Skin microbiome in AD
AD is characterized by a waxing and waning course, in which worsening of AD lesions (flares) occur periodically followed by a period of clinical improvement with treatment.5Microbes on the skin are thought to play an important role in skin homeostasis by modulating the immune system,8 providing colonization resistance against pathogens,9 or, in the case of Staphylococcus aureus (S. aureus), exacerbating skin inflammation.10The skin microbiome composition changes dramatically across the AD flare cycle, with microbial diversity dropping as disease severity increases, and a concomitant increase in abundance of S. aureus in the majority of cases.11 A predominant S. aureusstrain is often observed in the flare state, while heterogenousStaphylococcus epidermidis strain communities are seen in both the flare and post-flare states. On non-flare AD skin, microbial diversity remains high; however, the composition and functional potential of the AD subjects’ microbiome are distinct from healthy individuals, highlighting potential microbiome-related factors that could increase the susceptibility to flares.12 This altered microbiome signature in the non-flare skin of AD patients is more pronounced in individuals with high levels of circulating IgE and also associates with molecular changes in the skin surface microenvironmental niche.13 There is growing evidence that commensal microbes may play a mechanistic role in skin barrier repair and attenuate inflammation during AD via AhR-dependent signalling and glucocorticoid-related pathways.14, 15
The predominance of S. aureus strains in flares and relative reduction of commensal microbes are linked with skin barrier disruption and inflammation in AD skin.16 A meta-analysis reported that around 70% of individuals across all ages with AD are colonized with S. aureus on lesional skin, 30-40% in non-lesional skin, and 62% in the nares.17 AD patients have significantly higher odds of S. aureus colonization than healthy controls at all sites compared.17
There is a need for greater understanding of the initiators and drivers of the altered skin microbiome and why S. aureus predominates on AD skin. The contribution is likely complex and multifactorial, driven by Th2 skewing of the immune milieu, disruption to the skin barrier by mutations in genes such as FLG , and alterations in lipid and protein production in the epidermis that increase S. aureusadherence.18-20 There is also strong evidence from genetic disorders that either impact the immune system (such as inborn errors in immunity), or disrupt skin barrier function (such as congenital ichthyosis), resulting in overrepresentation ofStaphylococcus species as a common feature in microbial communities.21, 22 Thus, both host and microbiome factors likely contribute to AD pathogenesis in susceptible individuals.
S. aureus has long been associated with AD23and shown to demonstrate its pathogenic effects through various mechanisms such as disruption of barrier integrity, intrinsic host immune dysregulation, and expression of virulence genes. (Figure 1) Virulence factors such as α-toxin, protein A (SpA), lipotechoic acid (LTA), phenol soluble modulin (PSM)-α, and proteases can damage keratinocytes.24 While AD-associated S. aureusstrains have been isolated, several studies have investigated the differences between S. aureus strains isolated from children with AD as compared to other S. aureus strains to potentially identify specific strains that may contribute to AD skin disease.20, 25, 26
Viral-host interactions also mediate AD morbidity. AD patients with a history of S. aureus skin infections are at greater risk of eczema herpeticum (EH), a severe blistering cutaneous infection caused by herpes virus.27, 28 Some mouse and in vitrostudies have raised the possibility that S. aureus strains which produce staphylococcal toxic shock syndrome toxin-129or α-toxin30 may enhance viral entry. Conversely, AD host factors such as genetics, immune dysregulation, and barrier defects also increase susceptibility to EH compared to healthy individuals.
Differences in fungal composition have also been described in AD. Cross-sectional sequencing studies demonstrated lower relative abundances of Malassezia spp . Correspondingly, the higher prevalence of non-Malassezia fungi such as Cladosporium, Alternaria , and Aspergillus in AD skin12, 31may be attributed to environmental selection pressures asMalassezia spp grow best in lipid-rich conditions, which are deficient in AD skin. Certain Malassezia species perform protective functions, whereas others may induce skin inflammation.32, 33 A small study (n=17) demonstrated that Malassezia globosa, a commensal yeast from healthy adult volunteers, secretes a protease (Malassezia globosa Secreted Aspartyl Protease 1 - MgSAP1) which disrupts S. aureus biofilms by hydrolyzing SpA.34 A murine AD model, however, showed that M. pachydermatis, M. sympodialis , and M. furfur activated IL-23 and IL-17 inflammatory pathways, and this was corroborated by the presence of Malassezia -specific production of these cytokines by memory T cells in adult AD patients.33 These targeted studies were designed to investigate the specific inflammation-associated mechanisms exerted by commensal Malassezia spp, mostly in in vitro or in vivo models. There is, however, little direct data on how yeast interact with other skin microbiota or influence human host responses.
Native skin commensals also form part of the innate host defence against pathogenic organisms. S. epidermidis and Staphylococcus hominis are examples of coagulase-negative staphylococci (CoNS) which stimulate host production of antimicrobial peptides like cathelicidin and human beta defensin (HBD), and also directly produce PSM-γ and PSM-δ, which inhibit the growth of pathogenic bacteria.35
An observational sequencing study (n=100 adult patients) detected increased abundances of the Demodex folliculorum mite on AD skin compared to healthy controls.36 However, no further mechanistic studies were done to assess the role of Demodex mites in inducing or perpetuating skin inflammation in these patients. While house dust mites have been shown to trigger inflammation in AD murine models,37 there is little data to show that they trigger AD flares in AD patients. There is thus insufficient evidence to conclude if skin mites play a major role in AD development.
Skin microbiome markers of AD in early life
Early life exposure to the external environment may shape host-microbe interactions that play roles in skin health and atopic disease onset.38 Distinct fluctuations in neonatal and infant skin physiology in the first year of life influence skin microbiota colonization patterns. The skin microbiome in healthy neonates is characterized by the predominance of Bacillota (formerlyFirmicutes ), such as Staphylococcus andStreptococcus , and fewer Actinomycetota (formerlyActinobacteria ) (Cutibacterium andCorynebacterium ). A small, open, non-randomized, cross-sectional study found that soon after delivery, vaginally born neonates (n=4) have skin microbiota that are similar to their mothers’ vaginal fluid, predominantlyLactobacillus or Prevotella spp., while neonates born through Caesarean section (n=6) acquire mainly maternal skin commensals such as Staphylococcus, Corynebacterium, and Cutibacterium spp .39 In a subsequent publication where infants were followed up to 30 days of life, the authors reported that 4 Caesarean-born infants who were exposed to their mothers’ vaginal fluids at birth continued to exhibit skin microbiota characteristics similar to vaginal-born infants.6 A larger follow-on observational study recruited n=98 vaginal-born infants and n=79 Caesarean-born infants, 30 of whom were swabbed with maternal vaginal fluids after birth. Caesarean-born infants who underwent vaginal seeding had a gut microbiota developmental trajectory up to age 12 months that more closely resembled vaginal-born infants, compared to those who were not seeded, suggesting that early life vaginal microbiota transfer had a sustained effect up to 1-year post intervention.7 A subsequent double-blind randomized controlled trial of 10 vaginal-seeded infants vs 10 controls (seeded with sterile saline) found that vaginal-seeded infants had significantly reduced alpha diversity in the skin at day 1, as well as in transitional stool up to day 30 of life.8 There has, however, been no longitudinal data beyond 1 year of life, nor any data on whether this intervention had any impact on infant disease outcomes. Furthermore, as the Caesarean sections were all elective and non-emergent, there is little known about how other peripartum factors may influence maternal and infant microbiota and maternal-infant transmission.
Various factors such as maternal microbiome, mode of delivery, and early environmental exposure may influence the maturation trajectories of the infant skin microbiome, and from murine models, potentially infant skin immunology as well. A systematic review by Xiong et al found that studies examining the link between mode of delivery and AD risk harboured significant heterogeneity that could be attributed to differences in country, study design, and method of AD ascertainment.40 No definite conclusions can thus be made about the impact of delivery mode on AD risk.
Several longitudinal studies have demonstrated early skin microbial signals that may be predictive of AD onset. A small longitudinal study, nested within a birth cohort, found that infants who developed AD at age 1 year (n=10) had a much lower abundance of commensalStaphylococcus species detected by 16S rRNA gene sequencing at 2 months of age relative to controls (n=10) .41 A larger prospective birth cohort study (n=146) found that early S. aureuscolonization (by culture-based methods) in the first few months of life appeared to precede the onset of AD in infants.42Nakamura et al further observed that infants in the Chiba birth cohort who had S. aureus strains harbouring dysfunctional mutations in their Agr quorum-sensing (QS) system did not go on to develop AD (n=24), postulating that this functional S. aureus virulence mechanism may be important for AD pathogenesis.43Further studies are needed to examine the mechanistic pathways through which early differences in microbiota composition may influence later disease onset, such as the impact of microbiota on host immunological responses, skin barrier protein expression, and barrier function.
Marked pubertal differences in the skin microbiome have been demonstrated as children sexually mature through Tanner stages.44, 45 Similarly, the skin microbiome composition of young children with AD (2-12 years, n =59) differed from adolescents (13-17 years, n =13) and adults (18-62 years, n=56) versus age-matched nonatopic healthy controls.
Skin microbiome diversity was significantly higher in the non-lesional skin of AD children than in adolescents/adults, corresponding to similar patterns in healthy individuals.46 However,Staphylococcus was significantly more abundant in lesional as well as non-lesional skin of children and adolescents with AD. Skin commensals for AD and control subjects were also age-specific: such asStreptococcus spp enriched in children compared toCutibacterium and Corynebacterium in adults.
There are only a few observational studies describing changes in skin microbiome signatures in early infancy that may be associated with AD onset. Further mechanistic studies are needed to deconvolute the potential role of the skin microbiota in AD pathogenesis and their interactions with the host skin barrier and immune system in the pre-clinical disease state.