- Email: [email protected]
Measurement of cell death by oxidative stress in three-dimensional spheroids from trophoblast and in fragments of decidua tissue
Journal of Reproductive Immunology 85 (2010) 63–70
Contents lists available at ScienceDirect
Journal of Reproductive Immunology journal homepage: www.elsevier.com/locate/jreprimm
Measurement of cell death by oxidative stress in three-dimensional spheroids from trophoblast and in fragments of decidua tissue Regine-Susanne Theuerkauf a , Helmut Ahammer b , Monika Siwetz a , Christine Helige a , Gottfried Dohr a , Wolfgang Walcher c , José Ramón Palacio d , Paz Martinez d , Peter Sedlmayr a,∗ a b c d
Institut für Zellbiologie, Histologie und Embryologie, Medizinische Universität Graz, Harrachgasse 21, A-8010 Graz, Austria Insitut für Biophysik, Medizinische Universität Graz, A-8010 Graz, Austria Universitätsklinik für Frauenheilkunde und Geburtshilfe, Medizinische Universität Graz, A-8036 Graz, Austria Instituto de Biotecnologia y Biomedicina, Universidad Autónoma de Barcelona, E-08193 Bellaterra, Spain
a r t i c l e
i n f o
Article history: Received 30 October 2009 Received in revised form 15 January 2010 Accepted 26 January 2010 Keywords: Multicellular spheroids Hydrogen peroxide Viability Morphometry Embryo implantation
a b s t r a c t We report a new morphometric method for measurement of the amount of cell death in three-dimensional multicellular spheroids of the trophoblast-like cell line AC1-M59 and of cultured pieces of decidua tissue (decidua spheroids) in response to a cytotoxic agent. The viability of the spheroids was assessed by adding propidium iodide to the culture medium at the end of the toxic treatment. On fluorescence and brightfield images of serial cryosections the areas of propidium iodide fluorescence and the entire corresponding spheroids were measured by applying digital image processing and ratiometrical quantification. As an example, we evaluated the cytotoxic effect of hydrogen peroxide on both types of spheroids. The relative potency of hydrogen peroxide to induce tissue damage was assessed quantitatively for determination of the minimal concentration that leads to an increase in cytotoxicity. The method presented suggests general applicability for in vitro determination of toxicity against tissues. © 2010 Elsevier Ireland Ltd. All rights reserved.
1. Introduction There is some evidence of an impact of oxidative environment on the local immune system (Badia et al., 2008; Khadaroo et al., 2003). For studies of the immunomodulatory effect of oxidative stress on cells or tissues in culture it may be relevant not to exceed a limit of stress beyond which cell death is induced. Assessment of drug toxicity involves use of huge numbers of laboratory animals. Reduction of the number of animals to be sacrificed for this purpose is warranted for both ethical and economical reasons. For this purpose in vitro models have been developed. Monolayer
∗ Corresponding author. Tel.: +43 316 380 4234; fax: +43 316 380 9625. E-mail address: [email protected] (P. Sedlmayr). 0165-0378/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.jri.2010.01.004
cultures of cell lines have shortcomings in mimicking the native microenvironment. In contrast, cells grown under appropriate culture conditions can aggregate and form spheroids, multicellular three-dimensional compounds that display properties which are in many ways similar to intact tissues (reviewed in Mueller-Klieser, 1997; Verma et al., 2007). Cells within spheroids are exposed to non-uniform exposure to physical and chemical stress. Therefore, three-dimensional (3D) organized multicellular spheroid simulate more closely reaction of tissue to stress than cells grown in monolayer (Indovina et al., 2007). There is evidence that the microenvironment of the 3D culture accounts for the elevated level of resistance against drugs and toxic chemicals rather than inaccessibility to nutrients (Dhiman et al., 2005). The protein expression patterns may vary extensively between monolayer and spheroid cultures of the same cell line (Kumar et
64
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
al., 2008). The numerous applications of spheroids include their use as in vitro models for the study of drug toxicity or of cytotoxic response to irradiation (Dhiman et al., 2005; Hoffman, 1991; Lambert et al., 2006). Whereas determination of cell death in two-dimensional cultures or suspension cultures is comparatively easy and well established, the quantification of damage conveyed to a cellular compound or a piece of tissue is more difficult to assess accurately. Frequently used methods can be classified into measurement of factors released from viable cells, such as albumin or urea in the case of spheroids from liver cell lines (Verma et al., 2007), assessment of mitochondrial function using the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide) assay (Mosmann, 1983; Verma et al., 2007) or release of factors from dead cells, such as lactate dehydrogenase. Another approach uses assessment of tissue morphology following staining with hematoxylin and eosin, paying attention to organization of the aggregate, mitotic activity, and presence of necrosis. This late toxicity can be assessed using a quantitative outgrowth assay (Lambert et al., 2006). Most of the assays in use bear the handicap that the outcome depends not only on the toxicity but also on the volume of the spheroids, so these methods prove insufficiently sensitive for accurate assessment of the limits of cytotoxic activity of agents against spheroids or pieces of primary tissue with unequal volume. On the other hand, determination of toxicity against single cells after disaggregation of the tissue necessarily entails cell damage induced by the procedure of disaggregation. Moreover, none of these methods allows visualisation of the specific distribution of vital and damaged cells within intact spheroids/tissues. In order to overcome these limitations, we developed a novel assay for determination of tissue toxicity. This assay is based on automated morphometric analysis of sections of spheroids or also pieces of tissue which have been treated in culture with propidium iodide (PI) at the end of the exposure to the cytotoxic agent. In order to test the applicability of this method we used oxidative stress by hydrogen peroxide applied on spheroids of the choriocarcinoma/trophoblast hybrid cell line AC1-M59 and on cultured pieces of decidua tissue. Thereby, we determined the maximal concentration of hydrogen peroxide for a given period of exposure which does not lead to tissue damage in terms of cell death.
Frank, Aplagen GmbH, Baesweiler, Germany and was maintained in a 1:1 mixture of Dulbecco’s Modified Eagle Medium (DMEM) and Ham’s F12 culture media devoid of phenol red (Gibco BRL/Invitrogen, Vienna, Austria), containing 10% fetal calf serum, 2% l-glutamine, 1% penicillin/streptomycin, and 1% amphotericin B (PAA, Pasching, Austria). Every 10th -15th passage, unfused Jeg-3 cells were removed by challenging the culture with 100 M hypoxanthine and 5.7 M azaserine (Sigma–Aldrich, Vienna, Austria). Exponentially growing monolayers were gently detached by accutase (PAA) after removal of the medium and washing with Hanks’ balanced salt solution (HBSS; PAA). The resulting single cell suspension was cultivated in Petri dishes with non-adherent surfaces (Greiner Bio One, Kremsmünster, Austria) under permanent gentle shaking (Belly Button, Stovall Life Science Inc., Kehl, Germany) in a humidified CO2 incubator (5% CO2 , Heraeus Cytoperm 2, Thermo Electron LED GmbH, Vienna, Austria). After 24 h the cell spheroids were transferred into 25-ml glass spinner flasks (Glastechnische Werkstätte GmbH, Vienna, Austria) and stirred at 135 rotations per minute for further 48 h under identical culture conditions. Spheroids with a diameter of 300 m were selected under a dissecting microscope equipped with an ocular grid (Leica MZ6, Wetzlar, Germany). 2.3. Generation of spheroids from decidua tissue
The present study was approved by the Ethics Committee of the Medical University of Graz (No. 16-188 ex 04/05).
Human first trimester decidua parietalis was obtained from elective pregnancy terminations by curettage and vacuum suction between the 6th and 11th week of gestation at the Department of Obstetrics and Gynaecology, Medical University of Graz. Samples were checked for absence of invasive extravillous trophoblast cells by means of immunohistochemical staining for HLAG. Spheroid formation was accomplished by using the protocol of Grümmer et al. (1990). The tissue was cut into pieces of about 800 m under a dissecting microscope equipped with an ocular grid (Leica MZ6, Wetzlar, Germany), transferred to 25 ml glass spinner flasks and stirred at 135 rpm in a humidified incubator at 5% CO2 . The tissue explants were maintained in DMEM culture medium devoid of phenol red containing 10% FCS, 2% l-glutamine, 1% penicillin/streptomycin, 1% amphotericin B, 20 ng/ml progesterone, 10 ng/ml gestonoroncapronate (19-nor-17a-hydroxy-progesterone, a progesterone analogon and 300 pg/ml 17-beta-estradiol (kindly provided by Dr. Halfbrodt, Schering, Berlin, Germany). In addition to progesterone, gestonoroncapronate serving as a depot compound was included in the culture medium in order to reduce fluctuations of the hormone concentration. The medium was changed daily and the decidua spheroids were kept in culture for 3 days.
2.2. Generation of multicellular trophoblast spheroids
2.4. Toxic challenging of spheroids
We utilized a modified version of a previously published protocol (Helige et al., 2008). Briefly, the human trophoblast/Jeg-3 choriocarcinoma hybrid cell line AC1M59 (Schmitz, 2002) was generously provided by H-G.
Hydrogen peroxide (H2 O2 , Sigma–Aldrich, Vienna, Austria) was added to the culture medium of the spheroids for 48 h under continuous stirring at concentrations ranging from 0.05 mM to 1 mM. As controls, spheroids were either
2. Materials and methods 2.1. Ethics
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
65
Fig. 1. Brightfield and corresponding fluorescence images of cryosections of propidium iodide loaded AC1-M59 trophoblast-like cell spheroids. Original magnification 10×.
66
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
kept untreated or were killed by transfer into 50% ethanol (diluted in culture medium) for 30 min. 2.5. Propidium iodide loading and fluorescence microscopy For the purpose of viability testing the spheroids were washed with culture media devoid of phenol red and FCS. Subsequently, 50 g/ml propidium iodide (PI; Sigma–Aldrich, Vienna, Austria) was added for 15 min. After washing with HBSS, the spheroids were embedded in tissue freezing medium (OCT-Sakura, Sanova Pharma, Vienna, Austria) and frozen at −80 ◦ C. Serial cryosections were produced (thickness: 6 m, average distance between the sections evaluated: 18 m) and photomicrographs of the fluorescence and the brightfield image were taken on a fluorescence microscope (AxioPhot, Zeiss, Vienna, Austria) equipped with an HBO lamp (Osram, Zeiss, Vienna), a filter suitable for tetramethyl rhodamine isothiocyanate (TRITC), a 10× objective and a digital camera (AxioCam HRc, Zeiss, Vienna, Austria). All experiments were done thrice, and in each experiment three representative spheroids were analysed for each condition. 2.6. Image processing Although optical assessment of the images already allowed for rough estimation of the effects of H2 O2 stress on multicellular spheroids (Fig. 1), quantification of the results is desirable in order to distinguish faint differences and compare different samples and conditions. In order to develop a quantitative measure for toxicity several segmentation and image operations were carried out. 2.6.1. Segmentation of the propidium iodide fluorescence area Af The digital RGB color images of the fluorescent cells were segmented using k-means clustering with k = 2 in the RGB color space yielding red fluorescent clusters and an almost black background for each individual image (Fig. 2a). The digital RGB color transmission images represented the corresponding areal picture of the spheroid. The bi-colored images of red PI in the nuclei of dead cells against a black background were transformed into a binary form by setting the background cluster center to black and the fluorescence color center to white (Fig. 2b). 2.6.2. Segmentation of the total spheroid area At The segmentation of the spheroids was complicated because simple grey value thresholding or k-means clustering was not applicable. Relatively large areas inside the spheroid showed quite the same grey values as the background. Therefore, a more sophisticated algorithm was developed. As the colors of these images did not include any useful information, the images were transferred to gray value images (Fig. 3a). The variation of the overall image intensity of the entire image stack was minimized by applying the Lee operator with a 5 × 5 kernel yielding slightly low pass filtered and normalized images (Fig. 3b).
Fig. 2. Segmentation of a fluorescence image of a section of a propidium iodide loaded AC1-M59 cell spheroid. K-mean clustering with k = 2 was carried out in the RGB color space. The evolved cluster centers were set to black and white yielding a binary representation of the image. (a) Original color fluorescence image. (b) Segmented binary image, representing the total fluorescence area Af .
Subsequently, the images were converted and subjected to thresholding (Fig. 3c). In parallel, rough polygons were drawn interactively over the contours of spheroids in the Lee-filtered image (Fig. 3d) in order to exclude areas with increased dirt in the background from subsequent segmentation steps. The resulting regions of interest (ROI) were filled with white pixels serving as binary masks (Fig. 3e) for the Lee filtered and inverted images. All pixels of these filtered images which corresponded to black pixels in the masks were set to black (Fig. 3f) and a fixed threshold was applied to the overall stack of images, yielding binary representations of the spheroids (Fig. 3g). Large holes inside the spheroid areas were filled by a tenfold morphological closing (Fig. 3h) and binary hole filling (Fig. 3i) of all remaining holes smaller than 20% of the total image size, thereby preserving the filling of spheroid areas without erroneously filling the background of the images. Fig. 4 exemplarily shows the result by an overlay of the binary image over the original transmission image.
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
67
Fig. 3. Segmentation of a transmission image of a section of a AC1-M59 cell spheroid. The original color image was converted to a gray value image (a), Lee filtered (b) and inverted (c). A region of interest (d) and a mask image (e) were produced. The masked image (f) was thresholded (g), tenfold morphologically closed (h) and finally a hole filling of holes less then 20% of the total image size was applied (i).
68
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
Fig. 4. Overlay of the binary result over the original image. This overlay shows the result of a segmentation of a section of a trophoblast-like AC1-M59 cell spheroid. Despite of the relatively large areas in between the cells, it was possible to segment the compact spheroid area. The binary image was used to calculate the total spheroid area At .
2.6.3. Ratiometric quantification The binary images of both the fluorescence cells and the corresponding entire area of the spheroids were measured by counting white image pixels in the binary image representations. The total number of white image pixels in an image gave mathematically the total area of the fluorescence and the transmission images respectively. The quantitative toxicity factor S was calculated as the ratio of the total area of the fluorescent cells to the corresponding total area of the spheroids: S=
Af At
S:0≤S≤1
with Af the area of the fluorescent cells and At the corresponding total area of the two-dimensional disk representations of the spheroids. As S is dimensionless, it is equivalent if this calculation would be carried out in units of pixels or in the real physical units. S can theoretically show values in between zero and one, dependent on the tissue-specific quantitative relationship of nuclei on one hand and cytoplasm plus extracellular space on the other hand. The value zero means that there is really zero fluorescence in the image and the value one has the meaning that the total area of the spheroid is fluorescent. Therefore, the values around zero and also around one represent very extreme limits for S. 2.7. Measurement of lactate dehydrogenase (LDH) in the culture supernatant This was done for the purpose of comparison to a conventional methodology of determination of cell death. AC1-M59 cell spheroids were treated with H2 O2 as described above, after a period of 48 h culture supernatant was collected. The positive control for spheroid killing was treated with a final concentration of 1% Triton (Merck, Vienna, Austria) for 1 h at the end of the culture period.
Fig. 5. Diagrams showing the quantitative toxicity factor. The quantitative toxicity factor S for multicellular spheroids from the trophoblast-like cell line AC1-M59 (a) or spheroids from decidua tissue after 3 d in spinning culture (b) treated with different concentrations of hydrogen peroxide or with 50% ethanol as a positive control for killing of all cells is given as area of propidium iodide fluorescence per total area of the respective sections. Values are given as medians plus interquartile ranges. Asterisks indicate a p ≤ 0.001 in comparison to the untreated spheroids.
The culture supernatants were kept at −70 ◦ C until determination of LDH activity, which was done in 96-well plates using the LDH Cytotoxicity Detection Kit (Takara Bio Inc, Shiga, Japan) according to the manufacurer’s instructions. The absorbance of the samples was measured in duplicates at 492 nm using a photometer (anthos Reader 2010, Anthos, Salzburg, Austria), the reference wavelength was 620 nm. A total of three experiments were performed. 2.8. Statistical analysis As the data do not follow Gaussian distribution (as assessed by Komolgorov–Smirnov testing), we applied nonparametric two-tailed Mann–Whitney U-test for analysis of significance of differences in S between untreated and treated samples. The SPSS 16 software package was used for this purpose. 3. Results The viability of the spheroids was assessed by exclusion of PI from viable cells which in contrast rapidly labels
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
Fig. 6. Lactate dehydrogenase in the supernatant of cell spheroids. Alternative methodology for assessment of the amount of toxicity of hydrogen peroxide against AC1-M59 trophoblast-like cell spheroids. 1% triton is used as a positive control for killing. The graph indicates the medians and the ranges of three experiments.
the nuclei of dead cells (Fig. 1). At concentrations as low as 0.2 mM H2 O2 the amount of cell death as measured by the ratio of the total area of the fluorescent cells and the corresponding total area of the section of the trophoblast cell spheroids was not increased compared to the negative control. From 0.3 to 0.5 mM H2 O2 lethality rapidly increased in the trophoblast cell spheroids whereas at the same time the overall number, as well as the size of the surviving spheroids decreased (Fig. 5a). Due to complete decomposition of the trophoblast spheroids toxicity analysis was not applicable at higher concentrations of H2 O2 . However, in the positive control (treated with 50% ethanol) PI fluorescence was apparent throughout the whole the multicellular spheroids without obvious decomposition. Thus, the nature of both the cell spheroids and the toxic agent may determine the applicability of this method. Spheroids from decidua tissue were considerably more resistant to oxidative stress, at 0.5 mM H2 O2 no increase of lethality was observed (Fig. 5b). For comparison, the outcome of measurement of lactate dehydrogenase in the supernatant of AC1-M59 cell spheroids stressed with H2 O2 is shown in Fig. 6. 4. Discussion In the present study we report a novel method for quantification of the amount of cell death in multicellular three-dimensional spheroids by quantitative assessment of the relative potency of different concentrations of a toxic agent to induce tissue damage. As a model for toxic damage we used hydrogen peroxide added on spheroids made from a choriocarcinoma/trophoblast hybrid cell line and of spheroids produced from pieces of decidua tissue. We found that decidua spheroids are considerably more resistant to oxidative stress than trophoblast spheroids. The conventional methodology of measurement of lactate dehydrogenase in the culture supernatant was comparatively inefficient in determining toxicity against trophoblast-like cell spheroids stressed with H2 O2 .
69
Reactive oxygen species are a highly reactive group of oxygen-containing molecules, leading to oxidative stress (reviewed in Sies, 1997). Oxidative stress is involved in the aetiology of defective embryo development (Guerin et al., 2001). Damage induced by oxidative stress has been hypothesized to play a role in spontaneous abortion, idiopathic recurrent pregnancy loss, defective embryogenesis, and drug-induced teratogenicity (Iborra et al., 2005). Plasma membrane integrity which is lost during necrosis but also in the late phase of apoptosis was probed by the exclusion of PI (Buenz et al., 2007). In order to allow a numerical assessment and comparison of the relative potency of toxic agents and sensitivity of spheroids of diverse origins we optimized a quantitative measurement of toxicity based on processing of fluorescent images of PI loaded cells. Our goal was to determine the maximal concentration of hydrogen peroxide which induces tissue stress without damage in terms of an increased fraction of dead cells within multicellular spheroids. An increase of dead cells with nuclear staining by PI across the whole section correlates well with the concentration of the toxic agent applied. We show our method to be extendable to analysis of toxicity towards cultured pieces of primary tissue. Whereas the method allows a very accurate determination of toxicity, it is rather time-consuming in comparison to conventional methods for determining vitality or mortification, Furthermore, it requires a fluorescence microscope. A limiting factor for its applicability is the diameter of the spheroids, as PI needs to penetrate sufficiently in order to stain dead cells also in the center of the spheroids. This is very probably dependent on the type of cell line used and on whether or not a matrix is applied as a scaffold for spheroid formation. Moreover, it is reported that most of the proliferating cells within a spheroid are located in a viable rim, whereas quiescent cells are located centrally and cells at a depth exceeding 160 m frequently become necrotic (Lambert et al., 2006). In the present study, however, the trophoblast spheroids were used at a uniform diameter of 300 m and decidua spheroids were used at 800 m × 300 m. Both types did not show necrotic centers and PI could completely penetrate the whole aggregate. The method we propose allows for analysis of spatial distribution of necrotic cells. Using the toxic effect of oxidative stress on spheroids of a choriocarcinoma/trophoblast hybrid cell line as a model system, it proved to be reproducible and economically reasonable. Further technical advancement may ease parallel handling of numerous probes and thus enable application for screening in toxicological studies and drug screening. Acknowledgements We thank Ms. Nina Flieser and Mr. Rudolf Schmied of the Institute of Cell Biology, Histology and Embryology and Ms. Renate Michlmaier of the Department of Gynaecology and Obstetrics, Medical University of Graz, Austria, for their technical assistance. We are grateful to Dr. Erwin Tafeit (Institute of Physiological Chemistry, Medical University of Graz) for advice on statistical analysis. This study
70
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
was supported by the European Commission (Network of Excellence “The Control of Embryo Implantation (EMBIC)”, FP6-512040, lead researchers involved: P.M., P.S.). References Badia, R., Iborra, A., Palacio, J.R., Antich, M., Martinez, P., 2008. The effect of oxidative environment on immunosuppressive properties of human seminal plasma. Am. J. Reprod. Immunol. 60, 354–360. Buenz, E.J., Limburg, P.J., Howe, C.L., 2007. A high-throughput 3-parameter flow cytometry-based cell death assay. Cytometry A 71, 170–173. Dhiman, H.K., Ray, A.R., Panda, A.K., 2005. Three-dimensional chitosan scaffold-based MCF-7 cell culture for the determination of the cytotoxicity of tamoxifen. Biomaterials 26, 979–986. Grümmer, R., Hohn, H.-P., Denker, H.-W., 1990. Choriocarcinoma cell spheroids: an in vitro model for the human trophoblast. Trophoblast Res. 4, 97–111. Guerin, P., El Mouatassim, S., Menezo, Y., 2001. Oxidative stress and protection against reactive oxygen species in the pre-implantation embryo and its surroundings. Hum. Reprod. Update 7, 175–189. Helige, C., Ahammer, H., Hammer, A., Huppertz, B., Frank, H.G., Dohr, G., 2008. Trophoblastic invasion in vitro and in vivo: similarities and differences. Hum. Reprod. 23, 2282–2291. Hoffman, R.M., 1991. Three-dimensional histoculture: origins and applications in cancer research. Cancer Cells 3, 86–92. Iborra, A., Palacio, J.R., Martinez, P., 2005. Oxidative stress and autoimmune response in the infertile woman. Chem. Immunol. Allergy 88, 150–162.
Indovina, P., Rainaldi, G., Santini, M.T., 2007. Three-dimensional cell organization leads to a different type of ionizing radiation-induced cell death: MG-63 monolayer cells undergo mitotic catastrophe while spheroids die of apoptosis. Int. J. Oncol. 31, 1473–1483. Khadaroo, R.G., Kapus, A., Powers, K.A., Cybulsky, M.I., Marshall, J.C., Rotstein, O.D., 2003. Oxidative stress reprograms lipopolysaccharide signaling via Src kinase-dependent pathway in RAW 264.7 macrophage cell line. J. Biol. Chem. 278, 47834–47841. Kumar, H.R., Zhong, X., Hoelz, D.J., Rescorla, F.J., Hickey, R.J., Malkas, L.H., Sandoval, J.A., 2008. Three-dimensional neuroblastoma cell culture: proteomic analysis between monolayer and multicellular tumor spheroids. Pediat. Surg. Int. 24, 1229–1234. Lambert, B., De Ridder, L., De Vos, F., Slegers, G., de Gelder, V., Van de Wiele, C., Thierens, H., 2006. Assessment of supra-additive effects of cytotoxic drugs and low dose rate irradiation in an in vitro model for hepatocellular carcinoma. Can. J. Physiol. Pharmacol. 84, 1021–1028. Mosmann, T., 1983. Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J. Immunol. Methods 65, 55–63. Mueller-Klieser, W., 1997. Three-dimensional cell cultures: from molecular mechanisms to clinical applications. Am. J. Physiol. 273, C1109–C1123. Sies, H., 1997. Oxidative stress: oxidants and antioxidants. Exp. Physiol. 82, 291–295. Verma, P., Verma, V., Ray, P., Ray, A.R., 2007. Formation and characterization of three dimensional human hepatocyte cell line spheroids on chitosan matrix for in vitro tissue engineering applications. In Vitro Cell. Dev. Biol.-Anim. 43, 328–337.
Contents lists available at ScienceDirect
Journal of Reproductive Immunology journal homepage: www.elsevier.com/locate/jreprimm
Measurement of cell death by oxidative stress in three-dimensional spheroids from trophoblast and in fragments of decidua tissue Regine-Susanne Theuerkauf a , Helmut Ahammer b , Monika Siwetz a , Christine Helige a , Gottfried Dohr a , Wolfgang Walcher c , José Ramón Palacio d , Paz Martinez d , Peter Sedlmayr a,∗ a b c d
Institut für Zellbiologie, Histologie und Embryologie, Medizinische Universität Graz, Harrachgasse 21, A-8010 Graz, Austria Insitut für Biophysik, Medizinische Universität Graz, A-8010 Graz, Austria Universitätsklinik für Frauenheilkunde und Geburtshilfe, Medizinische Universität Graz, A-8036 Graz, Austria Instituto de Biotecnologia y Biomedicina, Universidad Autónoma de Barcelona, E-08193 Bellaterra, Spain
a r t i c l e
i n f o
Article history: Received 30 October 2009 Received in revised form 15 January 2010 Accepted 26 January 2010 Keywords: Multicellular spheroids Hydrogen peroxide Viability Morphometry Embryo implantation
a b s t r a c t We report a new morphometric method for measurement of the amount of cell death in three-dimensional multicellular spheroids of the trophoblast-like cell line AC1-M59 and of cultured pieces of decidua tissue (decidua spheroids) in response to a cytotoxic agent. The viability of the spheroids was assessed by adding propidium iodide to the culture medium at the end of the toxic treatment. On fluorescence and brightfield images of serial cryosections the areas of propidium iodide fluorescence and the entire corresponding spheroids were measured by applying digital image processing and ratiometrical quantification. As an example, we evaluated the cytotoxic effect of hydrogen peroxide on both types of spheroids. The relative potency of hydrogen peroxide to induce tissue damage was assessed quantitatively for determination of the minimal concentration that leads to an increase in cytotoxicity. The method presented suggests general applicability for in vitro determination of toxicity against tissues. © 2010 Elsevier Ireland Ltd. All rights reserved.
1. Introduction There is some evidence of an impact of oxidative environment on the local immune system (Badia et al., 2008; Khadaroo et al., 2003). For studies of the immunomodulatory effect of oxidative stress on cells or tissues in culture it may be relevant not to exceed a limit of stress beyond which cell death is induced. Assessment of drug toxicity involves use of huge numbers of laboratory animals. Reduction of the number of animals to be sacrificed for this purpose is warranted for both ethical and economical reasons. For this purpose in vitro models have been developed. Monolayer
∗ Corresponding author. Tel.: +43 316 380 4234; fax: +43 316 380 9625. E-mail address: [email protected] (P. Sedlmayr). 0165-0378/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.jri.2010.01.004
cultures of cell lines have shortcomings in mimicking the native microenvironment. In contrast, cells grown under appropriate culture conditions can aggregate and form spheroids, multicellular three-dimensional compounds that display properties which are in many ways similar to intact tissues (reviewed in Mueller-Klieser, 1997; Verma et al., 2007). Cells within spheroids are exposed to non-uniform exposure to physical and chemical stress. Therefore, three-dimensional (3D) organized multicellular spheroid simulate more closely reaction of tissue to stress than cells grown in monolayer (Indovina et al., 2007). There is evidence that the microenvironment of the 3D culture accounts for the elevated level of resistance against drugs and toxic chemicals rather than inaccessibility to nutrients (Dhiman et al., 2005). The protein expression patterns may vary extensively between monolayer and spheroid cultures of the same cell line (Kumar et
64
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
al., 2008). The numerous applications of spheroids include their use as in vitro models for the study of drug toxicity or of cytotoxic response to irradiation (Dhiman et al., 2005; Hoffman, 1991; Lambert et al., 2006). Whereas determination of cell death in two-dimensional cultures or suspension cultures is comparatively easy and well established, the quantification of damage conveyed to a cellular compound or a piece of tissue is more difficult to assess accurately. Frequently used methods can be classified into measurement of factors released from viable cells, such as albumin or urea in the case of spheroids from liver cell lines (Verma et al., 2007), assessment of mitochondrial function using the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide) assay (Mosmann, 1983; Verma et al., 2007) or release of factors from dead cells, such as lactate dehydrogenase. Another approach uses assessment of tissue morphology following staining with hematoxylin and eosin, paying attention to organization of the aggregate, mitotic activity, and presence of necrosis. This late toxicity can be assessed using a quantitative outgrowth assay (Lambert et al., 2006). Most of the assays in use bear the handicap that the outcome depends not only on the toxicity but also on the volume of the spheroids, so these methods prove insufficiently sensitive for accurate assessment of the limits of cytotoxic activity of agents against spheroids or pieces of primary tissue with unequal volume. On the other hand, determination of toxicity against single cells after disaggregation of the tissue necessarily entails cell damage induced by the procedure of disaggregation. Moreover, none of these methods allows visualisation of the specific distribution of vital and damaged cells within intact spheroids/tissues. In order to overcome these limitations, we developed a novel assay for determination of tissue toxicity. This assay is based on automated morphometric analysis of sections of spheroids or also pieces of tissue which have been treated in culture with propidium iodide (PI) at the end of the exposure to the cytotoxic agent. In order to test the applicability of this method we used oxidative stress by hydrogen peroxide applied on spheroids of the choriocarcinoma/trophoblast hybrid cell line AC1-M59 and on cultured pieces of decidua tissue. Thereby, we determined the maximal concentration of hydrogen peroxide for a given period of exposure which does not lead to tissue damage in terms of cell death.
Frank, Aplagen GmbH, Baesweiler, Germany and was maintained in a 1:1 mixture of Dulbecco’s Modified Eagle Medium (DMEM) and Ham’s F12 culture media devoid of phenol red (Gibco BRL/Invitrogen, Vienna, Austria), containing 10% fetal calf serum, 2% l-glutamine, 1% penicillin/streptomycin, and 1% amphotericin B (PAA, Pasching, Austria). Every 10th -15th passage, unfused Jeg-3 cells were removed by challenging the culture with 100 M hypoxanthine and 5.7 M azaserine (Sigma–Aldrich, Vienna, Austria). Exponentially growing monolayers were gently detached by accutase (PAA) after removal of the medium and washing with Hanks’ balanced salt solution (HBSS; PAA). The resulting single cell suspension was cultivated in Petri dishes with non-adherent surfaces (Greiner Bio One, Kremsmünster, Austria) under permanent gentle shaking (Belly Button, Stovall Life Science Inc., Kehl, Germany) in a humidified CO2 incubator (5% CO2 , Heraeus Cytoperm 2, Thermo Electron LED GmbH, Vienna, Austria). After 24 h the cell spheroids were transferred into 25-ml glass spinner flasks (Glastechnische Werkstätte GmbH, Vienna, Austria) and stirred at 135 rotations per minute for further 48 h under identical culture conditions. Spheroids with a diameter of 300 m were selected under a dissecting microscope equipped with an ocular grid (Leica MZ6, Wetzlar, Germany). 2.3. Generation of spheroids from decidua tissue
The present study was approved by the Ethics Committee of the Medical University of Graz (No. 16-188 ex 04/05).
Human first trimester decidua parietalis was obtained from elective pregnancy terminations by curettage and vacuum suction between the 6th and 11th week of gestation at the Department of Obstetrics and Gynaecology, Medical University of Graz. Samples were checked for absence of invasive extravillous trophoblast cells by means of immunohistochemical staining for HLAG. Spheroid formation was accomplished by using the protocol of Grümmer et al. (1990). The tissue was cut into pieces of about 800 m under a dissecting microscope equipped with an ocular grid (Leica MZ6, Wetzlar, Germany), transferred to 25 ml glass spinner flasks and stirred at 135 rpm in a humidified incubator at 5% CO2 . The tissue explants were maintained in DMEM culture medium devoid of phenol red containing 10% FCS, 2% l-glutamine, 1% penicillin/streptomycin, 1% amphotericin B, 20 ng/ml progesterone, 10 ng/ml gestonoroncapronate (19-nor-17a-hydroxy-progesterone, a progesterone analogon and 300 pg/ml 17-beta-estradiol (kindly provided by Dr. Halfbrodt, Schering, Berlin, Germany). In addition to progesterone, gestonoroncapronate serving as a depot compound was included in the culture medium in order to reduce fluctuations of the hormone concentration. The medium was changed daily and the decidua spheroids were kept in culture for 3 days.
2.2. Generation of multicellular trophoblast spheroids
2.4. Toxic challenging of spheroids
We utilized a modified version of a previously published protocol (Helige et al., 2008). Briefly, the human trophoblast/Jeg-3 choriocarcinoma hybrid cell line AC1M59 (Schmitz, 2002) was generously provided by H-G.
Hydrogen peroxide (H2 O2 , Sigma–Aldrich, Vienna, Austria) was added to the culture medium of the spheroids for 48 h under continuous stirring at concentrations ranging from 0.05 mM to 1 mM. As controls, spheroids were either
2. Materials and methods 2.1. Ethics
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
65
Fig. 1. Brightfield and corresponding fluorescence images of cryosections of propidium iodide loaded AC1-M59 trophoblast-like cell spheroids. Original magnification 10×.
66
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
kept untreated or were killed by transfer into 50% ethanol (diluted in culture medium) for 30 min. 2.5. Propidium iodide loading and fluorescence microscopy For the purpose of viability testing the spheroids were washed with culture media devoid of phenol red and FCS. Subsequently, 50 g/ml propidium iodide (PI; Sigma–Aldrich, Vienna, Austria) was added for 15 min. After washing with HBSS, the spheroids were embedded in tissue freezing medium (OCT-Sakura, Sanova Pharma, Vienna, Austria) and frozen at −80 ◦ C. Serial cryosections were produced (thickness: 6 m, average distance between the sections evaluated: 18 m) and photomicrographs of the fluorescence and the brightfield image were taken on a fluorescence microscope (AxioPhot, Zeiss, Vienna, Austria) equipped with an HBO lamp (Osram, Zeiss, Vienna), a filter suitable for tetramethyl rhodamine isothiocyanate (TRITC), a 10× objective and a digital camera (AxioCam HRc, Zeiss, Vienna, Austria). All experiments were done thrice, and in each experiment three representative spheroids were analysed for each condition. 2.6. Image processing Although optical assessment of the images already allowed for rough estimation of the effects of H2 O2 stress on multicellular spheroids (Fig. 1), quantification of the results is desirable in order to distinguish faint differences and compare different samples and conditions. In order to develop a quantitative measure for toxicity several segmentation and image operations were carried out. 2.6.1. Segmentation of the propidium iodide fluorescence area Af The digital RGB color images of the fluorescent cells were segmented using k-means clustering with k = 2 in the RGB color space yielding red fluorescent clusters and an almost black background for each individual image (Fig. 2a). The digital RGB color transmission images represented the corresponding areal picture of the spheroid. The bi-colored images of red PI in the nuclei of dead cells against a black background were transformed into a binary form by setting the background cluster center to black and the fluorescence color center to white (Fig. 2b). 2.6.2. Segmentation of the total spheroid area At The segmentation of the spheroids was complicated because simple grey value thresholding or k-means clustering was not applicable. Relatively large areas inside the spheroid showed quite the same grey values as the background. Therefore, a more sophisticated algorithm was developed. As the colors of these images did not include any useful information, the images were transferred to gray value images (Fig. 3a). The variation of the overall image intensity of the entire image stack was minimized by applying the Lee operator with a 5 × 5 kernel yielding slightly low pass filtered and normalized images (Fig. 3b).
Fig. 2. Segmentation of a fluorescence image of a section of a propidium iodide loaded AC1-M59 cell spheroid. K-mean clustering with k = 2 was carried out in the RGB color space. The evolved cluster centers were set to black and white yielding a binary representation of the image. (a) Original color fluorescence image. (b) Segmented binary image, representing the total fluorescence area Af .
Subsequently, the images were converted and subjected to thresholding (Fig. 3c). In parallel, rough polygons were drawn interactively over the contours of spheroids in the Lee-filtered image (Fig. 3d) in order to exclude areas with increased dirt in the background from subsequent segmentation steps. The resulting regions of interest (ROI) were filled with white pixels serving as binary masks (Fig. 3e) for the Lee filtered and inverted images. All pixels of these filtered images which corresponded to black pixels in the masks were set to black (Fig. 3f) and a fixed threshold was applied to the overall stack of images, yielding binary representations of the spheroids (Fig. 3g). Large holes inside the spheroid areas were filled by a tenfold morphological closing (Fig. 3h) and binary hole filling (Fig. 3i) of all remaining holes smaller than 20% of the total image size, thereby preserving the filling of spheroid areas without erroneously filling the background of the images. Fig. 4 exemplarily shows the result by an overlay of the binary image over the original transmission image.
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
67
Fig. 3. Segmentation of a transmission image of a section of a AC1-M59 cell spheroid. The original color image was converted to a gray value image (a), Lee filtered (b) and inverted (c). A region of interest (d) and a mask image (e) were produced. The masked image (f) was thresholded (g), tenfold morphologically closed (h) and finally a hole filling of holes less then 20% of the total image size was applied (i).
68
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
Fig. 4. Overlay of the binary result over the original image. This overlay shows the result of a segmentation of a section of a trophoblast-like AC1-M59 cell spheroid. Despite of the relatively large areas in between the cells, it was possible to segment the compact spheroid area. The binary image was used to calculate the total spheroid area At .
2.6.3. Ratiometric quantification The binary images of both the fluorescence cells and the corresponding entire area of the spheroids were measured by counting white image pixels in the binary image representations. The total number of white image pixels in an image gave mathematically the total area of the fluorescence and the transmission images respectively. The quantitative toxicity factor S was calculated as the ratio of the total area of the fluorescent cells to the corresponding total area of the spheroids: S=
Af At
S:0≤S≤1
with Af the area of the fluorescent cells and At the corresponding total area of the two-dimensional disk representations of the spheroids. As S is dimensionless, it is equivalent if this calculation would be carried out in units of pixels or in the real physical units. S can theoretically show values in between zero and one, dependent on the tissue-specific quantitative relationship of nuclei on one hand and cytoplasm plus extracellular space on the other hand. The value zero means that there is really zero fluorescence in the image and the value one has the meaning that the total area of the spheroid is fluorescent. Therefore, the values around zero and also around one represent very extreme limits for S. 2.7. Measurement of lactate dehydrogenase (LDH) in the culture supernatant This was done for the purpose of comparison to a conventional methodology of determination of cell death. AC1-M59 cell spheroids were treated with H2 O2 as described above, after a period of 48 h culture supernatant was collected. The positive control for spheroid killing was treated with a final concentration of 1% Triton (Merck, Vienna, Austria) for 1 h at the end of the culture period.
Fig. 5. Diagrams showing the quantitative toxicity factor. The quantitative toxicity factor S for multicellular spheroids from the trophoblast-like cell line AC1-M59 (a) or spheroids from decidua tissue after 3 d in spinning culture (b) treated with different concentrations of hydrogen peroxide or with 50% ethanol as a positive control for killing of all cells is given as area of propidium iodide fluorescence per total area of the respective sections. Values are given as medians plus interquartile ranges. Asterisks indicate a p ≤ 0.001 in comparison to the untreated spheroids.
The culture supernatants were kept at −70 ◦ C until determination of LDH activity, which was done in 96-well plates using the LDH Cytotoxicity Detection Kit (Takara Bio Inc, Shiga, Japan) according to the manufacurer’s instructions. The absorbance of the samples was measured in duplicates at 492 nm using a photometer (anthos Reader 2010, Anthos, Salzburg, Austria), the reference wavelength was 620 nm. A total of three experiments were performed. 2.8. Statistical analysis As the data do not follow Gaussian distribution (as assessed by Komolgorov–Smirnov testing), we applied nonparametric two-tailed Mann–Whitney U-test for analysis of significance of differences in S between untreated and treated samples. The SPSS 16 software package was used for this purpose. 3. Results The viability of the spheroids was assessed by exclusion of PI from viable cells which in contrast rapidly labels
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
Fig. 6. Lactate dehydrogenase in the supernatant of cell spheroids. Alternative methodology for assessment of the amount of toxicity of hydrogen peroxide against AC1-M59 trophoblast-like cell spheroids. 1% triton is used as a positive control for killing. The graph indicates the medians and the ranges of three experiments.
the nuclei of dead cells (Fig. 1). At concentrations as low as 0.2 mM H2 O2 the amount of cell death as measured by the ratio of the total area of the fluorescent cells and the corresponding total area of the section of the trophoblast cell spheroids was not increased compared to the negative control. From 0.3 to 0.5 mM H2 O2 lethality rapidly increased in the trophoblast cell spheroids whereas at the same time the overall number, as well as the size of the surviving spheroids decreased (Fig. 5a). Due to complete decomposition of the trophoblast spheroids toxicity analysis was not applicable at higher concentrations of H2 O2 . However, in the positive control (treated with 50% ethanol) PI fluorescence was apparent throughout the whole the multicellular spheroids without obvious decomposition. Thus, the nature of both the cell spheroids and the toxic agent may determine the applicability of this method. Spheroids from decidua tissue were considerably more resistant to oxidative stress, at 0.5 mM H2 O2 no increase of lethality was observed (Fig. 5b). For comparison, the outcome of measurement of lactate dehydrogenase in the supernatant of AC1-M59 cell spheroids stressed with H2 O2 is shown in Fig. 6. 4. Discussion In the present study we report a novel method for quantification of the amount of cell death in multicellular three-dimensional spheroids by quantitative assessment of the relative potency of different concentrations of a toxic agent to induce tissue damage. As a model for toxic damage we used hydrogen peroxide added on spheroids made from a choriocarcinoma/trophoblast hybrid cell line and of spheroids produced from pieces of decidua tissue. We found that decidua spheroids are considerably more resistant to oxidative stress than trophoblast spheroids. The conventional methodology of measurement of lactate dehydrogenase in the culture supernatant was comparatively inefficient in determining toxicity against trophoblast-like cell spheroids stressed with H2 O2 .
69
Reactive oxygen species are a highly reactive group of oxygen-containing molecules, leading to oxidative stress (reviewed in Sies, 1997). Oxidative stress is involved in the aetiology of defective embryo development (Guerin et al., 2001). Damage induced by oxidative stress has been hypothesized to play a role in spontaneous abortion, idiopathic recurrent pregnancy loss, defective embryogenesis, and drug-induced teratogenicity (Iborra et al., 2005). Plasma membrane integrity which is lost during necrosis but also in the late phase of apoptosis was probed by the exclusion of PI (Buenz et al., 2007). In order to allow a numerical assessment and comparison of the relative potency of toxic agents and sensitivity of spheroids of diverse origins we optimized a quantitative measurement of toxicity based on processing of fluorescent images of PI loaded cells. Our goal was to determine the maximal concentration of hydrogen peroxide which induces tissue stress without damage in terms of an increased fraction of dead cells within multicellular spheroids. An increase of dead cells with nuclear staining by PI across the whole section correlates well with the concentration of the toxic agent applied. We show our method to be extendable to analysis of toxicity towards cultured pieces of primary tissue. Whereas the method allows a very accurate determination of toxicity, it is rather time-consuming in comparison to conventional methods for determining vitality or mortification, Furthermore, it requires a fluorescence microscope. A limiting factor for its applicability is the diameter of the spheroids, as PI needs to penetrate sufficiently in order to stain dead cells also in the center of the spheroids. This is very probably dependent on the type of cell line used and on whether or not a matrix is applied as a scaffold for spheroid formation. Moreover, it is reported that most of the proliferating cells within a spheroid are located in a viable rim, whereas quiescent cells are located centrally and cells at a depth exceeding 160 m frequently become necrotic (Lambert et al., 2006). In the present study, however, the trophoblast spheroids were used at a uniform diameter of 300 m and decidua spheroids were used at 800 m × 300 m. Both types did not show necrotic centers and PI could completely penetrate the whole aggregate. The method we propose allows for analysis of spatial distribution of necrotic cells. Using the toxic effect of oxidative stress on spheroids of a choriocarcinoma/trophoblast hybrid cell line as a model system, it proved to be reproducible and economically reasonable. Further technical advancement may ease parallel handling of numerous probes and thus enable application for screening in toxicological studies and drug screening. Acknowledgements We thank Ms. Nina Flieser and Mr. Rudolf Schmied of the Institute of Cell Biology, Histology and Embryology and Ms. Renate Michlmaier of the Department of Gynaecology and Obstetrics, Medical University of Graz, Austria, for their technical assistance. We are grateful to Dr. Erwin Tafeit (Institute of Physiological Chemistry, Medical University of Graz) for advice on statistical analysis. This study
70
R.-S. Theuerkauf et al. / Journal of Reproductive Immunology 85 (2010) 63–70
was supported by the European Commission (Network of Excellence “The Control of Embryo Implantation (EMBIC)”, FP6-512040, lead researchers involved: P.M., P.S.). References Badia, R., Iborra, A., Palacio, J.R., Antich, M., Martinez, P., 2008. The effect of oxidative environment on immunosuppressive properties of human seminal plasma. Am. J. Reprod. Immunol. 60, 354–360. Buenz, E.J., Limburg, P.J., Howe, C.L., 2007. A high-throughput 3-parameter flow cytometry-based cell death assay. Cytometry A 71, 170–173. Dhiman, H.K., Ray, A.R., Panda, A.K., 2005. Three-dimensional chitosan scaffold-based MCF-7 cell culture for the determination of the cytotoxicity of tamoxifen. Biomaterials 26, 979–986. Grümmer, R., Hohn, H.-P., Denker, H.-W., 1990. Choriocarcinoma cell spheroids: an in vitro model for the human trophoblast. Trophoblast Res. 4, 97–111. Guerin, P., El Mouatassim, S., Menezo, Y., 2001. Oxidative stress and protection against reactive oxygen species in the pre-implantation embryo and its surroundings. Hum. Reprod. Update 7, 175–189. Helige, C., Ahammer, H., Hammer, A., Huppertz, B., Frank, H.G., Dohr, G., 2008. Trophoblastic invasion in vitro and in vivo: similarities and differences. Hum. Reprod. 23, 2282–2291. Hoffman, R.M., 1991. Three-dimensional histoculture: origins and applications in cancer research. Cancer Cells 3, 86–92. Iborra, A., Palacio, J.R., Martinez, P., 2005. Oxidative stress and autoimmune response in the infertile woman. Chem. Immunol. Allergy 88, 150–162.
Indovina, P., Rainaldi, G., Santini, M.T., 2007. Three-dimensional cell organization leads to a different type of ionizing radiation-induced cell death: MG-63 monolayer cells undergo mitotic catastrophe while spheroids die of apoptosis. Int. J. Oncol. 31, 1473–1483. Khadaroo, R.G., Kapus, A., Powers, K.A., Cybulsky, M.I., Marshall, J.C., Rotstein, O.D., 2003. Oxidative stress reprograms lipopolysaccharide signaling via Src kinase-dependent pathway in RAW 264.7 macrophage cell line. J. Biol. Chem. 278, 47834–47841. Kumar, H.R., Zhong, X., Hoelz, D.J., Rescorla, F.J., Hickey, R.J., Malkas, L.H., Sandoval, J.A., 2008. Three-dimensional neuroblastoma cell culture: proteomic analysis between monolayer and multicellular tumor spheroids. Pediat. Surg. Int. 24, 1229–1234. Lambert, B., De Ridder, L., De Vos, F., Slegers, G., de Gelder, V., Van de Wiele, C., Thierens, H., 2006. Assessment of supra-additive effects of cytotoxic drugs and low dose rate irradiation in an in vitro model for hepatocellular carcinoma. Can. J. Physiol. Pharmacol. 84, 1021–1028. Mosmann, T., 1983. Rapid colorimetric assay for cellular growth and survival: application to proliferation and cytotoxicity assays. J. Immunol. Methods 65, 55–63. Mueller-Klieser, W., 1997. Three-dimensional cell cultures: from molecular mechanisms to clinical applications. Am. J. Physiol. 273, C1109–C1123. Sies, H., 1997. Oxidative stress: oxidants and antioxidants. Exp. Physiol. 82, 291–295. Verma, P., Verma, V., Ray, P., Ray, A.R., 2007. Formation and characterization of three dimensional human hepatocyte cell line spheroids on chitosan matrix for in vitro tissue engineering applications. In Vitro Cell. Dev. Biol.-Anim. 43, 328–337.