If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Correspondence: Axel Roers, Institute for Immunology, Faculty of Medicine Carl Gustav Carus, Dresden University of Technology, Fetscherstrasse 74, 01307 Dresden, Germany.
Loss of FLG causes ichthyosis vulgaris. Reduced FLG expression compromises epidermal barrier function and is associated with atopic dermatitis, allergy, and asthma. The flaky tail mouse harbors two mutations that affect the skin barrier, Flgft, resulting in hypomorphic FLG expression, and Tmem79ma, inactivating TMEM79. Mice defective only for TMEM79 featured dermatitis and systemic atopy, but also Flgft/ft BALB/c congenic mice developed eczema, high IgE, and spontaneous asthma, suggesting that FLG protects from atopy. In contrast, a targeted Flg-knockout mutation backcrossed to BALB/c did not result in dermatitis or atopy. To resolve this discrepancy, we generated FLG-deficient mice on pure BALB/c background by inactivating Flg in BALB/c embryos. These mice feature an ichthyosis phenotype, barrier defect, and facilitated percutaneous sensitization. However, they do not develop dermatitis or atopy. Whole-genome sequencing of the atopic Flgft BALB/c congenics revealed that they were homozygous for the atopy-causing Tmem79matted mutation. In summary, we show that FLG deficiency does not cause atopy in mice, in line with lack of atopic disease in a fraction of patients with ichthyosis vulgaris carrying two Flg null alleles. However, the absence of FLG likely promotes and modulates dermatitis caused by other genetic barrier defects.
Over the past two decades, structural defects of the epidermal skin barrier emerged as a cause of immune dysregulation, leading not only to dermatitis but also to allergy and asthma (
). Barrier-defective skin allows abnormal penetration of antigen, which is then presented in the context of an abnormal immune milieu resulting from innate immune responses of stressed skin cells (
Standardized tape stripping prior to patch testing induces upregulation of Hsp90, Hsp70, IL-33, TNF-α and IL-8/CXCL8 mRNA: new insights into the involvement of 'alarmins'.
). While differentiating into corneocytes, KCs express filament aggregating protein (FLG). Full-length pro-FLG is proteolytically cleaved into FLG monomers that bind to and bundle keratin filaments, thereby compacting the cell (
). Conversely, sequencing of the FLG gene in patients with asthma and AD revealed one or two null alleles in 23% of cases, establishing genetic FLG deficiency as a major factor predisposing for the atopy associated with AD (
A homozygous nonsense mutation in the gene for Tmem79, a component for the lamellar granule secretory system, produces spontaneous eczema in an experimental model of atopic dermatitis.
A homozygous nonsense mutation in the gene for Tmem79, a component for the lamellar granule secretory system, produces spontaneous eczema in an experimental model of atopic dermatitis.
). The Tmem79ma mutation was separated from the Flgft allele in two independent studies addressing the individual contributions of the two genetic defects to the atopic phenotype. Both found that isolated loss of TMEM79 results in eczema and increased IgE (
A homozygous nonsense mutation in the gene for Tmem79, a component for the lamellar granule secretory system, produces spontaneous eczema in an experimental model of atopic dermatitis.
Targeted complete inactivation of Flg in C57BL/6 mice resulted in the IV-like phenotype, as expected, but resulted in no atopy even after backcrossing to BALB/c mice (
). In striking contrast, Flgft/ft BALB/c congenic mice showed skin inflammation with important features of AD, including eosinophilia and accumulation of type-2 innate lymphoid cells (ILC2s), high IgE, and spontaneous asthma (
). These findings established Flgft/ft BALB/c congenic mice as a highly relevant model for the human atopy associated with skin barrier defects.
The human FLG and mouse Flg genes are encoded within a large cluster of genes with roles in skin barrier function, the epidermal differentiation cluster, and are in strong linkage disequilibrium with the entire epidermal differentiation cluster. To determine whether genetic variability within the epidermal differentiation cluster and adjacent regions may account for the phenotypic difference between the Flgft/ft BALB/c congenic (
) mice, we generated FLG-deficient mice on a pure BALB/c background and analyzed them for skin phenotype and atopy. We also performed whole-genome sequencing for comparison of the Flgft/ft BALB/c congenics with our new BALB/c Flg−/− line.
Results
Generation of BALB/c Flg−/− mice
For inactivation of the Flg gene, two guide RNA sequences were chosen. Guide 1 targeted the upstream, nonrepetitive portion of exon 3, whereas guide 2 bound each of the repeat regions of exon 3 (Figure 1a). The guides were microinjected into BALB/c zygotes, and PCR identified three mutant alleles that harbored large deletions of the upstream nonrepetitive region plus at least one repeat (Supplementary Figure S1b and c). The mutations were bred to homozygosity, and total tail and ear skin samples from mice aged 8 weeks were analyzed by western blot, demonstrating an absence of both multimeric and monomeric FLG (Figure 1b). This result was confirmed by immunofluorescence analysis (Figure 1c). Mice homozygous for allele 1 (Supplementary Figure S1b) were used in subsequent experiments. Long-Read Single Molecule Real-Time Sequencing (Pacific Biosciences of California, Menlo Park, CA) revealed that this line carried a large (8,122 base pair) out-of-frame deletion (Figure 1a). In summary, we generated FLG-deficient mice on a pure BALB/c background.
Figure 1Generation of Flg−/− mice on a pure BALB/c background.(a) Strategy for CRISPR/Cas9-mediated targeting of the Flg gene. (b) Expression of FLG in ear skin as determined by western blot analysis of ear skin from mice aged 8 weeks. The expected size for full-length pro-FLG is approximately 500 kDa and approximately 30 kDa for the FLG monomer. Representative of eight mice. (c) Immunofluorescence analysis of FLG expression (green) in tail skin of ctrl (Flgwt/-) and Flg−/− mice aged 8 weeks. Nuclei were counterstained with DAPI (red). Representative images from four mice of each genotype. Dotted lines mark the basement membrane and the border of epidermis to cornified layer. Bar = 100 μM. β-Act, β-actin; bp, base pair; ctrl, control.
Although normal immediately after birth, homozygous Flg-deficient pups showed dry and scaly skin, hyperlinearity, and annular constrictions of tail skin a few days later (Figure 2a and b ). This phenotype ameliorated after few weeks, and the skin appeared macroscopically normal by age 4 weeks. There was no evidence of pruritus, and the fur of adult mice was normal. Flg−/− ears were smaller (not shown) and thicker (Figure 2c) as had been described for the flaky tail strain (
). H&E staining of skin sections revealed a marked reduction of keratohyalin granules in the upper epidermis of the Flg−/− mice (Figure 2a), consistent with lack of FLG as their major constituent (
). Except for the loss of these granules, Flg−/− skin appeared normal with no evidence of inflammation. Transepidermal water loss (TEWL) was unchanged compared with that of the controls in younger Flg−/− mice (aged 8–10 weeks) but was increased in older FLG-deficient (aged 16–18 weeks) mice compared with that in age-matched controls (Figure 2d).
Figure 2Skin phenotype of BALB/c Flg−/− mice.(a) Representative macroscopic and microscopic images of neonatal (d4) BALB/c ctrl and Flg−/− ear skin. Lower panels show the H&E staining of the epidermis. Insets from larger images, shown in Supplementary Figure S4. Arrowheads denote keratohyalin granules of ctrl skin. Images are representative of at least 10 mice. Bar = 100 μM. (b) Representative macroscopic images of BALB/c ctrl and Flg−/− tails mice aged 3 weeks. (c) Ear thickness of BALB/c littermate ctrl (n = 14) and Flg−/− (n = 13) mice aged 8 weeks, both male and female groups. Means ± SD, ∗∗∗∗P0.0001. (d) TEWL of shaved BALB/c ctrl and Flg−/− mice back skin. Means ± SD, ∗∗∗P0.001. ctrl, control; d, day; ns, not significant; TEWL, transepidermal water loss.
Collectively, we find that loss of FLG on a pure BALB/c background results in dry and scaly skin in neonates. In adult mice, skin and fur are macroscopically normal. Increased TEWL occurs in older but not in younger BALB/c Flg−/− mice, indicative of a skin barrier defect.
BALB/c Flg−/− mice skin features mild immune activation and altered microbiome
To test whether the loss of FLG in the suprabasal epidermis leads to detectable responses of the entire skin or specifically of cells of lower epidermal strata, we performed RNA sequencing of total skin and of FACS-sorted CD45-negative ITGA6 (CD49f)+ basal KCs of young adult control and BALB/c Flg−/− mice (Supplementary Figure S2c).
In accordance with the moderate changes observed macroscopically and histologically, only mild transcriptional changes were observed in female mice, whereas no significant changes occurred in male BALB/c Flg−/− mice (Figure 3a). Total skin transcriptomes of mutant mice showed significant enrichment of gene sets associated with skin stress responses (Supplementary Figure S2b), including UVB response, mTORC1 signaling, and MYC signaling (
). The FLG-deficient KCs showed enrichment of stress response gene sets as well (UVB response) but also showed an enrichment of inflammation-related genes (Figure 3a and Supplementary Figure S2d). For example, several genes encoding the components of the major histocompatibility complex II antigen presentation machinery, including H2-Aa, H2-Ab1, H2-Ea, CD74, and H2-DMb1, were upregulated (Supplementary Figure S2e), although individual genes (Cd74 and H2-Aa) did not reach significance in quantitative real-time reverse transcriptase–PCR analysis (Figure 3d). Flow cytometric analysis of skin cell suspensions showed a doubling of major histocompatibility complex II‒expressing KCs in female mice (Figure 3e). Patches of brightly major histocompatibility complex II‒positive KCs were recently identified as critical structures in antimicrobial T-cell responses in the skin (
Figure 3Alteration of gene expression and composition of the microbiota of Flg−/− skin.(a) CD49f+ keratinocyte RNASeq (Supplementary Figure S2), 286 significantly deregulated transcripts in Flg−/− mice versus those in the ctrls (black dots). Female Flg−/− mice (n = 4) and ctrls (n = 5) aged 8–12 weeks. (b) Significant enrichment of STAT6-regulated genes in Flg−/− keratinocytes (GSEA). (c) Enrichment of type-2 immune response gene set in Flg−/− keratinocytes (GSEA). (d) QRT-PCR for selected genes on sorted keratinocyte RNA. Fold change for individual female Flg−/− mice relative to that of the mean of six ctrls. (e) Frequency of CD45− epidermal cells expressing MHC II (flow cytometry) in Flg−/− and ctrl females aged 8–10 weeks; n = 8 both groups. (f) Relative fold change of Il1β mRNA in total ear skin RNA from Flg−/− mice aged 8 weeks and 20 weeks versus that from the ctrls; n = 8 in both groups. (g) Skin microbiome Shannon diversity index of female Flg−/− mice aged 8 weeks and the ctrls. (h) Phylogeny and abundance of taxa significantly depleted in Flg−/− mice. Heatmap displays log2 fold change of each ASVs. Means ± SD, ∗P 0.05, ∗∗P < 0.01. ASV, amplicon sequence variant; ctrl, control; ES, enrichment score; FDR, false discovery rate; GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; MHC, major histocompatibility complex; NES, normalized enrichment score; NOM P, nominal P-value; ns, not significant; Padj, adjusted P-value; QRT-PCR, quantitative real-time reverse transcriptase–PCR; STAT, signal transducer and activator of transcription; WT, wild type.
Interestingly, sets of signal transducer and activator of transcription 6‒inducible genes were clearly enriched in the mutant KC transcriptomes (Figure 3b) as was the gene set type-2 immune response (Figure 3c). Among the upregulated type-2 genes was Retnla, encoding RELMα (Figure 3a and d and Supplementary Figure S2e). RELMα is a marker of M2 macrophage differentiation and was also shown to function as an important KC antibacterial effector protein (
). These findings indicate a spontaneous low-level type-2 immune response, activated by the hallmark type-2 cytokines IL-4 and IL-13 that signal through signal transducer and activator of transcription 6 downstream of the IL-4R.
Quantitative real-time reverse transcriptase–PCR analysis of total ear skin revealed that whereas IL1β mRNA levels were unchanged in young adults, they were clearly increased in mice aged 20 weeks (Figure 3f), suggesting that spontaneous immune activation in the skin increases with age, a finding well in line with our observation of increased water loss in older but not in young mice.
To determine whether the absence of FLG affects the skin microbiome, we swabbed the skin of female BALB/c Flg−/− mice aged 8 weeks and that of the control mice cohoused in the same cages and performed 16s ribosomal RNA sequencing. As expected for mouse skin, Bacteroidetes, Proteobacteria, and Firmicutes were the dominant phyla without major changes in their abundance in mutants versus in controls (Supplementary Figure S2f). BALB/c Flg−/− mice showed a reduction of microbial diversity compared with that in control mice as indicated by Shannon diversity index (Figure 3g). Several taxa of the families Lachnospiraceae or Muribaculaceae (previously known as S24-7) (
) were dramatically reduced in Flg−/− versus in control mice (Figure 3h). In addition, we observed significant reductions of a few specific species of the Ruminococcaceae, Lactobacillaceae, and Rikenellaceae families. The alterations of skin microbiome observed in Flgft/ft BALB/c congenics were clearly more profound with a striking over-representation of Firmicutes than observed in these results.
Collectively, we find that loss of FLG in mice results in mild immune activation with spontaneous activation of type-2 immunity in the skin. The mutant skin features reduced diversity of skin microbiota, and local immune activation seems to increase with age.
BALB/c Flg−/− mice skin shows slight increase in T-cell compartment but otherwise unchanged immune cell populations
We quantified the immune cells in BALB/c Flg−/− mice skin by flow cytometric analysis of skin cell suspensions. Because we observed an increase in barrier defect and Il1β transcript levels with age in FLG-deficient mice (Figures 2d and 3f), we included young (aged 8 weeks) as well as older (aged ≥16 weeks) mice. Importantly, CD45+ hematopoietic cells were not increased in relation to nonhematopoietic skin cells (mostly KCs) in neither young nor older mice, confirming the absence of inflammatory infiltration observed histologically (Figure 4a). Whereas Flgft/ft BALB/c congenics feature larger eosinophil, mast cell, and ILC2 populations (
), we did not observe any increase in eosinophils, mast cells, neutrophils, or ILC2s (gating strategy in Supplementary Figure S3) in BALB/c Flg−/− mice versus those in the control mice at any age (Figure 4b and g). Likewise, dendritic cell, macrophage, and Langerhans cell populations were unchanged; similarly, CD86 expression by skin CD11c+ cells was not altered (Figure 4c, d, e, and i). We observed a moderate increase in T cells, which was most likely accounted for by the expansion of the γδ T-cell population (Figure 4f and h) that we consistently observed in young and older mice.
Figure 4Flow cytometric analysis of immune cells in skin cell suspensions from Flg−/− and ctrl female mice aged 8 weeks (young) and 16+ weeks (old) (n = 5–6 in both groups).(a) Fraction of CD45+ cells in viable ear skin cells. (b) Fractions of CD11b+SiglecF+ Eos, CD11b+Gr-1hi Neutr, and cKit+FceRI+ MCs among CD45+ cells. (c) Fraction of CD11c+CD64− DCs and CD11c+CD11b+F4/80+CD64+ Mϕs among CD45+ cells. (d) Mean CD86 expression on CD45+CD11c+ cells. (e) Fraction of XCR1+ cDC1 and SIRPα+ cDC2 cells among CD11c+CD64− DCs. (f) Fractions of total CD3+ T cells and γδ T cells among CD45+ cells. (g) Fraction of CD3−Thy1+IL7Ra+ICOS+CD25+ ILC2s among CD45+ cells. (h) Fractions of CD4+ and CD8+ T cells among CD45+ cells. (i) Fraction of EpCAM+ LCs among CD45+ epidermal cells. Cell suspensions were generated by digestion of epidermal sheets. Means ± SD, ∗P0.05. cDC, conventional dendritic cell; ctrl, control; DC, dendritic cell; Eo, eosinophil; ILC2, type-2 innate lymphoid cell; LC, Langerhans cell; MC, mast cell; MFI, mean fluorescent intensity; Mϕ, macrophage; Neutr, neutrophil; ns, not significant.
In summary, FLG-deficient mice on pure BALB/c background do not feature increased numbers of total immune cells in the skin. Although we detected some increase in skin T-cell numbers, these mice do not display the key inflammatory changes reported for young Flgft/ft BALB/c congenics, in particular the accumulations of ILC2s, eosinophils, and neutrophils (
BALB/c Flg−/− mice did not show overt skin inflammation at any time point (Figure 2a and Supplementary Figure S4). To test the Flg mutants for systemic atopy, total serum IgE was quantified. In contrast to IgE levels in Flgft/ft BALB/c congenics (
), IgE levels of our BALB/c Flg−/− mice at ages 8, 12, and 25 weeks did not differ significantly from the levels in the controls (Figure 5a). Because the microbiome is an important trigger factor for AD with important roles for Staphylococcus aureus (reviewed in
), we tested whether colonization with S. aureus would result in skin inflammation and systemic atopy. Flg−/− mice were painted on the shaved trunk (five times at 2-day intervals) with a suspension of an S. aureus CC1 isolated from a patient with AD and were monitored for 7 weeks thereafter; however, no overt skin inflammation (not shown) or increase in total IgE (Figure 5b) were observed. Because skin colonization can be impeded by resident flora, we repeated the colonization with previous systemic antibiotic treatment, but this regime also did not trigger dermatitis or enhanced IgE production (Figure 5b). To test whether the altered cutaneous immune milieu of Flg−/− mice (Figure 3a–d) would drive abnormal IgE responses, we epicutaneously immunized BALB/c Flg−/− mice with ovalbumin (OVA). OVA-specific IgE was not significantly different between mutant and control littermates 7 weeks later (Figure 5c). We next administered OVA to the skin of mice, Flg−/− and control, that had received an adoptive transfer of OVA-specific DO11.10 transgenic (
). Because our finding that TEWL increased in older (aged 16–18 weeks) but not in younger mice (aged 8–10 weeks) indicated that the barrier defect develops with age, we included mice of both age groups. Application of OVA to the skin of DO11.10 T-cell recipients did not cause any skin alteration in wild-type or in younger Flg−/− recipients (not shown), whereas marked skin inflammation was elicited in Flg-deficient‒recipients mice aged 16 weeks (Figure 5d). To further investigate the differences in the development of skin inflammation in DO11.10 cell‒recipient FLG-deficient mice but not in wild-type‒recipient mice, we assessed DO11.10 cellular responses in the skin draining lymph nodes (Figure 5e). In the lymph nodes of Flg-deficient mice but not in the control recipient wild-type mice, there was vigorous cell proliferation with elevated frequencies of IL-4-eGFP+ cells (gating strategy in Supplementary Figure S5) indicating enhanced OVA-specific T helper type 2 cellular response in the skin draining lymph node of Flg−/− mice (Figure 5e).
Figure 5Loss of FLG in mice does not result in atopy.(a) Serum IgE of Flg−/− (n = 5–8) and ctrl mice (n = 5) at indicated age. (b) Serum IgE of Flg−/− (n = 5–8) and ctrl (n = 5–6) mice aged 14–16 weeks colonized with Staphylococcus aureus ± ABX. (c) OVA-specific IgE of Flg−/− mice (n = 14) and ctrls (n = 7) aged 16 weeks treated with OVA. (d–e) DO11.10 transfer experiment. (d) Left: treated skin representative images on d5; right: dermatitis score (Supplementary Figure S5), n = 5–6. (e) Proliferative response and IL-4 expression of DO11.10 cells. Left: Frequency and # prolif. transgenic cells (gating, Supplementary Figure S5), right: Frequency and number of (#) IL-4 reporter‒expressing cells. n = 5 in both groups and 2 LNs per mouse. Data are representative of two experiments. Represented means ± SD. ∗P < 0.05, ∗∗P < 0.01. ABX, antibiotic pretreatment; ctrl, control; d, day; LN, lymph node; ns, not significant; O.D., optical density; OVA, ovalbumin; prolif., proliferating.
In summary, we show that loss of FLG in mice of pure BALB/c background does not result in increased total serum IgE, neither spontaneously nor after colonization with S. aureus. In addition, epicutaneous immunization did not elicit more specific IgE in BALB/c Flg−/− mice than in control mice but did induce enhanced T-cell proliferation and skin inflammation in the recipients of antigen-specific transgenic T cells.
Accurate whole-genome sequencing reveals that Flgft/ft BALB/c congenic mice are in fact Flgft/ftTmem79ma/ma double mutants, explaining their atopic phenotype
Our finding that complete loss of FLG in mice of pure BALB/c background did not result in detectable atopy, contrasting with the robust atopy of Flgft/ft BALB/c congenics (
), suggested that further genetic variation within the congenic interval in addition to the Flg loss-of-function mutation is essential for the development of atopy in FLG-deficient mice. We therefore generated a highly accurate whole-genome sequence of the Flgft/ft BALB/c congenic strain and of our newly generated BALB/c Flg−/−- mice using Pacific Biosciences of California Long-Read Single Molecule Real-Time Sequencing (
). Chromosome 3 alignments were used for variant analysis. In the Flgft/ft BALB/c congenic sequence, we identified the Flg 5303delA mutation, as expected, and the sequence of the new BALB/c Flg−/− line revealed the size of the deletion we had introduced by CRISPR/Cas9-mediated mutagenesis (Figure 1). Alignment of the new chromosome 3 sequence of Flgft/ft BALB/c congenic and BALB/c Flg−/− mice showed that chromosome 3 (and all other chromosomes, not shown) of the two strains were nearly identical to those of our BALB/c Flg−/− de novo assembly but differed extensively within a 51.3 megabase congenic interval, which was in roughly equal parts derived from C3H and from C57BL/6 in the Flgft/ft BALB/c congenic strain (Figure 6), reflecting its history of backcrossing first to C57BL/6 (
). Within the congenic interval, we identified 610 genes with functional annotation, 12 of which differed by deleterious mutations between the Flgft/ft BALB/c congenic and the BALB/c Flg−/− strain, whereas 143 genes differed by amino acid exchange mutations predicted to potentially impact structure or function (Supplementary Table S1).
Figure 6Presence of the Tmem79ma gene variant in the Flgft/ft BALB/c congenic strain. Chromosome 3 sequence representation for Flgft/ft BALB/c congenic strain (
) and in our BALB/c Flg−/− strain. The average difference between the two strains was 1 variation per 9,694 bp outside of the congenic interval (QV score of 39.9) and 1 variation per 145 bp within the interval (QV score of 21.6). Flg gene of the Flgft/ft BALB/c congenic strain (
) harbors the Flgft mutation (1 bp deletion in Flg repeat 6), whereas our BALB/c Flg−/− strain carries an 8,122 bp (out-of-frame) deletion in Flg exon 3. The Flgft/ft BALB/c congenic strain is homozygous for the atopy-associated Tmem79ma mutation, reported to be crossed out before initiation of backcrossing to BALB/c (
) in the CH3-derived interval of the Flgft/ft BALB/c congenics. The Flgft mutation originally arose on a strain (partially C3H) with matted hair to yield the flaky tail mouse (
identified the Y280Stop mutation of Tmem79, a gene linked with Flg on chromosome 3, to be responsible for the matted phenotype. To separate the Flgft and the Tmem79ma mutations,
also identified a human TMEM79 missense SNP associated with human AD. Flgft/ft C57BL/6 mice (assumed to be wild type for Tmem79) were backcrossed extensively to BALB/c to yield the Flgft/ft BALB/c congenic strain that is characterized by a prominent atopic phenotype, including eczema, high IgE, and spontaneous asthma (
). However, our sequencing result shows that the Flgft/ft BALB/c congenic strain is homozygous for the Tmem79ma mutation, indicating that either the two mutations had not been successfully separated, which is unlikely given the clear sequencing results and absence of atopy in the initial B6 Flgft/ft mice (
), or that the Tmem79maFlgft haplotype was accidentally reintroduced during the subsequent backcrossing to BALB/c. Genotyping of the current Flgft/ft BALB/c congenic mice for the Tmem79ma and Flgft mutations confirmed our sequencing data, with both Tmem79 and Flg mutations detected.
Collectively, we show that the decisive difference between nonatopic BALB/c Flg−/− mice and the atopic Flgft/ft BALB/c congenic strain is the presence of the atopy-causing Tmem79ma mutation in the latter.
Discussion
To resolve the unexplained discrepancy between prominently atopic FLG-deficient Flgft BALB/c congenic mice (
) and to determine the effects of FLG deficiency on skin immune regulation, we generated FLG-deficient mice on a pure BALB/c background by inactivating the Flg gene in BALB/c embryos. These mice do not develop systemic atopy. We sequenced the genome of the atopic Flgft BALB/c congenics and discovered that these mice harbor the atopy-causing matted Tmem79ma mutation. These findings shed light on the hitherto puzzling and conflicting published results on skin barrier‒defective mouse models.
In accordance with the phenotype of Flg-knockout mice (
), our BALB/c Flg−/− strain shows that in mice, the loss of FLG results in dry and scaly skin only postnatally, whereas skin and fur are macroscopically normal throughout adult life. Although the mutant skin featured a barrier defect with increased TEWL in old (but not in young) mice, histologically, a substantial reduction of keratohyalin granules is the only prominent change. Loss of FLG alone does not result in the spontaneous development of overt inflammation.
However, FLG deficiency is clearly not immunologically inert. Slight alterations of gene expression in BALB/c Flg−/− skin gene expression suggested mild stress responses and immune activation also reflected in a moderate expansion of T cells. However, the marked increases in ILC2s, eosinophils, and mast cells that characterize the Tmem79ma/maFlgft/ft BALB/c congenics (
The microbiome of BALB/c Flg−/− mice skin clearly differs from that of the control mice skin; however, these changes are less pronounced and are different from the gross skin microbiome alteration in Tmem79ma/maFlgft/ft BALB/c congenics (
). The effects of FLG deficiency on skin microbiota may either be a direct consequence of the lack of FLG and the resulting changes of skin pH, moisture, and turnover or texture of the cornified layer or could be a result of the subtle immune activation in the BALB/c Flg−/− skin. In contrast, the dysbiosis we found in BALB/c Flg−/− skin may also be the trigger of the mild stress response and immune activation, including increased expression of RELMα in these mice. We speculate that RELMα may contribute to the innate antimicrobial defense of the barrier-defective skin (
To test whether the barrier defect correlated with enhanced protein antigen penetration, we epicutaneously immunized FLG-deficient mice. No inflammation was observed in controls or in FLG-deficient mice aged 8 weeks, whereas Flg−/− mice aged 16 weeks developed eczema and enhanced OVA-specific T-cell responses, mirroring the development of the barrier defect with age, as determined by quantification of TEWL.
Importantly, we now show that the atopic phenotype of Flgft/ft BALB/c congenics (
) is likely caused by the Tmem79ma/ma mutation that is linked to the Flgft allele and was not known to be present in this strain. Mice with isolated TMEM79 deficiency develop a largely similar phenotype (
A homozygous nonsense mutation in the gene for Tmem79, a component for the lamellar granule secretory system, produces spontaneous eczema in an experimental model of atopic dermatitis.
, who rescued the atopic phenotype of flaky tail (Tmem79ma/maFlgf/ft) mice by transgenic expression of intact TMEM79. Although not causing atopy alone, lack of Flg may modulate skin inflammation and atopy of TMEM79-deficient mice. The comparison of Tmem79ma/maFlgft/ft BALB/c congenic mice (
) is confounded by the difference in strain background. The differences in skin phenotype between these mice could be caused by the absence or presence of FLG or by additional genetic variation in the respective congenic interval. Genes significantly altered in Tmem79ma/maFlgft/ft mice compared with those in BALB/c mice include barrier genes (Supplementary Table S1) but also include genes with functions in the immune system.
FLG deficiency, the cause of IV, is arguably the most important barrier defect associated with atopy (
). Our finding that loss of FLG is not sufficient to cause eczema and atopy in mice is in line with lack of atopy in a fraction of patients with IV homozygous for FLG null alleles. Our result highlights that atopy-promoting variants of other genes are required to synergize with FLG deficiency in causing atopy. Defects of other barrier genes linked to FLG in the epidermal differentiation cluster and adjacent regions may contribute to human atopy. Mouse models with defects of one or several skin barrier genes, including the Tmem79ma/maFlgft/ft BALB/c congenics, will be instrumental in understanding the immune dysregulation associated with a compromised skin barrier.
Materials and Methods
Mice
Mice were kept under specific pathogen-free conditions at the Experimental Centre, Faculty of Medicine Carl Gustav Carus, Dresden University of Technology (Germany). All experiments were done according to the German animal welfare law approved by the Landesdirektion Dresden (reference number DD25-5131/474/24). BALB/c mice were obtained from Janvier (Le Genest-Saint-Isle, France). Flgft/ft BALB/c congenic mice were kept at Trinity College Dublin (Ireland) and genotyped for mutations in Flgft(5303delA) and Tmem79ma (p.Y280∗) as described by
Generation of Flg−/− mice on a pure BALB/c background
Zygotes were isolated from superovulated female BALB/c AnNCrl mice at E0.5. A total of 526 single zygotes were microinjected with Cas9 RNPs prepared from 5.25 μM NLS-Cas9 (Toolgen, Seoul, South Korea), 1.73 μM crRNA1 (unique target site: 5′ GCTGGCAAAAGCATATTATG-3′), 1.73 μM crRNA2 (target site in each of the Flg repeat regions: 5′ GTCAGCGCAAGATCAGGCTC-3′), and 6.9 μM tracrRNA (Integrated DNA Technologies, Coralville, IA). A total of 187 embryos developed to the two-cell stage and were transferred into pseudopregnant foster mothers.
Quantification of TEWL
Mice that had been shaved on their back 24 hours earlier were anesthetized (ketamine and xylazine), and TEWL was measured using an MDD4 device (CK electronic, Cologne, Germany) according to the manufacturer’s instructions. Each value was acquired as a mean of three individual measurements.
Flow cytometric analysis of the ear skin
Ears were digested with an enzyme mixture containing Liberase (25 μg/ml, Liberase Research Grade, Sigma-Aldrich, St. Louis, MO), hyaluronidase (0.5 mg/ml Sigma, 100 mg hyaluronidase from bovine testes, Sigma-Aldrich), and DNaseI (200 U/μl, DNase I grade II, from bovine pancreas, Sigma-Aldrich) for 1 hour at 37 °C. Disassociated tissue was filtered, washed, and pelleted by centrifugation. Cells were resuspended in 75 μl of Fc blocking medium (5 μg/ml purified anti-mouse CD16/32 antibody, BioLegend [San Diego, CA] in FACS buffer) and incubated for 20 minutes on ice. After centrifugation, cells were resuspended in 75 μl of FACS buffer containing antibodies and incubated on ice for 45 minutes. Analysis was performed on a BD FACSAria III Cell Sorter (BD Biosciences, San Jose, CA), and FlowJo (TreeStar, Ashland, OR) was used for data analysis.
Quantification of total and antigen-specific IgE
Total mouse serum IgE levels were measured using the ELISA MAX Standard Set Mouse IgE kit (BioLegend) according to the manufacturer’s instructions. All individual standards and mouse serum samples were analyzed in triplicates, with means of triplicates displayed in graphs. For analysis of OVA-specific serum IgE, 96-well MaxiSorp ELISA plates (Nunc, Roskilde, Denmark) were coated with 2 μg/ml of OVA solution in carbonate buffer, and ELISA was performed using the ELISA MAX Standard Set Mouse IgE kit (BioLegend).
Whole-genome sequencing
High-molecular-weight genomic DNA was extracted from the snap-frozen spleen of Flgft/ft BALB/c congenics (
) and BALB/c Flg−/− mice, and long insert libraries were prepared using the SMRTbell Express Template Prep Kit 2.0 (Pacific Biosciences of California). Sequencing was done on a PacBio Sequel Instrument (Pacific Biosciences of California).
Data availability statement
Primary data are available to any researcher on request. RNA sequencing data were deposited under the Gene Expression Omnibus accession numbers GSE158028 and GSE164623. De novo assemblies of chromosome 3 generated by Pacific Biosciences of California Long-Read Single Molecule Real-Time Sequencing and annotations can be downloaded from https://bds.mpi-cbg.de/hillerlab/Bat1KPilotProject/.
This work was funded by Deutsche Forschungsgemeinschaft grants RO2133/9-1 to AR and DA 1311/3-1 to ADah in the setting of FOR2599 and was supported by the Federal Ministry of Education and Research (grant 01IS18026C). The majority of the work was performed in Dresden, Germany. We thank Werner Müller for helpful discussion. Christina Hiller, Livia Schulze, Madelaine Rickauer, and Tobias Häring provided expert technical assistance.
Supplementary Figure S1Generation and genotyping of BALB/c Flg−/− mice.(a) Numbers of BALB/c zygotes injected with Cas9-RNPs, transferred embryos, number of viable offspring resulting from these transfers, and numbers of founders harboring large deletions involving cleavage by both guides. (b) PCR result for mutant allele 1, resulting from deletion downstream of the guide 1 target site and one of the guide 2 target sites with loss of Flg repeats. The deletion results in absence of the FOR-REV1 product and alters the length of product FOR-REV2. (c) Strategy for PCR-based identification of mutant Flg alleles. Separate reactions were performed with primer pairs FOR-REV1 and FOR-REV2. bp, base pair; WT, wild type.
Supplementary Figure S2Alteration of gene expression and composition of the microbiota of Flg−/− skin supplement.(a) MA plot representing the RNA Seq analysis of whole back skin of four female Flg−/− mice aged 8–12 weeks and four littermate controls. (b) Top 10 significantly enriched pathways in the Flg−/− back skin, as determined by GSEA. (c) Gating strategy and postsort reanalysis of the flow cytometric purification of basal (CD49f+) KCs for bulk RNA Seq. (d) Top nine significantly enriched pathways in the Flg−/− CD49f+ KCs, as determined by GSEA. (e) Heatmap represents the expression of genes involved in antigen presentation and the expression of Retnla gene (relative expression, individual read counts corrected for the mean expression of the respective gene); P < 0.05 for all genes. (f) Relative abundance of microbial phyla in cheek skin swabs from female BALB/c Flg−/− aged 8 weeks and control cohoused females. ES, enrichment score; FDR, false discovery rate; FSC-A, forward scatter area; FWER, family-wise-error rate; GS, gene set; GSEA, Gene Set Enrichment Analysis; KC, keratinocyte; KO, knockout; MA, M&A scale; NES, normalized enrichment score; NOM P val, nominal P-value; Padj, adjusted P-value; RNA Seq, RNA sequencing; SSC-A, side scatter area; SSC-H, side scatter height; val, value; WT, wild type.
Supplementary Figure S4Ear skin histology of BALB/c Flg−/− and control mice. H&E staining. Insets represent the images shown in Figure 2a. Images are representative of at least 10 mice. Bar = 100 μM.
Supplementary Figure S5DO11.10 transfer experiment: gating strategy and clinical score.(a) Gating of proliferating and IL-4–expressing OVA-specific transgenic DO11.10 donor T cells in skin draining lymph node upon epicutaneous OVA administration. (b) Eczema severity score used to macroscopically quantify dermatitis in BALB/c Flg−/− and control recipients of OVA-specific transgenic DO11.10 T cells epicutaneously treated with OVA (Figure 5d). OVA, ovalbumin.
Standardized tape stripping prior to patch testing induces upregulation of Hsp90, Hsp70, IL-33, TNF-α and IL-8/CXCL8 mRNA: new insights into the involvement of 'alarmins'.
A homozygous nonsense mutation in the gene for Tmem79, a component for the lamellar granule secretory system, produces spontaneous eczema in an experimental model of atopic dermatitis.