In Vivo Visualization of Collagen Transdermal Absorption by a Combined Second-harmonic Generation and Two-photon Excited Fluorescence Method

The transdermal administration of collagen is an important method used for wound healing and skin regeneration. However, due to the limitations of previous approaches, the process and degree of collagen transdermal absorption could only be quantitatively and qualitatively assessed in vitro. In the present study, we introduced a novel approach combining second-harmonic generation with two-photon excited uorescence to visualize the dynamics of collagen transdermal absorption in vivo. The high resolution images showed that the exogenous recombinant human collagen permeated through the epidermis, reached the dermis and formed reticular structures in real time. We also validated these ndings through traditional in vitro skin scanning and histological examination. Thus, our approach provides a reliable method of measurement for the real-time evaluation of collagen absorption and treatment effects. this study, we used in vivo SHG-TPEF skin imaging to track the dynamic process of the transdermal absorption of exogenous collagen in a mouse model. We observed that exogenous collagen gradually the dermal skin and formed a reticular structure, which was also conrmed by traditional in vitro skin imaging. This is the rst study to use a noninvasive approach to visualize the absorption of exogenous collagen in skin in vivo, which could extend our understanding of the process of skin and provide the feasible technology for the clinical assessment of drugs for external use.


Introduction
The skin is the largest organ in the human body and plays a key role in protecting our bodies from the varied external environment [1]. It is mainly composed of three layers that span from super cial to deep, including the epidermis, dermis, and hypodermis [2]. The dermal skin contains components of the extracellular matrix, such as collagen bers, that contribute to the strength and toughness of the skin [3]. As the main component of connective tissue, collagen is the most abundant protein in skin, comprising 25% to 35% of the protein content in the whole body [4]. Collagen is also widely used in cosmetics and biomedical products [5,6]. After exogenous collagen is smeared on the skin, it permeates through the cuticle to the dermis and forms a reticular structure to support the structure of the dermis and supply more extracellular matrix to keep the skin elastic [7]. Thus, we need to dynamically visualize the process of the transdermal absorption of exogenous collagen and measure the extent of collagen absorption with qualitative and quantitative approaches.
To date, various technologies have been applied to measure the biomedical administration of collagen on skin, including chemical assays [ 8 ], tissue inspection [9], isotope tracing [10] and circulating diffusion chamber [11], which can only be used to quantitatively examine the transdermal absorption of exogenous collagen in vitro. However, these approaches could not dynamically track the process of transdermal absorption in vivo and evaluate the real-time effects of biomedical administration. These limitations could be overcome by nonlinear optical microscopy, such as two-photon excitation uorescence (TPEF) [12,13], and second harmonic generation (SHG) [14,15] Pittet, 2014 #16}]. Among such technologies, SHG is a well-established method to directly visualize anisotropic biological tissues that possess large hyperpolarizabilities, such as collagen [16].
Moreover, SHG imaging has many advantages, such as a higher signal-to-noise ratio than auto uorescence imaging, the detection intrinsic signals without extrinsic dyes and the utilization of nonlinear excitation from molecular uorescence [17]. Nonetheless, few studies have examined the effects of drug administration on the skin using SHG because the challenges in using it to distinguish endogenous from exogenous collagen. Recently, two-photon excited uorescence (TPEF) imaging has been widely applied to visualize the dynamics of intrinsic molecules and cells in living specimens through conventional labeling via staining or uorescent proteins [18]. Due to the divergent mechanisms of SHG and TPEF, we speculate that they could be combined to provide complementary information for the dynamic observation of biological processes. However, few studies have reported the in vivo imaging of the transdermal absorption of collagen with a combination of SHG and TPEF (SHG-TPEF).
Therefore, there is an urgent need to obtain in vivo high-resolution images of brillar collagen during the process of transdermal absorption using SHG-TPEF imaging, which could provide a novel approach for estimating the extent of drug absorption and further assessing the treatment effects of dermal drugs.
To address these issues, we rst constructed the mouse model of transdermal absorption through administrating exogenous recombinant human collagen (R-hc) which was proved to be effective in the skin absorption in our previous studies [19]. Second, we set up the platform of in vivo SHG-TPEF skin imaging. Third, we tracked the dynamic process of collagen absorption of short-term administration by in vivo SHG-TPEF imaging. Finally, we applied Franz-type diffusion method to evaluate the collagen absorption at different time points as validations.

Experimental
Animals Healthy BALB/c mice were purchased from Vital River Laboratories (China). All the animals used in this study were male adults (6-8 weeks, 20±2 g body weight). Mice were kept in a speci c pathogen-free environment with freely available food and water and a 12 h light-dark cycle. In this study, ve mice were used at each time point, each of which received a topical dose of 1mg/ml of R-hc on a shaved back on the left side of the thoracic spine and a saline control area on the right. The animal experiments were supervised and approved by the Research Ethics Committee of the China Academy of Chinese Medical Sciences( ERCCACMS21-2106-12).
Quality control of R-hc R-hc: R-hc is expressed by eukaryotic cells and integrates the human-like collagen gene, which is a heterozygous type I and type III collagen gene, into the yeast chromosome. The hydroxyproline in R-hc was replaced by proline. In our previous study, we analyzed the effect of R-hc on the repair of laser skin damage in mice. The results showed that R-hc can signi cantly accelerate wound healing, shorten wound healing time and promote collagen production in wounds [19], which suggests that R-hc could be transdermally absorbed.
Fluorescent label: For uorescent labeling, a protein stock solution was prepared by mixing 100 mg R-hc freeze-dried powder (TianJin Irheaya Biological Technology Co., Ltd., Tianjin) with 10 mL 1 M phosphate buffer (pH 8.5~9.0) to produce the 1 mL protein labeling stock solution. To prepare the dye stock solution, anhydrous DMSO was added into a vial of iFluor TM 750 SE (AAT Bioquest, Ex/Em=749/775 nm) to produce a 10-20 mM stock solution, which was mixed well by vortexing. Then, 1 mL of the dye stock solution was added to the vial of protein solution (10 ml) and shaken thoroughly. The concentration of the protein was 10 mg/ml. The reaction mixture was rotated at 4°C for 12 h. The reaction mixture was loaded into a dialysis bag (<1 kd) and placed in a large beaker containing ultra-pure water in the dark. The water was changed every 4 h until no uorescence was detected in the ultra-pure water. The uorescent protein was freeze-dried into a powder for future use.

Immuno uorescence staining
The localization of collagen III and collagen I protein was assessed by immuno uorescence. The skin tissue was taken from the left side of the thoracic spine after treated with R-hc for 0 h, 1 h, 2 h, 3 h, 4 h, 5 h and 24h. The frozen sections were obtained from an ultralow temperature freezer after 30 min at room temperature. The sections were washed in 0.02 M PBS 3 times and then blocked with QuickBlock™ Blocking Buffer for Immunological Staining (Beyotime, China) for 1 h. A 1:500 dilution of anti-collagen III mouse monoclonal antibody and anti-collagen I rabbit polyclonal antibody (ab6310, ab34710; Abcam, USA) was added and incubated overnight at 4°C. After washing with PBS, the sections were incubated for 1 h at room temperature in the dark with a secondary antibody (Dylight 488-goat-anti-mouse and Dylight 594-goat-anti-rabbit; EARTHox, China). After immunostaining, the specimens were examined with a wideeld uorescence microscope (BX51, Olympus, Tokyo, Japan) tted with a digital camera (DP71; Olympus). The relative intensity of the staining was analyzed by Image Pro Plus 6.0 software (Media Cybernetics, Inc., USA).

In vivo SHG-TPEF skin imaging
The back hair was shaved from the mice, which were treated with iFluor-R-hc (1 mg/ml) for 0 h, 1 h, 2 h, 3 h, 4 h and 5 h at room temperature while avoiding light. At the end of the experiment, all mice were anesthetized with Iso urane and xed on a mouse plate while the back skin was xed with holder. The full-thickness skin tissue was scanned by a two-photon scanning confocal microscope (FV1000-MPE, Olympus, Japan).
A two-photon excited uorescence microscopy system was used for image acquisition based on the reverse detection of SHG and two-photon excited uorescence signals from the skin. The two-photon excited uorescence microscopy system was equipped with a water-immersion objective lens (PLAN 25X OB. W IMM, NA 1.05, Olympus) and a Ti:sapphire laser oscillator (wavelength: 690-1040 nm; repetition rate: 80 MHz, pulse width: 100 fs; MaiTai HP DS-OL, Spectra-Physics, Inc. CA). The SHG image acquisition was performed at an excitation wavelength of 950 nm with a 690 nm dichroic mirror and a 475/20 nm emission lter (center wavelength/bandwidth). For the observation of iFluor-R-hc, an excitation wavelength of 790 nm and a 605 nm-680 nm emission lter were used to isolate the SHG signal. The auto uorescence image was acquired with an excitation wavelength of 750 nm and a 495 nm-540 nm emission lter.
The images were obtained from the epidermis and the dermis and stored as a z-stack image sequence (step size of 1 μm in the z-axis). The image size was 512×512 pixels, and the scanning speed was 2 μm/pixel. The merging of the color channels was performed using FV1000 Viewer 3.1 software (Olympus). The three-dimensional image was performed using Imaris (Imaris ver. 8.4.2, Bitplane AG, Switzerland). To examine the collagen staining intensity, we calculated the SHG signal intensity, which has been considered a good predictor of collagen structure [20,21]. Thus, we calculated the signal intensity of SHG in the dermis at different time points after iFluor-R-hc administration and determined the collagen thickness based on the z-stack depth of the SHG signal.
In vitro SHG-TPEF skin imaging Whole skin tissue (1×0.5 cm 2 ) was removed at 0 h, 1 h, 2 h, 3 h, 4 h and 5 h after the imaging of the mice. The tissues were examined as frozen sections (10 μm). The lateral plane of the skin was exposed under the objective lens of a two-photon excited uorescence microscopy system. The above mentioned scanning process was repeated to observe the iFluor-R-hc transdermal absorption. The laser condition settings were consistent with those used for the above-mentioned in vivo SHG-TPEF skin imaging.

Hematoxylin and eosin staining
All frozen sections were subjected to hematoxylin and eosin (HE) staining for the histological study. Differential interference contrast images of the sections were captured using an Olympus BX51 microscope (PlanApo 20x/NA: 0.45).

The Franz-Type Diffusion
Mouse skin was mounted in the receptor compartment of a Franz-type diffusion cell (TP-6, HT, Jintao Instrument Technology Co., Ltd., Tianjin, China). Ten milligrams of iFluor-R-hc dissolved in normal saline or normal saline alone was added to the tight interface facing the mouse skin. The donor cap was covered with tin foil. A 0.9% sodium chloride buffer solution (pH 7.4) was used as the receptor solution. The receptor solution was stirred by a magnetic stirrer rotating at 300 rpm at 37℃. The samples were examined at 1 h, 2 h, 3 h, 4 h and 6 h. The amount of total protein in the receptor solution was measured by a BCA protein assay kit (Beyotime, China), and the uorescence intensity was measured with a uorescence microplate reader(Synergy H1Hybrid Multi-Mode Reader, BioTek Instruments, Inc., USA) at Ex/Em= 640/670 nm.

Gel electrophoresis
The amount of total protein in the receptor solution was separated by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE). The samples were stained with Coomassie Brilliant Blue R-250 stain (P0017F, Beyotime Biotechnology, China) then scanned with ChemiDoc™ Touch Imaging System (Bio-Rad Laboratories, Inc., USA) .

Statistical analysis
The results were statistically analyzed by one-way ANOVA and differences between groups identi ed using Tukey's post-hoc test using GraphPad Prism v.6.0 (GraphPad software, Inc., La Jolla, Calif.). The difference was considered statistically signi cant at p<0.05. Post-hoc analysis was performed with p<0.05 as the threshold for statistical signi cance.

R-hc transdermal absorption
After 24 h of R-hc administration, we observed that the uorescence intensity of type III and type I collagen were signi cantly increased in the dermis relative to that observed in the control group (Fig. 1A). At the same time, we calculated the ratio of type III and type I collagen, compare to the control group the rate of increase of type III collagen was more notable. We also observed that type III collagen was interwoven into the reticular structures in the super cial dermis (Fig. 1B). The long ber content of type III collagen in the R-hc treatment group was signi cantly increased compared with that in the control group (Fig. 1C). These ndings suggest that R-hc could be transdermally absorbed and supplement type III and type I collagen in the dermis.

SHG-TPEF skin imaging
We characterized the structure of the skin and the iFluor-R-hc distribution from the epidermis to the dermis. Speci cally, in the epidermis, we used TPEF imaging at an excitation wavelength of 750 nm to observe the stratum spinosum and stratum basale ( Fig. 2A). Then, in the dermis, we separately used SHG and TPEF imaging at excitation wavelengths of 950 nm and 790 nm, respectively, to visualize the dynamic process of iFluor-R-hc absorption (Fig. 2B ). Here, we used two-dimensional (2D) images and three-dimensional (3D) images to illustrate the scanning results. In the 2D images, we scanned the histological structures of the skin by utilizing the SHG signal (λ=950 nm) and iFluor-R-hc by utilizing the TPEF signal (λ=790 nm) (Fig. 2B). Moreover, as shown in Fig. 2B, we revealed intrinsic collagen, and the exogenous iFluor-R-hc was mutually linked into reticular structures in the dermis.
Tracking the transdermal absorption of iFluor-R-hc using SHG-TPEF imaging in vivo In this paper, the infrared uorescent probe is used to label the collagen to be detected, which can avoid the auto uorescence of the skin, reducing background noise [22].We rst analyzed the mouse skin at the 0 h time point and showed a small amount of noise in the epidermis and the dermis and TPE auto uorescence in the hair shaft (Fig. 3A), which were considered to represent the baseline. Over time, we observed that the TPEF signal gradually increased (Fig. 3A), which suggests that iFluor-R-hc permeated through the cuticle and reached the dermis. Then, it formed reticular structures in the epidermis after 4 h (Fig. 3A). Statistically, the iFluor intensity signi cantly increased after 2 h administration of iFluor-R-hc (Fig. 3B), which suggests that R-hc was gradually absorbed after 2 h of application. Moreover, compared with the baseline, both the SHG signal intensity and collagen thickness were signi cantly increased after 4 h and 5 h of administration of iFluor-R-hc ( Fig. 3C and 3D In addition, we found that iFluor-R-hc could permeate and disperse around hair follicles after 3 h of administration (Fig. 4). This result is consistent with the results of our previous in vitro experiments that the protein rst accumulated in the hair follicle and then absorbed into the dermis [23], which suggests that R-hc may be absorbed into the dermis through the hair follicle.
To test whether our ndings were the result of false positive signals in the in vivo SHG-TPEF imaging, we performed histological analysis. The results revealed that there is an increase tendency of HE staining but no statistical difference between each group (Fig. 5A). However, we found that both iFluor and the SHG intensity of iFluor-R-hc were signi cantly increased after 4 h and 5 h of administration of iFluor-R-hc ( Fig.   5B and 5C), which was consistent with the ndings obtained from the in vivo SHG-TPEF imaging.
We used immunohistochemical methods to distinguish the changes of type I collagen and type III collagen with the time of R-hc administration in dermis. The results showed that the uorescence intensity of type III collagen in the dermis began to increase signi cantly after the 4h of R-hc administration, and the type I collagen increased signi cantly after 5 hours (Fig 6A, B and C). The ratio of type III collagen to type I collagen increased signi cantly after 2 h of R-hc administration (Fig 6D), suggesting an increase in the content of type III collagen in the dermis and it grows with the time of administration of R-hc.
In vitro tracking of the transdermal absorption of iFluor-R-hc Next, we used traditional in vitro Franz diffusion cells to test whether in vivo SHG-TPEF imaging could sensitively and accurately track transdermal absorption, which could con rm the generality of our novel approach. Speci cally, compared with the baseline (0 h), we rst observed the signi cantly increased uorescence intensity of iFluor-R-hc after 3 h of administration (Fig. 7A). Moreover, the iFluor-R-hc group showed signi cantly increased protein concentrations after 2 h of administration of iFluor-R-hc (Fig. 7B).
Compared with the control group, the iFluor-R-hc group showed both signi cantly increased iFluor intensity and iFluor-R-hc concentration after 4 h of administration of iFluor-R-hc ( Fig. 7C and 7D). These ndings were consistent with the observations of the in vivo SHG-TPEF imaging, which suggests that our novel approach could sensitively and dynamically visualize collagen transdermal absorption in vivo. In addition, in the collected liquid, we found a clear target protein band about 55 kD after 4 h of administration in the iFluor-R-hc group (Fig.7E), which suggests that R-hc can pentrate the full layer of skin after 4 h administration of iFluor-R-hc.

Discussion
In this study, we used a novel noninvasive approach to dynamically capture the process of collagen transdermal absorption in vivo by combining SHG with TPEF. Both two-photon uorescence imaging and second harmonic imaging are second-order nonlinear processes. In addition to the needs of different lters addition, TPEF and SHG are almost identical on the optical detection system, the obtained information for both complement and con rm each other, so it has received much attention in recent years. In this paper, the infrared uorescent probe is used to label the collagen , which can avoid the auto uorescence of the skin and reducing background noise [22]. Over time, we observed that collagen permeated through the hair follicle, reached the dermis and formed reticular structures in the dermis after 4 h of administration of exogenous iFluor-R-hc, which was also con rmed by the in vitro examinations. To the best of our knowledge, this approach is the rst to be used for the in vivo visualization of the transdermal absorption of biochemically administered collagen, and these ndings provide a technological method for the clinical assessment of transdermal drugs in the future.
A previous study showed that compounds with a molecular weight greater than 50 kDa penetrate the dermis with di culty because of blockage by the stratum corneum [24] , but whether macromolecules can penetrate the skin is still not clear. In our study, we used in vivo SHG-TPEF skin imaging to track the process of the transdermal absorption of exogenous collagen, which is approximately 55 kDa in size. The iFluor uorescence signal was rst detected in the hair follicle 3 h after coating R-hc on the skin, which con rms that macromolecules could be absorbed by the skin. Moreover, this combined technology has many advantages, such as noninvasiveness, the absence of conventional dyes and good penetration of biological tissues [17]. Thus, it allows the long-term observation of dynamic changes at the same skin site, which can be extended to the development of an in vivo real-time assessment method for cosmetics and biomedical administration.
The results of the examination of R-hc transdermal absorption provided a quantitative approach for the in vivo assessment of the transdermal absorption of protein drugs through the combined measurement of collagen density and absorption. Speci cally, we rst demonstrated that a typical application of R-hc increased type III collagen and collagen ber growth, which suggests that R-hc could increase skin elasticity. Second, we used TPEF combined with SHG imaging to study the manner and depth of R-hc absorption in vivo. Since the stratum corneum acts as a physical barrier in skin on the surface of the epidermis [25], it has an impact on the penetration e ciency of R-hc. However, the results of the in vivo transdermal absorption model showed that the uorescence intensity of type III collagen was signi cantly increased after 24 h of R-hc administration, which suggests that iFluor-R-hc could still effectively permeate the epidermal skin. Potts and colleagues found that an increased molecular weight increased lipophilicity and permeability, which suggests that the penetration of drugs into the skin is tightly associated with the molecular characteristics of the drug [26]. Lin and colleagues also considered that the effect of lipophilicity is a more dominant factor than the molecular weight in the passive diffusion permeability [27] . We observed the uorescence signal of iFluor in the hair follicle after 3 h of Rhc application by in vivo SHG-TPEF imaging, and the dermis was interlaced into a reticular network after 4 h. Therefore, we assume that the osmotic effect of R-hc in the dermis could be attributed to higher permeability.
HE staining is a classic method to detect the pathophysiological changes of skin, but it can only observe the skin's thickness changes in various layers, in ammatory cell invasion and other indicators, but the absorption & distribution route and collagen content, which are directly related to wound healing or drug e cacy assessment, can not be observed. In our study, we examined SHG intensity to measure the content of collagen in the dermal skin. SHG imaging provides increased resolution and contrast when examining collagen than typical histological staining analysis [28]. Moreover, it is also a feasible and fast method that could be applied to the imaging of thicker tissue without xing, slicing and staining [29]. SHG intensity could accurately re ect the in vivo changes in collagen. Speci cally, in the examination of HE staining of R-hc, we found that the amount of R-hc gradually increased after iFluor-Rhc administration and reached a peak at the 4 h time point. Similar results were also observed in the analysis of Franz-type diffusion, which was consistent with the ndings of the in vivo SHG-TPEF skin imaging. These ndings indicate that SHG imaging could provide reliable measurements for the real-time evaluation of collagen absorption.
Collagen is one of the core constituents that maintains the elasticity of skin; thus, reducing collagen in the skin would result in wrinkled and abby skin [30]. In our study, we observed that R-hc gradually permeated the dermal skin and formed the reticular structure after 3 h of administration. This suggests that exogenous collagen could not only supplement skin collagen to meet the needs of anti-aging cosmetics but also form granulation tissue together with intrinsic collagen to ll wounds in tissue.
SHG-TPEF imaging allows us to observe the manner of transdermal absorption and the aggregation sites of R-hc in vivo, but some limitations still exist for this technology. For example, individual differences in cortical thickness might cause variation in the scanning results, which could have an impact on examining SHG intensity. Alternatively, the auto uorescence of tissue might affect the detection of exogenous collagen labeled with uorescent probes. A previous study showed that the excitation wavelength of tissue auto uorescence was between 720 nm and 880 nm, and the overall signal sensitivity (420-490 nm) was affected by the convolution of the spectral transmittance and the detector e ciency [31]. Therefore, the use of infrared dyes could be useful for the uorescent labeling of exogenous proteins, and bandpass lters could be adjusted to reduce the effects of auto uorescence on skin scanning.

Conclusion
In this study, we used in vivo SHG-TPEF skin imaging to track the dynamic process of the transdermal absorption of exogenous collagen in a mouse model. We observed that exogenous collagen gradually permeated the dermal skin and formed a reticular structure, which was also con rmed by traditional in vitro skin imaging. This is the rst study to use a noninvasive approach to visualize the absorption of exogenous collagen in skin in vivo, which could extend our understanding of the process of skin absorption and provide the feasible technology for the clinical assessment of drugs for external use.

Declarations
Ethics approval and consent to participate: The animal experiments were supervised and approved by the Research Ethics Committee of the China Academy of Chinese Medical Sciences( ERCCACMS21-2106-12).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.