Biocompatibility of a Novel Cyanoacrylate Based Tissue Adhesive: Cytotoxicity and Biochemical Property Evaluation

Cyanoacrylate (CA) is most widely used as a medical and commercial tissue adhesive because of easier wound closure, good cosmetic results and little discomfort. But, CA-based tissue adhesives have some limitations including the release of cytotoxic chemicals during biodegradation. In previous study, we made prepolymerized allyl 2-CA (PACA) based tissue adhesive, resulting in longer chain structure. In this study, we investigated a biocompatibility of PACA as alternative tissue adhesive for medical application, comparing with that of Dermabond® as commercial tissue adhesive. The biocompatibility of PACA was evaluated for short-term (24 hr) and long-term (3 and 7 days) using conventional cytotoxicity (WST, neutral red, LIVE/DEAD and TUNEL) assays, hematoxylin-eosin (H&E) and Masson trichrome (MT) staining. Besides we examined the biochemical changes in cells and DNA induced by PACA and Dermabond® utilizing Raman spectroscopy which could observe the denaturation and conformational changes in protein, as well as disintegration of the DNA/RNA by cell death. In particular, we analyzed Raman spectrum using the multivariate statistical methods including principal component analysis (PCA) and support vector machine (SVM). As a result, PACA and Dermabond® tissue adhesive treated cells and tissues showed no difference of the cell viability values, histological analysis and Raman spectral intensity. Also, the classification analysis by means of PCA-SVM classifier could not discriminate the difference between the PACA and Dermabond® treated cells and DNA. Therefore we suggest that novel PACA might be useful as potential tissue adhesive with effective biocompatibility.


Introduction
Tissue adhesive is an attractive alternative to the traditional wound closure techniques such as sutures and staples. It should allow rapid adhesion, close apposition of wound edges, and maintenance of strong wound cover for a sufficient time [1]. Good tissue adhesives should be simple, effective, safe, rapid, painless and biodegradable with minimal tissue toxicity. They should also result in an optimal cosmetic appearance of the resultant scar [2,3].
Cyanoacrylates (CA) possess some of these properties and can be applied in medicine with good cosmetic outcomes. The commercial CA tissue adhesives for medical applications are octyl-2-CA (DermabondH, Johnson & Johnson/Ethicon, Somerville, NJ) and n-butyl-CA (HistoacrylH, B. Braun, Melsungen, Germany) which are longer chain derivatives. The length of the alkyl chain is important because the toxicity can be reduced with increased carbon number in the alkyl chain [4][5][6][7]. Medical applications must be non-toxic, no harmful side effects. Therefore, assessment of the cell viability and cytotocixity is a necessary step in the evaluation of biocompatibility [8]. In addition, in order to avoid overestimation or underestimation of the toxicity of biomaterials, more than one assay should be used to determine cytotoxicity assay, as this would increase the reliability of the results obtained [9,10].
Raman spectroscopy is a powerful analytical technique that is rapid, label-free, non-invasive, non-destructive and has high sensitivity, which can be used for the analysis of biological samples [11]. This is a well-established tool used in many research fields to directly investigate the molecular compositions and structures of the biological samples [12,13]. This technology represents highly multiplexed biochemical information of DNA, RNA, proteins, and lipid content as well as conformation of the living cell by spectral shape or intensity [14]. Raman spectroscopy has been applied to analyze the toxic effects of polymeric nanoparticles and gold nanoparticles. It is well-suited technique to uptake studies of nanoparticles into cells, as well as the cytotoxicity study for drug delivery [15,16]. Yao et al reported the potential of Raman spectroscopy as distinguishing between a single apoptotic cell and carcinoma cell for monitoring apoptosis, evaluating the efficacy of anti-cancer drug induced apoptosis in gastric carcinoma cells [17]. The cytotoxic effects of toxic agent on living cells could be evaluated from Raman spectra of biochemical change related to cell death [18].
Previously, we reported that partial pre-polymerization of allyl 2-cyanoacrylate (PACA) causes longer chain structure leading to improvement the biocompatibility of common CA [19]. In this study, we evaluated and compared in vitro and in vivo biocompatibility of the PACA and commercial CA tissue adhesive DermabondH. The cytotoxicity on the PACA and DermabondH was tested with direct and indirect contact for time course on fibroblast cell culture, and changes in the biochemical property at a molecular level after the exposure to tissue adhesives using Raman spectroscopy with multivariate statistical methods including principal component analysis (PCA) and support vector machine (SVM). These results were discussed the change of protein and DNA related to the cell death. The biocompatibility in vivo on the PACA and DermabondH was confirmed by histological analysis of animal dermal tissues.

Cytotoxicity of tissue adhesives assessment for short time using four different assays
Mouse fibroblast L929 cells were treated with same dose of the PACA and DermabondH tissue adhesives (5 ml/10 5 cells) for 24 hr by direct contact. Cytotoxicity was assessed by the Water-soluble tetrazolium salt (WST), Neutral red (NR), LIVE/DEAD and TUNEL assays. The results were observed in experiments performed in triplicate. Both the PACA and DermabondH tissue adhesives had similar cell viability values compared to the controls ( Fig. 1A and 1B). The WST and NR assays showed 54-57% cell viability for the cells treated with the PACA tissue adhesive and 53-57% cell viability for the cells treated with the DermabondH tissue adhesive. This result was confirmed by LIVE/DEAD viability/cytotoxicity assay using confocal microscopy. Representative images were selected from the results of one set experiment among three experiments. There was no difference in cell viability between the PACA and DermabondH tissue adhesive treated cells (Fig. 1C). Fig. 1D shows the results of the TUNEL assay, apoptotic cell detections of TdT-Fluor in the control and two tissue adhesives treatment were compared to confirm apoptosis and quantify cell death. DermabondH tissue adhesive treatment induced small scale DNA fragmentation compared to the PACA tissue adhesive.

Cytotoxicity of tissue adhesives assessment for time course using direct and indirect contact test
To evaluate the cytotoxicity for time course, the cells were seeded to a 12 well plate and maintained at 37uC for 24 hr. The PACA and DermabondH were added directly to the center of each well for times courses (24 hr, 48 hr and 72 hr) and cell viability was assessed by WST assay. Cell viability (Fig. 1E) (24-72 hr). In the indirect contact method, PACA and DermabondH were allowed to medium for 24 hr at 37uC before cell treatment. Mouse fibroblast L929 cells were exposed elution medium of PACA and DermabondH for time courses. Cell viability (Fig. 1F) showed more than 80% cell viability on the time course. Indirect contact method showed no cytotoxicity both PACA and DermabondH.

Evaluation of in vivo biocompatibility from incision rat model
To evaluate the in vivo biocompatibility, we performed histological analysis. Figure 2 shows the representative H&E and MT staining images at day 7 after incision and PACA or DermabondH treatments. The tissue adhesive-treated group showed the healing of incision area, whereas the non-treated group was not sealed. Also, both tissue adhesive treatment groups showed the similar distribution of collagen in the dermal layer. However PACA group showed less pathologic inflammation than DermabondH group. Fig. 3A shows the mean Raman spectra of the control and two adhesive treated cells (a: control group, b: PACA group, c: DermabondH group) and the mean spectral differences between the control and tissue adhesives (d: control-PACA group, e: control-DermabondH group, f: PACA-DermabondH group). It is clear that all Raman spectra were significantly decreased when the cells were treated with PACA and DermabondH as compared to control cells. Raman intensity of the PACA and DermabondH treated cells is almost similar. We selected some specific Raman spectra of compared the changes in their spectral intensities. The changes of spectra at 725 cm 21 (adenine), 778 cm 21 (cytosine/ thymine ring breathing of the DNA/RNA), 1002 cm 21 (symmetric ring stretching phenylalanine), 1096 cm 21 (phosphodioxy groups, PO 2 2 , of the DNA backbone) [24], 1257 cm 21 (amide III b-sheet of proteins), 1656 cm 21 (amide I a-helix of proteins) are from cells ( Table 1). The main spectra intensity changes related to the proteins can be observed at 1002 cm 21 , 1257 cm 21 , and 1656 cm 21 [17,18,23]. The Raman spectral intensities of tissue adhesive treated cells were significant decrease in the protein vibrations at 1002 cm 21 (43%), 1257 cm 21 (45%) and 1656 cm 21 (50%). The reduction of Raman spectral intensity corresponding to DNA/RNA vibrations at 725 cm 21 (36%) 778 cm 21 (33%) and 1096 cm 21 (46%) (Fig. 3B).

Effect of PACA and DermabondH tissue adhesives on Raman spectral of cells and DNA
Binding with cellular DNA is a crucial step in the mechanism of cytotoxicity. In order to assess the effect of tissue adhesive treatment on cellular DNA spectral signatures, we investigated the spectra of the genomic DNA. DNA was isolated from the cells to minimize the interfering signals from cytoplasm. Raman spectra of each DNA were obtained from an average of 15 spots with same dose of DNA samples. Polymerase chain reaction (PCR) products b-actin was used as the internal control (Fig. 4C). Fig. 3A shows Raman spectra obtained from control DNA and two tissue adhesive treated DNA, a: control group, b: PACA group and c: DermabondH group and the mean spectral differences between the control DNA and tissue adhesives treated DNA, d: control-PACA group, e: control-DermabondH group, f: PACA-Derma-bondH group. All Raman spectral intensity decreased compared to the control DNA ( Fig. 4A and B). The changes of spectra at 759 cm 21 (ring breathing tyrosine) 799 cm 21 (P-O-C symmetric stretching), 911 cm 21 (deoxyribose), 1156 cm 21 (deoxyribose and phosphate) 1257 cm 21 (thymine and adenine), 1470 cm 21 (guanine) and 1585 cm 21 (guanine and adenine) were from DNA ( Table 1). The reduction of Raman spectral intensity at 799 cm 21 (30%), 1257 cm 21 (26%) and 1470 cm 21 (30%) (Fig. 4B). We found no differences in the two adhesive for the cytotoxicity effect.

PCA-SVM classifiers for evaluating biocompatibility of PACA and DermabondH tissue adhesives on Raman spectral of cells and DNA
To estimate the cytotoxicity of two tissue adhesives, the PCA-SVM analysis was applied to Raman spectra of cells and DNA. The first step for the discrimination analysis was to determine the most relevant PC in PCA analysis of Raman spectra for cells and DNA, prior to input for SVM analysis. PCA analysis was performed at cytotoxicity associated with six or principal wavelengths of Raman spectra for each group. Only two PCs (more than 99% of the variance) were retained for SVM analysis; PC1 and PC2 for cells and PC1 and PC3 for DNA. The SVM classifier with linear discriminant function was used to discriminate between cytotoxic and non-cytotoxic tissue adhesives by means of the capability of Raman spectroscopy. For Raman spectroscopy based cytotoxicity assessment in cells (Fig. 5A), the tissue adhesive treated cells showed the significant difference with 97.14% sensitivity, 94.44% specificity and 96.23% accuracy compared to the controls (n = 18, Table 2). However, the proposed SVM classifier with linear kernel could not discriminate between the PACA and DermabondH treated cells, due to high correlation between two tissue adhesives. For Raman spectroscopy-based cytotoxicity assessment in DNA (Fig. 5B), the tissue adhesive treated DNA showed the significant difference with 100% sensitivity, 100% specificity and 100% accuracy compared to the controls (n = 14, Table 2). Similar to the cells there was no significant difference between the PACA and DermabondH treated DNA samples.

Discussion
The cytotoxicity assessment is the first process for examining the biocompatibility of biomaterials. Particularly, it is essential for developing new biomedical materials that modulate cell viability and to identify potential cytotoxic side effects of biomedical devices. This study was designed to investigate the cytotoxicity of new pre-polymerized of CA-based tissue adhesive (PACA) using a direct and indirect contact test in vitro cell culture. Direct contact testing of novel biomaterials with in vitro cell cultures may provide an early indicator of potential problems in vivo [20]. The tissue adhesives release toxic substances, so cell cytotoxicity was tested by indirect contact. Mouse L929 fibroblasts which are the main cellular component of dense connective tissues were used for cytotoxicity evaluation of tissue adhesive. Furthermore, L929 fibroblasts are the cells recommended by the international standard ISO-10993-5 for in vitro biocompatibility evaluations.
We comparatively assessed in vitro and in vivo cytotoxicity of new tissue adhesive PACA and commercially available tissue adhesive DermabondH. The four cytotoxicity assays were employed to assess the toxicity of tissue adhesives in vitro at short time and showed that the PACA and DermabondH are similar cytotoxicity (Fig. 1A-D). Although the DermabondH tissue adhesive treatment induced small scale DNA fragmentation from apoptotic cell death, the marked incorporation of FITC, compare to the PACA tissue adhesive (Fig. 1D). Also, the direct and indirect contact tests with the time course showed similar cell survival between PACA and DermabondH ( Fig. 1E and F). PACA tissue adhesive showed the most effective cell viability. The finding of PACA tissue adhesive suggests that the polymerization process by heat treatment could decrease the cell cytotoxicity [21]. In vivo histological analysis was performed to assess long-term biocompatibility (Fig. 2). The PACA and DermabondH treatment group were enhanced wound healing than the positive control group, whereas the PACA group showed less inflammation compared to the DermabondH group. These finding suggests that the PACA tissue adhesive led to superior skin    tissue closure, less inflammation, and less foreign body reaction compared to the DermabondH. Raman spectroscopy technique is based on the inelastic scattering of laser photons by the molecules of the biological samples. Thus, Raman spectrum represents a molecular fingerprint of their composition. Some studies evaluated the cytotoxic changes in response to adverse effects of chemicals agents on cells using Raman spectroscopy. This technique has the advantage of overcoming the limitations of conventional cytotoxicity assays for identification and quantification of chemicals agents [14,22]. Furthermore, Raman spectroscopy could distinguish the different stages of the cell cycle from the cells as well as identify unique DNA and protein peaks of dead cells in the culture within only few minutes [23,24].
In this study, Raman spectroscopy was used to detect and identify the change in cellular components including protein, DNA/RNA, lipid and carbohydrates related to the cytotoxicity by tissue adhesive treatment. Individual L929 cells showed specific Raman spectra of cellular components by tissue adhesives treatment ( Table 1). The spectra of cells treated with PACA and DermabondH tissue adhesives showed large decrease in the spectra corresponding to DNA/RNA vibrations at 725 cm 21 , 778 cm 21 , 799 cm 21 , 1257 cm 21 , 1470 cm 21 , and 1096 cm 21 and significant decrease intensity corresponding to protein vibrations at 1002 cm 21 , 1257 cm 21 , and 1656 cm 21 .
The most of important the decrease of the spectra was correlated with positions to DNA/RNA including DNA phosphodioxy group PO 2 2 stretching, DNA backbone O-P-O asymmetric stretching and DNA bases cytosine, thymine and adenine ( Table 1). The phosphate group of the phosphodioxy PO 2 2 stretching is sensitive to DNA structure [25]. A decrease in the spectral intensity of DNA structure, DNA backbone and DNA bases is likely to be affected by DNA strand breaks and nuclear fragmentation during the cell death [18]. These changes of intensity might be caused by the destruction of the ring structure, indicating disintegration of the DNA/RNA related to the cell death [26][27][28].
Besides the decrease in the Raman spectra of DNA/RNA, the Raman spectra of cells indicated changes in protein vibrations. The spectral intensity at 1002 cm 21 corresponding to the ring breathing of phenylalanine showed 43% reduction and that in the spectra at 1257 cm 21 corresponding to cell death showed 45% reduction. A decrease in the protein content in dying cells followed by the induction of the apoptosis proteins for cleavage and slicing is suggested by the decrease in most Raman spectra of proteins including phenylalanine (1002 cm 21 and 621 cm 21 ), tyrosine (853 cm 21 ) and C-C backbone stretching a-helix (932 cm 21 ) [26]. The spectral intensity at 1656 cm 21 corresponding to amide I ahelix showed 50% reduction. This spectra was observed in Raman spectra of dead cells, indicating conformational changes of proteins [18,[26][27][28]. A decrease in Raman spectra corresponding to proteins might indicate denaturation and conformational changes in proteins related to the cell death. This change involves with molecular mechanisms of membrane proteins signaling with the biochemical changes of the nucleus [18]. We suggest that the spectral regions of most dramatic changes, 725 cm 21 , 778 cm 21 , 1096 cm 21 , 1002 cm 21 , 1257 cm 21 and 1656 cm 21 could be used a maker of the cell death for cytotoxicity study.
PCA is a multivariate statistical technique used in bioinformatics. It analyzes a data matrix that represents the observations taken by several intercorrelated variables. The main goal of PCA technique is to extract dominant information from the complicated data matrix and to express the features as a set of new orthogonal variables called a PC. We compared the cytotoxicity between the control and tissue adhesive treated cells and DNA by combining Raman spectral information with PCA-SVM classifier as bioinformatics technique [28,29]. PCA-SVM analysis with linear discriminant function led to more than 99% of the variance from both cells and DNA. A two-dimensional plot (PC1, PC2) for cells and (PC1, PC3) for DNA clearly showed the no difference  between PACA and DermabondH tissue adhesives (Fig. 5). The result suggests that PACA tissue adhesive showed almost similar cytotoxicity effect on cells and DNA compared to commercial DermabondH tissue adhesive.
In summary, we have presented the biocompatibility of PACA tissue adhesive for clinical application using conventional cytotoxicity (WST, NR, LIVE/DEAD and TUNEL) assays and Raman spectroscopy with PCA-SVM analysis. Additionally, we evaluated the changes in biochemical properties using Raman spectroscopy. The Raman spectra indicated significant changes in biochemical properties related to nucleic acids and proteins by the tissue adhesive treatment. The reduction of Raman intensity is responsible for conformational changes of proteins, structure change of DNA and degradation of DNA related to the cell death. The cytotoxicity of PACA tissue adhesive in all methods does not suffer by comparison with commercial tissue adhesive Derma-bondH. Therefore, these finding indicate that PACA tissue adhesive could be useful an alternative tissue adhesive for skin wound sealing.

WST assay
Mouse L929 fibroblasts (1610 5 cells) were seeded in a 12 well plates and incubated for 24 hr. Half maximal inhibitory concentration (IC 50 ) for DermabondH is 5 ml/10 5 cells. The cells were incubated in direct contact with PACA and DermabondH (5 ml/ 10 5 cells) for 24 hr at 37uC. Cell cytotoxicity was determined using a Cell counting-8 kit (Sigma Aldrich.St. Louis, MO, USA). OD 450 was recorded utilizing the Synergy HT multi-mode microplate instrument (BioTek, Winooski, VT, USA) and converted to cell viability percentage as compared to the control. To examine the time course of cell cytotoxicity, the cells were incubated in direct contact with PACA and DermabondH for 24 hr, 48 hr and 72 hr. The cells were determined for cytotoxicity assessment using Cell counting-8 kit.

NR assay
The NR assay determines the accumulation of the NR dye (Sigma Aldrich. St. Louis, MO, USA) in lysosomes of viable cells. The cells were incubated for 2-3 hr with culture medium including 0.005% NR following the exposure to PACA and DermabondH. The cells were then washed with PBS and 150 ml desorbing fixative solution (EtOH/AcCOOH, 50%/1%) was added followed by gentle shaking for 10 min so that complete dissolution was achieved. OD 595 was recorded using the multimicroplate reader.

LIVE/DEAD viability/cytotoxicity assay
Two color fluorescent cell cytotoxicity assay (LIVE/DEAD viability/cytotoxicity kit; Molecular Probes, Eugene, OR, USA) was used to confirm the results obtained from the WST assay.
After the exposure of 24 hr tissue adhesives, the cells were incubated with a mixture of 8 mM ethidium homodimer (EtdD-1; red color) and 2 mM calcein acetoxymethyl (calcein-AM; green color) in PBS for 30-45 min at room temperature (RT). Stained samples were washed with PBS. Images were collected using a Ziess LSM-700 confocal microscopy system (Thornwood, NY, USA). The numbers of viable (green) and non-viable (red) cells were counted automatically with Image J software (National Institutes of Health, Bethesda, MD, USA). All experiments were performed in triplicate.

Elution assay
Same volume (5 ml/ml) of PACA and DermabondH were added in RPMI medium and then collected elution mediums of PACA and DermabondH for 24 hr. Control sample was not exposure with tissue adhesive. The cells were seeded in a 12 well plate and incubated for 24 hr. The cells were treated with elution mediums containing final 5% FBS. Following incubations for 24 hr, 48 hr and 72 hr, the cells were determined for cytotoxicity assessment by Cell counting-8 kit. OD 450 was measured with a multi-microplate reader. The results of each sample were averaged and were expressed as a percentage of the control.

In vivo studies
Fourteen young male Sprague-Dawley rats (250-350 g) were purchased from Orient Bio Korea (Seoul, South Korea). The rats were housed in a controlled environment with temperatures of 22-24uC and a 12-hr light/dark cycle and fed commercial rat chow and water ad libitum. Adequate injection and anesthesia were performed to minimize pain or discomfort. All experimental animals were cared for and treated according to the ''Guide for the Care and Use of Laboratory Animals'' issued by the Korea University School of Medicine, Seoul, South Korea (IRB; KUIACUC-2012-139). All animals were anesthetized with inhaled isoflurane (1-5%) and then injected rompun 5-10 mg/kg and ketamine 30 mg/kg were intramuscularly. After anesthesia, hair from the animal's back was removed with electric clippers was applied for several minutes. The area was washed with wet cotton and dried with a gauze sponge. Two 2-cm long incisions were made on the back of each rat. The optimum PACA and DermabondH were applied as a monolayer over the incision without additional treatment, according to instructions for the commercial tissue adhesive. Negative control group without an incision, and a positive control group was treated only 70% ethyl alcohol after the incision. Rats were sacrificed at day 7 after surgery to examine and compare the tissue reactions between the four groups. Each tissue specimen was fixed in 10% buffered formalin, embedded in paraffin, sectioned into sections 5 mm thick, and stained with H&E and MT to observe inflammatory cell infiltration and collagen distribution. The stained tissues were observed under an optical microscope.

Isolation of genomic DNA
The cells were incubated with PACA and DermabondH for 24 hr at 37uC. Cells were harvested by trypsinization, washed once in PBS. DNA preparation was performed by High Pure PCR Template Preparation Kit (Indianapolis, IN, USA). Before starting the purification reaction, warm up the Elution buffer to 70uC. 200 ml of sample material was added 200 ml Binding buffer and 40 ml proteinase K. It was mixed immediately and incubated at 70uC for 10 min. After adding 100 ml isopropanol, the suspension was mixed properly and the sample was loaded into High filter tube. Centrifugation was performed at 8000 g for 1 min. After centrifugation, the High filter tube was combined with new collection tube and 500 ml of Inhibitor Removal buffer was added. Centrifugation was performed again at 8000 g for 1 min. Then the High filter tube was combined again with another new collection tube and washed two times with 500 ml of Wash buffer. After discarding the flow through liquid, the entire High pure assembly was centrifuged additional 10 sec at full speed and collection tube was discarded. Pre-warmed Elution buffer was added in order to elute the DNA and then centrifuged for 1 min at 8000 g. DNA concentration was calculated by the Nano-100 Micro-spectrophotometer (Allsheng, Hangzhou city, China).

Polymerase Chain Reaction (PCR)
To confirm the PCR products b-actin, amplified from same concentration of DNA samples, DNA fragment (776 bp) was generated by PCR from the genomic DNA using primers b-actin F (59-ATGGGTCAGAAGGACTCCTACG-39) and b-actin R (59-AGTGGTACGACCAGAGGCATAC-39) (Bioneer, Daejeon, South Korea). PCRs were conducted at 94uC (30 sec), 60uC (30 sec) and 72uC (1 min) for 27 cycles using the GeneAmp PCR system 9700 (applied biosystems, Foster City CA, USA). An aliquot of the reaction was separated on a 0.8% agarose gel and the intensity of signals was quantified with GelDoc2000 system (Bio-Rad, Hercules, CA, USA).

Raman spectroscopic measurements
For the Raman analysis, the cells were seeded in gold coated substrates and incubated for 24 hr. The cells were treated with PACA and DermabondH (5 ml/10 5 cells) and then further incubated for 24 hr. The cells were washed twice with PBS and fixed for 20 min in 4% paraformaldehyde in PBS at 4uC and washed with 5 ml PBS. To Raman analysis of DNA, DNA prepared with DNA preparation kit and measured DNA concentration using Nano-100 Micro-spectrophotometer. Same concentration of DNA samples were added in gold coated substrates. Raman spectra were acquired using the SENTERRA confocal Raman system (Bruker Optics Inc., Billerica, MA, USA) equipped with a 785 nm diode laser source (100 mW before objective) and the resolution of 3 cm 21 . A 1006 air objective (MPLN N. A. 0.9, Olympus), which produced a laser spot size of ,1 mm was used to collect Raman signals and focus laser on a single cell. Raman spectra of cell and DNA were calculated as the average of fifteen measured. All Raman measurements are recorded with accumulation time of 60 sec in the 600-1750 cm 21 range, and spectral acquisitions were carried out using the OPUS software (Bruker, Optics Inc., Billerica, MA, USA).

Data analysis
PCA analysis was used to extract the major trends in each group. The determined principal vectors including Raman spectral intensities at 725 cm 21 , 778 cm 21 , 1002 cm 21 , 1096 cm 21 , 1257 cm 21 and 1656 cm 21 for cells and Raman spectral intensities at 759 cm 21 , 799 cm 21 , 911 cm 21 , 1156 cm 21 , 1256 cm 21 , 1470 cm 21 , and 1585 cm 21 for DNAs were used as the transfer function to determine the principal component (PC) scores for each spectrum of each group. Multi-SVM classifier with one-against-all approach based hierarchical structure was considered to be the three-class problem to classify the control, PACA and DermabondH treated cells and DNAs. The performance of SVM classifier was assessed by the sensitivity (SEN), specificity (SPE), and accuracy (ACC), where SEN is used for evaluating the ability of the classifier to detect the cytotoxicity of cells and DNAs, SPE is for the ability of the classifier to specify control cells and DNAs and ACC is for the accuracy of the classifier. PCA-SVM algorithm was implemented with the MATLAB software (MathWorks Inc., Natick, MA, USA).

Statistics
Summary data were reported as mean 6 SD. The results of WST and NR assays were obtained from eight sets of the same experiments. The results of LIVE/DEAD and TUNEL assays were obtained from three sets of the same experiments. Differences between each group were evaluated using the Mann-Whitney U test. P,0.05 was considered statistically significant.