Blood–brain barrier integrity is unaltered in human brain cortex with diabetes mellitus

Brain Research 954 (2002) 311–316 www.elsevier.com / locate / bres

Short communication

Blood–brain barrier integrity is unaltered in human brain cortex with diabetes mellitus Jiapei Dai a,b , *, Gijs F.J.M. Vrensen c , Reinier O. Schlingemann a a

Ocular Angiogenesis Group, Department of Ophthalmology, Academic Medical Center, Amsterdam, The Netherlands b The Netherlands Ophthalmic Research Institute, Meibergdreef 47, 1105 BA Amsterdam, The Netherlands c Leiden University Medical Center, Leiden, The Netherlands Accepted 4 June 2002

Abstract Diabetes-related cognitive dysfunction has been recognized for many years in humans, but the pathogenesis of this condition is poorly understood. Evidence from animal studies suggests that altered function of the blood–brain barrier (BBB) could be a potential cause contributing to this disease. This study aimed to investigate whether the permeability of the BBB is affected in the brains of persons with diabetes mellitus (DM). On postmortem prefrontal and temporal cortex of diabetic patients and controls, immunohistochemical stainings were carried out using specific antibodies against three proteins (PAL-E, IgG and albumin), which are considered as markers for the vascular permeability status of the BBB. Rare or no PAL-E staining was found in the capillaries of the prefrontal and temporal cortex parenchyma, in both DM and control materials. IgG and albumin were localized in and directly around blood vessel walls in the prefrontal and temporal cortex. No obvious differences in the staining pattern of IgG and albumin were observed between brain samples of persons with DM and controls. This study suggests that the BBB in diabetic patients is well maintained.  2002 Elsevier Science B.V. All rights reserved. Theme: Neural basis of behavior Topic: Cognition Keywords: Diabetes mellitus; Vascular permeability; Blood–brain barrier; PAL-E; IgG; Albumin

Diabetes-related cognitive dysfunction has been recognized for many years in humans. The pathogenesis of this condition is poorly understood, but increasing evidence from animal studies suggests that altered function and structure of the blood–brain barrier (BBB) could be a potential cause [1]. Under normal conditions, the BBB minimizes the entry of unwanted molecules into brain. This restriction is mainly accomplished by tight junctions between adjacent endothelial cells, the ionic charges on the surface of endothelial cells, and by limited pinocytotic activity [2]. Under diabetic conditions, chronic uncontrolled hyperglycemia and complications commonly associated with diabetes mellitus (DM), such as transient cerebral ischemia and systemic hypertension may disrupt the BBB integrity, therefore allowing molecules which are normally *Corresponding author. Tel.: 131-20-5666-101; fax: 131-20-5666121. E-mail address: [email protected] (J. Dai).

confined to the blood to enter into brain parenchyma. Although functional and structural BBB changes in cerebral microvessels have been observed in long-term streptozotocin-induced diabetic animals [3], no studies on the vascular permeability of the BBB have been carried out in diabetic humans. In the present study, we investigated, by using postmortem brain tissue and immunohistochemistry for an endothelium-specific antigen PAL-E [4–6], and two plasma proteins, IgG and albumin, which are considered as markers for a change of the vascular permeability [7–10], whether the permeability of BBB is affected in the diabetic patients. Brain tissues (prefrontal and / or temporal cortex) from 12 diabetic patients (five cases with diabetes mellitus type I and seven with diabetes mellitus type II) and 12 age-, sex- and postmortem delay-matched controls were collected by the Netherlands Brain Bank (NBB, Coordinator Dr. R. Ravid) by autopsy. According to the protocol of the NBB, and in agreement with the Declaration of Helsinki,

0006-8993 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved. PII: S0006-8993( 02 )03294-8

J. Dai et al. / Brain Research 954 (2002) 311–316

312

specific permission for brain autopsy and use of the brain and medical records for research purposes were obtained either from the patients themselves or from partners or relatives. Following removal of the brain from the skull, small pieces of mid prefrontal and temporal cortex gyrus (about 131 cm) were dissected and snap-frozen in liquid nitrogen and stored till use at 270 8C. For the diabetic persons the age ranged between 48 and 87 years (mean 70.3 years) and postmortem delay ranged between 3:20 (h:min) and 10:20 (mean 5:39). For the control persons the age ranged between 45 and 88 years (mean 72.1 years) and postmortem delay ranged between 3:45 (h:min) and 8:50 (mean 6:51). Clinicopathological information is given in Table 1. The brain tissue was cut into 10-mm cryo-sections on a cryostat. The sections were fixed in cold acetone for 10 min. Sections were stained by immunohistochemistry for the endothelial PAL-E antigen, a marker for non-BBB endothelium [4–6], and IgG and albumin, two plasma proteins serving as endogenous tracers indicative of increased vascular permeability [7–10]. A direct immunoperoxidase staining procedure was carried out for IgG and an indirect one for PAL-E and albumin. All incubations were performed at room temperature unless stated otherwise. In order to reduce non-specific staining, sections

were pre-incubated with 10% normal horse serum in TBS (0.05 M Tris 10.9% NaCl, 10.05% saponin, pH 7.6) for 15 min. Then sections were incubated for 1 h and overnight at 4 8C with the following antibodies: mouse monoclonal anti-human IgG antibodies conjugated to HRP (Dako, P0214, 1:800), mouse monoclonal antibody PAL-E (1:1000), mouse monoclonal EN-4 (Sanbio, The Netherlands) recognizing CD31 (platelet endothelial cell adhesion molecule-1) [11], and rabbit polyclonal antibodies recognizing albumin (Dako A001, 1:4000). Sections were rinsed in TBS for 3310 min and incubated with the secondary antibody (horse anti-mouse IgG and goat anti-rabbit IgG conjugated to biotin as appropriate) for 1 h. Then the sections were rinsed in TBS for 335 min and incubated in a streptavidin–biotinylated horseradish peroxidase complex (ABC) for 1.5 h. Sections were rinsed in TBS for 335 min and incubated in 0.05% DAB activated with 0.01% H 2 O 2 for 10 min, the reaction was terminated by rinsing the sections with tap water. The sections incubated with anti-IgG were rinsed in TBS for 3310 min and directly stained with DAB after overnight incubation. Counter staining was performed with hematoxylin. Sections were washed and coverslipped in Entellan. Sections were viewed by two independent observers. Immunohistochemical staining for EN-4 in the sections close to

Table 1 Clinicopathological data of diabetes mellitus subjects and controls NBBn

Sex

Age (years)

BW (g)

CSF (pH)

PMD (h)

Diabetes mellitus type I 94-012 M 97-114 F 97-025 M 98-049 M 97-145 M

64 87 79 87 48

1050 1022 1158 1379 1438

6.26 6.17 6.67 6.80 6.34

3:40 4:20 7:05 7:25 6:55

Diabetes mellitus type II 90-102 M 95-015 F 96-072 F 97-110 F 98-022 M 95-101 F 96-010 M

49 78 84 54 78 73 63

1426 1005 1021 1089 1245 1304 1250

6.17 6.75 6.68 6.14 6.72 6.38 6.37

4:25 5:15 3:20 3:55 6:00 5:30 10:20

Controls 95-011 97-008 94-051 92-032 94-060 91-123 96-084 94-074 91-116 95-093 98-104 93-133

62 88 79 87 45 71 78 85 54 78 74 64

1352 1159 1250 1262 1440 1135 1330 925 1600 1440 1207 1448

7.33 6.74 6.62 7.11 6.71 8.10 6.60 6.95 6.46 6.96 6.95 6.90

6:35 3:45 7:15 8:00 8:50 7:25 7:30 5:11 5:15 7:00 7:25 8:09

M F M M M F F F F M F M

Clinical diagnosis, neuropathology, cause of death AD, dementia AD, mortification Myocardial infarctions, renal insufficiency Cardiac arrest MS, uncontrolled diabetes mellitus AD, Presenile dementia AD, aspiration pneumonia AD, cachexia and dehydration AD, Presenile dementia, pneumonia AD, urinary tract infection, shock Heart failure Myocardial infarction, lung emphysema Metastasized adenocarcinoma Urinary tract infection, cachexia Myocardial infarction Pancreas carcinoma with metastases Kidney insufficiency, sepsis, shock Kidney insufficiency, cardiac arrest Terminal pulmonary emphysema, heart failure Pneumonia Brain tumor (between cerebellum and pons) Cardiac decompensation, pulmonary embolism Unstable pectoral angina, necrosis of the intestine Chronic myelocytic leukemia, thrombocytopenia

NBBn, Netherlands Brain Bank number; M, male; F, female; y, years; BW, brain weight; g, gram; CSF, cerebrospinal fluid; PMD, postmortem delay; AD, Alzheimer’s disease; MS, multiple sclerosis.

J. Dai et al. / Brain Research 954 (2002) 311–316

313

Table 2 Semiquantitative analysis of IgG and Albumin (Alb) staining in prefrontal cortex (PFC) and / or temporal cortex (TC) of patients with diabetes mellitus (DM, type I and type II) and in controls Diabetes mellitus type I NBBn

94-012 97-114 97-025 98-049 97-145

Controls

PFC

TC

NBBn

IgG

Alb

IgG

Alb

nd nd nd 11 111

nd nd nd 1 1

11 1 11 nd nd

11 11 11 nd nd

Diabetes mellitus type II NBBn

90-102 95-015 96-072 97-110 98-022 95-101 96-010

95-011 97-008 94-051 92-032 94-060

PFC

TC

IgG

Alb

IgG

Alb

nd 111 nd 1 111

nd 111 nd 1 11

111 11 1 2 nd

111 11 2 1 nd

Controls

PFC

TC

NBBn

IgG

Alb

IgG

Alb

1 111 11 1 1 11 1

1 111 1 111 11 11 11

2 nd nd 1 111 nd nd

2 nd nd 1 111 nd nd

91-123 96-084 94-074 91-116 95-093 98-104 93-133

PFC

TC

IgG

Alb

IgG

Alb

2 111 11 2 11 11 11

2 111 11 11 1 11

2 111 1 11 11 11

2 111 1 2 1 11

NBBn, Netherlands Brain Bank number. 2, rare or no staining around capillaries; 1, a little diffusion staining around a few number of capillaries; 11, moderate diffusion staining around moderate number of capillaries; 111, more diffusion and intensive staining than moderate around a large number of capillaries as shown in an example in Fig. 1D,F. nd, not determined.

PAL-E, IgG and albumin staining sections was used to identify the microvasculature. Based on EN-4 staining, the staining pattern and vascular distribution of PAL-E, IgG and albumin in the brain vasculature, interstitium, and parenchyma was compared between the brains of persons with DM and controls. A semiquantitative analysis of IgG and albumin (Alb) staining in prefrontal and / or temporal cortex of patients with diabetes mellitus (DM) and controls was carried out as shown in Table 2. Rare or no PAL-E staining was found in the BBB capillaries of prefrontal and temporal cortex in persons with DM and in controls (Fig. 1A,B). No difference in PAL-E staining was observed between the two groups. As described previously [4], the non-BBB capillaries of the choroid plexus showed staining of PAL-E, serving as a positive control of PAL-E staining in these tissue samples. IgG and albumin staining was localized along the basal lamina of the microvasculature and some IgG staining was observed in the area surrounding blood vessels within the adjacent parenchyma (Fig. 1C–F). Many glial cells were also positive for albumin in some DM and control cases (Fig. 1C,D), but not for IgG (Fig. 1E,F). Also for these endogenous plasma markers of vascular permeability, we found no differences in the staining pattern in the prefrontal and temporal cortex from the brains of persons with DM and controls, suggesting that extravasation of these plasma proteins was not higher in the diabetic persons than in controls (Fig. 1C–F). The observation on albumin and IgG staining in the samples of the individuals with DM (types I and II) and the controls are given in Table 2.

PAL-E is an endothelium-specific molecule present on the endothelial luminal surface, in the cytoplasm, and in endothelial pinocytotic vesicles, which are involved in transcellular transport. PAL-E is extensively expressed in vessels of most tissues, but is absent in tissues with blood–tissue barriers such as normal retina, brain and testis [4]. In diabetic retinopathy and in brain tumors, both conditions associated with loss of the BBB, PAL-E is expressed in capillaries [5,6]. Therefore, PAL-E staining appears a useful cellular vascular marker for the absence or loss of vascular blood–tissue barriers, indicating absence or loss of BBB integrity in normal tissues or pathological conditions. Another histological method to detect loss of the BBB is the use of endogenous tracers: an increase of the vascular permeability of the BBB results in increased leakage of plasma proteins such as albumin and IgG to the brain parenchyma or retina, as demonstrated in various studies in animals and humans [7–10]. These proteins are therefore useful as endogenous markers of increased permeability in BBB vasculature. It should be emphasized that several factors such as head trauma, etc., that are related to the cause of death may affect vascular permeability to IgG and albumin. Therefore we selected patients who had DM but did not have obvious factors affecting the BBB. We also found that postmortem delay would not affect the PAL-E, albumin and IgG staining in our study. Although most DM patients were diagnosed as Alzheimer’s disease (AD) and the typical neuropathological changes were found, however, according to a previous report [10] and an un-

314

J. Dai et al. / Brain Research 954 (2002) 311–316

J. Dai et al. / Brain Research 954 (2002) 311–316

315

Fig. 1. Micrographs of immunohistochemical staining for PAL-E (A,B), albumin (C,D) and IgG (E,F) in the cortex of patients with diabetes mellitus (DM, left) and in controls (right). Rare PAL-E staining was detected in very few capillaries and small vessels of the cortex in DM and control (arrow in A and B). No obvious differences in albumin and IgG staining were observed between DM and control samples: albumin staining patterns were similar in the capillaries of the cortex (arrows) in DM (C) and control (D), and many glial cells were positive (arrowhead). IgG staining pattern in the capillaries of the cortex (arrows) in DM (E) and control (F) is also similar. Bar52 mm for (A)–(F).

published study on the brain cortex of AD without DM by our group, no obvious change of vascular permeability to IgG and albumin was found, suggesting that AD may not be an important factor to affect the results. Disruption of the blood–retinal barrier (BRB) plays an important role in the development of diabetic retinopathy, and often causes loss of vision in patients with DM. In a recent study, we demonstrated that many blood vessels in the retina of patients with DM have increased expression of PAL-E [12], co-localizing with perivascular staining of endogenous fibrinogen, IgG and albumin. In the retina of control persons no perivascular staining for these endogenous tracers was observed [6]. These findings are consistent with the retinal vascular leakage observed clinically and by ‘fluorescein angiography’ in patients with diabetic retinopathy. In the present study, we were unable to detect similar changes in cerebral vessels in persons with diabetes. We did not observe a change in PAL-E staining in the brains of persons with DM as compared to controls, nor in the degree of extravasated IgG and albumin. In contrast to the complete absence of perivascular staining of these endogenous tracers in the retina of controls [9], a pattern of limited perivascular staining was observed in the present study around brain blood vessels in both controls and diabetic persons. It is unknown whether this extravascular staining around brain vessels represents a difference in permeability between the BBB and BRB or has another cause. Nevertheless, our results clearly indicate that the integrity of the BBB is not grossly changed in the prefrontal and temporal cortex of persons with DM. To our knowledge, this is the first study on BBB function in humans with DM. Although this study only investigated the brain cortex of a limited number of persons with DM, our findings are interesting in the light of the proposed pathogenetic mechanisms of cognitive dysfunction in persons with DM. These mechanisms are derived from work on experimental models of DM in small rodents. Several of these studies observed an increased permeability of the BBB, especially in a short-term diabetic animal model [13]. Other animal studies showed that the BBB permeability to albumin in long-term diabetic rats could be normalized following insulin treatment [14]. The variable results in animal studies may be due to the differences in the experimental models and might also depend on the time scale of the experiments and the limitations of the methods used to measure the permeability of the BBB. The situation in our diabetic patients, who most likely have long standing DM and some form of

treatment, may be similar to that in long-term animal models treated with insulin. Although cognitive dysfunction was not systemically tested in these patients, according to medical records, most patients with DM had cognitive dysfunction, as indicated in Table 1, and some patients were diagnosed as AD. Since epidemiological studies have provided evidence that DM almost doubles the risk of dementia [15], it is possible that DM may contribute to the clinical syndrome in dementia patients such as AD. Although our study fails to provide evidence for a gross dysfunction of the BBB in DM, a mechanism proposed to be causal in diabetes-related cognitive dysfunction, it is possible that in living diabetic patients BBB dysfunction may be intermittent and transient, e.g., in phases of uncontrolled hypo- or hyperglycemia and this could be related to diabetic brain damage and cognitive dysfunction.

Acknowledgements This work was supported by ‘Landelijke Stichting voor Blinden en Slechtzienden’ and by ‘Diabetes Fonds Nederland’ (Grants 95.103 and 99.050). We are grateful to Professor Dr. D. F. Swaab of the Netherlands Institute for Brain Research for his comments on the manuscript. The brain material was obtained from the Netherlands Brain Bank (Coordinator Dr. R. Ravid).

References [1] A.D. Mooradian, Central nervous system complications of diabetes mellitus—a perspective from the blood–brain barrier, Brain Res. Brain Res. Rev. 23 (1997) 210–218. [2] N.R. Saunders, M.D. Habgood, K.M. Dziegielewska, Barrier mechanisms in the brain. I. Adult brain (brief review), Clin. Exp. Pharmacol. Physiol. 26 (1999) 11–29. [3] J. Jakobsen, G.M. Knudsen, M. Juhler, Cation permeability of the blood–brain barrier in streptozotocin-diabetic rats, Diabetologia 30 (1987) 409–413. [4] R.O. Schlingemann, G.T.A.M. Bots, S.G. Van Duinen, D.J. Ruiter, Differential expression of endothelium-specific antigen PAL-E in vasculature of brain tumors and preexistent brain capillaries, Ann. NY Acad. Sci. 529 (1988) 111–114. [5] S. Leenstra, D. Troost, P.K. Das, N. Claessen, A.E. Becker, D.A. Bosch, Endothelial cell marker PAL-E reactivity in brain tumor, developing brain, and brain disease, Cancer 72 (1993) 3061–3067. [6] R.O. Schlingemann, P. Hofman, L. Anderson, D. Troost, R. Van der Gaag, Vascular expression of endothelial antigen PAL-E indicates absence of blood–ocular barriers in the normal eye, Ophthalmic Res. 29 (1997) 130–138.

316

J. Dai et al. / Brain Research 954 (2002) 311–316

[7] P.J. Robinson, Measurement of blood–brain barrier permeability (Review), Clin. Exp. Pharmacol. Physiol. 17 (1990) 829–840. [8] K.C. Ma, Y. Olsson, Structural and vascular permeability abnormalities associated with lacunes of the human brain (Review), Acta Neurol. Scand. 88 (1993) 100–107. [9] S.A. Vinores, Assessment of blood–retinal barrier integrity, Histol. Histopathol. 10 (1995) 141–154. [10] H.M. Wisniewski, A.W. Vorbrodt, J. Wegiel, Amyloid angiopathy and blood–brain barrier changes in Alzheimer’s disease, Ann. NY Acad. Sci. 826 (1997) 161–172. [11] F. Aroca, W. Renaud, C. Bartoli, C. Bouvier-Labit, D. FigarellaBranger, Expression of PECAM-1 / CD31 isoforms in human brain gliomas, J. Neurooncol. 43 (1999) 19–25.

[12] R.O. Schlingemann, P. Hofman, G.F. Vrensen, H.G. Blaauwgeers, Increased expression of endothelial antigen PAL-E in human diabetic retinopathy correlates with microvascular leakage, Diabetologia 42 (1999) 596–602. [13] B. Oztas, M. Kucuk, Blood–brain barrier permeability in streptozotocin-induced diabetic rats, Med. Sci. Res. 15 (1982) 645–646. [14] W.T. Stauber, S.H. Ong, R.S. McCuskey, Selective extravascular escape of albumin into the cerebral cortex of the diabetic rat, Diabetes 30 (1981) 500–503. [15] A. Ott, R.P. Stolk, F. van Harskamp, H.A. Pols, A. Hofman, M.M. Breteler, Diabetes mellitus and the risk of dementia: The Rotterdam Study, Neurology 53 (1999) 1937–1942.