Glycome profiling by lectin microarray reveals dynamic glycan alterations during epidermal stem cell aging

Abstract Aging in the epidermis is marked by a gradual decline in barrier function, impaired wound healing, hair loss, and an increased risk of cancer. This could be due to age‐related changes in the properties of epidermal stem cells and defective interactions with their microenvironment. Currently, no biochemical tools are available to detect and evaluate the aging of epidermal stem cells. The cellular glycosylation is involved in cell–cell communications and cell–matrix adhesions in various physiological and pathological conditions. Here, we explored the changes of glycans in epidermal stem cells as a potential biomarker of aging. Using lectin microarray, we performed a comprehensive glycan profiling of freshly isolated epidermal stem cells from young and old mouse skin. Epidermal stem cells exhibited a significant difference in glycan profiles between young and old mice. In particular, the binding of a mannose‐binder rHeltuba was decreased in old epidermal stem cells, whereas that of an α2‐3Sia‐binder rGal8N increased. These glycan changes were accompanied by upregulation of sialyltransferase, St3gal2 and St6gal1 and mannosidase Man1a genes in old epidermal stem cells. The modification of cell surface glycans by overexpressing these glycogenes leads to a defect in the regenerative ability of epidermal stem cells in culture. Hence, our study suggests the age‐related global alterations in cellular glycosylation patterns and its potential contribution to the stem cell function. These glycan modifications detected by lectins may serve as molecular markers for aging, and further functional studies will lead us to a better understanding of the process of skin aging.


| INTRODUC TI ON
The epidermis is the first barrier of our body that protects us from infection and dehydration. The epidermis consists of the interfollicular epidermis (IFE) and its appendages (hair follicles: HFs, sebaceous and sweat glands) and is replenished by distinct populations of stem cells (Gonzales & Fuchs, 2017;Rognoni & Watt, 2018). The IFE is renewed by stem cells located in the basal layer, which give rise to stratified squamous epithelium. HF stem cells reside in a specialized structure, called the bulge, and contribute to cyclic regeneration of HFs. Stem cells in the IFE and HFs are largely independent of each other during homeostasis, but they possess plasticity to change their fates in response to injury (Gonzales & Fuchs, 2017;Rognoni & Watt, 2018). The epidermis is separated from the dermis by the basement membrane enriched in the extracellular matrix, which regulates stem cell property and fates (Chermnykh, Kalabusheva, & Vorotelyak, 2018;Watt & Fujiwara, 2011).
Glycosylation is a reaction that proteins or lipids are modified with glycans (Varki, 2009). The protein glycosylation involves stepwise addition and removal of glycans, primarily mediated by glycosyltransferases and glycosidases (Spiro, 2002). The presence of glycans determines the structure, stability, and localization of glycoproteins, which affect a wide variety of biological processes, such as development (Haltiwanger & Lowe, 2004), tumorigenesis (Ohtsubo & Marth, 2006) and inflammation (Varki & Gagneux, 2015). Glycans are required for stem cell regulations by modulating signaling molecules that govern self-renewal and differentiation of stem cells (Nishihara, 2018). As glycans are located at the cell surface, they have been utilized as biomarkers, for example, pluripotent status of mouse embryonic stem cells (Adewumi et al., 2007;Muramatsu & Muramatsu, 2004;Muramatsu & Muramatsu, 2004) and human induced pluripotent stem cells (Tateno et al., 2011). Given the role of glycans in diverse biological and biochemical processes, glycosylation might play an important role in the process of stem cell aging.

However, the glycosylation state of stem cells in aged mammalian
tissues remains largely uncharacterized.
The glycome analysis of tissue stem cells has been challenging, as tissue stem cells are rare and large amounts of samples are required for the structural analysis of glycans by mass spectrometry. Lectin microarray, a platform for high-throughput glycome analysis, enables a comprehensive glycan profiling even from a relatively small number of cells (Kuno et al., 2005). Lectins are a class of glycan-binding proteins that recognize various glycan structures (Hirabayashi, 2004).
In lectin microarrays, a series of lectins with various glycan-binding specificities are immobilized on a glass slide (Hirabayashi, Yamada, Kuno, & Tateno, 2013). Lectin-glycan interactions are quantitatively measured as fluorescent signals after incubation with fluorescence-labeled samples in the lectin microarray (Kuno et al., 2005).
Using this technology, glycoproteins isolated from various biological samples can be utilized for glycome analysis without the liberation of glycans (Tateno et al., 2007).
In our current study, we performed a comprehensive glycome analysis of IFE and HF stem cells in the old mouse skin by using lectin microarray consisting of 96 lectins with various glycan-binding specificities (Tateno et al., 2011). We found that epidermal stem cells undergo global changes in their glycosylation patterns during aging, with decreased mannose and increased sialic acid (Sia) modifications. By overexpressing glycogenes in vitro, we recapitulated the old-type glycome patterns in epidermal stem cells, which led to a decline in the proliferation capacity.
We thus propose functional implications of glycans in stem cell regulation.

| Distinct glycosylation patterns between young and old epidermal stem cells
To analyze the glycosylation state of epidermal stem cells during aging, IFE and HF stem cells were isolated from wild-type C57BL/6 mice at 2 months (young, N = 4) and 22-24 months (old, N = 3) of age and subjected to lectin microarray ( Figure 1a). IFE stem cells (α6-integrin+/CD34−/Sca1+) and HF stem cells (α6-integrin+/CD34+) were separated by flow cytometry based on their differential expression of cell surface markers (Figure 1b,c). In old mouse skin, we detected significantly lower number of HF stem cells as previously reported (Matsumura et al., 2016), whereas the number of IFE stem cells remained unchanged ( Figure S1). We asked whether their cell surface glycans were affected by aging. The hierarchical clustering of lectin microarray data showed that young and old samples were

| Classes of lectins that differentially identified glycans in young and old stem cells
For the identification of lectins that were differentially bound to glycan structures between young and old stem cells, statistical analysis was performed using the mean normalized signals obtained from lectin microarray. Several classes of lectins were significantly changed (p < 0.01) between young and old stem cells ( Figure 3 and Table S1), and we categorized them based on their glycan-binding specificities.
Notably, lectins of similar functional classes were detected with significant differences in both IFE and HFs. Taken together, our lectin microarray analysis identified common sets of lectins that recognize age-dependent glycan changes in IFE and HF stem cells: decreased mannose-binding lectins and increased Sia-binding lectins during aging.

| Detection of age-related glycan changes in epidermal stem cells by flow cytometry using rHeltuba and rGal8N
To test the ability of lectin-directed detection of glycans in living stem cells, we employed flow cytometry analysis in young and old stem cells upper left, and 5d, left). The signals of rGal8N were inhibited by lactose, confirming the specific binding of rGal8N to glycans. In the HFs, however, there were no significant differences in the rGal8N signal intensity between young and old stem cells (Figure 5b, upper right, and 5d, right). One possible interpretation is that glycolipids, which content may differ between HF and IFE stem cells, had been detected by rGal8N in live HF cells and masked the difference between young and old HF stem cells. Taken together, these data indicate that both rHeltuba and rGal8N lectin probes successfully detected glycan changes in freshly isolated IFE stem cells by flow cytometry.

| Upregulation of sialyltransferase and mannosidase genes in old epidermal stem cells
To address which enzymes are responsible for age-related glycosylation changes in epidermal stem cells, we performed gene expression analysis using RT 2 profiler PCR array of mouse glycosylation-related genes. RNAs isolated from IFE stem cells at 2 months (young, N = 3) or 22-24 months (old, N = 3) of age were used for quantitative PCR.
Man1a is an α-1,2 mannosidase and is responsible for the removal of mannose residues to initiate the complex-type N-glycan formation F I G U R E 3 List of lectins significantly changed between young and old epidermal stem cells. (a, b) Lectins bound differentially to the young or old interfollicular epidermis (IFE) (a) and hair follicles (HF) (b). Statistically significant differences are calculated by unpaired Student's t test and p < 0.01 are selected. Lectins are categorized based on their binding specificities. Data are shown with t-values. Also, see Table S1. (Varki, 2009), which matches with the decreased signals of mannose-binding lectins in old IFE stem cells (Figure 3). Similarly, we also found an increased expression of Man1a, St3gal2, St6gal1 in the old HF stem cells ( Figure S2 and Table S2). Thus, glycan changes of epidermal stem cells during aging are possibly mediated by the changes in glycosyltransferase and glycosidase expressions with age.

| D ISCUSS I ON
In vivo sign of aging in the skin can be observed at the tissue and organismal levels; however, the molecular aspects of aging at the stem cell level remains elusive. In our current study, we performed a high-throughput lectin-based glycan profiling on murine epidermal stem cells and revealed their dynamic glycan alterations during aging. We propose a concept, "glycome shift" as a new molecular factor of epidermal stem cell aging (Figure 6c): high mannose-type N-glycans are globally replaced by α2-3/6 sialylated complex-type N-glycans with age. Intriguingly, overexpression of three glycogene(s) (Man1a, St3gal2, St6gal1) recapitulated the aging glycan patterns and impaired the growth of primary keratinocytes, suggesting that the glycans could be one of the drivers of age-related decline in the proliferation ability of epidermal stem cells. The identified lectins, the mannose-binding rHeltuba and the α2-3Sia-binding rGal8N can be used as probes to visualize, select, or remove aged stem cells, with implications in future applications for regenerative therapy and diagnosis of skin aging. We also provide a proof of concept that our lectin microarray platform (Tateno et al., 2011) can successfully analyze the glycome of adult tissue stem cells, which are rare in tissue (≤1% of total skin cells) and their biochemical properties are not well-characterized due to technical difficulties.
As glycosylation plays a critical role in cell-cell and cell-matrix interactions, the changes in glycans on the surface of epidermal stem cells might affect their ability to interact with neighboring stem cells, other cell types (e.g., fibroblasts, immune cells, and blood vessels), basement membrane and signaling molecules, all of which are essential components for maintaining the skin integrity. It will be interesting in the future to identify core proteins in which differential glycosylation takes place and to reveal the functional importance and biological meaning of glycosylation in age-related skin dysfunction.
An aged skin exhibits declined wound healing ability, which is in part caused by impaired crosstalk between epidermal stem cells and dendritic epithelial T cells (Keyes et al., 2016). Given that several immune cells, including dendritic cells, have mannose-binding receptors in the epidermis (Wollenberg et al., 2002), the decreased mannose in old IFE stem cells that we observed here (Figure 6c) could be associated with the defective stem cell-immune cell interaction in aged skin.
Our study showed an increase in α2-3 and α2-6 sialylation along with the expression of the corresponding sialyltransferase (St3gal2 and St6gal1) in old IFE stem cells (Figure 6c). In agreement with our findings, sialylation was reported to be increased in the aged mouse muscle (Hanisch et al., 2013). The upregulation of sialyltransferases has also been suggested as a potential aging marker in human, which shows a higher activity of St6gal1 in the plasma of individuals above 80 years of age (Catera et al., 2016). In addition, an α2-6 sialylation and the expression of St6gal1 were upregulated during epithelial to mesenchymal transition and tumor formation (Lu et al., 2014;Swindall et al., 2013).
In human pluripotent or mesenchymal stem cells, a higher sialylation is associated with a greater potential of stem cells (Hasehira et al., 2012;Tateno et al., 2011;Wang et al., 2015). The observed differences in the sialylation patterns might be due to the differences in cell types, species, or target proteins, indicating a diverse role of sialylation in the process of aging. Future studies using conditional knock-out or overexpression of differentially expressed glycosyltransferases in the mouse epidermis will directly address the role of sialylation in the context of epidermal stem cell aging.

| Mice
All animal procedures were conducted following animal experimentation guidelines approved by the Institutional Animal Experiment Committee at the University of Tsukuba. Young (2-month-old) and old (22-24-month-old) C57BL/6 mice were purchased from Charles River Laboratories or Japan SLC. Both male and female mice were used for experiments. All the experimental mice were housed in Laboratory Animal Resource Center, University of Tsukuba prior to experiments.

| Isolation of epidermal stem cells by flow cytometry
Mouse dorsal and ventral skin were dissected and the subcutaneous and fat tissues were removed from the dermal side of the skin. The skin was incubated in 0.25% trypsin/versene overnight at 4°C and for

F I G U R E 6
Gene expression analysis of glycosylation-related genes using RT 2 Profiler PCR array. (a) The volcano plot represents fold change and p-values on x-and y-axis, respectively. The vertical red and blue lines represent a fold-change cutoff of ≥1.5. N = 3 for young mice, N = 3 for old mice. Also, see Table S2. (b) Lists of differentially expressed sialyltransferase and mannosidase genes. (c) Schematic representation of the putative glycan changes during epidermal stem cell aging.

| Membrane protein isolation and quantification
Hydrophobic fractions containing membrane proteins were prepared using the CelLytic MEM Protein Extraction kit (Sigma-Aldrich) following the manufacturer's protocols. Proteins were quantified using a micro BCA assay kit (Thermo Fisher Scientific). Protein amounts ranging from 15 to 30 μg were obtained from 100,000-300,000 IFE or HF stem cells.

| Lectin microarray
The high-density lectin microarray was produced according to the method previously described (Tateno et al., 2011). The protein concentration was adjusted to 2 μg/ml with PBST [10 mM PBS (pH 7.4), 140 mM NaCl, 2.7 mM KCl, 1% Triton X-100] and was labeled with Cy3-N-hydroxysuccinimide ester (GE Healthcare). Cy3-labeled F I G U R E 7 Aging-associated glycogene overexpression leads to an impaired keratinocyte growth. (a) Scheme of the glycogene overexpression using the lentivirus system. (b) The qRT-PCR of Man1a, St3gal2, St6gal1 mRNA expression at 4 days after blasticidin selection (N = 3). Lenti-EGFP is used as a control. Data are shown as means ± SD. Mann-Whitney test. *p < 0.05. (c, d) Confirmation of glycan changes by lectin blotting using the horseradish peroxidase (HRP)-labeled lectins, rHeltuba (c) and rGal8N (d). One microgram of protein from three independent experiments is applied on each lane. Data are shown as means ± SD. Students t test. ***p < 0.001. **p < 0.01. *p < 0.05. The signal intensities of bands with indicated size are quantified. (e) Proliferation assay of primary keratinocytes after overexpressing glycogenes. The x-axis represents the time points, and the y-axis represents the absorbance at 450 nm. Absorbance is measured at 0, 1, 3, and 5 days post-infection. Data are shown as means ± SD. Students t test. ***p < 0.001. **p < 0.01. *p < 0.05. (f) Representative images of the primary keratinocytes infected with lenti-EGFP, or a combination or single lenti-Man1a, -St3gal2, and -St6gal1 at day 0 and 5.
proteins were diluted with probing buffer [25 mM Tris-HCl (pH 7.5), 140 mM NaCl, 2.7 mM KCl, 1 mM CaCl 2 , 1 mM MnCl 2 , and 1% Triton X-100] to 0.5 μg/ml and were incubated with the lectin microarray at 20°C overnight. Samples were washed with probing buffer for three times, and fluorescence images were captured using a Bio-Rex scan 200 evanescent-field-activated fluorescence scanner (Rexxam Co.

Ltd.).
The obtained signals were mean-normalized and subjected to unsupervised hierarchical clustering, followed by a heat map analysis.
The lectin signals of triplicate spots were averaged for each sample and normalized relative to the mean value of 96 lectins. The mean nor-

| Detection of lectin binding to epidermal stem cells by flow cytometry
Recombinant lectins (rHeltuba, rGal8N) were labeled with R-Phycoerythrin (PE) using Phycoerythrin Labeling Kit -NH2 (Dojindo) according to the manufacturer's protocol. The single-cell solution was prepared as described above and resuspended in 1% BSA (Sigma-Aldrich, A3059) without using the serum. Cells were stained with Lectin-PE for 1 hr at 4°C, at the following concentra-

| RT 2 profiler mouse glycosylation PCR arrays
Total RNAs were isolated using the RNeasy micro kit (QIAGEN), according to the manufacturer's protocol. The integrity of the isolated RNA was assessed by using RNA Pico Chips and Agilent 2100 bioanalyzer (Agilent Technologies). RNA samples with the RNA integrity number above 8 were used for further analysis. The cDNA from IFE and HF stem cells were synthesized from 50 and 5 ng of mRNA, respectively, using the RT 2 PreAMP cDNA Synthesis Kit (QIAGEN, 330451) followed by pre-amplification using the pathway-specific primer mix for mouse glycosylation (QIAGEN, PBM-046Z).
The relative mRNA expressions of 84 genes regulating mouse glycosylation were analyzed using RT 2 Profiler™ PCR Arrays (QIAGEN, PAMM-046Z) according to the manufacturer's instructions. The cDNA template prepared above was mixed with RT 2 SYBR Green qPCR Master Mix (QIAGEN, 330501) and nuclease-free water. The cDNA mixture of 25 μl was applied to each well of the PCR arrays that contain the preloaded primer mix for each gene. The real-time PCR amplification and detection were performed using a Bio-Rad CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad). Amplification cycle was used as following: activation of DNA Taq polymerase at 95°C for 10 min, followed by 40 cycles of denaturation at 95°C for 15 s and annealing for 1 min at 60°C. The threshold cycle (C t ) was used for PCR array quantification. The threshold values were set similarly across all the PCR array used in the analysis, and the baseline was defined by using the automated baseline option of the machine. Gene whose C t cycle was more than 35 was set as undetectable. C t values of biological replicates obtained from the real-time PCR array analysis were used for the ∆∆ C t -based fold-change calculations. For the data analysis, web-based PCR data analysis provided from the data analysis center of QIAGEN was used. Samples were normalized using automatic normalization from the five housekeeping genes (Actb,B2m,Gapdh,Gusb,Hsp90ab1) in the PCR array. An appropriate correction was also made during the web-based data analysis for the pre-amplification step.
Gene expression whose fold is greater than 1.5 was selected.
Mouse keratinocytes were seeded at 50,000 cells in a 24-well culture plate coated with collagen IV (Sigma). One day later, keratinocytes were infected with lentivirus along with 4 µg/mL polybrene for 16 hr. The 300 µl of medium containing 100 µl each of the glycogenes (Man1a, St3 Gal2, St6gal1) or 300 µl of medium containing lenti-EGFP were used. The medium was changed after 16 hr and the infected keratinocytes were selected by using blasticidin at a concentration of 1 µg/ml.

| qRT-PCR
Mouse keratinocytes RNA was isolated using the RNeasy micro kit (QIAGEN) and real-time RT-PCR was performed using iTaq

| Cell proliferation assay
Keratinocyte proliferation was measured by using the Cell-Counting Kit-8 (CCK-8, Dojindo, Japan) following the manufacturer's instructions. In brief, blasticidin-selected keratinocytes were seeded in triplicate in a collagen-IV-coated flat-bottom 96well plate at 2,000 cells/well. Keratinocytes were grown in the E-medium and analyzed at 0, 1, 3, and 5 days after infection. Ten microliter of CCK-8 reagent was incubated for 2 hr, and the absorption of the samples was measured at 450 nm using xMark microplate reader (Bio-Rad).
For microscope analysis, blasticidin-selected keratinocytes were seeded at 5,000 cells/well in a collagen-IV-coated 24-well plate.
Images were acquired by using the Evos FL cell imaging system (Thermo Fisher Scientific) at indicated time points.

ACK N OWLED G EM ENTS
We thank Dr. J. Kobayashi (Hokkaido University) and Ms. K.
Kawazoe for their critical discussion and technical assistance, and the Animal Resource Center at the University of Tsukuba for excellent mouse care. We would like to thank Dr. M. Kato, Dr. H. Suzuki, and Dr. Y. Watanabe (University of Tsukuba) for their help in lentivirus production. This work was supported by AMED-PRIME, AMED (JP19gm6110016), The Nanotech Career-up Alliance N.R.P to A.S., and research grants from the Mitsubishi Foundation, the Nakatomi Foundation, the Sumitomo Foundation and Hoyu Science Foundation to A.S.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest. Hiromi Yanagisawa https://orcid.org/0000-0002-7576-9186