 0.96% and cells in the S phase were reduced to 14.98 ± 0.38%. In contrast, the percentage of cells in G2 phase was not different between the by guest, on March 9, 2020 www.jlr.org Downloaded from transfected groups of cells. Additionally, the expression of cell cycle regulators was measured in HEKn to verify the effect of S1P lyase suppression-induced G1 growth arrest on keratinocytes. The mRNA levels of the cell cycle regulators did not change ( Figure 4D). Although the transcriptional levels of these genes were comparable to those of the control groups, the protein levels of these regulators were altered ( Figure 4E). The protein levels of p21 and p27 were both induced. However, no significant change was found in the levels of other regulators including cyclin D1, cyclin D3, CDK2, and CDK6. These findings suggest that S1P lyase suppression by siRNA induced G1 growth arrest via p21 and p27 induction and increased cellular S1P levels.
We next assessed whether suppression of S1P lyase induces differentiation of HEKn cells.
Suppression of S1P lyase markedly downregulated keratin5 and keratin16 ( Figure 5A). In contrast, the protein levels of keratin1, keratin10, involucrin, and loricrin were elevated at all differentiation phases including early, late and terminal steps ( Figure 5B). Specifically, unlike in SLI treated cells in which only keratin1 and involucrin were increased, keratin10 and loricrin were both elevated in siSGPL1-treated cells. Compared to SLI effect, the expression of differentiation markers was enhanced after siRNA transcription. In addition, cell cycle inhibitors and keratinocyte differentiation markers were increased even in the presence of interleukins ( Figure 5C). These results demonstrate that SGPL1 suppression in keratinocytes induces growth arrest and cell differentiation despite the hyperproliferative environment induced by interleukins.

S1P lyase inhibition by SLI ameliorates imiquimod-induced psoriasis skin lesions in mice
To examine whether SLI could improve psoriasis in vivo, we induced psoriatic effect on BALB/c mice by administering IMQ. Back skin treated with IMQ showed pathological changes on the epidermis.
After treatment for 8 days, IMQ treatment induced psoriasis-like symptoms including erythema and scaling were developed ( Figure 6A). We found that cyclosporine A, a positive control (30), reduced the by guest, on March 9, 2020 www.jlr.org Downloaded from degree of erythema and scaling, and SLI alleviated the psoriasis lesion on the back skin as compared to the IMQ-treated controls ( Figure 6A). Histological analysis demonstrated that the epidermis of the skin lesions was thickened by IMQ when compared to the vehicle-treated controls ( Figure 6B). In contrast, cyclosporine A and SLI at all doses reduced the epidermal thickness significantly ( Figure 6B, C).
Cumulative scores combining the degree of erythema, scaling, and skin thickness according to the PASI demonstrated that SLI treatment significantly ameliorated pathological conditions of IMQ-induced psoriasis ( Figure 6D and Table 1). However, we could not find the changes in ear thickness among groups.
These results suggest that SLI effectively ameliorates IMQ-induced psoriasis in mice.

Downloaded from
Discussion S1P has been reported to be involved in diverse cellular functions including cell proliferation, immune suppression and cardiovascular functions (31). Subsequently, it has recently been stated that S1P lyase that regulates S1P levels by degradation, is associated with various S1P-mediated cellular functions (19). S1P lyase can be attributed to regulate intracellular pools of S1P. It terminates sphingolipid metabolism by catalyzing S1P to hexadecanal and phosphoethanolamine. Inhibition of S1P lyase can increase cellular S1P levels in various cell types. S1P degradation pathway caused by S1P lyase occurs ubiquitously in various cell types. S1P lyase-deficient mice exhibited considerably elevated S1P levels in circulation and liver (21,32) as well as in the immune system (1, 22). In skin, heterozygous S1P lyase-deficient mice developed acanthosis/orthokeratotic hyperkeratosis because S1P inhibits epidermal acanthosis and regulates non-allergic skin inflammation in mouse model (23). However, the effects of S1P lyase inhibition which increases S1P levels on keratinocyte proliferation and differentiation in animal model have not been studied in detail.
In this study, we demonstrated that (1) S1P elevation by pharmacologic and genetic inhibition of S1P lyase decreased cell proliferation by arresting cell cycle at G1 phase; (2) increased S1P levels induced differentiation of keratinocytes by upregulation of genes involved in differentiation; (3) inhibition of S1P lyase induced differentiation and inhibits growth rate in hyperproliferative keratinocytes; (4) SLI ameliorated imiquimod-induced psoriasis in mouse model. Our results suggest that S1P lyase could be an important regulation point for keratinocyte differentiation and proliferation in hyperproliferative skin disease, psoriasis.
A variety of signaling pathways relating to keratinocyte differentiation have been discovered.
Differentiation of human epidermal keratinocytes by intracellular Ca 2+ forms a physical or chemical barrier on both outside and inside of the skin. S1P has been known to regulate keratinocyte proliferation and differentiation as a signaling molecule, inducing Ca 2+ release from the ER and Golgi apparatus (31, by guest, on March 9, 2020 www.jlr.org Downloaded from 33). Although the receptor that binds to S1P directly on the surface of ER has been reported, the exact mechanism of Ca 2+ release from the ER by S1P is not elucidated yet. (8,15,34,35). S1P functions both as an extracellular ligand for G-protein-coupled receptors (S1P1~5) and as an intracellular second messenger. Unlike PLC dependent pathway via S1P3, S1P causes transient activation of ERK and inactivation of Akt/protein kinase B via S1P2 and inhibits cell proliferation (33). S1P phosphatase encoded by SGPP1 is another enzyme that regulates cellular S1P by catalyzing the dephosphorylation of S1P to form sphingosine. A recent report demonstrated that keratinocytes isolated from the skin of sgpp1 -/mice had elevated levels of intracellular S1P and increased expression of genes associated with differentiation (36). Additionally, accumulation of S1P by Sphk1 activator induced differentiation of keratinocytes (37). Thus, modulation of the enzymes metabolizing S1P levels leads to keratinocyte differentiation. Previously, the findings that growth arrest by upregulation of p21 and p27 induces squamous differentiation were reported (28, 29). In consistent with these reports, we found that pharmacological and genetic inhibition of S1P lyase upregulated p21, p27 and ultimately arrested cell cycle (Figure 2, 4).
In contrast, there are conflicting results showing that epidermal differentiation does not necessarily have to undergo growth arrest prior to differentiation (39,40). Even with these conflicting results, both opinions agreed that regulation of cell cycle is importantly associated with differentiation. This controversial point could be due to the degree or phase of growth arrest but it is not clear at this moment.
Although our results demonstrated that SLI treatment and siRNA transfection induce an increase in the levels of S1P in HEKn cells, the resulting expression pattern of differentiation markers was different.
While SLI only increased the protein levels of keratin1 and involucrin, which are the markers of early and late differentiation respectively, suppression of S1P lyase by SGPL1-specific siRNA showed upregulation of keratin1, keratin10, involucrin, and loricrin. This inconsistency could be due to higher cellular S1P levels with SLI treatment than those found with SGPL1 suppression. However, exact reason is not clear at this moment and we cannot exclude the possibility of the compound-specific effect.
Since cell confluence can initiate the differentiation of keratinocytes, cell could be differentiated even in low calcium concentration which is a normal growth condition. To exclude this possibility, we grew the cells not to reach confluence until day 5 and tried to distinguish normal growth condition (low Ca 2+ ) from differentiating condition (high Ca 2+ ). We found alteration of expression of S1P metabolizing genes represented by upregulated synthetic Sphk1 and no change in degrading S1P lyase ( Figure 1B). Indeed, altered S1P metabolism led to increased cellular S1P. These results suggest that S1P lyase could be a regulation point for S1P during differentiation.
We examined the effects of S1P on cell proliferation under hyperproliferative condition which is one of the characteristics of psoriasis. Lesions on psoriatic skin activate T cells, secreting pro-inflammatory In addition, the skin regularly treated with IMQ caused skin lesion showing abnormal keratinocyte proliferation and differentiation similar to psoriasis in human. Administration of SLI at various doses resulted in ameliorated symptoms of psoriasis such as erythema, scaling and epidermal thickness in mouse skin. In contrast, there was no effect on ear thickness by SLI. Since ear thickness is the parameter to indicate the extent of severity of inflammation, SLI has no effect on anti-inflammatory effect. Previous reports demonstrated that the inducible heterozygous S1P lyase-deficient mice induced skin irritation caused by increased S1P and hyperplasia (45). This conflicting result could be due to systemic S1P increase by heterozygous whole body SGPL1 deficiency and possible inter-organ interaction. However, the exact mechanism is not clear and deserves further study. Therefore, inhibition of S1P lyase not only arrested cell cycle but improved pathophysiological aspects of psoriasis. Whether modulation of S1P lyase is directly involved in psoriasis therapy is not clear at this moment and deserves further study.
Taken together, this study showed that the differentiation of keratinocytes is induced by the inhibition of S1P lyase and elevated cellular S1P. We found that cell cycle was arrested at G1 phase and led to the differentiation of keratinocytes. Cytokine-mediated hyperproliferation of keratinocytes was inhibited by S1P lyase inhibition. Consequently, our study suggests that inhibition of S1P lyase could be an important regulator of psoriasis.