THE ANATOMICAL RECORD 218~466-469(1987) Pericytes in Human Cerebral Microvasculature C.R. FARRELL, P.A. STEWART, C.L. FARRELL, AND R.F. DEL MAESTRO Department of Anatomy, University of Toronto, Toronto, Ontario M5S 1A8 (C. R.F., PA.S.), and Brain Research Laboratory, Departments of Clinical Neurological Sciences and Biophysics, University of Western Ontario, Victoria Hospital, London, Ontario N6A 4G5 (C.L.F., R.F.D.), Canada ABSTRACT Two classes of pericytes are thought to exist in cerebral microvasculature, granular and agranular. This classification is based on the presence or absence of cytoplasmic, lysosome-like granules of variable size and appearance. The pericytes in structurally normal human brain tissue from 17 patients, male and female, ranging in age from 14 to 77, were examined. Light microscopic examination of single sections revealed that 67% of pericyte profiles contained granules in both sexes, and this ratio did not change with age. Following the serial reconstructions of 80 individual pericytes, it was found that all contained characteristic cytoplasmic granules. These data show that if truly agranular pericytes do exist in human cerebral microvasculature, they must constitute less than 5% of the population. Pericytes are a class of perivascular cells enclosed within the basement membrane of capillaries. Pericyte function is not well understood even though they have been identified in a number of tissues and are probably common to all microvascular beds. A great deal of confusion regarding the form and distribution of pericytes in the central nervous system (CNS) is evident in the literature. Some authors make no distinction among different types of pericytes (Castejon, 1984; Kristensson and Olsson, 1973; Le Beux and Willemot, 1980). Others refer to two types of pericytes, one containing numerous large, dense granules, which are autofluorescent, acid phosphatase positive, and PAS positive (Jeynes, 19851, and the other apparently without granules, although there is no consistency in the terminology used by different authors. Various distinctions between these two types include: agranular vs. granular pericytes (Jeynes, 1985; Mat0 and Ookawara, 1981); “ordinary” vs. phagocytic pericytes (van Deurs, 1976); pericytes containing numerous primary and secondary lysosomes vs. those with few lysosomes (Lafarga and Palacios, 1975); and those laden with granules vs. unladen pericytes (Sumner, 1982). The idea that there are two classes of pericytes came from the observation that, in single sections, a little more than half of the pericyte profiles contain granules and the rest do not. An alternative explanation, however, is that granules may be present in every pericyte but are not evenly distributed in the cytoplasm, and so would not be seen in every section. The form and distribution of pericytes in human brain are not known. The current study was undertaken to describe pericytes in human cerebral cortex and to re-examine the hypothesis that there are two classes of pericytes with the use of serial reconstruction. MATERIALS AND METHODS Structurally normal cerebral cortex was obtained a t biopsy from patients undergoing lobectomy for removal 0 1987 ALAN R. LISS, INC. of a tumor or corticectomy for treatment of intractable epilepsy.’ The samples were chosen as far from the site of pathology as possible (Stewart et al., 1986) and were judged to be normal. Standard light microscopic neuropathological evaluation of tissue texture, cell density, and organization of appropriate cells for the sites selected was performed. There was a n absence of vacuolization in the tissue around the cells indicating that edema (cytotoxic or vasogenic) was absent. No abnormal cell infiltration or alteration of vessels and endothelial cells in number or topography was judged to be present. Tissue was obtained from 14 tumor patients (13 glial tumors and 1 lymphoma) and 3 epileptic patients. Samples were obtained from the frontal lobe (5 patients), frontal-temporal area (2 patients), parietal lobe (1 patient), occipital lobe (2 patients), parieto-occipital area (2 patients), and the temporal lobe (5 patients). Patients ranged in age from 14 to 77 and consisted of 6 males and 11females. None of the patients had received prior chemotherapy or radiotherapy. Tissue samples were immediately immersed in fixative and cut into small fragments of approximately 1 mm3 within one minute of removal from the brain during surgery. Tissue was fixed in 2% glutaraldehyde and 2% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) and postfixed in osmium tetraoxide and embedded in epon (JBS). Thin sections were cut and stained with toluidine blue for light microscopic observations. Pericytes are undifferentiated mesenchymal cells associated with capillaries and enclosed within the vascu- Received October 2, 1986; accepted March 4, 1987. Address reprint requests to C.R. Farrell, Dept. of Anatomy, University of Toronto, Toronto, Ontario M5S 1A8, Canada. ’Human tissue was obtained through informed consent, and its use in this project was approved by the University of Toronto Human Rights Review Committee, February 27, 1984. PERICYTES IN HUMAN CEREBRAL MICROVASCULATURE lar basal lamina. In the light microscopic preparations used in this study, pericytes were identified by their close association with capillary walls (Fig. la,b). They were scored as granular or agranular depending on the presence or absence of granules. Control studies showed that the characteristic lysosome-like granules that stained dark purple in toliudine blue are also PAS positive. A second, more variable type of granule present in pericytes stained green in toluidine blue and were golden-brown in unstained sections indicating a lipofuscin component (Fig. lb). The effects of age and sex on the mean percentage of granular pericytes were examined using linear regression analysis. To test the hypothesis that there are two classes of pericytes, a total of 80 pericytes were serially reconstructed from 3 patients. Each series consisted of 30-32 continuous sections. Statistical procedures were based on the binomial distribution. The null hypothesis, that the agranular pericytes constitute at least 5% of the TABLE 1. Distribution of granular pericyte profiles in human cerebral cortex patients No. of Total no. of pericytes Mean%’ profiles w i t h granules Females’ 6 11 124 236 65.89 3.26 67.33 f 2.64 Total 17 360 66.98 rf: 1.88 Males2 ‘Mean % k standard error of the mean. 2No difference was found in % pericyte profiles with granules between males and females (p > 0.05). 467 pericyte population, was tested. Power was calculated using P = 100 (l-pn)%where p, representing the proportion of granular pericytes, was set at .95. RESULTS In single sections we found that 66.98% k 1.88 (S.E.M.) of pericyte profiles contained granules and the rest were apparently agranular. There were no significant (p > 0.05) differences due to age or sex in the distribution of granular or agranular profiles in the tissues studied (Table 1). A slight but not statistically significant increase in the percentage of granular pericyte profiles with increasing age was observed (Fig. 2). No qualitative differences in the pericytes were observed among the different patients. We serially reconstructed 80 pericytes and found that they all contained granules. Two basic types of inclusion bodies were observed: dense, amorphous, lysosome-like granules that stained purple in toluidine blue (Fig. la, arrowhead), and large green staining bodies that displayed a greater diversity in morphology (Fig. lb, arrowhead). Both types often occurred together in the same cell. When we began reconstructing pericytes from serial sections and failed to find any agranular ones, it became apparent that if “agranular” pericytes do exist, they must constitute only a small percentage of the population. We arbitrarily chose 5% as a working hypothesis and tested this using the formula shown above. This hypothesis was rejected at p < 0.02. Statistically this means more than 95% of the pericytes in the cortical tissue studied were granular and that if “agranular” pericytes do exist, they could constitute no more than 5% of the pericyte population. Figure 3a-c. shows a Fig. 1 . These micrographs show the close association of pericytes and capillary profiles in human cerebral cortex. Two types of granules are present in the cytoplasm of the pericytes in a and b. Dense, lysosome-like granules are indicated in a (arrowhead), and the larger, more heterogeneous variety of inclusion bodies indicated in b (arrowhead). The brilliant green color of this second type of body is not obvious from the black-and-white micrograph in b (1,200X 1. C.R. FARRELL, P.A. STEWART, C.L. FARRELL, AND R.F. DEL MAESTRO 468 " o % Pericyte 6o Profiles with Granules 4 0 e 0 tb 20 30 40 '- m 0 0 50 0 0 60 7.0 80 Age of Patients [years) Fig. 2. Graph showing the distribution of granular pericytes in human cerebral cortex. Male and female patients are represented by closed and open dots, respectively. The change in the percentage of pericytes containing granules is represented, and although a slight increase with age is evident, it is nonsignificant. sequential series of sections of the same pericyte in which granules are present in two sections but not in the third. DISCUSSION Two types of inclusion bodies in pericytes of the CNS have been described: large, spherical, lysosome-like electron dense bodies with homogeneous matrices (Cancilla et al., 1972; Castejon, 1984; Jeynes, 1985; Lafarga and Palacios, 1975; Mat0 and Ookawara, 1981), and large bodies with polymorphic matrices including dense granulations and possibly lipid vacuoles (Jeynes, 1985; Lafarga and Palacios, 1975) and lipofuscin aggregates (Kristensson and Olsson, 1973). In the present study both of these types were observed. The larger heterogeneous type of inclusion body displayed considerable diversity. Pale areas were sometimes observed in these bodies and they often assumed irregular shapes. In some cells these inclusion bodies were very large, occupying much of the cytoplasm and in some cases obscuring part of the nucleus. These two types of inclusion bodies occurred together in many of the cells analyzed. Pericytes may be a n important part of the blood-brain barrier acting as a "second line of defense" by phagocytosing molecules that pass the endothelial layer when it is stressed (Cancilla et al., 1972; van Deurs, 1976). Our finding that the majority (> 95%) of CNS pericytes are of the granular form strengthens this hypothesis that pericytes play a n important back-up role in the bloodbrain barrier. The phagocytic properties of pericytes in the brain have been well documented. Under experimental conditions such as acute hypertension (van Deurs, 19761, freeze-injury (Baker et al., 1971; Cancilla et al., 19721, and portocaval anastomosis (Sumner, 1982), in which the blood-brain barrier becomes leaky, granulecontaining pericytes have been observed to be actively involved in the phagocytosis of foreign proteins that entered the brain. The phagocytic role of CNS pericytes under pathological conditions such as neoplasms and allergic encephalomyelitis has also been established (Torack, 1961). Furthermore under such conditions peri- Fig. 3. A series of 3 micrographs of the same pericyte associated with a capillary from human cerebral cortex. In a, numerous granules (arrowhead) are present in the pericyte cytoplasm; in b, one granule may be present; in c, granules are completely lacking (1,200x). PERICYTES IN HUMAN CEREBRAL MICROVASCULATURE cytes and their lysosomes increase in number (Cancilla et al., 1972; Mat0 and Ookawara, 1981; Van Deurs, 1976). With advancing age granules have been observed by others to become more closely packed or replaced by large vesicles (Cammermeyer, 1970),but the phagocytic capabilities of pericytes also declines in aging (Mato and Ookawara, 1981). In a previous study we found a decrease in the number of pericytes with increasing age in human cerebral white matter but not gray matter (Stewart et al., 1986). Such factors may contribute to the pathology of aging by decreasing the ability to compensate for transient leaks in the blood-brain barrier. Since the number and size of lysosomelike granules in pericytes increases with age (Cammermeyer, 1970; Mat0 and Ookawara, 19811, the chances of sectioning a pericyte through a granule should also increase. The slight (but not statistically significant) increase in the percentage of pericyte profiles with granules may reflect this change and should be further investigated. Another possible function attributed to pericytes in the CNS is that of mechanical support of small blood vessels to counteract hydrostatic pressure. In one study acute hypertension caused the blood-brain barrier to break down, whereas the blood-retina1 barrier, which contains a higher density of pericytes, remained intact and undamaged (Laties et al., 1979). A contractile function of pericytes has also been postulated. Actin and myosin have been identified in the pericytes of rat brains (Le Beux and Willemot, 1980). Actin has been identified in retinal pericytes, and although it was also found to be present in retinal endothelial cells, filaments were more numerous in pericytes (Wallow and Burnside, 1980). Thus, if pericytes have the ability to contract, then the filaments may be involved in the regulation of capillary blood flow. However, the mere presence of actin and myosin is not proof of cellular contractility since these proteins play a central role in the phagocytic process (Silverstein et al., 1977). We have shown that the majority (> 95%)of pericytes in the CNS contain cytoplasmic granules, which are known to be associated with the phagocytic role of pericytes (as above). That we were unable to identify any “agranular” pericytes casts doubt on the hypothesis that they exist as a separate class from granular ones. If they do exist, they are rare. We suggest that the presence of filaments in pericytes may be a reflection of their phagocytic capabilities. In addition it is possible that CNS pericytes are also contractile as they are in skeletal muscle (Tilton et al., 1979) and so may play a role in vasoconstriction. Our failure to find any “agranular” pericytes also has implications for future work elucidating the functions of cerebral pericytes since interpretation of data is no longer constrained bv the necessity to explain two different-morphological types of CNS pericyt; 469 ACKNOWLEDGMENTS This work was supported by the National Cancer Institute (Canada) and the Brain Research Fund Foundation. R.F.D. is a recipient of the Canadian Life Insurance Medical Scholarship. The authors wish to thank Dr. J. Girvin for supplying tissue from epilepsy patients and Mrs. K. Hayakawa for her excellent technical assistance. The authors are grateful to Drs. J.J. Gilbert and J.F.C. Kaufmann for their neuropathological expertise. LITERATURE CITED Baker, R.N., P.A. Cancilla, P.S. Pollock, and S.P. Frommes (1971) The movement of exogenous protein in experimental cerebral edema. J. Neuropathol. Exp. Neurol., 80:668-679. Cammermeyer, J. (1970) The history of the microglial cell: A light microscopic study. In: Neurosciences Research, vol. 3. S. Ehrenpreis and O.C. Solnitzky, eds. New York, Academic Press, pp. 44170. Cancilla, P.A., R.N. Baker, P.S. Pollock, and S.P. Fommes (1972) The reaction of the central nervous system to exogenous protein. Lab. Invest., 26:376-383. Castejon, O.J. (1984) Submicroscopic changes of cortical capillary pericytes in human perifocal brain edema. J. Submicrosc. Cytol., 16:601-618. Jeynes, B. (1985) Reactions of granular pericytes in a rabbit cerebrovascular ishemia model. Stroke, 16:121-125. Kristensson, K., and Y. Olsson (1973) Accumulation of protein tracers in pericytes of the central nervous system following systemic injection in immature mice. Acta Neurol. Scand., 49:189-194. Lafarga, M., and G. Palacios (1975) Ultrastructural study of pericytes in the rat supraoptic nucleus. J. Anat., 120:433-438. Laties, A.M., S.I., Rapoport, and A. McGlinn (1979) Hypertensive breakdown of cerebral but not retinal blood vessels in rhesus monkey. Arch Opthalamol., 79:1511-1514. Le Beux, Y.V.I., and J. Willemot (1980) Actin and myosin-like filaments in retinal pericytes and endothelial cells. Invest. Opthalamol. Vis. Sci., 19:1433-1441. Mato, M., and S. Ookawara (1981) Influences of age and vasopresson on the uptake capacity of fluorescent granular perithelial cells (FGP) of small cerebral vessels of the rat. Amer. J. Anat., 162:4553. Silverstein, S.C., R.M. Steinman, and Z.A. Cohn (1977) Endocytosis. Ann. Rev. Biochem., 46:699-722. Stewart, P.A., M. Magliocco, K. Hayakawa, C.L. Farrell, R.F. Del Maestro, J. Girvin, J. Kaufmann, H. Vinters, and J. Gilbert (1987) Blood-brain barrier ultrastructural changes in the aging human. Microvasc. Res. (in press). Sumner, B.E.H. (1982) A quantitative study of vascular permeability to horseradish peroxidase, and the subsequent fate of ihe tracer, in rat brains after postcaval anastomosis. Neuropathol. Appl. Neurobiol., 8:117-133. Tilton, R.G., C. Kilo, J.R. Williamson, and D.W. Muich (1979) Differences in pericyte contractile function in rat cardiac and skeletal muscle microvasculature. Microvasc. Res. 18:336-352. Torack, R.M. (1961) Ultrastructure of capillary reaction to brain tumors. Arch Neurol., 5:416-428. van Deurs, B. (1976) Observations on the blood-brain barrier in hypertensive rats, with particular reference to phagocytic pericytes. J. Ultrastr. Res., 56:65-77. Wallow, H.I., and B. Burnside (1980) Actin filaments in retinal pericytes and endothelial cells. Invest. Opthalmol. Vis. Sci. 19:14331441.