The immune system constantly surveys the microenvironment and discriminates between harmless and potentially harmful components – a multifaceted task at barrier surfaces, such as the skin, that are constantly exposed to exogenous stimuli. There, a complex network of cellular and molecular pathways is employed, which allows the immune system to respond quickly and efficiently to harmful stimuli, while largely ignoring innocuous substances. Regulation of both innate and adaptive skin immunity is essential in preserving host integrity, thereby preventing inappropriate immune activation and pathology. T cells, as key actors of the adaptive immune system, are commonly identified by CD3 expression, and detect antigens through heterodimeric T cell receptors (TCRs) composed of either α and β or γ and δ chains (Morath and Schamel, 2020). Work over the last two decades has highlighted their importance in adult human skin, which harbors twice as many T cells as are seen in the circulation (Clark et al., 2006b). Multiple T cell subsets are involved in the defense against pathogens and tumors, and play a role in tissue homeostasis (e.g. hair follicle cycling, wound repair), but they also can cause inflammation and autoimmune diseases (Cruz et al., 2018; Ho and Kupper, 2019).

The developing conceptus is, in principle, protected from pathogens by the uterine barrier and maternal-derived antibodies while establishing a functioning immune network. Skin development is driven by a mutual inductive mechanism between ectoderm (epidermis) and mesoderm (dermis) (Carlson, 2012). The two skin compartments develop over several gestational periods (Ersch and Stallmach, 1999; Coolen et al., 2010). In the first weeks of gestation, the epidermis changes from a single layer of cells into a bilayered epidermis comprising basal cells and an embryonic and fetal-specific periderm (Holbrook and Odland, 1975). Incompletely keratinized cells are replaced by keratinocytes that differentiate while stratifying to form the fully functional epidermis, including the development of skin appendages and a functional stratum corneum between 15-24 weeks estimated gestational age (EGA). The initially dense cellular dermis is followed by augmented production of extracellular matrix components such as collagen fibers, which are detectable at 12 weeks of EGA, with distinguishable papillary and reticular dermis after 15 weeks of EGA (Carlson, 2012; Coolen et al., 2010). The first vessels in the dermis are visible from 11-13 weeks of EGA. The phenotype of cutaneous lymphatic and blood vasculature entirely develops during the second trimester and accumulation of subcutaneous fat begins (Coolen et al., 2010; Schuster et al., 2015; Smith et al., 1986; Colwell et al., 2003). In parallel to the structural changes during the in utero development, a diverse range of precursors and immune cells including T lymphocytes seed the skin (Schuster et al., 2009, 2012, 2014) and might be involved in tissue generation and regeneration (Botting and Haniffa, 2020).

In contrast to the well-studied T cells in adult human skin, extensive studies about the establishment of the T cell network in developing fetal human skin have been hampered by a limited amount of and access to fetal tissue, low absolute numbers of T cell per sample and a lack of dissociation methods that liberate viable cells with high yield but preserve T cell markers (Clark et al., 2006b; Schuster et al., 2012; Di Nuzzo et al., 2009; Sanchez Rodriguez et al., 2014). During recent developments in tridimensional visualization and analysis of early human development, high resolution single-cell methods [single-cell RNA-sequencing (scRNA-seq), mass cytometry (cytometry by time-of-flight; CyTOF)], imaging technologies (in situ transcriptomics) and computational methods including machine learning algorithms have changed biomedical research. They have facilitated the characterization of immune cell types, generation of diversity in antigen-specific recognition (e.g. TCR repertoire) and tissue compartmentalization immunity across age groups to generate a comprehensive atlas of the human immune system (Park et al., 2020; Mukhopadhyay, 2021; McGovern et al., 2017; Belle et al., 2017; Han et al., 2020) Nonetheless, the precise roles of immune cells such as T cells during skin development remains elusive. Using combinations of traditional and modern laboratory techniques, we carried out investigations on the origin, function and transcriptional profile of fetal skin T cells. Double-positive (DP) αβγδ T cells and single-positive (SP) αβ T cells with a naive phenotype were the predominant population followed by discrete subsets of SP γδ T cells, memory and regulatory T (TREG) cells (Reitermaier et al., 2021; Dhariwala et al., 2020).

Here, we provide versatile tools for the isolation and expansion of human fetal skin T cells, which enabled studying their complexity and heterogeneity using single-cell transcriptomics, TCR repertoire profiling, multiparametric flow cytometry and in situ immunofluorescence analyses. We thus layout possible future directions for advancing the understanding of skin immunity in early life.

Sort-purified CD3+ T cells were mixed with total fetal skin cells of the same donors for scRNA-seq (10x Genomics) (Fig. 2A) as described in detail previously (Reitermaier et al., 2021). Cell clustering using t-distributed stochastic neighbor embedding (t-SNE) enabled the distinction of major skin cell types including T cells. In total, 1,506 cells with an average of 1,260 unique genes per cell were successfully profiled and analyzed per cell. We detected 19,053 genes using a sequencing saturation of 88%. Based on their transcription level, 400 T cells [further classified according to their TCR expression profile into 184 SP αβ and 216 γδ T cells (100 thereof were DP αβγδ T cells and 116 SP γδ T cells)], 251 keratinocytes, 168 fibroblasts, 138 NK cells, 201 macrophages, 197 DCs and 151 erythrocytes were identified (Fig. 2B). Examination of the top cluster-specific genes in SP αβ and γδ T cells as well as DP αβγδ T cells, a recently described population (Reitermaier et al., 2021), revealed a differential distribution of CD3 subunit genes [CD3D (> 90%), CD3E and CD3G], CD4, CD8A and CD8B genes as well as genes characteristic for hematopoietic stem cells (CD7, CD34, and CD38), T cell precursors and naive T cells [CD2, CD62L (also known as SELL), and CCR7], and recent thymic emigrants (CD31, also known as PECAM1) (Fig. 2C,D).

Fig. 2.

scRNA-seq map of human fetal skin cells. (A) Outline of the workflow for single-cell transcriptome profiling of fetal skin biopsies, including dissociation, flow cytometric cell sorting and droplet-based scRNA-seq of cells. GEMs, gel beads-in-emulsion. (B) t-SNE clustering showing cell transcriptomes of two donors (20 and 22 weeks EGA). Each dot represents one cell (1506 in total). Cells are colored and grouped according to their transcription profile. (C,D) The intensity of purple color denotes the normalized level of CD3 subunit genes and hematopoietic as well as T cell precursor gene expression on t-SNE plots of three donors (17, 21 and 22 weeks EGA). Each point represents one cell. Violin plots showing the expression level and distribution of indicated genes.

Fig. 2.

scRNA-seq map of human fetal skin cells. (A) Outline of the workflow for single-cell transcriptome profiling of fetal skin biopsies, including dissociation, flow cytometric cell sorting and droplet-based scRNA-seq of cells. GEMs, gel beads-in-emulsion. (B) t-SNE clustering showing cell transcriptomes of two donors (20 and 22 weeks EGA). Each dot represents one cell (1506 in total). Cells are colored and grouped according to their transcription profile. (C,D) The intensity of purple color denotes the normalized level of CD3 subunit genes and hematopoietic as well as T cell precursor gene expression on t-SNE plots of three donors (17, 21 and 22 weeks EGA). Each point represents one cell. Violin plots showing the expression level and distribution of indicated genes.

T cell subsets – including naive, effector and memory – can be distinguished from one another via a combination of markers (CD4, CD8, CD45RA, CD45RO, CD62L, CCR7, etc.). A bi-dimensional heat map denotes high and low expression of selected surface markers on total CD3+ T cells and demonstrates a distinct marker expression profile of fetal and adult skin T cells as analyzed by flow cytometry (Fig. 4A). Confirming our previously reported results (Reitermaier et al., 2021), αβ and γδ T cell subsets in adult and fetal skin are differentially distributed (Fig. 4B). Furthermore, only fetal skin and intestine contains a T cell subset with a unique TCR co-expressing αβ and γδ chains (DP αβγδ T cells; Fig. 4B) (Reitermaier et al., 2021). Cell surface glycoproteins CD4 and CD8 serve as co-receptors with the TCR primarily for the interaction with the major histocompatibility complex (MHC) class II (MHC II) loaded with peptides derived from cytosolic proteins and MHC I with extracellular protein peptides, respectively. The percentage and expression level of CD4+ T cells was consistently, though not significantly, lower in fetal (75.8%±4.1%; n=8) compared with adult (78.3%±3.6%; n=8) skin. In contrast, repeatedly but not significantly more CD8+ T cells were present in fetal skin (26.5%±6.8%), with similar CD8 expression levels when compared with the CD8+ T cell population in adult skin (18.8%±1.7%) (Fig. 4A,C,F). The majority of T cells in adult skin had a memory phenotype (92.1%±2.6%) confirming reported results (Clark et al., 2006b), whereas only one-quarter of fetal skin T cells expressed CD45RO (23.3%±3.6%), thus extending results that were assessed by immunohistochemistry and recent (mass) cytometry analyses (Di Nuzzo et al., 2009; Reitermaier et al., 2021; Dhariwala et al., 2020) (Fig. 4A,C,F). Conversely, most fetal T cells were naive (67.1%±4.0%) compared with a small population of CD45RA+ T cells in adult skin (4.3%±1.6%) (Fig. 4A,C,F). The majority of resident CD45RO+ T cells in adult skin co-expressed the cutaneous lymphocyte antigen (CLA) (92.06%±2.1%), a skin lymphocyte homing receptor, whereas only a few CD45RA+ T cells (1.88%±0.516) were positive for this marker (Fig. 4A,D-F), corroborating previously reported findings in adults (Clark et al., 2006b). Small but distinct subsets of both CD45RO+CLA+ (7.9%±0.8%) and CD45RA+CLA+ (3.5±0.7%) T cells were identified in fetal skin (Fig. 4A,D-F). Of note, the minute population of CLA+ T cells in fetal skin precluded a more precise phenotyping. Further, we observed minor variabilities in the percentage of CLA+ cells between different fetal donors (Fig. 4D,E) even though they were of the same age (18 week EGA). We had similar findings in adult donors (36 years versus 38 years) (Fig. 4D,E).

Fig. 4.

T cells in fetal skin have a largely naive and proliferative phenotype. (A) Bi-clustering heat map exhibiting a T cell surface marker expression profile of freshly isolated total T cells of fetal (16-22 weeks EGA; n=9) and, for comparison, adult (30-50 years; n=5) human skin as analyzed by flow cytometry. Color scheme is based on marker expression ranging from 0% (blue) to 100% (yellow). (B,C) Representative contour plots illustrating freshly isolated CD3+ skin T cells expressing indicated markers in adult [44 years (B), 32 years (C)] and fetal [16 weeks EGA (B), 18 weeks EGA (C)] skin as analyzed by flow cytometry. (D) CLA expression on CD3+ T cells was analyzed by flow cytometry according to the isotype control in adult (36 years) and fetal (18 weeks EGA) skin. (E) Representative contour plots showing expression of CLA on memory and naïve (38 years versus 18 weeks EGA) T cell subsets in adult and fetal skin. (F) Bar graphs (mean±s.e.m.) revealing expression of indicated markers on CD3+ T cells in adult and fetal skin as analyzed by flow cytometry (n=8). Unpaired, two-tailed Student's t-test. ****P≤0.0005. ns, not significant.

Fig. 4.

T cells in fetal skin have a largely naive and proliferative phenotype. (A) Bi-clustering heat map exhibiting a T cell surface marker expression profile of freshly isolated total T cells of fetal (16-22 weeks EGA; n=9) and, for comparison, adult (30-50 years; n=5) human skin as analyzed by flow cytometry. Color scheme is based on marker expression ranging from 0% (blue) to 100% (yellow). (B,C) Representative contour plots illustrating freshly isolated CD3+ skin T cells expressing indicated markers in adult [44 years (B), 32 years (C)] and fetal [16 weeks EGA (B), 18 weeks EGA (C)] skin as analyzed by flow cytometry. (D) CLA expression on CD3+ T cells was analyzed by flow cytometry according to the isotype control in adult (36 years) and fetal (18 weeks EGA) skin. (E) Representative contour plots showing expression of CLA on memory and naïve (38 years versus 18 weeks EGA) T cell subsets in adult and fetal skin. (F) Bar graphs (mean±s.e.m.) revealing expression of indicated markers on CD3+ T cells in adult and fetal skin as analyzed by flow cytometry (n=8). Unpaired, two-tailed Student's t-test. ****P≤0.0005. ns, not significant.

Even though isolation and sorting of significant viable T cell numbers from freshly isolated fetal skin is possible, extensive studies remain challenging. Therefore, we aimed to set up a skin culture system suitable to expand viable T cells from fetal skin. To this end, we explored an ex vivo culture system previously used to expand T cells from both healthy and diseased adult human skin (Clark et al., 2006a), to investigate its applicability for fetal skin. In this setup, fresh fetal skin biopsies alongside adult skin controls were cut into small pieces and cultured on collagen-coated grids using two different culture media (RPMI plus serum and serum-free TexMACS) (Fig. 6A), in the presence and absence of IL2/15. T cells started to spill from the matrices on day 3 and proliferating clusters with an approximately similar size, both small and large in shape, were visible in cultures with fetal skin and adult control skin after 2 weeks (Fig. 6B, arrowheads). Without cytokines, only single T cells and markedly less proliferation was observed compared with cytokine-containing cultures, regardless of skin age and medium used (Fig. S3A,B). Expanded T cells from adult skin predominantly expressed CD45RO (Fig. 6C; Fig. S3B), whereas the majority of fetal skin T cells were positive for CD45RA (Fig. 6D; Fig. S3B). These data clearly demonstrate that the culture conditions do not facilitate the outgrowth of a particular T cell subset and rather reflect the situation before culture. Of note, in contrast to the small percentage of freshly isolated fetal T cells expressing CLA (Fig. 4), several expanded fetal T cells expressed this marker (Fig. 6D; Fig. S3B). Fibroblasts were regularly visible in cultures with RPMI upon 2-4 weeks, but scarcely in cultures with TexMACS (Fig. S3D). To determine signature cytokines for each T cell subset, supernatants collected at 1 and 4 weeks from fetal and adult skin cultures (TexMACS±IL2/15) were analyzed with a LEGENDplex bead array (Fig. 6E; Fig. S3C). TH1 and TH2 cytokines were not measurable in supernatants derived from fetal and adult skin explants and TexMACS medium only (Fig. S3C). In contrast, culture of skin biopsies with IL2/15 initiated secretion of TH1 and TH2-related cytokines but was more pronounced in supernatants from cultures with adult skin, particularly after 4 weeks of culture, whereas it appeared fairly unchanged at both time points in supernatants from fetal skin (Fig. 6E). Cytokines specific for TH9, TH17 and TREG cells could be observed in supernatants of both adult and fetal skin specimens cultured with or without cytokines after 1 week, but was more prominent in 4 week cultures of adult skin (Fig. 6E; Fig. S3C). Of note, high levels of the suppressor cytokine IL10 were measured in supernatants of fetal skin cultures after 1 week, regardless of whether cytokines were present or not in the culture medium, but these decreased after 4 weeks (Fig. 6E; Fig. S3C). The IL10 decline with culture duration (Fig. 6E) correlated with a decrease of TREG cells (Fig. 6F).

Fig. 6.

T cells can be expanded from fetal skin in vitro. (A) Ex vivo skin T cell expansion workflow. (B) Representative T cell clusters (arrowheads), indicating proliferation, are visible upon culturing of fetal (n=5) and adult (n=8) skin specimens on collagen-treated grids in the presence of IL2/15 and either RPMI or TexMACS medium after 2 weeks. Upper panels, nearby grid; lower panels, grid area. Scale bar: 200 μm. (C,D) Bar graphs (mean±s.e.m.) showing numbers of expanded CD3+ T cells of fetal (n=5) and adult (n=8) skin specimens with indicated culture conditions upon 2 weeks and flow cytometry analyses. (E) Heat map of signature cytokines for each T cell subset identified in supernatants of fetal and adult skin specimen cultures (TexMACS and IL2/15) and measured with a cytometric bead assay at 1 and 4 weeks (n=5/age group). The color bar demonstrates the z-score. (F) A significant decrease of TREG cells in fetal skin cell cultures with TexMACS medium and IL2/15 was observed after 2 weeks (n=7). Unpaired, two-tailed Student's t-test. **P≤0.006, ***P≤0.0004, ****P≤0.0001. nd, not detected; ns, not significant.

Fig. 6.

T cells can be expanded from fetal skin in vitro. (A) Ex vivo skin T cell expansion workflow. (B) Representative T cell clusters (arrowheads), indicating proliferation, are visible upon culturing of fetal (n=5) and adult (n=8) skin specimens on collagen-treated grids in the presence of IL2/15 and either RPMI or TexMACS medium after 2 weeks. Upper panels, nearby grid; lower panels, grid area. Scale bar: 200 μm. (C,D) Bar graphs (mean±s.e.m.) showing numbers of expanded CD3+ T cells of fetal (n=5) and adult (n=8) skin specimens with indicated culture conditions upon 2 weeks and flow cytometry analyses. (E) Heat map of signature cytokines for each T cell subset identified in supernatants of fetal and adult skin specimen cultures (TexMACS and IL2/15) and measured with a cytometric bead assay at 1 and 4 weeks (n=5/age group). The color bar demonstrates the z-score. (F) A significant decrease of TREG cells in fetal skin cell cultures with TexMACS medium and IL2/15 was observed after 2 weeks (n=7). Unpaired, two-tailed Student's t-test. **P≤0.006, ***P≤0.0004, ****P≤0.0001. nd, not detected; ns, not significant.

In this study, we used a combination of advanced approaches to further explore the complexity and heterogeneity of T cell subsets in fetal human skin. We describe methods for the isolation and expansion of rare fetal skin T cells. Further, single-cell transcriptomics and TCR repertoire profiling mapped the molecular state of fetal skin T cells, and multiparametric flow cytometry as well as in situ immunofluorescence analyses were integrated to validate the molecular-based profile, thus contributing to a better understanding of the particularities of fetal skin immunity.

The isolation of T cells from small fetal skin biopsies in sufficient numbers and with robust and reliable quality was a challenging first step towards studying their nature and function. To address this, we comparatively assessed T cell yields from fetal and adult human skin biopsies using several isolation techniques, identifying a combination of automated and enzymatic dissociation as the most efficient and reproducible procedure. Importantly, this methodology preserved pan surface markers such as CD3, CD4, CD8, CD45RA and CD45RO that are necessary for the phenotypic characterization of T cell subsets and will be also a useful tool for studying skin biopsies with limited patient material in the future.

Recognized for its high sensitivity (Zhao et al., 2018), we have used 10x Genomics for single-cell transcriptome profiling of fetal skin to resolve the cellular heterogeneity. The major cell types described in adult skin (Rojahn et al., 2020), were also identified during early development, although differing in relative abundance. Of note, other groups have identified B cells, precursors and other immune populations in human fetal skin (Botting and Haniffa, 2020; Dhariwala et al., 2020; Popescu et al., 2019), which were absent in our samples most likely due to the fact that the majority of cells were sorted CD3+ T cells that were mixed with total skin cells as our primary goal was to characterize T cells. We reported previously that beside conventional SP αβ and γδ T cells, a DP αβγδ fetal skin T cell population is unique to the early fetal period and is absent in the skin at the time of birth and in healthy adults (Reitermaier et al., 2021). All three T cell subsets showed gene expression for CD3 subunits, classical hematopoietic stem cells and T cell precursors. Multiparametric flow cytometry and in situ immunofluorescence analyses validated and expanded the transcriptome-based profile, and revealed that the majority of fetal skin T cells were positive for CD4, expressed markers that are characteristic for naive T cells (CD45RA, CD62L, CCR7) as well as hematopoietic stem cells (CD34 and CD38). Observations in sheep suggested a pathway of recirculating naive T cells within fetal skin to establish tolerance to self-antigens (Cahill et al., 1999). In support of this are published data that show CD45RA+CD8+CD62L−CCR7− T cells expressing the skin homing marker CLA in human cord blood lymphocytes (Zippelius et al., 2004). Although we observed a small CD45RA+CLA+ T cell population in fetal skin (Fig. 4B-D), it remains to be investigated whether they express CD4, CD8, CD62L and/or CCR7. We have previously reported that more than two-thirds of all naive T cells in fetal skin express CD31, indicative for recent thymic emigrants (Reitermaier et al., 2021), which is in line with a recent study reporting on elevated CD31 expression in CD45RO− fetal skin T cells (Dhariwala et al., 2020). Although our observation that DP αβγδ fetal skin T cells expressing CD31 are undetectable in the thymus suggest their extrathymic development (Reitermaier et al., 2021), the derivation of CD31+SP αβ and γδ T cells needs to be further explored.

Adult human skin is protected by two discrete populations of resident memory T cells and two distinct populations (CCR7+CD62L+ and CCR7+CD62L−) of recirculating skin-tropic (CLA+) T cells, each with different functional capacities (Clark et al., 2006b; Watanabe et al., 2015). Strikingly, fetal skin also harbors CLA+ memory T cells (Di Nuzzo et al., 2009) (Fig. 4B-D) but, in contrast to adult skin, in the absence of a reported pathology or any major infectious history. Their minute frequency correlates with the observation of low levels of the T cell-attracting chemokine CCL27 (also known as CTACK) (Morales et al., 1999) in fetal compared with adult (Schuster et al., 2012; Mildner et al., 2014) skin precluding an influx of memory T cells, if present, from the circulation (Zhang et al., 2014). Keratinocytes fail to produce CCL27 correlating with its negative staining in the first trimester. Strong specific CCL27 expression was observed only in the stratum corneum towards the end of the second trimester, indicating a differentiation-dependent regulation. These observations implied that seeding of fetal skin with naive T cells occurs independently of CCL27 (Schuster et al., 2012). The structural and functional immaturity of the epidermis in early fetal development may also explain why neither naive nor memory T cells have been identified in this compartment (Schuster et al., 2012; Di Nuzzo et al., 2009; Reitermaier et al., 2021; Dhariwala et al., 2020). Changes in the microenvironment and the keratinocyte differentiation process with increasing CCL27 levels in keratinocytes seem to favor a gradient of T cell influx, as naive and memory T cells were identified not only in the dermis but also in the epidermis after birth (Akgün et al., 2014). Of note, also in adult skin CCR7+CD62L−CLA+ migratory memory T cells were reported to be confined to the dermis and absent from the epidermis (Watanabe et al., 2015).

T cells are defined by their TCR sequences, facilitating the accomplishment of highly specific TCR-dependent antigen recognition. The antigen recognition triggers downstream signaling of T cells – a crucial biological process (Jung and Alt, 2004). The TCR repertoire hence represents a ‘footprint’ of the conditions faced by T cells that dynamically evolves according to the challenges that arise for the immune system. Consequently, profiling the TCR repertoire was of interest in our study. Unexpectedly, a relatively consistent TCR Vβ family repertoire was found, with small variations in the individual Vβ segments in fetal skin T cells and was comparable with T cells in adult skin. As our data with adult skin are in line with those obtained in a previous study, we furthermore demonstrate the robustness and reproducibility of this assay (Clark et al., 2012). Of note, an increase (5.2, 5.3, 7.1, 9, 17, 18 and 23) and a decrease (2, 3, 5.1, 13.1, 13.2, 16, 20 and 22) of particular Vβ families was observed in fetal compared with adult skin. However, further work is needed to understand this distinct expression profile. High-throughput sequencing of the CDR3 region indicated the occurrence of a diverse skin TCR repertoire, and an analogous distribution in fetal skin T cells between four unrelated donors. In addition, assessments of the rearrangement in the same donors showed quite similar TCR Vβ and TCR Jβ combinations. This is in contrast to previous studies reporting a skewed usage of TCR Vβ families during development (Rechavi et al., 2015; Carey et al., 2017). This discrepancy could be explained using different techniques and tissues. Together, our analyses are the first to unravel the TCR repertoire of fetal skin T cells. We are aware that the unknown functional relevance of TCR profiling hinders unbiased interpretation of the biology of T cells. Most recently, a tool (tessa) has been developed enabling mapping of the functional landscape of the TCR repertoires by combining scRNA-seq with TCR sequencing (Zhang et al., 2021). Its application could allow answer a variety of research questions regarding biology of fetal skin T cells in the future.

Given the limitations to obtain sufficient T cell numbers from fetal skin, we aimed to expand T cells from fetal skin ex vivo. We embarked on a method established for the expansion of T cells from adult human skin that has greatly facilitated studies of this important population (Clark et al., 2006a). Although both media (RPMI, TexMACS) favored the expansion of fetal skin T cells, we observed significantly better cell yields using TexMACS. Although T cell numbers isolated from different fetal donors varied, the kinetics of T cell expansion was comparable for all donors and, most importantly, maintained their naive phenotype. Of note, about 50% of expanded fetal T cells, irrespective of whether cultured with or without serum, expressed CLA. This is in contrast to the small population expressing this molecule at culture initiation and to a previous report showing that optimal CLA induction is only possible in the absence of serum (Armerding and Kupper, 1999). Our data may suggest an expansion of CLA+ T cells rather than its induction due to culture conditions and remains to be further investigated. Together, expanded T cells essentially maintain the properties they have in utero and can be used as a model for researchers studying basic biological processes.

Regarding signature cytokines for T cell subsets, we identified a shift towards TH1, TH2 and TH9 subsets in supernatants of adult but not fetal skin cultures. The high IL9 levels after 4 weeks of adult skin culture are either due to higher numbers of TH9 cells or, alternatively, due to the proliferation of other IL9-producing T cell subsets such as TH2 and/or TH17 cells which cannot be distinguished in fetal skin (Schlapbach et al., 2014). Of note, the higher levels of the cytokine IL22 in fetal cultures, irrespective of culture conditions and duration, suggest that SP γδ and/or DP αβγδ T cells may be the cellular source (Reitermaier et al., 2021; Mielke et al., 2013).

Today we know that the immune system is not fundamentally immature in early life, but simply differs from immune responses observed in later life. Recent studies indicate that a predisposition of the fetal immune system toward tolerance is assignable to both lymphocyte intrinsic and dendritic cell-dependent features (McGovern et al., 2017; Mold and McCune, 2012), which are involved to completely avoid or regulate responses to self, maternal or foreign antigens in utero. We have reported previously that IL10 concentrations in fetal skin single-cel