Epithelia separate organs and tissues from one another and from the environment. Hereby they act as physical barriers with protective and regulatory functions. The skin, in particular, is permanently affected by mechanical stress and had to develop structures and mechanisms to preserve integrity [1
]. Mature skin consists of two major layers: the epidermis and dermis. Occasionally, hypodermal tissue is added as a third skin layer. The epidermis is located superficially. It is assembled as stratified epithelium that is composed of tightly connected cells, namely keratinocytes, which are linked to the underlying dermal tissue via a basement membrane [2
]. The dermis, on the other hand, is a dense connective tissue consisting of fibrillary proteins embedded in a gel-like ground substance and approximately 20 times thicker than the epidermis [3
]. Finally, the hypodermis underneath is mainly made of fat-storing cells and serves as isolation and loose connection to cartilage and bones [4
From an engineering point of view, mature skin is a highly non-linear, anisotropic and viscoelastic material [5
]. While anisotropy leads to the well-described Langer’s lines as well as to an impact of sample orientation and location on elastic properties, viscoelasticity causes strain rate dependency of mechanical properties and the typical stress relaxation under prolonged loading [7
]. The highly non-linear stress-strain curves measured in tensile testing are often described as “J shaped” and originate from dermal microstructure [9
]. In the initial loading phase, the so-called ”toe region”, the material response, is very soft, and relatively high strain levels are reached with only small increase of tissue stress. Subsequently, fibrillary proteins are gradually oriented (heel region) and tightened (linear region). During this sequence, material stiffness increases by orders of magnitude and the material displays tremendous levels of strain stiffening [6
]. The mechanical properties of mature skin are most often discussed in terms of microstructure and properties of the collagen network in the dermis [12
]. This was convincingly demonstrated by electron microscopy and X-ray diffraction on skin at intermediate and high strain levels where collagen fibers progressively straightened and aligned to the direction of strain to be finally separated into fibrils shortly before failure [13
]. At lower strain levels, collagen fibers stay largely wavy and randomly orientated [11
]. Here, elastin and a gel-like matrix primarily made of glycosaminoglycans are considered to be responsible for the mechanical properties [14
]. All layers of the skin, but especially the dermis, develop extensively in a late fetal stage but also postnatally to their mature structure and composition. Cornification of the epidermis, decellularization of dermal tissue, cross-linking of collagen, and emergence of elastin should be mentioned here exemplarily [15
]. In general, knowledge on the mechanical properties of developing skin is scarce and, to the best of our knowledge, no mechanical study directly compared fetal and mature tissue.
Additionally, despite intense research, the complex mechanical response of mature full-thickness skin under physiological conditions is poorly understood. This holds especially true at comparatively small strains. In general, in vivo measurements are performed in this small strain regime but they are often difficult to reproduce whereas measurement devices for in vitro mechanical testing almost invariably perform much better in the high strain and high force regime [16
]. Therefore, many publications focused on excised human or animal tissue at high and possibly non-physiological stress levels to characterize the mechanical properties of skin [10
]. Here, monotonic or cyclic uniaxial tensile tests were used to analyze collagen fiber orientation and fracture as well as the influence of strain rate on viscoelasticity and anisotropy. Unfortunately, many publications on ex vivo skin samples give little or no information on sample conditions even though hydration has been shown to strongly influence mechanical behavior [11
]. In addition, cell viability was rarely explored. The role of cells for the mechanics of skin remains therefore largely obscure. This becomes even more important since homeostasis of the skin, as well as wound repair and pathogenesis of diseases, are particularly governed by intrinsic and extrinsic forces [23
As an alternative for ex vivo skin samples, many biomedical studies use isolated cells or epithelial cell monolayers to investigate mutations accounting for a variety of diseases like epidermolysis bullosa [24
]. To improve on the highly simplified and therefore artificial character of monolayers, multilayered skin equivalents can be formed [26
]. Such systems seem to largely mimic natural epidermal conditions. However, due to their dependence on rather stiff adhesion substrates and cultivation at the air-liquid interface, detailed mechanical characterization of such systems is still lacking [28
]. With ensured mechanical similarities, these model systems would greatly enhance our understanding of epidermal mechanics and cellular processes within.
For this reason, we developed a highly sensitive, environmental-controlled tensile testing machine. This instrument, which we call “tissue stretcher” for simplicity reasons, allowed, in a first direct approach, to evaluate and optimally compare the mechanical properties of a cell culture-based, multilayered epidermis model system with native tissue. Besides full-thickness adult rat skin, we used skin samples of 18-day-old rat embryos.
2. Materials and Methods
2.1. Tissue Stretcher Design
In this study, a custom-made stretching device for long-term, strain-controlled measurements of vital and even delicate tissue samples was developed and used. Adapted from our previously published stretcher for elastic silicone chambers [29
], the uniaxial movement is performed by a step motor that drives a linear translation stage (PA 24x29-30-SM02-Sonder, Steinmeyer, Germany). A custom-made LabVIEW-based (Version 2.0 FTDI, LabVIEW 2019, National Instruments, Austin, TX, USA) software tool enables the accurate spatiotemporal control of various parameters as, e.g., stretch frequency, amplitude and time interval between stretch cycles. The stage hosts a force transducer (SI-KG4 + Amplifier BAM21-LC, World Precision Instruments, Sarasota, FL, USA) for sensitive force measurements which can physically be connected to the grip by a titanium wire and a needle (Figure 1
). The specimen is mounted in a stationary holding frame on one side and in a fine, 5 mm wide clip made of steel sheet on the other side. During all measurements, the specimen is located in an environmentally controlled and medium-filled incubation chamber. A circulation thermostat (RMS6, Lauda, Lauda-Königshofen, Germany) flows heating liquid through the separated double-walled back of the incubation chamber to maintain a temperature of 37 °C. Additionally, an observation window on the front side allows a permanent visual control of the test specimen during all steps of the experiment.
2.2. Tissue Stretcher Calibration
To calibrate the force measurement of our tissue stretcher, cuboidal elastomer samples of the polydimethylsiloxane two-component formulation Sylgard 184 (Dow Chemical, Midland, MI, USA), a previously well-characterized material [30
], were prepared in certain mixture ratios and the Young’s modulus was determined via successive indentation as already described [32
]. The samples were mounted in the apparatus with 10 mm distance between the specimen holders and placed in the buffer-filled chamber unit. A prestretch of 3% (0.3 mm) was applied with an equilibration phase of 30 min. Subsequent straining was performed in two cycles of three successive stretch and release steps with 4% amplitude each (12% in total). Each step was held for 15 min. The dimension of the test samples in unloaded conditions were determined and the sectional area was used for calculation of the tensile stress σ as force per area unit with a Poisson’s ratio of 0.5. In the resulting stress-strain curve, the mean slope of linear regressions yielded the Young’s modulus.
2.3. Tissue Stretcher Noise Characterization
After system equilibration for at least 30 min, the holding grip was lowered and raised in buffer without any sample attached. Upper and lower amplitude were held for 1 min each; the full cycle was repeated four times. Force values were sampled at a rate of 10 Hz. For each sequence, histograms of all forces measured in the lower position and in the upper position, respectively, were taken. The cumulative normal distribution was fitted to these histograms. Noise value reported is the mean of all standard deviations from these fits. Amplitudes from these fits were used to measure buoyancy of the grip holder.
2.4. Generation of a Simplified Epidermis Equivalent (SEEs) as Multilayered Model System
Murine keratinocytes were isolated from wild type mice and routinely cultured in medium with low Ca2+
(50 µM) as described previously [33
]. To generate the multilayer epidermal model system, cells were grown on tissue culture plastic until confluence and switched to high Ca2+
medium (1.8 mM) to induce cell-cell contact formation. After 18–24 h, Ca2+
was depleted with EGTA (5 mM, E3889, Sigma, St. Louis, MO, USA) in PBS for 1 min and cells were trypsinized for 7 min (0.05%, 25300062, Thermo Fisher Scientific, Waltham, MA, USA). Subsequently, cells were replated in high density (4 × 105
keratinocytes per cm2
) in a total volume of 800 µL on fibronectin (20 µg mL−1
, 356008, Corning Inc., Corning, NY, USA) coated silicone elastomer chambers (4 cm², [30
]) made with the two-component silicone SORTA-Clear 12 (Smooth-On, Macungie, PA, USA) in a 1:1 ratio. After 6 h, dead and non-adhered cells were removed and fresh Ca2+
-containing medium was added. Resulting multilayer models were incubated overnight at 37 °C and 5% CO2
To transfer the simplified epidermis equivalents (SEEs) as free floating sheets in the tissue stretcher, one sidewall of the chamber was removed and the multilayered sheet was gently detached from the silicone using fine forceps. Subsequently, the SEE was folded 2 times and transferred on a C-shaped PVDF-membrane frame, which served as carrier and distance piece. Together they were mounted in grip and fixture of the holding frame. The lateral ligament of the PVDF frame allowed accurate length maintenance of the specimen. After mounting, the frame was cut and the prepared sample placed in the medium-filled incubation chamber. After connecting the motor-driven load cell with the moving grip, the needle was unblocked for subsequent analysis.
2.5. Preparation of Ex Vivo Full-Thickness Back Skin Explants from Fetal and Adult Rats
The tissue used in this study was isolated from animals prepared for other projects (Animal testing license: 81-02.04.2018.A90, LANUV NRW, Germany) and no additional animals were harmed. A pregnant rat (WISTAR, Janvier Labs, Le Genest-Saint-Isle, France) was sacrificed by CO2 and decapitated according to local regulations; 18-day-old fetuses were directly removed and decapitated before the full-thickness back skin was carefully excised and placed in cold Hank’s balanced salt solution (1× HBSS, H6648, Sigma, USA). To prevent the tissue from excessive mechanical stress during preparation, a removal of underlying hypodermal tissue was explicitly avoided. The explants were cut into rectangular pieces of 16 × 5 mm with its long axis in cranial-caudal orientation and stored in DMEM/F-12 medium supplemented with HEPES (15 mM, 11039021, Thermo Fisher Scientific, USA), penicillin (100 U mL−1), streptomycin (100 µg mL−1, 15140122, Thermo Fisher Scientific, USA) and FBS (10%, 10270106, Thermo Fisher Scientific, USA) at 37 °C until further use in stretching experiments within 8 h after isolation. For this, the fetal rat skin explants were transferred on a PVDF-membrane frame and mounted as described for SEEs above.
The adult skin samples were also isolated from the animal’s back. Therefore, a region of approximately 3 × 3 cm next to the spine was depilated and the used dimensions of 5 × 10 mm with long axis in cranial-caudal orientation were marked with waterproof ink. The skin was generously excised and placed in cold HBSS. Subsequently, tissue samples were cut to exact size and stored in medium at 4 °C until further use. All experiments with adult explants were equally performed at 37 °C after placing in the medium-filled chamber unit as described for fetal skin and SEEs. All measurements were started within 24 h after tissue isolation.
2.6. Tissue Thickness Determination
Tissue explants and SEEs were fixed in formaldehyde (3.7% in PBS, 1040021000, Sigma, USA) overnight at 4 °C (rat skins) or for 30 min at 37 °C (multilayer models). After brief washing in PBS, the adult rat skin was manually cut into thin slices with a sharp razor blade and placed laterally on a glass slide with some drops of buffer. Images of the slices and a calibrating scale were taken on a binocular microscope (Stemi 508, Zeiss, Jena, Germany) equipped with a camera (EOS 600D, Canon, Tokyo, Japan). The fetal rat skin was transferred in dissolved agarose (2% in PBS, A9539, Sigma, USA) at 60 °C, filled in cryomolds (10 × 10 × 5 mm, 4565, Tissue-Tek, Sakura Finetek, Alphen aan den Rijn, The Netherlands) and cured at room temperature. Subsequently, the agarose blocks were cut in 50 µm slices at the VT1000 S vibratome (Leica Microsystems, Wetzlar, Germany), collected on glass slides and imaged with a light microscope (Axio Vert.A1, Zeiss, Germany). Since the multilayered SEEs were too thin for proper sectioning, the fixed cell sheets were permeabilized with Triton-X 100 in PBS (1%, T8787, Sigma, USA) for 15 min at room temperature, washed with PBS and stained with Phallodin-Atto633 (1:300, 68825, Thermo Fisher Scientific, USA) for 4 h at room temperature. Cell nuclei were stained with NucBlue in PBS (R37606, Thermo Fisher Scientific, USA) for 1 h at room temperature. After washing with PBS, the samples were mounted in Fluoromount and covered with a cover glass. Confocal z-stacks of the full thickness layer were taken at a laser scanning microscope (LSM 880, Zeiss, Germany). Thicknesses were determined as mean of several independent measurements.
2.7. Tissue Stretch Protocol and Data Analysis
After transfer of the samples in the tissue stretcher and complete cut of the PVDF transfer frames, the starting sample length was set to 10 mm using a single-lens reflex camera (EOS600D, Canon, Japan) equipped with a macro objective (Macro lens EF-S 60 mm, Canon, Japan). After equilibration for approximately 30 min, the respective protocol was directly started. Data points were recorded every 0.1 s. Photos of the tissues were manually taken during static stretch protocol at certain time points for every cycle. The normalization of the initial force relaxation for each strain cycle was related to each cycle’s peak force. Accordingly, the mean and standard deviation of at least 3 different samples for every data point were calculated. Furthermore, the tissue force recovery was defined as the mean peak force of a certain cycle divided by the mean peak force of the first cycle. In order to apply cyclic tensile forces, we changed the straining protocol to a trapezoidal approximation of a sine wave for some experiments. Here, the tissues were cyclically loaded and unloaded with identical velocity compared to static stretch. The dwell time at loaded and unloaded position was set to 0.4 s to result in a stretch frequency of 0.3 Hz. For further analysis and visualization, only peak forces of every cycle were plotted.
2.8. Viability Assays of Fetal Rat Skin and SEEs
For viability analysis, fetal rat skin explants were stretched and subsequently transferred in medium with fluorescein diacetate (25 µg mL−1, F7378, Sigma, USA) and ethidium bromide (10 µg mL−1) for 15 min at 37 °C. A positive control was kept in medium supplemented with 1% sodium azide for 2 h prior to staining. Imaging was directly performed after a brief washing step with fresh medium at the LSM 880 confocal microscope (Zeiss, Germany) with a 20× plan apochromatic objective and appropriate filter settings. The SEEs were transferred in trypsin-EDTA solution (0.05%, 25300062, Thermo Fisher Scientific, USA) and incubated for 10 min. Once the cell sheets started to break into pieces by smooth agitation, the SEEs were mechanically dissolved by pipetting up and down. The cells were then centrifuged and the pellet either resuspended in PBS with FBS (10%, 10270106, Thermo Fisher Scientific, USA) or in PBS with EtOH (35%) as positive control. The following viability assay was performed with the LIVE/DEAD™ Fixable Cell Stain Kit (L34963, Thermo Fisher Scientific, USA) according to the manufacturer’s protocol and analyzed by flow cytometry (Cytoflex S, Beckman Coulter, Brea, CA, USA).
2.9. Data Analysis
Statistical analysis and plots of recorded data were performed using GraphPad Prism software (Version 9, GraphPad Software Inc., San Diego, CA, USA). Microscopic images of tissue sections were evaluated using Fiji ImageJ software (Version 1.53c, open source scientific image processing [34
]). Schemes and figures were designed using CorelDRAW software (Version 21, Corel, Ottawa, ON, Canada).
Due to their inherent fragility and often low reaction forces, biomechanical testing of delicate tissue samples or cell culture tissue models under physiological conditions can be quite challenging. The instrument developed here closes the gap between classic tensile test devices known from material sciences and very sensitive experimental approaches to measure mechanical properties on a molecular or cellular level [35
Our instrument exhibited a noise level of 22 µN at a sampling rate of 10 Hz. This noise in forces corresponds for simplified skin equivalents to a standard deviation in stresses of 100 Pa; for fetal skin, the corresponding value is 7.6 Pa. As can be seen, for example, in Figure 4
, this level of scatter is by far sufficient for tensile tests at physiologically relevant amplitudes. In many instances, relevant data can be averaged of much longer periods than 0.1 s. Therefore, most parameters can be measured with significantly higher accuracy. Moreover, many mechanical experiments on very soft tissues suffer from artifacts caused by clamping-induced cutting of the tissue. These we could circumvent for the softest samples by the use of a relatively stiff carrier made of PVDF membrane sheets. In addition, the frames facilitate a reproducible free tissue length at zero strain position based on their physiological length in situ. The lateral ligament of the PVDF transfer frames was cut after mounting of the sample and did, therefore, not affect the measurement. As a result, we never observed significant artifacts due to slipping or tearing of tissue at the fixtures.
In this study, the device enabled, for the first time, a direct comparison of mature full-thickness skin, fetal skin and an ECM-free epidermal model tissue (simplified epidermis equivalent—SEE) (Figure 3
and Figure 4
). Please note that under physiological or growth conditions, fetal skin and SEE are immersed in medium whereas the dermal side of adult skin is in contact with interstitial fluid while the epidermal side is exposed to air. Therefore, our aim of a direct comparison of adult skin, fetal skin, and epidermal equivalents necessitated an instrument design where samples are immersed in medium throughout the measurements. However, for adult skin alone, measurement in air and in medium are equally close to physiology. As measurement in air is much easier and can be performed with commercial equipment, most work on mature skin is done in (humidified) air. Nevertheless, especially in the low strain region, fluid contact and detailed nature of the fluid are important determinants for the mechanics of soft collagenous materials, in general, and skin, in detail [38
]. All mechanical analyses in our study were performed with hydrated tissue in a medium-filled chamber. Thus, it is not surprising that we measured tissue stresses on adult skin that are lower than most reported values at identical nominal strain.
Another important aspect in which our measurements on adult skin are set apart from the bulk of the literature is the strain regime. Whereas most measurements on adult skin were performed in the high strain linear regime, where collagen fibrils are almost uniformly aligned and intra- and interfibrillary sliding are the most important mechanisms of elasticity [5
], we confined our measurements to lower strains and, as a consequence, significantly lower stresses (Figure S2A
). Additionally, we were interested in both the initial, first response to mechanical stretch but also the history-dependent behavior of repeated stretch and release cycles. In accord with this choice, we did not precondition our samples as it is routinely done for measurements that focus on material response in the linear regime [6
In our experiments, we focused on physiological conditions and therefore stress levels below 30 kPa, which were largely ignored in previous studies (Figure 4
A–D). In fact, a recent, careful study of skin microstructure during stretch reported collagen fiber reorientation, stretching and sliding to be triggered only by stress above 1 MPa (corresponding to strains of more than 40%) whereas at lower deformations, straightening of the curved collagen fibers prevailed [11
As our samples were not preconditioned, the first force relaxation curve of adult skin differed from later ones (Figure 5
]). Although not fully understood, this history-dependent change in viscoelasticity is generally explained and modeled by a microstructural adaptation of ground substance and elastin fibers at low strain and a time-dependent increase in reference length of the collagen fibers at enhanced strain amplitudes [41
]. Our experiments confirm that this effect is still present after 1 h in rest position and supports the idea of a robust microstructural strain-dependent adaptation of dermal tissue. Furthermore, neither fetal rat skin nor SEEs showed a similar hysteresis (Figure 5
C,D). While our epidermal model system does not contain any ECM fibers, these mechanically relevant proteins are not yet entirely secreted and cross-linked in the highly cellularized fetal rat skin dermis [15
]. This immature condition of embryonic day 18 rat skin led to a markedly lower viscoelastic phenotype in previous shear creep tests and was surprisingly well modeled by soft gelatin gels [42
]. The gel-like characteristic was not only displayed by a notably fast force relaxation in repeated stretch cycles but also by a very low force recovery during 1 h rest at starting position (Figure 5
C and Figure 6
B,D). Besides, earlier histological studies have demonstrated severe fractures in the epidermis and dermis of fetal skin after creep tests until failure [43
], which argues for plastic deformations or, in other words, irreversible change of dermal microstructure. Based on well-established analytic models of poroelasticity [38
] and current simulation models [44
], the remaining small initial peaks in repetitive stretch cycles, followed by an almost instantaneous drop to basal force values (Figure 4
C), possibly reflect hydrodynamically driven water extrusion in an ultrasoft, poroelastic material.
Interestingly, tissue length is not equally affected and revealed an impressive recovery. Indeed, fetal skin samples reproducibly regain their initial length within the rest periods of the test protocol (Figure 4
E). This tissue strain recovery was similarly observed for epidermal equivalents in our experiments. Together with the absence of measurable equilibrium forces in fetal skin and the total absence of ECM itself in SEE, these observations implied an ECM-independent process. Although limitations in profile observations during tensile testing caused by the incubation chamber design prevented own verification, earlier studies have already demonstrated that the tissue dimension of rabbit skin was restored ex vivo after a long relaxation time [45
] but full recovery of mechanical properties was exclusively found in vivo after several hours of rest [18
]. Upon length reduction, suspended cell mono- and multilayers have previously shown an actomyosin-dependent mechanical adaption that results in fast tissue flattening and length recovery [46
]. This active process is dominated by structural changes in the actomyosin cortex and is responsible for stress relaxation and a change in tissue length [47
]. In accordance with this concept, keratinocytes of fetal epidermis or our epidermal model system actively restored their initial dimension while dead tissues were unable to regain their pre-stretch configurations (Figure 3
D; Figure S3
). Additionally, our SEEs displayed a high force recovery after a first stretch while mature skin showed a markedly larger decrease of tissue peak force (Figure 6
D). Interestingly, in the second half of the stretch protocol, both model systems recovered at a similar rate to the previous cycle. While during the first stretch, a formerly unstressed adult rat skin behaves more like a gel-like material similar to fetal skin, the normalized force relaxation of the following cycles is almost identical to the relaxation of ECM-free SEEs (Figure S4
). Although structurally simplified, we therefore conclude that in physiological regimes of repetitive force application, our multilayered SEEs mimic dermal mechanics and possess many features of full-thickness skin.
In the response to applied forces, a fundamental difference between mature connective tissue and living cells emerged when we applied comparatively fast cyclic stretch instead of a constant strain. While in SEEs, higher amplitudes did not affect force relaxation, we observed a surprisingly faster relaxation when the specimens were cyclically instead of statically stretched (Figure 7
D). However, cyclic stretch did not generally alter the relaxation behavior as shown for adult full-thickness skin in our observations here and earlier studies by others [48
]. On a subcellular scale, all cytoskeletal systems show a cyclic strain-induced reorientation in a specific time-dependent manner [49
]. In particular, in mechanically integrated cell monolayers, F-actin reorientation upon stretch goes along with cooperative reinforcement that is strongly dependent on intercellular junctions [29
]. Cytoskeletal deformations and subsequent dynamical regulation of actomyosin were identified to be responsible for structural integrity and stress relaxation of cell monolayers [47
]. When cyclic stretch is applied, the cytoskeleton fluidizes after a few minutes, followed by a substantial gain in order with increasing duration of cyclic straining [51
]. Consequently, we propose similar strain-induced adaptation processes in native epidermal tissue under physiological conditions.
In summary, we present here a direct mechanical comparison of simplified epidermal equivalents, fetal, and adult rat skin. Despite the structural differences between these samples, mechanical behavior exhibited surprising similarities that highlight the high value of epidermal equivalents as in vitro model systems for basic research in dermatology and epidermal cell biology.