Cytotechnology 4: 155-161, 1990. 9 1990KluwerAcademic Publishers. Printed in the Netherlands.
Automated surface area measurement of cultured cardiac myocytes M. Toraason, J.A. Krueger, K.A. Busch and P.B. Shaw Centers for Disease Control, National Institute for Occupational Safety and Health, Division of Biomedical and Behavioral Science, Cellular Toxicology Section, 4676 Columbia Parkway, Cincinnati, OH 45226, USA Received 12 January 1990; accepted in revisedform 9 May 1990
Key words: rat, heart, growth, norepinephrine
Abstract An automated method for rapidly measuring surface area of individual cardiac myocytes was used as an index of myocyte growth. Hearts from 2- to 4-day-old rats were digested by overnight incubation in cold trypsin solution. Enriched suspensions of myocytes were plated at 2 x 105 ceils/well in 12-well-culture plates. Cells were grown in M199 supplemented with 1%, 10% serum or 10% serum plus 10-7 M norepinephrine. On days 1-4 after plating, cells were fixed in Bouin's Solution and stained with Weigert's Iron Hematoxylin and Biebrich Scarlet-Acid Fuchsin. An inverted microscope, video camera and monitor were coupled to a video image processor (Image Technology Corp.). The enhanced image of stained heart cells was digitized, and perimeter, length, width and area of each selected cell were calculated. One hundred randomly selected cells were measured in each of eight wells from each treatment-day group. Areas of individual myocytes varied widely in culture dishes and the distribution was skewed toward larger cells. T h e standard deviation increased in proportion to an increase in mean cell area. A logarithmic transformation of the data normalized the data and yielded a more homogeneous variance. The geometric mean area of heart cells supplemented with 1% serum increased only slightly, but significantly, during four d a y s in culture. Geometric mean area of cells supplemented with 10% serum increased nearly four-fold. Supplementing cells with norepinephrine (10 -7 M) in addition to 10% serum did not induce a further increase in cell size. This technique has the potential to rapidly and objectively monitor heart cell growth following pharmacological or toxicological treatments.
Introduction Cardiac myocytes increase in size but not in number during the post-neonatal period (Cluzeaut and Maurer-Schultze, 1986; Claycomb, 1977). Hypertrophy of the myocardium can be induced by a variety of pharmaceutical and physical challenges (Marino et al., 1987; Meidell et al., 1986; Simpson, 1985; 1983). In attempting to
understand the cellular and molecular events associated with myocyte growth and hypertrophy, investigators have utilized cultured cardiac myocytes from neonatal and adult experimental animals. To a large extent, biochemical ahd physiological end points have been used to assess growth (Marino et al., 1987; Grynberg et al., 1986; Simpson, 1985; 1983). Densitometry has been used to assess over all growth of cultures
156 (Terracio and Douglas, 1982), but this technique does not provide information on the distribution of cell size within a population of ceils. With the advent of image analysis systems, however, morphological evaluations emphasizing individual cell size have played an increased role (Thurston et al., 1988; Travo et al., 1987; Nag et al., 1986). In addition to morphological assessment, methodologies have also been developed for objectively analyzing cell migration (Ryan and Mayfield, 1986). In our studies on growth of cardiac myocytes (Toraason and Krueger, 1988), we became aware of three obstacles to the automated measurement and statistical evaluation of surface area measurement. First, despite efforts to obtain pure cultures of cardiac myocytes through differential plating of cells, the cultures were not of a homogeneous cell type. Second, low contrast between cell image and background hampered image enhancement and objective demarcation of a cell perimeter. Third, after several days in culture, cell sizes are not normally distributed, which goes counter to assumptions made for the common statistical approaches used in data analysis. This report describes a selective staining procedure that singles out myocytes and allows automated measurement of surface area in non-homogenous cell cultures. A statistical approach for analyzing a population of cells Of heterogeneous size is also reported.
Materials and methods
Hearts were harvested aseptically from 2- to 4-day-old Sprague-Dawley rats from our own colony by a procedure previously described (Toraason et al., 1989). Atria and major vessels were trimmed from the ventricles which were minced into 1 mm 2 pieces, and refrigerated (4~ overnight in Hanks' balanced salt solution (Hazelton, Lenexa, KS) containing 0.1% crude trypsin (Sigma, St. Louis, MO). The following morning minced hearts were washed and then incubated in a trypsinization flask at 35~ for 25 min in M199 (Hazelton, Lenexa, KS) containing 10% fetal
bovine serum (Hyclone, Logan, UT) and 100 U/ml Penicillin-Streptomycin (Hazelton). Cardiac tissue was dispersed by rapid repipetting to produce a single cell suspension. The cell suspension was decanted from the flask leaving undigested tissue fragments behind. Approximately five million cells were obtained from each heart, and viability was consistently about 90% as determined by trypan blue exclusion. Enriched myocyte cultures were obtained by plating 40 to 60 ml of suspension containing 106 cells per ml in 175 cm 2 flasks for 0.5 hr. The suspension was removed, a cell count was performed, and cells were diluted to 2 x 105 cells per ml and plated in 16 mm wells in 12-well plates. Media was changed 24 hrs after plating and cells received M199 supplemented with 1% serum, 10% serum or 10% serum and 10-7 M norepinephrine (Sigma). Subsequently, the media was changed every day. Preservation and staining of cultured Cardiac myocytes were done by standard procedures. In brief, cells were washed with sodium phosphate buffer and placed in Bouin's solution for 1 hr at 560C. After washing in tap water for 10 min, cells w e r e stained with Weigert's Iron Hematoxylin for 10 min. Following a second wash in tap water, cells were stained with Biebrich ScarletAcid Fuchsin, then rinsed with deionized/distilled water and allowed to air dry. Preserving and staining reagents were obtained from Sigma. Area of stained heart cells was determined using an image analyzer (model 300) manufactured by Image Technology Corporation (Deer Park, NY). Heart cells were magnified 200x by a Nikon Diaphot microscope and the image was transferred via a video camera to a television monitor. Software developed by Image Technology Corporation and run on an IBM PC allowed enhancement of the individual video image of cells. The enhanced i m a g e w a s digitized, which allowed separation and selection of individual cells. Two or more myocytes were often connected by pseudopods and had.to be separated in the digitized image so as not to be measured as a single cell. Non-muscle cells (fibroblasts and endothelial cells) also had to be eliminated from
157 the analysis. This was aided by the darker staining of myocytes relative to non-muscle cells. Occasionally contrast between muscle and nonmuscle cells was too difficult to discern on the monitor. In such cases, it was necessary to view the cells through the microscope objective. Darkly red stained myocytes were easily distinguishable from the lightly stained non-muscle cells, which then could be eliminated from the digitized image. In order to calculate area, perimeter, length, and width, the image analysis system stores selected images on the video monitor in a memory panel of 419 • 512 pixels (picture points). Each enhanced heart cell image corresponds to a series of horizontal and vertical pixels that are converted to area (l.tm2) based on calibration of the system using an object of known dimension. Using this system, many cells could be measured simultaneously and it took about 10 min to meas-
ure 100 ceils in a well. Statistical analysis of the area data involved an investigation into the distributional properties of the heart cell area measurements. Logarithmic transformation of the area measurement yielded a suitable response variable for analysis of variance (ANOVA). To ensure that a logarithmic transformation was necessary and satisfactory, the Bartlett-Kendall test (Anderson and McLean, 1974, p. 19) for homogeneity of within-well variances was performed for both untransformed and transformed data. Data were also tested for normality using the Kolmogorov-Smimov test (Sokaland Rohlf, 1981). In a preliminary statistical study of similar data, both intrawell and interwell components of variability were examined for their distributional properties. A logarithmic transformation homogenized both intrawell and interwell variances in treatments with differ-
Fig. 1. Myocytessupplementedwith 10% fetalbovine serum one day (A) and four days (B) after plating. Myocyteswere fixed and
stained as noted in text. Magnificationwas 150x.
158 ent average cell sizes. The log transformation had nearly normal distributions of both types of random variation. The log-transformed cell areas were analyzed by ANOVA for day in culture and treatment differences. Individual differences were identified using Tukey's test (P
Automated surface area measurement of cultured cardiac myocytes M. Toraason, J.A. Krueger, K.A. Busch and P.B. Shaw Centers for Disease Control, National Institute for Occupational Safety and Health, Division of Biomedical and Behavioral Science, Cellular Toxicology Section, 4676 Columbia Parkway, Cincinnati, OH 45226, USA Received 12 January 1990; accepted in revisedform 9 May 1990
Key words: rat, heart, growth, norepinephrine
Abstract An automated method for rapidly measuring surface area of individual cardiac myocytes was used as an index of myocyte growth. Hearts from 2- to 4-day-old rats were digested by overnight incubation in cold trypsin solution. Enriched suspensions of myocytes were plated at 2 x 105 ceils/well in 12-well-culture plates. Cells were grown in M199 supplemented with 1%, 10% serum or 10% serum plus 10-7 M norepinephrine. On days 1-4 after plating, cells were fixed in Bouin's Solution and stained with Weigert's Iron Hematoxylin and Biebrich Scarlet-Acid Fuchsin. An inverted microscope, video camera and monitor were coupled to a video image processor (Image Technology Corp.). The enhanced image of stained heart cells was digitized, and perimeter, length, width and area of each selected cell were calculated. One hundred randomly selected cells were measured in each of eight wells from each treatment-day group. Areas of individual myocytes varied widely in culture dishes and the distribution was skewed toward larger cells. T h e standard deviation increased in proportion to an increase in mean cell area. A logarithmic transformation of the data normalized the data and yielded a more homogeneous variance. The geometric mean area of heart cells supplemented with 1% serum increased only slightly, but significantly, during four d a y s in culture. Geometric mean area of cells supplemented with 10% serum increased nearly four-fold. Supplementing cells with norepinephrine (10 -7 M) in addition to 10% serum did not induce a further increase in cell size. This technique has the potential to rapidly and objectively monitor heart cell growth following pharmacological or toxicological treatments.
Introduction Cardiac myocytes increase in size but not in number during the post-neonatal period (Cluzeaut and Maurer-Schultze, 1986; Claycomb, 1977). Hypertrophy of the myocardium can be induced by a variety of pharmaceutical and physical challenges (Marino et al., 1987; Meidell et al., 1986; Simpson, 1985; 1983). In attempting to
understand the cellular and molecular events associated with myocyte growth and hypertrophy, investigators have utilized cultured cardiac myocytes from neonatal and adult experimental animals. To a large extent, biochemical ahd physiological end points have been used to assess growth (Marino et al., 1987; Grynberg et al., 1986; Simpson, 1985; 1983). Densitometry has been used to assess over all growth of cultures
156 (Terracio and Douglas, 1982), but this technique does not provide information on the distribution of cell size within a population of ceils. With the advent of image analysis systems, however, morphological evaluations emphasizing individual cell size have played an increased role (Thurston et al., 1988; Travo et al., 1987; Nag et al., 1986). In addition to morphological assessment, methodologies have also been developed for objectively analyzing cell migration (Ryan and Mayfield, 1986). In our studies on growth of cardiac myocytes (Toraason and Krueger, 1988), we became aware of three obstacles to the automated measurement and statistical evaluation of surface area measurement. First, despite efforts to obtain pure cultures of cardiac myocytes through differential plating of cells, the cultures were not of a homogeneous cell type. Second, low contrast between cell image and background hampered image enhancement and objective demarcation of a cell perimeter. Third, after several days in culture, cell sizes are not normally distributed, which goes counter to assumptions made for the common statistical approaches used in data analysis. This report describes a selective staining procedure that singles out myocytes and allows automated measurement of surface area in non-homogenous cell cultures. A statistical approach for analyzing a population of cells Of heterogeneous size is also reported.
Materials and methods
Hearts were harvested aseptically from 2- to 4-day-old Sprague-Dawley rats from our own colony by a procedure previously described (Toraason et al., 1989). Atria and major vessels were trimmed from the ventricles which were minced into 1 mm 2 pieces, and refrigerated (4~ overnight in Hanks' balanced salt solution (Hazelton, Lenexa, KS) containing 0.1% crude trypsin (Sigma, St. Louis, MO). The following morning minced hearts were washed and then incubated in a trypsinization flask at 35~ for 25 min in M199 (Hazelton, Lenexa, KS) containing 10% fetal
bovine serum (Hyclone, Logan, UT) and 100 U/ml Penicillin-Streptomycin (Hazelton). Cardiac tissue was dispersed by rapid repipetting to produce a single cell suspension. The cell suspension was decanted from the flask leaving undigested tissue fragments behind. Approximately five million cells were obtained from each heart, and viability was consistently about 90% as determined by trypan blue exclusion. Enriched myocyte cultures were obtained by plating 40 to 60 ml of suspension containing 106 cells per ml in 175 cm 2 flasks for 0.5 hr. The suspension was removed, a cell count was performed, and cells were diluted to 2 x 105 cells per ml and plated in 16 mm wells in 12-well plates. Media was changed 24 hrs after plating and cells received M199 supplemented with 1% serum, 10% serum or 10% serum and 10-7 M norepinephrine (Sigma). Subsequently, the media was changed every day. Preservation and staining of cultured Cardiac myocytes were done by standard procedures. In brief, cells were washed with sodium phosphate buffer and placed in Bouin's solution for 1 hr at 560C. After washing in tap water for 10 min, cells w e r e stained with Weigert's Iron Hematoxylin for 10 min. Following a second wash in tap water, cells were stained with Biebrich ScarletAcid Fuchsin, then rinsed with deionized/distilled water and allowed to air dry. Preserving and staining reagents were obtained from Sigma. Area of stained heart cells was determined using an image analyzer (model 300) manufactured by Image Technology Corporation (Deer Park, NY). Heart cells were magnified 200x by a Nikon Diaphot microscope and the image was transferred via a video camera to a television monitor. Software developed by Image Technology Corporation and run on an IBM PC allowed enhancement of the individual video image of cells. The enhanced i m a g e w a s digitized, which allowed separation and selection of individual cells. Two or more myocytes were often connected by pseudopods and had.to be separated in the digitized image so as not to be measured as a single cell. Non-muscle cells (fibroblasts and endothelial cells) also had to be eliminated from
157 the analysis. This was aided by the darker staining of myocytes relative to non-muscle cells. Occasionally contrast between muscle and nonmuscle cells was too difficult to discern on the monitor. In such cases, it was necessary to view the cells through the microscope objective. Darkly red stained myocytes were easily distinguishable from the lightly stained non-muscle cells, which then could be eliminated from the digitized image. In order to calculate area, perimeter, length, and width, the image analysis system stores selected images on the video monitor in a memory panel of 419 • 512 pixels (picture points). Each enhanced heart cell image corresponds to a series of horizontal and vertical pixels that are converted to area (l.tm2) based on calibration of the system using an object of known dimension. Using this system, many cells could be measured simultaneously and it took about 10 min to meas-
ure 100 ceils in a well. Statistical analysis of the area data involved an investigation into the distributional properties of the heart cell area measurements. Logarithmic transformation of the area measurement yielded a suitable response variable for analysis of variance (ANOVA). To ensure that a logarithmic transformation was necessary and satisfactory, the Bartlett-Kendall test (Anderson and McLean, 1974, p. 19) for homogeneity of within-well variances was performed for both untransformed and transformed data. Data were also tested for normality using the Kolmogorov-Smimov test (Sokaland Rohlf, 1981). In a preliminary statistical study of similar data, both intrawell and interwell components of variability were examined for their distributional properties. A logarithmic transformation homogenized both intrawell and interwell variances in treatments with differ-
Fig. 1. Myocytessupplementedwith 10% fetalbovine serum one day (A) and four days (B) after plating. Myocyteswere fixed and
stained as noted in text. Magnificationwas 150x.
158 ent average cell sizes. The log transformation had nearly normal distributions of both types of random variation. The log-transformed cell areas were analyzed by ANOVA for day in culture and treatment differences. Individual differences were identified using Tukey's test (P
