B-vitamin deficiency causes hyperhomocysteinemia and vascular cognitive impairment in mice Aron M. Troen*, Melissa Shea-Budgell, Barbara Shukitt-Hale, Donald E. Smith, Jacob Selhub, and Irwin H. Rosenberg Jean Mayer U.S. Department of Agriculture Human Nutrition Research Center on Aging, Tufts University, 711 Washington Street, Boston, MA 02111-1524 Communicated by Leon E. Rosenberg, Princeton University, Princeton, NJ, June 5, 2008 (received for review July 20, 2007)

In older adults, mildly elevated plasma total homocysteine (hyperhomocysteinemia) is associated with increased risk of cognitive impairment, cerebrovascular disease, and Alzheimer’s disease, but it is uncertain whether this is due to underlying metabolic, neurotoxic, or vascular processes. We report here that feeding male C57BL6/J mice a B-vitamin-deficient diet for 10 weeks induced hyperhomocysteinemia, significantly impaired spatial learning and memory, and caused a significant rarefaction of hippocampal microvasculature without concomitant gliosis and neurodegeneration. Total hippocampal capillary length was inversely correlated with Morris water maze escape latencies (r ⴝ ⴚ0.757, P < 0.001), and with plasma total homocysteine (r ⴝ ⴚ0.631, P ⴝ 0.007). Feeding mice a methionine-rich diet produced similar but less pronounced effects. Our findings suggest that cerebral microvascular rarefaction can cause cognitive dysfunction in the absence of or preceding neurodegeneration. Similar microvascular changes may mediate the association of hyperhomocysteinemia with human age-related cognitive decline. cerebrovascular 兩 homocysteine 兩 mouse 兩 nutrition

C

erebrovascular disease is a major cause of cognitive decline in old age. All of its manifestations, including stroke, white matter disease, cerebral large vessel disease (atherosclerosis), and small vessel disease (arteriosclerosis) can lead to vascular cognitive impairment (VCI), ranging in severity from subtle neuropsychological deficits to frank dementia (1). Although these conditions may occur on their own, they frequently coexist with other neurodegenerative pathologies, such as Alzheimer’s disease. The fact that Alzheimer’s-type dementia shares many of the same risk factors with cerebrovascular disease suggests that at least some aspects of age-related cognitive decline, cerebrovascular, and neurodegenerative diseases share a vascular etiology (2). Mildly elevated plasma total homocysteine (tHcy) is an important risk factor for both cerebrovascular and Alzheimer’s disease. Numerous epidemiological studies find micromolar increments in tHcy to be associated with significantly increased risk of stroke, white matter disease, and cognitive dysfunctions that range in severity from mild cognitive impairment to Alzheimer’s disease (3). Several hypotheses have been advanced to explain how homocysteine might cause cognitive decline, generally invoking homocysteine-mediated neurotoxicity, vasotoxicity, or the inhibition of methylation reactions (4). While these mechanisms are plausible, the in vivo evidence for their occurrence is not convincing. Exposing animals to elevated plasma homocysteine through genetic, pharmacologic, and dietary manipulations has so far failed to produce clear evidence of homocysteinemediated neurotoxicity (5). There is stronger evidence for a direct link between impaired homocysteine metabolism and cerebrovascular disease. Structural and functional vascular changes have been observed in the cerebral arterioles of mice with genetic defects in homocysteine metabolism that were fed diets lacking in folate and high in methionine (6, 7). Cerebrovascular abnormalities have also been induced in WT mice and rats with experimental homocysteinemia (8, 9). However, the 12474 –12479 兩 PNAS 兩 August 26, 2008 兩 vol. 105 兩 no. 34

design of these studies makes it difficult to single out vitamin deficiency, excess methionine, or homocysteine as a primary cause of vascular dysfunction. Moreover, studies showing the association between vascular changes and cognition in experimental animals with hyperhomocysteinemia are lacking (5). To better understand the role of hyperhomocysteinemia in cognitive impairment, we examined the relationship between impaired homocysteine metabolism and neurodegenerative, cerebrovascular, and cognitive outcomes in a mouse model of dietary hyperhomocysteinemia. We fed control or homocysteine-inducing diets to male WT C57Bl6/J mice for 10 weeks. The control group consumed an AIN93M diet containing 0.33% methionine, 2-mg folic acid, 25-␮g cyanocobalamin (vitamin B12), and 7-mg pyridoxal L-phosphate (vitamin B6) per kg diet. Two different diets were formulated to induce hyperhomocysteinemia, the one through combined folate, vitamin B12, and vitamin B6 deficiency, the other through methionine enrichment with 1% L-methionine (10 g L-methionine/kg diet). Both diets induce hyperhomocysteinemia; however, they do so through markedly different metabolic impairments. B-vitamin deficiency inhibits homocysteine’s conversion to methionine or cysteine, causing it to accumulate while methionine is depleted. In contrast, high methionine intake drives the excessive synthesis of homocysteine and cysteine, but without limiting methionine. Before tissue harvest, spatial learning and memory were evaluated by the Morris water maze, and psychomotor function was evaluated by a battery of age-sensitive tests, including the accelerating rotarod. Using these tests, we have shown that the B-vitamin-deficient diet impairs ability to complete the Morris water maze but not psychomotor performance in Apolipoprotein-E-deficient mice (10). Blood homocysteine and B-vitamins were measured and hippocampal total capillary length and microglia numbers were quantified by unbiased stereology immunolabeling capillaries with an antibody to the endothelial glucose transporter type 1 (Glut-1) and microglia with an antibody to the microglial cell surface protein CD11b/Mac-1. For further detail, see Methods below. Results After 10 weeks of consuming the diets, control mice gained significantly more body weight (24.4 ⫾ 1.2 g mean ⫾ SD) than mice fed the high methionine (22.0 ⫾ 1.0 g) or B-vitamindeficient diets (22.1 ⫾ 1.3 g) (F ⫽ 13.8, P ⬍ 0.001). In contrast, brain weight was not affected by diet with a mean wet weight of 447 ⫾ 24 mg for controls, 452 ⫾ 6 mg for high methionine, and 450 ⫾ 13 mg for mice fed B-vitamin-deficient diets (F ⫽ 0.2, P ⫽ Author contributions: A.M.T., B.S.-H., J.S., and I.H.R. designed research; A.M.T., M.S.-B., and D.E.S. performed research; A.M.T., B.S.-H., J.S., and I.H.R. analyzed data; and A.M.T. wrote the paper. The authors declare no conflict of interest. Freely available online through the PNAS open access option. *To whom correspondence should be addressed. E-mail: [email protected]. This article contains supporting information online at www.pnas.org/cgi/content/full/ 0805350105/DCSupplemental. © 2008 by The National Academy of Sciences of the USA

www.pnas.org兾cgi兾doi兾10.1073兾pnas.0805350105

Table 1 Impact of homocysteinemia-inducing diets on blood, brain, and cognition Diet Outcome measures Blood biochemistry Folate (ng/ml) B12 (pmol/ml) PLP (B6 pmol/ml) tHcy (␮mol/L) Brain anatomy Hippocampal capillary length (␮m) Hippocampal microglia number (N) Cognitive function Mean baseline escape latency day 1 (s) Mean acquired escape latency day 3 (s) Mean reversal escape latency day 4 (s) Probe trial—latency to escape position (s) Psychomotor function Rotarod latency (s)

ANOVA

Control

High methionine

B-vitamin deficient

F

p

107 (39.2)a 18,861 (5,157)a 42.6 (2.5)a 5.2 (0.7)a

144.7 (44.8)b 15,386 (8564)a 35.1 (8.2)b 13.9 (3.6)a

28.3 (6)c 5,260 (2,742)b 4.0 (0.8)c 35.2 (28.7)b

34.8 14.4 172.1 8.4

⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001

6.2 ⫻ 105 (0.6 ⫻ 105)a 2,633 (235)a

4.8 ⫻ 105 (1.0 ⫻ 105)b 1,277 (216)b

4.4 ⫻ 105 (0.6 ⫻ 105)b 1,961 (579)c

8.5 68.8

0.004 ⬍0.001

30.7 (9.8) 18.5 (7.1) 19.6 (9.6)a 6.8 (3.9)a

34.1 (9.7) 20.8 (11.2) 30.5 (7.2)ab 12.2 (6.1)ab

32.9 (10.8) 15.6 (7.1) 31.9 (13.0)b 15.9 (8.7)b

0.3 1.1 4.2 4.6

0.758 0.354 0.025 0.018

32.9 (14.5)

35.2 (17.6)

35.6 (18.1)

0.07

0.93

Values given are mean (SD). Values marked across rows with the same superscript a, b, and c are not significantly different; values indicated only by different letters (a, b, or c) are significantly different from each other with at least P ⬍ 0.05 by one-way ANOVA with Tukey’s honest squares differences post hoc test. PLP, pyridoxal 5⬘-phosphate; tHcy, plasma total homocysteine.

Troen et al.

to adapt to the new escape position and they reverted to their mean baseline latencies of ⬎30 s. Similarly, during the probe trial on day 4, the mean latency to the escape position of the B-vitamin-deficient group was twice that of the controls (see Table 1). Mice fed the high methionine diet displayed similar cognitive impairments, but these were not statistically different from either the control or the B-vitamindeficient groups (see Table 1). Because diet did not affect swim speed, a nearly identical curve is observed when data are plotted as the path length swum in each trial [supporting information (SI) Fig. S1B]. As swim speed and psychomotor performance (see Table 1) were not affected by diet, they cannot explain the diet-induced differences in Morris water maze performance. On the probe trial, latency to escape position was the only parameter to differ statistically by diet. Nevertheless, a comparison of performance on the probe trials by diet suggests a subtle spatial dimension to the impaired performance of B-vitamin deficiency. Control mice demonstrated a trend for shorter latencies to the escape position (see Table 1 and Fig. S2) and more crossings of the escape position on days 2 to 4. Further60 50 40 30 20 10 0 1 2 3 4 Day 1

1 2 3 4 Day 2

1 2 3 4 Day 3

1 2 3 4 Day 4

Fig. 1. Morris water maze Learning Curves. The figure presents daily escape latencies on each of four daily trials.Control diet ⫽ blue circles; high methionine diet ⫽ green squares; B-vitamin-deficient diet ⫽ red triangles. The curves show that following 3 days of training, all mice were capable of learning and retaining the escape task, regardless of diet. However, in contrast to mice fed the control diet, mice fed the B-vitamin-deficient diet were impaired in their ability to learn the new escape position on the fourth day of testing when the position of the escape platform was reversed. PNAS 兩 August 26, 2008 兩 vol. 105 兩 no. 34 兩 12475

MEDICAL SCIENCES

Seconds

0.79). The appearance of all mice was normal and no morbidity was observed. As expected, the B-vitamin-deficient diet significantly decreased plasma vitamin levels. Mean plasma folate and vitamin B12 decreased to one quarter, and vitamin B6 decreased to one tenth of control concentrations (Table 1). The high methionine diet also produced significant, but different, changes in plasma vitamin status. Mean fasting folate levels increased significantly by 35% from the control concentration and vitamin B6 levels decreased significantly by 18%. Mean vitamin B12 levels after methionine feeding also decreased by the same percentage, but this change did not reach statistical significance (see Table 1). The mean fasting plasma tHcy concentration in mice fed the high methionine diet increased to nearly threefold higher than in mice fed the control diet (13.9 versus 5.2 ␮M). The mean fasting plasma tHcy concentration in mice fed a B-vitamindeficient diet increased to nearly sevenfold higher than in mice fed the control diet (35.2 versus 5.2 ␮M). The mean plasma tHcy concentration in the B-vitamin-deficient group was significantly higher than those of both the control and high methionine fed groups; however, the difference between the control and high methionine groups did not reach statistical significance (see Table 1). B-vitamin deficiency was associated with a significant cognitive impairment and spatial learning and memory in the Morris water maze test. A similar but less pronounced and nonsignificant deficit was observed in the group fed the high methionine diet. Fig. 1 shows the learning curves for escape latency from the Morris water maze over the course of the 3-day training period and performance on the reversal test on day 4. All three groups acquired the task over the first 3 days and learned to escape from the maze. On the first day of training, the mean escape latencies were similar for all groups (see Table 1). All groups improved with training and by day 3 the mean escape latencies had reached a similar threshold. However, a diet-induced cognitive deficit was revealed on the day 4 reversal task, which requires that the mice retain the escape strategy from the previous days while rapidly learning and remembering the new (reversed) position of the escape platform. On the day 4 reversal task the control group rapidly learned the new platform position and their escape latency did not differ from their previously acquired level. In contrast, B-vitamin-deficient mice were impaired in their ability

Latency to escape position (sec)

25 20 15 10 5 0 300,000

400,000

500,000

600,000

700,000

Hippocampal Capillary Length (micron)

Fig. 2. Cognitive performance on Morris water maze correlates with hippocampal capillary length. Chart shows that hippocampal capillary length strongly predicts escape latencies on the Morris water maze day 4 reversal probe trial (the shorter the latency, the better the performance). Note that there is little overlap between control and treatment groups. Decreases in hippocampal capillary length were highly correlated with longer latencies to the escape position during the Morris water maze reversal probe trial (r ⫽ ⫺0.757, P ⬍ 0.001). Blue circles ⫽ control diet; green squares ⫽ methioninerich diet; red diamonds ⫽ B-vitamin-deficient diet.

more, on days 3 and 4 they spend significantly ⬎25% of their swimming in the quadrant in which the escape platform is located (above chance levels), indicating that these mice are employing a spatial search strategy to escape from the maze, even on day 4. In contrast, changing the position of the escape platform on day 4 results in significantly longer latencies to the escape position for B-vitamin-deficient mice than for the other groups (P ⫽ 0.018, see Table 1). Moreover, on average B-vitamindeficient mice make 24% fewer crossings of the escape platform position on day 4 than on day 3, compared with mice fed control and high methionine diets, who maintain a similar number of escape position crossings on both days (1% and 7% fewer crossings, respectively). Finally, unlike control mice, who spend ⬎25% of their time in the quadrant where the escape platform is located, mice fed B-vitamin-deficient or high methionine diets do not spend significantly ⬎25% of their search time (chance level) in the escape quadrant (see Fig. S2). Gross brain anatomy appeared normal on microscopic examination (Fig. S3). There was no evidence of gliosis, which typically accompanies neurodegeneration (11, 12). Instead, the B-vitamin-deficient and high methionine diets significantly suppressed the population of microglial cells in hippocampus. Compared to controls, B-vitamin deficiency resulted in a ⬇25% decrease in microglia. High methionine suppressed microglia even more severely, reducing their estimated numbers by ⬇50% (see Table 1 and Fig. S4). High methionine and B-vitamin deficiency also resulted in a significant rarefaction of hippocampal microvasculature (see Table 1). Compared to controls, high methionine resulted in an ⬇23% reduction in mean total capillary length and B-vitamin deficiency resulted in an ⬇30% reduction (see Table 1 and Fig. S4). Reduced capillary length was highly correlated with both higher plasma tHcy (r ⫽ ⫺0.631, P ⫽ 0.007) and with longer latencies to the escape position during the probe trial (r ⫽ ⫺0.757, P ⬍ 0.001; Fig. 2). Longer latencies to escape position were also highly correlated with higher plasma tHcy (r ⫽ 0.721, P ⬍ 0.001) and remained highly significant even after excluding the two cases with the most severe hyperhomocysteinemia (r ⫽ 0.539, P ⫽ 0.002) (Fig. S5). In contrast with these results, and despite significant effects of diet on the number of hippocampal microglial cells, the number of microglia was only weakly correlated with homocysteine (r ⫽ 0.334, P ⬍ 0.01) and did not correlate with cognitive outcomes. 12476 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0805350105

Discussion Our examination of the impact of hyperhomocysteinemiainducing diets on brain anatomy and vascular integrity and cognition in mice allows us to make several important observations. First, dietary deficiencies of folate, vitamin B12, and B6 intake that lead to moderate hyperhomocysteinemia can also result in cognitive impairment. That this occurs in an inbred mouse strain that is not genetically predisposed to cerebrovascular or neurodegenerative disease demonstrates that such metabolic imbalances may not only sensitize the brain to a primary neurological insult, but are sufficient to induce cognitive dysfunction even in the absence of a pre-existing challenge. Moreover, these impairments were associated with mild to moderate hyperhomocysteinemia at homocysteine concentrations that are comparable to those that confer risk of cognitive impairment in humans [ie, reference (13)]. In this respect, findings in this model may be compared to human populations where similar nutritional deficiencies and hyperhomocysteinemia are observed to associate with cognitive deficits. Second, the rarefaction of brain capillaries following a relatively short-term dietary challenge demonstrates that brain microvasculature is sensitive to the activity of pathways that metabolize homocysteine. Third, the correlation between cognitive and microvascular deficits demonstrated in this model, in the absence of overt neurodegeneration, constitutes evidence that a metabolically driven change in cerebral microvascular circulation could account for the strong association of B-vitamin deficiency, hyperhomocysteinemia, and cognitive dysfunction in humans. The present mouse model exhibits several salient features of human age-related cognitive dysfunction. First, cerebral microvascular abnormalities are common in elders, and although they are also found in cognitively intact individuals, they are often more severe in those with dementia (14–16). Indeed, rarefaction of cortical capillary beds has recently been reported in Alzheimer’s disease (17). Second, Glut-1 expression is decreased in postmortem brain tissue from demented individuals (18–20). These structural and biochemical abnormalities are likely to contribute to impaired blood flow and the integrity of the physical blood brain barrier, impeding the brain’s supply of oxygen, energy, and nutrients, and the removal of toxins from the brain (21). The observation that relatively acute disruptions of homocysteine metabolism can induce such a dramatic change in microvascular density, in the apparent absence of the neurodegenerative and other pathologies that are associated with hyperhomocysteinemia in humans, suggests that microvascular changes may be sufficient to impair cognition even without the more extensive pathology that would be expected to arise from chronic exposure. To the extent that cerebral microvasculature in mice is susceptible to disrupted homocysteine metabolism, it is possible that similar homocysteine-related microvascular damage contributes to cognitive impairment in humans. However, this has been difficult to establish, not the least because of poor consensus on neuropathological methods of classifying and quantifying cerebrovascular lesions (1, 22). Recent advances in unbiased stereology (23) now provide a rigorous and objective (albeit tedious) method of quantifying structural parameters of the microvasculature in postmortem human (24) and animal brains (25). However, vascular and cognitive outcomes have not yet been correlated by these methods. In this respect, the model described here offers a new system for studying the relation of cerebral microvascular changes to cognition in VCI. Our finding of decreased total capillary length following diet-induced hyperhomocysteinemia enlarges significantly upon findings of structural and biochemical abnormalities in other animal studies of experimental homocysteinemia, by showing that B-vitamin deficiency not only changes existing capillaries, Troen et al.

Troen et al.

without neurodegeneration, astrogliosis, or demyelination (10). Such studies are consistent with a capacity of homocysteine to contribute to cognitive dysfunction by potentiation of a primary neurodegenerative insult. However, they do not support the hypothesis that direct neurotoxicity is the primary mechanism leading to homocysteine-related cognitive dysfunction. Thus, in the absence of neuropathology, alternative mechanisms, such as cerebrovascular dysfunction, may be more important for homocysteine-related cognitive impairment. Here, the apparent absence of neurodegeneration does not rule out the selective death of brain microglia and capillary endothelial cells. However, it is also possible that the observed reduction in capillary length and microglia numbers reflects a decreased capacity to proliferate and maintain normal cell numbers. Indeed, folate deficiency has been shown to slow neurogenesis in WT mice without causing neuronal cell death (41). Both folate deficiency and excess methionine can slow the synthesis of nucleic acids, thereby inhibiting proliferation. Theoretically, such a phenomenon might have a greater impact on vascular and glial cells that, in contrast to the relatively quiescent neurons, retain their proliferative potential in the adult brain (42, 43). Although we do not know whether the capillary rarefaction in this study was the result of homocysteine-mediated endothelial death or vascular remodeling, it is notable that adult cerebral microvasculature is capable of rapid and reversible angiogenic changes in response to metabolic stimuli. For example, rats exposed to decreased oxygen for 3 weeks to simulate a high altitude atmosphere exhibit a massive increase in cerebral microvascular density, which returns to baseline within 3 weeks following restoration of normal atmospheric oxygen (43). Similar vascular plasticity occurs in response to brain activity (26, 28). This type of plasticity is reminiscent of the rapid microvascular response to diet in this study, as well as the restoration of normal Glut-1 expression levels following the alleviation of short-term hyperhomocysteinemia in rats (9). Clearly, the extent to which the adult brain retains such microvascular plasticity could affect brain and cognitive integrity during aging (44, 45). In this context, our findings raise the intriguing possibility that impaired homocysteine metabolism might impair microvascular plasticity, thereby contributing to the accrual of age- or diseaserelated damage. Similarly, diet-induced decreases in microglial cells could also present a challenge to brain. During brain trauma, microglia are involved in a protective inflammatory response by mobilization to clear damage in brain parenchyma, such as plaques and debris, from degenerating cells. This may be one reason why the number of brain microglia increase in normal aging (46). However, chronic and persistent glial activation—gliosis—can cause damage through oxidative stress and other mechanisms. Accordingly, microglial suppression could have beneficial or adverse consequences for brain integrity, depending on the specific context (12). The severe suppression of microglia by high methionine intake in association with very mild hyperhomocysteinemia, as compared with a more modest suppression of microglia by B-vitamin deficiency despite more severe hyperhomocysteinemia, suggests that while hyperhomocysteinemia may be common to different metabolic imbalances, such as low B-vitamin and excess methionine, different neurological outcomes may be determined by the primary metabolic disruption and not by homocysteine. Taking account of the differential impact of methionine and B-vitamin deficiency on capillary length and microglia numbers makes it possible to distinguish between the effects of different subtypes of hyperhomocysteinemia in the same model animal. In conclusion, by isolating early cerebral microvascular changes from neurodegenerative processes, such as neuronal cell death and gliosis, we can define more precisely the mechanisms underlying cerebral microvascular disease, independent of or PNAS 兩 August 26, 2008 兩 vol. 105 兩 no. 34 兩 12477

MEDICAL SCIENCES

but induces remodeling and rarefaction of the capillary bed. Kim and colleagues (8) fed folate-deficient diets to 6-month old male rats for 8 weeks. In these rats, folate deficiency was accompanied by mild hyperhomocysteinemia and ultrastructural changes in cerebral capillaries, including endothelial damage with mitochondrial swelling and disintegration, swelling of pericytes, basement membrane thickening and fibrosis, and occasional perivascular detachment. Similarly, Lee and colleagues (9) found that feeding 8-week old rats a folate-deficient diet containing supplemental homocysteine for 4 weeks resulted in a mild hyperhomocysteinemia and a 24% reduction in brain Glut-1 expression, as well as similar reductions in other endothelial markers in whole brain homogenates. Previously, these findings could be understood to imply that glucose transport might be down-regulated in individual capillaries in response to hyperhomocysteinemia. Indeed, brain Glut-1 expression adapts rapidly in response to changes in brain metabolism and energy demand (26–28), and pharmacologic injections of homocysteine that induce severe hyperhomocysteinemia have been shown to inhibit brain energy and glucose metabolism in rats (29). However, inhibited brain energy metabolism following pharmacological administration of homocysteine coincides with severe hyperhomocysteinemia. In contrast, the more modest increases in plasma homocysteine induced in the present study better reflect the epidemiological associations of mild hyperhomocysteinemia and cognitive impairment. Based on structural evidence from the present study, we argue that the reduced abundance of brain Glut-1 and other vascular proteins found in other in vivo models (9) is likely to reflect capillary loss, which could then have a secondary impact on brain energy metabolism and neurovascular coupling (14, 30, 31). Furthermore, the striking correlation that we found between hippocampal microvascular length and escape latency in the water maze suggests that these observations may be functionally significant. This interpretation is consistent with data from a study of rats showing a similar correlation between the severity of ultrastructural damage to hippocampus capillaries and performance on the Morris water maze after microvascular damage was induced by chronic occlusion of the common carotid artery (32). Nevertheless, this correlation alone cannot prove that that the observed microvascular changes in hippocampus mediate the cognitive impairment. In this study, B-vitamin-deficient mice were significantly impaired in their ability to learn a new platform location and subtly impaired in using spatial strategies to find the platform compared to the control group. These decrements are similar to those observed in aged animals in that they also show difficulty in learning a new platform location during reversal training (33) and lack of a spatial preference when compared to young animals when tested in a probe trial (ie, less time spent in the training quadrant and fewer platform crossings) (34, 35). Although spatial performance on the Morris water maze is highly sensitive to hippocampal integrity (36), additional brain regions contribute to performance on this task (37). Thus, dysfunction in other brain systems may also contribute to the overall impairment that is induced by B-vitamin deficiency. Previous studies have failed to detect neuronal cell death in the hippocampus of WT mice fed hyperhomocysteinemiainducing diets (38, 39). Indeed, even the direct injection of a large dose of homocysteine into the hippocampus of WT mice did not kill neurons (40). Dietary hyperhomocysteinemia has only been found to enhance neuronal cell death in mice that are predisposed to neurodegeneration, such as the transgenic mutant amyloid precursor protein model of Alzheimer’s disease (38) or in mice rendered pharmacologically susceptible to Parkinson’s disease (39). Similarly, feeding neurocompromised Apolipoprotein E-deficient mice B-vitamin-deficient diets induces both hyperhomocysteinemia and cognitive impairment

before the onset of irreversible neurodegeneration. Given the relatively rapid induction of these changes, this model of homocysteine-related VCI should facilitate the study of their relation to downstream neurological and cognitive outcomes.

means across dietary treatments by ANOVA, with Tukey’s Honest Squares Differences posthoc adjustment for multiple comparisons. Psychomotor function was evaluated by a battery of tests, including rod walking, wire suspension, plank walking, and the accelerating rotarod (49).

Methods

Biochemical Assays. After 10 weeks of consuming the diets, mice were fasted overnight and killed by exsanguination under isoflurane anesthesia. Blood was collected by heart puncture into heparinized tubes on ice. Plasma was separated within an hour of collection and frozen at ⫺80°C for further analysis. Folate and B12 were measured using the Quantaphase II radioassay kit (Bio-Rad Laboratories ). Pyridoxal 5⬘-phosphate (vitamin B6) was determined by the tyrosine decarboxylase apoenzyme method (53). Plasma total homocysteine was determined by HPLC (54).

Animals and Diets. All animal procedures were approved by the Institutional Animal Care and Use Committee in accordance with the National Institutes of Health Guidelines. Housing and feeding procedures were described in ref. 47. Briefly, weanling male C57Bl6/J mice were purchased from Jackson Laboratories and maintained at our animal facility. They were systematically assigned to three groups of similar mean body weights of at least 10 mice per group, housed individually and fed for 10 weeks with diets formulated with vitaminfree, ethanol-precipitated casein and the appropriate control and experimental vitamin mix (Harlan Teklad, Madison WI,). The control group consumed an AIN93M diet containing 0.33% methionine, 2-mg folic acid, 25-␮g cyanocobalamin (vitamin B12), and 7-mg pyridoxal L-phosphate (vitamin B6) per kg diet. Two different diets were formulated to induce hyperhomocysteinemia: one through combined folate, vitamin B12 and vitamin B6 deficiency; the other through methionine enrichment with 1% L-methionine (10-g L-methionine/kg diet). Both diets induce hyperhomocysteinemia; however, they do so through markedly different metabolic impairments. B-vitamin deficiency inhibits homocysteine’s conversion to methionine or cysteine, causing it to accumulate while methionine is depleted. In contrast, high methionine intake drives the excessive synthesis of homocysteine and cysteine but without limiting methionine. Mice were allowed free access to water and were group pair-fed to ensure that all of the mice had similar food intake (48). All diets contained 1% sulfathiozole (10 g/kg diet. SIGMA), a nonabsorbed sulfa drug that inhibits folate formation by gut bacteria, to ensure that the animal’s only source of available folate was from the diet. Cognitive Testing. Spatial learning and memory were evaluated by the Morris water maze and psychomotor function was evaluated by a battery of agesensitive tests during the last week of the experiment (49). Using these tests, we have shown that in Apolipoprotein-E deficient mice specific cognitive deficits on the Morris water maze but not on psychomotor performance result from the B-vitamin-deficient diet (10). The Morris water maze is a well validated and highly sensitive test of rodent cognitive function, and it is particularly sensitive to hippocampal dysfunction (50 –52). The maze requires a mouse to use spatial learning to find a hidden platform submerged 1 cm below the surface of the water in a circular white pool filled with water made opaque by the addition of powdered milk. The mouse must use distal cues that are located outside of the maze to effectively locate and remember the location of the escape platform from previous trials. Accurate navigation is rewarded with escape from the water onto the platform, which is colored white and therefore hidden from sight. In this study we used a 4-day testing protocol, as previously described (10). Briefly, mice underwent 3 days of training in the maze of five trials each day, followed by a fourth and final day of testing in which the position of the escape platform was changed, thereby requiring the mouse to retain the learned escape strategy but to quickly learn the new escape position (Fig. S6). The fifth and last trial on days 2 to 4 was a probe trial in which the platform was removed and the mouse allowed to swim for 60 s to assess spatial strategies before the trial was stopped. Performances were videotaped and analyzed with image tracking software (HVS Image). Mean daily escape latencies and latency to the escape position on the probe trial were analyzed for differences between the

1. Hachinski V, et al. (2006) National Institute of Neurological Disorders and StrokeCanadian Stroke Network vascular cognitive impairment harmonization standards. Stroke 37:2220 –2241. 2. Breteler MM (2000) Vascular involvement in cognitive decline and dementia. Epidemiologic evidence from the Rotterdam Study and the Rotterdam Scan Study. Ann N Y Acad Sci 903:457– 465. 3. Troen A, Rosenberg I (2005) Homocysteine and cognitive function. Semin Vasc Med 5:209 –214. 4. Selhub J, Bagley LC, Miller J, Rosenberg IH (2000) B vitamins, homocyseine, and neurocognitive function in the elderly. Am J Clin Nutr 71:614S– 620S. 5. Troen AM (2005) The central nervous system in animal models of hyperhomocysteinemia. Prog Neuropsychopharmacol Biol Psychiatry 29:1140 –1151. 6. Baumbach GL, Sigmund CD, Bottiglieri T, Lentz SR (2002) Structure of cerebral arterioles in cystathionine beta-synthase-deficient mice. Circ Res 91:931–937. 7. Devlin AM, et al. (2004) Effect of Mthfr genotype on diet-induced hyperhomocysteinemia and vascular function in mice. Blood 103:2624 –2629. 8. Kim JM, Lee H, Chang N (2002) Hyperhomocysteinemia due to short-term folate deprivation is related to electron microscopic changes in the rat brain. J Nutr 132:3418 – 3421.

12478 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0805350105

Brain Histology and Stereology. Following exsanguination, brains from half of the mice were rapidly removed and fixed in 1% neutral buffered formalin/ 30% sucrose solution until they sank. Fixed brains were embedded together in gelatin blocks, from which serial sections 40-␮m thick were prepared in the sagittal plane through the entire brain (Neuroscience Associates). Cerebrovascular capillaries were visualized with a polyclonal rabbit antibody against the endothelial Glut-1 (catalog no. RDI-GLUT1, Research Diagnostics), which specifically labels capillaries and not larger vessels (55, 56). Microglia were labeled with a monoclonal rat antibody against cell surface protein CD11b/Mac-1, which is expressed by all brain microglia (catalog no. MCA74G, Serotec) (57). Immunolabeling was followed by routine immunohistochemistry (ABC Vector Elite kit, Vectorlabs). Stereological analysis was performed using a computerized stereology system driving a Zeiss Axioskop microscope equipped with a high-resolution color video camera and a motorized stage for movement in the x-y-z axes (Stereologer, Systems Planning and Analysis). For each measurement, we used consecutive sections spaced 200 ␮m apart through the hippocampus of one hemisphere. The hippocampal formation was defined according to the mouse brain atlas of Franklin and Paxinos (58) to include the oriens, pyramidal, stratum radiatum, and dentate gyrus granular and molecular layers, circumscribed by the corpus callosum, fimbria, and hippocampal fissure. Hippocampal microglia were counted at 40⫻ magnification using the optical fractionator probe (57). Hippocampal total capillary length was estimated at 20⫻ magnification using the cycloid probe (23, 25). Using this probe, the number of intersections between the system-generated cycloid probes and capillaries are counted. The probability of an intersection between probe and capillary is directly proportional to the total length of the capillaries traversing the region of interest. Measurements were carried out blind to treatment by a single investigator (M.S.B.) according to counting parameters established during pilot experiments. Statistics. Data were analyzed using the statistical software SPSS 12.0 for descriptive statistics, Pearson correlations, and comparison of differences between the means by ANOVA with Tukey’s honest squares differences posthoc adjustment for multiple comparisons. ACKNOWLEDGMENTS. We thank Ms. Laura Burns for technical assistance with the behavioral experiments. This project was supported by U.S. Department of Agriculture cooperative agreement 58 –1950-9 – 001. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of the U.S. Department of Agriculture.

9. Lee H, Kim HJ, Kim JM, Chang N (2004) Effects of dietary folic acid supplementation on cerebrovascular endothelial dysfunction in rats with induced hyperhomocysteinemia. Brain Res 996:139 –147. 10. Troen AM, et al. (2006) The cognitive impact of nutritional homocysteinemia in Apolipoprotein-E deficient mice. J Alzheimers Dis 9:381–392. 11. Ladeby R, et al. (2005) Microglial cell population dynamics in the injured adult central nervous system. Brain Res Brain Res Rev 48:196 –206. 12. Streit WJ, Walter SA, Pennell NA (1999) Reactive microgliosis. Prog Neurobiol 57:563– 581. 13. Kado DM, et al. (2005) Homocysteine versus the vitamins folate, B6, and B12 as predictors of cognitive function and decline in older high-functioning adults: MacArthur Studies of Successful Aging. Am J Med 118:161–167. 14. Farkas E, Luiten PG (2001) Cerebral microvascular pathology in aging and Alzheimer’s disease. Prog Neurobiol 64:575– 611. 15. Ince P, Neuropathology Group of the MRC CFAS Multi Center Study (2001) Pathological correlates of late-onset dementia in a multicentre, community-based population in England and Wales. Neuropathology Group of the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). Lancet 357:169 –175.

Troen et al.

38. Kruman II, et al. (2002) Folic acid deficiency and homocysteine impair DNA repair in hippocampal neurons and sensitize them to amyloid toxicity in experimental models of Alzheimer’s disease. J Neurosci 22:1752–1762. 39. Duan W, et al. (2002) Dietary folate deficiency and elevated homocysteine levels endanger dopaminergic neurons in models of Parkinson’s disease. J Neurochem 80:101–110. 40. Kruman II, et al. (2000) Homocysteine elicits a DNA damage response in neurons that promotes apoptosis and hypersensitivity to excitotoxicity. J Neurosci 20:6920 – 6926. 41. Kruman II, et al. (2005) Folate deficiency inhibits proliferation of adult hippocampal progenitors. Neuroreport 16:1055–1059. 42. Wirenfeldt M, et al. (2005) Reactive microgliosis engages distinct responses by microglial subpopulations after minor central nervous system injury. J Neurosci Res 82:507– 514. 43. Harik N, et al. (1996) Time-course and reversibility of the hypoxia-induced alterations in cerebral vascularity and cerebral capillary glucose transporter density. Brain Res 737:335–338. 44. LaManna JC, Chavez JC, Pichiule P (2004) Structural and functional adaptation to hypoxia in the rat brain. J Exp Biol 207:3163–3169. 45. Riddle DR, Sonntag WE, Lichtenwalner RJ (2003) Microvascular plasticity in aging. Ageing Res Rev 2:149 –168. 46. Mouton PR, et al. (2002) Age and gender effects on microglia and astrocyte numbers in brains of mice. Brain Res 956:30 –35. 47. Troen AM, et al. (2003) The atherogenic effect of excess methionine intake. Proc Natl Acad Sci USA 100:15089 –15094. 48. Wu D, et al. (2007) Dietary supplementation with white button mushroom enhances natural killer cell activity in C57BL/6 mice. J Nutr 137:1472–1477. 49. Shukitt-Hale B, Mouzakis G, Joseph JA (1998) Psychomotor and spatial memory performance in aging male Fischer 344 rats. Exp Gerontol 33:615– 624. 50. Crawley JN (2003) Behavioral phenotyping of rodents. Comp Med 53:140 –146. 51. Bailey KR, Rustay NR, Crawley JN (2006) Behavioral phenotyping of transgenic and knockout mice: practical concerns and potential pitfalls. Ilar J 47:124 –131. 52. Karl T, Pabst R, von Horsten S (2003) Behavioral phenotyping of mice in pharmacological and toxicological research. Exp Toxicol Pathol 55:69 – 83. 53. Shin YS, Rasshofer R, Friedrich B, Endres W (1983) Pyridoxal-5⬘-phosphate determination by a sensitive micromethod in human blood, urine and tissues; its relation to cystathioninuria in neuroblastoma and biliary atresia. Clin Chim Acta 127:77– 85. 54. Araki A, Sako Y (1987) Determination of free and total homocysteine in human plasma by high-performance liquid chromatography with fluorescence detection. Journal of Chromatography 422:43–52. 55. Choeiri C, Staines W, Messier C (2002) Immunohistochemical localization and quantification of glucose transporters in the mouse brain. Neuroscience 111:19 –34. 56. Kalaria RN, et al. (1988) The glucose transporter of the human brain and blood-brain barrier. Ann Neurol 24:757–764. 57. Long JM, et al. (1998) Stereological estimation of total microglia number in mouse hippocampus. J Neurosci Methods 84:101–108. 58. Franklin K, Paxinos G (1997) The Mouse Brain in Stereotaxic Coordinates (Academic, San Diego) 2nd Ed.

MEDICAL SCIENCES

16. de la Torre JC (1997) Cerebromicrovascular pathology in Alzheimer’s disease compared to normal aging. Gerontology 43:26 – 43. 17. Kitaguchi H, et al. (2007) Capillary beds are decreased in Alzheimer’s disease, but not in Binswanger’s disease. Neurosci Lett 417:128 –131. 18. Horwood N, Davies DC (1994) Immunolabelling of hippocampal microvessel glucose transporter protein is reduced in Alzheimer’s disease. Virchows Arch 425:69 –72. 19. Mooradian AD, Chung HC, Shah GN (1997) GLUT-1 expression in the cerebra of patients with Alzheimer’s disease. Neurobiol Aging 18:469 – 474. 20. Simpson IA, et al. (1994) Decreased concentrations of GLUT1 and GLUT3 glucose transporters in the brains of patients with Alzheimer’s disease. Ann Neurol 35:546 – 551. 21. de la Torre JC (1997) Hemodynamic consequences of deformed microvessels in the brain in Alzheimer’s disease. Ann N Y Acad Sci 826:75–91. 22. Pantoni L, et al. (2006) Postmortem examination of vascular lesions in cognitive impairment: a survey among neuropathological services. Stroke 37:1005–1009. 23. Calhoun ME, Mouton PR (2000) Length measurement: new developments in neurostereology and 3D imagery. J Chem Neuroanat 20:61– 69. 24. Bouras C, et al. (2006) Stereologic analysis of microvascular morphology in the elderly: Alzheimer disease pathology and cognitive status. J Neuropathol Exp Neurol 65:235– 244. 25. Lee GD, et al. (2005) Stereological analysis of microvascular parameters in a double transgenic model of Alzheimer’s disease. Brain Res Bull 65:317–322. 26. Duelli R, Maurer MH, Kuschinsky W (1998) Decreased glucose transporter densities, rate constants and glucose utilization in visual structures of rat brain during chronic visual deprivation. Neurosci Lett 250:49 –52. 27. Duelli R, et al. (2001) Correlation between local glucose transporter densities and local 3-O-methylglucose transport in rat brain. Neurosci Lett 310:101–104. 28. Choeiri C, et al. (2005) Glucose transporter plasticity during memory processing. Neuroscience 130:591– 600. 29. Streck EL, et al. (2003) Impairment of energy metabolism in hippocampus of rats subjected to chemically-induced hyperhomocysteinemia. Biochim Biophys Acta 1637:187–192. 30. de la Torre JC, et al. (1992) Chronic cerebrovascular insufficiency induces dementia-like deficits in aged rats. Brain Res 582:186 –195. 31. Deane R, Zlokovic BV (2007) Role of the blood-brain barrier in the pathogenesis of Alzheimer’s disease. Curr Alzheimer Res 4:191–197. 32. De Jong GI, et al. (1999) Cerebral hypoperfusion yields capillary damage in the hippocampal CA1 area that correlates with spatial memory impairment. Neuroscience 91:203–210. 33. Gage FH, Dunnett SB, Bjorklund A (1984) Spatial learning and motor deficits in aged rats. Neurobiol Aging 5:43– 48. 34. Frick KM, et al. (1995) Age-related spatial reference and working memory deficits assessed in the water maze. Neurobiol Aging 16:149 –160. 35. Rapp PR, Rosenberg RA, Gallagher M (1987) An evaluation of spatial information processing in aged rats. Behav Neurosci 101:3–12. 36. McDonald RJ, White NM (1994) Parallel information processing in the water maze: evidence for independent memory systems involving dorsal striatum and hippocampus. Behav Neural Biol 61:260 –270. 37. Devan BD, Goad EH, Petri HL (1996) Dissociation of hippocampal and striatal contributions to spatial navigation in the water maze. Neurobiol Learn Mem 66:305–323.

Troen et al.

PNAS 兩 August 26, 2008 兩 vol. 105 兩 no. 34 兩 12479

Supporting Information Troen et al. 10.1073/pnas.0805350105

A) Daily Escape Latencies 60 50

Seconds

40 30 20 10 0 1

2

3

4

1

2

Day 1

3

4

1

2

Day 2

3

4

1

Day 3

2

3

4

Day 4

B) Daily Path Length 10

Meters

8

6

4

2

0 1

2

3

4

Day 1

1

2

3 Day 2

4

1

2

3 Day 3

4

1

2

3

4

Day 4

Fig. S1. Learning curves for the Morris Water Maze showing (A) daily escape latencies on each of four daily trials, and (B) swim path length. Blue circles ⫽ control diet; green squares ⫽ high methionine diet; red triangles ⫽ B-vitamin-deficient diet. The learning curves show that following 3 days of training, all mice were capable of learning and retaining the escape task, regardless of diet. However, in contrast to mice fed the control diet, mice fed the B-vitamin-deficient diet were impaired in their ability to learn the new escape position on the fourth day of testing when the position of the escape platform was reversed. The fact that the curves for escape latency and path length are nearly identical illustrates that this impairment is because of slower escape latencies rather than different swimming speeds.

Troen et al. www.pnas.org/cgi/content/short/0805350105

1 of 6

A) Latency to the escape position 20

*

Latency to escape position (sec)

18 16 14 12 10 8 6 4 2 0 P1

P1

P2

Day 2

Day 3

Day 4

B) Number of crossings through the escape platform position 10

Escape position crossings (N)

9 8 7 6 5 4 3 2 1 0 P1

P1

P2

Day 2

Day 3

Day 4

C) Percent time spent in escape quadrant

Time in Escape Quadrant (%)

45 40 35 30 25 20 15 Q4

Q4

Q2

Day 2

Day 3

Day 4

Fig. S2. Figure shows performance on Morris Water Maze probe trials on days 2 to 4 when the escape platform was removed from the pool with respect to latency to the escape position (A), number of position crossings (B), and percent of total trial time spent searching for the platform in the escape quadrant (C). Control diet, blue bars on left; high methionine diet, green bars in center; B-vitamin-deficient diet, red bars on right. P1 and Q4 designate the escape platform position and quadrant on training days 2 and 3. P2 and Q2 designate the escape platform position and quadrant on day 4. Error bars in panels A and B represent standard error. Error bars in panel C represent the 95% confidence interval of the mean. Mean time spent in the quadrant is significantly different from chance when the boundary of the lower confidence intervals is greater than 25%.

Troen et al. www.pnas.org/cgi/content/short/0805350105

2 of 6

Glut-1

Mac-1

Control Diet

High Methionine Diet

B-vitamin Deficient Diet

Fig. S3. Representative immunolabelling of brain capillaries (left) and microglia (right) in the CA1 region of hippocampus in mice fed control, high methionine, and B-vitamin-deficient diets. Sections are stained with DAB (brown) and lightly counterstained with methyl green. Note that the control and homocysteinemic brain sections are qualitatively very similar. Scale bar: 100 microns for images of Glut-1 labeled capillaries; 50 microns for Mac-1 labeled microglia

Troen et al. www.pnas.org/cgi/content/short/0805350105

3 of 6

700,000

3,500

600,000

a

B b

500,000

400,000

300,000

200,000

100,000

0

3,000

b

Hippocampal Microglial Cells (N)

Hippocampal Capillary Length (μM)

A

a

2,500

c 2,000

1,500

b

1,000

500

0

Fig. S4. Charts show the effect of experimental diet on estimated capillary length (A) and microglia numbers (B) in hippocampus. Control diet, blue bar on left; high-methionine diet, green bar in middle; B-vitamin-deficient diet, red bar on right; error bars represent standard deviation. Bars with different superscripts (a, b, and c) are significantly different from each other by one-way ANOVA, P ⬍ 0.001

Troen et al. www.pnas.org/cgi/content/short/0805350105

4 of 6

25.0

40.0

A

B

20.0

Latency to Escape Position (sec)

Latency to Escape Position (sec)

35.0

15.0

10.0

5.0

0.0

R = -0.757, p