Signaling sphingolipids are biomarkers for atopic dermatitis prone to disseminated viral infections

Abstract

Background:

Life-threatening viral diseases such as eczema herpeticum (EH) and eczema vaccinatum (EV) occur in less than 5% of individuals with atopic dermatitis (AD). The diagnosis of AD, however, excludes all individuals with AD from smallpox vaccination.

Objective:

To identify circulatory and skin lipid biomarkers associated with EH and EV.

Methods:

Stratum corneum and plasma samples from 15 AD subjects with a history of EH, 13 age- and gender-matched AD subjects without EH history, and 13 healthy non-atopic controls (NA) were analyzed by liquid chromatography tandem mass spectrometry for sphingolipid content. Sphingosine-1-phosphate (S1P) and ceramide levels were validated in plasma samples from the ADVN/ADRN repository (12 NA, 12 AD, 23 EH) and plasma from 7 EV and 7 matched AD subjects. S1P lyase was down-regulated in human primary keratinocytes to evaluate its effect on herpes simplex virus 1 (HSV-1) replication in vitro.

Results:

The stratum corneum of EH patients demonstrated significantly higher levels of free sphingoid bases than NA patients, indicating enhanced sphingolipid turnover in keratinocytes (p<0.05). Plasma from two independent cohorts of EH patients had a significantly increased S1P/ceramide ratio in EH subjects versus AD and NA subjects (p<0.01). The S1P level in plasma from EV subjects was twice its level in AD subjects (mean=1533 vs 732 pmol/ml, p<0.001). Downregulation of S1P lyase expression with silencing RNA led to an increased S1P level and doubled HSV-1 titer in keratinocytes.

Conclusion:

Our data point to long-term abnormalities in the S1P signaling system as a biomarker for previous disseminated viral diseases and a potential treatment target in recurring infections.

Capsule summary

In patients with a history of eczema vaccinatum and eczema herpeticum, sphingolipid homeostasis is substantially perturbed in favor of S1P-mediated signaling and persists for many years after the last disease episode.

Introduction

Atopic dermatitis (AD) is the most common inflammatory skin disease, affecting up to 25% of children and 7% of adults worldwide.^1,2 It is associated with significant impairment in quality of life and other atopic co-morbidities, such as asthma and food allergy. A small subset (<5%) of AD has increased propensity to life-threatening disseminated viral infections including eczema herpeticum (EH) and eczema vaccinatum (EV). Beyond increased morbidity and hospitalization, these patients are excluded from smallpox vaccination. Furthermore, due to the risk of viral dissemination, the Centers for Disease Control and Prevention recommends that smallpox vaccination should be excluded in all patients with AD and their immediate families. Since AD affects more than 25% of children, EH and EV are significant public health problems. Biomarkers that distinguish this subset of AD at risk for disseminated viral infection are needed because the majority of patients with AD would benefit from smallpox vaccination in the event of a bioterrorist attack.

NIAID/NIH established the Atopic Dermatitis Vaccinia Network (ADVN) to investigate mechanisms underlying disseminated viral infections in AD.³ Investigations in ADVN suggested that AD and EH are complex diseases associated with epithelial skin barrier dysfunction and aberrant anti-viral immune responses influenced by environmental triggers.^4,5 The exact pathogenic mechanisms, however, leading to disseminated herpes simplex virus (HSV) infections in patients with AD are unknown. AD patients with filaggrin gene mutations, in particular, carry a greater than 10 odds ratio for development of EH, highlighting the importance of the skin barrier.⁶ The skin barrier is formed by the organized expression and assembly of proteins and lipids. While systemic skin genomic and protein traits have received some attention in relation to EH and EV,^7–9 skin lipids as well as circulatory lipids in EV and EH have not been studied.

The sphingosine-1-phosphate (S1P) signaling system is a critical signaling pathway that allows the resurgence of latent viral infections.^10–19 In addition to its role as an important regulator of cell survival and proliferation, S1P controls the egress of lymphocytes from lymph nodes.^10,11 S1P-mediated signaling is reported to promote replication of hepatitis C¹² and influenza^13,14 viruses, and sphingosine kinase 2 (SPHK2) permits persistence of lymphocytic choriomeningitis virus Cl 13.¹⁵ A separate line of investigation pointing to the importance of the S1P signaling system in viral replication and resurgence from the latent state comes from the increased number of disseminated viral diseases reported in multiple sclerosis patients who receive fingolimod (FTY720, an analog of sphingosine).^11,16–19 FTY720 is converted in the human body into the phosphorylated form (FTY720-phosphate) by SPHK2. It evenly distributes between the circulation and lymph nodes and, therefore, disrupts S1P-established gradient and gradient-controlled egress of lymphocytes and promotes S1P receptor internalization and degradation,²⁰ thereby causing immunosuppression. As AD can be complicated by disseminated herpes simplex virus 1 (HSV-1) infection and is considered a contraindication for smallpox vaccination, we explored the existence of biomarkers associated with the S1P signaling system in EH and EV patients. Using stratum corneum and plasma samples from EH and EV subjects as well as gene manipulations with human keratinocytes and HSV-1 in vitro, we investigated the association between sphingolipid turnover/S1P signaling system and disseminated viral skin diseases.

Methods

Study subjects

Skin and plasma samples were obtained from 15 adult subjects with active AD who had a prior history of three and more episodes of EH, 13 subjects with active moderate to severe AD, and 13 healthy non-atopic (NA) individuals with no personal or family history of atopy and skin diseases at National Jewish Health (NJH Cohort). Plasma samples were also selected from the ADVN/Atopic Dermatitis Research Network (ADRN) repository in the University of Rochester, NY, which included 12 NA subjects, 12 AD subjects, 11 EH subjects who had one to two prior EH episodes, and 12 EH subjects who had three and more prior EH episodes. All groups were matched for age, gender, and race. A summary of NJH and ADRN repository cohort subject characteristics is provided in Tables E1 and E2.

A caucasian, non-hispanic female was identified in the ADRN/ADVN registry as a subject that was enrolled into the study as an AD patient in 2006, and later had her first EH episode in 2008. The patient was subsequently enrolled in ADRN studies as a patient with more than 3 EH episodes. Plasma samples collected in 2006–2021 from this patient were analyzed in this study. The patient was 24 years old at the time of initial study enrollment.

In addition, plasma samples were obtained from seven AD patients with a prior history of EV stored at Oregon Health & Science University in Portland, OR, and seven AD subjects, who were also vaccinated against smallpox but did not develop EV, that were matched by gender, race, age, and time post-vaccination at the moment of blood draw. A summary of EV cohort subjects is presented in Table E3.

The patients in this study were part of the prospective, clinical mechanistic study registered at ClinicalTrials.gov, identifier: NCT03038932. The protocol was approved by the Western Institutional Review Board, protocol number 20161695. All subjects gave written informed consent prior to participation in the study.

Sample processing and statistics

Information about skin tape strip collection, human primary keratinocyte culture and in vitro experiments, tape strip amd plasma processing, description of mass pectrometric analysis of lipids, and statistics used is provided in the Supplementary Material.

Results

Lipids from stratum corneum of EH subjects suggest increased sphingolipid turnover in the skin

In these experiments, sphingolipidomic analyses were performed at NJH on stratum corneum samples from skin tape strips (STS) collected from non-lesional skin of NA controls, AD patients who never experienced EH, and AD patients with a prior history of EH (three episodes or more). Lipidomic analysis revealed that STS of AD subjects with a history of recurring EH have a distinct lipid profile as compared to AD subjects without EH, as well as healthy controls. EH subjects have the highest increase in the proportion of short-chain non-hydroxy fatty acid sphingosine (NS) ceramides (CER) (as presented by N-16:0-NS-CER in Fig 1A) and the strongest decline in the proportion of the long-NS-ceramides (as presented by N-26:0-NS-CER in Fig 1B) in comparison to NA subjects. EH subjects are also characterized by a global decrease in the ratio between omega-esterified hydroxy fatty acid sphingosine (EOS) ceramides and NS-ceramides (Fig 1C). Of note, out of all NS-ceramide molecular species, only a short-chain N-16:0-NS-ceramide was significantly increased when comparing either AD or EH group skin samples to the NA group (Fig 1A and Tables E4 and E5). Furthermore, in contrast to AD subjects studied in this report and previously,^21,22 non-lesional skin of EH subjects had increased levels of sphingomyelin (Fig 1D), with a short-chain 16:0-sphingomyelin molecular species contributing the most to this upregulation (Fig 1E).

Fig 1. Non-lesional skin of EH subjects demonstrate disturbed sphingolipid profile.

The levels of short-chain (N-16:0-) (A), long-chain (N-26:0-) NS-CER (B), and EOS-CER/NS-CER ratio (C) were the most abnormal in skin samples of EH subjects as compared to AD subjects and NA controls. (D) The stratum corneum of EH but not AD subjects demonstrate substantially increased total sphingomyelin content, with all sphingomyelin molecular species being unselectively increased (E). All lipid molecular species were quantified by targeted LC-ESI-MS/MS and normalized by sample total protein content. Some data were expressed as relative percentage within each lipid subclass. For the total list of analyzed molecular species, see Supplemetary Tables E4 and E5. Here and further, where applicable, data are presented as box and whiskers plots with individual values shown, with box indicating 25–75% interqualrtile range and whiskers showing minimum and maximum values. * p<0.05, ** p<0.01, *** p<0.001. This and further data are compared using one-way Anova with a correction for multiple comparisons (Tukey) unless otherwise specified. EH non-lesional skin – n=13, AD non-lesional skin – n=13, NA skin – n=15.

The content analysis of free sphingoid bases in the non-lesional stratum corneum of EH subjects revealed an unusual phenomenon, namely, the increased levels of free sphingosines, but not dihydrosphingosines and phytosphingosine, in comparison to AD-only subjects (Figs 2 and E1). This phenomenon is particularly interesting in terms of increased levels of 17-carbon sphingosine (Figs 2B and E1), as it is known to be formed through metabolic turnover of phytoceramides (NP-ceramides) (Fig E1 and reference²³). We also found that the stratum corneum of EH subjects, but not AD-only subjects, had decreased levels of non-hydroxy fatty acid phytosphingosine ceramides (NP-ceramides), while the levels of NS- and non-hydroxy fatty acid dihydrosphingosine ceramides (NDS-ceramides) were not changed (Fig E1). Altogether, these data demonstrate the strong dysregulation of sphingolipid metabolism in the skin of EH subjects and indicate increased sphingolipid turnover in the skin of these subjects, as the increased levels of free sphingosines were detected in EH skin samples.

Fig 2. Non-lesional skin samples of EH subjects have increased levels of free sphingoid bases.

Non-lesional stratum corneum even-chain (C18-) (A) and odd-chain (C17-) (B) sphingosines are upregulated in EH skin compared to AD skin. Each sphingoid base was quantified by targeted LC-ESI-MS/MS and normalized by sample total protein content. EH non-lesional skin – n=13, AD non-lesional skin – n=13, NA skin – n=15.

Plasma ceramides separate EH subjects into a unique AD subgroup

In an attempt to find circulatory markers associated with EH that also support dysregulated sphingolipid metabolism in EH, we performed a sphingolipidomic analysis of plasma collected at NJH from the same subjects who provided stratum corneum samples for analysis. First, we found little or no increase in circulatory S1P levels in plasma from EH subjects (Fig 3A). However, plasma from EH subjects had a decreased level of ceramides, a S1P signaling counterpart (Fig 3B), that resulted in an increased S1P/ceramide ratio (Fig 3C) in these samples. Both ceramide and S1P levels in plasma from AD subjects were approximately the same as in healthy NA subjects. Both sphingosine-based (NS-) and dihydrosphingosine-based (NDS-) ceramides in plasma were affected in EH subjects (Fig 3D and E).

Fig 3. Atopic subjects with a history of EH are distinct from AD subjects without history of EH by circulatory levels of ceramides and S1P.

Plasma lipids from two cohorts of subjects (A-E – NJH cohort; F-H – ADRN repository cohort) were extracted and analyzed by the LC-ESI-MS/MS for the content of S1P and ceramides. Both cohorts demonstrate that EH patients have a slight increase in circulatory levels of S1P (A,F) with a concomitant significant decrease in ceramide content (B,G), that leads to a substantial shift in S1P / ceramide ratio (C,H). Both NS-ceramides (ceramides, D) and NDS-ceramides (dihydroceramides, E) are affected in plasma samples of EH subjects. Note that plasma samples from EH subjects with prior history of one to two EH episodes (Gr1–2) or three and more episodes (Gr3+) are identical in sphingolipid abnormalities (F-H). NJH cohort: NA – n=13; AD - n=12; EH - n=15. ADRN repository cohort: NA – n=12; AD – n=12; EH Gr1–2 – n=11; EH Gr3+ - n=12. Analyzed species of NS-ceramides in the ADRN repository cohort are the same as presented in Figure 3D and E.

To replicate our observations obtained using patient samples from the NJH cohort, we performed a similar analysis of separate plasma samples from the NIAID/NIH ADRN/ADVN repository located at the University of Rochester, NY. Plasma samples from 12 NA subjects, 12 AD subjects, 11 EH subjects who had one to two prior EH episodes, and 12 EH subjects who had three and more prior EH episodes were procured and analyzed in a blinded fashion. All subjects were matched by age, gender, and race. The analysis of samples from the ADRN repository confirmed our original observations. Interestingly, plasma NS-ceramide levels in both groups of EH subjects were about half of the levels found in NA and AD subjects (Fig 3G). Further, the trend for upregulation of S1P levels seen in EH plasma from the NJH cohort (Fig 3A) was more clearly demonstrated in samples from the ADRN repository (Fig 3F). As a result, all EH subjects had significantly upregulated plasma S1P/NS-ceramide ratios irrespective of whether they had one-two or three and more historic episodes of EH (Fig 3H). AD subjects without EH did not differ from control subjects by any parameter. Importantly, AD severity had no effect on plasma sphingolipids as both moderate-to-severe AD (NJH cohort) and mild AD (ADRN Repository cohort) had exactly the same S1P-to-ceramide ratio (Fig 4). Our data, therefore, demonstrate significant dysregulation of sphingolipid homeostasis in EH that is manifested in plasma years after the last EH episode.

Fig 4. AD severity does not affect plasma S1P / ceramide ratio.

EASI scores are substantially different in NJH (moderate to severe) and ADVN repository (mild to moderate) cohorts. However, AD severity has no effect on plasma S1P to total NS-ceramide ratio in these subjects. Analyzed species of NS-ceramides in the NJH and ADVN cohorts are the same as presented in Figure 3D and E.

Dysregulated plasma sphingolipid profile is manifested in an AD subject before developing EH

In an attempt to see if dysregulated S1P-to-ceramide ratio in AD can be used as a predictor of susceptibility to EH, we performed a search in the ADRN repository for any AD subjects that changed their diagnosis from AD to EH after enrollment into ADRN studies. The ADRN biobank repository has collected plasma and serum samples from AD and EH subjects for more than fifteen years. In the entire repository, we found only one subject who was enrolled at NJH as an AD patient in 2006 and then developed EH in 2008. Plasma from that subject was collected in 2006 (AD diagnosis), then in 2017, 2019, and 2021 (all years had an EH diagnosis with three or more episodes by 2017). As shown in Fig 5, the plasma S1P-to-ceramide ratio in this single patient prior to EH development was as high as after developing EH, and, in most years, this ratio was at the higher end of 25%−75% interquartile range determined for the AD subjects tested at NJH (shadowed rectangles in Figures 3C and 5).

Fig 5. S1P-to-ceramide ratio in plasma of a patient who clinically transitioned from AD to EH category.

Plasma was collected from the subject who enrolled in ADRN studies in 2006 as AD patient, then was diagnosed with EH in 2008 and later enrolled in ADRN studies as EH subject who had three or more episodes of EH. Lipids were extracted and analyzed for the S1P and ceramide content by the ESI-LC-MS/MS. Years indicate when plasma was collected. Shadowed boxes provide 25–75% interquartile range of ratio values determined for the NJH cohort as reported in Fig 4C. Analyzed species of NS-ceramides are the same as presented in Figure 3D and E,

AD patients with a history of EV have increased levels of plasma S1P

To test if abnormalities in sphingolipid homeostasis are characteristic not only to EH but also to another cutaneous disseminated viral infection in AD, we analyzed plasma from EV subjects. Plasma from subjects with a history of EV was analyzed and compared to age, gender, and race-matched samples from AD subjects (who were vaccinated against smallpox but did not develop EV) collected at the same time. As shown in Fig 6A, plasma from each analyzed EV subject had about twice the level of S1P and dihydro-S1P (DHS1P, endogenous S1P analog) as in matched AD subjects. NS-ceramide and NDS-ceramide levels in plasma were also twice as high in EV subjects versus AD subjects (Fig 6B). There was no selectivity in the increase between molecular species of ceramides when comparing ceramides in plasma from EV and AD subjects (data not shown). These data demonstrate that plasma from AD subjects who experienced EV contains a unique lipid signature over 30 years after the EV episode.

Fig 6. AD patients with a history of EV have upregulated levels of plasma S1P and ceramides.

(A) S1P and DHS1P levels in plasma samples from AD and EV subjects. (B) Ceramide (NS-CER) and dihydroceramide (NDS-CER) levels in plasma samples from AD and EV subjects. Sphingolipid levels were measured by LC-ESI-MS/MS. n=7 for each group. Statistical difference was determined using a two-tailed Student’s t-test. Analyzed species of NS- and NDS-ceramides are the same as presented in Figure 3D and E.

Downregulation of S1P lyase increases HSV-1 viral titer in human keratinocytes in vitro

S1P lyase (SGPL1) is a global regulator of sphingolipid turnover and intracellular S1P levels. To test if the increase in the intracellular S1P facilitates viral replication in keratinocytes, we inhibited the expression of SGPL1 in primary human keratinocytes by silencing RNA (siRNA). Keratinocyte cultures were induced to differentiate in the presence of 1.3 mM CaCl₂ for 5 days then were infected with HSV-1 (0.1 multiplicity of infection (MOI)). Viral particle numbers in cell lysates were assessed twenty-four hours after infection. As additional controls, we examined HSV-1 replication in keratinocytes with inhibited expression of sphingosine kinases 1 and 2 (SPHK1/SPHK2), two major enzymes that control S1P levels in addition to S1P lyase, or in keratinocytes treated with scrambled control siRNA. As shown in Fig 7, downregulation of SGPL1 expression drastically increased intracellular levels of S1P and DHS1P in 5-day Ca²⁺-differentiated keratinocytes, while silencing of SPHK1/2 only slightly decreased the levels of these molecules (Fig 7A). Reverse transcription polymerase chain reaction (RT-PCR) analysis of cell lysates for the expression of HSV-1 glycoprotein D revealed doubling of HSV-1 viral particles in the lysate from cells with silenced SGPL1 (Fig 7B) that was also confirmed by the viral plaque assay (Fig 7C). These experiments demonstrate a positive relationship between the intracellular S1P level and the HSV-1 replication or infectivity rate in keratinocytes and point to S1P lyase as an important enzyme involved in viral replication.

Fig 7. S1P promotes herpes simplex virus 1 replication in keratinocytes in vitro.

Primary human keratinocytes were silenced with control non-targeting siRNA, SGPL1, SPHK1 or SPHK2 siRNA, and induced to differentiate for five days in the presence of 1.3mM CaCl₂. Cells for lipid analysis were taken at this point. Then, keratinocytes were infected with herpes simplex virus 1 (0.1 MOI) and cell lysates were prepared 24 hours post-infection. Inhibition of S1P lyase (SGPL1) expression drastically upregulates intracellular S1P and DHS1P levels (A) and the number of viral particles in cells lysates as measured by RT-PCR (B) or plaque assay (C). Data from one of the two independent experiments are shown, each done in triplicate. *** - p < 0.0001 versus corresponding control, one-way ANOVA.

Discussion

Disseminated viral skin diseases such as EH and EV are the most devastating complications of AD and often result in hospitalization or death. AD is known to be linked to impaired metabolism of skin sphingolipids with a special impact on skin ceramides that represent the majority of skin lipids and play a key role in defining skin barrier properties.^24–27 However, until now, it was not known whether EH is a unique AD endotype or simply a more severe form of AD that is characterized by a global shift toward short-chain ceramides and a decrease in the proportion of EOS-ceramides to NS-ceramides in the skin.^22,28,29 To fill this gap in knowledge, we have performed an analysis of stratum corneum lipids from non-lesional skin of EH patients and compared them to skin samples from matched AD subjects and healthy NA controls. All EH patients in the study did not have an active episode of EH at the time of skin sampling but reported a history of three or more EH episodes throughout their life.

What we found was only partially consistent with the concept that EH is the most severe form of AD. Thus, changes in the stratum corneum ceramides, which are characteristic for AD, were the most pronounced in EH patients (Fig 1). However, one parameter, namely, free sphingoid base levels in the skin was uniquely increased only in EH skin samples. In contrast to our earlier work (published in references^21,22 and Fig 2), which demonstrated that non-lesional skin of AD subjects usually had free sphingoid base levels at the lower end of values characteristic for healthy controls, stratum corneum samples of EH subjects had higher levels of free sphingoid bases than the AD without EH group (Fig 2). Furthermore, this phenomenon was also observed for sphingosine with 17 carbon atoms (C17-sphingosine) (Fig 2B). Usually, sphingoid bases with odd chain lengths are rarely present in mammalian cells. However, in differentiating keratinocytes, the formation of odd chain-length precursors of sphingoid bases actively occurs through a metabolic turnover of phytoceramides to phytosphingosine, then to phytosphingosine-1-phosphate and finally to pentadecanal and ethanolamine-phosphate, with S1P lyase catalyzing the last step of phytoceramide biotransformations.²³ Formed pentadecanal can be further converted to pentadecanoic fatty acid that, in turn, can enter sphingolipid biosynthesis initiated by serine palmitoyltransferase (SPT) that condensates fatty acid coenzyme A(s) (CoA) and serine and, hence, can be present as C17-sphingoid base in all newly formed sphingolipids (reference³⁰ and Fig E1). As phytoceramides represent a substantial portion of skin ceramides, a high amount of C17-sphingosine accompanied by a substantial decline of NP-ceramides in EH skin (Fig E1) clearly indicates active ceramide turnover through (phyto)sphingoid base-1-phosphates and their degradation by S1P lyase. In the current study, we found that this process is more active in EH than in AD skin, providing evidence for the increased sphingolipid turnover in EH.

Reports of an increase in cases of different viral infections in multiple sclerosis patients who receive FTY720 provided mechanistic insight into our findings.^11,16–19 FTY720 is a prodrug that structurally resembles (phyto)sphingosine. In the human host, FTY720 is converted to an analog of S1P, FTY720-phosphate, which binds with high affinity and activates four out of five S1P receptors (S1PR1,3,4,5).²⁰ In fact, the anti-inflammatory properties of FTY720 are determined, in part, by its ability in a phosphate form to downregulate S1PR1 expression through its ubiquitination and degradation and to block the egress of lymphocytes from lymph nodes. Continuously increased systemic and tissue levels of FTY720-phosphate in treated multiple sclerosis patients favor the emergence of multiple viruses from latent phase and/or their replication, including herpes simplex,³¹ varicella zoster,^16,32 and hepatitis C.¹⁹ This information led us to hypothesize that EH might be linked to either an elevated keratinocyte S1P level or S1P turnover which can promote intracellular and extracellular S1P-mediated signaling and favor viral replication. Multiple viruses can benefit from S1P-mediated signaling at several different levels. Thus, the activation of JAK/STAT1 signaling by S1P takes part in influenza virus propagation and cytopathogenicity.³³ Subsequently, the activation of intracellular S1P generation in response to viral infection^34,35 may lead to the increased histone acetylation and increased replication of viral genome through the inhibition of histone deacetylases as a result of S1P lyase-mediated S1P degradation and the formation of Δ2-hexadecenal that directly binds to histone deacetylases.^36,37 In fact, the increase in stratum corneum content of free sphingoid bases is consistent with this hypothesis, as S1P is undetectable in stratum corneum due to its degradation by lipid phosphate phosphatases during keratinocyte maturation and stratum corneum formation.

To test our hypothesis, we used siRNA to inhibit the expression of major enzymes that control the S1P level (SGPL1, SPHK1, and SPHK2) within Ca²⁺-differentiated primary human keratinocytes and then infected those keratinocytes with HSV-1. While the inhibition of SPHK1 and SPHK2 expression led to a slight decrease in intracellular S1P level, it did not affect HSV-1 titer in infected keratinocytes (Fig 7). However, the inhibition of SGPL1 expression in keratinocytes dramatically increased the S1P level and doubled the HSV-1 titer in infected keratinocytes (Fig 7). Thus, our data directly confirms previously published information about the importance of S1P-mediated signaling in viral replication^12–15 and suggests that HSV-1 benefits from elevated S1P turnover and signaling in keratinocytes.

Having confirmed the importance of S1P in viral replication in the in vitro system, we investigated whether circulating biomarkers related to sphingolipid metabolism exist that are unique to EH or EV. S1P levels in plasma of EH subjects from the NJH cohort were minimally higher than in healthy NA subjects (Fig 3A) but were more pronounced in plasma samples from EH patients in the ADRN repository cohort (Fig 3F). Importantly, in blinded studies, circulatory ceramide levels were significantly lower in plasma from EH subjects in comparison to NA and AD subjects (Figures 3B and G). This observation is interesting from a signaling perspective, as S1P and ceramides are considered opposing elements in a pro-survival signaling network, where S1P promotes survival and proliferation, while ceramides act as potent inducers of apoptosis and necrosis.^38,39 Therefore, the observed shift in balance between S1P and ceramides in the plasma of EH subjects indicated an overall increase in S1P-mediated signaling, even without a substantial increase in circulatory levels of S1P. Of note, the difference in the severity scores in the AD groups between the NJH cohort (moderate to severe) and the ADRN repository cohort (mild) had no effect on plasma S1P and ceramide levels or the S1P/ceramide ratio (Figures 3 and 4). This additionally links observed lipid changes to a viral disease and not to AD severity. Also, our study indicates that the current clinical sub-division of EH subjects into two sub-groups based on the number of prior EH episodes (one to two episodes versus three or more episodes) is unnecessary, as both groups are biochemically identical, at least, from the point of dysregulation in homeostasis of plasma sphingolipids. Recently, Sakai and co-authors have described that serum levels of S1P correlate with AD severity.⁴⁰ This report is consistent with our observation. It should be noted that serum S1P concentration is at least double that in plasma levels as S1P is released by platelets during clotting, and serum levels of S1P reflect platelet reactivity rather than its level in circulation.⁴¹ However, this report suggests altered platelet function in AD and prompts a broader investigation into circulatory levels of S1P in AD.

To determine if the S1P-to-ceramide ratio can be used as a predictive biomarker of susceptibility to EH, we attempted to identify in NJH-ADRN repositories plasma samples from AD subject(s) who enrolled in NJH-ADRN clinical studies as AD without a history of EH and were subsequently diagnosed with EH. From these biobanks, which collected AD and EH samples for more than fifteen years, we identified only one subject falling into this category. Interestingly, S1P-to-ceramide ratio in the plasma of this single subject collected in 2006 before the first EH episode, was as high as in plasma collected 10–15 years later, when the subject had suffered from more than three EH episodes (Fig 5).

To further verify our hypothesis that S1P and S1P-mediated signaling are important drivers for increased viral infectivity or replication, we examined plasma samples collected from AD subjects with the prior history of EV (over 30 years ago) and compared them against matched AD controls who were also vaccinated but did not develop EV collected contemporaneously. We found that plasma levels of S1P in EV subjects were abnormally high^41,42 which may provoke a systemic inflammatory response and tissue damage.⁴³

Changes in plasma S1P and ceramide levels in EH and EV subjects were different in the direction and magnitude of changes. Yet, both EH and EV samples point to a disturbance in sphingolipid metabolism and S1P homeostasis provoked by two distinct viruses, leading to severe skin diseases when a hyperactivated type 2 immune response is present in the immunologic background. Our data, together with the resurgence of latent viral infections upon prolonged treatment with FTY720 in multiple sclerosis patients,^11,16–19 suggest that S1P-mediated signaling plays an important role in viral replication in the skin and that sphingosine kinases and S1P receptors may serve as potential therapeutic targets to control HSV-1 or vaccinia viral infections. Further studies are needed to determine whether changes in S1P/ceramide homeostasis precede or are a consequence of active HSV-1 or vaccinia viral infections and whether they are linked to a particular genomic variability in enzymes involved in sphingolipid metabolism.

Key Messages.

• High plasma levels of S1P or an increased plasma S1P-to-ceramide ratio are characteristic to disseminated viral diseases such as eczema vaccinatum and eczema herpeticum.

• Stratum corneum of AD subjects with a history of eczema herpeticum contains lipid molecular markers reflective of activated sphingolipid turnover.

LIST OF ABBREVIATIONS

AD

Atopic dermatitis

ADRN

Atopic Dermatitis Research Network

ADVN

Atopic Dermatitis Vaccinia Network

CER

Ceramide

CHO

Fatty aldehyde

CoA

Coenzyme A

EH

Eczema herpeticum

EOS ceramide

Omega-esterified hydroxy fatty acid sphingosine ceramide

EV

Eczema vaccinatum

FTY720

Fingolimod

HSV-1

Herpes simplex virus 1

LC-ESI-MS/MS

Liquid chromatography electrospray ionization tadem mass spectrometry

MOI

Multiplicity of infection

NA

Non-atopic controls

NDS-ceramide

Non-hydroxy fatty acid dihydrosphingosine ceramide

NJH

National Jewish Health

NP-ceramide

Non-hydroxy fatty acid phytosphingosine ceramide

NS-ceramide

Non-hydroxy fatty acid sphingosine ceramide

RT-PCR

Reverse transcription polymerase chain reaction

SGPL1

S1P lyase 1

SPHK

Sphingosine kinase

S1P

Sphingosine-1-phosphate

STS

Skin tape strips

siRNA

Silencing RNA