| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,1
* Department of Biological Chemistry and Israel Center for Neuronal Computation, The Hebrew University of Jerusalem, Jerusalem, Israel;
School of Chemistry, The Raymond and Beverly Sackler Faculty of Exact Sciences, Tel-Aviv University, Ramat Aviv, Israel; and
Tel Aviv Sourasky Medical Center, Human Brain Imaging Unit, The Wohl Institute for Advanced Imaging, Tel Aviv, Israel
4Correspondence: Department of Biological Chemistry and Israel Center for Neuronal Computation, The Hebrew University of Jerusalem, Jerusalem 91904, Israel. E-mail: soreq{at}cc.huji.ac.il
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: cholinergic neurotransmission • acetylcholinesterase • blood-brain barrier • aquaporin 4
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and accompanies neurodegenerative diseases (e.g., Alzheimer’s and Parkinson’s disease) (2
, 3)
. However, it remained unclear whether imbalanced cholinergic neurotransmission may by itself lead to chronic brain damage(s) and, if it does, which molecular, cellular, and physiological pathways are involved. To approach this question, we used transgenic (TgS) mice with neuronal overexpression of synaptic acetylcholinesterase (AChE-S) of human origin (4)
. The TgS mouse displays a complex central and peripheral phenotype including fluctuations of brain acetylcholine (ACh) levels, elevated high affinity choline transport, accelerated stress-related neuropathology associated with progressive cognitive deterioration, and hypersensitivity to drugs and to close head injury (5)
. Peripheral impairments include complex neuromuscular pathologies (6)
; at least some of the central and peripheral malfunctioning symptoms could be corrected by anticholinesterase or antisense treatments, suppressing the excess AChE activities (6
, 7)
. While this detailed picture largely points at a primary cholinergic imbalance due to neuronal AChE overexpression in the TgS mouse, the underlying changes in neuronal gene expression were never explored, and no integrative explanation was offered for this phenotype.
Prolonged AChE overexpression occurs in the mammalian brain under psychological stress (8)
, and various studies provide evidence that such stress can induce blood-brain barrier (BBB) disruption (9
10
11
12)
via yet unknown mechanisms. Others contested this hypothesis (13)
, emphasizing the complexity of the issue. Viewing the TgS mouse as an appropriate model system for exploring these processes, we initiated a study of its BBB functioning in an effort to unravel the origin for the complex TgS phenotype and find out the consequences of chronic cholinergic imbalances. This study was designed to address two main questions: is BBB functioning impaired in the TgS mouse and, if so, which molecular mechanism(s) may lead to this impairment.
To answer the first question, we turned to magnetic resonance imaging (MRI). MRI is a noninvasive technique with high spatial and temporal resolution that enables repeated inspections of single live subjects. Combined use of various parameters obtained from MRI (e.g., T1, T2, T2*, diffusion maps, and perfusion MRI) allows detailed characterization of brain water content and motion, as well as BBB status. For example, diffusion-weighted MRI (DWI) allows measurement of the apparent diffusion coefficient (ADC), which is markedly reduced in the acute ischemic brain (
30–40%) (14
, 15)
. T2 and ADC are widely used as a measure for various edema-related central nervous system (CNS) pathologies, such as ischemia (16
, 17)
. Moreover, injection of contrast-enhancing compounds (perfusion MRI) provides a dynamic measure of BBB function (18
, 19)
.
To link the physiological MRI measurements with the underlying molecular mechanisms and answer the second question, we used DNA microarrays. To improve the specificity of this analysis by limiting the neuronal complexity and focusing on a stress-relevant brain region, we selected the prefrontal cortex (PFC) (20)
. The PFC contributes substantially to cognitive and emotional functioning, both of which are disrupted under stress (21)
. Being densely populated with glucocorticoid receptors, the PFC participates in the regulatory response to stress of the hypothalamus-pituitary-adrenal axis (22)
. Functional MRI demonstrated impaired PFC performance in patients with post-traumatic stress disorder (PTSD) (23)
and survivors of the Tokyo subway terror attack (March 20, 1995) exposed to the poisonous anticholinesterase sarin showed reduction in gray matter volume in the medial PFC (24)
.
To combine our physiological and molecular studies of the consequences of chronic cholinergic abnormalities, we subjected live TgS mice to brain imaging analyses, tested their PFC gene expression using microarrays, and verified the results using molecular biology and immunohistochemistry tools. We observed reduced water ADCs and BBB abnormalities in the brain of TgS mice and found evidence supporting the involvement of specific alterations in brain channels and transporters, (e.g., CLCN3 and AQP4) in these aberrations. Our findings support the notion of causal association of primary cholinergic imbalances with transport, water diffusion, and BBB impairments and point to such brain impairments as the putative origin of progressive neurodeterioration phenotypes.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Body temperature of wild-type (wt) (n=2) and transgenic (n=4) mice was measured using a chrome/alume probe (type K) connected to a TES-1310 digital thermometer (TES Electrical Electronic Corp., Taipei, Taiwan).
Magnetic resonance imaging
MRI experiments were performed on an 8.4 T spectrometer (Bruker, Rheinstetten, Germany) equipped with a minimaging accessory capable of producing pulse gradients of up to 20 gauss/cm in each direction using an in-house built surface coil of 15 x 10 mm. The first MRI protocol ("diffusion protocol") included a T1-weighted image, T2-weighted image, and a series of diffusion-weighted MR images. The multi-slices T1- and T2-weighted images (5 slices of 1.35 mm thickness with a 0.2 mm gap) were acquired using the spin echo sequence with a field of view of 2.5 x 2.5 cm, matrix dimensions of 256 x 128 (transformed into a 256x256 matrix), and two averages. The repetition and echo times (TR/TE) of the T1- and T2-weighted MRI images were 600/20 and 3000/50 ms, corresponding to image acquisition times of 156 s and 13 min, respectively. The diffusion-weighted images were acquired using the stimulated echo diffusion pulse sequence with the following parameters: TR = 1800 ms, TE = 41.6 ms, 
= 8 ms, 
= 100 ms, two averages, field of view of 2.5 x 2.5 cm, matrix dimensions of 256x128 (transformed into a 256x256 matrix), and five slices of 1.35 mm thickness with a 0.2 mm gap. The diffusion gradient strength, G, was incremented from 0 to 9.23 gauss/cm in four steps, yielding a maximal b value of 3.79 x 105 s cm–2. The entire diffusion protocol was completed within 1 h.
The second MRI protocol ("perfusion protocol") included two T1-weighted images before the gadolinium diethylenetriaminepentaacetic acid (Gd-DTPA, Magnetol, Soreq Radiopharmaceuticals, Yavne, Israel) injection, a series of gradient echo images (T2* images) pre- and post-Gd-DTPA injection, and two T1-weighted images after the injection of the contrast agent. T1-weighted images were acquired with the same parameters as the T1-weighted images in the diffusion protocol. Acquisition of postcontrast T1-weighted MR images was initiated in all cases 11 min after Gd-DTPA injection. The series of T2*-weighted images involved single slice (1.35 mm thick), low angle (
=10) gradient echo MR images acquired with TR = 60 ms, TE = 15 ms, one average, a field of view of 2.5 x 2.5 cm, and matrix dimensions of 64 x 32, transformed into 64 x 64 images. The total acquisition time per image in that series was 1 s. At least 60 gradient echo images were acquired in each study. After the 10th image, 0.42 mmol/kg of Gd-DTPA was injected through the tail vein. The first 10 images were used to obtain an average baseline image. Again, the entire protocol was completed in less than 1 h.
To measure and compare Gd-DTPA blood clearance rates, we injected the tail vein of wt (n=4) and TgS (n=4) mice with 200 nmol/kg weight of Gd-DTPA diluted in phosphate-buffered saline (PBS). Tail blood samples were withdrawn 5, 11, and 16 min postinjection, diluted 1:25 in PBS, then 1:1000 in double-distilled water, and Gd-DTPA levels were measured using Inductively Coupled Plasma Mass Spectrometer (ICP-MS, Perkin Elmer, MA, USA). Similarly, prepared blood samples from naïve mice and PBS alone served as negative controls.
Image processing
ADC maps were produced using an in-house Matlab® program (The MathWorks Inc., Natick, MA, USA). The set of diffusion-weighted images were analyzed for each pixel according to Eq. 1
(25)
,
 | (1) |
is the gyromagnetic ratio, and G are the duration and amplitude of the diffusion gradient pulses, and is the time interval between the leading edges of these pulses. In Eq. 1
, factor b represents the total diffusion weighting and (–/3) is the effective diffusion time for the case of rectangular diffusion gradients. We extracted from the ADC maps the diffusion coefficients of the entire brain and of six different anatomical regions. The changes in signal intensities of the T2*-weighted images pre- and post-Gd-DTPA injection were used to extract the relative cerebral blood volume (rCBV), relative mean transit time (rMTT), and the relative cerebral blood flow (rCBF) using Eqs. 2
, 3
, and 4
, respectively.
 | (2) |
 | (3) |
 | (4) |
DNA microarrays
PFCs from groups of four wt and four TgS animals were dissected on ice. Total PFC RNA was extracted using the RNeasy mini kit (Qiagen, Hilden, Germany). MgU74Av2 (www.affymetrix.com) mouse microarrays were used according to MIAME (www.mged.org) as described in the Affymetrix manual (www.affymetrix.com/support/downloads/manuals/expression_s2_manual.pdf). Two independent preparations of pooled PFC RNA of wt and TgS brains (4 animals each) were hybridized to four standard Affymetrix mouse arrays (MgU74Av2) containing 12,450 different transcripts. After identical data filtering processing on the four arrays, we compared the wt vs. wt, TgS vs. TgS and the four possible combinations of wt vs. TgS profiles. Analysis involved fluidic station FS-400, MAS 5.0 software and GeneArray 2500 scanner (Affymetrix). Initial filtering procedures included an absence/presence (A/P) test, where each gene should receive at least one "P" in order to be included. Marginal (M) detection levels were counted as absent. Scatter plots were computed with the remaining genes (6000 of 12,450). Comparison tests included increase/decrease (I/D) test, and a log2 ratio of higher than 1 (or lower than –1). To identify genes that were commonly altered in all tests, we intersected the four sets of altered genes in each of the wt vs. TgS comparisons. Only genes for which the expression was similarly altered (displaying at least 2-fold difference in expression level) in the four different comparisons were regarded as "changed." Data mining and graphs were carried out using the Spotfire program (Spotfire, Somerville, MA, USA). All raw data files were submitted to GEO (http://www.ncbi.nlm.nih.gov/geo/). Accession numbers for wt-1, TgS-1, wt-2, and TgS-2 are GPL795, GPL796, GPL797, GPL798, respectively. To identify categories over-represented among the genes that were changed in the TgS brain, we performed Gene Ontologies (GO) classification using the EASE software (26)
with a P value of 0.1 (1-tailed Fisher exact probability test).
Real-time PCR
RNA template (100 ng) was used for one-step RT-PCR amplification protocol (SYBR Green I RNA amplification kit, Roche, Basel, Switzerland). Primers were designed using the LC Primer/Probe design software (Roche) and ordered from Sigma (Jerusalem, Israel). All primers are given in 5' to 3' direction (Table 1
).Standard curves of varying template concentrations were generated as described (27)
for each primer pair; amplification efficiency for each primer pair was calculated and used for relative quantification. Samples were normalized against -tubulin mRNA levels, which were not altered in TgS brains. All products were electrophoresed on 1% agarose gels and sequenced for verification. If more than one clear band was detected on agarose, primers were replaced or samples omitted.
|
Immunohistochemistry
Brains were dissected and transferred to fresh 4% paraformaldehyde in PBS solution for 7 days fixation. Sagittal paraffin-embedded sections were deparaffinized using xylene and a descending series of ethanol dilutions in double-distilled water (DDW). Antigens were retrieved by heating for 10’ in a microwave oven. Intrinsic horseradish-peroxidase (HRP) activity was quenched by immersing slides in 3% H2O2 for 30 min. Nonspecific binding was blocked by incubation in serum (1 h, room temp.). Primary antibodies (polyclonal goat anti-CLCN3, sc-17572; Santa Cruz Biotechnology, Santa Cruz, CA, USA; and mouse monoclonal AQP4, Serotec, Cergy Saint-Christophe, France) were used in 1:100 dilutions. Biotinylated donkey anti-goat secondary antibody (Jackson Immuno-Research, West Grove, PA, USA) was used to detect CLCN3 and HRP-anti-mouse to detect AQP4. Colorimetric enzymatic detection was done with the Elite ABC-VECTASTAIN kit (Vector Laboratories inc., Burlingame, CA, USA) and diaminobenzidine (DAB) nickel solution for CLCN3 or standard DAB (Sigma) detection for AQP4.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
(Fig. 1
A). This transgene is expressed in brain neurons but not in glia or muscle (28)
. Brain AChE activity was found to be at least 2-fold higher in TgS mice than controls and considerably more variable among individuals than in the parent strain (Fig. 1B
), compatible with the production of the murine stress-associated AChE variant AChR-R and variable learning and memory impairments among TgS mice (7)
. In preparation for the MRI experiments, we tested these animals’ response to anesthesia by measuring the body temperatures of TgS mice and matched controls. These measurements yielded higher body temperatures in the transgenic mice under anesthesia (Fig. 1C
), compatible with the fluctuating levels and altered regulation of ACh shown in the TgS brain by microdialysis (29)
, the impaired central responses of TgS mice to anticholinesterases (8)
, and their impaired circadian cycle when challenged by stress (7)
. Altogether, these findings suggest a pronounced yet variable cholinergic imbalance in the CNS of TgS mice.
|
TgS mice present altered brain diffusion processes
Diffusion-weighted MR images were taken to compare bulk MRI parameters in the TgS brains to those of matched controls. Others reported considerably higher sensitivity for the DWI approach compared with T2-weighted MRI for the evaluation of numerous CNS pathophysiological conditions (14
15
16
, 18
, 19
, 30)
. The imaging version of the simulated echo (diffusion sequence used for conducting DWI experiments and the formula for obtaining the diffusion coefficient (D)) from the signal decay in these experiments are shown in Fig. 2
A.
|
The T1- (data not shown) and T2-weighted MR images (Fig. 2B
, T2 images) of the two groups of mice predictably appeared similar. This excluded the possibility of gross morphological or water content differences between brains of wt vs. TgS mice, compatible with our histological observations and earlier findings (5)
. In view of the sensitivity of ADC parameters for detecting CNS pathologies associated with water mobility differences, we next measured ADC values (Fig. 2B
, diffusion maps). A statistically small decrease in the ADC values of the TgS brains was found in motor, somatosensory and piriform cortices and in the striatum (Fig. 2C, D
, *P<0.05, **P<0.01, 2-tailed Student’s t test). The differences observed in the cingulate cortex and the preoptic area were not significant, implying selective brain region regulation of the ADC. Average ADC values for the entire brain also showed a small but statistically significant decrease for the transgenic group (ADCbrain, Fig. 2E
). Note that the lower ADC values in the brain of TgS mice could not be due to lower body temperature; rather, they occurred in spite of the higher body temperatures seen in TgS mice under anesthesia (see Fig. 1C
, above).
Increased blood-brain barrier permeability in the TgS mouse
The decreased ADC values together with the apparent stress-related characteristics of the TgS animals further suggested altered regulation of BBB integrity. To test the physiological significance of these changes, we used contrast-enhanced MRI after i.v. injection (tail vein) of Gd-DTPA in a search for differences in perfusion rates and in BBB permeability between the two groups. The MRI protocol used in these experiments is schematized in Fig. 3
A. Figure 3B
shows the normalized brain signal obtained from the T2*-weighted images for the different mice studied pre- and post-Gd-DTPA injection. On average, signal intensities in wt mice were reduced to 65 ± 12% of the initial signal after Gd-DTPA injection, whereas signal intensities in TgS brains were reduced to 48 ± 7% of the initial signal within the same time frame (P<0.04). The calculated ratios of the rCBV, rMTT, and rCBF values of the wt and TgS mouse groups were 0.56 ± 0.21 (P<0.02), 1.03 ± 0.30 (P>0.05), 0.55 ± 0.20 (P<0.01), respectively.
|
Whereas wt mice responded to Gd-DTPA injection in closely similar patterns, TgS mice showed pronounced variability in this response (Fig. 3B
). These differences matched the inter-animal differences in brain AChE activities demonstrated above (Fig. 1B
). Moreover, for up to 100 s after Gd-DTPA injection, signal recovery in TgS brains was considerably slower than in wt mice (Fig. 3B
). This additional test supported the notion of attenuated leakage from the brain’s microvasculature of wt compared with TgS mice. Contrast-enhanced MRI thus supported the hypothesis that the BBB in wt mice was more efficient in preventing penetrance of Gd-DTPA than in the TgS mice. To ascertain that differential Gd-DTPA clearance rates from the circulation between the two strains was not causing these changes, we measured and compared blood Gd-DTPA clearance in wt and TgS animals and found no significant differences between the two (P>0.2, data not shown).
To test the steady-state integrity of the BBB, we have further measured T1 images in control and TgS mice at 15 min intervals, both before and 11 min after injection of the contrast agent. Figure 3C
shows the subtraction brain maps (100x(post-pre)/pre) of two representative wt (left) and two representative TgS mice (right). The subtraction brain maps reveal higher signals in the brains of TgS than with wt mice (Fig. 3D
), implying that more contrast agent molecules accumulated in the TgS brains. This observation further supports the notion of weaker protection from penetrance through the BBB in TgS mice compared with wt animals. In conclusion, TgS mice displayed elevated Gd-DTPA leakage into the brain, reminiscent of the findings of others for null mice for the mdr1a gene (31)
. These findings represent a difference in BBB permeability between the two mice, as the GdDTPA clearance from blood was found to be similar in the two strains of mice.
Variable gene expression patterns in the TgS prefrontal cortex
To gain insight into the gene expression patterns potentially underlying the observed differences in brain water ADCs and BBB functioning, we performed DNA microarrays. Focusing on the PFC as a stress-modulating brain region, we compared gene expression profiles of TgS and gender, age and strain-matched wt controls. The two wt preparations yielded very similar profiles (Fig. 4
A, B), with merely five genes (0.08% of 6000 of which the detection levels were significantly above background) showing a > 2-fold detection signal difference. The two TgS microarrays displayed more limited reproducibility, with 37 genes (0.62%) outside the 2-fold limit (Fig. 4A
, middle). This correlated with the relatively high individual behavioral differences among TgS compared with wt animals (7)
, and with the higher variability in their striatal AChE activity levels (Fig. 1B
) and body temperature under anesthesia (Fig. 1C
). Notwithstanding, the number of genes outside the 2-fold limit was well below the expected 2% for same-sample comparisons. In sharp contrast, comparison of PFC-expressed genes in wt with TgS mice revealed over 200 genes with a > 2-fold difference between the two groups in each of the four comparisons (Fig. 4A
, bottom). This yielded a list of 213 genes (3.6%) (Supplementary Table S1), of which 63 are of unknown identity, producing a list of 140 genes with known annotations. The greater variability between PFC-expressed genes in the TgS mouse, and the differences between TgS and wt mRNAs are highlighted in the hierarchic clustering of the expression arrays shown in Fig. 4B
.
|
Over-representation of gene classes associated with BBB features
The up-regulated genes in TgS compared with wt PFC (Supplementary Table S1) were analyzed according to their Gene Ontology annotations (http://www.geneontology.org) using EASE software (26)
. This analysis revealed over-represented terms in the molecular function (Fig. 5
A) and biological process (Fig. 5B
) ontology branches for TgS brains (Supplementary Table S2). Two major (in terms of number of representative genes) categories with highly significant differences included transcription-related processes and ion transport/channel activity-related processes.
|
Within the transcriptional-related class, we identified DNA binding transcription factors, trans-acting transcriptional activators/repressors, chromatin remodeling proteins, and hormone receptors. Changes in such proteins predictably yield secondary changes downstream, resulting in differential expression of many more genes, compatible with the microarray outcome.
Within the ion transport class were included potassium, chloride (e.g., CLCN3), calcium, and sodium channels as well as a potassium channel motif kinase (Mark2), suggesting diverse changes in the transport of various ions. Increasingly recognized nonenzymatic roles of AChE include cell adhesion properties, reflected in the structural similarities between AChE and neurotactin (32
, 33)
. Altered regulation of neuronal adhesion properties in the PFC could potentially affect cell volume and/or transport properties of the TgS brain, independently or jointly with the changes in ion transport. Compatible with this prediction, adhesion-related genes were over-represented in the TgS brain. These included neuropilin, T-cadherin, cadherin-2, contactin-1, PTEN (phosphatase and tensin homologue deleted on chromosome-10), and an L1-like protein, to name a few. Compatible with the recently reported involvement of AChE-S in apoptosis (34)
, the TgS PFC further showed elevated transcript levels of the TNF-related cell death gene TR2L (35)
, the p53-related imprinted neuronal gene Peg-3 (36)
, and the TIA-1 gene associated with RNA metabolism and alternative splicing-related apoptosis (37)
, among others. Overexpression of the identified gene classes could thus be directly related to both the cholinergic (ion transport) and noncholinergic (apoptosis, adhesion) roles attributed to AChE.
Several representatives of the overexpressed channel- or adhesion-related genes were selected for real-time RT-PCR analysis. Two genes (SCN8 and Kcnma) showed no significant increase in RT-PCR, but 11 of 13 demonstrated reproducible and significant overexpression in the TgS PFC compared with wt (Fig. 5C, D
), validating the microarray observations. Alpha-tubulin, which was selected for control based on our microarray results, remained unchanged in our real-time PCR analysis as well (Fig. 5C, D
, right).
Cellular accumulation of ion/water controllers in the TgS brain
To more deeply explore the significance of the changes observed in the microarray analysis, we decided to focus on two specific genes: CLCN3 and AQP4. Located on synaptic and lysosomal vesicles, the voltage-gated chloride channel CLCN3 contributes to their luminal acidification (38)
. CLCN3 was suggested to be involved in the regulation of cell volume (39)
, belonging to the family of volume-sensitive organic osmolyte and anion channels (VSOACs) (40)
. AQP4 was shown to be involved in brain water transport (41)
, ischemia (42)
, and edema (43)
as well as in the development and integrity of the BBB (44)
. Therefore, CLCN3 and AQP4 could both be potentially involved in the complex phenotype of the TgS mouse.
Microarray and RT-PCR analyses, which use total RNA as input, do not necessarily imply elevation of protein levels. They cannot distinguish between elevated mRNA (or protein) levels in all cells that express a certain gene or increased number of cells expressing that gene. To measure protein levels, we first used immunoblot analysis of total PFC protein extracts of wt and TgS mice for CLCN3 and AQP4. In both cases, elevated protein levels were observed (Fig. 5E
). To explore their cellular and subcellular distribution, we immunostained the CLCN3 and AQP4 proteins in brain sections from wt and TgS mice.
CLCN3 staining was considerably higher in the TgS brain, mainly in the brainstem, with 37% increase in the number of CLCN3-positive brainstem cells (P<0.01, 2-tailed Student’s t test, Fig. 6
A, B). Histochemical staining revealed that CLCN3 staining was observed in glia (57% of the labeled cells) and neurons (43%) in wt and TgS mice. CLCN3 distribution within wt brain cells showed cell circumference location whereas CLCN3 in TgS cells filled the entire cellular space (Fig. 6C-E
). Thus, the cholinergic imbalance in the TgS brain and/or the nonenzymatic features of AChE enhanced CLCN3 expression levels and modified its cellular and subcellular distribution in cells of neural and glial origin.
|
Staining intensity of the water channel AQP4 was significantly stronger in the brains of TgS compared with wt animals (Fig. 7
A–D), yet the number of labeled cells was similar (Fig. 7B
, left). Thus, excess of AChE-S intensified AQP4 expression without altering the number of expressing cells. Most cells in the wt brain showed punctuated cytoplasmic labeling. In contrast, cells in the TgS brain’s microvascular system were labeled in the cytoplasm and nucleus (Fig. 7E, F
). Histological H&E staining revealed no differences in cell number, density, or the incidence of necrotic or apoptotic cells between wt and TgS animals (data not shown), suggesting the occurrence of functional changes but not cell death.
|
The association of a large number of overexpressed channels and transporters in the TgS brains including, specifically, AQP4, and BBB abnormalities, suggested the possible involvement of AQP1, since it is known to be selectively expressed in choroids plexus of the lateral ventricles (45
, 46)
. We therefore performed immunostaining for AQP1 in wt and TgS brains, but found no significant differences in the expression levels or number of AQP1-expressing cells between the two strains (Supplementary Fig. S1).
The cholinergic imbalance, CLCN3 and AQP4 overexpression, and BBB impairments further showed spatial associations. Within the TgS brain, transgenic AChE-S and host AChE-R showed distribution patterns that mimic those of the normal brain, with particularly high expression levels in the basal nuclei and cortical periphery, respectively (Fig. 8
A, B). This matched the sum of the observed diffusion differences (Fig. 8B
). Intriguingly, AQP4 and CLCN3 accumulation (Fig. 8C
) matched these regions as well as the brain diffusion abnormalities observed in the TgS brain. While higher resolution tests and different experimental approaches would be required to establish the corresponding cellular events, these similarities were compatible with the hypothesis that excess AChE-S and the resultant cholinergic imbalance and cell adhesion changes induced elevations in water and ion transport, leading to consequent BBB promiscuity (Fig. 8D
). The differences in CLCN3 and AQP4 overexpression thus suggested causal involvement with the physiological phenotype and BBB malfunctioning of the TgS brains.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Lower ADC in TgS mice
The indistinguishable T1 and T2 images in TgS and wt mice suggest similar total brain water content. Nevertheless, the corresponding diffusion maps were significantly different for these two strains. ADC, a common parameter for assessing the motion of water in tissues, is extremely sensitive to cytotoxic edema during cerebral insults such as stroke (14
15
16)
and is used to evaluate water motion in other CNS pathologies (17
, 47)
. Similar to animal models of ischemia or stroke with elevated AQP4, TgS mice displayed lower baseline ADC values than their age-matched controls. As expected, the transgenic effect was less dramatic to the integrity of the brain than those of direct physical insults. Quantitative comparison of the observed ADC differences in TgS mice (4.5%) shows a limited effect compared with that observed in ischemic stroke (30–40%) (14
15
16)
. However, to the best of our knowledge this is the first report of baseline differences in ADC in an animal strain with transgenic overexpression of a single protein (48)
, highlighting the profound physiological consequences induced by the primary subtle overexpression of the AChE-S transgene.
Putative causes for the reduced ADC values
At the organismal level, lower body temperature could have caused ADC decreases. However, TgS mice presented higher body temperature than wt mice and demonstrated limited decreases in body temperature under the anesthetic conditions used in the MRI tests. The effect of brain tissue density was also considered. Richer dendritic arborisations could lead to smaller extracellular volumes and possibly higher restriction, and hence lower average ADC values. However, TgS mice suffer progressive loss of their dendritic arborisations (49)
. Together with the higher body temperature, this could potentially increase the ADC values, suggesting that the ADC decreases seen may in fact conceal yet larger functional discrepancies.
At the cellular level, one would expect faster motion of ions and water molecules into and out of TgS brain cells, with a possible change in cell volume regulation due to the up-regulated CLCN3 and other ion channels. In the spinal cord of TgS mice, we indeed observed increased cell volume of motoneurons (5)
, which might be explained by such changes. Even a subtle increase in the volume of brain cells will reciprocally reduce the extracellular space, thus restricting diffusion and leading to reduced ADC, as is the case in the ischemic brain (15)
.
At the molecular level, the ADC reduction observed in the brains of TgS mice might be due to the elevated mRNA and protein levels of ion and water channels, such as AQP4. AQP4 overexpression follows the course of the brain edema induced by ischemia (42)
and displays an inverse correlation with ADC after ischemia (50)
. In dystrophin null transgenic mice with mislocalized AQP4, the decrease in ADC that normally follows edema was delayed (51)
. This is compatible with the findings that AQP4 null mice show delayed brain edema after water intoxication (43)
and that TgS mice show exaggerated edema after head injury (5)
.
Transgenic effects on BBB integrity
In healthy mammals, injected contrast agents would transiently reach the brain’s microvasculation system, creating a short-lasting signal change. This was indeed the case in the wt FVB/N mice, demonstrating an apparently intact BBB. However, TgS animals displayed considerable variability in the extent and duration of brain signal change after perfusion of Gd-DTPA, compatible with the assumption of variably disrupted BBB properties in TgS mice: in several of these mice, larger quantities of the injected contrast agent apparently leaked out of the vascular system into the brain tissue itself. Changes in ADC values are often, but not always, accompanied by apparent BBB disruption.
Several pieces of evidence support the assumption that reduced ADC values together with the secondary feedback responses that occurred in the TgS brain were causally related to BBB disruption. First, the up-regulated ion channels and transporters in the TgS brain likely reflect altered control over ion transport. This could be due to the cholinergic imbalance, which mimics chronic stress conditions. Second, drug penetrance into the brain of FVB/N mice was shown to increase after confined swim stress (10)
; exposure to anticholinesterases, which also induces cholinergic imbalance, impairs BBB functioning (52)
. While this was difficult to reproduce in postmortem perfusion-based studies (13)
, human brain imaging demonstrated an apparent association of stress and AChE overexpression with BBB disruption (12)
. Recent studies further demonstrate cholinergic control over inflammatory reactions (53)
, which constitute an important element in BBB disruption (54)
. Acute stress was reported to increase BBB permeability through activation of brain mast cells (11)
, supporting the notion that cholinergic control is involved.
Third, modified adhesion elements, due to either AChE’s nonenzymatic features or changes induced in gene expression, could change the physical composition of the BBB. Our current findings thus outline a cascade of events initiating with cholinergic imbalance, proceeding with modified gene expression, ion and water transport, and cell adhesion, and culminating in chronic BBB disruption, which in turn promotes progressive neurodeterioration.
Profound impact on PFC gene expression
PFC neurons from TgS brains showed significant differences in their gene expression patterns, compatible with the involvement of the PFC in organismal responses associated with cholinergic imbalance (e.g., psychological stress). TgS and wt PFC microarrays demonstrated >200 transcripts with expression differences. For comparison, cerebral mRNA from transgenic mice overexpressing cyclooxygenase 2, which is involved in neuronal cell cycle activity and glutamate-mediated neurotoxicity, showed altered expression of only one gene (55)
. The more limited variety of PFC neurons could assist in detecting subtle expression differences; nevertheless, that so many genes showed modified expression levels supports the notion that cholinergic imbalance and/or excess of AChE-S exert profound changes on numerous cellular processes. The wt and TgS mice are inbred strains, with minimal inter-animal differences to be expected. However, TgS mice showed larger variability in PFC transcripts than wt mice, suggesting experience-derived contribution in addition to their genotype differences.
Distinct responding cell types and modes of up-regulation
The cell volume regulator channel CLCN3 is expressed in more cells in the TgS than in the wt brain. The water channel AQP4, however, is expressed in higher cellular levels but in a similar number of cells in the TgS and wt mice. In both cases, glial cells showed differences, although the AChE-S transgene is expressed only in neurons (7)
. Therefore, the observed phenomena reflected a tissue level response to the transgenic effects. The modified AQP4 expression patterns were especially intriguing because of AQP4 involvement with focal cerebral ischemia (42)
, brain edema under intoxication or ischemic stroke (43)
, and the integrity of the BBB (44)
. This is compatible with TgS mice being considerably more sensitive to closed head injury than wt mice (5)
, and suggests their impaired BBB might have been involved.
Changes observed in the expression of ion and water channels, transporters, and adhesion-associated proteins could explain the differences in the mobility of water and ions into and out of brain cells and in BBB integrity. These effects are likely to be manifested in hierarchically higher brain organization and/or functioning levels, proposing a putative mechanistic explanation to the noninvasive in vivo brain imaging that revealed changes in water ADCs and BBB permeability.
AQP4 and CLCN3 reflect a more general alteration
Being a water channel, AQP4 is intensively expressed in the brain’s microvascular epithelium; its levels increase in diseases associated with BBB disruption (56)
and its absence protects the mouse brain from damages characteristic of BBB malfunctioning (43)
. However, AQP4 is unlikely to account for the entire TgS phenotype. Rather, multiple small effects of numerous genes, proteins, and their complex interactions are likely involved. Ion channels such as CLCN3 or CLCN4 are induced in neurons and glia alike, probably magnifying neuron-glia interactions that are pivotal elements of BBB regulation (54)
. Adhesion proteins such as selectin, neuropilin, contactin, and PTEN, also known to increase in the ischemic brain (57
58
59
60)
, probably contribute as well. That subtle changes in the brain’s cholinergic balance suffice to disrupt basic transport processes supports the notion that this balance is an essential prerequisite for BBB maintenance.
Molecular and cellular basis of the TgS phenotype
By overexpressing transgenic synaptic AChE-S and host AChE-R, the TgS mouse represents a situation where stress-induced transcriptional activation occurred but did not involve the alternative splicing shift that limits AChE-S excess in the healthy mammalian brain (8)
. This would result in suppressed cholinergic signals into cholinoceptive neurons, unlike normal brain responses where soluble AChE-R limits acetylcholine levels throughout the brain, ameliorating cholinergic excitation of noncholinergic neurons carrying acetylcholine receptors (e.g., GABAergic, glutamatergic, and dopaminergic neurons). Our microarray tests demonstrate that such AChE-S excess induces massive changes in neuronal gene expression, presumably aimed at retrieving cholinergic tone sufficient for survival. These changes would likely facilitate neuronal capacities to respond to remaining acetylcholine molecules but that further impair cell adhesion processes. Together, these suffice to interfere with the maintenance of BBB functioning, which in turn may explain the progressive accumulation of stress-associated neuropathologies in the TgS brain and its consequent cognitive deterioration (5)
.
Cholinergic imbalances likely occur in patients with Alzheimer’s disease or in the brain of PTSD patients; BBB maintenance, which we found to be considerably more vulnerable than previously perceived (10)
, thus protects the mammalian brain from neurodeterioration, but is continuously subject to attacks of environmental origin.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
2 Present address: National Cancer Institute, 41 Library Dr., National Institutes of Health, Bethesda MD 20852, USA.
3 Present address: Department of Neurobiology, Duke University Medical Center, Durham, NC 27710, USA.
Received for publication August 22, 2004. Accepted for publication February 8, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
