Abstract
3D skin models have been explored as an alternative method to the use
of animals in research and development. Usually, human skin equivalents
comprise only epidermis or epidermis/dermis layers. Herein, we leverage
3D bioprinting technology to fabricate a full-thickness human skin
equivalent with hypodermis (HSEH). The collagen hydrogel-based
structure provides a mimetic environment for skin cells to adhere,
proliferate and differentiate. The effective incorporation of the
hypodermis layer is evidenced by scanning electron microscopy,
immunofluorescence, and hematoxylin and eosin staining. The
transcriptome results underscore the pivotal role of the hypodermis in
orchestrating the genetic expression of a multitude of genes vital for
skin functionality, including hydration, development and
differentiation. Accordingly, we evidence the paramount significance of
full-thickness human skin equivalents with hypodermis layer to provide
an accurate in vitro platform for disease modeling and toxicology
studies.
Subject terms: Tissue engineering, Skin models
__________________________________________________________________
We used 3D bioprinting to create a full-thickness human skin model with
a hypodermis layer, offering an in vitro tool for disease and
toxicology studies. This model replicates skin function and highlights
the hypodermis’ role in tissue development.
Introduction
Skin tissue engineering is a field which greatly evolved in the last 5
decades. Elucidation of skin physiopathology, together with
biotechnology evolution, allowed the development of an assortment of
skin substitute templates which are intended for medical approaches and
scientific purposes including in vitro modeling of skin disorders and
testing of pharmaceutical products^[46]1. Advanced techniques and
modern strategies of biofabrication led to the assembly of complex
tissue engineered skin models which efficiently mimic some skin
conditions, including atopic dermatitis, melanoma, and others, although
none of these fully recapitulate the heterogeneity of natural skin
biology and physiopathology^[47]2.
The expectation for bioengineered skin models is to recapitulate the
physiology of human skin to be applied as a non-clinical human model
with more predictability than animal models in the search for new
therapeutic strategies, or in the elucidation of the pathophysiology of
skin disorders^[48]3–[49]5. For this purpose, they should mimic the
natural biology and structure of the skin. Considering the importance
of each skin layer for its integrity and functionality, an ideal model
should present an equally 3-layered structure. However, most of the
currently available models present only 1 or 2-layered
tissues^[50]1,[51]2,[52]4,[53]6–[54]11. Although dermis and epidermis
have been proved to be efficiently mimicked, hypodermis is frequently
omitted^[55]4.
The diversity of the essential skin functions like immune and physical
protection, thermoregulation, water content regulation, and others, is
only possible because of the highly complex biology of the organ,
composed by the 3 main structural layers: epidermis, dermis, and
hypodermis^[56]12. Hypodermis, also known as subcutaneous layer, was
functionally underestimated for a long time. Nowadays, it is well-known
that the hypodermis is not just a storage of energy or a source of
padding and insulation for the body. It is also profoundly involved in
immune process and homeostasis, exerting a dynamic influence over the
other skin layers^[57]13. The involvement and importance of the
hypodermis in skin disorders such as metabolic syndrome, burns, and
scarring has been increasingly put in evidence^[58]14,[59]15.
Despite the acknowledged physiological importance of the hypodermis,
its integration into full-thickness skin models has been limited.
Previous studies have incorporated this layer by introducing mature
human primary adipocytes or by differentiating human adipose
mesenchymal stem cells into adipocytes. These models demonstrated that
the presence of adipocytes supports the maturation and differentiation
of epidermal cells, which is crucial for maintaining barrier
functionality and integrity^[60]13,[61]16. Recently, a 3-layered skin
model showed that adipocytes presence improved dermal-epidermal
junction quality and rete ridge-like structures when compared to a
bi-layered skin model^[62]5. Another study focused on the development
of a 3-layered skin model to obtain a reliable in vitro platform for
studying diabetic skin wound healing^[63]17. Additionally, 3-layered
bioprinted skin graft has been applied for skin wound healing in
different animal models^[64]8. In contrast to these approaches, our
goal was to develop a simplified, cost-effective full-thickness skin
model to assess the physiological relevance of in vitro systems for
creating reliable humanized tissues applicable in cosmetic,
pharmaceutical, and basic research. Our approach employs a
straightforward fabrication process using only human primary
keratinocytes, human primary fibroblasts, and adipocytes from
differentiated human mesenchymal stem cells embedded in a
collagen-based matrix. Through transcriptomic analysis, we unveiled the
significant impact of the hypodermis on gene expression for advancing
our 3D skin models. The Human Skin Equivalent with Hypodermis (HSEH)
developed in this study not only faithfully replicated the
architectural layers of native skin but also upregulated specific
markers associated with adipose tissue, skin differentiation, and
hydration, which are involved in key biological processes. The HSEH
emerged as a promising and ethically responsible tool for advancing
dermatological research and reducing the need for animal
experimentation.
Results and discussion
Collagen properties
As collagen is the major component of the extracellular matrix in the
dermis and hypodermis, we conducted a thorough characterization of this
material. Rheological measurements have indicated that the collagen
solution behaves differently in the presence or absence of cells
(Fig. [65]1a). In the absence of cells, the collagen solution exhibits
Bingham plastic-like behavior at low shear rates, attributed to the
entanglement of collagen molecules. However, once the shear rate
exceeds a critical threshold, the material undergoes a transition to
shear-thinning behavior. This occurs due to the alignment of protein
chains in the direction of flow, which increases with rising shear
rate^[66]18. In the cellular collagen solution, the cells hinder the
collagen aggregation, therefore it behaves as a non-Newtonian fluid
with a pseudoplastic behavior throughout the analyzed shear rate range.
Pseudoplastic materials are fluids that exhibit a decrease in viscosity
with an increase in the shear rate. In other words, they become less
viscous when subjected to higher shear forces. This shear-thinning
behavior is advantageous for bioinks because as the ink is extruded
through the syringe nozzle, shear occurs and viscosity decreases, thus
allowing the material to flow more easily^[67]19. This facilitates the
controlled extrusion of material through the fine printing tips and the
precise deposition of the desired patterns^[68]20. After deposition,
the viscosity of the bioink increases, assisting in maintaining the
integrity of the printed structure. In collagen, the shear-thinning is
caused by the disentanglement of collagen chains during flow^[69]21.
The presence of fibroblast into the collagen matrix slightly decreased
the viscosity values due to the distancing of collagen chains in the
presence of cells^[70]22.
Fig. 1. Rheological, mechanical, and morphological properties of collagen.
[71]Fig. 1
[72]Open in a new tab
a Shear rate sweep and (b) temperature sweep of non-crosslinked
collagen with and without fibroblasts. c Frequency sweep of crosslinked
collagen with and without fibroblasts. d Time sweep at 37 °C of
collagen solutions with and without fibroblasts. e Stress-strain curves
of crosslinked collagen without fibroblasts, dermis (collagen +
fibroblasts), and human skin equivalent (HSE). Scanning electron
microscopy (SEM) of the crosslinked collagen (f) without and (g) with
fibroblasts. Scanning electron micrographs of HSEH after 10 days of
maturation. h shows the full thickness images, (i) shows the hypodermis
region and (j) epidermis layer. In the graphs (a–d) error bars are
resultant from n = 3.
The crosslinking of the collagen was evaluated as a function of
temperature (Fig. [73]1b). Initially, the storage modulus (G’) was
lower than the loss modulus (G”), characteristic of a liquid-like
material^[74]23. At 31.6 ± 0.6 °C (crossover temperature), the storage
modulus of the collagen became higher than the loss modulus, indicating
a solid-like behavior. The presence of cells into the collagen matrix
practically did not affect the crossover temperature (29.7 ± 1.5 °C).
Therefore, we confirmed that the incubation temperature of 37 °C is
adequate to crosslink the collagen structure.
Both storage and loss moduli were also measured as a function of
angular frequency, after collagen reticulation (Fig. [75]1c). Both
constructs (collagen with and without cells) displayed predominant
elastic behavior (G’ > G”) throughout the angular frequency range,
which is desirable for application in 3D bioprinting. Elasticity is the
ability of a material to return to its original shape after
deformation. Therefore, the elastic collagen bioink can aid in
preserving the structural integrity of the printed construct^[76]24.
The presence of cells into the collagen structure significantly
improved both elastic and viscous behavior, which indicates that the
fibroblasts contributed to the overall stiffness of the construct.
Figure [77]1d displays the time sweep curves of collagen solutions with
and without cells at 37 °C. From the start, both collagen solutions
exhibited a solid-like structure, with the storage modulus surpassing
the loss modulus. Notably, the curve profiles for both samples were
highly similar. Within the initial 10 min at 37 °C, the most pronounced
increase in the storage modulus occurred, indicating that the material
becomes sufficiently tough to prevent drainage from the insert
(Fig. [78]1d – inset). However, the storage modulus continued to
slightly rise over the 60-minute duration, which represents the
incubation time before epidermal incorporation during the fabrication
of the human skin equivalent model. Although the collagen solutions
with and without cells displayed similar rheological profiles, the G’
values of the cell-laden gel were notably higher, attributed to the
contribution of fibroblasts. This observation is consistent with
findings from angular frequency sweep curves. This positive effect of
cell presence on the mechanical property was confirmed by compression
testing.
Stress-strain curves of reticulated collagen, dermis construct
(collagen + fibroblast) and HSE show 4 stages of deformation: (i)
elastic deformation, from which was extracted the elastic modulus; (ii
and iii) different levels of plastic deformation of the collagen
structure by alignment; and (iv) fractured construct (Fig. [79]1e). The
elastic modulus increased from 0.21 ± 0.08 kPa (collagen without cells)
to 41.8 ± 0.4 kPa for dermis construct and to 52.1 ± 1.8 kPa for HSE.
We observed that the presence of both fibroblasts and the epidermis
positively affected the elastic modulus. This increase is a result of
the interlocking network formed through cell-polymer and cell-cell
interactions. The elastic modulus of the 3D skin model developed here
is close to the human native skin (10–37 kPa)^[80]25,[81]26. SEM
micrographs show the fibrillar morphology of the collagen structure
after crosslinking (Fig. [82]1f) and the high density of fibroblasts
(Fig. [83]1g) attached to the collagen structure (dermis) of the 3D
skin construct.
Human skin equivalent with hypodermis morphological characterization
As already pointed out, new strategies for testing cosmetic or
dermatological products have arisen worldwide to replace animal tests.
Here, we focused on advancing 3D skin models by incorporating the
hypodermis layer. Our skin model comprises 3 layers: epidermis, dermis,
and hypodermis (Fig. [84]1h–[85]J). The hypodermis layer was
constructed with a thick portion of collagen type I and adipocytes
differentiated from hMSCs. Above this portion, the dermis equivalent
was built up by bioprinting a mixture of dermal fibroblasts with
collagen. Over the dermal portion, keratinocytes were deposited and
differentiated, thus forming the epidermis (Fig. [86]2).
Fig. 2. Morphological characterization of the bioprinted human skin
equivalent with hypodermis (HSEH).
[87]Fig. 2
[88]Open in a new tab
a Scheme of production of HSEH model. Created in BioRender. Figueira,
A. (2024) BioRender.com/x38z313. b Photo of apical view and (c) basal
portion of the HSEH, highlighting the hypodermis layer. d Hematoxylin
and eosin-stained histological sections of HSEH model. The first image
shows the full thickness image, the second shows the hypodermis region,
and the third image the epidermis layer. The regions marked with E
correspond to the epidermis, D to the dermis, and H to the hypodermis.
Skin equivalents serve as a highly valuable alternative to animal
testing and can be considered physiologically comparable to natural
skin^[89]27. Numerous companies have designed skin model alternatives
to assess the efficacy of pharmaceutical, skincare, and cosmetic
products in an in vitro platform, decreasing the dependency on animal
usage. While HSE research has made significant progress in
reconstructing dermal and epidermal layers using fibroblasts and
keratinocytes, there is a requirement for innovative co-culture systems
that can better emulate the intricacies of human anatomy^[90]28.
Distinguished from other research cohorts that employed in vitro skin
models by the incorporation of an explanted hypodermal layer
^[91]13,[92]28,[93]29, our model was fabricated using the
differentiated adipocyte cells from hMSC, using these cells our model
has the advantages of reproducibility, high throughput, and control
over internal structure, and shape compared to other
techniques^[94]9–[95]11.
To evaluate the structure of the HSEH, skin equivalents were harvested
on day 10. Pictures of the 3D skin model showed the formation of the
hypodermis layer on the basal portion. H&E staining (Fig. [96]2d) and
SEM images also confirmed the presence of the hypodermis, with visible
lipid droplets spaces in circular format and dense adipocyte population
as shown under the dermis. The dermal portion of the skin model
presented a homogenous distribution of the fibroblasts within the
collagen structure and the epidermis comprised stratified layers of
keratinocytes. In summary, the histological data confirmed the
successful biofabrication of a robust skin model containing 3
differentiated layers, that closely mimic the native skin tissue
structure. The achievement of a 3-layered model overcomes one notable
challenge within the realm of bioengineering in vitro skin
models^[97]6.
HSEH physiology and gene signature
To characterize the physiology and transcriptome profile of the HSEH,
we compared it to a two-layered HSE model. The difference between the
two models is the presence of the hypodermis in the HSEH, which is
absent in the HSE. Both models have identically constructed dermis and
epidermis layers.
One of the principal challenges in the context of 3D cell culture
pertains to the preservation of cell viability within the model^[98]30.
To assess cell viability and proliferation in our models, we measured
ATP production on both days 5 and 10 of skin differentiation, using it
as an indirect indicator of cell viability (Fig. [99]3a). After
luminescence measurement using the CellTiter-Glo® 3D cell viability
assay, the ATP values were obtained using a standard curve
(Supplementary Information, Fig. [100]1). The decrease in ATP values in
10 days of skin differentiation for both HSE and HSEH is likely
attributed to the fact that, at this point, the differentiation process
has stabilized, and the cells have entered a less metabolically active
state.
Fig. 3. HSEH cell viability, proliferation, and differentiation.
[101]Fig. 3
[102]Open in a new tab
a ATP measurement comparison between HSE and HSEH at days 5 and 10. b
Trans-epithelial electrical resistance (TEER) (Ω.cm^2) measured at days
5 and 10. c Full image of HSEH differentiation after 10 days. d High
magnification of hypodermis layer of HSEH showing the lipids droplets
inside the adipocytes in green. e High magnification of epidermis of
HSEH showing lipids labeled with bodipy in green, actin filaments
labeled with phalloidin-rhodamine in red, and cell nuclei labeled with
DAPI in blue. f HSEH epidermis labeling for involucrin (green), actin
filaments (red), and cell nuclei (blue). g Immunolabeling of
Cytokeratin 10 (red) in the epidermis of HSEH and (h) immunolabeling of
Cytokeratin 15 (green). Graphic symbols denotate significance levels:
*P ≤ 0.05. In the graph A dots represent n = 3 biologically independent
samples and in the graph (b), the dots represent n = 6.
To evaluate the integrity of the skin barrier, we performed TEER
measurements, which is a reliable methodology to study barriers
permeability and integrity. Here, we evaluated the HSEH in comparison
with HSE at days 5 and 10 of skin differentiation. This method involves
applying an alternating current (AC) across a membrane to gauge
electrical resistance^[103]31. Through this approach, we can assess the
condition of tight junctions and identify any indications of barrier
dysfunction. The results of skin integrity through TEER revealed that
HSEH exhibited significantly higher TEER values compared to HSE,
demonstrating a 50% increase (Fig. [104]3b). This distinctive trend was
consistently observed in measurements taken on both day 5 (p = 0.05)
and day 10 (p = 0.05). The HSEH presented values around 60 Ω.cm^2,
corroborating with the values found in the human skin explants^[105]32.
Compared to other complex full-thickness models, the HSEH presented
superior values, indicating that it better mimics human skin^[106]33.
The HSE (Supplementary Information, Fig. [107]2) and HSEH models were
subjected to an immunostaining assay to evaluate the epidermis
differentiation through protein expression, and bodipy staining to
assess the adipocyte lipid content on hypodermis layer.
After keratinocytes stop proliferating, they migrate upwards to form
the suprabasal layer, where they begin expressing differentiation
markers like cytokeratins 1 and 10 (CK1/CK10)^[108]34. Cytokeratin 15
(CK15), typically expressed in basal keratinocytes, plays a role in
maintaining the stem cell population before differentiation occurs.
Eventually, the keratinocytes reach the outermost layer of the
epidermis, where they form the cornified layer (stratum corneum) and
express terminal differentiation markers such as loricrin and
filaggrin^[109]34,[110]35. In Fig. [111]3g, the expression of CK10 is
evident as a marker of keratinocyte differentiation, while Fig. [112]3h
shows the distribution of cytokeratin 15, highlighting keratinocytes
that are on the path to differentiation. The presence of cytokeratin
corroborates with the data from the RNA sequencing (Supplementary
Information, Fig. [113]3) showing the upregulation of cytokeratin 1, 4,
10, and 15 of HSEH compared to HSE.
Involucrin, the major protein component of the cornified cell envelope
found in terminally differentiated epidermal cells^[114]36, was also
monitored by immunostaining. The involucrin was consistently expressed
in HSEH model, indicating its similarity with human skin tissue, and
pointing out its feasibility as a good marker for evaluation of skin
differentiation (Fig. [115]3f). In the same way, it was possible to
observe that the hypodermis was well formed after the differentiation
of hMSCs into adipocytes, which were successfully evidenced by bodipy
identification (Fig. [116]3e). It was possible to observe lipids
present in the dermis and epidermis as well.
The immunofluorescence images of the HSEH demonstrate remarkable
similarities to actual human skin^[117]37–[118]39. The outermost layer,
the stratum corneum, appears as a thin, brightly stained band due to
the presence of keratinized cells. The stratum spinosum shows a network
of cells with intercellular bridges, stained for proteins like
involucrin. The basal layer, adjacent to the dermis, contains more
rounded cells with prominent nuclei. All these features can be observed
in Fig. [119]3e that highlight the ultrastructure of HSEH epidermis.
The dermis contains dispersed fibroblasts, elongated within the
extracellular matrix (Fig. [120]3f). In the hypodermis, the most
prominent feature is the presence of adipocytes. These cells are
typically large, round, and have a characteristic ring-like appearance
due to the large lipid droplet within them that displaces the nucleus
to the periphery. Lipid-specific stains, such as Bodipy, highlight
these droplets, usually resulting in bright green staining.
Besides staining lipid droplets with bodipy to assess the adipogenic
differentiation of hMSCs within the 3D collagen type I hypodermal
layer, we also evaluated the expression of adipocyte markers at the
mRNA level (as shown in Fig. [121]4). The presence of PPARɤ, PPARβ, and
PDK4 as specific markers for adipocytes, was assessed in reconstructed
hypodermis after 10 days of culture using qPCR. hMSCs cultured in
adipogenic conditions within the collagen hydrogels exhibited
differentiation into adipocytes expressing the respective markers,
while the control HSE did not show this expression (as depicted in
Fig. [122]4c–e).
Fig. 4. Adipogenic differentiation of hMSCs and mRNA expression profile of
genes regulated during the adipogenesis.
[123]Fig. 4
[124]Open in a new tab
a, b 3D surface projections of a basal view of hypodermis. Cells were
stained with phalloidin (red) to mark the actin in the cytoplasm,
highlighting the boundaries of each cell, and with bodiby (green) to
mark the lipid droplets. Expression of (c) PPARɤ, (d) PPARβ and (e)
PDK4 genes analyzed by qRT-PCR in the HSE and HSEH. Error bars
represent standard deviation of means from at least 3 replicates
(n = 3). Graphic symbols denotate significance levels: *P ≤ 0.05,
**P ≤ 0.01, ***P ≤ 0.001.
A comprehensive in vitro human skin model should aim to recapitulate
the intricate structural composition of native skin, encompassing
various cell types. Our model endeavors to mirror the fundamental
architectural layers of the skin while emulating the physiological
characteristics of these layers in an in vitro context. The hypodermal
layer demonstrates the upregulation of distinctive markers associated
with adipose tissue (PPARɤ, PPARβ, and PDK4), while faithfully
reproducing the organizational arrangement and lipid accumulation
patterns for adipocytes. The PPARɤ and the mTORC1 is related with
upregulation of subcutaneous fat adiponectin secretion and the
glyceroneogenesis in the hypodermis^[125]40. The expression PDK4 is
closely associated with glyceroneogenesis^[126]41,[127]42, while PPARβ
showed to stimulate differentiation and lipid accumulation in
keratinocytes^[128]43. We also investigated other adipocyte-specific
morphology through 3D whole-mount fluorescence staining of bodipy in
differentiated hydrogels in vitro. The differentiated adipocytes were
evenly distributed throughout the hydrogels and displayed numerous
lipid droplets within their cytoplasm, as evidenced by the green color
(bodipy) staining (illustrated in Fig. [129]3c) and observed across the
entire hydrogel (as demonstrated in the Z-stack image in Fig. [130]4a,
b).
To fully understand the gene expression signature of our HSEH model in
comparison to the HSE model, we performed a transcriptomic evaluation
through an RNA-sequencing analysis. The next-generation RNA sequencing
(RNA-seq) provides unprecedented detail about the transcriptional
landscape of our in vitro model, not only allowing for precise
measurement of transcript expression levels in HSEH, but also showing
how our model is assembling in vitro the genes pathways of the skin.
Gene ontology (GO) enrichment analysis was performed to determine
pathways and functions associated with the identified Differentially
Expressed Gene (DEGs). Comparative transcriptome analysis revealed
distinct gene expression patterns, including pathways associated with
adipogenesis, regulation of epidermal differentiation, and collagen
synthesis and modulation. In detail, a gene was considered DEG when it
exhibited a p ≤ 0.005. According to this metric, out of the 28,163
analyzed genes, 1081 were identified as DEGs in the HESH compared to
HSE, with 61.98% upregulated and 38.02% downregulated (Fig. [131]5b).
The volcano plot highlighted the upregulation of genes mainly
associated to remodeling and organization of extracellular matrix and
lipid metabolism (Fig. [132]5a). The most upregulated gene was
Hyaluronan and Proteoglycan Link Protein 1 (HAPLN1), a protein involved
in crosslinking and stabilizing hyaluronic acid (HA) with aggregates of
proteoglycan monomers. The function of HAPLN1 is believed to influence
collagen and ECM contraction. A study was conducted to understand the
role of HAPLN1 in skin, finding that its expression decreases with age.
This decrease is associated with alterations in the extracellular
matrix (ECM) and consequent impairment of vascular integrity, which, in
turn, influences metastasis and inflammatory cell
infiltration^[133]44,[134]45. Given that HAPLN1 is essential for a
normally functioning ECM, the up regulation of this gene in the HSEH
model suggests that the presence of hypodermis enhances ECM structure
and functionality.
Fig. 5. Identification of DEGs comparing HSEH to HSE.
[135]Fig. 5
[136]Open in a new tab
a Volcano plot shows data for the transcriptome. Threshold in axis Y
indicates 1,5 and in axis X indicates log2 fold change. Blue dots
represent genes out of the threshold, red dots are the up and down
regulated genes and gray dots are non-significant genes. b
Part-to-whole chart indicate the percentage of up- and down-regulated
genes among the identified DEGs (Total: 1081 genes). c Biological
process analysis plot of the 12 most enriched process when comparing
HSEH with HSE. d The graph expresses the 20 most GO enriched process in
the HSEH. N = 3 biologically independent samples.
The functional analysis highlighted 12 significant biological processes
(BP) when we compared HSEH with HSE (Fig. [137]5c). In the enrichment
of gene clusters, dots represent term enrichment with color coding: red
indicating high enrichment, while gray indicating low enrichment. The
sizes of the dots represent the percentage of each row (GO category),
the graph confirming the influence of the hypodermis in ECM modulation.
Upregulated genes showed enrichment in terms related to ECM processes,
with the most significant being ‘collagen-containing extracellular
matrix,’ ‘external encapsulating structure’, and ‘extracellular matrix”
(Fig. [138]5c, d). These results demonstrate the robustness of our
3-layered model, as literature highlights the crucial interaction
between the hypodermis and dermis in maintaining skin homeostasis, as
recently reviewed^[139]46. Furthermore, we can assume that the presence
of a functional hypodermis ensures a robust dermal ECM, influencing key
dermal properties such as strength and elasticity^[140]46.
From the GO-enriched terms, we highlighted the molecular function
“collagen binding” (Fig. [141]6a) and the cellular component “collagen
trimer” (Fig. [142]6b), both crucial to the composition and reliability
of the skin ECM. Collagen trimerization and interactions with binding
proteins are essential for the integrity and functionality of the skin
ECM. The triple helix of the three collagen molecules provides
structural strength and stability, while binding proteins like
proteoglycans and glycoproteins enhance the framework and assist in the
organization and crosslinking of ECM fibers. Together, these processes
ensure the resilience, elasticity, and functionality of the
skin^[143]47.
Fig. 6. Heat map of the enrichment analysis.
[144]Fig. 6
[145]Open in a new tab
a Collagen binding. b Collagen trimmer. c Keratinocyte differentiation.
d Leptin and Adiponectin. Target genes are colored by increasing
statistical significance (turquoise to red). N = 3 biologically
independent samples.
Additionally, an enrichment in the biological process term “regulation
of keratinocyte differentiation” was observed (Fig. [146]6c),
confirming that the presence of the hypodermis influences not only the
dermis but also the outermost layer of the skin, the epidermis. The
expression of structural proteins such as cytokeratin 10 (KRT10) is
used as a marker for skin maturation and the establishment of an
effective barrier^[147]48. In this study, KRT10 was one of the
upregulated genes in HSEH, further suggesting that HSEH is more
effective in structuring the skin barrier.
As expected, the addition of hypodermis to the skin model promoted the
expression of genes related to lipid metabolism, as evidenced by the GO
enrichment on the canonical pathways of leptin and adiponectin
(Fig. [148]6d). The expression of these genes creates a more
physiologically relevant environment, enhancing the model’s robustness
and making it more suitable for studying a variety of skin disorders.
Adipokines like leptin and adiponectin act as hormones and cytokines,
playing crucial roles in skin homeostasis. They are involved in
structural functions such as angiogenesis and collagen production and
provide the skin with anti-inflammatory tools to combat pathogens,
thereby maintaining its structural and functional
integrity^[149]49,[150]50.
Our model showed that the subcutaneous layer, or hypodermis, played a
key role in the epidermal differentiation, ECM proteins expression, and
skin hydration, which are related to the regulation of the adipocytes
layer. Some reports have identified that the presence of hypodermis
caused a positive regulation in the differentiation of in vitro skin
models through the increased number of actively dividing basal cells,
the deposition of basement membrane proteins, and the increase of cells
transformed into fully mature keratinocytes, ultimately optimizing the
formation of a multi-layered epithelial structure^[151]13,[152]51.
The transcriptome data presented here highlights crucial pathways that
are enriched in the HSEH, demonstrating their significant involvement
in skin function. This achievement opens an opportunity to apply the
HSEH model in a wide range of assays, compromising it to the “3Rs”
principle. This principle encompasses Replacement, Reduction, and
Refinement in humane animal research^[153]13, and it is globally
recognized and firmly embedded in numerous national and international
regulations.
This groundbreaking research has revealed the development of a 3D skin
model remarkably resembling native human skin, composed of epidermis,
dermis and hypodermis. Using 3D bioprinting, we achieved automation to
fabricate the HSEH constructs with great cell organization. The
collagen hydrogel has provided structural, mechanical, and biological
cues to skin cells functions. This model faithfully replicated the
architectural layers of native skin and exhibited upregulation of
specific markers associated with adipose tissue, skin differentiation,
and skin hydration. The HSEH’s transcriptome profile unveiled essential
biological processes, such as epidermis development, ECM remodeling,
and lipid metabolism. Additionally, the model’s enriched pathways
highlighted its potential for various research applications, aligning
with the “3Rs” principle and regulatory measures aimed at reducing
animal experimentation. Overall, the HSEH model represents a promising
and ethically sound tool for advancing the field of dermatological
research.
Methods
Cell culture before biofabrication of constructs
Human mesenchymal stem cells (hMSC, PT-2501, Lonza) were cultured in
MSCBM Basal Media (Lonza) and MSCGM SingleQuots Supplement Kit (Lonza)
and were used up to passage 6. Human dermal fibroblasts cells provided
from BCRJ were extracted from foreskin of a black male with 2 years old
(HDFn, nh-skp-FB0040, Banco de Células do Rio de Janeiro, Brazil),
cultured in Dulbecco’s modified Eagle medium (DMEM 12-604, Lonza) with
4.5 g. L^−1 glucose and L-glutamine supplemented with 10% (w/v) fetal
bovine serum (FBS, VitroCell), and they were used up to passage 20.
Primary human keratinocytes provided from BCRJ were extracted from
foreskin of a black male with 7 years old (HEKn, nh-skp-KT0009, Banco
de Células do Rio de Janeiro, Brazil), cultured in EpiLife™ medium
(MEPI500CA, Gibco) with 60 µM Ca^+2 supplemented with human
keratinocyte growth supplement (HKGS S-001-5, Gibco), and they were
used up to passage 5. All cell culture were performed in an incubator
(Thermo) at 37 °C, humidified, 5% CO[2] environment.
Collagen preparation previous to biofabrication of constructs
For each well, 525 μL of collagen I from rat tail (Corning;
concentration of 3 mg. mL^−1) was mixed with 70 μL of HAM F-12 (Gibco;
10x), 70 μL of reconstitution buffer (10x), and 35 μL of FBS. The
reconstitution buffer was prepared by mixing 50 µM NaOH, 261 mM
NaHCO[3]and 200 mM HEPES (All from Sigma Aldrich). After mixing, the
solution was transferred to a volumetric flask and completed to 50 mL.
After preparation, the reconstitution buffer was sterilized with 0.2 µm
filter, aliquoted and kept at −20 °C until use. During collagen
preparation, the solution was kept on ice to prevent reticulation.
Human skin equivalent (HSE) and human skin equivalent with hypodermis (HSEH)
fabrication
Hypodermis fabrication
A thin layer of collagen (200 µL) was printed on a transwell insert
(Corning; 12 mm Transwell® with 0.4 µm pore polyester membrane insert)
using a commercial bioprinter (Genesis, 3D Biotechnology Solutions,
Brazil). The plate was incubated for 1 h at 37 °C for collagen
crosslinking. After this time, 3 × 10^5 of hMSC (in 500 µL of DMEM with
10% FBS) were manually deposited on top of the collagen (Fig. [154]2a).
The plate was incubated at 37 °C in an incubator with 5% CO[2] during
11 days for cell differentiation into adipocytes. Initially, DMEM with
10% FBS (Basal medium) was used as cell culture media (D-11,
Fig. [155]2a). After 1 day of cultivation (D-10), the culture medium
was changed for a differentiation medium that consist into Basal medium
supplemented with 1 µM of dexamethasone, 0.2 nM of indomethacin, 0.5 mM
of IBMX and 1.7 µM of insulin (all from Sigma-Aldrich). In the 3rd day
of cultivation (D-8, Fig. [156]2a), the DMEM with 10% FBS was
supplemented only with 1.7 µM of insulin and, in the next day (D-7,
Fig. [157]2a), the differentiation medium was used for cell adipocyte
differentiation. Finally, between 7 and 11 days of cultivation (D-4 to
D0, Fig. [158]2a) DMEM with 10% FBS with 1.7 µM of insulin was used.
The culture medium was placed inside (500 µL) and outside (1000 µL) the
insert.
Dermis fabrication
After 11 days of hypodermis biofabrication and differentiation, and at
the initial day of HSE fabrication (D0, Fig. [159]2a), 700 µL of
collagen solution containing 3 × 10^5 of fibroblasts were 3D bioprinted
on top of the insert with or without the hypodermis layer. The
construct was designed in the Thinker Cad software and consisted of 9
layers of concentric circles. The Simplify 3D software (Simplify 3D,
V5) was used for slicing and to control the bioprinter. The bioink was
placed in a 10 mL sterile syringe (Luer-Lock, BD) coupled to an
irrigation needle with a nozzle diameter of 0.7 mm (22 G, Injex). The
temperature of the syringe holder was set to 4 °C and the construct was
printed with speed of 60 mm s^−1. After the printing process, the plate
was incubated for 1 h at 37 °C for collagen crosslinking.
Epidermis fabrication
Right after the dermis crosslinking, for both HSEH and HSE, 4 × 10^5 of
keratinocytes were manually deposited on top of the insert using 200 μL
of cell culture media, composed of 100 µL of EpiLife™ and 100 µL of a
differentiation medium prepared by mixing: 100 mL of DMEM-F12, 100 µL
of 10 mg.mL^−1 insulin (Sigma, I2643) in 5 μM HCl (Sigma Aldrich)
400 µL of 1 μM hydrocortisone 21-hemisuccinate sodium salt (Sigma,
H4881) in ethanol (Sigma Aldrich) and DMEM (1:1), 25 µL of 2 μg.mL^−1
epidermal growth factor (Sigma, E9644) in DMEM, 100 µL of 5 mg.mL^−1
apo-transferrin (Sigma, T1147) in water, and 10 µL of 1 μM cholera
toxin (Sigma, C8052) in sterile purified water. Since this layer did
not demand ECM components, the cells were manually added and not
printed. The constructs were incubated for 24 h, and the medium in the
apical portion of the insert was aspirated to form an air liquid
interface, and the inserts were moved to a deep well plate (Falcon
355467) with a cell strainer (Falcon 352360). The differentiation
medium supplemented as described above was kept only outside of the
insert (9 mL) for the following 10 days, changing the media every 3
days.
Evaluation of the viscous properties by rheology
The rheological properties of the collagen were evaluated with and
without fibroblasts using a compact modular rheometer (MCR-102, Anton
Paar, Austria), controlled by the Anton Paar Rheoplus software
(RHEOPLUS/32 version 3.61). A cone-plate geometry with 50 mm of
diameter was positioned with 0.097 mm between the plates. Viscosity
measurements were performed as a function of shear rate (1.0 to
300 s^−1) at 25 °C, immediately after gel preparation (without
crosslinking), to simulate 3D bioprinting conditions. The storage (G’)
and loss (G”) moduli were measured as a function of temperature,
immediately after gel preparation, to evaluate temperature-driven
crosslinking. For temperature sweep, a constant angular frequency of
10 rad. s^−1, coupled with 1% of strain was applied, using a heating
rate of 2 °C.min^−1 and temperature ranging from 10 to 70 °C.
Furthermore, G’ and G” were measured as a function of oscillation
frequency (0.1 and 17 rad. s^−1) using crosslinked collagen after 10
days of cell culture. For these measurements, a plate-plate geometry
with 10 mm diameter and 1.0 mm gap was used. For the oscillation
frequency measurement, a constant strain of 1% was applied, which lies
within the linear viscoelasticity area. For time sweep, the collagen
solutions with and without fibroblasts were analyzed immediately after
gel preparation (without crosslinking), using temperature of 37 °C, 1%
strain, and angular frequency of 10 rad. s^−1. All measurements were
performed in triplicate. The graphs were plotted using GraphPad Prism,
version 9.5.1.
Mechanical characterization by compression testing
The mechanical properties of the constructs were evaluated in a texture
analyzer (TA.XTplus, Stable Micro Systems, UK) using a 500 N load cell.
The equipment was controlled by the Exponent TA.XTplus software,
version 6.2. The collagen constructs without cells were measured
immediately after crosslinking (1 h at 37 °C). In addition, the skin
model containing dermis (collagen with fibroblasts), and epidermis
(keratinocytes) was measured after 10 days of cell culture. A construct
containing only the dermis layer (collagen with fibroblasts) was also
fabricated and measured after 10 days of cell culture. The cylindric
constructs were compressed up to 100% strain with a probe diameter of
75 mm and compression rate of 0.05 mm. s^−1. The elastic modulus was
extracted from the initial linear region of the stress-strain curves,
using the OriginPro 2023 software. All measurements were performed in
triplicate.
Morphology assessment by scanning electron microscopy (SEM)
The morphology of the crosslinked collagen without cells and the cell
laden 3D skin construct were analyzed by field emission gun-SEM
(FEG-SEM Inspect F50, Thermo Fisher Scientific, USA). Secondary
electron (SE) images were acquired with acceleration voltage of 3 kV.
Sample preparation for SEM included: (i) washing with phosphate
buffered saline (DPBS) (Gibco); (ii) overnight fixation of the
constructs at 4 °C in a solution containing 2.5% of glutaraldehyde and
0.1 M of sodium cacodylate buffer with 1.5 mM of CaCl[2] (all from
Sigma Aldrich); (iii) dehydration in gradient ethanol solutions (30%,
50%, 70%, 90% and 100%), 10 min each; (iv) drying at 25 °C; (v) cutting
with blade for cross-sectional view; and (vi) deposition of a thin
layer of gold. The ImageJ software, version 1.54d, was used to add the
scale bar to all images, following calibration with a reference-scaled
image.
Barrier integrity analysis by trans-epithelial electrical resistance (TEER)
The barrier integrity of HSE and HSEH models was evaluated by TEER.
Measurements were taken using an Epithelial Volt/Ohm Meter (Millicell®
ERS-2 Voltohmmeter) and a pair of Ag/AgCl probes (Milipore). TEER
values were calculated according to Eq. [160]1:
[MATH: TEER=(Rsample−Rblank)×A :MATH]
1
where R[sample] is the resistance value for the skin model, R[blank] is
the resistance value of an insert without cultured cells, and A the
effective culture area (1.12 cm^2). Briefly, the differentiation medium
was removed, 2 mL of PBS were added in the basal region, and additional
300 µL of PBS were added in the apical region for the measurements. The
probe was placed such that one electrode was submerged in the upper
compartment and the other was submerged in the lower compartment. TEER
values were recorded on day 5 and day 10 of skin differentiation. These
measurements were performed in 3 batches of 3 reconstituted skin models
at each time. The graph was plotted using GraphPad Prism, version
9.5.1.
Cell viability by the ATP measurement
To evaluate the proliferation state of HSE and HSEH cultures, after day
5 and day 10 of skin differentiation, the models were harvested from
the inserts and placed in a white 96-well plate with 100 µL of culture
medium. The ATP of each tested condition was quantified using a
CellTiter-Glo® 3D Cell Viability Assay (Promega) following the
manufacturer’s instructions. The luminescence was recorded in a Glow
Max Plate Reader (Promega, USA). The graphs were plotted using GraphPad
Prism, version 9.5.1.
Cell differentiation by immunofluorescence
HSE and HSEH models were fixed at day 5 and day 10, for 4 h in 10%
formalin (Sigma Aldrich) and embedded in a cryopreservative medium
(OCT, Tissue Tek, Sakura) and immediately frozen at −20 °C, after which
the blocks were sectionized in 10 µm sections using a cryostat (Leica).
The sections were fixed on a glass slide for 10 min at ambient
temperature. To perform the immunolabeling, the sections were washed
three times with PBS, incubated with permeation solution (0.1% Triton
X-100), for 15 min at room temperature, washed 3 times with PBS, and
blocked for 2 h, with 2% bovine serum albumin (BSA, Sigma-Aldrich).
After blocking, samples were incubated with primary antibody
anti-involucrin (Thermo MA5-11803), anti-cytokeratin 10 (Thermo
MA5-13705), and anti-cytokeratin 15 (Abcam AB52816) overnight, at 4 °C,
in a humidify chamber. Subsequently to incubation with primary
antibody, samples were washed 3 times with PBS and incubated for 2 h
with secondary antibody anti-mouse (Thermo A110001 and A11032) and
anti-rabbit (Thermo A11034). Nuclei were stained with DAPI (Biotium)
for 5 min, actin fibers with phalloidin rhodamine (Thermo) and lipids
droplets with Bodipy (Cayman) for 40 min before image acquisition.
Images were acquired in a confocal laser scanning microscope (TCS SP8,
Leica) using the software LAS X 5.0.2. The ImageJ software, version
1.54d, was used to add the scale bar to all images, following
calibration with a reference-scaled image.
Structure by histology trough historesin embedding and hematoxylin and eosin
staining
The samples were fixed in 10% buffered formalin from the dilution of
37% formaldehyde solution (Sigma-Aldrich) in PBS at pH 7.4.
Subsequently, they were gradually dehydrated in ethanol (30, 50, 70,
80, 90, 95, and 100%) and embedded in historesin (Leica). Previous to
embedding, the ethanol of the samples was gradually replaced by the
historesin (ethanol:historesin ratio of 5:1, 4:1, 3:1, 2:1, 1:1, and
1:2). A rotating microtome (RM2235, Leica) was used to cut the samples,
with the aid of glass slides made from glass strips for ultramicrotome
(Ted Pella, USA) in a Glass Knife Maker (EM KMR3, Leica). Transverse
cuts were made with thickness of 5 µm and transferred to 76 × 26 mm
microscopy slides (Waldemar Knittel). For hematoxylin & eosin (H&E)
staining, the historesin sections were incubated at 40 °C in Harris
hematoxylin (Scientific ACS), for 15 min, and rinsed in running water
for 3 min. Subsequently, sections were incubated at 40 °C, for 15 min
in an aqueous solution of 0.5% eosin Y (Sigma-Aldrich) and 0.5%
Phloxine B (Sigma-Aldrich) and rinsed under running water for 3 min.
After drying, the slides were assembled with Entellan® new
(Sigma-Aldrich) and 24 × 50 mm coverts (Exacta) for further analysis in
an optical microscope (DM6, Leica). The ImageJ software, version 1.54d,
was used to add the scale bar to all images, following calibration with
a reference-scaled image.
Transcriptome Illumina library preparation and sequencing
To prepare the transcriptome libraries, approximately 150 ng of total
RNA from each sample was processed using the Illumina Stranded Total
RNA Prep, Ligation Kit with Ribo-Zero Plus (Illumina Inc., San Diego,
CA, USA), following the manufacturer’s instructions. Library quality
was validated with the D5000 ScreenTape Assay Kit for the TapeStation
4150 (Agilent) and quantified using qPCR with the QIAseq Library Quant
Assay Kit (Qiagen). The libraries were pooled in equimolar ratios
(650 pM) and submitted for paired-end sequencing (2 × 100 bp) on the
Illumina NextSeq 2000 platform at the Brazilian Biorenewables National
Laboratory (LNBR/CNPEM) in Campinas, Brazil.
FastQC^[161]52 was employed to assess the quality of the sequencing
reads. Skewer^[162]53 was then utilized to trim low-quality ends using
the parameters (--mode pe --end-quality 20 --mean-quality 20 --min 30),
as recommended by previously published guidelines^[163]54. The trimmed
reads were aligned to the GRCh38 genome using STAR^[164]55.
Subsequently, transcript abundance was quantified with HTSeq^[165]56.
Differential expression analysis was conducted with DESeq2^[166]57,
following the removal of unwanted variation via RUVSeq^[167]58. Pathway
enrichment analysis was carried out using clusterProfiler^[168]59 and
ReactomePA^[169]60,[170]61. Signature scoring was performed with
singscore^[171]62, using reference sets from REACTOME^[172]61,
MSigDB^[173]63 or WikiPathways^[174]64. The raw reads were deposited in
the Sequence Read Archive (SRA) under the access code PRJNA1139958.
Statistics and Reproducibility
The data were analyzed in GraphPad Prism software, version 9.5.1, using
t-student test nonparametric. Statistical significance was established
at a p value of less than 0.05. All data points were derived from two
or more biological replicates. Replicates were defined through
manufacturing of skin constructs on different days, utilizing cells
from distinct passage numbers, and employing varied collagen lots to
ensure comprehensive and diverse data representation.
Reporting summary
Further information on research design is available in the [175]Nature
Portfolio Reporting Summary linked to this article.
Supplementary information
[176]Supplementary Information^ (411.7KB, pdf)
[177]42003_2024_7106_MOESM2_ESM.pdf^ (74.7KB, pdf)
Description of Additional Supplementary Materials
[178]Supplementary Data 1^ (222.8KB, xlsx)
[179]Reporting summary^ (70.8KB, pdf)
Acknowledgements