2.1.

Chemical Agents

Considering the solubility of disaccharides and theoretical investigation, sucrose and maltose were selected as the representatives to compare with fructose, which has been reported to have a greater OCP among the three monosaccharides.¹⁶ Figure 1 shows the chemical structures, respectively.

Fig. 1

Chemical structure of sucrose, maltose, and fructose.

In this study, sucrose, maltose, and fructose were purchased from the Sinopharm Chemical Reagent Co., LTD (Shanghai, China). To evaluate the OPC of the OCAs, various agents with some concentration gradient were prepared for the ex vivo experiments. Their RIs were measured using an Abbe digital refractometer (WAY-2S, Shanghai YiCe Apparatus & Equipment Co., LTD, Shanghai, China). Since skin optical clearing methods have universality for various optical imaging techniques and the pH value of OCAs may produce influences on the fluorescence proteins or dyes, the pH values of each OCA with different concentrations were measured using pH-Indicator strips (Sigma-Aldrich Co., Shanghai, China). The RIs and the pH values of each agent at different concentrations (Con.) were listed in Table 1.

Table 1

The RIs and the pH value of each agent at different Con. (M refers to mol/L).

2.2.

Molecular Dynamics Simulation

MD simulation was employed to study the interactions between chemical agents and collagen. Here, the OCP of each agent was quantitatively evaluated by analyzing the number of hydrogen bonds formed between OCAs and collagen mimetic peptides. MD simulation software CHARMM (version-36, Chemistry at Harvard Macromolecular Mechanics, Harvard University, United States) with CHARMM22 force field and the carbohydrate force field was used. 1BKV (Protein Data Bank ID) and a regular GPO peptide ${[{(\mathrm{GPO})}_{10}]}_3$ (G, glycine; P, proline; and O, hydroxylproline) are representative collagen mimetic peptides for MD simulation.²⁹ The information of the computational hardware is as follows: Intel(R) Core(TM) i5 CPU at 2.80 GHZ, 6 GB of physical RAM, Linux. The MD simulation of all agents proceeded in parallel, and the runtime was about 17 h.

Each agent was applied to simulate with both 1BKV and GPO, respectively. On account of disaccharide molecular size limits, 10 disaccharide molecules (sucrose or maltose) were placed randomly around the collagen mimetic peptide at a radial distance of 12.0 Å away from the cylindrical axis of the peptides.²⁹ By contrast, 12 monosaccharide molecules (fructose) were placed in a similar way according to the previous study.¹⁶ Each production run lasted 600 picosecond (ps) and the coordinates were recorded every 1 ps. To prevent diffusion of the agent molecules away from the collagen, a cylindrical potential shell was applied, and the harmonic constraints were used to anchor the peptide. The detailed simulation process was reported by Hirshburg et al.²⁹

The average numbers of hydrogen bonds between each agent and mimetic peptides during 600 ps were analyzed after the entire simulation. The cutoff radius of hydrogen bonds was set to 2.4 Å.³² The number of hydrogen bonds formed between every molecule and collagen every 1 ps was counted to evaluate the skin OCP of sucrose, maltose, and fructose, respectively.

2.3.

Animal Preparation

Male Sprague-Dawley rats (60 g, $n=40$) were bought from Wuhan University Center for Animal Experiments (Wuhan, China). All animals were fed under a specific pathogen free level of feeding condition. The animal care and experimental procedures were approved by the Experimental Animal Management Ordinance of Hubei Province, China. Rats were anesthetized by an intraperitoneal injection of a mixture of 10% urethane and 2%-$\alpha$-chloralose with a dosage of $0.8\mathrm{mL}/100\mathrm{g}$. After rats were thoroughly unhaired with depilatory cream, they were placed on the experimental platform.

2.4.

Ex Vivo Experiments

Thereafter, we removed the dorsal skin and cleaned the subcutaneous fat, fascia, and blood completely. Then the fresh skin samples ($5\mathrm{cm}\times 5\mathrm{cm}$, $N=95$) were immersed in phosphate buffer solution and kept under 4°C until experiments were performed in $<24\mathrm{h}$. The fresh dorsal skin samples were randomly divided into 19 groups according to Table 1.

A micrometer (Chengdu ChengLiang Tools Group Co., Ltd., Chengdu, China) was used to measure the thickness of skin sandwiched between two glass slides at four different directions. To reduce artificial measurement error, the measurements were done by the same person. The spectrophotometer (Lambda 950, PerkinElmer, Shelton, Connecticut) with an integrating sphere (150MM RSA ASSY, Labsphere, North Sutton), a dual-beam system, was applied to measure the reflectance and transmittance of the samples. The diameter of the integrating sphere is 150 mm with a circle reflectance-entrance ($r=12.70\mathrm{mm}$) and a transmittance-entrance. The transmittance-entrance combines two semicircles ($r=8.11\mathrm{mm}$) and one rectangle ($w\times h=16.22\mathrm{mm}\times 7.78\mathrm{mm}$).³³ The spectrophotometer’s software WinLab was used to calibrate. Each measurement needed to scan a baseline, and 100% transmittance was performed without samples, 100% reflectance was performed by a reflectance standard (Labsphere), 0% transmittance and reflectance were in the absence of light to reflect the dark noises. After the skin samples were immersed into each solution for 45 min, the thickness of the skin samples and the reflectance and transmittance spectra were measured again.

Based on the reflectance, transmittance, and thickness of each sample, the inverse adding-doubling method was applied to calculate the reduced scattering coefficient without introducing correction factors. Here, the skin RI and single scattering anisotropy were set as 1.4 and 0.8.^34–36 Further, the reduced scattering ratio (RSR) was calculated as follows:²⁸

2.5.

Optical Coherence Tomography for Assessing Optical Clearing Potential In Vivo

The imaging principle of OCT indicates that it has the capacity to image the tissue within 1 to 3 mm depth in vivo,^37–39 which makes it a powerful tool to assess the optical clearing process in tissues.^22,23,40 In this study, we used a commercial OCT system (OCP930SR, Thorlab Inc., Newton, New Jersey) working at $(930\pm 5)\mathrm{nm}$ with $(100\pm 5)\mathrm{nm}$ full-width-half-maximum, with an optical power of 2 mW, a maximum image depth of 1.6 mm, an axial resolution of $6.2\mu \mathrm{m}$ (in air), and a lateral resolution about $10\mu \mathrm{m}$. The saturated sucrose (78.9%) or saturated fructose (67.1%) induced skin optical clearing efficacy was assessed quantitatively by analyzing the improvements in imaging depth and signal intensity at the given depth.

Here, 12 rats were used for in vivo experiments. Initially, the prepared rats were placed on the experimental platform. The OCT B-scan images were recorded through the intact dorsal skin, tape stripped skin, and 6, 12, 18, 24, 30, and 36 min after topically applied saturated sucrose or saturated fructose on the skin, respectively. In order to reduce the mirror reflection caused by OCA, the OCA was quickly wiped away prior to acquiring the images and then was reapplied. It took about 30 s to wipe away and reapply the OCA, and 30 s acquire the images.

To facilitate analysis, the acquired images were flattened to enable averaging of the signal intensity profiles over the whole image so as to reduce the effects of the heterogenetic distribution of cutaneous absorbers and scatters. The flattening procedure was carried out with two steps: finding the $z$ position of the skin surface at each lateral $x$ position in the image and then translating the column of pixels at that $x$ to bring the skin surface to a common axial position.⁵ Figure 2 shows an example of flattening the skin by the image analysis. What is worth noting is that the flattening procedure may distort the tissue structure somewhat because it shifts pixels so that the skin surface is aligned at one constant $z$-axis position. However, such spatial distortion has minimal effect on the deduced light penetration since this study only paid attention the one-dimensional penetration of light into the skin.⁵

Fig. 2

(a) Original OCT image and (b) computer-flattened image. Bright color indicates the strong OCT signal intensity.

3.1.

Hydrogen Bonds Formed Between Collagen and Agents

Previous studies indicated that the higher-order structures of collagen could be stabilized by the hydration shell and water-mediated hydrogen bonds.^28,32 Nevertheless, the formation of hydrogen bonds as well as hydrogen bond bridges between collagen and chemical agents could disrupt the hydration shell and water-mediated hydrogen bonds,^29,32 which would destabilize the higher-order structure of collagen. Thus, the propensity of chemical agents to form hydrogen bonds could be used to predict the OCP of agents. Figure 3 shows the hydrogen bonds formed between the three agents and collagen mimetic peptide, it is expected that the more hydrogen bonds and hydrogen bond bridges an agent forms, the more effectively the hydration shell and water-mediated hydrogen bonds are disrupted.

Fig. 3

Schematic representation of hydrogen bond formation of (a) fructose, (b) sucrose, and (c) maltose with representative collagen mimetic peptides, respectively.

The average numbers of hydrogen bonds formed between each agent and two collagen mimetic peptides per ps are calculated and listed in Table 2. We find that sucrose forms the most hydrogen bonds among the three agents. That is, sucrose has the strongest ability to disrupt the hydration shell and water-mediated hydrogen bonds. Previous research has shown that the RSR increases as collagen solubility increases.⁴¹ Hence, it predicts that sucrose has the best skin OCP, while maltose is the second best and fructose is the worst, respectively, among the three agents.

Table 2

Average number of hydrogen bonds formed.

3.2.

Ex Vivo Experiment for Reduced Scattering Ratio

Further, to verify the prediction of MD simulation, the reduced scattering coefficient before and after skin samples were immersed into each agent was calculated, and then the RSR was deduced. The slope of the RSR was used to define the OCP of each agent.²⁸ Figure 4 shows the RSR of skin samples treated with each agent at different concentrations for 45 min, and it can be seen that the RSR increases with the concentration of an agent for the same OCA. Considering three concentration regions (A: 1-2 M; B: 1-3 M; C: 1-6 M) for each agent, different slopes were calculated separately. The results show that there is an excellent correlation ($R^2$) between the experimental data and the fitted data for each agent. The slope in concentration region A is in a descending order of sucrose, fructose, and maltose. Likewise, in concentration region B, the slope of sucrose is larger than that of fructose. According to Ref. 28, the slope is defined as the OCP of a chemical according to all the fitting data points, and the OCPs of sucrose, maltose, and fructose are 1.05, 0.65, and 0.48, respectively. Thus, sucrose and fructose have the best and worst skin OCP among the three agents, which is consistent with the MD simulation results.

Fig. 4

RSR of ex vivo rat skin as a function of agent concentration. (a–c) The slope of each agent is obtained by linear regression analysis of RSR data in different saturated concentration regions (inset). The data are shown as $\text{mean}\pm \text{standard deviation}$ (*$p<0.05$).

It should be noted that different chemicals have different saturation concentrations, and OCAs-induced optical clearing efficacy increases with its concentration. Furthermore, statistical analysis at common concentrations and the highest concentrations for fructose and sucrose were made, respectively. The results show that the RSR values of sucrose at the same concentration are obviously larger than those of other two agents ($p<0.05$). However, there is no significant difference between the RSR of sucrose (3 M) and fructose (6 M) ($p>0.05$).

3.3.

Agent-Induced Improvements in OCT Imaging Depth and Signal Intensity

Although MD simulation and ex vivo experimental results demonstrated that sucrose has a better OCP than that of fructose, the skin optical clearing efficacy caused by saturated sucrose is almost the same as that by saturated fructose. Moreover, both are on the basis of the direct interactions between the dermal collagen and the OCAs, which neglects the barrier of the stratum corneum that limits the penetration of OCAs into the dermis. In this work, the in vivo rat skin optical clearing performances of saturated fructose and saturated sucrose were evaluated using OCT.

Figure 5 shows the typical OCT results recorded through the intact rat skin in vivo, tape stripped skin, and 6, 12, 18, 36 min after the topical application of fructose (top row) or sucrose (bottom row) on the skin, respectively. By comparing the initial and the tape stripping, we find no significant changes, whereas after topical application of fructose or sucrose, the OCT imaging depth and signal intensity in the deeper region are enhanced gradually as optical clearing develops.

Fig. 5

Computer-flattened OCT images of rat dorsal skin treated with saturated fructose or saturated sucrose, respectively.

Figures 6(a)–6(b) show the typical normalized OCT signal intensity profiles after treatment of saturated fructose or sucrose for different times, respectively. It is found that OCAs’ treatment can evidently increase the signal intensity. To quantitatively evaluate OCAs-induced changes in imaging depth, the depth at which the signal intensity attenuates to $1/e$ of the normalized OCT intensity was calculated, respectively. The imaging depth can reach up to 450 or $550\mu \mathrm{m}$ after treatment of saturated fructose or saturated sucrose for 36 min, respectively, but the initial imaging depth is only $200\mu \mathrm{m}$. Figure 6(c) demonstrates the changes in image depth versus treatment of OCA, and the slope of linear regression analysis was also used. The results show that the slope of saturated sucrose (0.33) is higher than that of saturated fructose (0.23), respectively. Further statistical analysis using one-way analysis of variance (ANOVA) with SPSS (version 16.0, IBM) shows that there exists an extremely significant difference ($p<0.01$) between the two slopes.

Fig. 6

Typical temporal dynamical profiles of normalized OCT signal intensity after treatment with (a) saturated fructose and (b) saturated sucrose, respectively. (c) The relative changes in imaging depth ($\mathrm{\Delta }D/D_0$) at different treatment time with the two saturated solutions. Data are $\text{mean}\pm \text{standard deviation}$, N.S. refers to no significant difference, * means significant difference ($p<0.05$), and ** means extremely significant difference ($p<0.01$).

Furthermore, we calculated the changes in signal intensity at 300 and $400\mu \mathrm{m}$ before and after OCA treatment for different times [see Figs. 7(a)–7(b)]. It demonstrates that both OCAs can significantly enhance the signal intensity at 300 and $400\mu \mathrm{m}$. The linear regression slopes of sucrose and fructose are 0.49 and 0.44 for the changes in signal intensity at $300\mu \mathrm{m}$, and 0.38 and 0.27 for that at $400\mu \mathrm{m}$, respectively. Further statistical analysis using one-way ANOVA shows that there is an extremely significant difference for that at $400\mu \mathrm{m}$ ($p<0.01$) between the linear regression slopes of fructose and sucrose, but no significant difference for that at $300\mu \mathrm{m}$ ($p>0.05$).

Fig. 7

Relative changes in signal intensity ($\mathrm{\Delta }I/I_0$) at the depth of (a) $300\mu \mathrm{m}$ and (b) $400\mu \mathrm{m}$ through the in vivo rat skin treated by saturated fructose and saturated sucrose, respectively. Data are $\text{mean}\pm \text{standard deviation}$, N.S. refers to no significant difference, * means significant difference ($p<0.05$), and ** means extremely significant difference ($p<0.01$).