In Vivo Analysis of Laser Preconditioning in Incisional Wound Healing of Wild-Type and HSP70 Knockout Mice With Raman Spectroscopy

Abstract

Background and Objective

Laser preconditioning augments incisional wound healing by reducing scar tissue and increasing maximum tensile load of the healed wound [.Wilmink et al. (2009) J Invest Dermatol 129(1): 205–216]. Recent studies have optimized treatments or confirmed results using HSP70 as a biomarker. Under the hypothesis that HSP70 plays a role in reported results and to better understand the downstream effects of laser preconditioning, this study utilized a probe-based Raman spectroscopy (RS) system to achieve an in vivo, spatiotemporal biochemical profile of murine skin incisional wounds as a function of laser preconditioning and the presence of HSP70.

Study Design/Materials and Methods

A total of 19 wild-type (WT) and HSP70 knockout (HSP70−/−) C57BL/ 6 mice underwent normal and laser preconditioned incisional wounds. Laser thermal preconditioning was conducted via previously established protocol (λ = 1.85 µm, H₀ =7.64 mJ/cm² per pulse, spot diameter = 5 mm, Rep. rate = 50 Hz, τ_p = 2 milliseconds, exposure time = 10 minutes) with an Aculight Renoir diode laser, with tissue temperature confirmed by real-time infrared camera measurements. Wound-healing progression was quantified by daily collection of a spatial distribution of Raman spectra. The results of RS findings were then qualified using standard histology and polarization microscopy.

Results

Raman spectra yielded significant differences (t-test; α =0.05) in several known biochemical peaks between WT and HSP70 (−/−) mice on wounds and in adjacent tissue early in the wound-healing process. Analysis of peak ratios implied (i) an increase in protein configuration in and surrounding the wound in WT mice, and (ii) an increased cellular trend in WT mice that was prolonged due to laser treatment. Polarization microscopy confirmed that laser treated WT mice showed increased heterogeneity in collagen orientation.

Conclusions

The data herein supports the theory that HSP70 is involved in normal skin protein configuration and the cellularity of early wound healing. Laser preconditioning extends cellular trends in the presence of HSP70. Despite study limitations, RS provided a non-invasive method for quantifying temporal trends in altered wound healing, narrowing candidates and design for future studies with clinically applicable instrumentation.

INTRODUCTION

Wound repair is a complex coordinated sequence of overlapping biochemical and cellular events, which under normal conditions, follows a specific and well-defined time course [1]. Various disorders create conditions that impair the normal sequence of wound repair [2] causing many patients to eventually develop chronic wounds [3,4]. Efforts to reduce wound morbidity and mortality have sparked investigation into pre-treating cells or tissues with an initial mild thermal stimulus (often referred to as preconditioning), eliciting a stress response that can protect the tissue from subsequent lethal stresses [5–7]. Recent results demonstrate that laser preconditioning tissue 6–12 hours prior to cutaneous injury (scalpel incision and laser ablation) significantly accelerates wound healing [8]. It was shown that (i) HSP70 upregulation is required for a efficient preconditioning response [9]; and (ii) HSP70 is a useful biomarker for therapeutic efficacy of improved wound healing after tissue preconditioning.

However, in order to translate the preconditioning protocol to a clinically relevant thermal modulation protocol for the treatment of chronic wounds, development of a non-invasive, in vivo method capable of analyzing tissue biochemistry is critical for optimization of laser parameters. Raman spectroscopy (RS) utilizes the vibrational level energy change of laser light as it scatters off chemical bonds to provide a molecular-specific biochemical fingerprint. Conventional RS for in vivo applications utilizes a fiber optic probe design with a fast acquisition protocol and automated pre-processing. Such a system represents the potential for real-time guidance, and subsequent decreases in cost and patient recovery time [10,11]. In our studies, spectral content may allow for optimization of laser treatment effects on the healing wound, while providing tissue level information about the wound itself.

Despite the possibility of a natural extension of ongoing studies in skin [11,12], remarkably few research groups have pursued the investigation of wound healing or tissue repair using optical spectroscopies. Zhang et al. [13] utilized infrared (IR) spectroscopy and confocal RS to visualize the keratin-rich reepithelialization front and the spatial distribution of elastin and collagen within the wound bed, respectively. Chan et al. analyzed an ex vivo human skin wound culture model with IR and confocal Raman spectroscopies over several days. Subsequent deconvolution of confocal mappings using factors with distinct elastin and collagen features allowed for spatial-temporal distribution of major structural proteins [14]. Crane et al. [15] demonstrated qualitative differences between healing and chronic wound spectra from wound biopsies of soldiers. Alimova et al. [16] analyzed the changes in spectral content as a function of incisional wound closure via traditional suture or experimental laser tissue welding from two different lasers at discrete time points in the healing process of four guinea pigs.

The work reported here differs from other reports of spectroscopic wound healing in several ways. The RS acquisition protocol for this laboratory study utilizes irradiances and collection times currently approved for clinical study, allowing for possible future translation and treatment optimization. Secondly, to the best of our knowledge, this work represents the first report of Raman sensitivity to spectral differences in the healing wound and neighboring tissue as a function of both genetic expression as well as treatment response.

This study seeks to evaluate a role for HSP70 in untreated and laser preconditioned wound healing by detecting in vivo differences in time course biochemistry via RS. Analysis of peak ratios from the wound-healing profiles of murine incisional wounds highlighted biochemical differences due to genetic expression of HSP70 and treatment response. The results of findings via RS were qualified using standard and polarization microscopy. Ultimately, RS shows promise towards a clinically relevant, non-invasive, in vivo, spatio-temporal biochemical analysis of cutaneous wound healing.

METHODS

Mouse Model and Preparation

A pre-conditioning and wounding protocol as described below was followed using C57BL/6 mice under the approval of and conducted in accordance with procedures established by the Vanderbilt Institutional Animal Care and Use Committee (IACUC). HSP70 (−/−) mice were the model described by Hunt et al. [17], acquired through the Mutant Mouse Regional Resource Centers (B6;129S7-Hspa1a/Hspa1b^tm1Dix/Mmcd). Nineteen adult female mice, 10 wild-type (WT) and nine HSP70 knockout [HSP70(−/−)], were anesthetized with isoflurane. While under anesthesia, dorsal fur was clipped and remaining fur was removed using depilating cream. The affected area was repeatedly rinsed and wiped down with alcohol to minimize any undesired reaction. Mice were returned to animal care and observed for 36–48 hours, during which observation of inflammation, pigmentation in the area to be treated, superficial wounds, or chemical reaction, resulted in exclusion of those animals from the study.

Upon routine inspection no <24 hours prior to treatment, the dorsal skin of anesthetized mice was again cleaned and disinfected, and target areas were surrounded with a series of manually created tattoos to ensure consistency of location throughout preconditioning, surgery, and wound-healing analysis. Tattoos were placed just below the epidermis using 28-gauge needles with India ink at no <1 cm away from the target location. To prevent bias due to anatomical variation, contralateral target locations were consistently chosen 1–1.5 cm caudal to the base of the skull and centered 0.8–1 cm lateral from the spine. Any improper tattooing or spreading of ink was considered grounds for study exclusion.

Laser Preconditioning and Surgery

Twelve hours prior to surgery, anesthetized mice underwent laser preconditioning on one of the two (randomly selected) target dorsal locations. Preconditioning protocol was conducted with an Aculight Renoir diode laser (Renoir; Lockheed Martin Aculight, Bothel, WA) coupled into a 600 µm diameter, multimode silica fiber according to Wilmink et al. [8] (λ = 1.85 µm, H₀ = 7.64 mJ/cm², spot diameter = 5 mm, Rep. rate = 50 Hz, τ_p = 2 milliseconds, exposure time = 10 minutes). During laser exposure, real-time tissue temperature traces were measured via infrared camera to within ± 0.18°C (A20 series with ThermaCAM Researcher 2.8, FLIR Systems, Portland, OR). In assuring consistency with results from Wilmink et al. [8], generated temperatures were verified to be approximately 43–44.58°C.

Twelve hours after preconditioning, mice were anesthetized with isoflurane, and the surgical area was sterilized with povidone-iodine solution and alcohol. Using tattoos as guides, two contralateral 1 cm full-thickness linear incisions were made with a #11 scalpel blade. Immediately after the incision, each end of the wound was closed using a 7.5-mm Michelle clip, ensuring proper opposition of wound edges. Clips were subsequently removed on post-operative day 5 (POD5).

Raman Spectroscopy: Acquisition and Processing

The portable Raman system consists of a mode locked diode laser (λ = 785 nm; Innovative Photonic Solutions, Mammoth Junction, NJ) coupled to a custom fiber optic probe (Emvision, Loxahatchee, FL). This probe consists of a single, central 400-µm diameter excitation fiber surrounded by seven equally spaced 300-µm diameter collection fibers with in-line filtering to reject 785 nm. Probe power output was confirmed at ~80 mW for each scan. Collection fibers were coupled to a f/1.8 holographic imaging spectrograph (Kaiser Optical Systems, Inc., Ann Arbor, MI), subsequently dispersing light onto a backilluminated, deep depletion, thermo-electrically cooled CCD (Model# 920- BRD, Andor Technologies Plc, Belfast, UK). Each Raman scan consisted of three accumulations of 3 seconds each to ensure maximum signal without saturation.

Spectral calibration and processing were performed according to Robichaux-Viehoever et al. [18]. Subsequent to preprocessing algorithms, spectra were then normalized to mean spectral intensity to account for overall intensity fluctuations. Normalized spectra were analyzed as a function of time and spatial location with regards to the intensity of the most prominent peaks and the relative intensity of their ratios.

To eliminate the influence of anatomical variations in skin, eleven Raman spectra (Fig. 1) were acquired from consistent sites on wounded and adjacent intact skin prior to preconditioning, after surgery, and daily thereafter until mice were culled for histology at days 4, 7, and 10. For peak ratio analyses, samples from intact skin immediately adjacent from the wound were separated from medial wound samples and lateral wound samples. At each time point, wounds were documented using digital photography.

Fig. 1.

Diagram of mouse model dorsal wound and Raman spectra collection location indicates grouping by anatomical locations and subsequent similar biochemical trends: Intact skin adjacent to wounds (dots); lateral wound samples (checkers); medial wound samples (solid black). Samples along the spine (white) were used to help establish stability of skin spectra as a function of spatial location and hair regrowth but not directly used in analysis.

Histology

After euthanasia, the entire dorsum of the mouse was excised and divided into 5-mm wide cross-sections of the wound and adjacent intact skin. Following 24-hour formalin fixation, histological cross-sections of the wounds were stained with hematoxylin and eosin (H&E). Slides were analyzed using a microscope with a polarizing filter. Under polarized light, newer, immature collagen displays less birefringence than its mature organized counterpart [19]. As a qualitative confirmation of Raman spectral analysis, images of the histology slides under standard and polarized light were examined for inflammatory cell abundance, epidermal thickness, the intensity of collagen birefringence, and collagen fiber orientation.

RESULTS

Figure 2 illustrates a representative time course (increasing in vertical axis) of Raman spectra of untreated WT skin healing from the lateral spot (Fig. 1, red) wound, starting prior to surgery, and ending on post-operative day 7 (POD7). The greatest intensity differences seen during the wound-healing time course (Fig. 2 gray bars) also correspond to the most prominent pre-operative (Pre-Op, bold baseline) peak intensities: 1,265, 1,304, 1,440, and 1,657 cm⁻¹ (significance in Table 1). The most notable changes in peak intensities relative to Pre-Op skin occurred between POD1 and POD3. The general shape of the spectrum returns to the baseline preoperative shape by POD 7. Interestingly, several peak intensity changes that occur between POD1 and POD3 still remain, or even increase by POD7. These results suggest that Raman spectra are sensitive to multiple distinct trends within the course of normal incisional wound healing.

Fig. 2.

A representative time course of Raman spectra from a single untreated wild-type (WT) lateral wound throughout the wound-healing process indicates changes within 2 days of operation and gradual peak intensity recovery by POD7. Increasing time points (pre-operative, POD 0–7) correlate to increasing y-axis value. Prominent peaks (gray bars) indicate changes throughout the time course at 1,265 cm⁻¹, 1,304 cm⁻¹, 1,440 cm⁻¹, and 1,657 cm⁻¹.

TABLE 1.

Chemical Bond Associations for Prominent Raman Peaks in Healing Wounds^*

Raman shift	Bond typea	Biochemical Assignmenta
1,265 cm−1	Amide III	Structural proteins
1,304 cm−1b	Amide III	Histones and nucleic acids
	CH2 twisting	Phospholipid membranes
1,440 cm−1	CH2 stretch	Protein side chains
1,657 cm−1	Amide I	Amino acid linkages

^*Only peaks with established bond associations are included. Context of other peaks requires investigation.

^aAssignments are drawn from bands reported for cell and tissue changes. See Refs. [–] for details.

^bBands exist at 1,304 cm⁻¹.

Figure 3 provides statistical support for the trends represented in Figure 2. Panels A and B show mean spectra (n = 10) from untreated wounds in WT mice at POD2 and POD7 relative to Pre-Op spectra. Bottom panels indicate the probability of significant difference between the mean spectra using a paired Student’s t-test for each wavenumber. Significance of the t-test is plotted as the probability of rejecting the null hypothesis (mathematically expressed as ‘‘1-P’’, where P is the probability of accepting the null hypothesis), such that significance increases as the value of ‘‘1-P’’ approaches 1. This test evaluates the null hypothesis that the intensity of a given wavenumber of Raman shift for a particular post-operative day is the same as that of its matching pre-operative spectrum. Peak intensities at 964, 1,265, and 1,304 cm⁻¹ exhibit statistically significant differences (P < 0.05) from Pre-Op spectra on POD2 but not on POD7. On the contrary, peak intensities at 1,118, 1,440, 1,657, and 1,748 cm⁻¹ do not exhibit statistically significant differences from Pre-Op spectra until POD7. These findings support Raman sensitivity to multiple trends seen in Figure 2.

Fig. 3.

Comparison of mean spectra on POD2 (A: blue dotted) and POD7 (B: blue dotted) relative to the mean Pre-Op spectrum (A,B: red solid). Paired t-tests (C,D: shown below their respective spectra as the probability of rejecting the null hypothesis) illustrate recovery or persistence of peak intensity differences. Low wavenumber peaks including 726 cm⁻¹, 964 cm⁻¹, 1,083 cm⁻¹, 1,118 cm⁻¹, and the doublet between 1,265 and 1,304 cm⁻¹ illustrate peak intensity change at POD2 (significance in C) followed by recovery of pre-operative peak intensity by POD7 (lack of significance in D). Peak intensity differences seen in 1,440 cm⁻¹, and 1,657 cm⁻¹, on POD2 persist and even increase by POD7.

In untreated WT wounds, POD2 represents the greatest spectral departure from Pre-Op normal tissue, prior to significant scab dehydration and sloughing (possible sources of variance). To examine how downstream changes are affected by HSP70 genetic expression and laser preconditioning, average spectra from the wound and adjacent skin were constructed for POD2. Similar in layout to Figure 3, Figure 4 illustrates the downstream effects of HSP70 within treatment groups, in which a point-wise two sample t-test evaluates the null hypothesis that the intensity of a given Raman shift within a WT mouse wound is the same as that of an HSP70 (−/−) wound. For both the treated and untreated groups, there is a highly significant difference (P < 0.001) between the Raman spectrum on POD2 of a WT and HSP70 (−/−) mouse for all prominent peaks.

Fig. 4.

Comparison of average WT (n = 10) and HSP70 (−/−) (n = 9) mouse wound Raman spectra on POD2 (A: preconditioned wounds; B: untreated wounds). Two-sample t-tests (C,D: shown below their respective spectra) indicate significant differences in normalized intensity at peaks of biological interest. Similar spectra results are found for preconditioned wounds (A) as those seen for untreated wounds (B), indicating native difference in tissue biochemistry due to HSP70 knockout.

Figure 5 illustrates the relative effect of preconditioning upon mean POD2 Raman spectra within genetic mouse type. Laser preconditioning creates numerous differences in the Raman spectra of WT mouse wounds, but not in those of the HSP70 (−/−) mice. In WT wounds, significant differences were seen between the intensity of untreated and preconditioned peaks at 723 and 1,083 cm⁻¹ , Marginally significant differences (significant at P < 0.10) in the peaks of wounds in WT mice were seen at 1,118, 1,265, 1,304, 1,440, and 1,657 with statistical significance (P < 0.05) observed in the shoulders of these peaks. Within HSP70 (−/−) mice, statistical significant differences in spectra are observed only in the shoulders of peaks at 1,083 cm⁻¹ (at 1,052 cm⁻¹ ) and 1,440 cm⁻¹ (at 1,489 cm⁻¹ ). Ultimately, preconditioning induced marginally significant differences in the intensity of prominent Raman peaks of the normal wound-healing time course (Fig. 2), but only in WT wounds.

Fig. 5.

Comparison of average preconditioned and untreated wound Raman spectra on POD2 [A: WT; B: HSP70 (−/−)]. Two-sample t-tests indicate few significant differences in normalized intensity at peaks of biological interest for WT samples (C), and no significant differences at peak locations for HSP70 (−/−) samples (D), suggesting that HSP70 in WT wounds is implicated downstream of preconditioning.

While the analyses shown in Figures 2–5 were limited to absolute peak intensity, peak ratios examine relative intensity differences, allowing for a further analysis of specific spectral changes. For these analyses, Raman spectra were grouped by sample location to prevent the confounding effects of natural anatomical variation in skin (Fig. 1). Examined ratios were limited to prominent peaks whose assignment has precedence in related contexts. Figure 6 illustrates the ratio of the normalized peaks at 1,440–1,657 cm⁻¹ with error bars (± 1 SD) as a function of time for lateral wound samples (Fig. 1, checkers). Within the context of wound healing, this ratio likely suggests the degree and quality of protein deposition and organization. WT and HSP70 (−/−) wounds exhibit consistent statistically significant differences (two sample t-test; P <0.01) through POD7 following a trend in both laser preconditioned (B) and untreated (A) wounds: ratio decrease until POD3 and a subsequent recovery and increase by POD7. Similar significant trends (not shown) exist for medial wound spectra (Fig. 1, solid black) and neighboring intact tissue (Fig. 1 dots) suggesting a persistent, genotype-specific tissue difference.

Fig. 6.

Paired comparisons of time course (POD) of average peak ratio intensity of 1,440:1,657 cm⁻¹ in lateral wound samples indicates significant differences between wounds in WT (A) and HSP70 (−/−) mice (B) for the first 6 days. No significant differences are seen in either WT or HSP70 (−/−) as a function of preconditioning. Error bars represent 1 SD (separation indicates P < 0.05). Trends suggest that HSP70 knockout correlates to constant tissue level biochemical differences in skin that persist throughout wound healing.

Figure 7 shows the normalized peak ratio of 1,304– 1,265 cm⁻¹ over time, indicating a significant difference (two sample t-test; P <0.05) between the early wound healing processes in WT mice and those lacking HSP70. Due to peak locations and subsequent biochemical assignments within this wound-healing context, this ratio suggests a measure of either cellularity or cellular activity. In untreated wounds genotype differences are statistically significant (P < 0.05) from Pre-Op until POD1; however, laser preconditioning of WT mouse wounds amplified this difference on both POD1 and POD2 (compare A to B). Moreover, laser preconditioned wounds exhibit a significant difference in peak ratio between WT and HSP70 (−/−) mouse wounds from Pre-Op spectra up to and including POD5 (B).

Fig. 7.

Paired comparisons of time course (POD) of average peak ratio intensity of 1,304:1,265 cm⁻¹ in lateral wound samples indicates significant differences between WT (A) and HSP70 (−/−) mice (B) prior to and following surgery. Preconditioning prolongs significant differences between WT and HSP70 (−/−) beyond POD1 (A vs. B). No significant differences are seen within either WT or HSP70 (−/−) as a function of preconditioning. Error bars represent 1 standard deviation (separation indicates P < 0.05). Trends suggest preconditioning prolongs the increase in ratio associated with HSP70.

Analysis of H&E stained wound cross-sections established the wound healing time course. Cross-sections were examined for inflammatory cell abundance, epidermal thickness, and wound area. Two blinded, independent pathology evaluations concurred that (i) cellular infiltration and wound closure completed prior to POD4, (ii) that the wounds were largely reepithelialized by POD7, and (iii) that most wounds were remodeling on POD10. There was no significant difference in epidermal thickness in adjacent normal skin of either WT or HSP70 (−/−) mice. Acute cellular inflammatory infiltrates were largely gone in all groups by POD4, when the wound environment was dominated by macrophages. Histological samples from POD7 indicated complete wound reepithelialization with little residual fibroplasia.

Qualitative histological analysis using polarization microscopy on POD4 and POD7 was utilized to confirm results from RS. To prevent bias from natural variation between animals, blinded analysis was conducted comparing preconditioned and untreated samples from the same mouse for collagen fiber birefringence intensity, location, and orientation. Reported trends were the result of integrated analysis of serial histological samples to exclude the influence of hair follicles on collagen organization. Figures 8 and 9 are paired, representative images that illustrate cumulative trends in qualitative analysis. Figure 8 illustrates the trends seen on POD4. Healing wounds in WT mice showed less collagen in the wound bed than HSP70 (−/−) counterparts; however, the organizational patterns of collagen were distinctly more heterogeneous in wounds of WT mice compared to HSP70 (−/−) mice. Both WT and HSP70 (−/−) preconditioned wounds show slightly more total collagen within the wound bed than their untreated counterparts. Figure 9 illustrates the effects of HSP70 deletion and laser preconditioning treatment on POD7. Preconditioned wounds of both WT and HSP70 (−/−) mice indicated more heterogeneity of collagen adjoining the repairing wound bed, relative to untreated counterparts. Untreated wounds demonstrated bright collagen birefringence; however, the fibers contributing to this birefringence intensity exhibited a strongly unidirectional distribution (summarized in Table 2).

Fig. 8.

Polarized microscopy of H&E stained wound cross-sections at POD4 [A: preconditioned WT; B: preconditioned HSP70 (−/−); C: untreated WT; D: untreated HSP70 (−/−)]. Samples indicate increased quantity and directionality of striated collagen in samples from HSP70 (−/−) mice (B,D) regardless of pretreatment. Samples from WT mice (A,C) show relatively less deposited collagen in the wound bed with a more heterogeneous directionality. Wound area highlighted by rectangle. Scale bar represents 100 µm.

Fig. 9.

Polarized microscopy of H&E stained wound cross-sections at POD7 [A: preconditioned WT; B: preconditioned HSP70 (−/−); C: untreated WT; D: untreated HSP70 (−/−)]. Samples indicate directional striated collagen in samples from HSP70 (−/−) mice (B,D) and untreated WT mice (C). Preconditioned samples (A,B) show more heterogeneous directionality than their untreated counterparts (C,D). Wound area highlighted by rectangle. Scale bar represents 100 µm.

TABLE 2.

Summary of Representative Histology Comparison

Time point	Mouse type	Treatment type	Inflammatory cell abundance	Epidermal thickness	Collagen birefringence intensity	Collagen fiber orientation
POD4	WT	Untreated	+	0	−	+
POD4	WT	Preconditioning	+	0	−	+
POD4	HSP70 (−/−)	Untreated	−	0	+	−
POD4	HSP70 (−/−)	Preconditioning	−	0	+	−
POD7	WT	Untreated	+	0	−	−
POD7	WT	Preconditioning	+	0	+	+
POD7	HSP70 (−/−)	Untreated	−	0	−	−
POD7	HSP70 (−/−)	Preconditioning	+	0	−	+

⁺Improved performance

⁻decreased performance; 0: no difference.

DISCUSSION

Probe-based RS represents a novel method for realtime, in vivo, quantitative analysis of wound healing. In what we believe to be the first report of the impact of differential genetic expression on wound healing using RS, RS resolved downstream tissue-level skin differences in a mouse model of HSP70 knockout. To a limited extent, the analysis of Raman peak ratios delineated the time course of differential wound-healing treatment response between genotypic groups. Though trends were observed, the current experiment was insufficient to resolve differences due to preconditioning within a given genotype. Given the novelty of this application, a dearth of literature precedence for Raman peak assignments limited analysis.

Trends exhibited in Figures 2 and 3 for WT incisional wound healing imply a spectral profile for known woundhealing phases. In murine wound healing, the changes seen between POD1 and POD3 occur during the inflammatory phase of wound healing, supported by histological absence of acute inflammatory cells by POD4. Spectral results seen at POD7 occur during early remodeling phase of murine wound healing, observed histologically as complete reepithelialization. While time course correlation of spectral trends to histology shows promise for the use of RS, further experimentation is necessary before a detailed spectral profile of wound-healing phases can be established.

Analysis of laser preconditioning required an understanding of spectral variations in time as a function of the presence or absence of HSP70. Figure 4 suggests that HSP70 knockout has a profound effect on skin biochemical composition. Examined in concert with peak ratios these results indicate that some of these differences persist throughout wound healing (Fig. 6), while others are only seen in certain phases (Fig. 7). Applying known peak assignments from Raman literature (Table 1), we can make preliminary associations of peak ratios to woundhealing significance. The ratio of peaks at 1,440– 1,657 cm⁻¹ compares the CH2 stretch of proteins, found in protein side chains, to the presence of the amide I bond, found near amino acid residue linkages. As a ratio, these wavenumbers yield information regarding the degree of conformation and organization of proteins relative to their quantity. In contrast, the ratio of peaks at 1,304– 1,265 cm⁻¹ suggests a measure of either cellularity or cellular activity. The 1,265 cm⁻¹ intensity is more often correlated to the amide III bond of structural proteins, while the 1,304 cm⁻¹ is associated with both histones and nucleic acids [20]. Alternative assignment of 1,304 cm⁻¹ to the CH₂ twisting [21,22] in phospholipid membranes [23], yields a ratio with similar implications on cellularity or cellular activity.

Raman spectroscopy suggests that the presence of HSP70 is involved with the downstream effects of laser preconditioning. Marginal differences seen in the prominent peak intensities of wounds in WT but not in HSP70 (−/−) mice as a function of preconditioning (Fig. 5) suggest that HSP70 forms a part of the known treatment response to laser preconditioning. Lesser known peaks and shoulders exhibited significant differences whose contextual significance remains to be explored. Moreover, this analysis averaged spatially independent samples, including the confounding factors of skin biochemical and anatomic variation. Eliminating this spatial variation, significant cellular differences were observed through the ratio peak intensity (1,304–1,265 cm⁻¹ , Fig. 7) between wounds in WT and HSP70 (−/−) mice, regardless of laser preconditioning. Laser preconditioning led to an increase of this elevated cellular trend by 2 days, fading appropriately as the reepithelialization phase of wound healing makes way for the remodeling phases. While the correlation of HSP70 upregulation to downstream preconditioning effects has been well established [8,9,24,25], the differential effects of the presence of HSP70 expression on the laser preconditioning therapy have not been quantified otherwise.

The extent and degree of experimental analysis was limited largely by the use of the only available mouse model of HSP70 (−/−). Analysis of the remodeling phase of wound healing (>7 days) using Raman spectra was limited because the C57BL/6 background strain develops abnormal skin pigment after 10 days in response to even mild abrasion. This occurs independently of either laser preconditioning or depilation (pilot study, data not shown). Pigmentation changes tissue optical properties and leads to Raman peaks that obfuscate analysis, ultimately preventing spectral confirmation of long-term preconditioning effects reported by Wilmink et al. [8].

The incisional wound model was originally chosen for consistency in preconditioning protocol parameters; however, the low volume of incisional wounds is suboptimal for histological analysis of processes such as inflammation, angiogenesis, fibroplasia, and epithelization. Therefore, histological analysis was limited to establishing the time course of wound healing to frame and qualify the RS results.

In polarization microscopy, collagen birefringence intensity correlates to fiber assembly, and collagen fiber arrangement indicates matrix remodeling in adaptation to mechanical stress. This can be seen macroscopically in scar tissue as large, poorly woven patterns in comparison to adjacent normal dermis. Fiber orientation heterogeneity in WT wound bed collagen (Fig. 8) suggests that HSP70 is implicated in collagen organization in the extracellular matrix. This would support both results by Wilmink et al. [8] (improved tensile strength and cosmetic appearance) and the woundhealing context assigned to the 1,440–1,657 cm⁻¹ peak ratio. Untreated wounds demonstrate bright, strongly unidirectional collagen birefringence on POD7 (Fig. 9), indicating less organization and potentially greater scarring. Because RS is an optical measure, instrumentation can be adapted to quantify biochemistry with sensitivity to polarization, allowing future analysis of these structural phenotypes in vivo.

In our studies, RS represents a plausible method to be developed for future optimization and tracking of laser preconditioning towards clinical applications, provided concurrent development of an RS peak library for healing skin. Spectral fingerprints of individual components, including collagens and elastin can allow for numerical analysis of percent composition [13,14]. During library development, RS can still be utilized to guide investigation without precise knowledge of contextual significance. Applying limited peak assignments from other contexts, we observed RS indicators of prolonged cellularity differences during the inflammation phase, and constant differences in protein organization comparing WT to HSP70(−/−). This detection of differences between genotypically different wounds led to our use of polarization microscopy, and supports the future investigation of RS as a method for the differential diagnosis of altered or chronic wound healing. Had the scope of our study allowed, this information could also have led to different placement of histological endpoints.

As a non-invasive technique to quantitatively probe tissue biochemistry, RS could guide time course evaluation of other studies, due to its clinical and laboratory applicability. Raman could assist in assessment or planning of dermal laser therapy efficacy, guiding timing and placement of crucial biopsies or repeated therapeutic applications. With proper developments of Raman signatures for wound healing, one could speculate that RS may help guide patient-specific care.

CONCLUSIONS

Raman spectral analysis is consistent with the hypothesis that HSP70 is necessary for normal skin composition and implicated in normal wound healing. This finding is consistent with the HSP family role as chaperone proteins responsible for the refolding and repair of damaged protein. It is important to recall that HSP70 is believed to be an intracellular protein, yet its expression will have an indirect effect on the extracellular matrix proteins that are produced within cells in the wound bed. Despite limitations due to the novelty of RS for wound healing and available mouse model physiology, RS indicated prolonged cellularity or cellular activity within the inflammatory phase, directing future investigations.

The non-invasive nature of RS allowed for the collection of a spatio-temporal distribution of biochemical content throughout the wound healing process; subsequently yielding a multivariate data set with statistically significant results, provided only 38 wounds. To the best of our knowledge, this article represents the first in vivo RS analysis of the wound healing time course as a function of model genetic expression and treatment, and as such, interpretation was limited to available information from similar contexts. Nevertheless, quantitative time courses from RS guided our analysis of laser preconditioning, directing future hypotheses and study design. Provided continual development of RS technology and an RS database for wound healing, the fast, non-invasive nature of probe-based RS may help to bridge laboratory and clinical studies. Such application could reduce necessary sample numbers and cost, ultimately assisting in efforts towards patient-tailored dermatology