Optical clearing agent increases effectiveness of photodynamic therapy in a mouse model of cutaneous melanoma: an analysis by Raman microspectroscopy

2.1. Cell line and animal model

B16F10 murine melanoma cells were purchased from the American Type Cell Collection (ATCC) and were cultured in Dulbecco's Modified Eagle Medium (DMEM) with 10% fetal bovine serum and 1% penicillin-streptomycin maintained at 37 °C in an incubator with 5% CO2. For the cutaneous animal model, one million cells were intradermally injected in the flanks of 6-week old BALB/c nude mice. All animal procedures were approved by the Ethics Committee for Animal Use of the São Carlos Institute of Physics (IFSC, from the University of São Paulo, São Carlos, SP-Brazil), document number 5500080618. The animals were examined daily for lethargy, abnormal posture, weight-loss and tumor volume, being euthanized if necessary.

2.2. Optical clearing agents

Our group has previously investigated [39] the optical clearing effect on skin and melanoma of several OCA mixtures such as glycerol at different concentrations and mixing solutions of PEG-400 and 1,2-propanediol at the investigated animal model. The mixture between PEG-400 and 1,2-propanediol in the ratio of 19:1, respectively, presented the best results and was chosen for use in all the present PDT experiments. They were acquired from the company Sigma-Aldrich.

2.3. Photodynamic therapy

The animals were anesthetized with inhaled isoflurane 5% for induction and 2% for maintenance. When the pigmented melanoma lesions achieved 1 mm in diamenter, chlorin e6 (Photodithazine, VETAGRAND Company, Russia) was injected intraperitoneally at a dose of 9.0 mg/kg. The drug-light interval was 2 h 30 min, as determined in a preliminary study of Photodithazine concentration at the melanoma lesion. OCA were applied topically (approximately 300 µL) 15 minutes before the irradiation with a gentle massage to improve its penetration through the skin. PDT irradiation was performed using a 660 nm diode laser, delivering a total fluence of 100 J/cm² in 17 minutes.

During PDT irradiation, an aluminum foil mask was used to avoid illumination of adjacent normal tissues. Four experimental groups (N=3) were assessed: i) control (no OCA and no PDT), ii) tumor with OCA only, iii) tumor with PDT only and iv) tumor with PDT + OCA. Using a USB2000 spectrophotometer (Ocean Optics, USA), the presence of PDZ and its relative concentration in the tumor was monitored by laser-induced fluorescence spectroscopy before and after the treatment with excitation at 532 nm (see Fig. S1). Mice cohorts were sacrificed immediately and at 7days-post-PDT and the tumor was removed for ex vivo analysis. The tumors were embedded in Tissue Tek Optimal Cutting-Temperature media (Sakura Finetek, Torrance, Canada) and frozen in liquid nitrogen for cryosection. Longitudinal cuts were prepared for analysis of the skin and tumor overall profile (top to bottom).

2.4. Raman microspectroscopy

Raman spectra of the 60 µm-thick cryosections were measured using a confocal Raman microscope (WITec Alpha 300 RAS, Ulm, Germany) with excitation wavelength at 785 nm and the detection range at 100–3200 cm⁻¹. The approximate diameter of the laser beam at the focus point is 0.65 µm. The spectra were collected with a 20x objective lens (Zeiss, Jena, Germany) with an integration time of 30 s and 3 accumulations for each spectrum (see Fig. S2). The duration of the Raman measurement for each sample was around 2 h, and since the tumor samples were cut through cryosectioning protocol, minimal biochemical changes are expected to occur during this analysis period. Three spectra were acquired for each depth within the tumor; measurements were taken between 25 and 2000µm, at 100 µm steps (Fig. 1). The data were processed using the Project-FOUR-4.1 (WITec, Ulm, Germany) and Origin (OriginLab Corporation, Northampton, Massachusetts, USA) softwares.

3.1. Raman analysis of tumors with and without OCAs

There is still no complete knowledge of the action of skin-clearing agents in vivo nor in melanoma. For this reason, it was initially investigated if it was possible to detect OCA-induced modifications through Raman signal, and if so, to track the depth achieved by OCA in the tissue. Figure 2 shows the Raman spectrum of the used clearing agent mixture of PEG-400 and 1,2-propanediol, which is in agreement with the ones reported in the literature [40].

Raman spectra from melanoma tumor slides with and without OCA treatment were recorded. Three main regions of the tumor were chosen for Raman analysis: 25-125 µm considering the most superficial region of the melanoma; 225-325 µm for the central region of the tumor and 525-725 µm for the deepest and internal region of the tumor. This choice was made in order to verify to what depth the OCA could permeate through the tumor, as well as to what depth the agents would cause biochemical changes to the melanoma.

Examples of the spectra obtained at different depths of the melanoma with and without OCA are presented in Fig. 3. The Raman peaks of OCA are in the same region as the spectrum of the tumor tissue with some peaks overlapping the same wavenumbers. Therefore, it was not possible to directly observe the OCA Raman peaks at the OCA-treated melanoma, neither a clear modification of the melanoma Raman spectrum after the OCA treatment (Fig. 3).

In general, the Raman spectra from tissues show noisy patterns and bands with low signal amplitude, instead of sharp clearly defined. The most characteristic peaks of both the control and the tumor with OCA for the most superficial region (25-125 µm) are presented in Table 1. In the spectrum of normal skin, the most characteristic peaks referred mainly to actin, collagen, and elastin are 857, 939, 1004, 1248, 1271, and 1452 cm⁻¹ [32]. There are also related peaks in the region of 400 to 1000 cm⁻¹, but with low amplitudes. The 1270-1300 cm⁻¹ band and the peak 1452 cm⁻¹ can be attributed to the saturated fatty acid of the epidermis ceramides and the cell membrane phospholipids [32]. Other peaks of unsaturated lipids appear as 1092,1271,1301,1452 and 1658 cm⁻¹ [32]. Comparing the spectra of normal skin and melanoma, it is possible to infer that some peaks in the region of 800-1000 cm⁻¹ and 1200- 1400 cm⁻¹ are slightly shifted, with different intensities [32,41]. The melanoma peaks are more similar to the normal skin in the most superficial region. Stremersch et al. [42] obtained Raman spectra using surface enhanced Raman spectroscopy (SERS) of B16-F10 melanoma cells, with gold nanoparticles. Brauchle et al. [43] used Raman Spectroscopy to distinguish different types of melanoma cells from melanocytes. Both in the Stremersch and Brauchle studies, the spectra of melanoma cells showed similar peaks better defined in 780, 980, 1000, 1050, and 1236 cm⁻¹ in the studies with nanoparticles [42,43]. The tumor Raman spectra in our experiments, on the other hand, present the peaks previously mentioned, but with the presence of noise, since no enhancement was used, and because there is a great variety of molecules present in the tumor that contribute to the Raman signal. The resulting signal collected is composed of all biomolecules present in the tissue (cells and extracellular matrix), in addition to the light-tissue interactions being much more complex when compared than those occurring in cells or solutions.

The highest change after OCA addition occurred in the 850-1130 cm⁻¹ region, where the Raman peaks of the C-C backbone of proteins and lipids are located [44], indicating that the vibrations of the protein and lipid skeleton and the conformation of proteins (collagen) are among the most affected by the optical clearing.

In the region of the tumor between 225 and 325 µm (Fig. 2), several Raman peaks also changed their amplitudes in the presence of OCA. No clear distinction was observed when comparing the intensities of the same peaks of tumor with and without OCA. The differences in intensities between the peaks occur in proteins, lipids, and nucleic acids (according to Table 1 references) and may be due to conformational changes in collagen, with the number of alpha-helices being changed and the proportion of beta-sheet and random coil conformations as well [48]. The OCA appears to cause dissociation of collagen fibrils into microfibrils, thus changing the packaging of collagen in the tumor [35,36,49]. Collagen fibers are complex and highly scattering structures, and the main bonding force that unites triple helices is hydrogen bonds [35,49]. OCA, which have several hydroxyl groups, destabilize the ordered structure of the fibers, dissociating them, but this can be easily reversible because the interactions are non-covalent bonds between the fibers [35]. Yeh et al. observed this fact in an in vitro experiment immersing a tissue in glycerol and later in phosphate-buffered saline (PBS) [50]. R. Millon et al. also studied the effect of some OCA like PBS and DMSO in vivo on a squamous epithelium [34]. Because OCA dehydrate the tissue, there is an increase in the osmolarity of the interstitial fluid, causing the water to escape from the cells and collagen fibers. This results in a decrease of tissue thickness, leading to a denser tissue with more orderly architecture [33,35,36,49]. Sdobnov et al. [51] performed a quantitative analysis of the ratio of the collagen Raman peaks intensity after the application of two types of OCA, glycerol, and iohexol (Omnipaque). The authors observed that the intensity of the peaks increased with the application of both agents to the pig skin dermis, which may result from a more compact structure of collagen fibers [51]. Nguyen et al. used Raman spectroscopy to verify the behavior of collagen with water, differentiating the age of the skin [52]. It was verified that there is a variation in the Raman signal, mainly due to the difference in intensity of some peaks related to the hydration marking of the biological sample [52,53]. Therefore, it is expected that the application of OCA in tumors will generate a change in the Raman signal of collagen.

Finally, in the region between 525 µm and 725 µm (Fig. 3), other peaks, that were not previously seen nor identified presented greater intensity. They correspond to proline, tyrosine and valine amino acids, amide III, proteins, lipids, and phosphate from nucleic acids [43–47].

Deeper than 725 µm, the spectra of the tumor with OCA showed little difference between the tumors with and without OCA. Since the difference between spectra is insignificant, this may indicate that the clearing agents did not penetrate above this depth and had little or no effect in the tumor and, consequently, on the light penetration. Yanina et al. [54] investigated different concentrations of glycerol solutions as OCA in an ex vivo model of pig skin. Maximum penetration, evaluated by confocal Raman microspectroscopy, down to the depth of 80 µm was achieved with a 60% glycerol and 40% water for 45 min of application. Sdobnov and collaborators [55] reported the increased depth Raman spectral changes of the hydrogen bound water, at the ex vivo porcine skin, with increased OCA treatment time, both for glycerol and Omnipaque clearing solutions. Our results are in agreement with these reported results on the decrease of light attenuation effect of the OCAs in skin, even though a direct comparison of the at-depth response is not possible, since distinct OCA compositions were used; and ex vivo and in vivo skin presents different OCA diffusion patterns. Lin et al. [56] compared the optical clearing effect of 3 different agents (glycerol, iohexol and polyethylene glycol) and serial concentrations at in vivo human skin. They report that the three investigated agents enhanced the optical clearing effect on skin, and that the dehydration mechanism, which occurs at the initial treatment period between 10-20 min, reduces the OC. In our in vivo cutaneous melanoma model, we have indirectly observed the enhanced optical clearing through the improved PDT response. This result is in agreement with our previous study where we investigated the use of OCA using optical coherence tomography at the same cutaneous melanoma model. Using the OCA previous to the OCT resulted in improved imaging up to 750 µm [38].

To check if it was possible to determine the maximum penetration depth of OCA in melanoma, a mathematical analysis of the spectra taken from melanoma slides with and without previous OCA treatment was performed. First, normalization was performed at the wavenumber of the highest amplitude of all spectra. Melanoma spectrum obtained from the most superficial region of the tumor (25-125 µm) was summed to the OCA spectrum. Each final spectrum was then compared with the spectrum taken at melanoma treated with the OCA. It was analyzed which peaks increased their intensities and whether it was possible to verify the presence of OCA by Raman spectroscopy. It was observed that some peaks, such as 850, 1082, and 1240 cm⁻¹ showed higher intensities, probably due to OCA presence. The main region of increased amplitude was over 1003 cm⁻¹, in which the peaks are more pronounced than in the melanoma without OCA. Even though these changes could be observed at the comparison between the processed spectra and the spectra from melanoma with OCA, they were not evident enough to directly detect the presence of OCA, mainly because melanoma itself already shows Raman peaks at the same region.

3.2. Raman analysis of tumors with and without OCAs right after PDT

Representative spectra (Fig. 4) taken at the melanoma lesion at different depths after PDT treatment with and without OCA.

At the more superficial region (25-125 µm), there was no difference in the spectra of tumors treated with PDT and with and without OCA topical application (Fig. 4). The spectra taken at this depth were similar within the same treatment group and also inter-groups, highly noisy and with a relevant decrease in the Raman signal when compared to the non-treated melanoma. This is probably a result of PDT destruction of the biomolecules and tumor structure in this region. Similar results between the groups can be explained by the fact that at the superficial layers of the tumor, no relevant light attenuation is observed, resulting in a homogenous PDT effect within this tumor volume and no changes in the response are observed with the OCA combination.

When evaluating the region between 225 µm and 325 µm (Fig. 4), the spectra from tumors treated only with PDT or PDT + OCA showed mostly changes in the intensity profile, no new Raman peaks nor relevant displacements were present. It is possible to observe that all the peaks showed their intensity decreased when the OCA was applied before PDT. The spectrum from the PDT group without OCA is more similar to the ones measured from non-treated tumors. The PDT + OCA spectrum is similar to the spectra collected at the most superficial part of the tumor. Based on these results, there is an indication that the OCA use has minimized the effect of light attenuation at the melanoma depth of 225-325 µm, resulting in a more effective PDT response similar to the superficial tumor region.

The spectra from the region between 525 µm and 725 µm (Fig. 4) of the PDT group, showed a dual behavior where there were spectra similar to the spectrum of a non-treated tumor and a tumor showing PDT response. On the other hand, the spectra from the PDT + OCA group at this depth were similar to the ones measured at the superficial region, i.e. with the PDT response. The variation in the shape of the spectra is due to the tumor heterogeneity, but also the heterogeneity of the photodynamic response at greater depths. Heterogeneous PDT response within the tumor means that non-treated regions with viable melanoma may result in tumor regrowth.

More similar spectra after PDT + OCA treatment at the studied depths down to 750 µm are indicative that a more homogeneous and effective PDT response was achieved. We hypothesize that the OCA, by improving the refractive index matching resulted and making tumors more optically homogeneous, lead to a more effective photodynamic response. This is a relevant result since the heterogeneity of tumor irradiation is one of the most important reasons that result in treatment failed, since islands of tumor cells remain viable. Based on these results, we hypothesize that OCA would also be beneficial to other pigmented lesions, as pigmented basal cell carcinoma, and also to thicker tumors, the last case, if other delivery methods are developed to increase OCA depth distribution.

3.3. Raman analysis of tumors with and without OCAs 7 days after PDT

In order to evaluate if the addition of the OCA to the subdose PDT treatment would result in a modified tumor regrowth, Raman spectra were also recorded at 7 days post-PDT. Figure 5 shows Raman spectra recorded from these regrown melanomas. No relevant changes were observed in the spectra from different depths and treated groups when compared to the non-treated tumors. In this sense, when a partial response is achieved after OCA are used in combination with PDT, which was planned for this study, the resulted regrown tumor did not show biochemical changes evident by Raman spectroscopy. This behavior should be also investigated by other methods such as immunohistochemistry and longer follow up period, but Raman spectroscopy provides a preliminary indicative.