Characterization of age-associated gene expression changes in mouse sweat glands

Sweating capacity is reduced in old mice

To evaluate the change of sweating capacity during aging, we studied 6 young (3 months old (m.o.)) and 6 old (28 m.o.) C57BL/6JN male mice using the iodine-starch sweat test [4]. The results revealed a 30% reduction of sweating spots in old relative to young mice (Figure 1A, 1B), indicative of reduced numbers of actively secreting SWGs in the old. The reduction of active SWG numbers is consistent with earlier reports in old mice and humans [7, 8, 14]. Histological analysis showed loosely distributed SWG clusters in old relative to young mice, but without significant reductions in cluster numbers (Figure 1C, upper panels). Notably, the lumens of sweat ducts (Figure 1C, arrows) and secretory coils (Figure 1C, dotted red circles as examples) were enlarged in old SWGs.

Identification of mRNAs enriched in SWGs

The sweating phenotype in old mice prompted us to analyze the transcriptomic alterations accompanying SWG aging. To do so, we first identified mRNAs enriched in SWGs by studying male ‘Tabby’ mice, in which all parts of the SWG are absent, including the secretory coil as well as dermal and epidermal ducts, but the skin epidermis and dermis remain intact [13]. Thus, we performed RNA-seq analysis of skin obtained from the footpads (where the mouse SWGs are located) of SWG-deficient Tabby mice and SWG-bearing WT mice in order to identify mRNAs preferentially expressed in SWGs (GSE249784).

Unlike the hind footpads, the fore footpads in mice lack hair follicles in the center [15]; thus, to avoid contamination of the hair follicle transcriptomes, we collected fore footpads from 3 m.o. Tabby male mice and age- and sex-matched WT mice (3 biological replicates each) for RNA-seq analysis (Figure 2A). At the established criteria of absolute fold change >2.0 and p-value < 0.01, 365 mRNAs were less abundant and 72 mRNAs more abundant in Tabby relative to WT footpads (Figure 2B and Supplementary Table 1). We focused on the 365 mRNAs showing reduced expression in the SWG-deficient Tabby footpads; we reasoned they would be enriched in mRNAs expressed in SWGs.

We narrowed down the list of underrepresented mRNAs by establishing a higher fold difference between Tabby and WT. The Eda mRNA in Tabby male footpads was ~4.5-fold lower than in WT footpads (Supplementary Table 1), and was virtually undetectable by in situ hybridization experiments [12]. Applying this criterion (Tabby/WT <4.5) yielded 171 underrepresented mRNAs in Tabby footpads (Supplementary Table 1); we designated these 171 mRNAs as ‘SWG-enriched’. As expected, GO annotation (using ShinyGO) showed that many of the SWG-enriched mRNAs were associated with exocrine system development, ion transport, and synaptic regulation, along with additional terms including tight junction assembly, stem cell differentiation, lipid transport, and cell-cell adhesion (Figure 2C).

Identification of core secretory mRNAs

From the 171 mRNAs, we further selected mRNAs encoding proteins associated with secretion based on GO annotations (Figure 2C). In particular, 47 mRNAs were identified encoding proteins implicated in secretory functions, from transcription factors to ion channels, ion transporters, and trans-synaptic signaling proteins (Figure 2D, 2E and Supplementary Table 1); we named them ‘core secretory’ mRNAs. Many such mRNAs expressed in SWGs were present in this core list – Foxa1, Best2, Foxc1, Foxi1, Kcnn4, Kcnk5, Kcnq1, Kcne3, Clic6, Slc26a4 (Pendrin), Slc12a2 (Nkcc1), and Chrm3 mRNAs were reported to be expressed specifically in SWGs in rodent skin [4, 15–18]. Interestingly, mice with mutations in genes Foxa1, Best2, and Foxc1, and mice with double mutations in genes Kcnn4 and Kcnk5, were reported to have reduced sweat secretion [4, 16, 19]. However, most core secretory mRNAs, including Atp6v1b1, Serca3, Kit, Fxyd2, Vipr1, Cadps2, Gsg1l, and Vamp1 mRNAs, have not been reported in SWGs (Figure 2D, 2E and Supplementary Table 1).

We then confirmed the SWG-specific localization of core secretory proteins by immunofluorescence staining (Figure 3). The transcription factor FOXA1 and the anion channel BEST2 were expressed in scattered sweat secretory cells (Figure 3A), as previously reported [4]. Some proteins were also identified that were novel to SWGs (Figure 3A). For example, ATP6V1B1, a proton pump, was specifically expressed on the luminal membrane of secretory cells; ATP2A3 (SERCA3) was found in the cytoplasm of the secretory coil, but not the sweat duct; while VAMP1, a synaptic vesicle-associated protein, was localized in KRT14+ myoepithelial cells surrounding the secretory coil, suggesting that myoepithelial cells may be an important target of neuronal signaling in SWGs.

FOXC1 was strongly expressed in nuclei of luminal cells in epidermal and dermal sweat ducts, which is consistent with its critical function in sweat duct differentiation (Figure 3B) [19]. Notably, FOXC1 protein was also observed in the nuclei of scattered secretory cells, although the expression levels were much lower in the secretory coil compared to the duct (Figure 3B). The above core secretory proteins analyzed by fluorescence immunohistology were specifically localized in the secretory coil and/or sweat duct, but not in skin epidermis or dermis (not shown).

Identification of altered SWG-enriched mRNAs in old mice

We then determined if the SWG-enriched, core secretory mRNAs changed with age. We used RNA-seq analysis to identify differentially expressed mRNAs in the footpads of young (3 m.o.) and naturally aged (28 m.o.) male C57BL/6JN mice (4 biological replicates per group; Figure 4A). We found lower levels of 264 mRNAs and higher levels of 410 mRNAs in old relative to young footpads, using absolute fold change >1.5 and p < 0.01 (Figure 4B and Supplementary Table 2). The list included 28 SWG-enriched mRNAs, 13 less abundant and 15 more abundant in old footpads compared to young (Figure 4C, Supplementary Table 2), many of them core secretory mRNAs (see below). Among the non-secretory, SWG-enriched mRNAs reduced with age, Oit1 mRNA encodes OIT1, a protein that promotes the integrity of the gastrointestinal tract [20], and Plin2 mRNA encodes PLIN2, a protein implicated in lipid droplet formation [21]. Among the mRNAs increasing with age, Stra6 mRNA encodes STRA6, a protein important for removing retinol from the retinol-retinol binding protein complex [22], and WFDC family proteins were proposed to be protease inhibitors [23]. The functions of these mRNAs in SWGs warrant further study.

Identification of altered core secretory mRNAs in old mice

Notably, 11 of the 28 altered SWG-enriched mRNAs in old mice encoded core secretory proteins. Foxa1, Best2, Chrm3, Atp2a3, Foxi1, and Slc26a4 mRNAs were less abundant, whereas Foxc1, Gsg1l, Cadps2, Rhcg, and Slc5a8 mRNAs were more abundant in old compared to young footpads (Figure 4C, arrows, and Supplementary Table 2). These mRNAs encode transcription factors (FOXA1, FOXI1, FOXC1), ion channels (BEST2, SLC26A4, RHCG), ion transporters (ATP2A3/SERCA3, SLC5A8), and trans-synaptic signaling proteins (CHRM3, GSG1L, CADPS2), most of them with important roles in SWGs or other secretory organs (Figure 2E and Discussion).

The RNA-seq data were validated using reverse transcription (RT) followed by quantitative real-time PCR (RT-qPCR) analysis using specific primer pairs (Supplementary Table 3). As expected, there were moderate but significant reductions in the levels of Foxa1, Best2, Atp2a3, Chrm3, Slc26a4, and Foxi1 mRNAs in old footpads (Figure 4D), while the levels of Foxc1 mRNA were significantly elevated in old footpads (Figure 4D). As a marker of tissue aging, the senescent marker p16 mRNA was elevated in old footpads by both RT-qPCR and RNA-seq analyses (Figure 4D and Supplementary Table 2).

We then assessed whether the alterations in mRNA expression in old footpad skin resulted in detectable changes at the protein level using immunofluorescence staining and quantification of signal intensities. Most proteins, including FOXA1, BEST2, and SERCA3, did not show detectable changes in signal intensity or expression patterns between young and old SWGs (data not shown). However, a significant increase was observed in the number of FOXC1+ secretory cells in old SWGs (Figure 5A, 5B). The intensity of FOXC1 signals in secretory cells was also elevated overall in the old, although this parameter did not reach statistical significance (Figure 5C). Notably, FOXC1+ cell numbers and FOXC1 signal intensities were indistinguishable between young and old sweat ducts (Figure 5D–5F).

Animal models

All mouse work was performed under an Animal Study Proposal (ASP#476-LGG-2024) approved by the Animal Care and Use Committee of the National Institute on Aging (NIA). C57BL/6JN male mice were imported from the NIA Aged Rodent Colonies (https://www.nia.nih.gov/research/dab/aged-rodent-colonies-handbook). Eda mutant Tabby mice (C57BL/6J A^w-J-Eda^Ta-6J/J, Strain#000338) were purchased from the Jackson Laboratory (Bar Harbor, ME), and bred in the NIA animal facility. All mice were provided with standard diet ad libitum and maintained under a 12-h/12-h light/dark cycle.

Starch-iodine sweat test

The protocol of iodine-starch sweat test was reported previously [4]. Briefly, iodine/alcohol solution (1 g iodine/50 ml ethanol) was applied to the surface of a hind footpad of experimental mice. Once dry, the surface was painted with starch-oil (10 g starch/10 ml castor oil). Purple sweating spots start to form in 1–2 min and peak at around 10 min. Images were taken at 10 min with a close-up mode of a Canon PowerShot G16 camera and quantified using the ImageJ plug-in cell counter.

Tissue collection from mice

Fore footpad skin was taken from 3 m.o. Tabby male mutant mice, as explained above (C57BL/6J A^w-J-Eda^Ta-6J/J, Strain#000338) and their wild-type (WT) littermates, as well as 3 m.o. and 28 m.o. (C57BL/6JN) male mice for histology and RNA isolation. For histological studies, tissues were embedded in a Tissue-Tek OCT compound (VWR, Cat#25608-930) on dry ice and stored in −80°C freezer or fixed in 10% formaldehyde (Ricca Chemical) for 16 h at 25°C, dehydrated, and embedded in paraffin. For RNA isolation, footpad skin was frozen on ice immediately after isolation and kept at −80°C until use.

RNA isolation, library preparation, mRNA sequencing, and RT-qPCR analysis

Total RNA was extracted from frozen footpad skin obtained from three 3 m.o. Tabby male mice and three 3 m.o. WT male mice using Qiagen RNeasy Fibrous Tissue Mini Kit (Qiagen, Cat#74704) with DNase treatment. Briefly, tissue samples were transferred to 2.0 ml Sarstedt tubes (Sarstedt, Cat#72.694.006) containing 800 μl of RLT (RNA lysis buffer) (Qiagen, Cat#74704) plus DTT (Dithiothreitol) (Teknova, Cat#D9750) buffer from the Qiagen kit, and added 1.0-mm diameter Zirconia beads (Biospec, Cat#11079110z) to make a total volume of 1.25 ml. The tube was processed using the Precellys 24 homogenizer, followed by centrifugation at 17,000 RCF (g) for 5 min at 4°C. The resulting homogenate was evenly distributed into two 1.5-ml Eppendorf tubes for Proteinase K treatment, and subsequently used for RNA isolation. The quality and quantity of RNA were assessed using the Agilent Bioanalyzer and the RNA 6000 Nano Kit (Agilent, Cat#5067-1511).

An aliquot of RNA (125 ng) was utilized for sequencing library preparation using the Illumina TruSeq Stranded mRNA Library prep kit (Illumina, Cat# 20020594), following the manufacturer’s protocol. Briefly, mRNAs were isolated using Poly-T oligomer-containing magnetic beads, and subsequent cDNA synthesis was carried out. The cDNAs were subjected to 3’ end adenylation and adapter ligation, and were then purified using AMPure beads (Beckman, Cat#A63881). The purified cDNAs were size-selected using SPRIselect beads (Beckman, Cat#B23318), underwent PCR enrichment, and were purified once more with SPRIselect beads to generate the final libraries. The quality and quantity of the libraries were assessed using the Agilent DNA 1000 Screen Tape on the Agilent Tapestation. Finally, paired-end sequencing of the libraries was conducted using the Illumina NovaSeq 6000 sequencer. The same protocols were employed for isolating RNA and preparing cDNA libraries for four 3-m.o. and four 28-m.o. C57BL/6JN male mice. We obtained ~50 million reads from Tabby and WT footpad libraries, and ~40 million reads from young and old C57BL/6JN footpad libraries. RNA-seq data were deposited in GEO (GSE249784).

The RNA-seq reads were aligned to mouse genome NCBIM38/mm10 with spliced transcripts alignment to a reference (STAR) software v2.7.0f_0328, and subread (version 2.10.5) was used to create gene counts. Normalization of counts and differential expression analysis of genes between experimental groups were performed using the DESeq2 package (versions 1.30.1 and 1.36.0) pipeline [34].

For mRNA profiling of Tabby and WT controls, we set the criteria of absolute fold change (FC) >2.0, p-value < 0.01, and base mean >30, for significant change. From underrepresented mRNAs in Tabby footpads, we collected mRNAs that show FC >4.5 (WT/Ta) and designated them ‘SWG-enriched’ mRNAs. We carried out GO annotation (ShinyGO: http://bioinformatics.sdstate.edu/go/) [35] with SWG-enriched mRNAs, and further collected mRNAs encoding protein involved in secretory function. We designated them ‘core secretory’ mRNAs.

For mRNA profiling of young and old footpad skin, we set the criteria of absolute FC >1.5, p-value < 0.01, and base mean >50, as a significant change. We further identified SWG-enriched mRNAs and core secretory mRNAs from the significantly altered mRNA list in old footpad skin by Venn diagrams (https://bioinformatics.psb.ugent.be/webtools/Venn/).

Following RNA isolation from young and old mouse footpad skin, 500 ng of total RNA was used for reverse transcription (RT) followed by real-time quantitative PCR (qPCR) analysis. For qPCR analysis, 0.1 μl of cDNA was used with 250 nM of primers (Supplementary Table 3) and KAPA SYBR^® FAST qPCR Kits (KAPA Biosystems). RT-qPCR analysis was carried out on a QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific) with a cycle setup of 20 s at 95°C, followed by 40 cycles of 1 s at 95°C, and 20 s at 60°C. For the analysis, relative RNA levels were calculated after normalizing to mouse Actb mRNA (which encodes the housekeeping protein β-actin), using the 2^−ΔΔCt method.

Histology, immunohistology, and quantification of signal intensities

For general histology, 5- to 8-μm paraffin sections were cut, deparaffinized, and stained with hematoxylin (Sigma-Aldrich, Cat#SLCL4498) and eosin (Sigma-Aldrich, Cat#SLCN7193) using the manufacturer’s standard protocol.

For immunofluorescence microscopy, 14-μm frozen sections were cut with a Leica CM3050S cryostat, attached to Silane-Prep Slides (Sigma-Aldrich, Cat#S4651-72EA), air-dried for 30 min, and stored at -20°C until use. Sections were fixed in 4% paraformaldehyde (PFA) (EMS, Cat#15710) for 6 min, rinsed in phosphate-buffered saline (PBS) (Sigma-Aldrich, Cat#P3813) containing 0.1% Triton X-100 (Sigma Aldrich, Cat#9036-19-5) for 15 min, and blocked with 10% donkey serum (DS) (Sigma-Aldrich, Cat#D9663-10 mL) in PBS + 0.1% Tween-20 (DS-PBST) (Sigma-Aldrich, Cat#P1379) for 1 h. Sections were incubated for 18 h at 4°C with primary antibodies diluted in DS-PBST (see Supplementary Table 3 for antibodies and dilutions). Sections were then rinsed with PBST and incubated with Alexa Fluor-conjugated secondary antibodies diluted in DS-PBST (see Supplementary Table 3) for 45 min at 25°C. Sections were mounted in ProLong Gold with DAPI (Invitrogen, Cat#P36935) and images taken with a Zeiss 980 LMS confocal microscope with settings of the channels of interest held constant. The z-stack function was used to create a z-stack image with slices width determined by the manufacturer’s “optimal” setting slices.

Signal quantification was performed using the ImageJ (https://imagej.net/ij/index.html) software following the method described by Shihan et al. [36] with modifications. Briefly, positive signals in the form of integrated density were measured. Maximum intensity (MAX) projections were created for each section. MAX projections for FOXC1 duct and secretory coil images were produced from 45 and 30 0.3-μm slices, respectively. Regions of interest (ROIs) for measurement were then identified by isolating, blurring, and thresholding the channel of interest. Each image set was processed with a unique constant threshold and blur setting. After creating a thresholding image, particle analysis was used to identify areas of interest. Two to four images were analyzed from each mouse for each unique image set. For FOXC1 quantification, >70 positive nuclei for ducts and >90 positive nuclei for secretory coil cells from each mouse were counted and averaged to obtain a final representative integrated density measurement. All measurements were adjusted to account for background signal by measuring the integrated density value of 6 randomly placed ROIs in control sections stained solely with secondary antibodies. ROIs used were of the approximately same size to the average positive ROIs. Representative measurements were then averaged to obtain an overall group average. Cells were counted for percentile quantification from MAX projections with the ImageJ plug-in cell counter. Statistical significance (^*p < 0.05; ^**p < 0.01; ^***p < 0.001) was established using Student’s t-test.