Cortical Metabolites as Biomarkers in the R6/2 Model of Huntington's Disease

Abstract

To improve the ability to move from preclinical trials in mouse models of Huntington's disease (HD) to clinical trials in humans, biomarkers are needed that can track similar aspects of disease progression across species. Brain metabolites, detectable by magnetic resonance spectroscopy (MRS), have been suggested as potential biomarkers in HD. In this study, the R6/2 transgenic mouse model of HD was used to investigate the relative sensitivity of the metabolite profiling and the brain volumetry to anticipate the disease progression. Magnetic resonance imaging (MRI) and ¹H MRS data were acquired at 9.4 T from the R6/2 mice and wild-type littermates at 4, 8, 12, and 15 weeks. Brain shrinkage was detectable in striatum, cortex, thalamus, and hypothalamus by 12 weeks. Metabolite changes in cortex paralleled and sometimes preceded those in striatum. The entire set of metabolite changes was compressed into principal components (PCs) using Partial Least Squares-Discriminant Analysis (PLS-DA) to increase the sensitivity for monitoring disease progression. In comparing the efficacy of volume and metabolite measurements, the cortical PC1 emerged as the most sensitive single biomarker, distinguishing R6/2 mice from littermates at all time points. Thus, neurochemical changes precede volume shrinkage and become potential biomarkers for HD mouse models.

Keywords

biomarker metabolite profiles metabolomics principal-least square discriminant analysis

Introduction

Biomarkers for Huntington's disease (HD) are actively being sought for following progression from prodromal periods through manifest disease (Paulsen et al, 2008). Genomic, proteomic, and metabolomic panels as well as individual proteins or metabolites from peripheral tissues and fluids have been proposed as biomarkers (Sturrock et al, 2010; Zuccato et al, 2010). Behavioral and neuropsychological tests and accompanying functional imaging, while applicable to humans, have few good parallels in mice (Ferrante, 2009). For brain, the tissue most affected in HD, imaging of volumetric changes provides the single best direct measure to date of progression in both humans and mice (Rizk-Jackson et al, 2010). Brain shrinkage, however, provides no data on the mechanisms mediating that loss. However, region-specific neurochemical information provided by magnetic resonance spectroscopy (MRS) can be related to underlying mechanisms on a molecular level and may reveal processes, which differ by disease stage. At high magnetic fields, ¹H MRS can reliably quantify up to 17 brain metabolites in mouse models of neurological disease (Tkac et al, 2007; Oz et al, 2010b). A different number of metabolites (from 3 to 17) can be reliably quantified in human brain based on the magnetic field strength (1.5 to 7 T), pulse sequence and data processing (Jenkins et al, 1998; Sturrock et al, 2010; Oz et al, 2010a). As a noninvasive measurement in both species, MRS has the potential to provide a scale for aligning progression in mouse models with the human disease.

A useful biomarker will be able to uncover early signs of disease onset before overt neurological dysfunction and to change quantitatively as disease unfolds. This study had two goals: to determine (1) if neurochemical changes in cortex and striatum are detectable before brain shrinkage, currently one of the most sensitive indicators of disease and (2) if brain metabolite information could be condensed into an one or two dimensional biomarker that could be used to track disease progression. We studied the R6/2 transgenic mouse, a severe and rapid model of HD that expresses a toxic fragment of exon 1 of the human HD gene with ∼150 CAG repeats (Mangiarini et al, 1996). In this mouse model, abnormalities in running wheel and climbing behavior parallel the appearance of nuclear inclusions as early as 4 to 4.5 weeks (Davies et al, 1997; Hickey et al, 2005). In our previous MRS study of R6/2 mice, we reported only striatal neurochemical changes at 8 and 12 weeks (Tkac et al, 2007). We now extend that finding to cover the whole lifespan (4 to 15 weeks) including both early and late stage disease and propose a method for combining the concentration measurements into a computed biomarker of disease progression using Partial Least Squares-Discriminant Analysis (PLS-DA). In addition, we use high-resolution MRI to assess volumetric changes in different brain regions of R6/2 mice.

Materials and methods

R6/2 Mice

All experiments were preapproved by the University of Minnesota Institutional Animal Care and Use Committee. Timed pregnant female C57B6/CBA mice previously mated with R6/2 males were obtained from the CHDI colony at Jackson Laboratories (Barr Harbor, ME, USA). R6/2 offspring contained 120 to 142 CAG repeats. Mice were housed in mixed genotype groups, 2 to 4 per cage, in enriched conditions on a 12:12 light dark cycle. Mice received daily fresh food mash in addition to standard pellet food, Purina 5001 (Menalled et al, 2009). Eight R6/2 and wild-type littermate mice (WT or LM) were scanned and imaged at 4, 8, 12, and 15 weeks. Behavioral assessments were performed 3 to 4 days after the magnet measurements. An additional cohort of 8 mice per genotype was raised as backups and substituted as needed if magnet times had to be rescheduled.

Genotyping was performed using Qiagen HotStar Taq Plus DNA Polymerase Kit (Valencia, CA, USA, Cat. no. 203605) according to established protocols (Mangiarini et al, 1996). Genotyping and CAG repeat number were also determined by Laragen (Los Angeles, CA, USA).

Magnetic Resonance Spectroscopy

All experiments were performed using a horizontal 9.4 T/31 cm magnet (Varian/Magnex, Oxford, UK) equipped with a 15-cm internal diameter gradient coil insert (450 mT/m, 200 μs), which included a powerful second-order shim set (Resonance Research, Inc., Billerica, MA, USA). The magnet was interfaced to a Varian INOVA console (Varian, Inc., Palo Alto, CA, USA). The protocol for collecting MR spectra was identical to that previously reported (Tkac et al, 2007). Spontaneously breathing mice were anesthetized using 1.5% isoflourane in a gas mixture (O₂:N₂O=1:1) for up to 2 hours per session. Mice were fixed in a cylindrical chamber and their temperature was maintained by temperature controlled circulating water. Respiration was continuously monitored using SAM PC (Small Animal Instruments, Inc., Stony Brook, NY, USA). Precise position of the volume of interest (VOI) was based on multislice fast spin echo images in coronal and sagittal planes. The size and location of the VOI was adjusted to avoid the ventricles. The MRS data were acquired from VOIs centered in left striatum (VOI=4.5 to 5.3 μL) and bilateral dorsal cerebral cortex (VOI=4.1 to 5.5 μL) using ultra-short echo-time STEAM sequence (echo time=2 ms, repetition time=5 seconds) combined with VAPOR water suppression. Broadband radio frequency pulses of the STEAM sequence (13.5 kHz) minimized the chemical shift displacement error <10% along each axis for the 3 p.p.m. chemical shift range. The homogeneity of the magnetic field was adjusted automatically using FASTMAP. Metabolites were quantified using LCModel with the macromolecule spectrum included in the basis set. The unsuppressed water signal was used as an internal concentration reference.

Brain water content was measured from 16-week-old mice as a difference between the initial brain wet weight and the dry weight after desiccation at 50°C for 7 days. Since no difference in brain water content between R6/2 and littermate mice was found (littermates 78.1±0.4%, n=10; R6/2 78.2±0.6%, n=8; P=0.9), a water content value of 78.1% was used for metabolite quantification. The concentrations of the following 17 metabolites were quantified: alanine (Ala), ascorbate (Asc), aspartate (Asp), creatine (Cr), phosphocreatine (PCr), γ-aminobutyric acid, glucose (Glc), glutamine (Gln), glutamate (Glu), glutathione (GSH), myo-inositol (myo-Ins), lactate (Lac), N-acetylaspartate (NAA), N-acetylaspartylglutamate, phosphoethanolamine (PE), taurine (Tau), glycerophosphocholine+phosphocholine (GPC+PC). Total creatine (Cr+PCr) and the macromolecule content were also reported.

Magnetic Resonance Imaging

For the volumetric measurements, coronal MR images were acquired using high-resolution fast spin echo imaging sequence (echo train length=16, echo spacing=8 ms, echo time=64 ms, repetition time=7 seconds, thk=0.3 mm, in-plane resolution=100 μm × 100 μm, number of slices=24, coronal orientation, number of averages=16, total measuring time 22 minutes) were obtained from the rostral pole of the frontal cortex to the caudal pole of the occipital cortex. Images were evaluated by user guided segmentation (Amira, Visage Imaging, San Diego, CA, USA) to determine regional brain volume changes in whole brain, cortex, striatum, lateral ventricles, and third ventricle. The boundaries of the cortex were defined ventrally by the corpus callosum and lateral ventricles. The striatum was defined by the corpus callosum dorsally and laterally, by the lateral ventricles medially and by the anterior commissure ventrally. Anterior brain measurements indicating overall brain size were summed from all structures in all image planes obtained, including ventricles.

Behavior and Physiology

During daylight hours, mice were placed in an upside-down wire mesh pencil holder for 5 minutes and videotaped (Hickey et al, 2005). At least two independent, blinded observers scored the videos for total time, latency and number of instances of climbing and rearing behaviors. Climbing was defined as all four paws off of the ground and on the wire mesh. Rearing was defined as 2 to 3 paws off of the ground, not necessarily on the wire mesh. Clasping behavior was also measured. Behavioral analysis was performed on both cohorts of mice. Periodic blood samples were taken by submandibular puncture. Plasma glucose measurements were made using an Aviva Accu-Chek glucometer (Roche Diagnostics, Indianapolis, IN, USA).

Statistics

Statistical analyses comparing R6/2 with littermates used two-way analyses of variance (ANOVAs) followed by Bonferroni post tests (GraphPad Prism, v. 4.0, GraphPad Software, La Jolla, CA, USA). For the metabolites, similar results were obtained by using two-tailed t-tests followed by False Discovery Rate analysis (not shown). Metabolite concentrations were compressed into two principal component (PC) variables using PLS-DA in the pls statistical package in R (Frank and Friedman, 1993). Matlab was used to plot Z scores (Z=(X−μ)/(σ/√n)) comparing mean metabolite levels in R6/2 (X) with mean metabolite levels of littermate controls (μ, σ, n). Two-tailed t-tests followed by False Discovery Rate analysis were applied to compare the efficacy of the various computed biomarkers.

Results

Behavior

Since motor behavior has been proposed as a sensitive biomarker in HD mice (Menalled et al, 2009), climbing behavior was assayed for assessment of the degree of functional impairment accompanying metabolite changes. This assay was chosen based on its simplicity and early ability to separate gene-positive mice phenotypically (Hickey et al, 2005). Two measures from the climbing assay successfully distinguished R6/2 mice from controls at the youngest ages but not at later disease stages. At 4 weeks, R6/2 mice took longer to initiate climbing and did not climb as frequently as the controls (Figures 1A and 1B). For both genotypes, the latency to climb continued to increase until all mice stopped climbing entirely at 15 weeks. In parallel, instances of climbing continued to decline with age. Despite the differences at young ages, the total time R6/2s spent climbing was never significantly different from controls (Figure 1C).

Figure 1

R6/2 (filled symbols) and littermate (open symbols) climbing behavior. (A) Variation in the latency to climb was attributable to both genotype and age, P<0.0001. (B) Variation in the number of climbing episodes was attributable to genotype and age, P=0.0024 and P<0.0001, respectively. (C) Variation in total time spent climbing was only attributable to age, P=0.0154. (D) Latency to rear varied by age, P=0.0156. (E) Variation in instances of rearing was due to both genotype and age, P<0.0001. (F) Variation in total time spent rearing was due to both genotype and age, P=0.0033. Data are mean±s.e.m. of 22, 17, 10, 6 R6/2 and 22, 20, 14, and 10 littermate mice at 4, 8, 12, and 15 weeks, respectively. In all figures, ∗P<0.05, ∗∗P<0.01, ∗∗∗P<0.001 in Bonferonni post tests after two-way analyses of variance (ANOVAs).

Rearing behavior distinguished the genotypes only at later stages of the disease (Figures 1D—F). Rearing in the R6/2 was normal at 4 weeks, in all measurements. Starting at 8 weeks and persisting throughout the study, R6/2 mice reared less often than their littermate controls. Total time spent rearing did not decline until 12 weeks. The latency to rear did not distinguish between the genotypes.

Additionally, common physiological measures were obtained throughout the study to gauge well being and overall health of the mice. Clasping and weight loss were evident at 12 weeks and older in R6/2 mice. The mean life expectancy for this R6/2 colony was 13.6 weeks. Blood glucose levels were measured periodically (6, 10 to 12 and 16 weeks). No evidence of diabetes was observed (R6/2 versus littermates: 0.13<P values<0.94, two-tailed Mann—Whitney test) despite a greater variation in R6/2s.

Brain Volume

Since loss of striatal volume is the most well-documented measure of prodromal changes in HD (Paulsen et al, 2008; Hobbs et al, 2010), MRIs were acquired in conjunction with MRS to document changes in mouse brain size (Figure 2; Supplementary Table 1). Between 4 and 8 weeks, increases in anterior brain, cortical, hypothalamic, and striatal volumes were observed in littermates but not in R6/2s. Loss of striatal, cortical, thalamic, hypothalamic, and overall anterior brain volume occurred in the R6/2s at 12 and 15 weeks while littermate volumes stabilized at the 8-week level. In striatum at 8 weeks, littermate volumes increased, while R6/2 volumes were stable, producing a significant difference. The lateral ventricles were larger initially and remained larger throughout the lifespan. No genotype-specific changes were observed in third ventricle size. Age had an independent effect on volumes of the anterior brain, thalamus, and hypothalamus. When mean striatal, cortical, and anterior brain volumes of R6/2s were normalized to those of the control brains, the rate of volume change in the striatum (−2.6%/week) and thalamus (−2.9%/week) exceeded that of hypothalamus, anterior brain, and cortex (−1.5, −1.6, and −1.7%/week, respectively).

Figure 2

Volumetric analysis for indicated brain regions from MRIs of R6/2 (filled symbols) and littermates (open symbols). For Figures 2 and 4, mean±s.d. of 8, 8, 8, 6 R6/2 or littermate mice at 4, 8, 12, and 15 weeks, respectively. Asterisks over data points indicate significant comparisons between genotypes. Asterisks over lines indicate significant comparisons between ages. All regional two-way analyses of variance (ANOVAs) (details Supplementary Table 1) were significant by genotype except for the third ventricle (P<0.003 accounting for 14% to 45% of the variation). Of the regions with genotype-dependent variation, age-related variations were also significant for striatum, anterior brain, hypothalamus, and thalamus (P<0.022 or better, 7% to 18% of the variation). Interactions between genotype and age were significant for striatum, hypothalamus, and thalamus (P<0.023 or better, 11% to 13% of the variation).

Figure 4

Metabolite changes in striatum and cortex of R6/2 (filled symbols) and wild-type (WT) littermate mice (open symbols). Asterisks over data points indicate the level of significance for the post hoc analysis of variance (ANOVA) comparisons between genotypes (∗P<0.05, ∗∗P<0.01, ∗∗∗P<0.001). Error bars indicate s.d. Significant variations (P<0.02) attributable to age (a), genotype (g), and their interaction (i) are indicated for each regional metabolite where appropriate (ANOVA details, Supplementary Table 2).

Magnetic Resonance Spectroscopy

Because of its ability to detect biochemical changes in vivo, MRS was obtained throughout the lifespan of the R6/2. The high spectral quality (Figure 3) routinely enabled quantification of 15 metabolites from both regions (Figure 4). The partial volume effect was minimized by adjusting the VOI size based on fast spin echo MRI in coronal and sagittal planes, taking into account progressive striatal and cortical atrophy and ventricle enlargement in R6/2 mice. Aspartate was not consistently detectable in both regions as was Asc in striatum, eliminating both from further consideration. These data were further analyzed using two-way ANOVAs to reveal genetic differences and age-related patterns of change for each metabolite (Figure 4; Supplementary Table 2). Metabolite changes in cortex and striatum of the R6/2 mice were largely parallel, with a few exceptions. In both regions, Tau, Gln, GPC+PC, Cr, and Cr+PCr increased and NAA decreased progressively with disease progression. Increases in myo-Ins and PCr were more pronounced and occurred earlier in the striatum than in cortex. Lactate increased sooner and to a greater extent in the cortex. PE decreased only in the striatum while Glu decreased and GSH increased only in the cortex. The earliest metabolite changes occurred in cortex at 4 weeks when NAA decreased significantly and Cr+PCr and Gln begin to move away from control levels. Ala and Glc significantly decreased in striatum at 12 weeks only (not shown). γ-Aminobutyric acid and N-acetylaspartylglutamate levels as well as the macromolecule content remained unchanged in both brain regions (not shown).

Figure 3

Representative ¹H MRS spectra from cortex (A) and striatum (B) of both wild-type (WT) and R6/2 mice. STEAM, echo time=2 ms, repetition time=5 seconds, number of transients=240. Inserts show the typical position and the size of the volume of interest (VOI). Arrows indicate general direction of metabolite change in R6/2 compared with wild-type littermates (LMs).

Comparisons between adjacent time points following the two-way ANOVAs suggested that different mechanisms may underlie changes at different stages of mouse HD. In the control mice, age-related changes in Tau and NAA were observed between 4 and 8 weeks, suggesting a few growth-related processes may still be detectable. In R6/2, only GPC+PC progressively increased significantly at each age with disease progression. Striatal Gln, Tau, myo-Ins and cortical Tau and NAA changed significantly in two of the three intervals hinting that the mechanisms underlying these shifts may also be progressive. But the elevations of Cr and PCr and decline in PE occurred between 4 and 8 weeks without further significant changes suggesting they stem from early disease-related events. In general, changes in striatum and cortex occurred in parallel with the notable exceptions of PCr, Glu, GSH, PE, and Lac.

For those metabolites that differed between genotypes, genotype explained on average 29.8% and 40.6% of the variation in striatum and cortex, respectively (Supplementary Table 2). For these same metabolites, age accounted for on average 17.1% and 13.3% of the variation for these regions. Similarly, interactions between genotype and age accounted for an average of 12.1% and 11.3% of the variation for striatum and cortex.

Partial Least Squares-Discriminant Analysis

Many of the individual metabolites that changed dramatically could be considered as biomarkers for disease progression. However, this disregards the information content from the metabolites that changed subtly, exhibited a wider variation in levels or that changed as a more complicated function of time. To use all of this information more effectively in biomarker construction, the steady-state concentrations were combined into PCs using PLS-DA. Partial Least Squares-Discriminant Analysis weights and combines variables (see loading plots Figures 5A, 5C, and 5E), producing a reduced set of computed variables known as PCs that facilitate separation of the groups. Only the data from R6/2 regions were used to generate the PCs in the PLS-DA model. Then, both R6/2 and littermate data were plotted in the PC space (Figure 5). Three separate PLS-DA analyses were performed, two using data for each region individually (striatum, Figures 5A and 5B; cortex, Figures 5C and 5D) and one combining data from both regions (Figures 5E and 5F).

Figure 5

Partial Least Squares-Discriminant Analysis (PLS-DA) analysis of striatal (A, B) and cortical (C, D) metabolites in separate principal component (PC) spaces or combined in one space (E, F). The loading plots provide the relative weights of each metabolite concentration used in computing PC1 and PC2 (A, C, E). Total creatine was used instead of individual values for Cr and PCr. The plots of PC space (B, D, E) contain the computed metabolite profile for each mouse at each time point. Solid symbols represent R6/2; open symbols represent littermate controls; squares represent cortex; and triangles represent striatum. Gray scale differs depending on age.

Principal component 1 and PC2 variables created from a loading plot generated using only striatal data with age as the independent variable (Figures 5A and 5B) were able to account for 65.1% and 9.4% of the variability, respectively. In the corresponding cortical plot (Figures 5C and 5D), PC1 and PC2 accounted for 61.7% and 8.4% of observed variability, respectively. Disease progression in each region could be tracked in that region's PC space, mostly by PC1. Finally, a loading plot generated from both cortical and striatal R6/2 data with both age and region as independent variables created a PC1 and PC2 set accounting for 63.0% and 10.0% of variability, respectively (Figures 5E and 5F). For the striatum, the 4 weeks R6/2 data clustered together within the range of the controls, as might be expected from the absence of major metabolite changes identified at 4 weeks (Figures 4 and 5B). In the cortex, the 4 weeks R6/2 values fell just on or outside the border of the control cluster (Figure 5D). Clusters of R6/2 data showed less overlap in the cortical PC space than in the striatal PC space. Overlap of data points at older ages suggests greater variability in disease progression with advancing status.

The regional similarities in progression can be realized from the PC space that resulted when data from both regions were combined in a single PLS-DA analysis (Figures 5E and 5F). In the latter analysis, PC2 tended to distinguish the regions. The similarity of the loading plots for PLS-DA from each region confirmed the highly parallel nature of changes in both regions, indicating mechanistically similar processes were at work in both brain areas. In the combined analysis, the regions were distinguished by metabolites that only changed in a single region (Glu, PE) and by consistently different absolute concentrations for these, NAA, Lac, and myo-Ins.

Biomarker Comparisons

To visualize comparatively the pattern of significant metabolite, volume and principle component changes, the mean values from R6/2 and littermates were combined into a Z score at each age (Figure 6A). This transformation and the large range of Z values highlighted metabolite levels that differed from controls by many standard deviations. Early changes in cortical metabolites at 4 weeks could be distinguished from the otherwise largely parallel nature of metabolite changes in both regions. Many of the cortical metabolite changes were also stronger earlier. Volume changes in striatum and cortex paralleled many of the metabolite changes. Z scores of the computed cortical PC1 variable detected and tracked disease progression earlier than any single striatal metabolite or regional volume measure and as well as or better than any cortical metabolite. Principal component 1 tracked disease progression in cortex more strongly than in striatum.

Figure 6

Comparisons among potential biomarkers. (A) Z scores compare mean metabolite concentrations, principle components, volumes, and other measures in R6/2 to littermates at each age. Z scores reflecting changes in R6/2 greater than (red) or less than (blue) three standard deviations from the control means are shown. Gray designates measurements in which metabolite levels were not distinguishable (nd) from spectral noise in four or more mice. PC1 and PC2 values are from region-specific analyses (Figures 5A—D). (B) Comparison of biomarker efficacy. T-test P values comparing R6/2 and littermates on each indicated measures at each age. Dotted lines represent P=0.05 and P=0.01 levels. All values falling below the P=0.01 line also satisfied a False Discovery Rate test set to accept one false positive in 50 comparisons. Boxed area of top graph is expanded in bottom graph. To be represented, a potential biomarker measure had P<0.01 at 8, 12, and 15 weeks (Supplementary Table 3). Metabolites not shown in this category due to crowding were striatal Cr, PCr, GCP+PC, Gln and myo-Ins, and cortical Cr and Cr+PCr. At 4 weeks, P values for these metabolites did not cluster near the P=0.01 line. CV, cortical volume; SV, striatal volume; ctx, cortex; str, straitum.

To determine the relative efficacy of these various measures as biomarkers of disease progression in the R6/2 mouse, comparisons were made for each measure at each age between R6/2s and littermates. T-tests were used to determine the sensitivity of each independent measure at each age. P values for the most sensitive measures illustrated that while many could distinguish R6/2 from littermates at 8 weeks or later, only a few were able to do so at 4 weeks (Figure 6B; Supplementary Table 3). Over all time points, the cortical PC1 measure was most sensitive and able to distinguish R6/2 mice from littermate controls.

Discussion

Anatomical, behavioral, and biochemical measurements were examined as possible biomarkers for disease progression in the R6/2 mouse model of HD. Although able to distinguish hypoactive R6/2s from littermates at young ages, the climbing assay was unable to maintain that distinction across multiple sessions and ages. Anatomically, decreases in striatal volume were most pronounced and preceded faster than those in cortex or other brain regions. Biochemically, the steady-state values of brain metabolites provided a rich and varied source of information regarding disease progression with many metabolites showing robust changes as disease progressed. Variability in the timing of changes in individual metabolites suggested that multiple asynchronous mechanisms may be involved. Although many individual metabolites tracked disease, combining metabolites from cortex into two pseudo-variables yielded a biomarker with the greatest ability to distinguish R6/2 mice from littermates.

A Single Behavioral Test Was Insufficient to Monitor Disease Progression

A single, representative motor behavior, the climbing assay, was chosen for its simplicity of execution and previous use, in the absence of a rotarod apparatus (Hickey et al, 2005). The climbing assay did not prove to be reliable or sensitive over the course of the R6/2 lifespan. Only some of the measurements observed in the climbing test tracked disease progression, consistent with previous reports (Hickey et al, 2005; Menalled et al, 2009; Pallier et al, 2009). The climbing test failed to distinguish between genotypes because all mice reduced their activity in parallel over repeated tests. This decline in activity might be attributable to a memory of the procedure or to a general loss of playfulness as the mice matured (Auger and Olesen, 2009). Mice had no motivation to climb other than endogenous curiosity. Combined, multiple behavioral tests with strong motivational aspects may be needed to construct a reliable, repeatable behavioral biomarker (Pallier et al, 2009).

Decreases in Brain Volume Reflected Lack of Growth and Degeneration

Our data showed that accurate volumetric analysis of mouse brain can be rapidly achieved in anesthetized animals. Long imaging sessions on fixed tissue, in situ or ex vivo, were not necessary to obtain accurate volume measurements (Sawiak et al, 2009; Zhang et al, 2010). Manual segmentation of the images using commercially available software was sufficient to extract volume measurements and achieve accuracy for large structures. By 12 weeks, volume loss was detectable in subcortical regions consistent with stereological and voxel-based morphometry studies (Kremer et al, 1990; Stack et al, 2005; Douaud et al, 2009).

Two processes could be distinguished in the volume data; the continued growth of the brain over the adolescent period in littermates and the shrinkage of brain tissue with disease in the R6/2 mice. Both striatal and total brain volume increase between adolescence (1 month) and adulthood (3 months) in the C57Bl/6 mouse (Koshibu et al, 2004). The absence of growth in R6/2 striatum over this period was reported previously but not specifically noted (Zhang et al, 2010). In contrast to striatum, the hypothalamus grew in all mice between 4 and 8 weeks. These observations are reminiscent of similar regional specificity of the requirement for normal huntingtin for neuronal survival early in development (Reiner et al, 2003). Since normal mouse huntingtin expression decreases between 5 and 7 weeks the absence of striatal growth during adolescence may reflect HTT effects on growth factor influenced neuronal growth in this time period (Zhang et al, 2003). Beyond 8 weeks, the rate of hypothalamic shrinkage was comparable to that of the striatum, suggesting these regions may be equally susceptible to mutant huntingtin-induced degeneration.

Metabolite Changes Reliably Anticipated Volume Changes

Despite the growth in brain size observed between 4 and 8 weeks in littermate mice, only three metabolites changed significantly in this same time period, Lac, NAA, and myo-Ins. As shown previously in the developing rat (Tkac et al, 2003), metabolites have largely reached adult levels by the fourth week of rodent life. These longitudinal shifts in control mice suggest that some maturational changes in metabolite levels occurred during the transition from adolescence to adulthood. By comparison in R6/2, early changes at 4 weeks and multiple changes evident by 8 weeks proceeded with varied time courses. The contrast between early changes in Cr and Gln, the progressive changes in GPC+PC, the delayed changes in Tau, NAA, and Lac and the intermittent rises in myo-Ins over the R6/2 lifetime suggested that multiple mechanisms may be responsible for the observed steady-state metabolite changes. The earliest (4 weeks) cortical metabolite changes in GSH, Cr+PCr, Gln, and Glu occurred before significant regional shrinkage (8 weeks).

In both regions, NAA decreased beginning first in cortex, reaching a plateau level by 12 weeks, comparable to earlier studies in both R6/2 and other HD mouse models (Jenkins et al, 2005; Tsang et al, 2006; Tkac et al, 2007). Taurine became progressively elevated in both regions beginning at 8 weeks, sooner than we previously reported (Tkac et al, 2007). Reciprocal Glu and Gln changes have been registered after NMDA receptor blockade in rat models of schizophrenia (Iltis et al, 2009), supporting the interpretation that alterations in astrocyte-neuronal glutamate cycling can be detected by MRS. The absence of such reciprocal changes in R6/2 striatum and the presence of an apparently progressive increase in Gln probably signify a different underlying mechanism. As examples, Gln may increase with gliosis and both Gln and Glu can serve as osmolytes (Pouwels et al, 1998). Increases in Gln were prominent in yeast transfected with mutant huntingtin, supporting the idea that additional mechanisms may cause this increase (Joyner et al, 2010). Both Cr and PCr increased in striatum and cortex, suggesting that additional disease-related mechanisms beyond energy balance were perturbing their steady-state levels. In contrast to the current study, total creatine was reported to decrease beginning at presymptomatic stages in humans (Sturrock et al, 2010). The observed increase in cortical Lac was not present in the previous study (Tkac et al, 2007) but changes in Lac, both increases and decreases, have been noted by others (Jenkins et al, 2005; Tsang et al, 2006).

All three detectable metabolites involved in phospholipid metabolism changed with disease progression in R6/2; GPC+PC and myo-Ins increased steadily in both regions and PE decreased in striatum. Only cortical PE remained at control levels, making PE (along with Glu) one of the few metabolites to behave differentially between the two regions. The GPC+PC increases were the most pronounced and consistent among all metabolites. Increases in GPC+PC, normally greater in white matter than in gray matter, have been associated with myelin breakdown (Pouwels et al, 1998; Kirov et al, 2009). Diffusion Tensor Imaging studies in early stage HD suggest axonal degeneration and white-matter disruption are indeed pathophysiological sequelae (Weaver et al, 2009). Although the regions of interest in this mouse study were largely composed of gray matter avoiding the corpus callosum, both regions are marbled with axons and white matter. Indeed, an increase in functional anisotropy of subcortical gray matter in HD patients suggests fiber tract disruption may occur throughout the brain (Douaud et al, 2009). Elevated myo-Ins, principally found within astrocytes, is considered a marker for gliosis and is elevated in a number of neurodegenerative diseases (Fisher et al, 2002; Oz et al, 2010a). Similarly, increases in myo-Ins and decreases in NAA have been found in both R6/2 mice and presymptomatic and early stage HD patients (Jenkins et al, 2005; Tsang et al, 2006; Sturrock et al, 2010). In the human 3T MRS study, choline increased in early stage HD, consistent with the mouse data (Sturrock et al, 2010).

The current data largely replicated MRS data from R6/2 striatum at 8 and 12 weeks from our previous study (Tkac et al, 2007), showing the repeatability and reliability of the MRS measurements as biomarkers. This stability was maintained despite an intervening 7 years in the data collection, differences in diet between the two studies and rederivation of R6/2 by Jackson Laboratories. CAG repeat numbers were comparable between the two cohorts. The primary difference with the previous paper was the current absence of progressive changes in antioxidants, Asc and GSH. In the current work, cortical GSH was higher at all ages but only became significant in post tests at 8 weeks. Such variations suggest the oxidative stress component of HD in development may be sensitive to unidentified environmental or physiological factors. Mice in the current study did not develop signs of diabetes, prevalent in other R6/2 lineages (Hurlbert et al, 1999). The diabetic status of the mice in the 2007 study is unknown.

Biomarker

In a comparative evaluation of the principle components, volumes and individual metabolites for biomarker effectiveness, the cortical PC1 emerged as the most efficacious variable for distinguishing R6/2 mice from their wild-type littermates at all time points, even at 4 weeks. Despite largely parallel overall changes, cortical metabolites changed earlier than striatal metabolites, as observed in postmortem mouse brain extracts (Tsang et al, 2006). In contrast, regional shrinkage, accumulation of aggregates, cell death, and other measures have appeared more prominent in striatum (Han et al, 2010). Only recently cortical shrinkage has been considered as a possible early sign of disease in adults (Rosas et al, 2005). The prominence of cortical changes may be characteristic of early onset disease. Psychotic behaviors and cognitive difficulties, symptoms suggestive of cortical dysfunction, are characteristic of juvenile HD (Ribai et al, 2007). Thus, the R6/2 may model juvenile HD, overlaying disease on top of developmental maturation to produce a severe phenotype with more cortical involvement. Future studies should compare volume, metabolite, and PC measures among more slowly progressive mouse models to determine if early cortical change and cortical PC1 sensitivity are universal features of HD.

Many of the detected individual metabolite changes were able to distinguish R6/2 mice from littermates and therefore could qualify as unique biomarkers. Since metabolite changes in R6/2s can bear similarities to those in other diseases, designating a single metabolite as a biomarker to track disease progression may lack specificity. N-acetylaspartate is easy to measure in clinical magnets and decreases in all neurodegenerative conditions (Choi et al, 2007). In human and mouse spinal cerebellar ataxia, myo-Ins increases and NAA decreases, comparable to HD (Oz et al, 2010a, 2010b). Since multiple metabolite levels shift, ratioing one metabolite to another may result in inaccurate perceptions of what is driving disease progression and disregards information from any other measured variable (Sturrock et al, 2010). If both metabolites increase, the ratio becomes insensitive. To overcome these disadvantages, metabolite profiles were collapsed into PCs. The computed loading values reflected the amount of a metabolite change weighed against how much it contributed to distinguishing among individual mice. The unbiased regional brain PLS-DA PCs optimally explained the longitudinal variation associated with disease, a method similarly applied to MRS of plasma samples (Underwood et al, 2006). By performing PLS-DA on metabolite concentrations derived by LCModel as opposed to raw spectral values (Tsang et al, 2006), the computed PCs have the potential to be applicable to data acquired from different magnets. The PC analysis could prove to be useful in comparing severity of disease and predicting longevity among different R6/2 mouse colonies and across different HD mouse models. However, ease of clinical MRS acquisition may dictate which metabolites can be combined into a valid HD biomarker.

Footnotes

Acknowledgements

The authors wish to thank Arif Hamid, Alexandra Helgeson, and Yodit Tesfaye for quantification of mouse behavior and Dr Michael Michlin for insightful discussions.

The authors declare no conflict of interest.

Notes

Supplementary Information accompanies the paper on the Journal of Cerebral Blood Flow & Metabolism website ()

References

Auger

Olesen

(2009) Brain sex differences and the organisation of juvenile social play behaviour. J Neuroendocrinol 21:519–25.

Choi

Dedeoglu

Jenkins

(2007) Application of MRS to mouse models of neurodegenerative illness. NMR Biomed 20:216–37.

Davies

Turmaine

Cozens

Difiglia

Sharp

Ross

Scherzinger

Wanker

Mangiarini

Bates

(1997) Formation of neuronal intranuclear inclusions underlies the neurological dysfunction in mice transgenic for the HD mutation. Cell 90:537–48.

Douaud

Behrens

Poupon

Cointepas

Jbabdi

Gaura

Golestani

Krystkowiak

Verny

Damier

Bachoud-Levi

Hantraye

Remy

(2009) In vivo evidence for the selective subcortical degeneration in Huntington's disease. Neuroimage 46:958–66.

Ferrante

(2009) Mouse models of Huntington's disease and methodological considerations for therapeutic trials. Biochim Biophys Acta 1792:506–20.

Fisher

Novak

Agranoff

(2002) Inositol and higher inositol phosphates in neural tissues: homeostasis, metabolism and functional significance. J Neurochem 82:736–54.

Frank

Friedman

(1993) A statistical view of some chemometrics regression tools. Technometrics 35:109–48.

Han

You

Kordower

Brady

Morfini

(2010) Differential vulnerability of neurons in Huntington's disease: the role of cell type-specific features. J Neurochem 113:1073–91.

Hickey

Gallant

Gross

Levine

Chesselet

(2005) Early behavioral deficits in R6/2 mice suitable for use in preclinical drug testing. Neurobiol Dis 20:1–11.

10.

Hobbs

Barnes

Frost

Henley

Wild

Macdonald

Barker

Scahill

Fox

Tabrizi

(2010) Onset and progression of pathologic atrophy in Huntington disease: a longitudinal MR imaging study. AJNR Am J Neuroradiol 31:1036–41.

11.

Hurlbert

Zhou

Wasmeier

Kaddis

Hutton

Freed

(1999) Mice transgenic for an expanded CAG repeat in the Huntington's disease gene develop diabetes. Diabetes 48:649–51.

12.

Iltis

Koski

Eberly

Nelson

Deelchand

Valette

Ugurbil

Lim

Henry

(2009) Neurochemical changes in the rat prefrontal cortex following acute phencyclidine treatment: an in vivo localized (1)H MRS study. NMR Biomed 22:737–44.

13.

Jenkins

Andreassen

Dedeoglu

Leavitt

Hayden

Borchelt

Ross

Ferrante

Beal

(2005) Effects of CAG repeat length, HTT protein length and protein context on cerebral metabolism measured using magnetic resonance spectroscopy in transgenic mouse models of Huntington's disease. J Neurochem 95:553–62.

14.

Jenkins

Rosas

Chen

Makabe

Myers

MacDonald

Rosen

Beal

Koroshetz

(1998) 1H NMR spectroscopy studies of Huntington's disease: correlations with CAG repeat numbers. Neurology 50:1357–65.

15.

Joyner

Matheke

Smith

Cichewicz

(2010) Probing the metabolic aberrations underlying mutant huntingtin toxicity in yeast and assessing their degree of preservation in humans and mice. J Proteome Res 9:404–12.

16.

Kirov II Patil

Babb

Rusinek

Herbert

Gonen

(2009) MR spectroscopy indicates diffuse multiple sclerosis activity during remission. J Neurol Neurosurg Psychiatry 80:1330–6.

17.

Koshibu

Levitt

Ahrens

(2004) Sex-specific, postpuberty changes in mouse brain structures revealed by three-dimensional magnetic resonance microscopy. Neuroimage 22:1636–45.

18.

Kremer

Roos

Dingjan

Marani

Bots

(1990) Atrophy of the hypothalamic lateral tuberal nucleus in Huntington's disease. J Neuropathol Exp Neurol 49:371–82.

19.

Mangiarini

Sathasivam

Seller

Cozens

Harper

Hetherington

Lawton

Trottier

Lehrach

Davies

Bates

(1996) Exon 1 of the HD gene with an expanded CAG repeat is sufficient to cause a progressive neurological phenotype in transgenic mice. Cell 87:493–506.

20.

Menalled

El-Khodor

Patry

Suárez-Fariñas

Orenstein

Zahasky

Leahy

Wheeler

Yang

MacDonald

Morton

Bates

Leeds

Park

Howland

Signer

Tobin

Brunner

(2009) Systematic behavioral evaluation of Huntington's disease transgenic and knock-in mouse models. Neurobiol Dis 35:319–36.

21.

Hutter

Tkac

Clark

Gross

Jiang

Eberly

Bushara

Gomez

(2010a) Neurochemical alterations in spinocerebellar ataxia type 1 and their correlations with clinical status. Mov Disord 25:1253–61.

22.

Nelson

Koski

Henry

Marjanska

Deelchand

Shanley

Eberly

Orr

Clark

(2010b) Noninvasive detection of presymptomatic and progressive neurodegeneration in a mouse model of spinocerebellar ataxia type 1. J Neurosci 30:3831–8.

23.

Pallier

Drew

Morton

(2009) The detection and measurement of locomotor deficits in a transgenic mouse model of Huntington's disease are task- and protocol-dependent: influence of non-motor factors on locomotor function. Brain Res Bull 78:347–55.

24.

Paulsen

Langbehn

Stout

Aylward

Ross

Nance

Guttman

Johnson

McDonald

Beglinger

Duff

Kayson

Biglan

Shoulson

Oakes

Hayden

(2008) Detection of Huntington's disease decades before diagnosis: the predict HD study. J Neurol Neurosurg Psychiatry 79:874–80.

25.

Pouwels

Kruse

Korenke

Mao

Hanefeld

Frahm

(1998) Quantitative proton magnetic resonance spectroscopy of childhood adrenoleukodystrophy. Neuropediatrics 29:254–64.

26.

Reiner

Dragatsis

Zeitlin

Goldowitz

(2003) Wild-type huntingtin plays a role in brain development and neuronal survival. Mol Neurobiol 28:259–76.

27.

Ribai

Nguyen

Hahn-Barma

Gourfinkel-An

Vidailhet

Legout

Dode

Brice

Durr

(2007) Psychiatric and cognitive difficulties as indicators of juvenile huntington disease onset in 29 patients. Arch Neurol 64:813–9.

28.

Rizk-Jackson

Stoffers

Sheldon

Kuperman

Dale

Goldstein

Corey-Bloom

Poldrack

Aron

(2010) Evaluating imaging biomarkers for neurodegeneration in pre-symptomatic Huntington's disease using machine learning techniques. Neuroimage 56:788–96.

29.

Rosas

Hevelone

Zaleta

Greve

Salat

Fischl

(2005) Regional cortical thinning in preclinical Huntington disease and its relationship to cognition. Neurology 65:745–7.

30.

Sawiak

Wood

Williams

Morton

Carpenter

(2009) Voxel-based morphometry in the R6/2 transgenic mouse reveals differences between genotypes not seen with manual 2D morphometry. Neurobiol Dis 33:20–7.

31.

Stack

Kubilus

Smith

Cormier

Del Signore

Guelin

Ryu

Hersch

Ferrante

(2005) Chronology of behavioral symptoms and neuropathological sequela in R6/2 Huntington's disease transgenic mice. J Comp Neurol 490:354–70.

32.

Sturrock

Laule

Decolongon

Dar

Coleman

Creighton

Bechtel

Reilmann

Hayden

Tabrizi

Mackay

Leavitt

(2010) Magnetic resonance spectroscopy biomarkers in premanifest and early Huntington disease. Neurology 75:1702–10.

33.

Tkac

Dubinsky

Keene

Gruetter

Low

(2007) Neurochemical changes in huntington R6/2 mouse striatum detected by in vivo 1H NMR spectroscopy. J Neurochem 100:1397–406.

34.

Tkac

Rao

Georgieff

Gruetter

(2003) Developmental and regional changes in the neurochemical profile of the rat brain determined by in vivo 1H NMR spectroscopy. Magn Reson Med 50:24–32.

35.

Tsang

Woodman

McLoughlin

Griffin

Tabrizi

Bates

Holmes

(2006) Metabolic characterization of the R6/2 transgenic mouse model of Huntington's disease by high-resolution MAS 1H NMR spectroscopy. J Proteome Res 5:483–92.

36.

Underwood

Broadhurst

Dunn

Ellis

Michell

Vacher

Mosedale

Kell

Barker

Grainger

Rubinsztein

(2006) Huntington disease patients and transgenic mice have similar pro-catabolic serum metabolite profiles. Brain 129:877–86.

37.

Weaver

Richards

Liang

Laurino

Sami

Aylward

(2009) Longitudinal diffusion tensor imaging in Huntington's disease. Exp Neurol 216:525–9.

38.

Zhang

Peng

Jahanshad

Hou

Jiang

Masuda

Langbehn

Miller

Mori

Ross

Duan

(2010) Longitudinal characterization of brain atrophy of a Huntington's disease mouse model by automated morphological analyses of magnetic resonance images. Neuroimage 49:2340–51.

39.

Zhang

Drozda

Chen

Ren

Mejia Sanchez

Leavitt

Cattaneo

Ferrante

Hayden

Friedlander

(2003) Depletion of wild-type huntingtin in mouse models of neurologic diseases. J Neurochem 87:101–6.

40.

Zuccato

Valenza

Cattaneo

(2010) Molecular mechanisms and potential therapeutical targets in Huntington's disease. Physiol Rev 90:905–81.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.49 MB