Proc. Nat. Acad. Sci. USA Vol. 73, No. 2, pp. 651-653, February 1976
Pathology
Clustering of replicating cells in aortic endothelium (cell replication/atherosclerosis/artery/endothelial injury)
STEPHEN M. SCHWARTZ AND EARL P. BENDITT Department of Pathology, University of Washington, Seattle, Wash. 98195
Contributed by Earl P. Benditt, November 7,1975
ABSTRACT Cell division in the aortic endothelium of the rat is not randomly distributed. Maps of the aortic surface show focal areas where the daily rate of replication may be in excess of 50%. This implies the existence of focal areas of rapid cell growth or rapid cell turnover and other areas where growth or turnover is largely absent.
five segments. The endothelium from each segment was removed as a sheet, mounted on glass slides, and processed for autoradiography. The final result of this procedure is a preparation of endothelium with the original distribution of cells preserved and with replicating cells identified by autoradiography. Mapping of Specimens. Samples were chosen for mapping based on the quality of preparation, including: (i) retention of most or all of the endothelial surface, (ii) absence of nonendothelial cells, (iii) good quality of staining of nuclei, and (iv) low background grain count. A total of 15 segments was completely mapped, including portions of both thoracic and abdominal aorta. These preparations were examined in the light microscope using a square-shaped reticule in a lOX wide-field eyepiece with a 40X objective. Each specimen was examined by moving the square field systematically across the preparation, recording the number of labeled cells and the total number of cells in each field. The location of each field was recorded on a grid. The grid thus provided a map of the distribution of labeled cells and the population density of the entire specimen. Evaluation of Data. The following parameters were obtained from each specimen: 2N* = total number of labeled cells; ZN = total number of cells; Tf = total number of fields. These numbers were in turn used to calculate the thymidine index (TI), the cell density (CD), and the mean number of labeled cells per field (N*):
Large and small blood vessels are lined on their inner surfaces by a continuous single layer of specialized cells, the endothelium. The endothelium modulates passage of fluid, solutes, and cells from the blood into the adjacent vessel wall and prevents adhesion of platelets (1, 2). Any breach in the integrity of the endothelium leads to local loss of these barrier functions. For this reason, an understanding of how the continuity of the endothelium is maintained becomes important. It is known that endothelial cells are lost into the blood stream in human beings (3-5). From animal studies, it is known that endothelial cells can multiply, as indicated by incorporation of tritiated thymidine into nuclei in the endothelium of normal animals (6-8) and by the regeneration of endothelium following partial denudation of the aorta in experimental animals (9). The rate of aortic endothelial cell replication under normal conditions has been studied in several species and has been estimated to be less than 1% per day (6-8). This low rate of incorporation has been taken to mean that the rate of cell turnover is low. This, however, assumes that cell loss and cell replication are uniformly and randomly distributed over the inner surface of the vessel. It is possible to accurately map essentially the entire surface of the aortic endothelium of small animals with respect to thymidine incorporation and hence to ascertain the distribution of replicating cells (8). We report here a quantitative study of the distribution of multiplying cells in the rat aortic endothelium. In the aorta the distribution of cells by this indicator of mitosis appears to be nonrandom and to present significant numbers of loci of high turnover and large areas of apparently low or possibly absent multiplication.
TI = 2N°/2;N
CD= =N/f N°
=
ZN°/f
TI, the thymidine index, was used as a measure of the replication rate. CD, the cell density, is a measure of the number of cells per unit area. N° is the average number of labeled cells per field.
The map data also allowed us to examine statistically the distribution of labeled cells. Data from each specimen were grouped according to the number of fields with 0 labeled cells, 1 labeled cell, two labeled cells, and so on up to the maximum number of labeled cells present in any one field (Table 1). The mean and the variance of the number of labeled cells per field (10, 11) were then used to test the fit of the distribution to a random distribution, the Poisson. Finally, the actual distribution of labeled cells was compared with the random distribution expected from a binomial distribution. Ten specimens were chosen at random. The actual number of fields showing four or more labeled cells was counted for all 10 specimens. The average cell density (CD) for these specimens was calculated as above. This average cell density, 143 cells per field for the 10 preparations, was used as a sample size in calculating the expected frequency of fields containing four or more labeled cells. This
MATERIALS AND METHODS Animals. The experiments were performed on 10 female Sprague-Dawley rats, about 150 g in weight and 3 months of age. The animals were born and raised in facilities at the University of Washington and were maintained on a standard rat diet (Ralston-Purina Co., St. Louis, Mo.). Labeling Schedule. The dosage schedule consisted of three doses of [methyl-3H]thymidine (New England Nuclear Corp., Boston, Mass.), each dose: 50 mCi/100 g at 8-hr intervals. This schedule was chosen to minimize diurnal variation and to provide large enough numbers of cells to permit accurate counting. In addition, if we assume that S phase is about 8 hr long, this schedule should approximate a 24-hr labeling interval. Tissue was fixed and prepared as previously described (8). The aorta from each animal was divided into 651
Proc. Nat. Acad. Sci. USA 73 (1976)
Pathology: Schwartz and Benditt
652
the binomial distribution with p = labeled cells = 0.6%; k (sample size) = 143; (number samples) = total number of fields in the
was done by generating mean fraction of of and n
10 specimens
=
1926.
RESULTS and focal clusters of replication Overall rate of cell replicating cells Examples of the mapped data are shown in Fig. la, b, and c. The overall thymidine index frequency (TI) in the 15 segments was 0.55% 4 0.1 SE. This low mean value is consistent with observations in other species (6, 7). The labeled cells, however, do not appear evenly distributed. As can be seen in Fig. 1, labeled cells appear clustered in regions of variable size and shape. The labeling frequency within these clusters reaches as high as 60% in individual fields. Probability that clusters could occur by chance The first question that must be asked is whether these clusters could arise by chance. If the clustering occurs as part of a random series of events (10, 11), the number of labeled cells per field should fit a Poisson distribution. For this distribution, the ratio of the variance to the mean, the coefficient of dispersion, approximates 1. Values in excess of 1 indicate nonrandom aggregation or "clustering." Values below 1 represent the opposite phenomenon (11). Table 1 shows the results of this analysis of the 15 mapped segments. The coefficient of dispersion (C) is greater than 1 in all but three specimens. Moreover, when the coefficient is calculated for the entire group of specimens, the value is 2.4, with P < 0.001 by the x2 test (10, 11). Thus we may conclude that the labeled cells are not randomly distributed and that at least some cells are contained in clusters. Other evidence supports the distinction between areas containing labeled cells and the rest of the aorta. If the clusters are real, then replicating cells should be distributed in only a portion of the total area of the aorta. In this case one would expect to find a greater number of fields with rela-
H,,
+;*S Al
-67%
c~~~~~~~~~~~~
FIG. 1. These maps represent frequency distributions for labeled cells in the endothelium of portions of thoracic aorta from three animals. In each case the long axis represents the long axis of the aorta and the long edges represent the dorsal or posterior margin. (a) The dividing cells form a belt around the circumference of the vessel. This belt clearly involves areas of the aortic surface distinct from any vessel branches. The only branches in this segment of the vessel are the intercostal arteries, and these are confined to the dorsal region. (b) A very different pattern occurs, with a zone of labeled cells forming a stripe parallel to the long axis of the vessel. One field shows a labeling frequency greater than 10%. (c) The striking feature is the presence of two adjacent heavily labeled fields, each with an excess of 50% of cells showing the label. Other than these two fields, the remainder of the specimen shows only scattered labeling of cells. The scale is represented by a heavysided square to the right of each figure; the sides of this square represent 0.28 mm, equivalent to the sides of the counting reticule. Each dot on the map represents a labeling frequency of 0.1% in the area of the reticule as it was moved over the specimen. Arrows indicate the craniad and caudad directions.
tively large numbers of labeled cells than if the labeled cells were randomly distributed on the aortic surface. We tested this by looking at the actual number of fields with four or more labeled cells, compared with the number expected
Table 1. Number of fields with one or more labeled cells
men
Total
Number of labeled cells per field
Speci1
1 53 2 30 23 3 4 13 24 5 24 6 76 7 90 8 9 23 10 89 11 17 58 12 13 70 14 21 30 15 Total 641
2
3
4
5
6
24 14 6 11 5 16 31 73 12 48 11 29 64 5 10 359
5 10
3
1
1
1
1
2 7 23 52 5 14 2 8 45 1 14 188
10 11 12 13 14 15 16 17
9
8
no. of
4 3 21 2 5
2 1 10
8
3
2
1 1 1
1
1
3
2
1
1
1 1
1
7 26 2 4 77
1
12
2 10
6
3
23
9
8
1
1
4_ 1
1
1 3
35
1
1
2
0
1
1 1
1
fields
N*
s2
C
344 190 401' 76
0.34 0.58 0.09 0.61 0.63 1.66 1.28 1.65 0.57 1.04 0.76 1.55 2.13 0.75 1.73 1.01
0.45 1.16 0.11 0.96 0.43
1.3 2.0 1.3 1.6 0.7 3.0 1.0 2.1 1.6
54 85 181 393 123 281 70 119 307 63 89 2776
4.92 1.30 3.53 0.89 2.22 1.49 1.42 4.00 1.16 6.24 2.41
2.1 2.0 0.9 1.9 1.6 3.6 2.4
Table 1 shows the distribution of labeled cells in terms of the number of fields with 1 labeled cell, 2 labeled cells, up to 17, the highest number of labeled cells found in any one field. The column labeled "Total no. of fields" represents the sum of all fields, including unlabeled fields. N*, s2, and C for the row labeled "Total" represent these statistics for the data gathered from all 15 specimens. N* = mean number of labeled cells per field; s2 = variance of N*; C = coefficient of dispersion = (s2)/(N*).
Pathology:
Schwartz and Benditt
Proc. Nat. Acad. Sci. USA 73 (1976)
653
from a binomial distribution. Data were gathered from the first 10 mapped specimens. The expected value was 33 fields; the actual value was 108 fields. Finally, it was noted that the number of cells per field was higher in fields containing labeled cells, 153, than in unlabeled fields, 139. The direction of this difference was consistent in maps from these 10 preparations and was statistically significant by C test with P < 0.001. However, the difference in cell densities is small and might be attributable to the increased probability of finding labeled cells in fields with more cells. To test this possibility, we examined the thymidine index of "dense" fields (defined as fields containing at least 1 SD more cells than an average field). The average value for the thymidine index of the "dense" fields was 1.2%, twice the thymidine index for the entire population, 0.55%. This difference is significant at the P < 0.005 level by the t test. Thus the mean number of cells per field is higher in fields containing labeled cells.
injury must include factors other than disturbed flow pattern. Finally, another problem must be considered. We have assumed that the clusters of labeled cells seen in the maps have fixed locations. This may not be true. It is conceivable that different regions of the endothelium would label at other times or that labeled cells, while produced at specific points, later migrate to other regions of the endothelial surface. Data acquired from only a single labeling interval cannot be used to answer this question. In summary, the initial impression that aortic endothelium is long-lived, based on a low rate of cell replication of the population as a whole, represents an oversimplification. Some areas of the aorta may in fact be very long-lived and show little evidence of cell replacement. Other areas, represented by the heavily labeled clusters shown here, either represent foci of residual but rapid growth or represent foci where cells are being rapidly lost and replaced.
DISCUSSION We can conclude that replicating endothelial cells are not randomly distributed on the aortic surface. Some portion of the replicating cells appear clustered in focal regions of the aortic surface. There are at least three possible interpretations of these phenomena: (i) Divson after pulse. Some aggregation of labeled cells could arise by virtue of the fact that division of cells occurs during the labeling interval. However, it is unlikely that this could account for areas of the vessel lining containing 10 to 100 times as many labeled cells as the average or the appearance, in the mapped data, of large numbers of adjacent fields containing labeled cells. (ii) Growth centers. The clusters could represent centers of vascular enlargement or remodeling. It is difficult to be certain that the total number of cells in the endothelial population is in a steady state. With a replication rate as low as 0.6%, the rate of increase in the total number of cells must be quite low. In young animals, clusters of replicating cells may represent focal areas of residual growth, and this presents a special problem for studies in growing animals. (iii) High turnover regions. The clusters could represent sites where cells are being both lost and replaced at high rates. This is consistent with evidence from other studies which indicate increased replication rates in the endothelium near areas of disturbed blood flow, particularly vascular branches (6, 7, 12). However, a pattern of localization only near branches does not appear when the entire surface is mapped as here. While some of the clusters may be located near orifices, other labeled regions were found extending completely about the circumference of the aorta and hence must include areas without branches. Furthermore, heavily labeled regions were found in the anterior portion of the thoracic aorta where no branches occur. Thus, if the labeled areas represent sites of spontaneous injury, the mechanism of
This investigation was supported by National Institutes of Health Grants HL-03174, GM-13543, and GM-00100. 1. Landis, E. M. & Pappenheimer, J. R. (1963) "Exchange of substance through the capillary walls," in Handbook of Physiology (Am. Physiological Soc., Washington), Section 2: Circulation, Vol. II, pp. 961-1034. 2. Stemerman, M. B., Baumgartner, H. R. & Spaet, T. H. (1971) "The subendothelial microfibril and platelet adhesion," Lab. Invest. 24, 179-186. 3. Herbeuval, H. & Fourot, M. (1964) "Etude comparative de l'endothdlium vasculaire et de sa basale, sur coupe et en leucoconcentration," C.R. Seances Soc. Biol. Ses FPl. 158, 137-141. 4. Bouvier, C. A., Maurice, P. A. & Roch, R. (1964) "Resultats obtenus par la leucoconcentration," Bull. Schweiz. Akad. Med. Wiss. 20, 15-26. 5. Spaet, T. H. & Gaynor, E. (1970) "Vascular endothelial damage and thrombosis," Adv. Cardiol. 4, 47-66. 6. Wright, H. P. (1970) "Endothelial turnover," in Vascular Factors and Thrombosis (F. K. Schattauer Verlag, Stuttgart), pp. 79-84. 7. Caplan, B. A. & Schwartz, C. J. (1973) "Increased endothelial cell turnover in areas of in vivo Evans blue uptake in the pig aorta," Atherosclerosis 17, 401-417. 8. Schwartz, S. M. & Benditt, E. P. (1973) "Cell replication in the aortic endothelium: a new method for study of the problem," Lab. Invest. 28,699-707. 9. Schwartz, S. M., Stemerman, M. B. & Benditt, E. P. (1975) "Studies on aortic intima. II. Repair of the aortic lining after mechanical denudation," Am. J. Pathol., in press. 10. Snedecor, G. W. & Cochran, W. G. (1967) in Statistical Methods (Ames, Iowa State University Press), 6th ed., pp. 228-238. 11. Sokal, R. R. & Rohlf, F. J. (1969) in Biometry (W. H. Freeman & Co., San Francisco), pp. 81-95, 560-575. 12. Caplan, B. A., Gerrity, R. G. & Schwartz, C. J. (1974) "Endothelial cell morphology in focal areas of in vivo Evans blue uptake in the young pig aorta. I. Quantitative light microscopic findings," Exp. Mol. Pathol. 21, 102-117.
Pathology
Clustering of replicating cells in aortic endothelium (cell replication/atherosclerosis/artery/endothelial injury)
STEPHEN M. SCHWARTZ AND EARL P. BENDITT Department of Pathology, University of Washington, Seattle, Wash. 98195
Contributed by Earl P. Benditt, November 7,1975
ABSTRACT Cell division in the aortic endothelium of the rat is not randomly distributed. Maps of the aortic surface show focal areas where the daily rate of replication may be in excess of 50%. This implies the existence of focal areas of rapid cell growth or rapid cell turnover and other areas where growth or turnover is largely absent.
five segments. The endothelium from each segment was removed as a sheet, mounted on glass slides, and processed for autoradiography. The final result of this procedure is a preparation of endothelium with the original distribution of cells preserved and with replicating cells identified by autoradiography. Mapping of Specimens. Samples were chosen for mapping based on the quality of preparation, including: (i) retention of most or all of the endothelial surface, (ii) absence of nonendothelial cells, (iii) good quality of staining of nuclei, and (iv) low background grain count. A total of 15 segments was completely mapped, including portions of both thoracic and abdominal aorta. These preparations were examined in the light microscope using a square-shaped reticule in a lOX wide-field eyepiece with a 40X objective. Each specimen was examined by moving the square field systematically across the preparation, recording the number of labeled cells and the total number of cells in each field. The location of each field was recorded on a grid. The grid thus provided a map of the distribution of labeled cells and the population density of the entire specimen. Evaluation of Data. The following parameters were obtained from each specimen: 2N* = total number of labeled cells; ZN = total number of cells; Tf = total number of fields. These numbers were in turn used to calculate the thymidine index (TI), the cell density (CD), and the mean number of labeled cells per field (N*):
Large and small blood vessels are lined on their inner surfaces by a continuous single layer of specialized cells, the endothelium. The endothelium modulates passage of fluid, solutes, and cells from the blood into the adjacent vessel wall and prevents adhesion of platelets (1, 2). Any breach in the integrity of the endothelium leads to local loss of these barrier functions. For this reason, an understanding of how the continuity of the endothelium is maintained becomes important. It is known that endothelial cells are lost into the blood stream in human beings (3-5). From animal studies, it is known that endothelial cells can multiply, as indicated by incorporation of tritiated thymidine into nuclei in the endothelium of normal animals (6-8) and by the regeneration of endothelium following partial denudation of the aorta in experimental animals (9). The rate of aortic endothelial cell replication under normal conditions has been studied in several species and has been estimated to be less than 1% per day (6-8). This low rate of incorporation has been taken to mean that the rate of cell turnover is low. This, however, assumes that cell loss and cell replication are uniformly and randomly distributed over the inner surface of the vessel. It is possible to accurately map essentially the entire surface of the aortic endothelium of small animals with respect to thymidine incorporation and hence to ascertain the distribution of replicating cells (8). We report here a quantitative study of the distribution of multiplying cells in the rat aortic endothelium. In the aorta the distribution of cells by this indicator of mitosis appears to be nonrandom and to present significant numbers of loci of high turnover and large areas of apparently low or possibly absent multiplication.
TI = 2N°/2;N
CD= =N/f N°
=
ZN°/f
TI, the thymidine index, was used as a measure of the replication rate. CD, the cell density, is a measure of the number of cells per unit area. N° is the average number of labeled cells per field.
The map data also allowed us to examine statistically the distribution of labeled cells. Data from each specimen were grouped according to the number of fields with 0 labeled cells, 1 labeled cell, two labeled cells, and so on up to the maximum number of labeled cells present in any one field (Table 1). The mean and the variance of the number of labeled cells per field (10, 11) were then used to test the fit of the distribution to a random distribution, the Poisson. Finally, the actual distribution of labeled cells was compared with the random distribution expected from a binomial distribution. Ten specimens were chosen at random. The actual number of fields showing four or more labeled cells was counted for all 10 specimens. The average cell density (CD) for these specimens was calculated as above. This average cell density, 143 cells per field for the 10 preparations, was used as a sample size in calculating the expected frequency of fields containing four or more labeled cells. This
MATERIALS AND METHODS Animals. The experiments were performed on 10 female Sprague-Dawley rats, about 150 g in weight and 3 months of age. The animals were born and raised in facilities at the University of Washington and were maintained on a standard rat diet (Ralston-Purina Co., St. Louis, Mo.). Labeling Schedule. The dosage schedule consisted of three doses of [methyl-3H]thymidine (New England Nuclear Corp., Boston, Mass.), each dose: 50 mCi/100 g at 8-hr intervals. This schedule was chosen to minimize diurnal variation and to provide large enough numbers of cells to permit accurate counting. In addition, if we assume that S phase is about 8 hr long, this schedule should approximate a 24-hr labeling interval. Tissue was fixed and prepared as previously described (8). The aorta from each animal was divided into 651
Proc. Nat. Acad. Sci. USA 73 (1976)
Pathology: Schwartz and Benditt
652
the binomial distribution with p = labeled cells = 0.6%; k (sample size) = 143; (number samples) = total number of fields in the
was done by generating mean fraction of of and n
10 specimens
=
1926.
RESULTS and focal clusters of replication Overall rate of cell replicating cells Examples of the mapped data are shown in Fig. la, b, and c. The overall thymidine index frequency (TI) in the 15 segments was 0.55% 4 0.1 SE. This low mean value is consistent with observations in other species (6, 7). The labeled cells, however, do not appear evenly distributed. As can be seen in Fig. 1, labeled cells appear clustered in regions of variable size and shape. The labeling frequency within these clusters reaches as high as 60% in individual fields. Probability that clusters could occur by chance The first question that must be asked is whether these clusters could arise by chance. If the clustering occurs as part of a random series of events (10, 11), the number of labeled cells per field should fit a Poisson distribution. For this distribution, the ratio of the variance to the mean, the coefficient of dispersion, approximates 1. Values in excess of 1 indicate nonrandom aggregation or "clustering." Values below 1 represent the opposite phenomenon (11). Table 1 shows the results of this analysis of the 15 mapped segments. The coefficient of dispersion (C) is greater than 1 in all but three specimens. Moreover, when the coefficient is calculated for the entire group of specimens, the value is 2.4, with P < 0.001 by the x2 test (10, 11). Thus we may conclude that the labeled cells are not randomly distributed and that at least some cells are contained in clusters. Other evidence supports the distinction between areas containing labeled cells and the rest of the aorta. If the clusters are real, then replicating cells should be distributed in only a portion of the total area of the aorta. In this case one would expect to find a greater number of fields with rela-
H,,
+;*S Al
-67%
c~~~~~~~~~~~~
FIG. 1. These maps represent frequency distributions for labeled cells in the endothelium of portions of thoracic aorta from three animals. In each case the long axis represents the long axis of the aorta and the long edges represent the dorsal or posterior margin. (a) The dividing cells form a belt around the circumference of the vessel. This belt clearly involves areas of the aortic surface distinct from any vessel branches. The only branches in this segment of the vessel are the intercostal arteries, and these are confined to the dorsal region. (b) A very different pattern occurs, with a zone of labeled cells forming a stripe parallel to the long axis of the vessel. One field shows a labeling frequency greater than 10%. (c) The striking feature is the presence of two adjacent heavily labeled fields, each with an excess of 50% of cells showing the label. Other than these two fields, the remainder of the specimen shows only scattered labeling of cells. The scale is represented by a heavysided square to the right of each figure; the sides of this square represent 0.28 mm, equivalent to the sides of the counting reticule. Each dot on the map represents a labeling frequency of 0.1% in the area of the reticule as it was moved over the specimen. Arrows indicate the craniad and caudad directions.
tively large numbers of labeled cells than if the labeled cells were randomly distributed on the aortic surface. We tested this by looking at the actual number of fields with four or more labeled cells, compared with the number expected
Table 1. Number of fields with one or more labeled cells
men
Total
Number of labeled cells per field
Speci1
1 53 2 30 23 3 4 13 24 5 24 6 76 7 90 8 9 23 10 89 11 17 58 12 13 70 14 21 30 15 Total 641
2
3
4
5
6
24 14 6 11 5 16 31 73 12 48 11 29 64 5 10 359
5 10
3
1
1
1
1
2 7 23 52 5 14 2 8 45 1 14 188
10 11 12 13 14 15 16 17
9
8
no. of
4 3 21 2 5
2 1 10
8
3
2
1 1 1
1
1
3
2
1
1
1 1
1
7 26 2 4 77
1
12
2 10
6
3
23
9
8
1
1
4_ 1
1
1 3
35
1
1
2
0
1
1 1
1
fields
N*
s2
C
344 190 401' 76
0.34 0.58 0.09 0.61 0.63 1.66 1.28 1.65 0.57 1.04 0.76 1.55 2.13 0.75 1.73 1.01
0.45 1.16 0.11 0.96 0.43
1.3 2.0 1.3 1.6 0.7 3.0 1.0 2.1 1.6
54 85 181 393 123 281 70 119 307 63 89 2776
4.92 1.30 3.53 0.89 2.22 1.49 1.42 4.00 1.16 6.24 2.41
2.1 2.0 0.9 1.9 1.6 3.6 2.4
Table 1 shows the distribution of labeled cells in terms of the number of fields with 1 labeled cell, 2 labeled cells, up to 17, the highest number of labeled cells found in any one field. The column labeled "Total no. of fields" represents the sum of all fields, including unlabeled fields. N*, s2, and C for the row labeled "Total" represent these statistics for the data gathered from all 15 specimens. N* = mean number of labeled cells per field; s2 = variance of N*; C = coefficient of dispersion = (s2)/(N*).
Pathology:
Schwartz and Benditt
Proc. Nat. Acad. Sci. USA 73 (1976)
653
from a binomial distribution. Data were gathered from the first 10 mapped specimens. The expected value was 33 fields; the actual value was 108 fields. Finally, it was noted that the number of cells per field was higher in fields containing labeled cells, 153, than in unlabeled fields, 139. The direction of this difference was consistent in maps from these 10 preparations and was statistically significant by C test with P < 0.001. However, the difference in cell densities is small and might be attributable to the increased probability of finding labeled cells in fields with more cells. To test this possibility, we examined the thymidine index of "dense" fields (defined as fields containing at least 1 SD more cells than an average field). The average value for the thymidine index of the "dense" fields was 1.2%, twice the thymidine index for the entire population, 0.55%. This difference is significant at the P < 0.005 level by the t test. Thus the mean number of cells per field is higher in fields containing labeled cells.
injury must include factors other than disturbed flow pattern. Finally, another problem must be considered. We have assumed that the clusters of labeled cells seen in the maps have fixed locations. This may not be true. It is conceivable that different regions of the endothelium would label at other times or that labeled cells, while produced at specific points, later migrate to other regions of the endothelial surface. Data acquired from only a single labeling interval cannot be used to answer this question. In summary, the initial impression that aortic endothelium is long-lived, based on a low rate of cell replication of the population as a whole, represents an oversimplification. Some areas of the aorta may in fact be very long-lived and show little evidence of cell replacement. Other areas, represented by the heavily labeled clusters shown here, either represent foci of residual but rapid growth or represent foci where cells are being rapidly lost and replaced.
DISCUSSION We can conclude that replicating endothelial cells are not randomly distributed on the aortic surface. Some portion of the replicating cells appear clustered in focal regions of the aortic surface. There are at least three possible interpretations of these phenomena: (i) Divson after pulse. Some aggregation of labeled cells could arise by virtue of the fact that division of cells occurs during the labeling interval. However, it is unlikely that this could account for areas of the vessel lining containing 10 to 100 times as many labeled cells as the average or the appearance, in the mapped data, of large numbers of adjacent fields containing labeled cells. (ii) Growth centers. The clusters could represent centers of vascular enlargement or remodeling. It is difficult to be certain that the total number of cells in the endothelial population is in a steady state. With a replication rate as low as 0.6%, the rate of increase in the total number of cells must be quite low. In young animals, clusters of replicating cells may represent focal areas of residual growth, and this presents a special problem for studies in growing animals. (iii) High turnover regions. The clusters could represent sites where cells are being both lost and replaced at high rates. This is consistent with evidence from other studies which indicate increased replication rates in the endothelium near areas of disturbed blood flow, particularly vascular branches (6, 7, 12). However, a pattern of localization only near branches does not appear when the entire surface is mapped as here. While some of the clusters may be located near orifices, other labeled regions were found extending completely about the circumference of the aorta and hence must include areas without branches. Furthermore, heavily labeled regions were found in the anterior portion of the thoracic aorta where no branches occur. Thus, if the labeled areas represent sites of spontaneous injury, the mechanism of
This investigation was supported by National Institutes of Health Grants HL-03174, GM-13543, and GM-00100. 1. Landis, E. M. & Pappenheimer, J. R. (1963) "Exchange of substance through the capillary walls," in Handbook of Physiology (Am. Physiological Soc., Washington), Section 2: Circulation, Vol. II, pp. 961-1034. 2. Stemerman, M. B., Baumgartner, H. R. & Spaet, T. H. (1971) "The subendothelial microfibril and platelet adhesion," Lab. Invest. 24, 179-186. 3. Herbeuval, H. & Fourot, M. (1964) "Etude comparative de l'endothdlium vasculaire et de sa basale, sur coupe et en leucoconcentration," C.R. Seances Soc. Biol. Ses FPl. 158, 137-141. 4. Bouvier, C. A., Maurice, P. A. & Roch, R. (1964) "Resultats obtenus par la leucoconcentration," Bull. Schweiz. Akad. Med. Wiss. 20, 15-26. 5. Spaet, T. H. & Gaynor, E. (1970) "Vascular endothelial damage and thrombosis," Adv. Cardiol. 4, 47-66. 6. Wright, H. P. (1970) "Endothelial turnover," in Vascular Factors and Thrombosis (F. K. Schattauer Verlag, Stuttgart), pp. 79-84. 7. Caplan, B. A. & Schwartz, C. J. (1973) "Increased endothelial cell turnover in areas of in vivo Evans blue uptake in the pig aorta," Atherosclerosis 17, 401-417. 8. Schwartz, S. M. & Benditt, E. P. (1973) "Cell replication in the aortic endothelium: a new method for study of the problem," Lab. Invest. 28,699-707. 9. Schwartz, S. M., Stemerman, M. B. & Benditt, E. P. (1975) "Studies on aortic intima. II. Repair of the aortic lining after mechanical denudation," Am. J. Pathol., in press. 10. Snedecor, G. W. & Cochran, W. G. (1967) in Statistical Methods (Ames, Iowa State University Press), 6th ed., pp. 228-238. 11. Sokal, R. R. & Rohlf, F. J. (1969) in Biometry (W. H. Freeman & Co., San Francisco), pp. 81-95, 560-575. 12. Caplan, B. A., Gerrity, R. G. & Schwartz, C. J. (1974) "Endothelial cell morphology in focal areas of in vivo Evans blue uptake in the young pig aorta. I. Quantitative light microscopic findings," Exp. Mol. Pathol. 21, 102-117.
