Senescent Dermal Fibroblasts Decrease Stemness in Basal Keratinocytes in a Bioengineered Model of Human Full-Thickness Skin

Intrinsic skin ageing is characterized by decreased regenerative capacity and impaired differentiation of epidermal keratinocytes resulting in a thinning of the epidermis, which contributes to a compromised barrier function. Cell senescence is a hallmark of ageing, and senescent fibroblasts, keratinocytes, melanocytes and others accumulate in ageing skin (Low et al., 2021Low E. Alimohammadiha G. Smith L.A. Costello L.F. Przyborski S.A. von Zglinicki T. et al.How good is the evidence that cellular senescence causes skin ageing?.Ageing Res Rev. 2021; 101456https://doi.org/10.1016/j.arr.2021.101456Crossref Scopus (43) Google Scholar). Accumulation of senescent dermal fibroblasts has been widely hypothesized as a cause of dermal and epidermal ageing. This is because senescent cells secrete a wide range of bioactive molecules termed the Senescence-Associated Secretory Phenotype (SASP) that induce ‘ageing-like’ bystander effects in neighbouring cells and tissues and because ageing could be postponed by anti-senescence interventions in many organs (Kirkland and Tchkonia, 2017Kirkland J.L. Tchkonia T. Cellular Senescence: A Translational Perspective.EBioMedicine. 2017; 21: 21-28https://doi.org/10.1016/j.ebiom.2017.04.013Abstract Full Text Full Text PDF PubMed Scopus (672) Google Scholar). However, the specific evidence for senescent fibroblasts causing epidermal ageing is still not strong (Low et al., 2021Low E. Alimohammadiha G. Smith L.A. Costello L.F. Przyborski S.A. von Zglinicki T. et al.How good is the evidence that cellular senescence causes skin ageing?.Ageing Res Rev. 2021; 101456https://doi.org/10.1016/j.arr.2021.101456Crossref Scopus (43) Google Scholar). Bioengineered skin equivalents capture the complexity of human skin in vitro and enable interactions between different cell populations to be studied. A small number of 3D skin-equivalents have been developed that test the impact of senescent dermal fibroblasts (Diekmann et al., 2016Diekmann J. Alili L. Scholz O. Giesen M. Holtkötter O. Brenneisen P. A three-dimensional skin equivalent reflecting some aspects of in vivo aged skin.Experimental Dermatology. 2016; 25: 56-61https://doi.org/10.1111/exd.12866Crossref PubMed Scopus (44) Google Scholar; Janson et al., 2013Janson D. Rietveld M. Willemze R. El Ghalbzouri A. Effects of serially passaged fibroblasts on dermal and epidermal morphogenesis in human skin equivalents.Biogerontology. 2013; 14: 131-140https://doi.org/10.1007/s10522-013-9416-9Crossref PubMed Scopus (28) Google Scholar; Weinmüllner et al., 2020Weinmüllner R. Zbiral B. Becirovic A. Stelzer E.M. Nagelreiter F. Schosserer M. et al.Organotypic human skin culture models constructed with senescent fibroblasts show hallmarks of skin aging.npj Aging and Mechanisms of Disease. 2020; 6: 1-7https://doi.org/10.1038/s41514-020-0042-xCrossref PubMed Scopus (47) Google Scholar). Results were not unequivocal: in a collagen-based matrix, senescent fibroblasts induced hallmarks of intrinsic skin ageing including epidermal thinning, reduction of basal keratinocyte proliferation, impairment of epidermal differentiation and loss of barrier function (Weinmüllner et al., 2020Weinmüllner R. Zbiral B. Becirovic A. Stelzer E.M. Nagelreiter F. Schosserer M. et al.Organotypic human skin culture models constructed with senescent fibroblasts show hallmarks of skin aging.npj Aging and Mechanisms of Disease. 2020; 6: 1-7https://doi.org/10.1038/s41514-020-0042-xCrossref PubMed Scopus (47) Google Scholar). In a collagen–glycosaminoglycan–chitosan scaffold, senescent fibroblasts reduced epidermal filaggrin as well as dermal collagen and elastin expression but had no effect on epidermal thickness (Diekmann et al., 2016Diekmann J. Alili L. Scholz O. Giesen M. Holtkötter O. Brenneisen P. A three-dimensional skin equivalent reflecting some aspects of in vivo aged skin.Experimental Dermatology. 2016; 25: 56-61https://doi.org/10.1111/exd.12866Crossref PubMed Scopus (44) Google Scholar). Finally, in a fibroblast-derived matrix in a polyester permeable support (Janson et al., 2013Janson D. Rietveld M. Willemze R. El Ghalbzouri A. Effects of serially passaged fibroblasts on dermal and epidermal morphogenesis in human skin equivalents.Biogerontology. 2013; 14: 131-140https://doi.org/10.1007/s10522-013-9416-9Crossref PubMed Scopus (28) Google Scholar), late PD (senescent) fibroblasts caused a thinner dermis and minor changes in the epidermal expression of keratin 6 and 10, but no changes in epidermal thickness, keratinocyte proliferation or basement membrane formation. A main limitation of previous skin equivalents is variability due to the use of complex protocols and of exogenous extracellular matrix (ECM) proteins. We recently developed a robust full-thickness skin equivalent that is highly reproducible due to the use of an inert porous polystyrene membrane (Alvetex®) as scaffold, commercially available cells, and defined low-serum media. The scaffold allows human dermal fibroblasts (HDF) to produce their own endogenous ECM proteins, which alleviates the need for exogenous collagen, and fully supports the differentiation and stratification of the epidermis (Costello et al., 2019Costello L. Fullard N. Roger M. Bradbury S. Dicolandrea T. Isfort R. et al.Engineering a Multilayered Skin Equivalent: The Importance of Endogenous Extracellular Matrix Maturation to Provide Robustness and Reproducibility.Methods Mol Biol. 2019; 1993: 107-122https://doi.org/10.1007/978-1-4939-9473-1_9Crossref PubMed Scopus (8) Google Scholar; Roger et al., 2019Roger M. Fullard N. Costello L. Bradbury S. Markiewicz E. O’Reilly S. et al.Bioengineering the microanatomy of human skin.Journal of Anatomy. 2019; 234: 438-455https://doi.org/10.1111/joa.12942Crossref PubMed Scopus (93) Google Scholar). Here, we bioengineered Alvetex-based human full thickness skin models containing variable fractions of senescent fibroblasts (from 0 to 100% seeding density, see suppl. Fig. S1). Dermal equivalents were matured for 28 days followed by formation of an epidermal layer for 14 days (see Supplementary Material & Methods). Four independent experiments were performed varying the senescent fibroblast seeding density between 0 and 20% (experiments 1 to 3) or between 0 and 100% (experiment 4). Models from every experiment were divided in two and technical repeats (including independent sectioning, staining, imaging and quantification) were done in separate labs with lab B being fully blinded to the fraction of senescent fibroblasts in the sample. Incorporation of senescent fibroblasts in the dermal component induced SASP components already at 20% seeding density (suppl. Fig. S2). Accordingly, there was a decrease of total collagen in the skin equivalents created with higher senescent HDF frequencies (suppl. Fig. S3). Epidermal thinning with increased senescent cell load (suppl. Fig. S4) was significant with p<0.05 (Fig. 1A). Qualitatively, there was no evidence for changed epidermal differentiation as judged by co-staining for cytokeratins K10 and K14 (suppl. Fig. S5). To test whether senescent dermal fibroblasts induce senescence in basal keratinocytes we measured cell and nuclear size, KI67 and HMGB1 expression (also in suprabasal cells). There were no changes in basal keratinocyte size (Fig. 1B), basal keratinocyte nuclear size (Fig. 1C), or HMGB1 positivity of either basal (Fig. 1D, suppl. Fig. S6A) or suprabasal (Fig 1G) keratinocytes, and there was even an increase in KI67-positive keratinocytes with increasing load of senescent fibroblasts (Fig. 1E, suppl. Fig. S6C). However, lower frequencies of basal keratinocytes were positive for the epithelial stemness marker p63 under high senescence loads (Fig. 1F, suppl. Fig. S6B). In contrast, frequencies of p63-positive keratinocytes in the suprabasal layer increased with senescent load (Fig. 1H). A 2way ANOVA on the untransformed data revealed a systematic observer bias for most parameters (Suppl. Tab S1). However, when the same analysis was performed on the normalised data, observer effects were mostly cancelled out (suppl. Tab S2). This suggested variation between experiments as the main variation source, while intra-experimental and inter-observer factors resulted in a bias in absolute values, but not in the ability to detect experimentally induced differences. Specifically, results appeared not dependent on observer blinding. In conclusion, although senescent fibroblasts induced a SASP similar to observations in aged skin (Low et al., 2021Low E. Alimohammadiha G. Smith L.A. Costello L.F. Przyborski S.A. von Zglinicki T. et al.How good is the evidence that cellular senescence causes skin ageing?.Ageing Res Rev. 2021; 101456https://doi.org/10.1016/j.arr.2021.101456Crossref Scopus (43) Google Scholar) and in collagen-based 3D skin models (Weinmüllner et al., 2020Weinmüllner R. Zbiral B. Becirovic A. Stelzer E.M. Nagelreiter F. Schosserer M. et al.Organotypic human skin culture models constructed with senescent fibroblasts show hallmarks of skin aging.npj Aging and Mechanisms of Disease. 2020; 6: 1-7https://doi.org/10.1038/s41514-020-0042-xCrossref PubMed Scopus (47) Google Scholar), they did not induce keratinocyte senescence. However, they did cause a reduction of the stemness marker p63 specifically in the basal layer, which might be the cause of the observed epidermal thinning. Matrix composition may be a major modifier of the effect of senescent fibroblasts onto the epidermal equivalent. Ageing-like epidermal changes were reported in collagen-based skin models containing senescent fibroblasts (Weinmüllner et al., 2020Weinmüllner R. Zbiral B. Becirovic A. Stelzer E.M. Nagelreiter F. Schosserer M. et al.Organotypic human skin culture models constructed with senescent fibroblasts show hallmarks of skin aging.npj Aging and Mechanisms of Disease. 2020; 6: 1-7https://doi.org/10.1038/s41514-020-0042-xCrossref PubMed Scopus (47) Google Scholar) but not in a skin model using polyester-based material (Janson et al., 2013Janson D. Rietveld M. Willemze R. El Ghalbzouri A. Effects of serially passaged fibroblasts on dermal and epidermal morphogenesis in human skin equivalents.Biogerontology. 2013; 14: 131-140https://doi.org/10.1007/s10522-013-9416-9Crossref PubMed Scopus (28) Google Scholar). It is well known that cellular senescence not only reshapes the matrix (Mavrogonatou et al., 2023Mavrogonatou E. Papadopoulou A. Pratsinis H. Kletsas D. Senescence-associated alterations in the extracellular matrix: deciphering their role in the regulation of cellular function.Am J Physiol Cell Physiol. 2023; 325: C633-C647https://doi.org/10.1152/ajpcell.00178.2023Crossref PubMed Scopus (7) Google Scholar), but extracellular matrix properties significantly impact on the senescent cell phenotype including SASP intensity and composition (Blokland et al., 2020Blokland K.E.C. Pouwels S.D. Schuliga M. Knight D.A. Burgess J.K. Regulation of cellular senescence by extracellular matrix during chronic fibrotic diseases.Clin Sci (Lond). 2020; 134: 2681-2706https://doi.org/10.1042/CS20190893Crossref PubMed Scopus (67) Google Scholar). Taken together, our data reveal the context- and probably matrix-dependency of the effects of senescent fibroblasts in human full thickness skin equivalents. Senescent fibroblasts may drive skin intrinsic ageing by different mechanisms depending on their environment, suggesting that besides interventions that reduce cell senescence there may also be others, specifically those impacting on the extracellular matrix that might have potential to reduce skin ageing. Datasets related to this article can be found at: https://data.mendeley.com/datasets/w8d49vjnkt/1, an open-source online data repository hosted at Mendeley Data. Author SP acts as a technical consultant with the company Reprocell Europe. Proctor&Gamble co-funded part of the BBSRC grant # BB/S006710/1 to TvZ. This study was performed in accordance with the Declaration of Helsinki. No human or animal experiments were performed in this study. Conceptualization. TvZ Data Curation: TvZ lead, EL, LS support Formal Analysis: EL, LS, EF, TvZ equal Funding Acquisition: SP, TvZ equal Investigation: EL, LS equal leads, EF support Methodology: LS lead, EL, EF support Project Administration: SM lead, SP, TvZ support Resources: LS lead, EL, EF support Supervision: SP, SM, TvZ equal Validation: EL, LS lead, TvZ support Visualization: EL, LS, EF, TvZ equal Writing - Original Draft Preparation: TvZ Writing - Review and Editing: All equal Neonatal human dermal fibroblasts (HDFn) were cultured in either Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 5ml of L-Glutamine (G7513-100ML), 5ml Penicillin streptomycin (PS) and 50ml Foetal Bovine Serum (FBS) Sigma (F9665-500ML) or Human Fibroblast Expansion Basal Medium (Fisher Scientific UK, M106500) supplemented with Low Serum Growth Supplement (LSGS) and (Fisher Scientific UK, S00310) and Gentamicin (Fisher Scientific, G418 Sulfate) at a 37°C humidified atmosphere with 5% CO2. Cells were bought from Life Technologies, ThermoFisher Scientific in 2017 (Lot number 1366434) and were regularly validated (by morphology and onset of replicative senescence at PD˜50) and mycoplasm-tested (bi-monthly). Neonatal human epidermal keratinocytes (HEKn) were cultured in Epilife Medium with 60 μM calcium (Fisher Scientific UK, MEPI500CA) supplemented with Human Keratinocyte Growth Supplement (Fisher Scientific UK, S0015) and Gentamicin (Fisher Scientific, G418 Sulfate) at a 37°C humidified atmosphere with 5% CO2. Cells were bought from Life Technologies, ThermoFisher Scientific in 2021 (Lot number 2018512) and were bimonthly tested for mycoplasms. To induce cellular senescence, fibroblasts were subjected to 20 Gy ionising X ray radiation. Cells were then maintained for at least 10 days to allow a fully senescent phenotype to develop (Passos et al., 2010Passos J.F. Nelson G. Wang C. Richter T. Simillion C. Proctor C.J. et al.Feedback between p21 and reactive oxygen production is necessary for cell senescence.Mol Syst Biol. 2010; 6: 347https://doi: 10.1038/msb.2010.5Crossref PubMed Scopus (718) Google Scholar). Cell numbers remained constant for at least 30 days after irradiation. The senescent phenotype was confirmed by absence of KI-67 staining, low levels of HMGB1 and laminB1 and increased nuclear size (suppl. Fig. S1). Full thickness human skin equivalents were generated using Alvetex technology as described in (Roger et al., 2019Roger M. Fullard N. Costello L. Bradbury S. Markiewicz E. O’Reilly S. et al.Bioengineering the microanatomy of human skin.Journal of Anatomy. 2019; 234: 438-455https://doi.org/10.1111/joa.12942Crossref PubMed Scopus (93) Google Scholar), with minor protocol modifications to incorporate senescent cells into the dermal compartment. To generate dermal equivalents containing senescent fibroblasts, neonatal and senescent fibroblasts were combined at various ratios and seeded onto Alvetex scaffolds, with the final number of cells seeded kept consistent according to the size of the Alvetex insert (0.5M total cells for 12 well inserts, 0.17M total cells for 24 well inserts). Following seeding, all dermal models were maintained as described in (Roger et al., 2019Roger M. Fullard N. Costello L. Bradbury S. Markiewicz E. O’Reilly S. et al.Bioengineering the microanatomy of human skin.Journal of Anatomy. 2019; 234: 438-455https://doi.org/10.1111/joa.12942Crossref PubMed Scopus (93) Google Scholar) for 28 days, at which point HEKn were seeded onto the constructs to form full thickness skin models. These were maintained for 14 days at the air-liquid interface before harvesting for downstream analysis. Full thickness human skin equivalents were unclipped from the inserts, rinsed with PBS and fixed in 4% paraformaldehyde for 2 hours at room temperature. The models were then dehydrated in a graded ethanol series for 15 minutes each (30%, 50%, 70%, 80%, 90%, 95%, and 100%). Models were transferred to plastic cassettes and immersed in Histoclear for 30 minutes followed by a mixture of Histoclear and melted wax with (1:1 ratio) for at least 30 minutes in the 60°C oven. Then, models were moved to full melted wax for 60 minutes before being fully embedded in wax in moulds. Slides were deparaffinised and hydrated through a series of ethanols, before washing in distilled water, Slides were incubated in Mayer's Haematoxylin (Sigma-Aldrich) for 5 mins, before washing for 30 seconds in distilled water. Slides were then dehydrated in 70% ethanol for 30 seconds before incubating in alkaline alcohol for 30 seconds to blue the nuclei. Slides were further dehydrated in 95% ethanol, before incubation in Eosin for 30 seconds. Slides were finally washed in 95% ethanol before further dehydration and clearing in 100% ethanol and Histoclear. Slides were mounted onto glass coverslips using Histomount and imaged using a Nikon E800 microscope at 200x magnification. After deparaffinisation, antigen retrieval and washing, samples were incubated in blocking solution; 1:60 [Normal Goat Serum]: [0.1% BSA/PBS] or 1:4 [Normal Calf Serum]: [0.4% Triton/PBS] for 1 hour at room temperature followed by incubation with primary antibody overnight at 4°C (suppl. Tab. S3). The samples were washed in PBS 3 times for 5 minutes each followed by an hour of secondary antibody incubation at room temperature. Samples were mounted onto microscope slides with VECTASHIELD® Antifade Mounting Medium with DAPI (H-1200-10). The total collagen content in punch biopsies was measured by the QuickZyme Total Collagen Assay (QuickZyme Biosciences) according to the manufacturer’s instructions. Punch biopsies were used to standardise the area when the samples were taken. The assay measures total collagen content in ug/mL, so the values were normalised to the control value for each experiment to enable comparison (and account for variation between experiments). Media samples were collected at the end of culture by removing 1 mL media from each well/plate and storing at -80°C. Samples were shipped frozen to Eve Technologies (Calgary, Canada) for selected cytokine arrays to be performed: Human Cytokine Proinflammatory Focused 15-Plex Discovery Assay® Array (HDF15) and Human MMP and TIMP Discovery Assay® Array for Cell Culture and non-blood samples (HMMP/TIMP-C,O). Embedded samples were divided in two and separately processed and analysed in independent labs. Lab/observer B was fully blinded to the senescent fibroblast density in the model. Pilot experiments showed a high degree of variation between estimates by independent observers. Therefore, observers agreed on the following operationalisations before the actual experiments:1.Epidermal thickness was measured on HE-stained sections. A grid was overlaid onto the images at regular 50 μm intervals, and the epidermal thickness was measured using the straight-line tool in ImageJ. A total of 300 measurements were made across the entire length of the model avoiding only areas that showed embedding artefacts, measured from the bottom of the basal cells to the bottom of the stratum corneum as exemplified in suppl. Figure S7.2.Basal cell and nuclear size was measured on anti-CK10/14-stained IF images. To be included, cells had to touch the basal membrane, be completely identifiable and non-overlapping. 300 cells per sample were measured using the circumference tool in ImageJ.3.Immunofluorescence (IF): After finding significant inter- and intra-observer variation when positive cells were manually identified, nuclei were identified as positive when their mean fluorescence values (MGV, measured in ImageJ) was greater than average background plus 2SD. Average background was calculated separately for each image from 5 clearly negative cells. About 300 cells per experimental were measured on images. of positive cells were as of all cells in the or To show effects of and technical data were as et al., and variability in cell Cell Biol. 2020; Google Scholar). The effects of variable seeding of senescent dermal fibroblasts were by size measurements thickness, basal cell and nuclear per experiment and lab were used as data while for cell fractions by per image the data To ANOVA was out on and normalised data of SASP components in the of 3D full thickness skin equivalents generated with no or 20% senescent HDF as measured by cytokine were at the end of Data independent Figure image Fig. Total collagen content in the dermal equivalent in 3D skin models measured using the QuickZyme Total Collagen Data normalised to the control models. 3 models per analysis results in a negative of Figure image images of full thickness 3D skin models. Senescent cell seeding on of the Figure image of full thickness 3D skin models. Senescent cell seeding on of the 50 Figure image of full thickness models with either or 100% senescent cell seeding density for and 50 Figure image of epidermal thickness due to Figure image Tab. ANOVA values on values between labs A and between senescent fibroblast seeding values were all A cell cell nuclear in a Tab. ANOVA values for normalised values between labs A and between senescent fibroblast seeding values were all A cell cell nuclear in a Tab. for Antifade Mounting Medium with DAPI in a Tchkonia and T. J. and in PubMed Scopus Google The by and Project to TvZ and BBSRC # to The had no role in study in the analysis and of in the of the and in the to the article for TvZ is the for this

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.