Human organ rejuvenation by VEGF-A: Lessons from the skin

Abstract

Transplanting aged human skin onto young SCID/beige mice morphologically rejuvenates the xenotransplants. This is accompanied by angiogenesis, epidermal repigmentation, and substantial improvements in key aging-associated biomarkers, including ß-galactosidase, p16^ink4a, SIRT1, PGC1α, collagen 17A, and MMP1. Angiogenesis- and hypoxia-related pathways, namely, vascular endothelial growth factor A (VEGF-A) and HIF1A, are most up-regulated in rejuvenated human skin. This rejuvenation cascade, which can be prevented by VEGF-A–neutralizing antibodies, appears to be initiated by murine VEGF-A, which then up-regulates VEGF-A expression/secretion within aged human skin. While intradermally injected VEGF-loaded nanoparticles suffice to induce a molecular rejuvenation signature in aged human skin on old mice, VEGF-A treatment improves key aging parameters also in isolated, organ-cultured aged human skin, i.e., in the absence of functional skin vasculature, neural, or murine host inputs. This identifies VEGF-A as the first pharmacologically pliable master pathway for human organ rejuvenation in vivo and demonstrates the potential of our humanized mouse model for clinically relevant aging research.

Aged human skin is rejuvenated by grafting onto young mice, for which increased VEGF-A expression is both required and sufficient.

INTRODUCTION

If one follows Sinclair’s persuasive argument that aging is an ultimately fatal disease whose progress can be slowed and reversed (1), and views aging as a druggable and reprogrammable target (2), dissecting the key drivers of human organ aging and developing effective molecular strategies to prevent or even reverse it surely constitutes one of the most fundamental missions of biomedical research. In this endeavor, clinically relevant aging research models are critically needed in which not only the key drivers of human organ aging can be identified but also the most promising senescence prevention and human tissue rejuvenation strategies, e.g., through senolytic drugs (3), can be tested preclinically.

Human skin is ideally suited as such a preclinical aging research model (4) but is rarely used by mainstream aging research for this purpose. Yet, aging of the human body becomes nowhere sooner and more immediately visible than in skin changes and hair graying (5). While massive industry efforts therefore cater to the ancient human desire to halt or reverse the phenotype of aging skin, success at this frontier has remained moderate at best, and many product claims of in vivo rejuvenation of human skin are typically insufficiently substantiated (2). Nevertheless, the molecular mechanisms that underlie extrinsic and intrinsic skin aging in vivo are becoming increasingly understood, albeit mostly in nonhuman animal models of uncertain clinical relevance (4). However, the fact that aged murine tissues can be rejuvenated in vivo by systemic exposure to (incompletely defined) factors present in the blood of young animals (6, 7) demonstrates that the rejuvenation of aged mammalian tissue is possible in principle (2, 6–8).

Notably, this includes human skin. Previously, we had shown that grafting aged human skin to immunocompromised young mice reverts several aging-associated parameters in the epidermis of the human xenotransplants (9, 10). Yet, it is unknown whether the observed skin rejuvenation effects extend beyond the epidermis, and the molecular mechanisms that underlie this striking epidermal rejuvenation phenomenon have remained elusive. Examining this accessible, experimentally pliable, and clinically relevant model for human organ rejuvenation in vivo, the present study hoped to identify druggable targets for human organ rejuvenation. This led us to identify vascular endothelial growth factor A (VEGF-A) as a key driver of human organ rejuvenation in vivo (for more background information, see the Supplementary Introduction).

RESULTS

A young mouse host environment morphologically rejuvenates aged human epidermis

First, we sought to independently reproduce the striking epidermal rejuvenation phenomenon of aged human skin transplanted onto young nude mice we had reported long ago (9, 10) in a new, more severely immunocompromised mouse strain (SCID/beige). Thirty days after engrafting aged human skin onto 14 ± 3.2–month–old SCID/beige mice [“old-in-old” design (OiO)] (Fig. 1 and table S1), the phenotype and all assessed aging-associated readout parameters of aged human epidermis (11) had been fully maintained (Figs. 2 and 3 and fig. S1) compared with the pretransplantation status. Thus, the transplantation procedure as such (skin wounding and wound healing) had no rejuvenation effect on aged human skin.

Fig. 1. Overview: experimental design.

(A) Aged skin was transplanted onto young (OiY) or old (OiO) mice. Similarly, young skin was transplanted on to young (YiY) or old (YiO) mice. (B) Each group of mice was separated to three groups treated with anti–VEGF-A, VEGF-A, or BSA. Qualitative and quantitative (immuno-)histomorphometry were performed before and 1, 2, and 4 weeks after transplantation. (C) Experimental manipulations in aged human skin transplanted onto young SCID mice (OiY) for 1 or 12 months, along with selected changes in key aging readouts. OiY mice were treated with anti–VEGF-A antibodies, while control groups and OiO mice received intradermal injection of VEGF-A–loaded nanoparticles. (D) Organ culture of aged human skin or epidermal sheets, with/without VEGF-A added to the serum-free culture medium.

Fig. 2. Epidermal and dermal parameters of human aged skin before and after transplantation onto old and young mice.

Before transplantation: Rete-ridge structures were clearly observed in sections of human young skin, whereas old skin is characterized by a marked flattening of the dermoepidermal junction. Increased number of blood vessels and organized collagen in the dermis of the young skin in contrast to the aged one. After transplantation: (A) Increased epidermal thickness (N = 4 young donors, 4 old donors, 5 OiY mice, and 5 OiO mice), (B) Proliferation (N = 4 old donors, 8 OiO mice, and 6 OiY mice) and (C) melanocytes (N = 4 old donors, 9 OiO mice, and 8 OiY mice) in OiY mice compared with pretransplanted aged skin and OiO transplants. (D) p16^ink4a expression (N = 3 old donors, 6 OiO mice, and 6 OiY mice) in aged skin before transplantation in OiO and OiY mice and the absence in OiY mice. (E) PGC1α expression in aged skin before transplantation in OiO and OiY (N = 4 old donors, 8 OiO mice, and 6 OiY mice). (F) Expression of SIRT1 (N = 4 old donors, 7 OiO mice, and 6 OiY mice). (G) MTCO-1 (N = 4 old donors, 7 OiO mice, and 7 OiY mice). (H) Quantitation. Data were assessed by IHC from four individual donors. Four areas were evaluated per section, and three sections were analyzed per mouse. After the Shapiro-Wilk test, unpaired Student’s t test: *P < 0.05, **P < 0.01, and ***P < 0.001 or nonparametric Mann-Whitney U test: ^###P < 0.001. EP, epidermis; DER, dermis; SC, stratum corneum; SG, stratum granulosum; SS, startum spinosum; SB, stratum basale; PL, papillary layer; RL, reticular layer; H&E, hematoxylin and eosin; N.S., not significant. Scale bars, 50 μm.

Fig. 3. Biomarkers related to epidermal skin aging.

(A) The absence of epidermal filaggrin (N = 4 old donors, 8 OiO mice, and 8 OiY mice) in aged skin before transplantation and in OiO mice versus reappearance in YiO mice. (B) COL17A1 expression (N = 3 old donors, 7 OiO mice, and 7 OiY mice) along the basement membrane of pre-engrafted skin and in the OiO compared to the OiY xenotransplants and (C) MMP1 (N = 3 old donors, 7 OiO mice, and 6 OiY mice). (D) A structural disorganization and decrease in collagen fibers in the pretransplanted aged skin just as in OiO mice, while complete recovery along the dermis of OiY mice (N = 4 old donors, 7 OiO mice, and 6 OiY mice). (E) Quantitation. Data were assessed by IHC from three individual donors. Four areas were evaluated per section, and three sections were analyzed per mouse. After the Shapiro-Wilk test, unpaired Student’s t test: *P < 0.05, and **P < 0.01. EP, epidermis; DER, dermis; SS, startum spinosum. Scale bars, 50 μm.

In sharp contrast, these aging-associated features were reversed when old human skin was xenotransplanted onto young SCID/beige mice [“old-in-young” design (OiY)] (Figs. 1A, 2, and 3, and fig. S1), thus confirming human skin rejuvenation in vivo (10) in a different host mouse strain. Specifically, OiY human skin transplants showed a 2.4-fold increase in epidermal thickness (P < 0.05) and 3.1-fold higher epidermal keratinocyte proliferation in situ (P < 0.05) (Fig. 2, A and B). Epidermal rete ridge structures, whose flattening is a characteristic sign of human skin aging (9), had reappeared in OiY but not in OiO xenotransplants (Fig. 2A). Because this corresponds well to what we had seen in young nude mice (9, 10), the degree and quality of immune suppression in the host mice (12) appear to be irrelevant for the occurrence of human epidermal rejuvenation in vivo. However, we cannot exclude that some element of residual innate immune functions of the murine host might have affected the investigated antiaging results.

In addition, we saw twofold more melan A⁺ epidermal melanocytes in OiY compared with OiO human skin xenotransplants (P < 0.01) (Fig. 2C). This notable pigmentary phenomenon may be more important in a skin aging context than we had originally realized (9) because senescent melanocytes were recently reported to operate as drivers of human epidermal aging (13).

Human epidermal rejuvenation in vivo is confirmed at the molecular level

Human epidermal rejuvenation was confirmed at the molecular level: Besides a sevenfold decrease in senescence-associated β-galactosidase (β-Gal) (14) activity in the epidermis of OiY compared with OiO human skin transplants and pre-engrafted aged skin (P < 0.05) (fig. S2, A to E), the key senescence-associated marker, cyclin-dependent kinase inhibitor 2A, (p16^INK4a) (15), decreased even 10-fold in the epidermis of OiY compared with OiO human xenotransplants (P < 0.001) (Fig. 2D). In addition, three other key parameters that progressively decline in aging human skin, peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1α), sirtuin-1 (SIRT1), and mitochondrially encoded cytochrome c oxidase (MTCO-1) (11, 16–18), were significantly up-regulated in the epidermis of OiY compared with OiO xenotransplants and pretransplantation aged human skin (Fig. 2, E to H).

The characteristic aging-associated decline of the skin barrier (11) was also reversed, as indicated by a 2.9-fold increase in the protein expression of filaggrin, a key hygroscopic protein that determines epidermal barrier function, in OiY xenotransplants (P < 0.01) (Fig. 3A). Moreover, epidermal expression of the antiaging collagen 17A/BP180, which is critical to protect epithelial stem cells from aging (19, 20), significantly increased along the basement membrane of OiY human epidermis in vivo compared to its low expression level in pre-engraftment aged human skin and OiO xenotransplants (Fig. 3B). In addition, the high level of epidermal matrix metallaprotease-1 (MMP1) expression typically seen in aged human skin (11) was significantly reduced in the OiY human skin transplants (Fig. 3C).

The key regulator of skin oxidative damage responses, the transcription factor nuclear factor–like 2 (NRF2) (17, 18), whose activity declines during skin aging (21), increased 5.3-fold (P < 0.001) (fig. S1A). This corresponded to a significantly increased protein expression of the key downstream antioxidant enzymes regulated by NRF2 (22) in OiY human skin compared with the pretransplantation status of aged human skin: Heme oxygenase-1 (HO-1) (22) showed a 4.5-fold increase (fig. S1B), and peroxiredoxin 1 (PRDX) (23) showed a 5-fold increase (fig. S1C), while glutathione reductase (GSR) (23) increased 4.2-fold (fig. S1, D and E). This suggests that transplantation into a young mouse environment may lead to an enhanced ability of aged human skin to cope with oxidative stress (17, 18).

Human skin rejuvenation by a young mouse environment extends to the dermis and is associated with enhanced angiogenesis and reduced inflammaging

Next, we asked whether skin rejuvenation extended to the dermis of aged human skin transplants and is associated with angiogenesis. We found a 2.5-fold increase in thick dermal collagen bundles in OiY skin (P < 0.05), while the abundance of thin collagen filaments decreased in OiY xenotransplants compared with pre-engraftment aged skin (Fig. 3, D and E). This is important because thick collagen bundles decline in both number and diameter during extrinsic and intrinsic skin aging (20). Moreover, the histochemically assessed abundance of elastin expression, which also declines during skin aging (24), was significantly higher in OiY compared with OiO and in pre-engraftment skin (P < 0.05) (fig. S3A). Similarly, staining of hyaluronic acid–binding protein (HABP), another key parameter in human skin aging whose dermal content declines with progressing senescence (25), was more intense in OiY compared with OiO (P < 0.01) and to pre-engraftment aged skin (P < 0.01) (fig. S3, B and C).

OiY human xenotransplants also showed a 2.4-fold increase in the number of human blood vessel cross sections, a key vascularization parameter indicative of angiogenesis (26), 4 weeks after transplantation compared with their pretransplantation status (P < 0.05) (Fig. 4A). The increased angiogenesis within the human skin transplants could have arisen from the much better vascularized young mouse skin environment (fig. S4, A to C). However, double immunostaining of endothelial cells with species-specific CD31 antibody showed a 2.8-fold increase in human compared with murine blood vessels (P < 0.01) in the OiY xenotransplants. Instead, the number of CD31⁺ murine blood vessel cross sections in the transplant bed of OiY mice remained essentially unaltered (Fig. 4B). Thus, the young mouse environment promoted angiogenesis from resident human endothelial cells within the aged human skin transplants.

Fig. 4. Density of human dermal blood vessels before and after transplantation onto old and young mice.

(A) A significantly increased number of CD31⁺ blood vessels in OiY mice compared with OiO and pretransplanted skin. (B) Double staining of human (red; arrows) and murine (green; arrowheads). EP, epidermis; DER, dermis. Scale bars, 50 μm. (C) RNA-seq analyses in OiY mice. DEGs in the first week versus pretransplant included numerous transcripts related to angiogenesis and for 2 and 4 weeks after transplantation. (D) SCTST on the RNA-seq data identified six trajectory clusters. (E) Volcano plots of the distribution of –log₁₀ (P values) versus the gene expression fold changes. Genes with fold change >2 and P value <0.05 are indicated in green, and genes with fold change <−2 and P value <0.05 are indicated in red. The up-regulated DEGs related to promoting angiogenesis are overimposed on the volcano plot, as well as the down-regulated DEGs that are known to inhibit angiogenesis. (F) qRT-PCR time series profiles of the up-regulation of HIF1A, as well as angiogenesis-related genes such as CXCL1 and CXCL5, MMP9, PGF, LCN2, and ESM1. The red and blue lines mark the expression trajectories of two different patients.

Double immunostaining for the mast cell-specific marker, tryptase, and human HLA-A, HLA-B, and HLA-C also demonstrated a significantly increased number of human dermal mast cells in the OiY versus OiO setting (P < 0.001) (fig. S5, A to C). This could be relevant in the current context because mast cells can promote angiogenesis, including by the secretion of VEGF-A (27), and participate in extracellular matrix remodeling.

In addition, RNA sequencing (RNA-seq) analysis demonstrated a significant threefold decrease in CXCL9 gene expression, an independent, most recently identified key marker and driver of tissue inflammaging in both the human and murine systems (28), in OiY versus pre-engraftment aged skin (P < 0.03). This indicates that the OiY design also reduced human skin inflammaging, a recognized major driver of skin aging (28, 29).

“Young” human skin maintains its phenotype within an aged mouse environment

Although this is clinically irrelevant, we were curious to learn whether “young” human skin undergoes accelerated aging following transplantation into old mice [“young-in-old” (YiO) xenotransplants]. A 10-fold increase in β-Gal cells in the aged murine epidermis around the young human xenotransplants confirmed the senescent nature of the mouse host environment (fig. S2, F to I). However, no significant changes in the examined aging-related readout parameters, such as p16^INK1a expression, epidermal proliferation and melanocyte count, filaggrin expression, and dermal blood vessel density, were seen between YiO and “young-in-young” (YiY) xenotransplants 1 and 3 months after transplantation (figs. S6, A to G, and S7, A to F). Thus, “young” human skin appears to be relatively resistant to potential senescence-promoting signals released by an aged mouse environment, at least under the current assay conditions and for the time span examined.

Gene expression profiling correlates skin rejuvenation with up-regulated VEGF-A signaling and angiogenesis

To explore the unknown molecular mechanisms underlying human skin rejuvenation in vivo in a hypothesis-free manner, we performed RNA-seq–based gene expression profiling of full-thickness OiY human skin transplants (Fig. 4, C and D). To obtain a dynamic transcriptomics pattern, human skin from two different aged donors was analyzed before transplantation and 1, 2, and 4 weeks after transplantation. The number of differentially expressed genes (DEGs) between OiY and aged skin before transplantation was highest in the first week after transplantation (Fig. 4, C and D). Stable clustering of time series trajectories (SCTST) analysis of the transcriptomic dynamics identified six trajectory clusters (Fig. 4D) that covaried through the time points, with the highest variation being observed in the first week. Intriguingly, most of these clusters highlighted angiogenesis-related genes.

In the first week after aged human skin transplantation onto young mice (OiY) versus the pretransplantation status, DEGs revealed an up-regulation of many transcripts of genes related to angiogenesis and/or hypoxia pathways, such as VEGF-A, ANGPTL4, HIF1A, HIF3A, and CXCL5 (30–32), as well as TGFB3 and IGF (but not HGF) (Fig. 4E). Volcano plots were used to visualize statistically significant changes in the transcription of angiogenesis-regulatory genes in OiY xenotransplants compared with the pre-engrafted aged skin. Multiple angiogenesis-related genes were identified among the up-regulated DEGs between pretransplantation and 1 week after transplantation (Fig. 4E): ESM-1 and Apelin, whose transcription is up-regulated by VEGF-A (33, 34) and which are important regulators of endothelial cell biology (33–36) and wound healing (33, 36); PGF (37), which elicits strong angiogenic responses similar to those promoted by VEGF-A protein (37, 38); CXCL1 and CXCL5 (both of which promote angiogenesis and are prominently expressed during wound healing) (31); and the aging-associated metalloproteinase MMP9 (39), a key mediator of tissue remodeling and angiogenesis (39). In contrast, several genes known to inhibit angiogenesis showed reduced transcription in OiY human skin compared with aged skin before transplantation, namely, DKK2 and SEMA3A (40–42) (Fig. 4E).

Collectively, this suggested a key role of angiogenesis-related signaling pathways in the observed human skin in vivo rejuvenation. This hypothesis was further explored by quantitative reverse transcription polymerase chain reaction (qRT-PCR) analysis, which confirmed a selective up-regulation of VEGF-A mRNA, but not of VEGFB or any of the VEGF receptors, in OiY human skin (Fig. 4F). HIF1A and other selected angiogenesis-related genes (37, 43–45), namely, the proangiogenic chemokines CXCL1 and CXCL5 (31), as well as MMP9, PGF, LCN2, and ESM1, were also confirmed independently to be up-regulated by qRT-PCR in aged human skin transplants (OiY) at 1 and 2 weeks after transplantation compared with aged human skin before transplantation (Fig. 4F). Thus, an up-regulation of VEGF-mediated angiogenesis pathways preceded the start of the morphologically detectable human skin rejuvenation process.

Human skin rejuvenation by a young mouse environment is lost over time with progressing host age

To understand the long-term dynamics of the observed skin rejuvenation phenomenon, we also asked whether it persists while host mice age progressively. To this end, OiY xenotransplants were allowed to age for more than 1 year on initially young, progressively aging mice. This revealed that all the key morphological and molecular signs of human skin rejuvenation observed in the OiY human skin xenotransplants 30 days after transplantation (Figs. 2 and 3and figs. S1 to S3) were lost/reversed over time: After about 1 year on aging SCID/beige mice, the xenotransplants essentially displayed the aging profile seen in aged human skin before transplantation onto young mice (Fig. 5 and figs. S8 and S9). For example, the number of human β-Gal⁺ cells and MMP1 immunoreactivity, which had declined significantly in OiY xenotransplants, returned to the high levels seen in aged human skin before transplantation, while the up-regulation of PGC1α, SIRT1, and MTCO-1 expression seen after 1 month in the OiY xenotransplants vanished after 12 months (Fig. 5 and fig. S2, F to H). Thus, human skin rejuvenation by a young mouse environment gets lost over time while host mice age progressively.

Fig. 5. Biomarkers related to epidermis of aged skin in pretransplanted skin in YiO and after 12 months on the mice.

Expression of (A) MMP1 (N = 3 old donors, 4 OiY mice after 1 month, and 6 OiY mice after 12 months), (B) PGC1α (N = 3 old donors, 5 OiY mice after 1 month, and 6 OiY mice after 12 months), (C) SIRT1 (N = 3 old donors, 5 OiY mice after one month, and 6 OiY mice after 12 months), and (D) MTCO-1 (N = 3 old donors, 4 OiY mice after 1 month, and 6 OiY mice after 12 months) in epidermis of pretransplanted aged skin, in OiY and in 12-month-aged transplants. (E) Quantitation. Four areas were evaluated per section, and three sections were analyzed per mouse. After the Shapiro-Wilk test, Student’s t test: *P < 0.05. EP, epidermis; DER, dermis. Scale bars, 50 μm.

A young mouse environment up-regulates human VEGF-A expression in the human xenotransplants

Therefore, we next investigated the localization of intracutaneous human and mouse VEGF-A transcripts by in situ hybridization (ISH) using human- and mouse-specific probes. This revealed prominent human VEGF-A mRNA expression in human epidermal keratinocytes and in blood vessel–associated dermal cells within the OiY transplants (Fig. 6, A and B). The number of human VEGF-A mRNA⁺ cells significantly increased in the human epidermis (OiY compared to OiO) (Fig. 6, A and B), further confirming our RNA-seq results. Instead, mouse VEGF-A mRNA was almost undetectable in human skin OiY or OiO xenotransplants by ISH.

Fig. 6. VEGF-A RNA and protein in OiO and OiY xenotransplants.

(A) In situ hybridization for human VEGF-A in OiO and OiY transplants (N = 3 OiY mice and 3 OiO mice). (B) Quantification of the number of hVEGF-A⁺ cells in n = 3 xenotransplants per group. Means ± SEM, n = 9 microscopic fields per xenotransplant. For each xenotransplant, n = 3 nonconsecutive sections were evaluated, and for each section, n = 3 microscopic fields were analyzed. (C and D) VEGF-A protein expression in OiO (N = 4 mice) versus OiY (N = 3 mice). (E) Quantitative analysis. Data were assessed by IHC from three individual donors. Four areas were evaluated per section, and three sections were analyzed per mouse. After the Shapiro-Wilk test, Student’s t test or nonparametric Mann-Whitney test for ISH: *P < 0.05, and **P < 0.01. Scale bars, 50 μm.

However, qualitatively more murine VEGF-A mRNA was detected in the peri-transplant epidermis and blood vessels of young mouse recipients compared with old host mice (Fig. 6A). Therefore, given that murine VEGF-A can bind with high affinity to human VEGF receptors (46), in our model, both human and murine VEGF-A protein could have promoted human skin rejuvenation in vivo. Additional mouse-derived factors that up-regulate the expression of genes known to promote VEGF-A transcription/production may contribute to the latter’s induction in the xenotransplants. Because hypoxia-inducible factor α (HIFα) (47), PGC1α (48), LCN2 (Lipocalin-2) (49), and FUT2 (fucosyltransferase 2) (50), which were up-regulated in the OiY xenotransplants, all increase VEGF-A mRNA and/or protein expression, searching next for mouse-derived secreted mediators that stimulate the transcription of these genes would seem particularly productive when attempting to identify non-VEGF factors emanating from a young mouse host.

Because mRNA and protein abundance do not always correlate (51), we also determined the VEGF-A protein expression profile in the xenotransplants. This demonstrated significantly higher human and murine VEGF-A protein expression in OiY (Fig. 6C) compared with OiO human xenotransplants (Fig. 6, C to E). Quantitative immunohistomorphometry revealed an 11-fold increase in human VEGF-A protein expression by differentiated epidermal keratinocytes [double immunohistochemistry (IHC) and immunofluorescence (IF) for cytokeratin 10 and human VEGF-A] (P < 0.001) (Fig. 7A) (52), a 6.3-fold increase by human CD68⁺ macrophages (P < 0.001) (Fig. 7B) (53), and an 11-fold higher human VEGF-A protein expression by human CD42b⁺ platelets (P < 0.001) (Fig. 7, C and D) (54) in OiY versus OiO xenotransplants and pretransplantation aged skin [Note that keratinocytes (52), macrophages (53), and platelets (54) can all synthesize VEGF-A protein.] Collectively, these transcript and protein expression data pointed to VEGF-A–mediated angiogenesis as a key event in the rejuvenation of aged human skin in a young mouse environment.

Fig. 7. Expression of VEGF-A by keratinocytes, macrophages, and platelets in pretransplanted aged skin, in OiO, and in OiY xenotransplants.

Double IHC staining revealed decreased VEGF-A protein expression (A) by cytokeratin 10 (expressed by keratinocytes) (N = 3 old donors, 5 OiO mice, and 7 OiY mice), (B) by CD68 (macrophages) (N = 3 old donors, 5 OiO mice, and 7 OiY mice), and by (C) CD42b (platelets) (N = 4 old donors, 5 OiO mice, and 5 OiY mice). (D) Quantitative analysis. Data were assessed by IHC from three individual donors. Four areas were evaluated per section, and three sections were analyzed per mouse. Nonparametric Mann-Whitney U test: ^##P < 0.01 and ^###P < 0.001. Scale bars, 50 μm.

VEGF-A is required for the rejuvenation of aged human skin in vivo

Therefore, we next asked whether VEGF-A protein is critical for the rejuvenation of aged human skin by injecting antibodies intradermally into human OiY xenotransplants that neutralize both human and murine VEGF-A proteins (55) (Fig. 8, A to J). As controls, appropriate isotype immunoglobulin G (IgG) controls, anti-Hepatocyte growth factor– or anti–insulin-like growth factor 1 (IGF1)–blocking antibodies, or a small-molecule human/mouse transforming growth factor–β1 (TGF-β1) inhibitor were injected (table S2 and Fig. 1C). The controls was chosen since all four growth factors potently promote angiogenesis in various systems (56–59) and are generated constitutively in substantial quantities in human and mouse skin and at increased levels under wound healing conditions, including posttransplantation wound healing. Activity/functionality of all used antibodies was demonstrated by reduced/absent phosphorylation of the relevant key downstream molecules of each of the respective signaling pathways (VEGF, IGF-1, HGF, and TGF-β) (fig. S10, A to E).

Fig. 8. Epidermal and dermal parameters of human aged skin before and after transplantation onto young mice treated with VEGF-blocking antibodies.

Each group included 11 OiY mice transplanted with skin from six human donors (age range, 77 to 83 years; mean, 81). Injections to OiY decreased skin aging–related biomarkers compared to OiY injected with isotype control as follows: (A) epidermal thickness, (B) proliferation (Ki-67⁺ keratinocytes), (C) number of melanocytes (Melan-A⁺ cells), (D) PGC1α, (E) SIRT1, (F) mast cells (c-KIT⁺), and (G) CD31⁺ blood vessels, as well as the (H) proportion of thick and thin collagen fibers and (I) collagen 17A. (J) Quantitation. For all tested markers: N = 4 old donors, 6 OiY mice injected with control Abs, and 8 OiY mice injected with VEGF-blocking Abs. Data were assessed by IHC from four individual donors. Four areas were evaluated per section, and three sections were analyzed per mouse. After the Shapiro-Wilk test, Student’s t test: *P < 0.05, **P < 0.01, and ***P < 0.001, or nonparametric Mann-Whitney U test: ^##P < 0.01. EP, epidermis; DER, dermis; SG, stratum granulosum; SB, stratum basale; PL, papillary layer; RL, reticular layer; Abs, antibodies. Scale bars, 50 μm.

As expected for VEGF-A–blocking antibodies (56), the number of CD31⁺ blood vessel cross sections decreased twofold (P < 0.05) after antibody injection compared with OiY with nonspecific antibodies (Fig. 8, G and J), thus attesting to antibody functionality. Only VEGF-A–neutralizing antibodies robustly prevented rejuvenation of aged human skin xenotransplants (OiY) in vivo (Fig. 8, A to J): Antibody administration significantly reduced the increase in epidermal thickness by 30% (P < 0.01) (Fig. 8, A and J), epidermal proliferation by 90% (Fig. 8, B and J), and epidermal barrier-associated filaggrin expression (P < 0.01) (fig. S11, A and E). The number of melanocytes was significantly reduced 1.8-fold compared with the isotype control–injected xenotransplants in the OiY setting (Fig. 8, C and J), as was the epidermal expression of PGC1α (threefold, P < 0.01) (Fig. 8, D and J), SIRT1 (2.5-fold, P < 0.01) (Fig. 8, E and J) (28, 30), and NRF2 (1.8-fold, P < 0.05) (fig. S11, B and E). The number of epidermal and dermal c-Kit⁺ cells was also significantly reduced [7.2-fold (P < 0.05) and 5-fold (P < 0.05), respectively] (Fig. 8, F and J). Moreover, the number of thick collagen bundles significantly decreased, while that of thin collagen filaments increased (Fig. 8, H and J) compared with OiY treated with nonspecific antibodies. In comparison to isotype control–injected OiY human skin, several other skin aging readout parameters no longer showed rejuvenation-associated changes after VEGF-A neutralization antibody injection, such as collagen 17A (threefold reduction, P < 0.01) (Fig. 8, I and J), elastin staining intensity (fig. S11, C and E), and HABP expression (3.9-fold reduction, P < 0.05) (fig. S11, D and E).

On the gene expression level, RNA-seq analysis showed a down-regulation of the transcription of major proangiogenic genes in aged human skin xenotransplants when these were treated with anti–VEGF-A protein compared with nonspecific antibodies (fig. S12). Instead, the TGF-β, HGF, or IGF-1 inhibition experiments failed to significantly affect the investigated skin rejuvenation readout parameters (fig. S12, B and C). Thus, only VEGF-A protein was required for the observed human skin rejuvenation phenotype in vivo.

VEGF-A is required and suffices for rejuvenation of aged human skin in vivo

This was further investigated by gain-of-function experiments, which probed whether human VEGF-A protein alone is also sufficient to induce rejuvenation of aged human skin in an old mouse environment (OiO). For this, PLGA nanoparticles were loaded with either recombinant human VEGF-A protein or bovine serum albumin (BSA; as a negative control) (60) and injected intradermally into OiO xenotransplants in vivo.

In the OiO setting, intradermal injection of PLGA–VEGF-A–loaded nanoparticles sufficed to induce a 2.2-fold increase in epidermal thickness (P < 0.05) (Fig. 9) and a 6.5-fold increase in epidermal proliferation (P < 0.01) compared to PLGA-BSA–loaded controls (Fig. 9B). Filaggrin expression was up-regulated 3.4-fold (P < 0.05) (Fig. 9C). The number of melan A⁺ melanocytes also increased significantly (P < 0.03) (Fig. 9D), as did the number of c-Kit⁺ cells (P < 0.05; Fig. 9E) and CD31⁺ blood vessel cross sections (P < 0.05) in aged human skin xenotransplants treated with PLGA–VEGF-A–loaded nanoparticles within an old mouse host environment (OiO) (Fig. 9F). A decrease in thin collagen filaments (P < 0.01) was contrasted by an increase in thick collagen filaments (P < 0.05) was also observed (Fig. 9, G and H).

Fig. 9. Epidermal and dermal parameters of human aged skin before and after transplantation onto young mice treated with VEGF-A loaded nanoparticles.

Intradermal injections to OiO mice increased skin aging–related biomarkers compared to OiO injected with BSA-loaded PLGA nanoparticles and to aged skin before transplantation. The biomarkers were as follows: (A) epidermal thickness (N = 3 old donors, 7 OiO mice injected with PLGA-VEGF, and 6 OiO mice injected with PLGA-BSA), (B) proliferation (Ki-67⁺ keratinocytes) (N = 3 old donors, 7 OiO mice injected with PLGA-VEGF, and 6 OiO mice injected with PLGA-BSA), and (C) differentiation (filaggrin) (N = 3 old donors, 7 OiO mice injected with PLGA-VEGF, and 7 OiO mice injected with PLGA-BSA), (D) number of melanocytes (Melan-A⁺ cells) (N = 3 old donors, 8 OiO mice injected with PLGA-VEGF, and 6 OiO mice injected with PLGA-BSA), (E) mast cells (c-Kit⁺) (N = 3 old donors, 8 OiO mice injected with PLGA-VEGF, and 7 OiO mice injected with PLGA-BSA), (F) CD31⁺ blood vessels (N = 3 old donors, 7 OiO mice injected with PLGA-VEGF, and 6 OiO mice injected with PLGA-BSA), and (G) increased thick dermal collagen bundles while thin collagen filaments decreased and quantitation (N = 3 old donors, 7 OiO mice injected with PLGA-VEGF, and 7 OiO mice injected with PLGA-BSA). (H) Quantitation. Data were assessed by IHC from four individual donors. Four areas were evaluated per section, and three sections were analyzed per mouse. After the Shapiro-Wilk test, Student’s t test: *P < 0.05 and **P < 0.01, or nonparametric Mann-Whitney U test: ^#P < 0.05. EP, epidermis; DER, dermis; SG, stratum granulosum; SB, stratum basale; PL, papillary layer; RL, reticular layer. Scale bars, 50 μm.

RNA-seq analyses demonstrated that VEGF-A protein–loaded nanoparticles exerted the expected proangiogenic effects in aged human skin xenotransplants, showing up-regulated transcription of angiogenesis-regulatory factors such as STAT3, LCN2, HIF1A, CLC2A1, VEGF-A, FLT3, PRKCB, CXCL1, ID2, CXCL2, FUT2, and NRF2 (fig. S12A) (31, 41, 47, 48, 50, 61, 62). The old SCID/beige mice used in our study had a significantly higher serum cortisol level than the young mice (1.7 ± 0.5 versus 1.09 ± 0.2 μg/dl, P < 0.05). Thus, VEGF-A protein exerted its skin rejuvenation effects even in the presence of aging-promoting higher serum cortisol levels (excess glucocorticoid serum levels promote skin thinning, impair wound healing, and may accelerate skin aging) (63). In summary, VEGF-A protein–mediated signaling is not only required but also sufficient for human skin rejuvenation in vivo, even in an old mouse environment.

VEGF-A improves human skin aging parameters ex vivo, independent of functional murine vasculature, innervation, systemic, or other mouse-derived factors

Last, we asked whether functional vasculature, skin innervation, or other systemic factors provided by a murine host, and secreted signals from a mouse environment, are critically required for the observed skin rejuvenation effects of VEGF-A, as this would obviously limit the clinical relevance of our findings. Therefore, recombinant human VEGF-A protein was administered to organ-cultured aged full-thickness human skin (64) or to enzymatically prepared human epidermal sheets (Fig. 1D) in serum-free supplemented medium as previously described (64). This experimental design excludes all confounding vascular, neural, and systemic factors, as well as extracutaneous cell populations (63), not only from mice but also those impacting on human skin under clinical conditions.

Even under these very restrictive ex vivo conditions, human VEGF-A protein significantly improved multiple epidermal and dermal skin aging parameters of old human skin over 6 days of organ culture compared with the corresponding expression level of aged human skin before VEGF-A stimulation ex vivo (figs. S13 to S15). For example, VEGF-A protein significantly increased protein expression of SIRT1 (P < 0.05) (fig. S15A), PGC1α (P < 0.05) (fig. S15B), and MTCO-1 (P < 0.05) (fig. S15C) in aged human epidermis ex vivo. These short-term ex vivo pilot experiments show that exogenous human recombinant VEGF-A protein alone exerts substantial human skin rejuvenation properties, independent of systemic or neural inputs or a young mouse host environment (for details, see the Supplementary Results).

DISCUSSION

Here, we provide the first evidence that VEGF-A–mediated signaling is both required and sufficient for rejuvenation of a relatively fast-aging human organ, skin, at both the level of morphology and molecular aging markers. This identifies a pharmacologically pliable master pathway for human organ rejuvenation. The fact that the rejuvenation effect is seen not only in vivo but also ex vivo in human skin organ-cultured in the absence of serum and in the presence of antibiotics demonstrates that neural, microbial, and systemic contributions (including extracutaneous hormonal, nutritional, microbial, and metabolic ones) are dispensable for VEGF-A–induced rejuvenation of aged human skin.

However, it is unlikely that VEGF-A is the only factor involved in the skin rejuvenation phenomenon since the VEGF-A–neutralizing antibodies did not completely abrogate the rejuvenation effect. This renders it plausible that other elements emanating from mouse skin could contribute to rejuvenation of the aged human skin xenotransplant (see below) yet are not essential for this phenomenon to occur. Yet, because we were seeking to identify a single pharmacologically targetable factor for which well-tested drugs (including peptidomimetics) are already available, thus facilitating the clinical implementation of our study outcomes, we decided to focus exclusively on the role of VEGF-A, one of the best-defined growth factors (65).

Our study provides strong human data support for the “angiogenesis hypothesis of aging,” which postulates a decline in angiogenic factor production and subsequently reduced capillary density underlying physiological, intrinsic tissue aging, and that tissue aging may therefore be reversed by angiogenesis-promoting therapy (66, 67). Our clinically relevant data, which were generated directly in human skin, largest and fastest-aging human organ, rather than in mice (65–67), also underscore that human skin aging, just like human hair graying (5), is reversible, at least temporarily, and is druggable even in the human system in vivo (2, 67).

The current data in human skin are perfectly in line with a very recent mouse study by Grunewald et al. (65), which states that aging-related phenotypes are alleviated or even reversed in VEGF-treated mice as a result of angiogenic and geroprotective activities of VEGF, although skin was not examined. Our study now provides the first evidence that the antiaging effects of VEGFR-mediated signaling observed in different murine organs (65) translate to the human body’s largest organ. However, our study goes beyond this by showing that recombinant human VEGF-A protein induces skin rejuvenation in aged human skin even under organ-culture conditions, i.e., in the absence of a functional vasculature and in serum-free culture medium. The human skin organ culture data do not exclude that systemic or neural inputs affect the human skin xenotransplant rejuvenation phenomena observed in vivo. However, our ex vivo data clearly show that systemic or neural inputs are not required to induce an early molecular skin rejuvenation signature for which human VEGF itself suffices. This indicates that VEGF-A protein does not act only by promoting blood flow and enhancing angiogenesis, thereby improving access to blood-borne senolytic hormones, growth factors, metabolites, polyamines, and antioxidants. It deserves to be studied, next, whether the most recently identified VEGF signaling insufficiency in aging mice as a result of increased production of decoy receptors (65) is also at play during human skin aging. If yes, this would make it even more compelling to promote VEGF-A–dependent signaling as an antiaging strategy.

Mechanistically, we identify VEGF-A protein by loss- and gain-of-function experiments as the one required and sufficient senolytic signal that a young mouse host environment sends to rapidly initiate the rejuvenation of aged human skin xenotransplants in vivo. A mechanistic key role of angiogenesis and hypoxia-related pathways in this is also supported by our gene expression profiling data. Yet, our experiments do not permit us to state with certainty whether locally and/or systemically released murine VEGF-A protein or intracutaneously released human VEGF-A protein from the skin xenotransplant is the main tissue rejuvenation driver and what controls VEGF-A protein production and secretion in young versus old mice. However, young mice have a much higher VEGF level than old mice, and VEGF protein expression in the human xenotransplants also increases (Fig. 6). Therefore, it is plausible to postulate that initially, young mouse VEGF-A stimulates human VEGFR (68), which leads to a positive feedback loop to further promote the secretion of VEGF in aged human skin [note that VEGF-A is known to induce a VEGFR-dependent positive feedback loop (69)]. The increased level of combined murine and human VEGF then likely triggers a cascade of proangiogenic and skin rejuvenation events, as summarized in Fig. 10.

Fig. 10. Schematic representation of human aged skin before and after transplantation onto young SCID beige mice.

(A) Schematic illustration showing the characteristic changes seen in human skin aging process with decreased epidermal thickness and proliferation, flattened epidermal rete ridges, senescent dermal fibroblasts, epidermal keratinocytes and melanocytes, impaired angiogenesis, decreased upstream and downstream regulators involved in the induction of VEGF-A, and increased biomarker signatures of skin aging. (B) Summary of the proposed rejuvenation cascade of aged human skin after transplantation to young mice (OiY). High-level mouse VEGF-A may initiate this cascade by stimulating human VEGFRs in the xenotransplant, leading to a positive feedback loop that enhances the secretion of human VEGF-A within in the xenotransplant and triggers the indicated cascade of proangiogenic and skin rejuvenation signaling events. The rejuvenation effect can be antagonized significantly, although not entirely abrogated, by neutralizing antibodies that recognize both human and mouse VEGF-A. Thus, additional signals besides VEGF-A secretion that emanate from young mice may contribute to the up-regulation of VEGF-A within the human, namely, mouse-derived signals that up-regulate HIFα, PGC1α, LCN2, and FUT2 expression in human skin. Abs, antibodies; MMP1, matrix metallopeptidase 1; ROS, reactive oxygen species; SEMA3A, semaphorin 3A; SASP, senescence-associated secretory phenotype; RR, rete ridges.

While our ex vivo experiments demonstrate that human VEGF-A alone suffices to induce the early molecular signature of skin rejuvenation in aged human skin, we also observed an up-regulation of HIF1α (47), PGC1α (48), LCN2 (49), and FUT2 (50) expression in human skin in vivo (exclusively in the OiY design). Therefore, a conceivable additional mechanism for how VEGF may be up-regulated within aged human skin is via (as yet unknown) secreted young mouse–derived senolytic signals that up-regulate human HIFα (47), PGC1α (48), LCN2 (49), and FUT2 (50) (see Fig. 10), which older mice are no longer capable of secreting. Our data suggest that as yet unknown secreted factors from young mice stimulated VEGF-A production in old human skin, and the latter increased VEGF-A expression drives morphological and biochemical changes that rejuvenate the aged human skin. While the rejuvenation effect could in part be the result of increased access to as yet unknown “antiaging factors” present in the blood of young mice, our ex vivo experiments suggest that human VEGF-A alone suffices to induce at least some molecular rejuvenation profile, despite the very short duration of human skin organ culture (6 days). However, our ex vivo data (figs. S13 to S15) suggest that these non–VEGF-A, young mouse–derived signals are not indispensable in humans for skin rejuvenation.

Our transcriptomic data suggest that several other pathways besides VEGF-A–related ones are up-regulated in OiY xenotransplants, such as hypoxia-related pathways, HIF1 and NRF2 signaling (Fig. 4). Moreover, the striking decline in the OiY xenotransplants of CXCL9 expression, a recently identified key marker of tissue inflammaging (28), and the increase in the number of epidermal melanocytes (Fig. 2C) suggest that the OiY design also counteracted cutaneous inflammaging and melanocyte senescence, recognized major drivers of skin aging (13, 28, 29). These additional pathways deserve systematic follow-up so as not to miss additional druggable targets for human organ rejuvenation. Moreover, in-depth mechanistic follow-up studies are needed to clarify exactly how VEGF-A protein exerts its rejuvenation effects on defined aging readout parameters (for example, along the cascade postulated in Fig. 10), whether these effects are directly or indirectly mediated, and which VEGF-A–producing cells are the most important ones in the context of tissue rejuvenation.

We have ruled out that human skin rejuvenation in vivo resulted from transplantation-associated wound healing phenomena and show that human skin rejuvenation is reversed over time as the host mice progressively age—even though a senescent mouse environment as such does not promote aging of young human skin in vivo. This intriguing observation suggests that the rejuvenation effect of a young mouse environment on human skin itself is transitory, that the aging of host mice increasingly robs the human xenotransplants of their rejuvenating host milieu, and/or that a senescing mouse environment cannot actively support human skin rejuvenation. The reversed phenotype does appear to be mainly due to progressive loss of angiogenic capacity of the murine host, namely, a reduced level of murine VEGF-A in old mouse skin (Figs. 6 to 10). Indeed, wounds from aged mice contain significantly less VEGF compared with young animals (70). Of course, we cannot exclude that this coincides with other skin aging–associated changes. Overall, however, our data support that the declining VEGF-A level is the dominant reason for the observed reversibility of human skin rejuvenation over time [this likely also explains why none of the rejuvenation phenomena were seen when aged human skin was transplanted onto old mice (OiO)]. This observation only adds to the credibility of the striking rejuvenation phenomenon reported here, since it partially imitates a crossover clinical trial design and further underscores the attractiveness of this mouse model for long-term preclinical human aging research studies in vivo.

Together, our data invite the working hypothesis that in OiY mice, initially, high levels of murine VEGF-A protein released by a young mouse environment stimulate VEGFRs in the human transplant (68), activate NRF2 (71), and initiate downstream signaling pathways relevant to angiogenesis and antioxidant responses to protect cells from damage caused by reactive oxygen species (ROS) activity (Fig. 10, A and B). In addition, VEGF signaling also up-regulates proangiogenetic gene signatures including ESM-1, Apelin, PGF, SLC2A1, CXCl1, STAT3, and PRKB gene expression (Fig. 10, A and B). To prove that mouse-derived VEGF-A initiates the rejuvenation cascade, the current in vivo experiments would have to be repeated in the presence of mouse-specific VEGF-A antibodies. Also, the complexity of VEGF-A biology and of the observed rejuvenation phenomena makes it important to systematically search for additional young mouse–derived signals that contribute to the rejuvenation effects.

We demonstrate here that VEGF-A protein also directly affects intracutaneous aging mechanisms such as replicative senescence, as well as the proliferation and differentiation of epidermal keratinocytes. VEGF is known to promote wound repair beyond stimulating angiogenesis by, for example, direct effects on melanocytes and keratinocytes (72, 73). However, the exact molecular mechanisms through which VEGF-A exerts its skin rejuvenation effects on multiple different levels remain to be determined. Moreover, the field is now challenged to dissect how exactly a single growth factor manages at the molecular level, directly or indirectly, the feat of potently up-regulating the expression of these diverse key antiaging players such as p16^ink4a, SIRT1, PGC1α, and collagen 17A, and likely also inflammaging and melanocyte senescence. The complex gene expression changes observed here suggest that other growth factors, cytokines, hormones, and extracellular vesicle–derived signals besides VEGF-A may well contribute to promoting the human skin rejuvenation phenomena in vivo reported here.

Our data raise the pertinent question whether VEGF-A protein itself or VEGF-A mimetic drugs could become new senolytic therapeutics (1, 72). However, important safety concerns arising from the well-recognized pathogenic role of VEGF-A, for example, in tumor-associated angiogenesis (49) and several autoimmune diseases (74), need to be overcome before clinical pilot studies can be justified. This must be carefully considered also in view of the known adverse effects of long-term VEGF-A overexpression because chronic VEGF-A receptor overstimulation over many years could involve considerable long-term risks (75, 76). For human skin rejuvenation purposes, it would therefore seem the most prudent strategy to explore intermittent pulse therapy regimes and develop topically applicable VEGF-A mimetics with minimal systemic absorption, such as mimetic peptides that are rapidly degraded intracutaneously, or topical agents that enhance intracutaneous production and secretion of endogenous VEGF-A protein and/or up-regulate the skin expression of cognate receptors.

Moreover, our study raises the important question of how long-term anti-VEGF therapy, e.g., with antibodies such as bevacizumab, which has now been practiced in clinical medicine for about 15 years, affects human organ aging, including and well beyond the skin. Our findings strongly encourage one to systematically examine this in-depth in patients under long-term anti-VEGF therapy.

Our observation that a single growth factor is both required and sufficient to rapidly rejuvenate a human organ in vivo and ex vivo supports the concept that the progressive relative lack of VEGF-A protein, not only the increased release of soluble VEGF receptors recently reported in aging murine organs (65), is at least one key driver of human organ aging. This provides invaluable guidance for rational senolytic drug design also beyond the skin (1, 3, 77) (for more details, see the Supplementary Discussion).

Last, our study also illustrates the power of humanized mouse models for preclinical human aging research. The field has long struggled to reliably translate the data and concepts generated in nonhuman or cell culture–based models favored by mainstream aging research into clinical reality. The increased use of our relatively simple but highly instructive, experimentally pliable, and clinically relevant xenotransplantation in vivo assay in the future would overcome this important hurdle in human aging research and greatly facilitate testing how candidate senolytics, such as VEGF-A mimetics, affect an aged human model organ (skin) in vivo.

MATERIALS AND METHODS

As previously described (9, 10), human split-thickness skin (0.4-mm) samples from non–ultraviolet (UV)–exposed upper thigh (chosen to minimize UV-induced skin photoaging as a potential study confounder) were collected from old volunteers. These human skin samples were either snap frozen in liquid nitrogen, fixed in 10% saline-buffered formalin overnight, reserved for histological stains, or xenotransplanted onto 53 old (14 ± 3.2 months) (OiO) versus 98 young (2 ± 1.1 months old) (OiY) SCID/beige mice (C.B-17/IcrHsd-Prkdc^scidLyst^bg-J) (12) (Envigo, Jerusalem). Human skin samples of young volunteers xenotransplanted onto old mice (YiO) were used as an additional control. Twelve “young” adult (43 ± 5 years old) and 14 “aged” (83 ± 5 years old) human donors were enrolled in this study after institutional review board approval and written patient consent (for details, see table S1 and Fig. 1, A and B).

VEGF-A–targeting functional manipulations

Four weeks after transplantation, OiY mice were intradermally injected with anti–VEGF-A protein, anti-HGF or anti-IGF1 antibodies, or goat IgG as isotype control (table S2 and Fig. 1C). Alternatively, the small-molecule TGF-β1 inhibitor SB431542 (78) was injected intradermally.

The downstream readouts of each pathway were tested by IHC staining and quantitative immune-histomorphometry. This confirmed at the protein level that all four antibodies neutralized their respective target proteins in all treated xenotransplants and were thus fully functional as follows: p-Smad2 in the TGF-β signaling was down-regulated by TGF-ß–neutralizing antibody (78), p-Akt in the VEGF signaling pathway by VEGF-neutralizing antibody (26), p-IRS in the IGF-1 pathway by the IGF-1 antibodies (79), and p-c-Met and p-STAT3 as direct downstream molecules in HGF signaling by the corresponding antibodies (80) (fig. S10).

VEGF-A–loaded nanoparticles were injected intradermally on days 6 and 15 after engraftment into skin xenotransplants from three aged donors (ages 80 to 92 years; mean, 86) on 32 old SCID/beige mice (mean age, 13.0 months) (tables S1 and S3) (for details and explanations, see the Supplementary Materials and Methods).

Human skin organ culture

Fifteen 4-mm fragments of full-thickness skin from two female human skin donors (ages 76 and 81 years) were organ cultured at the air-liquid interface in serum-free Williams’ E medium supplemented with antibiotics, insulin, l-glutamine, and hydrocortisone as described (56, 64) and treated ex vivo for 6 days with VEGF-A protein (0.1 μg/1 ml) (Fig. 1D) (see the Supplementary Materials and Methods).

Histochemistry, IHC, and quantitative immunohistomorphometry

Human skin samples were analyzed by histochemistry, IHC, and IF microscopy as described (Fig. 1, B and D) (16) using the following established aging-associated readout parameters: β-Gal activity (14), SIRT1 (17, 18), PGC1α (17, 18), p16^ink4a (15), NRF2 (71), HO-1 (22), peroxireduxin (23), GSR (23), MMP1 (20), MTCO-1 (16), thick/thin collagen bundles (20), collagen 17A/BP180 (19), elastin (24), and HABP (25).

In addition, we performed single or double IHC/IF for the following markers: Ki-67 (proliferation) (10); melan A (melanocytes) (13); filaggrin (epidermal barrier) (11); CD31 (endothelial cells) (56); c-Kit; tryptase/HLA-A, HLA-B, and HLA-C (mast cells) (27); VEGF-A (47); VEGF-A/cytokeratin-10 (52); VEGF-A/CD68 (macrophages) (53); and VEGF-A/CD42b (platelets) (54). Human skin xenotransplants were examined before and 30 days after transplantation (for details on primary and secondary antibodies, controls, and explanations, see the Supplementary Materials and Methods, tables S4 and S5).

VEGF-A ISH

ISH was carried out in human skin samples obtained before and 30 days posttransplantation by using the Single-plex RNAscope Fast Red Assay (Advanced Cell Diagnostics, Newark, CA, USA) and custom-designed, species-specific ISH probes (from Advanced Cell Diagnostics to exclude species cross-reactivity) for mouse VEGF-A mRNA, human VEGF-A mRNA, human PPIB, mouse PPIB (positive controls), and DapB (negative control) (for details and explanations, see the Supplementary Materials and Methods).

Gene expression profiling

Total RNA was extracted from full-thickness xenotransplants using the miRNeasy Micro Kit (Qiagen, Germany), and RNA concentration was determined using a Qubit RNA BR assay kit (Thermo Fisher Scientific, USA). Transcriptomic analyses were performed by qRT-PCR or RNA-seq and subsequent computational biology analysis (for details, including data processing, pathway analysis, and data mining such as determination of SCTST, see the Supplementary Materials and Methods).

Serum cortisol measurement

Serum from six young mice (2 months) and six old mice (14 months) was obtained and analyzed immediately for serum cortisol using the Immulite analyzer and DPC kits (Siemens Healthineers, Germany) (for details, see the Supplementary Materials and Methods).

Patient and public involvement in research

Patients or the public was not involved in the design, conduct, reporting, or dissemination plans of this preclinical research project.

Statistical analysis

Data are presented as the means ± SEM or fold change of means ± SEM; P values of <0.05 were regarded as significant. Gaussian distribution of the data was analyzed using the Shapiro-Wilk test. Significant differences were analyzed using either the unpaired Student′s t test (comparison between one set of data) or one-way analysis of variance (ANOVA; comparison between multiple sets of data) for parametric data, or the Mann-Whitney test (comparison between one set of data and sham or vehicle) for nonparametric data, or the Kruskal-Wallis test and the Dunn’s test (comparison between multiple sets of data). The relevant statistical analysis method and the n (e.g., number of donors, tissue sections, or microscopic fields) used for each dataset reported here are listed in the corresponding figure legends.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S15

Tables S1 to S5

References

View/request a protocol for this paper from Bio-protocol.