Effects of acute and chronic hyperglycemia on the neurochemical profiles in the rat brain with streptozotocin-induced diabetes detected using in vivo1H MR spectroscopy at 9.4 T

Authors

  • Wen-Tung Wang,

    1. Hoglund Brain Imaging Center, University of Kansas Medical Center, Kansas City, Kansas, USA
    Search for more papers by this author
  • Phil Lee,

    1. Hoglund Brain Imaging Center, University of Kansas Medical Center, Kansas City, Kansas, USA
    2. The Department of Molecular & Integrative Physiology, University of Kansas Medical Center, Kansas City, Kansas, USA
    Search for more papers by this author
  • Hung-Wen Yeh,

    1. The Department of Biostatistics, University of Kansas Medical Center, Kansas City, Kansas, USA
    Search for more papers by this author
  • Irina V. Smirnova,

    1. The Department of Physical Therapy and Rehabilitation Science, University of Kansas Medical Center, Kansas City, Kansas, USA
    Search for more papers by this author
  • In-Young Choi

    1. Hoglund Brain Imaging Center, University of Kansas Medical Center, Kansas City, Kansas, USA
    2. The Department of Molecular & Integrative Physiology, University of Kansas Medical Center, Kansas City, Kansas, USA
    3. The Department of Neurology, University of Kansas Medical Center, Kansas City, Kansas, USA
    Search for more papers by this author

Address correspondence and reprints requests to In-Young Choi, Hoglund Brain Imaging Center, University of Kansas Medical Center, 3901 Rainbow Boulevard, Kansas City, KS 66160, USA. E-mail: ichoi@kumc.edu

Abstract

J. Neurochem. (2012) 121, 407–417.

Abstract

Chronic hyperglycemia could lead to cerebral metabolic alterations and CNS injury. However, findings of metabolic alterations in poorly managed diabetes in humans and animal models are rather inconsistent. We have characterized the cerebral metabolic consequences of untreated hyperglycemia from the onset to the chronic stage in a streptozotocin-induced rat model of diabetes. In vivo1H magnetic resonance spectroscopy was used to measure over 20 neurochemicals longitudinally. Upon the onset of hyperglycemia (acute state), increases in brain glucose levels were accompanied by increases in osmolytes and ketone bodies, all of which remained consistently high through the chronic state of over 10 weeks of hyperglycemia. Only after over 4 weeks of hyperglycemia, the levels of other neurochemicals including N-acetylaspartate and glutathione were significantly reduced and these alterations persisted into the chronic stage. However, glucose transport was not altered in chronic hyperglycemia of over 10 weeks. When glucose levels were acutely restored to euglycemia, some neurochemical changes were irreversible, indicating the impact of prolonged uncontrolled hyperglycemia on the CNS. Furthermore, progressive changes in neurochemical levels from control to acute and chronic conditions demonstrated the utility of 1H magnetic resonance spectroscopy as a non-invasive tool in monitoring the disease progression in diabetes.

Abbreviations used:
Ala

alanine

Asp

aspartate

BBB

blood–brain barrier

bHB

β-hydroxybutyrate

Cho

choline

CMRglc

cerebral metabolic rate of glucose consumption

Cr

creatine

CRLB

Cramér–Rao lower bound

CTL

control

DM

diabetes mellitus

FID

free induction decays

GABA

γ-aminobutyric acid

Glc

glucose

Glu

glutamate

Gln

glutamine

GPC

glycerophosphoryl-choline

Kt

apparent Michaelis–Menten kinetic constant

LCModel

linear combination of model spectra of metabolite solutions in vitro

Ins

myo-inositol

Lac

lactate

MRS

magnetic resonance spectroscopy

NAA

N-acetylaspartate

NAAG

N-acetylaspartylglutamate

PCr

phosphocreatine

RF

radiofrequency

Ser

serine

STEAM

stimulated echo acquisition mode

STZ

streptozotocin

Tau

taurine

tCr

creatine + phosphocreatine

TE

echo time

TM

mixing time

Tmax

apparent maximal transport rate

TR

repetition time

Uncontrolled chronic hyperglycemia, resulting from absolute insulin deficiency (type 1 diabetes) or insulin resistance without or with insulin deficiency (type 2 diabetes), is one of the leading causes of diabetic complications in a number of organs (Porte et al. 2003). Hyperglycemia-induced (or associated) metabolic and vascular disturbances are known to afflict the CNS, increasing risks of stroke, seizures, diabetic encephalopathy and cognitive compromise (McCall 1992). These pathological conditions may result from alterations in cerebral energy homeostasis and metabolism possibly through mechanisms including changes in osmolar gradients in hyperglycemia (Stevens et al. 1993), hormonal regulation (Nurjhan et al. 1985), glucose utilization (Duelli et al. 2000), oxidative stress (Sharma et al. 2010) and the levels of ketone bodies (Felig 1974; Balasse and Fery 1989). However, the effects of diabetes on the CNS, that is, the metabolic consequences associated with hyperglycemia, and the processes increasing the CNS injury have been reported rather inconsistently in both humans and animal models (Biessels et al. 1994; Isales et al. 1999), probably due to the degree of severity and different durations of the disease.

Metabolic alterations reflected in changes of neurochemical levels can occur in the early stages of diabetes prior to pathologic changes in brain structures that occur later in the disease, such as brain atrophy detected using magnetic resonance imaging (Araki et al. 1994; Brands et al. 2007; van Harten et al. 2007; Kumar et al. 2008; Tiehuis et al. 2008). Proton magnetic resonance spectroscopy (1H MRS) allows non-invasive detection of neurochemical levels in vivo. Monitoring metabolic alterations provides unique information on the disease status and progression as the levels of metabolites reflect pathophysiological processes at molecular or cellular levels. Diabetes-related biochemical changes such as the elevated levels of brain glucose (Glc), myo-inositol (Ins) and choline (Cho) in the human cortex have been reported using 1H MRS (Bruhn et al. 1991; Kreis and Ross 1992).

Rodents with streptozotocin (STZ)-induced diabetes are the commonly used animal models of type 1 diabetes. Diabetic complications in the CNS of these animal models have been reported including altered memory and cognitive functions (Biessels et al. 1999; McCall 2004). Previously, changes of neurochemical levels have been reported by measuring a few neurochemicals such as Cho, total creatine [tCr = creatine (Cr) + phosphocreatine (PCr)], and N-acetylaspartate (NAA) in the brains of rats with 28 weeks of STZ-induced diabetes using 1H MRS (Biessels et al. 2001). Recently, the effect of 4 weeks of STZ-induced diabetes on a larger number of neurochemicals in the rat brain was assessed at 9.4 T (Duarte et al. 2009). Duarte and colleagues reported that 4 weeks of hyperglycemia resulted in significant increases in the levels of Cr, Ins, taurine (Tau), NAA, β-hydroxybutyrate (bHB) and glycerophosphoryl-choline (GPC), while the levels of GSH and N-acetylaspartylglutamate (NAAG) were decreased compared with the controls. Upon restoration of the glucose levels to euglycemia via insulin administration, most of the altered neurochemical levels were restored to control levels except for Ins, which showed a sustained increase. While this study provided important information on the disease progression, systematic characterization of metabolic changes in the course of the disease development from acute to chronic hyperglycemic conditions has not been reported, nor any further neurochemical alterations beyond 4 weeks of uncontrolled hyperglycemia. The systematic characterization will provide better understanding of the disease progression and the impact of uncontrolled hyperglycemia on the brain especially at a very early stage in diabetes when subclinical and clinical presentations of diabetes have not developed yet.

The aim of this study was to non-invasively characterize the metabolic consequences of uncontrolled hyperglycemia on the brain in an animal model of type 1 diabetes through the various stages of diabetes from onset to long term. We sought to determine the neurochemical changes that are indicative of the mechanisms and processes during the disease progression in a longitudinal manner, that the progressive nature of CNS injury in respect to stages of diabetes could be better described. In addition, the long-term effect of uncontrolled diabetes on glucose transport across the blood–brain barrier (BBB) was also investigated as inconsistent findings during chronic hyperglycemia have been reported (Cornford et al. 1995; Lapidot and Haber 2001; Akman et al. 2010).

Methods

Animals

All study procedures were in compliance with the guidelines for the care and use of laboratory animals at the University of Kansas Medical Center and approved by the Institutional Animal Care and Use Committee. Nineteen male Sprague–Dawley rats (303 ± 34 g) (Harlan, Indianapolis, IN, USA) were used for the study. Diabetes was induced by an intraperitoneal injection of STZ (65 mg/kg; Sigma, St Louis, MO, USA) in 10 mM sodium citrate buffer, pH 4.5 (Loganathan et al. 2006). Diabetes was confirmed by measuring non-fasting blood glucose levels (≥ 11.1 mM) 3 days after the STZ injection using a glucometer based on a glucose oxidase method (Accu-Check Active system, Roche Diagnostic, Switzerland) and a rapid multi-assay analyzer (GM7 MicroStat; Analox Instruments, Lunenburg, MA, USA).

For longitudinal MR scans, the rats were anesthetized by 1–2% isoflurane in air and oxygen (1 : 1) delivered through a nose cone. Two ear bars and a bite bar were used for three-point stereotactic fixation of the animal. The animal body temperature was maintained at 37 ± 0.5°C using a rectal temperature probe and a temperature controller (Cole-Palmer, Vernon Hills, IL, USA) and a blanket with warm water circulation. The respiratory rates of the animals were also monitored using a pneumatic pillow sensor (SA Instruments, Stony Brook, NY, USA).

1H MRS and quantification of metabolites

All experiments were performed on a 9.4 T MR system composed of gradient coils (120 mm i.d., maximum gradient strength = 0.4 T/m) with second order shim coils (maximum shim strength = 0.4 T/m2) (Magnex Sceintific, Abingdon, UK) and a Varian INOVA console (Varian Inc., Palo Alto, CA, USA). A quadrature radiofrequency (RF) surface coil with two geometrically decoupled loops (diameter = 18 mm) was used to transmit/receive at 400 MHz. Gradient-echo images (TE/TR = 2.8/110 ms) were acquired to determine a voxel position of 6 × 3×5 mm3 in the cortex and hippocampus region. The water signal was efficiently suppressed using variable power RF pulses with optimized relaxation delays (VAPOR) technique (Tkac et al. 1999). The voxel was localized using ultra-short echo-time stimulated echo acquisition mode (STEAM) sequence (TE/TM/TR = 2/20/5000 ms) with asymmetric RF pulses combined with outer volume suppression (Tkac et al. 1999). First- and second-order shim currents were adjusted for the volume of interest using the fast automatic shimming technique by mapping along projections (FASTMAP; Gruetter and Tkac 2000), resulting in line widths of the unsuppressed water signal in the range of 13–18 Hz.

Baseline MR scans were performed before the STZ injection, thus each animal served as its own control (CTL). Subsequent scans were performed on the third day after STZ-injection or 1 day after the development of hyperglycemia (DM1) followed by five additional scans on every other week thereafter for up to 10 weeks (2 weeks: DM15; 4 weeks: DM29; 6 weeks: DM43; 8 weeks: DM57; and 10 weeks: DM71). The animal body weight and blood glucose levels were measured prior to each MR scan. MRS data were acquired as 20 blocks of free induction decays (FIDs), each consisting of eight averages. Each FID block was corrected for any frequency shift and all FID blocks were averaged and corrected for residual eddy current effects using the unsuppressed reference water signal before quantification.

Metabolite concentrations were quantified using linear combination of model spectra of metabolite solutions in vitro (LCModel) software (Provencher 1993), in which the in vivo spectra were fitted by superposition of a set of in vitro basis spectra by using a constrained regularization algorithm. The unsuppressed water signal measured from the same voxel was used as an internal concentration reference. A total of 21 neurochemicals were quantified from each 1H MR spectrum: alanine (Ala), aspartate (Asp), ascorbate), bHB, Cho, Cr, γ-aminobutyric acid (GABA), Glc, glutamate (Glu), glutamine (Gln), GSH, GPC, phosphorylcholine, Ins, lactate (Lac), macromolecule, NAA, NAAG, PCr, phosphorylethanolamine, serine (Ser), and taurine (Tau). Cramér–Rao lower bound (CRLB), an error estimate of metabolite concentrations from LCModel, less than 20% was used for all the metabolites except Ala, bHB, and Ser, whose concentrations were quantified with CRLB ≤ 30%.

Determination of glucose transport kinetics

Glucose transport kinetics experiments were performed on five diabetic rats about 2 weeks after the completion of DM71 MR measurements (i.e. DM82–DM87). Two catheters were inserted into tail veins: one for intravenous infusion of glucose and the other for insulin. Another catheter was inserted into the tail artery for taking blood samples for blood gas (pO2, pCO2) and glucose analyses. Then animals were anesthetized with isoflurane as in the longitudinal study, intubated and ventilated with a pressure-driven ventilator (Ugo Basile North America Inc., Collegeville, PA). The animal body temperature and respiratory rates were monitored, and its end-tidal CO2 was monitored with a capnometer (Datex Instrumentarium, Helsinki, Finland). During MR experiments, glucose and insulin were concurrently infused via tail veins (Pelligrino et al. 1990). The infusion rates of the d-glucose (32% w/v) and insulin (2 U/mL) solution were adjusted to reach the target blood glucose levels starting from hyperglycemia (> 28.5 mM) to euglycemia (4.2–9.2 mM) and mild hypoglycemia (< 4.2 mM) according to the previous study (Choi et al. 2001). Blood samples were taken from the tail artery to measure the glucose levels of the whole blood using the glucometer. 1H MRS data were acquired after each target glucose level was stable for over 20 min. The physiological parameters were maintained within the normal range throughout the experiments.

Glucose transport kinetic parameters across the BBB were determined using the reversible Michaelis–Menten model (Gruetter et al. 1998). The reversible model is based on the three compartmental model: the blood circulation compartment, the BBB assuming BBB functions as a single layer membrane, and the brain aqueous phase that is separated from metabolic pool where glucose is consumed. At the metabolic steady state, this model suggests a linear relation between brain glucose and plasma glucose concentrations described by the following equation (Gruetter et al. 1998; Choi et al. 2001):

image(1)

In this equation, Gbrain denotes the brain glucose concentration (in μmol/g), Gplasma plasma glucose concentration (in mM), Tmax the apparent maximal transport rate (in μmol/g/min), Kt the apparent Michaelis–Menten kinetic constant (in mM), CMRglc the cerebral metabolic rate of glucose consumption (in μmol/g/min) and Vd the volume of the physical distribution space of glucose in the brain (0.77 mL/g) (Lund-Andersen 1979; Gjedde and Diemer 1983).The values of the kinetic parameters were estimated using a computer program written in R software (R Development Core Team, http://www.r-project.org). Confidence intervals of the parameters were determined using a non-parametric bootstrapping method that randomly samples the same total number of data points as the original data set with replacement and repeating the analysis 10 000 times.

Statistical analysis

The data were first summarized using descriptive statistics. Mean and standard deviation were calculated for concentrations of brain glucose and other metabolites before and after diabetes induction. For blood glucose levels, the Kaplan–Meier estimates were used and the medians were reported because a number of rats showed high glucose levels beyond the detection limit (600 mg/dL or 33.3 mM) of the glucometer used to measure routine blood glucose levels (Acu-Check Active), resulting in censored observations.

To assess the effect of uncontrolled hyperglycemia on neurochemical levels over time, we applied the mixed-effects anova models with first-order autoregressive correlation structure for repeated measures. Each DM condition was compared to the baseline and the type I error rate was controlled by the Dunnett’s procedure. The Benjamini–Yekuteli’s procedure was applied to control the 0.05 false discovery rate among the 21 metabolites. Separate mixed models were fitted to the data of baseline, hyperglycemia, and euglycemia following glycemic normalization, and pair-wise comparisons were performed. All data are presented as mean ± standard deviation obtained from a given number (n) of animals. Differences were considered significant at p-values < 0.05.

Results

Rats injected with STZ developed hyperglycemia on the third day after the injection, as expected for this model (Loganathan et al. 2007; Lenzen 2008) and maintained plasma glucose level at over 28 mM for the duration of the experiment, until week 10 (DM71). There was no significant change in the animal body weight while the variations were up to -8%. Neurochemical profiles of 21 metabolites were measured longitudinally from the cerebral cortex and hippocampus region of STZ-induced diabetic rats. The rats were evaluated at the onset of hyperglycemia (DM1) and during progression (every 2 weeks following DM1). Thirteen out of 21 metabolites showed distinctive patterns of alterations through the acute and chronic phases of the disease. In particular, the levels of brain glucose, ketone bodies and neurochemicals related to osmoregulation were significantly altered as soon as animals developed hyperglycemia (DM1) and the alterations were sustained until the last scan time point, up to 10 weeks of diabetes. About 4 weeks after the development of hyperglycemia, significant changes in the levels of a neurotransmitter (i.e. Asp), a major antioxidant (i.e. GSH) and a neuronal marker (i.e. NAA) were observed. While chronic uncontrolled diabetes of up to 10 weeks led to sustained alterations of a few key neurochemicals such as Ala, Ins, NAA and Gln, the glucose transport across the BBB was not altered.

Figure 1 shows the MR spectra obtained on a rat before (CTL) and after STZ-induced diabetes (DM1, DM15, and DM57). Excellent water suppression allowed the detection of a distinctive brain glucose signal at 5.23 ppm, showing clear increases in brain glucose levels on DM1, DM15 and DM57 compared with CTL. Brain and blood glucose levels before (CTL) and after STZ-induced diabetes (DM) are listed in Table 1. Changes of glucose levels in the brain after the development of hyperglycemia showed a similar trend to those in blood. In acute hyperglycemia (DM1), brain glucose levels showed a sharp increase, coinciding with the rapid increase of blood glucose levels. In chronic hyperglycemia (DM15 and later), glucose levels remained relatively stable at the elevated levels at over 160% to up to 180% in the brain, and at over 300% to up to about 360% in blood (Table 1).

Figure 1.

 Representative in vivo1H MR spectra acquired from the rat brain before (CTL) and after the development of hyperglycemia via STZ injection at the following time points: DM1, DM15, and DM57. The glucose signal at 5.23 ppm clearly shows increases of brain glucose levels with diabetes progression. The inset represents a typical location of the volume of interest (6 × 3 × 5 mm3) in the cerebral cortex and hippocampus region.

Table 1.   Rat brain and blood glucose levels before and after STZ injection
 CTLDM1DM15DM29DM43DM57DM71
  1. CI, confidence interval; NA, not available.

Brain Glc (μmol/g)3.2 ± 0.86.3 ± 1.068.2 ± 2.48.3 ± 1.98.3 ± 1.478.9 ± 2.018.3 ± 1.8
Change (%)97160164163180162
Blood Glc (mM)7.325.232.2> 33.333.130.929.5
Change (%)255345> 357356333311
95% CI(6.5, 7.8)(23.1, 27.2)(27.3, NA)(29.6, NA)(31.2, NA)(26.9, NA)(27.6, NA)
No. censored data0157665

Figure 2 shows changes of the metabolite concentrations over 10 weeks after the development of hyperglycemia monitored in the same rats before (CTL) and at different time points after STZ injection (n = 13). The CRLB values for PCr, Gln, Glu, Ins, NAA and Tau were below 5% and those in other metabolites were below 20% except those with very low concentrations such as Ala, bHB and Ser. The CRLB values (average ± standard deviation) for Ala, bHB and Ser at seven measurement points, that is, from CTL to DM73, were (Ala: 12 ± 4%, 13 ± 5%, 14 ± 3%, 15 ± 6%, 16 ± 6%, 20 ± 7%, 17 ± 6%), (bHB: 26 ± 4%, 20 ± 6%, 18 ± 7%, 18 ± 9%, 14 ± 6%, 12 ± 4%, 17 ± 8%), and (Ser: 15 ± 4%, 17 ± 5%, 18 ± 5%, 17 ± 6%, 20 ± 6%, 20 ± 6%, 20 ± 5%), respectively.

Figure 2.

 Concentrations of brain metabolites in the cerebral cortex and hippocampus region of diabetic rats before (CTL) and after the development of hyperglycemia (DM1, DM15, DM29, DM43, DM57, DM71), where *, **, and *** denote p < 0.05, p < 0.01, and p < 0.001 adjusted by the Dunnett’s procedure (n = 13), in comparison with CTL, respectively.

Three distinct patterns of the time course of metabolite levels were identified: initial increase at DM1 and plateau at DM15 (pattern 1), initial decrease at DM1 and plateau at DM15 (pattern 2), and initial no change and late decrease at DM29 (pattern 3). Metabolites with the pattern 1 included bHB (+106%, p = 0.03 on DM15), GPC (+33%, p < 0.001 on DM1), Ins (+7%, p = 0.02 on DM1), Tau (+8%, p = 0.01 on DM1), and Gln (+10%, p = 0.01 on DM1). The metabolite with the pattern 2 included Lac (−11%, p = 0.002 on DM15). The metabolites with the pattern 3 included Ala (−22%, p = 0.01 on DM43), Asp (−20%, p = 0.003 on DM29), GSH (−18%, p = 0.02 on DM29), NAA (−6%, p = 0.001 on DM43), and NAA + NAAG (−6%, p = 0.03 on DM29).

Among the metabolites with significantly elevated levels, the levels of bHB, GPC, Ins, and Tau showed rapid increases in acute hyperglycemia (DM1), reached a plateau on DM15 and maintained at the elevated levels thereafter. For example, bHB concentrations showed an increase of 100% from DM15 to 201% on DM57 (p < 0.001). Upon the development of hyperglycemia, the levels of Lac were significantly decreased by over 19% (p < 0.002) on and after DM15 and up to 42% (p < 0.001) on DM57. The concentrations of Gln showed transient increases on DM1 and DM29 and returned to the CTL levels thereafter. The Glu concentrations were overall maintained at a similar level with a trend of decreases and the decreases were statistically significant only on DM57 (p = 0.03). Nine (ascorbate, Cr, PCr, phosphorylcholine, GABA, macromolecule, NAAG, phosphorylethanolamine and Ser) of 21 measured metabolites did not show any significant changes during STZ-induced hyperglycemia.

Figure 3 shows quantitative comparison of the metabolite levels measured on CTL, DM71, and about 2 weeks after DM71 with glycemic normalization. When the glycemic levels were normalized to euglycemia on DM82–87 via insulin administration, the concentrations of brain Glc, bHB, Lac, GSH, Asp, GPC and Tau were restored back to their CTL levels. The brain glucose signal at 5.23 ppm measured after glycemic normalization was also similar to those from the CTL (not shown). However, the levels of Ala, Ins and NAA were not restored back to their CTL levels. The level of Ala was further decreased after glycemic normalization (p = 0.002), from 19% (p = 0.007) reduction on DM71 to 39% (p = 0.003) reduction after glycemic normalization compared with the CTL level. The decreased level of NAA on DM71 was not influenced by the glycemic normalization and remained lower than that on CTL (−9.6%, p = 0.03). While Ins on DM71 showed a significant increase of 18% compared with that on CTL (p < 0.001), the level of Ins after glycemic normalization was 9% higher than that of CTL (p = 0.003), showing a trend of restoring to the CTL level. While the levels of Cr, PCr and Gln at DM71 were not significantly different from those of CTL, the glycemic normalization led to significant increases of PCr by 18% (p < 0.001) and Gln by 61% (p < 0.001), and a decrease of Cr by 18% (p = 0.004) as compared with those of CTL.

Figure 3.

 Quantitative comparison of the metabolite levels obtained from the rat brain on CTL (n = 13), on DM71 (DM71-hyperglycemia; n = 13), and after glycemic normalization (DM82-87-euglycemia; n = 5). * (#), ** (##), and *** (###) denote p < 0.05, p < 0.01, and p < 0.001, in comparison with CTL, respectively.

To characterize the patterns of neurochemical level changes over disease progression of diabetes, the levels of two most significantly altered neurochemicals, NAA and Ins, were plotted against brain Glc levels (Fig. 4). All three plots, NAA/Ins versus Glc, Ins versus Glc and NAA versus Glc, showed clear separation of the CTL conditions from the DM conditions (DM1, DM29, and DM71). The plot of the ratio (NAA/Ins) versus Glc showed further separation between acute (DM1) and chronic (DM29 and DM71) stages in hyperglycemia. In the chronic stage, neurochemical levels remained largely the same.

Figure 4.

 Diabetes staging through plots of NAA/Ins, Ins and NAA versus brain Glc separating DM data points from CTL. The levels of NAA and Ins, and their ratio, and brain glucose acquired from the cortex and hippocampus region characterize three separate stages in the disease progression: normal (CTL), acute (DM1) and chronic (DM29 and DM71) conditions.

The relationship between brain and plasma glucose levels obtained from glucose transport kinetics experiment is shown in Fig. 5. The glucose levels measured after DM71 showed an apparent linear relationship over a wide range of plasma glucose concentration up to 47 mM, which is consistent with the reversible Michaelis–Menten model for glucose transport across the BBB. The best fit to the data resulted in apparent Kt of 0.73 mM (0.00–3.47 mM, 95% confidence interval) and Tmax/CMRglc of 1.94 (1.76–2.11) with the reversible model (Table 2). In comparison with the kinetic parameters from the control rats in the previous work (Duarte et al. 2009) that had very close experimental design to the current work except for the choice of anesthesia, that is, isoflurane and α-chloralose in the work by Duarte et al. and isoflurane alone in this study, no significant differences were found in the Tmax/CRMglc (t70 = 1.00, p = 0.32). Direct comparison in Kt was not calculated because negative values of Kt were discarded during the confidence interval estimation, resulting in skewed parameter distributions. However, the 95% confidence intervals were nearly completely overlapping, suggesting the difference of Kt between the current and previous works is not likely significant.

Figure 5.

 The linear relationship between brain glucose and plasma glucose concentrations in the chronic stage of the STZ-induced diabetes (DM71, n = 5) measured using 1H MRS when the plasma glucose level was at the steady state. The solid line shows the best fit of the reversible Michaelis–Menten model of glucose transport (eqn 1). The kinetic parameters of glucose transport were estimated using the reversible Michaelis–Menten model and listed in Table 2.

Table 2.   Glucose transport kinetic constants calculated using the reversible Michaelis–Menten glucose transport model
 Reversible Michaelis–Menten model
Kt (mM)Tmax/CMRglc
  1. The data from Duarte et al. (2009) were measured from male Sprague–Dawley rats under α-chloralose anesthesia.

DM-STZ (current)0.73 (0.00–3.47)1.94 (1.76–2.11)
Control (Duarte et al. 2009) 1.23 (0.00–3.79)1.77 (1.48–2.07)

Discussion

This study demonstrates the effect of uncontrolled hyperglycemia on the CNS during the disease progression in an animal model of STZ-induced diabetes. Using highly resolved in vivo1H MRS at 9.4 T, levels of over 20 neurochemicals were quantified from the living intact rat brain before and during acute and chronic phases of STZ-induced diabetes in a longitudinal manner. We were able to monitor progressive alterations of a number of key neurochemicals directly related to the disease processes, providing a reference to the severity of diabetes-induced changes to the CNS and disease staging.

The ability to measure the brain glucose levels in the living brain tissue has been identified as a unique contribution of in vivo1H MRS to the field of neurochemistry. Current study showed that the onset of hyperglycemia in STZ-induced diabetes significantly increased brain Glc levels and altered the concentration of a number of neurochemicals related to osmotic regulation (e.g. GPC, Ins and Tau) and ketone bodies (e.g. bHB) (Heilig et al. 1989). As hyperglycemia persisted over 4 weeks, levels of additional neurochemicals including GSH and NAA were altered, which suggests increased oxidative stress and deterioration of neuronal integrity, respectively. Marked decreases in l-glucose space under chronic hyperglycemia starting at 2 weeks post-STZ injection have been reported in the rat brain (McCall 1992), which was attributed to decreased non-carrier–mediated diffusion of glucose across the BBB (Ergul et al. 2009). Others have shown that mechanical hyperalgesia, associated with peripheral diabetic neuropathy, developed in the STZ-injected rat after 4 weeks only, indicating prolonged exposure to high glucose levels was necessary for the mechanical hyperalgesia phenotype to develop (Malcangio and Tomlinson 1998). Contributing to autonomic nervous system dysfunction in diabetes, mesenteric axonopathy was apparent in rats 6–12 weeks post-STZ injection (Lien et al. 1991).

Increased levels of ketone bodies such as bHB, a characteristic feature of uncontrolled diabetes, were observed from the onset of hyperglycemia to the chronic stage of diabetes indicating ketosis in the diabetic brain in agreement with previous studies (Mans et al. 1988; Lapidot and Haber 2002). The increased ketone bodies in the diabetic liver are known to be exported to other organs, possibly including the brain. After being taken up into the brain via monocarboxylic acid transport, bHB could be utilized in the healthy brain as an alternative fuel mainly in the glia. However, in diabetes, where brain bHB levels are increased such as in this study, ketone bodies can provide only modest amount of fuel (Hawkins et al. 1986). Reliable quantification and longitudinal monitoring of bHB in the diabetic brain using 1H MRS may allow us to study the presence and effect of ketosis, and resulting complications including cerebral edema in a non-invasive manner. For example, elevated bHB is known to decrease the release of Ala in muscles and thus decrease the plasma Ala level, probably leading to a decreased level of Ala in the brain as shown in the current study. Furthermore, our finding on concurrent decreases in Lac levels could be explained by the inhibitory effect of elevated bHB on glycolysis (Robinson and Williamson 1980). While the significantly decreased Lac levels observed in the present study agree with a study using a rabbit model (Lapidot and Haber 2002), other studies using rat models reported variable Lac levels ranging from significant increases (Salceda et al. 1998) to insignificant changes (Mans et al. 1988; van der Graaf et al. 2004; Duarte et al. 2009).

Hyperglycemia leads to osmolar gradients across cell membrane, triggering alterations in cell volume regulation that shifts water from the intracellular fluid space to the extracellular fluid space and subsequently results in contraction of the cell volume. Adjustment of cell osmolarity can be accomplished by accumulating or releasing organic osmolytes, which are molecules that can adjust intracellular osmolarity without interfering with cellular functions. Therefore, the osmolar gradients due to the elevated glucose levels in hyperglycemia may lead to increased levels of the intracellular osmolytes including GPC, Ins and Tau to promote the maintenance of neuronal and glial cell volume, as observed in the current study. A previous work showed that increases in the levels of the intracellular osmolytes can be resulted from the osmolar gradients induced by chronic hyponatremia (Soupart et al. 2002).

When the changes of osmolyte levels reached a plateau on DM29 in this study, significant decreases in NAA and GSH levels occurred concurrently. In addition, glycemic normalization via insulin administration after 10 weeks of diabetes was able to restore the GSH level but not that of NAA, indicating potential irreversible neuronal damage due to prolonged hyperglycemia. These changes could be due to compromised osmolar regulation leading to impaired cell volume regulation. This notion is supported by the hypothesis that sustained hyperglycemia results in defective cell volume regulation and in turn leads to neural dysfunction (Hansen 2001).

When the plasma glucose levels were restored to the euglycemic level, concentrations of most neurochemicals including bHB, GPC, Glc, Tau, Asp, GSH and Lac were recovered to their baseline levels indicating that changes in the levels of those neurochemicals were reversible. Unlike other osmolytes (GPC and Tau), Ins levels remained elevated after the glycemic normalization, possibly due to slow efflux of Ins from the brain cells. This notions is supported by the previous findings of a slow reduction of brain organic osmolyte levels, particularly Ins, following their accumulation in the brain after a regulatory cell volume increase in animals and cultured glial cells (Strange 1992). Further evidence comes from a study showing an extremely slow decrease of accumulated Ins in the brain of human infants undergoing rehydration therapy (Lee et al. 1994). The elevated Ins levels in glycemic normalization in the current study is also consistent with elevated Ins levels in diabetic patients with relatively good glycemic controls (Kreis and Ross 1992). A significant increase of PCr and the concomitant decrease of Cr with no changes of tCr suggest decreased energy utilization and increased energy buffers following the glycemic normalization.

Our data of the steady-state glucose transport following 10 weeks of uncontrolled hyperglycemia showed no significant alterations of glucose transport kinetic parameters compared with those in control animals. This result is consistent with previous work in both experimental diabetes in animals (Jacob et al. 2002; Duarte et al. 2009) and poorly controlled diabetes in humans (McCall 2004; Seaquist et al. 2005) in that chronic hyperglycemia did not alter glucose transport across the BBB.

The alteration patterns of neurochemical levels due to STZ-induced hyperglycemia in the current study are in general consistent with a previous study with a similar study design (Duarte et al. 2009). Yet, the current study demonstrates major differences from the previous report in alterations of additional metabolites such as Ala, Asp, and Lac at the similar duration of hyperglycemia. Another significant finding is the effect glycemic normalization on the levels of Ala, NAA, NAAG, and NAA + NAAG. Although it is unclear, these differences could be due to the differences of the sample size (n = 13 vs. n = 6), study design (longitudinal vs. cross-sectional), brain region (cortex and hippocampus vs. hippocampus), and study duration (10 weeks vs. 4 weeks) in the current study in comparison with the previous study by Duarte and colleagues.

In conclusion, the study results provide insights into the effect of uncontrolled hyperglycemia resulting in dynamic changes of neurochemical levels in the rat brain in acute and chronic hyperglycemia in STZ-induced diabetes using high resolution in vivo1H MRS. The hyperglycemia accompanying diabetes led to significant, rapid (after 1 day of hyperglycemia) changes in the levels of osmolytes from the onset to the acute stage followed by increased oxidative stress and neuronal dysfunction in the chronic stage. Furthermore, the disease progression could be assessed through altered levels of neurochemicals altered throughout the time course of hyperglycemia. An example of NAA and Ins plots versus Glc demonstrate the reliable separation between CTL and DM groups indicating the usefulness of these neurochemical measures in assessing the disease progression in both pre-clinical and clinical studies. Moreover, this non-invasive technique allowed monitoring changes of metabolites in diabetic brain practically as soon as hyperglycemia develops, when any clinical and subclinical features of diabetes are not evident.

Acknowledgements

We are grateful to Rajprasad Loganathan for his help with the STZ injection protocol. This study was supported by the NIH grant (R21 DK081079 to Dr. Choi). The HBIC is partly supported by NIH (C76 HF00201 and P30 HD002528) and the Hoglund Family Foundation. The authors have no conflict of interest for this study.

Ancillary