Comparative Effects of Glucose- and Mannitol-Induced Osmolar Stress on Blood—Brain Barrier Function in Ovine Fetuses and Lambs

Abstract

We examined the effects of hyperglycemic hyperosmolality on blood-brain barrier (BBB) permeability during development. We hypothesized that the barrier becomes more resistant to hyperglycemic hyperosmolality during development, and the immature BBB is more resistant to glucose than to mannitol hyperosmolality. We quantified the BBB response to hyperosmolality with the blood-to-brain transfer constant (K_i) in immature fetuses, premature, and newborn lambs. K_i increased as a function of increases in osmolality. A segmented regression model described the relationship between K_i and osmolality. At lower osmolalities, changes in K_i were minimal but after a threshold, increases were linear. We examined responses of K_i to hyperglycemic hyperosmolality by comparing the thresholds and slopes of the second regression segments. Lower thresholds and steeper slopes indicate greater vulnerability to hyperosmolality. Thresholds increased (P < 0.05) during development in pons and superior colliculus. Thresholds were higher (P < 0.05) during glucose than mannitol hyperosmolality in thalamus, superior colliculus, inferior colliculus and medulla of premature lambs, and in cerebrum and cerebellum of newborns. We conclude that BBB permeability increased as a function of changes in glucose osmolality, the barrier becomes more resistant to glucose hyperosmolality in two brain regions during development, and the barrier is more resistant to glucose than to mannitol hyperosmolality in some brain regions of premature and newborn lambs.

Keywords

α-aminoisobutyric acid barrier development glucose immature hseep

Introduction

The blood—brain barrier (BBB) is a selective diffusion barrier that maintains central nervous system homeostasis and limits the entry of substances that could alter neuronal function (Betz and Goldstein, 1986; Bradbury, 1985). We have previously examined the ontogeny of BBB function measured quantitatively with a small hydrophilic molecule, α-aminoisobutyric acid (AIB), during fetal and postnatal development in normal sheep (Stonestreet et al, 1996). Although the BBB exhibited ontogenic decreases in permeability from 60% of gestation to maturity, the barrier was relatively impermeable to AIB at all ages during perinatal development (Stonestreet et al, 1996). Even though we have shown that the BBB is relatively impermeable during development under normal conditions, limited information is available concerning the vulnerability of the BBB to insults and injury during development (Stolp et al, 2005). In addition, the effects of adverse factors on the BBB most likely change during development (Stolp et al, 2005). We have previously shown that during acute mannitol-induced hyperosmolality, BBB permeability increased in response to changes in systemic osmolality and became more resistant to hyperosmolality during development (Stonestreet et al, 2006).

Fetuses of diabetic women who develop hyperglycemia during pregnancy, infants with congenital diabetes, infants who are small for gestational age, premature infants, premature infants exposed to corticosteroids, and infants with sepsis are at high risk for dysregulation of glucose homeostasis and, potentially hyperosmolality (Dweck, 1976; Grylack et al, 1984; Wu, 1996). Hyperglycemic-hyperosmotic stress potentially could affect BBB function in immature subjects. In adult rodents and human subjects, diabetes has been shown to compromise the integrity of the BBB (Allen and Bayraktutan, 2009; Hawkins et al, 2007; Huber et al, 2006; Starr et al, 2003). The effect of glucose-induced hyperosmolality on BBB function has not been studied in immature subjects.

Cerebral glucose consumption is particularly high in immature subjects because of the large size of the neonatal brain in relation to body weight (Otto Buczkowska et al, 2001). Glucose is transported into the brain (Devaskar et al, 1991), where it is metabolized. Based on the known active transport of glucose into brain (Devaskar et al, 1991) and the Staverman reflection coefficient theory (Staverman, 1951), which states that water movement due to osmolality declines with increasing solute transport or permeability across membranes. We postulated that mannitol-induced osmotic stress has greater effects on BBB permeability than glucose-induced osmotic stress because large quantities of glucose are transported into and metabolized by the immature brain (Otto Buczkowska et al, 2001), whereas mannitol is neither transported into nor metabolized by brain. Knowledge of the effects of glucose-induced osmotic stress on BBB function in the fetus and neonate is important because glucose dysregulation is very common in premature infants who are at high risk for brain injury (Dweck, 1976; Grylack et al, 1984; Vohr et al, 2000; Wu, 1996). Blood-brain barrier dysfunction could potentially represent a component of brain injury in these at risk premature infants.

In this study, we examined the effects of exposure to hyperglycemic hyperosmolality as a means to test the relative vulnerability of the BBB in immature subjects to this specific stress. We tested the following hypotheses during acute hyperglycemic hyperosmolality: (1) BBB permeability measured by the blood-to-brain transfer constant with AIB increases as a function of changes in glucose-induced systemic plasma osmolality in immature subjects. (2) The BBB becomes more resistant to the effects of a systemic hyperglycemic-hyperosmotic stress during development. (3) The immature barrier is more resistant to glucose-than to mannitol-induced hyperosmolality.

Materials and methods

This study was conducted after approval by the Institutional Animal Care and Use Committees of Brown University and Women and Infants Hospital of Rhode Island according to the National Institutes of Health Guidelines for use of experimental animals.

Animal Preparation

Surgery was performed under halothane anesthesia as previously described in the fetuses at 60% of gestation (full term sheep gestation is 150 days), premature lambs at 90% of gestation, and newborn lambs (Stonestreet et al, 2000, 2003, 2004). Briefly, in the fetuses at 60% of gestation, catheters were placed in the subclavian vein for intravenous infusions and isotope administration, and subclavian artery and advanced to the thoracic aorta for blood sample withdrawal, heart rate, and blood pressure monitoring. An amniotic fluid catheter was placed as a referent for fetal arterial blood pressure. Catheters were placed in a femoral artery and vein in the ewes.

The surgery on the premature lambs was performed on fetuses in utero to avoid the stress of surgery on the day of study. Polyvinyl catheters were placed into a brachial vein and artery in fetuses at 90% of gestation for the same purposes as described above for the fetuses at 60% of gestation. An endotracheal tube was placed to facilitate immediate suctioning and ventilation at delivery.

Two to three day-old lambs were intubated under ketamine (10 mg/kg) and maintained with 0.75% to 2.0% halothane anesthesia. Catheters were placed into a brachial vein and brachial artery for the purposes described above.

Study Groups

Experiments were performed, after recovery from surgery for 3 days (range 2 to 4 days) in the fetuses at 60% of gestation, 3 days in the premature lambs (range 2 to 7 days), and 1 day (range 1 to 2 days) in the newborn lambs. The fetuses at 60% of gestation were a mean of 88 (range 87 to 88) days of gestation, preterm lambs were 138 (range 136 to 142) days of gestation and the newborn lambs 4 (range 3 to 5) days of age on the day of study. Although the relative maturation of the ovine and human brain cannot be compared precisely, the sheep brain at 80% to 85% of gestation is generally thought to be similar to the newborn infant at term (Back et al, 2006; Gunn et al, 1997). Hence, the relative brain maturation of the fetuses at 60% of gestation would be approximately similar to preterm human infants at 22 to 25 weeks of gestation (Back et al, 2006), the premature lambs at 90% of gestation similar to full term to several week-old infants and the newborn lambs to several weeks to a month-of-age infants.

The fetuses at 60% of gestation, premature lambs delivered at 90% of gestation, and newborn lambs received glucose plus sodium chloride (NaCl) as a 50% glucose and 0.5 M NaCl solution or placebo (0.154 M NaCl). Sodium chloride was added to the glucose solution to limit decreases in plasma and brain NaCl concentrations resulting from the glucose administration (Cserr et al, 1987a; Stonestreet et al, 2004). The study groups included 14 glucose-treated and 7 placebo-treated fetuses at 60% of gestation, 11 glucose-treated and 5 placebo-treated premature lambs delivered at 90% of gestation, and 17 glucose-treated and 6 placebo-treated newborn lambs. The placebo-treated control fetuses and lambs were historic control subjects from our previous studies (Stonestreet et al, 1996, 2006). We used the historic control data to avoid repeating the placebo-treated control experiments to minimize unnecessary animal usage. This was justified because the studies were performed using the same experimental design and methodology as in the experimental studies of the glucose-treated sheep performed for the current study. The brain tissue water content values and the comparison BBB permeability in the mannitol-treated sheep also were from animals in our earlier work (Stonestreet et al, 2003, 2004, 2006).

In this study, we did not study fetal sheep in utero at 90% of gestation because glucose infusions resulted in severe metabolic acidosis and death 20 minutes after the infusions were begun (Stonestreet et al, 2004). In contrast, the fetuses at 60% of gestation tolerated the glucose infusions. Hence, we studied surfactant-treated, ventilated, premature lambs at 90% of gestation that tolerated the glucose infusions.

Experimental Protocol and Methodology

The fetuses, premature, and newborn lambs were studied as previously described (Stonestreet et al, 2006). Baseline determinations were obtained and the glucose or placebo solutions were administered as initial rapid intravenous injections followed by continuous infusions to achieve increases in systemic plasma osmolality (Stonestreet et al, 2004). The osmolar loads were selected to produce both a wide range of osmolalities for subjects within each group and steady-state elevations in plasma osmolality within each subject for the duration of the study. Before each study, we targeted a specific osmolality value ranging between 325 and 438 mosmol/kgH₂O to ensure that the final BBB permeability measures represented a wide range of plasma osmolalities.

In the fetuses at 60% of gestation, glucose plus NaCl or placebo infusions were administered to the fetuses and ewes, because we found that we could not maintain a specific osmolality in the fetuses unless the ewes also received glucose (Stonestreet et al, 2004). For the fetal studies, the ewes were administered initial glucose infusions with a range of 285 to 1,200 ml over 34 minutes followed by continuous infusions with a range of 0 to 768 ml over 1 hour. The fetuses were administered an initial glucose infusion with range of 0 to 6 ml over 10 minutes followed by continuous infusions with a range of 0 to 19 ml over 1 hour.

The initial intravenous infusions followed by continuous infusions of glucose or placebo were administered directly to the premature and newborn lambs. The premature lambs were administered an initial glucose infusion with a range of 20 to 80 ml over 10 minutes followed by continuous infusions with a range of 5 to 166 ml over 1 hour. The newborn lambs were administered an initial glucose infusion with a range of 116 to 198 ml over 10 minutes followed by a continuous infusion with a range 116 to 197 ml over 1 hour.

Plasma osmolality values were obtained, before the glucose infusions at baseline, and sequentially after the end of the initial glucose infusion and during the continuous infusions. Blood gases, pH, heart rate, and arterial blood pressure were measured at baseline, 30, and 60 minutes of study.

Analytic Methods

BBB function was measured in the fetuses, premature, and newborn lambs with α-[¹⁴C]-AIB (Dupont-New England Nuclear, Boston, MA, USA) as previously described (Stonestreet, 1996 et al, 2006). Briefly, after baseline determinations were obtained and the initial infusion of glucose, or placebo administered, [¹⁴C]-AIB was injected intravenously. Arterial plasma concentrations of [¹⁴C]-AIB were obtained at fixed intervals before and after the injection as previously reported (Stonestreet et al, 1996). Brain vascular volume was determined in three additional separate fetuses and lambs in each group treated under the same protocol with ¹⁴C-polyethylene glycol (¹⁴C-PEG, Amersham, England), which was injected intravenously 2 minutes before the end of the study (Stonestreet et al, 2000). The brain vascular volume values did not differ between glucose and 0.5 M NaCl or placebo groups and were similar to reported values (Stonestreet et al, 1996). Brain parenchymal tracer concentration was determined at the end of the experiment.

For the fetal studies, a hysterotomy was performed under intravenous ketamine (15 to 40 mg/kg) anesthesia and the fetus withdrawn from the uterus and decapitated (Stonestreet, 1996 et al, 2000). The ewe was then euthanized with pentobarbital (100 to 200 mg/kg). A similar procedure was used in the lambs.

The brains were dissected into the following regions: cerebral cortex, hippocampus, cerebellum, thalamus, superior colliculus, inferior colliculus, pons, medulla, and cervical spinal cord. The plasma and tissue samples were prepared and the radioactivity of the samples quantified (Stonestreet, 1996 et al, 2006). Knowledge of the plasma concentration profile and the concentration of tracer in the parenchyma allows calculation of the blood-to-brain transfer constant K_i (Ohno et al, 1978; Stonestreet et al, 1996).

The blood-to-brain transfer constant K_i (μl/g Brain per minute) is given by (Equation 1)

K_{i} = A_{b r} / \int_{0}^{t} c_{p} (T) d T .

(1)

Where A_br is the amount of tracer that crossed the BBB from blood to brain during the tracer study (dpm/g), and c_p is the tracer concentration in plasma (dpm/μl) at the time t (min). A_br is obtained by correcting the total amount of isotope measured in the tissue A_m (dpm/g) for that residual part remaining in the brain vasculature space, which is measured by [¹⁴C]-PEG. Thus, A_br = A_m -V_pc_p, where V_p is the plasma volume of brain tissue (ml/g) and c_p is the concentration of tracer in the terminal plasma sample (dpm/g). V_p= A^†_m/c^†_p, where A^†_m and c^†_p have the same definitions as A_m and c_p above except that they apply to [¹⁴C]-PEG (Cserr et al, 1987b; Stonestreet et al, 1996, 2006).

Heart rate, mean arterial blood pressure, amniotic fluid pressures, arterial pH and blood gases, hematocrits, plasma osmolalities, and brain tissue water content values were measured as previously described (Stonestreet et al, 1996, 2003, 2004, 2006).

Calculations

The osmolar load represented the total amount of solute contained in the initial plus the continuous infusions administered to the ewes, fetuses, premature, and newborn lambs. The total osmolar load to the ewe plus the fetus was the sum of the osmolar loads to the ewe and fetus. In the lambs, the total osmolar load was calculated in the same manner (Table 1).

Table 1

Weight, duration of osmolar exposure, and total osmolar load in glucose-infused subjects by study group

	Fetus	Preterm lamb	Newborn lamb
	60% gestation	90% gestation	3-5 days of age
Weight, kg	0.47±0.02	3.28±0.26*	5.0±28*+
Duration of osmolar exposure, minutes	102±3	73±1*	74±0.3*
Total osmolar load, ewe plus fetus or lamb, mosmol	3424±291	276±62*	462±44*

Values are mean ± s.e.m.; glucose-infused fetuses at 60% of gestation n = 14, preterm lambs at 90% of gestation n = 11, lambs at 3–5 days of age n =17.

Total osmolar load was the amount of solute contained in the initial and continuous infusions, which were given to the ewe plus fetus, preterm, and newborn lambs.

P < 0.05 versus fetus at 60% of gestation, + P < 0.05 versus preterm lamb at 90% of gestation.

Statistical Analysis

One-way analysis of variance was used to compare the weight of the sheep and the total osmolar load among the groups (Table 1). Serial measurements were compared within the glucose-infused groups, and among the glucose- and placebo-infused groups by analysis of variance for repeated measures. We also compared plasma osmolality concentrations between the glucose-infused and our reported-mannitol infused (Stonestreet et al, 2006) fetal sheep at 60% of gestation, preterm lambs, and newborn lambs by three-factor analysis of variance for repeated measures with time, age groups, and glucose versus mannitol infusions as the factors. Similar analyses were used to compare the regional brain water contents among the placebo-, glucose-, and mannitol-infused sheep from our earlier studies (Stonestreet et al, 2003, 2006). If a significant difference was found by analysis of variance, the Neumann—Keuls post hoc test was used. All data were expressed as mean ± s.e.m. P<0.05 was considered statistically significant unless otherwise specified.

Modeling Blood-to-Brain Transfer Constant Versus Plasma Osmolality in the Glucose-Infused Sheep

We found that the time-averaged plasma osmolality values represented a statistically higher correlation with the K_i values than the final plasma osmolality values (Stonestreet et al, 2006). The time-averaged plasma osmolality values were calculated individually for each animal during the continuous infusions. The scattergrams (Figures 1 –3) represent the blood-to-brain transfer constant (K_i) plotted on the y axis against the time-averaged plasma osmolality values for the duration of the study for each animal on the x axis. The scattergrams showed an increase in the K_i values as plasma osmolarity increased. However, the rate of change in K_i was not constant throughout the range of values. Therefore, we determined the relationship between K_i and plasma osmolality among age groups within each brain region, with a segmented regression model to fit a linear regression model for each segment with a separate slope, and an unknown break point between the two lines (Stonestreet et al, 2006). The estimated break point between the lines is termed the ‘threshold’ (Cserr et al, 1987b; Stonestreet et al, 2006).

Figure 1

(A) Superior colliculus. (B) Pons. The closed circles represent the glucose-infused sheep and the open circles represent the historic data from the placebo-infused sheep in our previous study (Stonestreet et al, 2006). Closed circles represent new experimental data from the current experiments. The y axis represents the blood-to-brain transfer constant (K_i) and the × axis the time-averaged plasma osmolality values for the duration of the study. K_i plotted against plasma osmolality in the superior colliculus and pons for fetuses at 60% of gestation, preterm lambs at 90% of gestation, and newborn lambs at 2 to 6 days of age. The solid lines represent the first and second segments from the segmented regression model as described under statistical analysis and the point between the two lines represent the threshold. The dashed lines represent the same parameters from our previous study examining the effects of mannitol-induced hyperosmolality (Stonestreet et al, 2006). Individual data points for the mannitol-induced hyperosmolality in the fetuses and lambs not shown for simplicity. *P <0.05 indicates that the value of the threshold for glucose (break point between the solid lines) is significantly higher than the value of the threshold for mannitol (break point between the dashed lines). ⁺P <0.05 indicates that the value of the threshold for glucose is significantly higher in the newborn than in the fetuses at 60% of gestation.

Figure 2

(A) Cerebral cortex. (B) Hippocampus. K, plotted against plasma osmolality in the cerebral cortex and hippocampus. Groups, symbol legends, and description of the lines as for Figure 1. *P < 0.05 indicates that the value of the threshold for glucose (break point between the solid lines) is significantly higher than the value of the threshold for mannitol (break point between the dashed lines).

Figure 3

(A) Cerebellum. (B) Medulla. K, plotted against plasma osmolality in the cerebellum and medulla. Groups, symbol legends, and description of the lines as for Figure 1. *P < 0.05 indicates that the value of the threshold for glucose (break point between the solid lines) is significantly higher than the value of the threshold for mannitol (break point between the dashed lines). ⁺⁺P <0.05 indicates that the slope of the second regression segment was significantly higher in mannitol-than in glucose-treated newborn lambs, implying greater sensitivity to the mannitol-than to glucose-hyperosmotic stress after the threshold had been reached.

We fit equation 2 to the relationship between K_i (y axis) and plasma osmolality (x axis), which is equivalent, to equation 2 in our previous report (Stonestreet et al, 2006):

y = b_{1} \times m i n i m u m (x, a) + b_{2} \times m a x i m u m (x, a) + c

(2)

where b₁ and b₂ are the coefficients of the slopes of the first and second segments, ‘a’ is the threshold between the two segments, and c is a constant (Shuai et al, 2003). In this and our former work (Stonestreet et al, 2006), we used the NONLIN Program in SAS to fit the models (Schabenberger, 1998).

We examined the relationship between K_i and plasma osmolality during development by comparing the thresholds and slopes of the second regression segment among the fetuses and lambs within each brain region. Lower thresholds and steeper slopes of the post-threshold segments indicate greater vulnerability to the hyperglycemic-hyperosmotic stress within a given brain region (Stonestreet et al, 2006). Pairwise comparisons of the thresholds and slopes were made for the three age groups of fetuses and lambs within each brain region. There were three pairwise comparisons for the age groups within each brain region (Figures 1 –3; Table 4). Therefore, a modified Bonferronitype procedure, with an approximate two-sample t-test that adjusted for the appropriate α values, was used to determine the critical t values to compare the thresholds and slopes within the brain regions as suggested by Holm (Glantz, 2002; Stonestreet et al, 2006).

Comparison of Ki Versus Plasma Osmolality Model Parameters Between the Glucose- and Mannitol-Infused Sheep

We compared the effects of glucose and mannitol hyperosmolality on the relationship between K_i and plasma osmolality by fitting the segmented regression model described in equation 2 as follows (equation 3 and 4 for glucose and mannitol, respectively):

\begin{aligned} G l u c o s e : K_{i} = b_{11} \times m i n i m u m (x, a_{1}) \\ + b_{21} \times m a x i m u m (x, a_{1}) + c_{1} \end{aligned}

(3)

\begin{aligned} M a n n i t o l : K_{i} = b_{21} \times m i n i m u m (x, a_{2}) \\ + b_{22} \times m a x i m u m (x, a_{2}) + c_{2} \end{aligned}

(4)

where the b's refer to the coefficients of the slopes, a's are the thresholds between the two segments, and c's are the constants as in equation 2 above and in our previous publication (Stonestreet et al, 2006). We applied permutation tests suggested by Manly (2006) to compare the thresholds a₁ and a₂ and the second slopes b₂₁ and b₂₂ for each of the three age groups and brain regions. We used this method because the placebo-infused fetuses at 60% of gestation, preterm, and newborn lambs provided the basal ‘control’ values both in the current study (Figures 1 –3, open circles) and our former work (Stonestreet et al, 2006). For example, when we compared the effects of the glucose and mannitol hyperosmolality on the relationship of the K_i values with the plasma osmolalities in the fetal sheep at 60% of gestation, 16 fetuses were exposed to the mannitol infusions, 14 to the glucose infusions, and 7 to the placebo infusions. However, the same seven sheep exposed to the placebo infusions provided the basal ‘control’ values for both the glucose-(current work: open circles, Figures 1 –3) and mannitol-infused fetal sheep (open circles: Figures 2–5 from our previous publication) (Stonestreet et al, 2006). Both data sets contain the same seven placebo-infused basal ‘control’ sheep and, hence, cannot be considered completely independent but partially dependent. Consequently, the estimated slopes b₂₁ and b₂₂ and threshold values a1 and a2 are not independent. Thus, the comparisons of the thresholds and the slopes of the second segment could not be compared using traditional t-tests, as the covariances between them could not be computed when the models are fitted separately.

We describe the permutation tests to compare the thresholds for cerebrum in the fetuses at 60% of gestation as follows. First, we fit the models which yielded a threshold of a₁ = 346 for glucose, and of a₂ =334 for mannitol and the difference between them was 12. From 30 observations on the mannitol- and glucose-exposed fetal sheep, a random selection of 16 observations was made. These 16 observations combined with the seven observations from the placebo-infused fetal sheep provided a new mannitol data set. The remaining 14 observations combined with the same seven observations from the placebo-infused fetal sheep provided a new glucose data set. For each new data set, the segmented regression models were fitted yielding threshold values and slopes and their differences were determined. This process was repeated 5,000 times. In nonlinear model fitting methods, the program may not always converge as observed in the current analysis. The 5,000 permutations resulted in a number of convergent cases, which varied between 500 and 4,000. For each convergent case, we computed the difference between a₂ and a₁. The proportion of such differences exceeding 12 or below 12 constituted the P value for the test.

We verified the stability of the procedure by using two methods. In the first method, the number of permutations was increased to 10,000 in some select sample cases. In the second, some of the results obtained using the first method were compared with results using t-tests for comparisons of the thresholds and slopes among the different age groups, which did not require knowledge of the covariances. This analysis was also repeated using the permutation test method. The rate of convergent cases and the P values for the 5,000 and 10,000 permutations demonstrated excellent concordance. The P values from t-tests and permutation tests also yielded similar results supporting the same statistical conclusions. Consequently, this procedure could be used to compare both the thresholds (Table 4) and slopes between the fetuses at 60% of gestation, preterm, and newborn lambs exposed to glucose- and mannitol-induced hyperosmolality.

Results

Weight increased with age as expected (Table 1). The total osmolar load administered to the fetuses plus ewes, and lambs are summarized in Table 1.

The glucose-treated sheep within each group demonstrated stable elevations in plasma osmolality during the studies as we have reported (Stonestreet et al, 2004). The plasma osmolality values during the studies were higher (P<0.05) in the newborn lambs than in the fetuses at 60% of gestation. Examination of a wide range of plasma osmolality values (Figures 1 –3) within each group of sheep was essential to determine the effects of changes in osmolalities on BBB function. We also compared the plasma osmolality values from the glucose-treated sheep with the mannitol-exposed fetuses at 60% of gestation, preterm, and newborn lambs from our previous report (Stonestreet et al, 2006), and did not find differences in the plasma osmolality values during the studies between the sheep exposed to the glucose and mannitol infusions (F =0.49, P =0.94). The plasma glucose concentrations of the preterm and newborn lambs were higher (P <0.05) than the fetuses at 60% of gestation (Stonestreet et al, 2004).

Table 2 contains the physiological variables for the glucose-treated sheep before (baseline) and during the glucose infusions. There were no significant differences in the baseline physiological variables among the placebo-, mannitol- (data not shown), and the glucose-treated sheep (Table 2). Arterial pH decreased in the fetuses and newborn lambs during the glucose infusions. Small decreases also were observed in arterial oxygen tension in the newborn lambs, and increases in carbon dioxide tension in the fetuses and newborn lambs. Heart rate decreased in the fetuses and increased in newborn lambs during the glucose infusions. Mean arterial blood pressure decreased in the preterm lambs.

Table 2

Arterial pH, blood gases, heart rate and mean arterial blood pressure by study group

		Baseline	Glucose infusion
Time, min	Groups	–30	30	60
pH	Fetus: 60% gestation	7.38±0.01	7.32±0.02*	7.30±0.02*
	Preterm lamb	7.41±0.02	7.36±0.03	7.32±0.03
	Newborn lamb	7.40±0.01	7.26±0.02*	7.23±0.04*
PO₂, mm Hg	Fetus: 60% gestation	26±0	24±1	25±1
	Preterm lamb	109±9	110±11	99±9
	Newborn lamb	88±2	76±3*	75±5*
PCO₂, mm Hg	Fetus: 60% gestation	48±1	61±2*	66±3*
	Preterm Lamb	38±2	33±2	35±3
	Newborn lamb	41±1	51±2*	52±4*
Heart rate, beats/min	Fetus: 60% gestation	200±3	189±4*	195±5
	Preterm lamb	163±3	183±6	171±12 (10)
	Newborn lamb	196±10	225±9*	211±11 (16)
Mean arterial blood pressure, mmHg	Fetus: 60% gestation	38±2	38±3	38 ±2
	Preterm lamb	60±3	51±6*(10)	46± 5*(9)
	Newborn lamb	75±3 (15)	75±3 (14)	73±3 (13)

Values mean ± s.e.m.; glucose-infused: fetuses at 60% of gestation, n = 14, preterm lambs at 90% of gestation, n = 11, lambs at 3–5 days of age, n = 17, unless indicated in parenthesis. * P < 0.05 versus baseline. Values shown are for the experimental glucose-infused sheep; values for the placebo-infused and mannitol-infused sheep have been previously reported (Stonestreet et al, 2006) and are not shown.

The blood-to-brain transfer constant K_i increased as a nonlinear function of glucose-induced plasma osmolality in the all brain regions and age groups. At the lower osmolality values, the change in K_i was minimal but after a threshold was reached, the increase (P<0.05) in K_i was linear. The slopes of the first and second regression segments did not differ among any of the age groups in any brain region (Table 3).

Table 3

Values of the slopes and the standard errors of the first (b₁) and second (b₂) regression segments in the brain regions of the glucose-treated sheep by study group

Brain regions	Fetuses at 60% of gestation		Premature lambs		Newborn lambs
Brain regions	b ₁	b₂	b₂	b₂	b₁	b ₂
Cerebral cortex	0.004±0.032	0.120±0.039	0.010±0.024	0.010±0.040	0.014±0.010	0.096±0.045
Hippocampus	0.033± 0.057	0.150± 0.037	0.008± 0.020	0.110± 0.035	0.018± 0.007	0.057± 0.031
Cerebellum	0.019± 0.097	0.310 ± 0.118	0.010± 0.062	0.140± 0.066	0.025± 0.014	0.050± 0.090
Thalamus	0.043± 0.013	0.082± 0.051	0.007± 0.015	0.028± 0.027	0.005± 0.005	0.045± 0.037
Superior colliculus	0.011± 0.059	0.150± 0.037	0.009± 0.032	0.075 ± 0.035	0.019 ± 0.008	0.087± 0.033
Inferior colliculus	0.009±0.053	0.180±0.066	0.008±0.029	0.052 ± 0.031	0.020 ± 0.009	0.069±0.042
Pons	−0.017± 0.070	0.110 ± 0.048	0.006± 0.033	0.042 ± 0.031	0.009 ± 0.006	0.094± 0.044
Medulla	0.039± 0.018	0.250± 0.110	0.005± 0.020	0.050 ± 0.035	0.013 ± 0.006	0.030± 0.028
Cervical spinal cord	−0.024± 0.032	0.160± 0.064	0.002± 0.024	0.056 ± 0.026	0.006± 0.005	0.037± 0.022

There were no statistically significant differences in the b₁ or b₂ slopes within each brain region among the fetuses at 60% of gestation, premature lambs, and newborn lambs.

Figure 1 illustrates the pattern of the K_i values as plasma osmolality increased in the superior colliculus and pons of the fetuses at 60% of gestation, preterm, and newborn lambs. The solid lines illustrate the changes in K_i as a function of hyperglycemic hyperosmolality for the new experimental studies, whereas the dashed lines are shown for comparison and represent summary data from our former study examining the effects of mannitol hyperosmolality on the changes in K_i (Stonestreet et al, 2006). In the superior colliculus and pons, the thresholds for osmotic barrier opening were significantly higher in the newborn lambs than in the fetuses at 60% of gestation. The threshold was also significantly higher in superior colliculus of the glucose- than mannitol-treated preterm lambs.

Figure 2 illustrates the pattern of change in K_i as plasma osmolality increased in the cerebral cortex and hippocampus of the fetuses at 60% of gestation, preterm, and newborn lambs. There were no significant differences in the threshold values or slopes of the second segments in the cerebral cortex or hippocampus among any of the glucose-treated age groups. The line designations are as for Figure 1. The threshold was significantly higher in cerebral cortex of the glucose- than mannitol-treated newborn lambs.

Figure 3 shows the pattern of change in K_i as plasma osmolality increased in the cerebellum and medulla. There were no significant differences in the threshold values or slopes of the second segments in the cerebellum or medulla among any of the glucose-treated age groups. The line designations are as for Figure 1. The thresholds were significantly higher in the cerebellum of the glucose- than mannitol-treated newborn lambs and in the medulla of the glucose- than mannitol-treated preterm lambs. The slope of the second regression segment was significantly higher in the medulla of the mannitol- than glucose-treated newborn lambs, implying greater sensitivity to the mannitol- than glucose-hyperosmotic stress after the threshold had been reached (Figure 3, dashed line for mannitol versus solid line for glucose). Otherwise, there were no significant differences between the slopes of the second regression segment in any of the other brain regions between the glucose- and mannitol-treated sheep (Figures 1 –3; Stonestreet et al, 2006).

Similar patterns of change in K_i versus plasma osmolality were observed in the thalamus, inferior colliculus, and cervical spinal cord (graphic data not shown).

Table 4 compares the thresholds by age group and brain regions in the sheep treated with the glucose infusions from the current study and the mannitol infusions from our previous publication (Stonestreet et al, 2006). The thresholds were significantly higher in the thalamus, superior colliculus, inferior colliculus, and medulla of the glucose- than mannitol-treated preterm lambs, and higher in the cerebrum and cerebellum of the glucose- than mannitol-treated newborn lambs.

Table 4

Plasma osmolality values of the estimated thresholds in brain regions of the sheep treated with glucose or mannitol by study group

Brain regions	Fetuses at 60% of gestation		Premature lambs		Newborn lambs
	Glucose	Mannitol	Glucose	Mannitol	Glucose	Mannitol
Cerebral cortex	346±57	334 ±102	385 ±24	344 ± 18	398 ±15*	323 ±18
Hippocampus	337 ± 22	334 ± 71	386 ±17	344 ± 33	398 ±22	366 ±10
Cerebellum	342 ±21	334 ±82	376 ± 36	343 ±13	398 ±10*	322 ±34
Thalamus	375 ±18	340 ±18	379 ±67*	303 ±58	409 ±18	403 ±6
Superior colliculus	338 ±19	346 ± 15	380 ± 38*	332 ±28	403 ±13⁺	400 ±9
Inferior colliculus	338 ±19	362 ±5	380 ±51*	330 ±50	393 ±28	352 ±23
Pons	336 ± 25	339 ±18	380 ± 72	377 ±25	408 ±10+	406 ±6
Medulla	372 ±8	359 ±7	386 ±42*	332 ±23	399 ±45	377 ± 9
Cervical spinal cord	359± 12	334 ±139	380 ± 35	362 ±27	398 ±19	371 ±26

Plasma osmolality values are mosmol/H₂O. Mannitol values from previous publication (Stonestreet et al, 2006).

P < 0.05 versus mannitol for the same age group

P < 0.05 versus fetuses at 60% of gestation for the glucose treatment group.

The regional brain water content was lower in the glucose- and mannitol- than in the placebo-infused sheep (Table 5). The regional water content was lower in cerebral cortex and cerebellum of the mannitol- than in the glucose-treated fetuses at 60% of gestation, and in cerebral cortex and medulla of the mannitol- than in the glucose-treated premature lambs. The changes in regional brain water contents did not differ significantly between the mannitol- and glucose-infused newborn lambs. The final plasma osmolality values did not differ significantly among the groups of the glucose- and mannitol-infused sheep.

Table 5

Brain water content values in brain regions of placebo-, glucose- and mannitol-treated sheep

Brain regions	Fetuses at 60% of gestation			Premature lambs			Newborn lambs
Water content	Placebo	Glucose	Mannitol	Placebo	Glucose	Mannitol	Placebo	Glucose	Mannitol
Cerebral cortex	9.77 ± 0.12	8.41 ±0.12*	8.07 ± 0.11*⁺	5.62 ±0.07	4.92 ±0.15*	4.43 ±0.10*⁺	5.14 ± 0.12	4.01 ±0.08*	3.98 ±0.10*
Cerebellum	9.03 ±0.23	7.64 ±0.16*	7.09 ± 0.13*⁺	4.57 ± 0.05	3.95 ±0.09*	3.67 ±0.07*	4.30 ± 0.12	3.45 ±0.06*	3.36 ± 0.07*
Medulla	7.23 ±0.14	6.04 ±0.11*	5.73 ±0.08*	3.61 ±0.04	3.11 ±0.09*	2.76 ±0.13*⁺	3.32 ±0.12	2.63 ±0.06*	2.50 ± 0.06*
Osmolality	296±2	377 ± 5*	375 ±2*	303 ±1	395 ±13*	384 ± 9*	301 ±3*	407 ±10*	409 ±10*

Values are mean ± s.e.m. Brain water content given in ml/g per dry wt. Plasma osmolality given in mosmol/kg H₂O for the final values the end of the studies.

Values represent results from historical animals obtained from our previous work (Stonestreet et al, 2003, 2004).

P<0.05 versus placebo for the same age group

P<0.05 versus glucose for the same age group

Discussion

The purpose of the present study was to determine the relative permeability of the BBB in immature animals in response to a hyperglycemic-hyperosmotic stress during development, and to determine if the BBB is more resistant to glucose- than to mannitol-induced hyperosmolality (Stonestreet et al, 2006). There are three major findings in this study. (1) BBB permeability increases as a function of changes in systemic glucose-induced hyperosmolality in fetal sheep and lambs, but the rate of change in K_i is not constant throughout the range of plasma osmolalities and is best described by a segmented regression model (Stonestreet et al, 2006). (2) The BBB becomes more resistant to glucose-induced hyperosmolality during development only in two select brain regions, the superior colliculus and pons. (3) The barrier is more resistant to glucose than to mannitol hyperosmolality in some brain regions of premature and newborn lambs, but not in fetuses at 60% of gestation.

Arterial pH decreased slightly in the fetal sheep at 60% of gestation and in the newborn lambs because of a mild respiratory acidosis. The decreases in arterial pH were similar to those observed during mannitol-induced hyperosmolality in the same age groups of sheep (Stonestreet et al, 2006). Likewise, the increases in arterial carbon dioxide tensions in the fetuses at 60% of gestation and in the newborn lambs during the glucose infusions were similar to those during the mannitol infusions (Stonestreet et al, 2006). Although hypercapnia results in increased penetration of sucrose into the brains of fetal and newborn sheep and adult rabbits, the carbon dioxide levels in the previous reports were much higher than during glucose (Table 1) and mannitol hyperosmolality in our studies (Cameron et al, 1969; Evans et al, 1976; Stonestreet et al, 2006). The increases in the arterial carbon dioxide tensions in the fetuses at 60% of gestation and in the newborn lambs were similar and analogous to those we reported during the mannitol infusions (Stonestreet et al, 2006). Hence, the elevations in carbon dioxide tension should not affect the age-related comparisons of barrier permeability between the fetuses at 60% of gestation and newborn lambs, or between the glucose- and mannitol-infused sheep in the same age groups (Table 4).

We have previously shown that the BBB demonstrates ontogenic decreases in permeability in normo-osmotic ovine fetuses from 60% of gestation up to 3 years of age in the adult (Stonestreet et al, 1996), and that the BBB becomes more resistant to the effects of a mannitol-induced hyperosmotic stress during development (Stonestreet et al, 2006). The current study extends our work by examining the developmental regulation of BBB permeability to a glucose-induced hyperosmotic stress. Similar to our findings with mannitol-induced hyperosmolality, hyperglycemic-hyperosmotic stress affects barrier function to some extent in all brain regions and age groups of sheep because the pattern of change in K_i versus plasma osmolality was described by a segmental regression model and an osmotic threshold was identified in all brain regions in each age group.

We cannot be certain of the mechanism by which hyperglycemic hyperosmolality impaired barrier function in our fetal sheep and lambs. Previous work suggested that the ‘threshold effect’ results from osmotic opening of pores created between adjacent endothelial cells resulting from opening of tight junctions, and water-filled channels at the blood-brain interface, and that increased diffusion occurs via a leak pathway (Cserr et al, 1987b; Ziylan et al, 1984). These phenomena could account for the increases in K_i resulting from glucose- and mannitol-induced osmotic stresses (Stonestreet et al, 2006). Furthermore, consistent with the decreases in brain water contents in the glucose- and mannitol-infused sheep (Table 5), osmotic stress-related increases in barrier permeability potentially could have resulted from shrinkage of endothelium at the BBB associated with brain dehydration (Rapoport et al, 1980). The phenomenon has previously been characterized as ‘osmotic opening’ of the BBB when hyperosmotic substances were applied directly to the pial surface (Cserr et al, 1987b; Rapoport et al, 1972) or injected into the carotid artery (Blasberg et al, 1980; Ziylan et al, 1984). Therefore, the ‘threshold’ that we identified between the two lines, presumably is similar to the osmotic opening of the BBB after the application of topical substances or the intracarotid infusions (Blasberg et al, 1980; Cserr et al, 1987b; Rapoport et al, 1972; Ziylan et al, 1984). Before reaching the threshold, the glucose-induced increases in osmolality resulted in small increases (Figures 1 –3) in K_i such that barrier function was in the near normal range (Cserr et al, 1987b). It is also important to point out that the changes in K_i that we have determined with AIB as the tracer are not dependent on changes in cerebral blood flow because AIB is not a blood flow-dependent substance (Fenstermacher, 1984; Stonestreet et al, 1996).

In this and our former work (Stonestreet et al, 2006), we examined the responses of the BBB for any two age groups within each brain region by comparing the thresholds and slopes of the second regression segments. In contrast to our hypothesis, and the developmental increases in barrier permeability observed with mannitol-induced hyperosmolality, the estimated osmotic thresholds demonstrated limited developmental increases during glucose-induced hyperosmolality with higher thresholds confined to the superior colliculus and pons in the newborn lambs compared with the fetuses at 60% of gestation (Stonestreet et al, 2006). Although we found significantly higher thresholds only in the superior colliculus and pons, inspection of Table 4 suggests that the values for the thresholds generally tended to be higher in the premature and newborn lambs than in the fetuses at 60% of gestation. These findings are important because there is very little information regarding the susceptibility of the BBB to injury during development (Stolp et al, 2005). The greater sensitivity of select brain regions to glucose-induced hyperosmolality in the fetuses at 60% of gestation compared with newborn lambs suggests greater vulnerability of the barrier to this stress in the more immature subjects in some brain regions.

Glucose is actively transported across the BBB by a glucose transporter, GLUT-1, which is located in endothelial cells and closely associated with the interendothelial junctional complexes (Devaskar et al, 1991; Vannucci, 1994; Vorbrodt et al, 2001). GLUT-1 expression has been reported to be higher in the brain and cerebral microvessels of fetuses and newborns than in adult rodents (Devaskar et al, 1991; Vannucci, 1994; Vorbrodt et al, 2001). Thus, if GLUT-1 expression increases with age and, hence, larger amounts of glucose potentially are transported into the brain, such a phenomenon could in part account for the smaller effect of glucose hyperosmolality on the threshold values in some brain regions of the newborn lambs compared with the fetuses at 60% of gestation (Table 4). Moreover, it is also important to point out that acute hyperglycemia in rats does not increase GLUT-1 levels in cerebral microvessels (Simpson et al, 1999).

There is increasing survival of extremely premature infants at the lower limits of viability, who potentially could be inadvertently exposed to hyperosmolality, which predisposes them to intraventricular hemorrhage (El-Metwally et al, 2000; Glasgow et al, 1983; McDonald et al, 1989; Thomas, 1976). We speculate that if the immature BBB exhibits osmotic stress-related increases in permeability, then impaired BBB function, short of complete barrier breakdown, e.g., hemorrhage, could also predispose infants to brain damage by facilitating entry of toxic substances into brain parenchyma that could alter neuronal function. The mechanism(s) by which the BBB is more vulnerable to glucose-hyperosmotic stress in some regions of the brain in the fetuses than in the newborn lambs cannot be determined in our study. However, although we have identified differential developmental regulation in the expression of the some tight junction proteins, their expression was not necessarily lower early in fetal development (Duncan et al, 2009). Nonetheless, based on our findings in this and our previous work (Stonestreet et al, 2006), the tight junctions of the BBB could be more vulnerable to injury at an early time in fetal development than after birth.

Premature infants often exhibit glucose dysregulation and are at high risk for brain injury (Dweck, 1976; Grylack et al, 1984; Vohr et al, 2000; Wu, 1996). The major patterns of brain injury in premature infants are intraventricular/periventricular hemorrhage and periventricular leukomalacia (Pierson et al, 2007; Vohr and Ment 1996; Volpe, 2003). However, brain injury has also been reported in the cerebral cortical gray matter, thalamus, basal ganglia, globus pallidus, and cerebellar dentate nucleus in association with periventricular leukomalacia (Pierson et al, 2007). We cannot determine from our study if impaired barrier function associated with hyperglycemic-hyperosmotic stress contributes to brain damage in premature infants. Nonetheless, we observed increased barrier permeability in all of the brain regions (Table 4) that we examined in the fetuses at 60% of gestation and premature lambs, suggesting that impaired barrier function could represent a component of injury in premature infants exposed to a hyperglycemic hyperosmolality in such brain regions as the gray matter, cerebellum, and thalamus.

Our findings of impaired barrier function with hyperglycemic-hyperosmotic stress are consistent with work in adult rodents and humans, demonstrating that diabetes can impair BBB integrity (Allen and Bayraktutan, 2009; Hawkins et al, 2007; Huber et al, 2006; Starr et al, 2003). Increased BBB permeability was detected in patients with type II diabetes (Starr et al, 2003). Similarly, streptozotocin-induced diabetes in adult rats was associated with increases in permeability to small molecules, but similar to our findings, the changes were region specific with the largest changes occurring in the midbrain (Huber et al, 2006). The increases in BBB permeability in streptozotocin-induced diabetes appear to be related to a loss of tight junction proteins possibly resulting from increases in plasma matrix metalloproteinase activity and not from hyperglycemia alone (Hawkins et al, 2007). We cannot be certain if impaired barrier function resulting from hyperglycemic-hyperosmotic stress also could be related to a loss of tight junction proteins resulting from increases in matrix metalloproteinase activity because these variables were not determined in our studies.

We also compared the BBB responses between the different age groups of sheep exposed to the glucose- and mannitol-hyperosmotic stresses (Stonestreet et al, 2006). To make valid comparisons between the glucose- and mannitol-hyperosmotic fetuses, preterm, and newborn lambs, we needed to establish that the patterns of change in plasma osmolality during the studies were similar for the two hyperosmotic treatments. Our analysis confirmed that the patterns of change in plasma osmolality during the studies were similar between glucose- and mannitol-treated sheep (Stonestreet et al, 2004, 2006). The osmotic threshold values were significantly higher in the thalamus, superior colliculus, inferior colliculus, and medulla, and in the cerebral cortex and cerebellum of the glucose- than mannitol-exposed preterm and newborn lambs, respectively. In addition, the slope of the second regression segment was significantly steeper in the medulla of the mannitol- than glucose-treated newborn lambs, implying greater sensitivity of this region to mannitol- than glucose-hyperosmotic stress after the threshold was reached. Consistent with the Staverman reflection coefficient theory (Staverman, 1951) and work suggesting that if an osmotic solute penetrates the cell membrane, cell shrinkage is less than shrinkage resulting from an equal concentration of an impermeable solute (Rapoport et al, 1972), osmotic barrier modification occurred at lower plasma osmolalities in the mannitol- than in the glucose-treated preterm and newborn lambs (Table 4). Thus, the effective osmotic pressure at the BBB was probably greater for mannitol than for glucose exposure. The lower brain water contents in the mannitol- than in glucose-infused groups (Table 5) support the contention that concentrated solutions impair BBB function by shrinking cerebrovascular endothelial cells, and widening the interendothelial tight junctions, and that the damaging effects are inversely proportional to the ability of a solute to penetrate cell membranes (Rapoport et al, 1972; Rapoport et al, 1980).

In summary, we conclude that BBB permeability increases as a function of changes in glucose-induced plasma osmolality in fetal sheep and lambs. The BBB becomes relatively more resistant to glucose hyperosmolality only in the pons and superior colliculus during development, and the BBB is more resistant to glucose- than to mannitol-osmotic stress in some brain regions of premature and newborn lambs.

Footnotes

Acknowledgements

The authors acknowledge and are thankful to the late Helen F Cserr, PhD for her superb scientific guidance at the original conception of these studies.

The authors declare no conflict of interest.

References

Allen

Bayraktutan

(2009) Antioxidants attenuate hyperglycaemia-mediated brain endothelial cell dysfunction and blood-brain barrier hyperpermeability. Diabetes Obes Metab 11:480–90

Back

Riddle

Hohimer

(2006) Role of instrumented fetal sheep preparations in defining the pathogenesis of human periventricular white-matter injury. J Child Neurol 21:582–9

Betz

Goldstein

(1986) Specialized properties and solute transport in brain capillaries. Annu Rev Physiol 48:241–50

Blasberg

Gazendam

Patlak

Shapiro

Fenstermacher

(1980) Changes in blood-brain transfer parameters induced by hyperosmolar intraocarotid infusion and by metastatic turmor growth. In: The Cerebral Microvasculature ( Eisenberg

Suddith

, eds). New York: Plenum, 307–19

Bradbury

(1985) The blood-brain barrier. Transport across the cerebral endothelium. Circ Res 57:213–22

Cameron

Davson

Segal

(1969) The effect of hypercapnia on the blood-brain barrier to sucrose in the rabbit. Yale J Biol Med 42:241–7

Cserr

DePasquale

Patlak

(1987a) Regulation of brain water and electrolytes during acute hyperosmolality in rats. Am J Physiol 253:F522–9

Cserr

DePasquale

Patlak

(1987b) Volume regulatory influx of electrolytes from plasma to brain during acute hyperosmolality. Am J Physiol 253:F530–7

Devaskar

Zahm

Holtzclaw

Chundu

Wadzinski

(1991) Developmental regulation of the distribution of rat brain insulin-insensitive (Glut 1) glucose transporter. Endocrinology 129:1530–40

10.

Duncan

Sadowska

Stonestreet

(2009) Ontogeny and the effects of exogenous and endogenous glucocorticoids on tight junction protein expression in ovine cerebral cortices. Brain Res 1303:15–25

11.

Dweck

(1976) Neonatal hypoglycemia and hyperglycemia: two unique perinatal metabolic problems. Postgrad Med 60:118–24

12.

El-Metwally

Vohr

Tucker

(2000) Survival and neonatal morbidity at the limits of viability in the mid 1990s: 22 to 25 weeks. J Pediatr 137:616–22

13.

Evans

Reynolds

Saunders

(1976) The effect of hypercapnia on a blood-brain barrier mechanism in foetal and newborn sheep. J Physiol 255: 701–14

14.

Fenstermacher

(1984) Volume Regulation of the Central Nervous System. New York: Raven Press, 383–404

15.

Glantz

(2002) Primer of Biostatistics 5th edn. 92–5

16.

Glasgow

Boeckx

Miller

MacDonald

August

Goodman

(1983) Hyperosmolality in small infants due to propylene glycol. Pediatrics 72:353–5

17.

Grylack

Chu

Scanlon

(1984) Use of intravenous fluids before cesarean section: effects on perinatal glucose, insulin, and sodium homeostasis. Obstet Gynecol 63:654–8

18.

Gunn

de Haan

Williams

Gluckman

(1997) Dramatic neuronal rescue with prolonged selective head cooling after ischemia in fetal lambs. J Clin Invest 99:248–56

19.

Hawkins

Lundeen

Norwood

Brooks

Egleton

(2007) Increased blood-brain barrier permeability and altered tight junctions in experimental diabetes in the rat: contribution of hyperglycaemia and matrix metalloproteinases. Diabetologia 50:202–11

20.

Huber

VanGilder

Houser

(2006) Streptozotocin-induced diabetes progressively increases blood-brain barrier permeability in specific brain regions in rats. Am J Physiol Heart Circ Physiol 291:H2660–8

21.

Manly

BFZ

(2006) Randomization, Bootstrap and Monte Carlo Methods in Biology. Atlanta: Chapman and Hall/CRC, 1-20pp

22.

McDonald

Martha

Jr Kerrigan

Clarke

Rogol

Blizzard

(1989) Treatment of the young child with postoperative central diabetes insipidus. Am J Dis Child 143:201–4

23.

Ohno

Pettigrew

Rapoport

(1978) Lower limits of cerebrovascular permeability to nonelectrolytes in the conscious rat. Am J Physiol 235:H299–307

24.

Otto Buczkowska

Szirer

Jarosz-Chobot

(2001) Glucose homeostasis in children. I. Regulation of blood glucose. Przegl Lek 58:20–4

25.

Pierson

Folkerth

Billiards

Trachtenberg

Drinkwater

Volpe

Kinney

(2007) Gray matter injury associated with periventricular leukomalacia in the premature infant. Acta Neuropathol 114:619–31

26.

Rapoport

Fredericks

Ohno

Pettigrew

(1980) Quantitative aspects of reversible osmotic opening of the blood-brain barrier. Am J Physiol 238:R421–31

27.

Rapoport

Hori

Klatzo

(1972) Testing of a hypothesis for osmotic opening of the blood-brain barrier. Am J Physiol 223:323–31

28.

Schabenberger

(1998) Nonlinear regression with the SAS system 1–16

29.

Shuai

Zhou

Yost

(2003) Using segmented regression models to fit soil nutrient and soybean grain yield changes due to liming. Environ Stat 8: 240–52

30.

Simpson

Appel

Hokari

Oki

Holman

Maher

Koehler-Stec

Vannucci

Smith

(1999) Blood-brain barrier glucose transporter: effects of hypo- and hyperglycemia revisited. J Neurochem 72:238–47

31.

Starr

Wardlaw

Ferguson

MacLullich

Deary

Marshall

(2003) Increased blood-brain barrier permeability in type II diabetes demonstrated by gadolinium magnetic resonance imaging. J Neurol Neurosurg Psychiatry 74:70–6

32.

Staverman

(1951) The theory of measurement of osmotic pressure. Recl Trav Chim Pays-Bas 70:344–52

33.

Stolp

Dziegielewska

Habgood

Lane

Potter

Saunders

(2005) Breakdown of the blood-brain barrier to proteins in white matter of the developing brain following systemic inflammation. Cell Tissue Res 320:369–78

34.

Stonestreet

McKnight

Sadowska

Petersson

Oen

Patlak

(2000) Effects of duration of positive-pressure ventilation on blood-brain barrier function in premature lambs. J Appl Physiol 88:1672–7

35.

Stonestreet

Oen-Hsiao

Petersson

Sadowska

Patlak

(2003) Regulation of brain water during acute hyperosmolality in ovine fetuses, lambs and adults. J Appl Physiol 94:1491–500

36.

Stonestreet

Patlak

Pettigrew

Reilly

Cserr

(1996) Ontogeny of blood-brain barrier function in ovine fetuses, lambs and adults. Am J Physiol 271: R1594–601

37.

Stonestreet

Petersson

Sadowska

Patlak

(2004) Regulation of brain water during acute glucose-induced hyperosmolality in ovine fetuses, lambs, and adults. J Appl Physiol 96:553–60

38.

Stonestreet

Sadowska

Leeman

Hanumara

Petersson

Patlak

(2006) Effects of acute hyperosmolality on blood-brain barrier function in ovine fetuses and lambs. Am J Physiol Regul Integr Comp Physiol 291:R1031–9

39.

Thomas

(1976) Hyperosmolality and intraventricular haemorrhage in premature babies. Acta Paediatr Scand 65:429–32

40.

Vannucci

(1994) Developmental expression of GLUT1 and GLUT3 glucose transporters in rat brain. J Neurochem 62:240–6

41.

Vohr

Ment

(1996) Intraventricular hemorrhage in the preterm infant. Early Hum Dev 44:1–16

42.

Vohr

Wright

Dusick

Mele

Verter

Steichen

Simon

Wilson

Broyles

Bauer

Delaney-Black

Yolton

Fleisher

Papile

Kaplan

(2000) Neurodevelopmental and functional outcomes of extremely low birth weight infants in the national institute of child health and human development neonatal research network, 1993-1994. Pediatrics 105:1216–26

43.

Volpe

(2003) Cerebral white matter injury of the premature infant-more common than you think. Pediatrics 112:176–80

44.

Vorbrodt

Dobrogowska

Tarnawski

(2001) Immunogold study of interendothelial junction-associated and glucose transporter proteins during postnatal maturation of the mouse blood-brain barrier. J Neurocytol 30:705–16

45.

(1996) Infant of diabetic mother: a continuing challenge for perinatal-neonatal medicine. Zhonghua Min Guo Xiao Er Ke Yi Xue Hui Za Zhi 37:312–9

46.

Ziylan

Robinson

Rapoport

(1984) Blood-brain barrier permeability to sucrose and dextran after osmotic opening. Am J Physiol 247:R634–8.