- Email: [email protected]
Discrimination of a transformation phenotype in Syrian golden hamster embryo (SHE) cells using ATR-FTIR spectroscopy
Toxicology 258 (2009) 33–38
Contents lists available at ScienceDirect
Toxicology journal homepage: www.elsevier.com/locate/toxicol
Discrimination of a transformation phenotype in Syrian golden hamster embryo (SHE) cells using ATR-FTIR spectroscopy Michael J. Walsh a,1 , Shannon W. Bruce b , Kamala Pant b , Paul L. Carmichael c , Andrew D. Scott c , Francis L. Martin a,∗ a b c
Centre for Biophotonics, Lancaster Environment Centre, Lancaster University, Lancaster LA1 4YQ, UK BioReliance Corporation, Medical Center Drive, Rockville, MD, USA Safety and Environmental Assurance Centre, Unilever Colworth Science Park, Bedfordshire, UK
a r t i c l e
i n f o
Article history: Received 4 December 2008 Accepted 5 January 2009 Available online 9 January 2009 Keywords: ATR-FTIR spectroscopy Benzo[a]pyrene Linear discriminant analysis Morphological transformation Principal component analysis Syrian hamster embryo cells
a b s t r a c t Primary Syrian hamster embryo (SHE) cells might be used to assess morphological transformation following treatment with chemical carcinogens. We employed attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy to interrogate SHE colonies, as complex biomolecules absorb in the mid-infrared (IR; = 2–20 m) giving vibrational spectra associated with structure and function. Earlypassage SHE cells were cultured (pH 6.7) in the presence or absence of benzo[a]pyrene (B[a]P; 5.0 g/ml). Unstained colonies were applied to an ATR crystal, and vibrational spectra were obtained in the ATR mode using a Bruker Vector 22 FTIR spectrometer with Helios ATR attachment. These were individually baseline-corrected and normalised. Spectra were then analysed using principal component analysis (PCA) plus linear discriminant analysis (LDA). PCA was used to reduce the dataset dimensions before LDA was employed to reveal clustering. This determined whether wavenumber–absorbance relationships expressed as single points (scores) in ‘hyperspace’ might on the basis of multivariate distance reveal biophysical differences associated with morphologies in vehicle control (non-transformed or transformed) or carcinogen-treated (non-transformed or transformed) cells. Retrospectively designated SHE colonies (following staining and microscopic analysis) clustered according to whether they were vehicle control (non-transformed), B[a]P-treated (non-transformed) or transformed (control and B[a]P-treated). Scores plots pointed to a B[a]P-treated phenotype and derived loadings plots highlighted distinguishing markers in control transformed vs. B[a]P-treated transformed; these were mostly associated with Amide I, Amide II and phosphate stretching (asymmetric and symmetric) vibrations. Combined application of ATR-FTIR spectroscopy and unsupervised (PCA)/supervised (LDA) may be a novel approach to scoring SHE colonies for morphological transformation. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Distinguishing non-transformed from transformed foci (colonies) of primary Syrian golden hamster embryo (SHE) cells cultured (at pH 6.7) in the presence or absence of test agent has been used to screen compounds for potential carcinogenicity (LeBoeuf et al., 1996). This has been achieved by microscopic assessment of colony morphology allowing the identification of those colonies exhibiting features of morphological transformation (including, randomly orientated criss-crossed, three-dimensional colony growth characteristics, with colonies composed of basophilic
∗ Corresponding author. Tel.: +44 1525 594505; fax: +44 1524 593192. E-mail address: [email protected] (F.L. Martin). 1 Present address: 5247 Beckman Institute, 405 North Mathews Avenue, Urbana, IL 61801, USA. 0300-483X/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.tox.2009.01.003
stacked cells). However, the process of scoring colonies is subjective and there is a need for an approach that reliably identifies these changes. Furthermore, the SHE assay suffers from a lack of biological understanding of the fundamental mechanism(s) involved in the transformation process. Fourier-transform infrared (FTIR) microspectroscopy can generate an IR ‘biochemical-cell fingerprint’ in cellular systems (German et al., 2006a). For example, a signature of susceptibility-to-adenocarcinoma, according to zonal location in the prostate, has been proposed using IR analysis (German et al., 2006a). Similarly, putative stem cells of the cornea appear to possess identifying IR spectral characteristics (Bentley et al., 2007; German et al., 2006b). IR microspectroscopy may also be applied to monitor cell cycle (Hammiche et al., 2005) or response to toxic insult (Barber et al., 2006). Biomolecules absorb in the mid-IR (2–20 m) giving rise to vibrational spectra. This allows the measurement of a range of cellular components including protein conformation, DNA, RNA, lipids
34
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
and glycogen content; these are often characteristic of cell type and/or pathology (Walsh et al., 2007a). IR absorbance bands include Amide I (1650 cm−1 ), Amide II (1550 cm−1 ), Amide III (1260 cm−1 ), asymmetric phosphate (as PO2 − ; 1225 cm−1 ), symmetric phosphate (s PO2 − ; 1080 cm−1 ) and glycogen (1030 cm−1 ) (Walsh et al., 2007b, 2008). Alterations in absorbance bands (i.e., shape, shifts and/or intensity changes) are indicative of intracellular alterations. The Amide I band is mainly due to the (C O) bond whereas the Amide II peak associated with ␦ (N–H) and (C–N) protein bonds (Goormaghtigh et al., 2006; Mantsch and Chapman, 1996). Attenuated total reflection-FTIR (ATR-FTIR) spectroscopy facilitates the interrogation of specimens with a spatial resolution of ≈250 m × 250 m. This gives a high signal-to-noise ratio (SNR), with fast acquisition of IR spectra (Walsh et al., 2007b). Spectroscopic studies generate large data sets that subsequently require computational data reduction approaches in order to identify variance e.g., principal component analysis (PCA) (Walsh et al., 2007a). With PCA, each spectrum is converted into a single point (or score) in n-hyperspace and groups of scores may be viewed in three-dimensions rotated on principal components (PCs). Scores are spatially separated proportional to spectral similarity. The objective identification of morphologically transformed colonies remains a key determinant in the conduct of the SHE assay and there are no biomarkers that define the underlying progression in response to either DNA-reactive genotoxic or non-genotoxic carcinogens. In this pilot study, we set out to interrogate unstained and unidentified SHE colonies that had been cultured with or without a prior exposure to benzo[a]pyrene (B[a]P). We investigated whether such colonies, subsequently categorised by conventional visual scoring as control (untransformed), B[a]P-treated (untransformed), control (spontaneous transformed) or B[a]P-treated (transformed) colonies, could be segregated following interrogation with ATRFTIR spectroscopy and PCA–LDA. 2. Materials and methods 2.1. Primary Syrian golden hamster embryo (SHE) cell transformation assay (pH 6.7) The clonal transformation assay was carried out using SHE cells following a 7day exposure with or without test compound. The SHE transformation assay was designed to allow the expression of morphologically transformed clonal SHE cells seeded onto feeder cells, after exposure for 7 days to B[a]P (5.0 g/ml) as well as vehicle control (DMSO) treatments in the absence of a supplemental exogenous mammalian metabolic activation system. The final concentration of the vehicle in treatment medium did not exceed 0.2% (v/v). Following treatment, the cells were cultured to determine the cytotoxic effects of treatment and the induction of morphological transformation (Kerckaert et al., 1996). 2.2. Preparation of feeder cells Cryopreserved SHE cells from a tested and approved lot were thawed and grown to 50–90% confluency in growth flasks (2–4 days). On Day 1 of the assay, feeder cells were detached and suspended in culture medium in a growth flask on wet ice. They were then X-ray irradiated (approximately 5000 rad); 5 plates containing only feeder cells were cultured long term to confirm that they were no longer viable or capable of replication. Following irradiation, the cells were re-suspended in culture medium, counted using a haemocytometer and the cell concentration adjusted to 0.67 × 104 cells/ml in culture medium; 6.0 ml of this suspension was placed into each 60-mm culture dish containing a 1 cm × 1 cm low-E reflective glass slide. 2.3. Preparation of target cells and treatment The day after seeding feeder cells, a second vial of SHE cells from the same lot was thawed and seeded in a growth flask for approximately 5 h. After the 5-h incubation period, the target cells were detached, counted with a haemocytometer, and diluted with culture medium to a concentration that yielded approximately 25–45 colonies/dish. Two milliliters of the target cell suspension was placed into each culture dish containing glass slide and feeder cells. Dishes were incubated at 37 ± 1 ◦ C in a humidified atmosphere of 10 ± 0.5% CO2 for approximately 24 h prior to treatment. The pH of the culture medium at the time of preparation was measured and adjusted to 6.7 ± 0.05 prior to use in the assay. The pH of the dosing solutions was
checked after at least 4 h of undisturbed incubation at 37 ± 1 ◦ C with approximately 10 ± 0.5% CO2 in air (LeBoeuf and Kerckaert, 1987). Test article stock solution was prepared by dissolving B[a]P in DMSO (500×) at 2.5 mg/ml. This test solution was diluted 1:250 with culture medium to yield a 2× desired final concentration. Prior to treatment, 4.0 ml medium was removed from each culture dish. Four milliliters 2×B[a]P in 0.4% solvent in culture medium was then added to each treatment dish. Each culture dish in the solvent control group received 4 ml of 0.4% solvent in culture medium. All culture dishes were incubated (undisturbed) at 37 ± 1 ◦ C in a humidified atmosphere of 10 ± 0.5% CO2 for 7 days. All dishes with feeder-cells only were replenished with 6 ml complete medium on the treatment day. 2.4. Scoring colonies for transformation Following the 7-day incubation, the culture medium was removed from each dish. The SHE cell colonies in dishes with glass slides were fixed with methanol. The slides were then air-dried and placed in slide holders and marked for orientation. 2.5. Attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy Unstained, but macroscopically visible, SHE colonies adhered to 1 cm × 1 cm low-E glass microscope slides (Kevley Technologies, Chesterland, OH, USA) were interrogated using ATR-FTIR spectroscopy. IR spectra were acquired using the Bruker Vector 22 FTIR spectrometer with Helios ATR attachment containing a diamond crystal (Bruker Optics Ltd, Coventry, UK). Using a CCTV camera attached to the ATR crystal, 5 random points per colony were interrogated. Data was collected in ATR mode and spectra (8 cm−1 spectral resolution, co-added for 32 scans) were converted into absorbance using Bruker OPUS software. Sodium dodecyl sulphate was used to clean the ATR crystal prior to the first spectral analysis of each colony. Each spectrum had a background absorption automatically subtracted, was baseline-corrected and normalised to the Amide I absorbance peak (1650 cm−1 ) using OPUS software. This normalisation function during spectral pre-processing also removed any influence of cell density or absorbance. Average absorbance spectra were then derived using OPUS software, resulting in one absorbance spectrum per colony. 2.6. Computational analysis Raw spectra were processed employing principal component analysis (PCA) performed using the Pirouette software package (Infometrix Inc., Woodinville, WA, USA). PCA scores plots were derived and various principal components (PCs) were examined to see which gave best segregation. This was conducted without knowledge of colony designation into one of four categories i.e., control (untransformed), B[a]P-treated (untransformed), control (spontaneous transformed) or B[a]P-treated (untransformed). PCA loadings plots were then derived to determine spectral variance on the selected PCs. Using PCA alone and following retrospective designation of interrogated colonies into the aforementioned categories, remarkable clustering was obtained (data not shown). PCA was used for preliminary data reduction and the output was processed using linear discriminant analysis (LDA) (Martin et al., 2007) incorporating our previously-published cluster vector approach (German et al., 2006a). PCA–LDA scores and loadings plots were derived for the biochemicalcell fingerprint region (1850 cm−1 to 900 cm−1 ). A PCA-LDA loadings plot displays the coefficients by which each of the original wavenumber variables must be multiplied to obtain the hyperspace vector passing through the median of a chosen cluster. It thereby picks out those spectral bands that are primarily responsible for the discrimination of those spectra from all the spectra taken as a whole. 2.7. Retrospective scoring Following IR-spectral acquisition, the glass slides were stained with Giemsa and air-dried. Using a stereo microscope, the colonies were scored. Each colony was evaluated and recorded as either normal or morphologically transformed. Normal colonies tend to be in a monolayer and contain cells with an organised, often flowing, pattern of growth with minimal cell criss-crossing, particularly where the cells are at a confluent density (Fig. 1A and B). Morphologically transformed colonies contain cells arrayed in an extensive randomly oriented, three-dimensional, stacked growth pattern, with criss-crossing of cells at the perimeter and in the interior of the colony (Fig. 1C and D). Cells in morphologically transformed colonies are frequently more basophilic than their normal counterparts and have increased nuclear/cytoplasmic ratios (Isfort et al., 1994), irrespective of their phenotypic origin (e.g., epithelial, fibroblast, etc.) (LeBoeuf et al., 1990a, 1990b). Colonies where the cell growth was too sparse to assess morphology or where the majority of the colony was missing were not assessed for morphological transformation.
3. Results Two independent experiments were conducted in which SHE cells were cultured in the presence or absence of 5 g/ml B[a]P. In Experiment 1, unstained colonies were identified and per colony,
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
35
Fig. 1. Photomicrographs of foci (colonies) of cultured primary Syrian golden hamster embryo (SHE) cells (pH 6.7). Each colony was evaluated and recorded as either (A, B) normal or (C, D) morphologically transformed.
Fig. 2. Interrogation using ATR-FTIR spectroscopy of SHE cells cultured in the presence or absence of 5 g/ml B[a]P treatment (Experiment 1). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were identified and per colony, 5 spectra were acquired: (A) average spectra of all the colony averages per retrospectively identified category [i.e., control (n = 6 colonies), B[a]P-treated (n = 17), spontaneous transformed (n = 2) and B[a]P-treated transformed n = 3)]; (B) resultant loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ); and (C) PCA–LDA scores plot, classed by category i.e., control (untransformed; yellow diamond), B[a]P-treated (untransformed; blue diamond), control (transformed; closed red star) or B[a]P-treated (transformed; open red star) (each score corresponds to the average spectrum derived from one colony). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
36
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
5 spectra were acquired (32 co-additions, 8 cm−1 spectral resolution). Fig. 2A shows the average spectra of all the colony averages per retrospectively identified category [i.e., control (n = 6), B[a]P-treated (n = 17), spontaneous transformed (n = 2) and B[a]Ptreated transformed n = 3)]. Visually, there was an apparent shift in the centroid of the Amide I and Amide II peaks, while compared to the average control spectra, the remaining categories showed spectral alterations from <1400 cm−1 ; the most apparent regions coincided with spectral regions around 1225 cm−1 (as PO2 − ) and 1080 cm−1 (s PO2 − ). The loadings plot (Fig. 2B) showed that wavenumbers 1716 cm−1 to 1697 cm−1 (C O stretching), 1620 cm−1 [Amide I right-hand shoulder (RHS)], 1585 cm−1 [C–C stretching (protein)], 1404 cm−1 [C–H bend (protein)] and 1029 cm−1 [C–O–H bend (glycogen)] were the main loadings segregating out the B[a]P-treated transformed, and 1759 cm−1 [C O stretching (lipid)], 1654 cm−1 [C O stretch (Amide I)], 1554 cm−1 [C–N stretch N–H bend (Amide II)], 1288 cm−1 to 1215 cm−1 , 1153 cm−1 [C–O stretch (carbohydrates)], 960 cm−1 [C–O, C–C stretch (nucleic acids/protein phosphorylation)] segregating out the control transformed, whereas 1500 cm−1 (Amide I/II trough), 1090 cm−1 and 1080 cm−1 (s PO2 − ) separated both out equally. In line with the scores plot (Fig. 2C), the B[a]P-treated, but untransformed, colonies were the most biochemically average i.e., intermediate in phenotype between control (untransformed) and transformed (control and B[a]P-treated). Along the linear discriminants LD1 and LD2, clear cluster segregation of the untransformed control colonies, untransformed B[a]P-treated colonies and all five transformed (control and B[a]P-treated) colonies was observed. In Experiment 2, similar spectral characteristics were noted. Again, 5 spectra per unstained colony were acquired and the average derived. Fig. 3 shows the average spectra of all the derived colony averages for the control (untransformed; n = 100), B[a]Ptreated (untransformed; n = 70), control transformed (n = 4) and B[a]P-treated transformed (n = 5) categories. A subtle shift in the centroids of the Amide I and Amide II peaks in the average spectra derived from the latter three categories compared to the untransformed control spectrum is noted; however, the majority of the readily identifiable differences are noticeable <1400 cm−1 and, again spectral regions associated with 1225 cm−1 and 1080 cm−1 appeared as distinguishing characteristics. Fig. 4 shows the resultant scores plot along linear discriminants LD1 and LD3; in this much larger sample size, there was clear (despite some overlap) segregation between the control
Fig. 3. Average spectra of all the colony averages per retrospectively identified category [i.e., control (n = 100 colonies), B[a]P-treated (n = 70), spontaneous transformed (n = 4) and B[a]P-treated transformed n = 5)] derived using ATR-FTIR spectroscopy (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells in the presence or absence of 5 g/ml B[a]P treatment. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were fixed with methanol and per colony, 5 spectra were acquired.
Fig. 4. PCA–LDA scores plot classed by category i.e., control (untransformed; yellow diamond), B[a]P-treated (untransformed; blue diamond), control (transformed; closed red star) or B[a]P-treated (transformed; open red star) (each score corresponds to the average spectrum derived from one colony) (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells in the presence or absence of 5 g/ml B[a]P treatment. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were fixed with methanol and per colony, 5 spectra were acquired. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
(untransformed) and the B[a]P-treated (untransformed) phenotypes. Perhaps because of the larger dataset, the transformed (control and B[a]P-treated) scores did not segregate away from the untransformed phenotypes; however, they (except for one control transformed colony) were clustered towards the B[a]Ptreated (untransformed). The PCA–LDA loadings plot (Fig. 5) identified the following wavenumbers as segregating out the control transformed category from all four categories (control, B[a]P-treated, control transformed and B[a]P-treated transformed): these were 1762 cm−1 (also B[a]P-treated transformed), 1697 cm−1 (C O stretching) (most pronounced), 1681 cm−1 (Amide I LHS) (also B[a]P-treated transformed), 1616 cm−1 (Amide I/II trough) (also B[a]P-treated transformed) and 1504 cm−1 . Of interest was the observation that <1400 cm−1 (the DNA/RNA region), no prominent distinguishing weightings were noted for spontaneous transformed colonies; additionally, 1543 cm−1 [C–N stretch N–H bend (Amide II)] was observed as a major contributor to the segregation of B[a]P-treated transformed colonies. For B[a]P-treated transformed
Fig. 5. Loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ) in order to identify the wavenumbers segregating out one category from all four categories (i.e., control, B[a]P-treated, spontaneous transformed and B[a]P-treated transformed). The y-scale is a measure of how much variation each wavenumber contributes to the relevant cluster vector (Walsh et al., 2007b).
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
Fig. 6. Comparison of transformed phenotype with its corresponding nontransformed category (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were identified and per colony, 5 spectra were acquired: (A) loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ) in order to identify the wavenumbers segregating out the spontaneous transformed category; (B) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony); (C) loadings plot following PCA–LDA derived in order to identify the wavenumbers segregating out the B[a]P-treated transformed category; (D) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony).
colonies the following additional wavenumbers were noted as contributors to segregation: 1415 cm−1 (in-plane ring vibrations of NH and CH deformation modes), 1396 cm−1 [C–H stretching (protein)], 1338 cm−1 , 1315 cm−1 (Amide III), 1238 cm−1 [C–N stretch N–H bend (Amide III)], 1222 cm−1 (as PO2 − ), 1176 cm−1 , 1153 cm−1 [C–O stretching (carbohydrates)], 1137 cm−1 , 1118 cm−1 [C–O vibrations (RNA)], 1053 cm−1 [C–O vibrations (DNA/lipids)] and 1006 cm−1 (the most pronounced contributor to variance in this region). PCA–LDA of control untransformed vs. transformed colonies identified that the following wavenumbers contributed to segregation of these two categories: 1697 cm−1 [C O stretching (RNA)], 1624 cm−1 (Amide I RHS), 1589 cm−1 and 1583 cm−1 (Amide I/II trough), 1546 cm−1 [C–N stretch N–H bend (Amide II)] and 1500 cm−1 (Amide II RHS); there followed a series of smaller loadings between 1076 cm−1 and 1010 cm−1 (Fig. 6A). In Fig. 6B, the scores for transformed colonies were offset from those that were untransformed. In such scores plots, closeness implies similarity and distance dissimilarity; thus the control transformed colonies clearly clustered together and the vast majority of the untransformed colonies segregated away. However, there were a small number (n = 7; out of 100) of such latter colonies that seemed to co-cluster with the transformation phenotype. In Fig. 6C and B[a]P-treated transformed colonies were distinguished from B[a]P-treated untransformed colonies by 1627 cm−1 (Amide I RHS), 1558 cm−1 [C–N stretch N–H bend (Amide II)], 1543 cm−1 [C–N stretch N–H bend (Amide II)], 1516 cm−1 (Amide II RHS), 1504 cm−1 , 1419 cm−1 [in-plane ring vibrations of NH and CH deformation modes (protein)], 1353 cm−1 , 1234 cm−1 (as PO2 − ), 1215 cm−1 , 1188 cm−1 , 1168 cm−1 [as CO–O–C (lipids)], 1134 cm−1 , 1076 cm−1 (s PO2 − ), 1006 cm−1 and 929 cm−1 . In addition, there was marked clustering of the untransformed vs. transformed phenotypes (Fig. 6D) with the majority of the B[a]P-treated colonies segregated together. However, there was also co-clustering to transformed colonies with a large number of untransformed phenotypes (n = 23; out of 70). When the untransformed control and B[a]P-treated colonies were compared using PCA–LDA, the following distinguish-
37
Fig. 7. Effect of B[a]P treatment on phenotype (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were identified and per colony, 5 spectra were acquired: (A) loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ) in order to identify the wavenumbers segregating out the B[a]P-treated category; (B) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony); (C) loadings plot following PCA–LDA derived in order to identify the wavenumbers segregating out the B[a]P-treated transformed category; (D) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony).
ing wavenumbers were observed: 1747 cm−1 [C O stretch (lipids)], 1627 cm−1 (Amide I RHS), 1554 cm−1 [C–N stretch N–H bend (Amide II)] and 1543 cm−1 [C–N stretch N–H bend (Amide II)], 1500 cm−1 and 1481 cm−1 (in-plane ring vibrations of NH and CH deformation modes), 1230 cm−1 (as PO2 − ), 1134 cm−1 and 1114 cm−1 [O–H bending (glycoproteins)], and 1037 cm−1 [C–O–H bend (glycogen)] (Fig. 7A). Of note was the observation that there appeared to be two clusters of B[a]P-treated transformed scores; one (n = 26) co-clustered with control (untransformed) colony scores and another (n = 44) that very clearly segregated away (Fig. 7B). Several wavenumbers segregated spontaneous transformed from B[a]P-treated transformed, and these included 1716 cm−1 (C O stretching), 1701 cm−1 (C O stretching), 1654 cm−1 [C O stretch (Amide I)], 1635 cm−1 (Amide I RHS), 1593 cm−1 (Amide I/II trough), 1573 cm−1 (Amide II LHS), 1543 cm−1 [C–N stretch N–H bend (Amide II)], 1512 cm−1 (Amide II RHS), 1492 cm−1 , 1411 cm−1 [in-plane ring vibrations of NH and CH deformation modes (protein)], 1338 cm−1 , 1284 cm−1 , 1269 cm−1 , 1245 cm−1 , 1226 cm−1 (as PO2 − ), 1199 cm−1 , 1184 cm−1 , 1130 cm−1 , 1029 cm−1 [C–O–H bend (glycogen)], 1006 cm−1 , 975 cm−1 [C–O, C–C stretch (DNA)] and 933 cm−1 (Fig. 7C). It was notable that there was a very clear segregation of scores derived from these two phenotypes (Fig. 7D). 4. Discussion In early-passage cultures, the SHE assay can be used as an in vitro transformation assay for the detection of genotoxic and nongenotoxic carcinogens and morphologically transformed colonies form tumours when injected into isologous animals (Maire et al., 2007). As an in vitro assay to predict rodent and human carcinogenesis, the reduced pH SHE assay offers several advantages, such as it encompasses the multi-step process of carcinogenesis, transformation occurs at a high enough rate to allow for statistical evaluation, SHE cells can be cryopreserved and they are metabolically competent so thus may bio-activate pro-carcinogens (LeBoeuf et al., 1996; Mauthe et al., 2001). However, difficulties remain with regards to objectively scoring whether a colony is transformed or not, in that
38
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
it takes a trained eye to detect the morphological changes and a colony might only be part-transformed or might visibly appear untransformed. Different SHE colony types that were subsequently designated by visual examination (vehicle control or B[a]P-treated; transformed or untransformed) were obtained in this study. Following interrogation using ATR-FTIR spectroscopy, derived IR spectra were subjected to PCA–LDA, which reduces the sources of intra-category variance but maximises inter-category differences. These analyses resulted in discrimination between different colony categories in two independent experiments. Our objective was to ascertain whether the application of mid-IR spectroscopy might be a novel approach to further the predictive power of the SHE assay (Engelhardt et al., 2004), and whether multivariate analyses might identify IR spectral-derived biomarkers of a carcinogenic signature. Chemical groupings that appear important for segregation of colony categories, consistently indicated in loadings plots following PCA–LDA, were the Amide I, Amide II and the 1500 cm−1 to 1200 cm−1 region. Characteristic IR spectra were obtained from fixed SHE colonies and these exhibited biomolecular chemical entities similar to other cell types (Hammiche et al., 2005; German et al., 2006a, 2006b; Barber et al., 2006; Bentley et al., 2007). There was a clear underlying effect induced by B[a]P treatment and, this was highlighted by pronounced segregation of untransformed control and B[a]P-treated colony scores. An important question arising from this work is what distinguishes spontaneous transformation from B[a]P-associated changes? Spontaneous transformation in control cultures (vs. control untransformed) was associated with loadings mostly located in 1700 cm−1 to 1500 cm−1 region (i.e., associated with Amide I and Amide II); while there was a marked absence of major loadings in the 1500 cm−1 to 1200 cm−1 region (i.e., the DNA/RNA region). In contrast, B[a]P-treated transformed colonies (vs. B[a]P-treated untransformed) gave rise to loadings in both the 1700 cm−1 to 1500 cm−1 region and the 1500 cm−1 to 1200 cm−1 region. This might suggest that induced molecular alteration in the 1700 cm−1 to 1500 cm−1 region are critical towards morphological transformation (i.e., they occur in both spontaneous and B[a]P-treated) whereas those associated with the 1500 cm−1 to 1200 cm−1 region are more compound-specific. FTIR spectroscopy with subsequent multivariate analysis (PCA–LDA) of SHE colonies could allow for an objective classification approach for morphological transformation. Identifying underlying biomolecular alterations that were signatures of genotoxic or non-genotoxic carcinogens would serve to enhance the predictive power of the SHE cell transformation assay (pH 6.7). This non-destructive approach might conveniently be combined with retrospective staining and conventional visual scoring. A future study will examine a panel of positive and negative control test agents in order to ascertain whether such distinctions might be made and validated. Conflict of interest statement None of the authors have financial interests or any other interests that might be prejudicial to the interpretation or presentation of the work contained within this manuscript.
Acknowledgement This work was funded by Unilever. References Barber, J.L., Walsh, M.J., Hewitt, R., Jones, K.C., Martin, F.L., 2006. Low-dose treatment with polybrominated diphenyl ethers (PBDEs) induce altered characteristics in MCF-7 cells. Mutagenesis 21, 351–360. Bentley, A.J., Nakamura, T., Hammiche, A., Pollock, H.M., Martin, F.L., Kinoshita, S., Fullwood, N.J., 2007. Characterization of human corneal stem cells by synchrotron infrared micro-spectroscopy. Mol. Vis. 22, 237–242. Engelhardt, G., Schwind, K.-R., Muler, B., 2004. The testing of chemicals in the Syrian hamster embryo (SHE) cell transformation assay for assessment of carcinogenic potential. Toxicol. In Vitro 18, 213–218. German, M.J., Hammiche, A., Ragavan, N., Tobin, M.J., Cooper, L.J., Matanhelia, S.S., Hindley, A.C., Nicholson, C.M., Fullwood, N.J., Pollock, H.M., Martin, F.L., 2006a. Infrared spectroscopy with multivariate analysis potentially facilitates the segregation of different types of prostate cell. Biophys. J. 90, 3783–3795. German, M.J., Pollock, H.M., Zhao, B., Tobin, M.J., Hammiche, A., Bentley, A., Cooper, L.J., Martin, F.L., Fullwood, N.J., 2006b. Characterization of putative stem cell populations in the cornea using synchrotron infrared microspectroscopy. Invest. Ophthalmol. Vis. Sci. 47, 2417–2421. Goormaghtigh, E., Ruysschaert, J.M., Raussens, V., 2006. Evaluation of the information content in infrared spectra for protein secondary structure determination. Biophys. J. 90, 2946–2957. Hammiche, A., German, M.J., Hewitt, R., Pollock, H.M., Martin, F.L., 2005. Monitoring cell cycle distributions in MCF-7 cells using near-field photothermal microspectroscopy. Biophys. J. 88, 3699–3706. Isfort, R.J., Cody, D.B., Doerson, C., Kerckaert, G.A., LeBoeuf, R.A., 1994. Alterations in cellular differentiation, mitogenesis, cytoskeleton and growth characteristics during Syrian hamster embryo cell multistep in vitro transformation. Int. J. Cancer 59, 114–125. Kerckaert, G.A., Isfort, R.J., Carr, G.J., Aardema, M.J., LeBoeuf, R.A., 1996. A comprehensive protocol for conducting the Syrian hamster embryo cell transformation assay at pH 6.70. Mutat. Res. 356, 65–84. LeBoeuf, R.A., Kerckaert, G., 1987. Enhanced morphological transformation of early passage Syrian hamster embryo cells cultured in medium with a reduced bicarbonate concentration and pH. Carcinogenesis 8, 680–697. LeBoeuf, R.A., Kerckaert, G.A., Aardema, M.J., Gibson, D.P., 1990a. Multistage neoplastic transformation of Syrian hamster embryo cells cultured at pH 6.70. Cancer Res. 50, 3722–3729. LeBoeuf, R.A., Kerckaert, G.A., Aardema, M.J., Gibson, D.P., Brauninger, R., Isfort, R.J., 1990b. The pH 6.7 Syrian hamster embryo cell transformation assay for assessing the carcinogenic potential of chemicals. Mutat. Res. 356, 85–127. LeBoeuf, R.A., Kerckaert, G.A., Aardema, M.J., Gibson, D.P., Brauninger, R., Isfort, R.J., 1996. The pH 6.7 Syrian hamster embryo cell transformation assay for assessing the carcinogenic potential of chemicals. Mutat. Res. 356, 85–127. Maire, M.A., Rast, C., Landkocz, Y., Vasseur, P., 2007. 2,4-Dichlorophenoxyacetic acid: effects on Syrian hamster embryo (SHE) cell transformation, c-Myc expression. DNA damage and apoptosis. Mutat. Res. 631, 124–136. Mantsch, H.H., Chapman, D., 1996. Infrared Spectroscopy of Biomolecules. John Wiley & Sons, New York. Martin, F.L., German, M.J., Wit, E., Fearn, T., Ragavan, N., Pollock, H.M., 2007. Identifying variables responsible for clustering in discriminant analysis of data from infrared microspectroscopy of a biological sample. J. Comput. Biol. 14, 1176–1184. Mauthe, R.J., Gibson, D.P., Bunch, R.T., Custer, L., 2001. The Syrian hamster embryo (SHE) cell transformation assay: review of the methods and results. Toxicol. Pathol. 29 (Suppl. l), 138–146. Walsh, M.J., German, M.J., Singh, M., Pollock, H.M., Hammiche, A., Kyrgiou, M., Stringfellow, H.F., Paraskevaidis, E., Martin-Hirsch, P.L., Martin, F.L., 2007a. IR microspectroscopy: potential applications in cervical cancer screening. Cancer Lett. 246, 1–11. Walsh, M.J., Singh, M.N., Pollock, H.M., Cooper, L.J., German, M.J., Stringfellow, H.F., Fullwood, N.J., Paraskevaidis, E., Martin-Hirsch, P.L., Martin, F.L., 2007b. ATR microspectroscopy with multivariate analysis segregates grades of exfoliative cervical cytology. Biochem. Biophys. Res. Commun. 352, 213–219. Walsh, M.J., Fellous, T.G., Hammiche, A., Lin, W.-R., Fullwood, N.J., Grude, O., Bahrami, F., Nicholson, J.M., Cotte, M., Susini, J., Pollock, H.M., Brittan, M., Martin-Hirsch, P.L., Alison, M.R., Martin, F.L., 2008. Fourier transform infrared micro-spectroscopy identifies symmetric PO2 − modifications as a marker of the putative stem cell region of human intestinal crypts. Stem Cells 26, 108–118, doi:10.1634/stemcells.2007-0196.
Contents lists available at ScienceDirect
Toxicology journal homepage: www.elsevier.com/locate/toxicol
Discrimination of a transformation phenotype in Syrian golden hamster embryo (SHE) cells using ATR-FTIR spectroscopy Michael J. Walsh a,1 , Shannon W. Bruce b , Kamala Pant b , Paul L. Carmichael c , Andrew D. Scott c , Francis L. Martin a,∗ a b c
Centre for Biophotonics, Lancaster Environment Centre, Lancaster University, Lancaster LA1 4YQ, UK BioReliance Corporation, Medical Center Drive, Rockville, MD, USA Safety and Environmental Assurance Centre, Unilever Colworth Science Park, Bedfordshire, UK
a r t i c l e
i n f o
Article history: Received 4 December 2008 Accepted 5 January 2009 Available online 9 January 2009 Keywords: ATR-FTIR spectroscopy Benzo[a]pyrene Linear discriminant analysis Morphological transformation Principal component analysis Syrian hamster embryo cells
a b s t r a c t Primary Syrian hamster embryo (SHE) cells might be used to assess morphological transformation following treatment with chemical carcinogens. We employed attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy to interrogate SHE colonies, as complex biomolecules absorb in the mid-infrared (IR; = 2–20 m) giving vibrational spectra associated with structure and function. Earlypassage SHE cells were cultured (pH 6.7) in the presence or absence of benzo[a]pyrene (B[a]P; 5.0 g/ml). Unstained colonies were applied to an ATR crystal, and vibrational spectra were obtained in the ATR mode using a Bruker Vector 22 FTIR spectrometer with Helios ATR attachment. These were individually baseline-corrected and normalised. Spectra were then analysed using principal component analysis (PCA) plus linear discriminant analysis (LDA). PCA was used to reduce the dataset dimensions before LDA was employed to reveal clustering. This determined whether wavenumber–absorbance relationships expressed as single points (scores) in ‘hyperspace’ might on the basis of multivariate distance reveal biophysical differences associated with morphologies in vehicle control (non-transformed or transformed) or carcinogen-treated (non-transformed or transformed) cells. Retrospectively designated SHE colonies (following staining and microscopic analysis) clustered according to whether they were vehicle control (non-transformed), B[a]P-treated (non-transformed) or transformed (control and B[a]P-treated). Scores plots pointed to a B[a]P-treated phenotype and derived loadings plots highlighted distinguishing markers in control transformed vs. B[a]P-treated transformed; these were mostly associated with Amide I, Amide II and phosphate stretching (asymmetric and symmetric) vibrations. Combined application of ATR-FTIR spectroscopy and unsupervised (PCA)/supervised (LDA) may be a novel approach to scoring SHE colonies for morphological transformation. © 2009 Elsevier Ireland Ltd. All rights reserved.
1. Introduction Distinguishing non-transformed from transformed foci (colonies) of primary Syrian golden hamster embryo (SHE) cells cultured (at pH 6.7) in the presence or absence of test agent has been used to screen compounds for potential carcinogenicity (LeBoeuf et al., 1996). This has been achieved by microscopic assessment of colony morphology allowing the identification of those colonies exhibiting features of morphological transformation (including, randomly orientated criss-crossed, three-dimensional colony growth characteristics, with colonies composed of basophilic
∗ Corresponding author. Tel.: +44 1525 594505; fax: +44 1524 593192. E-mail address: [email protected] (F.L. Martin). 1 Present address: 5247 Beckman Institute, 405 North Mathews Avenue, Urbana, IL 61801, USA. 0300-483X/$ – see front matter © 2009 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.tox.2009.01.003
stacked cells). However, the process of scoring colonies is subjective and there is a need for an approach that reliably identifies these changes. Furthermore, the SHE assay suffers from a lack of biological understanding of the fundamental mechanism(s) involved in the transformation process. Fourier-transform infrared (FTIR) microspectroscopy can generate an IR ‘biochemical-cell fingerprint’ in cellular systems (German et al., 2006a). For example, a signature of susceptibility-to-adenocarcinoma, according to zonal location in the prostate, has been proposed using IR analysis (German et al., 2006a). Similarly, putative stem cells of the cornea appear to possess identifying IR spectral characteristics (Bentley et al., 2007; German et al., 2006b). IR microspectroscopy may also be applied to monitor cell cycle (Hammiche et al., 2005) or response to toxic insult (Barber et al., 2006). Biomolecules absorb in the mid-IR (2–20 m) giving rise to vibrational spectra. This allows the measurement of a range of cellular components including protein conformation, DNA, RNA, lipids
34
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
and glycogen content; these are often characteristic of cell type and/or pathology (Walsh et al., 2007a). IR absorbance bands include Amide I (1650 cm−1 ), Amide II (1550 cm−1 ), Amide III (1260 cm−1 ), asymmetric phosphate (as PO2 − ; 1225 cm−1 ), symmetric phosphate (s PO2 − ; 1080 cm−1 ) and glycogen (1030 cm−1 ) (Walsh et al., 2007b, 2008). Alterations in absorbance bands (i.e., shape, shifts and/or intensity changes) are indicative of intracellular alterations. The Amide I band is mainly due to the (C O) bond whereas the Amide II peak associated with ␦ (N–H) and (C–N) protein bonds (Goormaghtigh et al., 2006; Mantsch and Chapman, 1996). Attenuated total reflection-FTIR (ATR-FTIR) spectroscopy facilitates the interrogation of specimens with a spatial resolution of ≈250 m × 250 m. This gives a high signal-to-noise ratio (SNR), with fast acquisition of IR spectra (Walsh et al., 2007b). Spectroscopic studies generate large data sets that subsequently require computational data reduction approaches in order to identify variance e.g., principal component analysis (PCA) (Walsh et al., 2007a). With PCA, each spectrum is converted into a single point (or score) in n-hyperspace and groups of scores may be viewed in three-dimensions rotated on principal components (PCs). Scores are spatially separated proportional to spectral similarity. The objective identification of morphologically transformed colonies remains a key determinant in the conduct of the SHE assay and there are no biomarkers that define the underlying progression in response to either DNA-reactive genotoxic or non-genotoxic carcinogens. In this pilot study, we set out to interrogate unstained and unidentified SHE colonies that had been cultured with or without a prior exposure to benzo[a]pyrene (B[a]P). We investigated whether such colonies, subsequently categorised by conventional visual scoring as control (untransformed), B[a]P-treated (untransformed), control (spontaneous transformed) or B[a]P-treated (transformed) colonies, could be segregated following interrogation with ATRFTIR spectroscopy and PCA–LDA. 2. Materials and methods 2.1. Primary Syrian golden hamster embryo (SHE) cell transformation assay (pH 6.7) The clonal transformation assay was carried out using SHE cells following a 7day exposure with or without test compound. The SHE transformation assay was designed to allow the expression of morphologically transformed clonal SHE cells seeded onto feeder cells, after exposure for 7 days to B[a]P (5.0 g/ml) as well as vehicle control (DMSO) treatments in the absence of a supplemental exogenous mammalian metabolic activation system. The final concentration of the vehicle in treatment medium did not exceed 0.2% (v/v). Following treatment, the cells were cultured to determine the cytotoxic effects of treatment and the induction of morphological transformation (Kerckaert et al., 1996). 2.2. Preparation of feeder cells Cryopreserved SHE cells from a tested and approved lot were thawed and grown to 50–90% confluency in growth flasks (2–4 days). On Day 1 of the assay, feeder cells were detached and suspended in culture medium in a growth flask on wet ice. They were then X-ray irradiated (approximately 5000 rad); 5 plates containing only feeder cells were cultured long term to confirm that they were no longer viable or capable of replication. Following irradiation, the cells were re-suspended in culture medium, counted using a haemocytometer and the cell concentration adjusted to 0.67 × 104 cells/ml in culture medium; 6.0 ml of this suspension was placed into each 60-mm culture dish containing a 1 cm × 1 cm low-E reflective glass slide. 2.3. Preparation of target cells and treatment The day after seeding feeder cells, a second vial of SHE cells from the same lot was thawed and seeded in a growth flask for approximately 5 h. After the 5-h incubation period, the target cells were detached, counted with a haemocytometer, and diluted with culture medium to a concentration that yielded approximately 25–45 colonies/dish. Two milliliters of the target cell suspension was placed into each culture dish containing glass slide and feeder cells. Dishes were incubated at 37 ± 1 ◦ C in a humidified atmosphere of 10 ± 0.5% CO2 for approximately 24 h prior to treatment. The pH of the culture medium at the time of preparation was measured and adjusted to 6.7 ± 0.05 prior to use in the assay. The pH of the dosing solutions was
checked after at least 4 h of undisturbed incubation at 37 ± 1 ◦ C with approximately 10 ± 0.5% CO2 in air (LeBoeuf and Kerckaert, 1987). Test article stock solution was prepared by dissolving B[a]P in DMSO (500×) at 2.5 mg/ml. This test solution was diluted 1:250 with culture medium to yield a 2× desired final concentration. Prior to treatment, 4.0 ml medium was removed from each culture dish. Four milliliters 2×B[a]P in 0.4% solvent in culture medium was then added to each treatment dish. Each culture dish in the solvent control group received 4 ml of 0.4% solvent in culture medium. All culture dishes were incubated (undisturbed) at 37 ± 1 ◦ C in a humidified atmosphere of 10 ± 0.5% CO2 for 7 days. All dishes with feeder-cells only were replenished with 6 ml complete medium on the treatment day. 2.4. Scoring colonies for transformation Following the 7-day incubation, the culture medium was removed from each dish. The SHE cell colonies in dishes with glass slides were fixed with methanol. The slides were then air-dried and placed in slide holders and marked for orientation. 2.5. Attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy Unstained, but macroscopically visible, SHE colonies adhered to 1 cm × 1 cm low-E glass microscope slides (Kevley Technologies, Chesterland, OH, USA) were interrogated using ATR-FTIR spectroscopy. IR spectra were acquired using the Bruker Vector 22 FTIR spectrometer with Helios ATR attachment containing a diamond crystal (Bruker Optics Ltd, Coventry, UK). Using a CCTV camera attached to the ATR crystal, 5 random points per colony were interrogated. Data was collected in ATR mode and spectra (8 cm−1 spectral resolution, co-added for 32 scans) were converted into absorbance using Bruker OPUS software. Sodium dodecyl sulphate was used to clean the ATR crystal prior to the first spectral analysis of each colony. Each spectrum had a background absorption automatically subtracted, was baseline-corrected and normalised to the Amide I absorbance peak (1650 cm−1 ) using OPUS software. This normalisation function during spectral pre-processing also removed any influence of cell density or absorbance. Average absorbance spectra were then derived using OPUS software, resulting in one absorbance spectrum per colony. 2.6. Computational analysis Raw spectra were processed employing principal component analysis (PCA) performed using the Pirouette software package (Infometrix Inc., Woodinville, WA, USA). PCA scores plots were derived and various principal components (PCs) were examined to see which gave best segregation. This was conducted without knowledge of colony designation into one of four categories i.e., control (untransformed), B[a]P-treated (untransformed), control (spontaneous transformed) or B[a]P-treated (untransformed). PCA loadings plots were then derived to determine spectral variance on the selected PCs. Using PCA alone and following retrospective designation of interrogated colonies into the aforementioned categories, remarkable clustering was obtained (data not shown). PCA was used for preliminary data reduction and the output was processed using linear discriminant analysis (LDA) (Martin et al., 2007) incorporating our previously-published cluster vector approach (German et al., 2006a). PCA–LDA scores and loadings plots were derived for the biochemicalcell fingerprint region (1850 cm−1 to 900 cm−1 ). A PCA-LDA loadings plot displays the coefficients by which each of the original wavenumber variables must be multiplied to obtain the hyperspace vector passing through the median of a chosen cluster. It thereby picks out those spectral bands that are primarily responsible for the discrimination of those spectra from all the spectra taken as a whole. 2.7. Retrospective scoring Following IR-spectral acquisition, the glass slides were stained with Giemsa and air-dried. Using a stereo microscope, the colonies were scored. Each colony was evaluated and recorded as either normal or morphologically transformed. Normal colonies tend to be in a monolayer and contain cells with an organised, often flowing, pattern of growth with minimal cell criss-crossing, particularly where the cells are at a confluent density (Fig. 1A and B). Morphologically transformed colonies contain cells arrayed in an extensive randomly oriented, three-dimensional, stacked growth pattern, with criss-crossing of cells at the perimeter and in the interior of the colony (Fig. 1C and D). Cells in morphologically transformed colonies are frequently more basophilic than their normal counterparts and have increased nuclear/cytoplasmic ratios (Isfort et al., 1994), irrespective of their phenotypic origin (e.g., epithelial, fibroblast, etc.) (LeBoeuf et al., 1990a, 1990b). Colonies where the cell growth was too sparse to assess morphology or where the majority of the colony was missing were not assessed for morphological transformation.
3. Results Two independent experiments were conducted in which SHE cells were cultured in the presence or absence of 5 g/ml B[a]P. In Experiment 1, unstained colonies were identified and per colony,
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
35
Fig. 1. Photomicrographs of foci (colonies) of cultured primary Syrian golden hamster embryo (SHE) cells (pH 6.7). Each colony was evaluated and recorded as either (A, B) normal or (C, D) morphologically transformed.
Fig. 2. Interrogation using ATR-FTIR spectroscopy of SHE cells cultured in the presence or absence of 5 g/ml B[a]P treatment (Experiment 1). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were identified and per colony, 5 spectra were acquired: (A) average spectra of all the colony averages per retrospectively identified category [i.e., control (n = 6 colonies), B[a]P-treated (n = 17), spontaneous transformed (n = 2) and B[a]P-treated transformed n = 3)]; (B) resultant loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ); and (C) PCA–LDA scores plot, classed by category i.e., control (untransformed; yellow diamond), B[a]P-treated (untransformed; blue diamond), control (transformed; closed red star) or B[a]P-treated (transformed; open red star) (each score corresponds to the average spectrum derived from one colony). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
36
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
5 spectra were acquired (32 co-additions, 8 cm−1 spectral resolution). Fig. 2A shows the average spectra of all the colony averages per retrospectively identified category [i.e., control (n = 6), B[a]P-treated (n = 17), spontaneous transformed (n = 2) and B[a]Ptreated transformed n = 3)]. Visually, there was an apparent shift in the centroid of the Amide I and Amide II peaks, while compared to the average control spectra, the remaining categories showed spectral alterations from <1400 cm−1 ; the most apparent regions coincided with spectral regions around 1225 cm−1 (as PO2 − ) and 1080 cm−1 (s PO2 − ). The loadings plot (Fig. 2B) showed that wavenumbers 1716 cm−1 to 1697 cm−1 (C O stretching), 1620 cm−1 [Amide I right-hand shoulder (RHS)], 1585 cm−1 [C–C stretching (protein)], 1404 cm−1 [C–H bend (protein)] and 1029 cm−1 [C–O–H bend (glycogen)] were the main loadings segregating out the B[a]P-treated transformed, and 1759 cm−1 [C O stretching (lipid)], 1654 cm−1 [C O stretch (Amide I)], 1554 cm−1 [C–N stretch N–H bend (Amide II)], 1288 cm−1 to 1215 cm−1 , 1153 cm−1 [C–O stretch (carbohydrates)], 960 cm−1 [C–O, C–C stretch (nucleic acids/protein phosphorylation)] segregating out the control transformed, whereas 1500 cm−1 (Amide I/II trough), 1090 cm−1 and 1080 cm−1 (s PO2 − ) separated both out equally. In line with the scores plot (Fig. 2C), the B[a]P-treated, but untransformed, colonies were the most biochemically average i.e., intermediate in phenotype between control (untransformed) and transformed (control and B[a]P-treated). Along the linear discriminants LD1 and LD2, clear cluster segregation of the untransformed control colonies, untransformed B[a]P-treated colonies and all five transformed (control and B[a]P-treated) colonies was observed. In Experiment 2, similar spectral characteristics were noted. Again, 5 spectra per unstained colony were acquired and the average derived. Fig. 3 shows the average spectra of all the derived colony averages for the control (untransformed; n = 100), B[a]Ptreated (untransformed; n = 70), control transformed (n = 4) and B[a]P-treated transformed (n = 5) categories. A subtle shift in the centroids of the Amide I and Amide II peaks in the average spectra derived from the latter three categories compared to the untransformed control spectrum is noted; however, the majority of the readily identifiable differences are noticeable <1400 cm−1 and, again spectral regions associated with 1225 cm−1 and 1080 cm−1 appeared as distinguishing characteristics. Fig. 4 shows the resultant scores plot along linear discriminants LD1 and LD3; in this much larger sample size, there was clear (despite some overlap) segregation between the control
Fig. 3. Average spectra of all the colony averages per retrospectively identified category [i.e., control (n = 100 colonies), B[a]P-treated (n = 70), spontaneous transformed (n = 4) and B[a]P-treated transformed n = 5)] derived using ATR-FTIR spectroscopy (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells in the presence or absence of 5 g/ml B[a]P treatment. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were fixed with methanol and per colony, 5 spectra were acquired.
Fig. 4. PCA–LDA scores plot classed by category i.e., control (untransformed; yellow diamond), B[a]P-treated (untransformed; blue diamond), control (transformed; closed red star) or B[a]P-treated (transformed; open red star) (each score corresponds to the average spectrum derived from one colony) (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells in the presence or absence of 5 g/ml B[a]P treatment. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were fixed with methanol and per colony, 5 spectra were acquired. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)
(untransformed) and the B[a]P-treated (untransformed) phenotypes. Perhaps because of the larger dataset, the transformed (control and B[a]P-treated) scores did not segregate away from the untransformed phenotypes; however, they (except for one control transformed colony) were clustered towards the B[a]Ptreated (untransformed). The PCA–LDA loadings plot (Fig. 5) identified the following wavenumbers as segregating out the control transformed category from all four categories (control, B[a]P-treated, control transformed and B[a]P-treated transformed): these were 1762 cm−1 (also B[a]P-treated transformed), 1697 cm−1 (C O stretching) (most pronounced), 1681 cm−1 (Amide I LHS) (also B[a]P-treated transformed), 1616 cm−1 (Amide I/II trough) (also B[a]P-treated transformed) and 1504 cm−1 . Of interest was the observation that <1400 cm−1 (the DNA/RNA region), no prominent distinguishing weightings were noted for spontaneous transformed colonies; additionally, 1543 cm−1 [C–N stretch N–H bend (Amide II)] was observed as a major contributor to the segregation of B[a]P-treated transformed colonies. For B[a]P-treated transformed
Fig. 5. Loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ) in order to identify the wavenumbers segregating out one category from all four categories (i.e., control, B[a]P-treated, spontaneous transformed and B[a]P-treated transformed). The y-scale is a measure of how much variation each wavenumber contributes to the relevant cluster vector (Walsh et al., 2007b).
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
Fig. 6. Comparison of transformed phenotype with its corresponding nontransformed category (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were identified and per colony, 5 spectra were acquired: (A) loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ) in order to identify the wavenumbers segregating out the spontaneous transformed category; (B) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony); (C) loadings plot following PCA–LDA derived in order to identify the wavenumbers segregating out the B[a]P-treated transformed category; (D) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony).
colonies the following additional wavenumbers were noted as contributors to segregation: 1415 cm−1 (in-plane ring vibrations of NH and CH deformation modes), 1396 cm−1 [C–H stretching (protein)], 1338 cm−1 , 1315 cm−1 (Amide III), 1238 cm−1 [C–N stretch N–H bend (Amide III)], 1222 cm−1 (as PO2 − ), 1176 cm−1 , 1153 cm−1 [C–O stretching (carbohydrates)], 1137 cm−1 , 1118 cm−1 [C–O vibrations (RNA)], 1053 cm−1 [C–O vibrations (DNA/lipids)] and 1006 cm−1 (the most pronounced contributor to variance in this region). PCA–LDA of control untransformed vs. transformed colonies identified that the following wavenumbers contributed to segregation of these two categories: 1697 cm−1 [C O stretching (RNA)], 1624 cm−1 (Amide I RHS), 1589 cm−1 and 1583 cm−1 (Amide I/II trough), 1546 cm−1 [C–N stretch N–H bend (Amide II)] and 1500 cm−1 (Amide II RHS); there followed a series of smaller loadings between 1076 cm−1 and 1010 cm−1 (Fig. 6A). In Fig. 6B, the scores for transformed colonies were offset from those that were untransformed. In such scores plots, closeness implies similarity and distance dissimilarity; thus the control transformed colonies clearly clustered together and the vast majority of the untransformed colonies segregated away. However, there were a small number (n = 7; out of 100) of such latter colonies that seemed to co-cluster with the transformation phenotype. In Fig. 6C and B[a]P-treated transformed colonies were distinguished from B[a]P-treated untransformed colonies by 1627 cm−1 (Amide I RHS), 1558 cm−1 [C–N stretch N–H bend (Amide II)], 1543 cm−1 [C–N stretch N–H bend (Amide II)], 1516 cm−1 (Amide II RHS), 1504 cm−1 , 1419 cm−1 [in-plane ring vibrations of NH and CH deformation modes (protein)], 1353 cm−1 , 1234 cm−1 (as PO2 − ), 1215 cm−1 , 1188 cm−1 , 1168 cm−1 [as CO–O–C (lipids)], 1134 cm−1 , 1076 cm−1 (s PO2 − ), 1006 cm−1 and 929 cm−1 . In addition, there was marked clustering of the untransformed vs. transformed phenotypes (Fig. 6D) with the majority of the B[a]P-treated colonies segregated together. However, there was also co-clustering to transformed colonies with a large number of untransformed phenotypes (n = 23; out of 70). When the untransformed control and B[a]P-treated colonies were compared using PCA–LDA, the following distinguish-
37
Fig. 7. Effect of B[a]P treatment on phenotype (Experiment 2). The SHE transformation assay (pH 6.7) was designed to allow the expression of the transformed morphology of clonal SHE cells. Unstained colonies on 1 cm × 1 cm low-E glass microscope slides were identified and per colony, 5 spectra were acquired: (A) loadings plot following PCA–LDA derived for the biochemical-cell fingerprint region (1850 cm−1 to 900 cm−1 ) in order to identify the wavenumbers segregating out the B[a]P-treated category; (B) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony); (C) loadings plot following PCA–LDA derived in order to identify the wavenumbers segregating out the B[a]P-treated transformed category; (D) PCA–LDA scores plot classed by category (each score corresponds to the average spectrum derived from one colony).
ing wavenumbers were observed: 1747 cm−1 [C O stretch (lipids)], 1627 cm−1 (Amide I RHS), 1554 cm−1 [C–N stretch N–H bend (Amide II)] and 1543 cm−1 [C–N stretch N–H bend (Amide II)], 1500 cm−1 and 1481 cm−1 (in-plane ring vibrations of NH and CH deformation modes), 1230 cm−1 (as PO2 − ), 1134 cm−1 and 1114 cm−1 [O–H bending (glycoproteins)], and 1037 cm−1 [C–O–H bend (glycogen)] (Fig. 7A). Of note was the observation that there appeared to be two clusters of B[a]P-treated transformed scores; one (n = 26) co-clustered with control (untransformed) colony scores and another (n = 44) that very clearly segregated away (Fig. 7B). Several wavenumbers segregated spontaneous transformed from B[a]P-treated transformed, and these included 1716 cm−1 (C O stretching), 1701 cm−1 (C O stretching), 1654 cm−1 [C O stretch (Amide I)], 1635 cm−1 (Amide I RHS), 1593 cm−1 (Amide I/II trough), 1573 cm−1 (Amide II LHS), 1543 cm−1 [C–N stretch N–H bend (Amide II)], 1512 cm−1 (Amide II RHS), 1492 cm−1 , 1411 cm−1 [in-plane ring vibrations of NH and CH deformation modes (protein)], 1338 cm−1 , 1284 cm−1 , 1269 cm−1 , 1245 cm−1 , 1226 cm−1 (as PO2 − ), 1199 cm−1 , 1184 cm−1 , 1130 cm−1 , 1029 cm−1 [C–O–H bend (glycogen)], 1006 cm−1 , 975 cm−1 [C–O, C–C stretch (DNA)] and 933 cm−1 (Fig. 7C). It was notable that there was a very clear segregation of scores derived from these two phenotypes (Fig. 7D). 4. Discussion In early-passage cultures, the SHE assay can be used as an in vitro transformation assay for the detection of genotoxic and nongenotoxic carcinogens and morphologically transformed colonies form tumours when injected into isologous animals (Maire et al., 2007). As an in vitro assay to predict rodent and human carcinogenesis, the reduced pH SHE assay offers several advantages, such as it encompasses the multi-step process of carcinogenesis, transformation occurs at a high enough rate to allow for statistical evaluation, SHE cells can be cryopreserved and they are metabolically competent so thus may bio-activate pro-carcinogens (LeBoeuf et al., 1996; Mauthe et al., 2001). However, difficulties remain with regards to objectively scoring whether a colony is transformed or not, in that
38
M.J. Walsh et al. / Toxicology 258 (2009) 33–38
it takes a trained eye to detect the morphological changes and a colony might only be part-transformed or might visibly appear untransformed. Different SHE colony types that were subsequently designated by visual examination (vehicle control or B[a]P-treated; transformed or untransformed) were obtained in this study. Following interrogation using ATR-FTIR spectroscopy, derived IR spectra were subjected to PCA–LDA, which reduces the sources of intra-category variance but maximises inter-category differences. These analyses resulted in discrimination between different colony categories in two independent experiments. Our objective was to ascertain whether the application of mid-IR spectroscopy might be a novel approach to further the predictive power of the SHE assay (Engelhardt et al., 2004), and whether multivariate analyses might identify IR spectral-derived biomarkers of a carcinogenic signature. Chemical groupings that appear important for segregation of colony categories, consistently indicated in loadings plots following PCA–LDA, were the Amide I, Amide II and the 1500 cm−1 to 1200 cm−1 region. Characteristic IR spectra were obtained from fixed SHE colonies and these exhibited biomolecular chemical entities similar to other cell types (Hammiche et al., 2005; German et al., 2006a, 2006b; Barber et al., 2006; Bentley et al., 2007). There was a clear underlying effect induced by B[a]P treatment and, this was highlighted by pronounced segregation of untransformed control and B[a]P-treated colony scores. An important question arising from this work is what distinguishes spontaneous transformation from B[a]P-associated changes? Spontaneous transformation in control cultures (vs. control untransformed) was associated with loadings mostly located in 1700 cm−1 to 1500 cm−1 region (i.e., associated with Amide I and Amide II); while there was a marked absence of major loadings in the 1500 cm−1 to 1200 cm−1 region (i.e., the DNA/RNA region). In contrast, B[a]P-treated transformed colonies (vs. B[a]P-treated untransformed) gave rise to loadings in both the 1700 cm−1 to 1500 cm−1 region and the 1500 cm−1 to 1200 cm−1 region. This might suggest that induced molecular alteration in the 1700 cm−1 to 1500 cm−1 region are critical towards morphological transformation (i.e., they occur in both spontaneous and B[a]P-treated) whereas those associated with the 1500 cm−1 to 1200 cm−1 region are more compound-specific. FTIR spectroscopy with subsequent multivariate analysis (PCA–LDA) of SHE colonies could allow for an objective classification approach for morphological transformation. Identifying underlying biomolecular alterations that were signatures of genotoxic or non-genotoxic carcinogens would serve to enhance the predictive power of the SHE cell transformation assay (pH 6.7). This non-destructive approach might conveniently be combined with retrospective staining and conventional visual scoring. A future study will examine a panel of positive and negative control test agents in order to ascertain whether such distinctions might be made and validated. Conflict of interest statement None of the authors have financial interests or any other interests that might be prejudicial to the interpretation or presentation of the work contained within this manuscript.
Acknowledgement This work was funded by Unilever. References Barber, J.L., Walsh, M.J., Hewitt, R., Jones, K.C., Martin, F.L., 2006. Low-dose treatment with polybrominated diphenyl ethers (PBDEs) induce altered characteristics in MCF-7 cells. Mutagenesis 21, 351–360. Bentley, A.J., Nakamura, T., Hammiche, A., Pollock, H.M., Martin, F.L., Kinoshita, S., Fullwood, N.J., 2007. Characterization of human corneal stem cells by synchrotron infrared micro-spectroscopy. Mol. Vis. 22, 237–242. Engelhardt, G., Schwind, K.-R., Muler, B., 2004. The testing of chemicals in the Syrian hamster embryo (SHE) cell transformation assay for assessment of carcinogenic potential. Toxicol. In Vitro 18, 213–218. German, M.J., Hammiche, A., Ragavan, N., Tobin, M.J., Cooper, L.J., Matanhelia, S.S., Hindley, A.C., Nicholson, C.M., Fullwood, N.J., Pollock, H.M., Martin, F.L., 2006a. Infrared spectroscopy with multivariate analysis potentially facilitates the segregation of different types of prostate cell. Biophys. J. 90, 3783–3795. German, M.J., Pollock, H.M., Zhao, B., Tobin, M.J., Hammiche, A., Bentley, A., Cooper, L.J., Martin, F.L., Fullwood, N.J., 2006b. Characterization of putative stem cell populations in the cornea using synchrotron infrared microspectroscopy. Invest. Ophthalmol. Vis. Sci. 47, 2417–2421. Goormaghtigh, E., Ruysschaert, J.M., Raussens, V., 2006. Evaluation of the information content in infrared spectra for protein secondary structure determination. Biophys. J. 90, 2946–2957. Hammiche, A., German, M.J., Hewitt, R., Pollock, H.M., Martin, F.L., 2005. Monitoring cell cycle distributions in MCF-7 cells using near-field photothermal microspectroscopy. Biophys. J. 88, 3699–3706. Isfort, R.J., Cody, D.B., Doerson, C., Kerckaert, G.A., LeBoeuf, R.A., 1994. Alterations in cellular differentiation, mitogenesis, cytoskeleton and growth characteristics during Syrian hamster embryo cell multistep in vitro transformation. Int. J. Cancer 59, 114–125. Kerckaert, G.A., Isfort, R.J., Carr, G.J., Aardema, M.J., LeBoeuf, R.A., 1996. A comprehensive protocol for conducting the Syrian hamster embryo cell transformation assay at pH 6.70. Mutat. Res. 356, 65–84. LeBoeuf, R.A., Kerckaert, G., 1987. Enhanced morphological transformation of early passage Syrian hamster embryo cells cultured in medium with a reduced bicarbonate concentration and pH. Carcinogenesis 8, 680–697. LeBoeuf, R.A., Kerckaert, G.A., Aardema, M.J., Gibson, D.P., 1990a. Multistage neoplastic transformation of Syrian hamster embryo cells cultured at pH 6.70. Cancer Res. 50, 3722–3729. LeBoeuf, R.A., Kerckaert, G.A., Aardema, M.J., Gibson, D.P., Brauninger, R., Isfort, R.J., 1990b. The pH 6.7 Syrian hamster embryo cell transformation assay for assessing the carcinogenic potential of chemicals. Mutat. Res. 356, 85–127. LeBoeuf, R.A., Kerckaert, G.A., Aardema, M.J., Gibson, D.P., Brauninger, R., Isfort, R.J., 1996. The pH 6.7 Syrian hamster embryo cell transformation assay for assessing the carcinogenic potential of chemicals. Mutat. Res. 356, 85–127. Maire, M.A., Rast, C., Landkocz, Y., Vasseur, P., 2007. 2,4-Dichlorophenoxyacetic acid: effects on Syrian hamster embryo (SHE) cell transformation, c-Myc expression. DNA damage and apoptosis. Mutat. Res. 631, 124–136. Mantsch, H.H., Chapman, D., 1996. Infrared Spectroscopy of Biomolecules. John Wiley & Sons, New York. Martin, F.L., German, M.J., Wit, E., Fearn, T., Ragavan, N., Pollock, H.M., 2007. Identifying variables responsible for clustering in discriminant analysis of data from infrared microspectroscopy of a biological sample. J. Comput. Biol. 14, 1176–1184. Mauthe, R.J., Gibson, D.P., Bunch, R.T., Custer, L., 2001. The Syrian hamster embryo (SHE) cell transformation assay: review of the methods and results. Toxicol. Pathol. 29 (Suppl. l), 138–146. Walsh, M.J., German, M.J., Singh, M., Pollock, H.M., Hammiche, A., Kyrgiou, M., Stringfellow, H.F., Paraskevaidis, E., Martin-Hirsch, P.L., Martin, F.L., 2007a. IR microspectroscopy: potential applications in cervical cancer screening. Cancer Lett. 246, 1–11. Walsh, M.J., Singh, M.N., Pollock, H.M., Cooper, L.J., German, M.J., Stringfellow, H.F., Fullwood, N.J., Paraskevaidis, E., Martin-Hirsch, P.L., Martin, F.L., 2007b. ATR microspectroscopy with multivariate analysis segregates grades of exfoliative cervical cytology. Biochem. Biophys. Res. Commun. 352, 213–219. Walsh, M.J., Fellous, T.G., Hammiche, A., Lin, W.-R., Fullwood, N.J., Grude, O., Bahrami, F., Nicholson, J.M., Cotte, M., Susini, J., Pollock, H.M., Brittan, M., Martin-Hirsch, P.L., Alison, M.R., Martin, F.L., 2008. Fourier transform infrared micro-spectroscopy identifies symmetric PO2 − modifications as a marker of the putative stem cell region of human intestinal crypts. Stem Cells 26, 108–118, doi:10.1634/stemcells.2007-0196.