| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,1
* Department of Pathology and Laboratory Medicine and
Department of Psychiatry, Boston University School of Medicine, Boston, Massachusetts, USA;
Biological Technologies, Wyeth Research, Cambridge, Massachusetts, USA
1Correspondence: Department of Pathology and Laboratory Medicine, Boston University School of Medicine, 715 Albany St., Rm. L804, Boston, MA 02118 USA. E-mail: jbluszta{at}bu.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: microarray • memory • nutrition • development • pregnancy
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
2
3)
. This dietary compound is necessary to support several physiological functions. Normal growth and function of all cells require choline as the metabolic precursor of phosphatidylcholine – the major structural component of cellular membranes (2
, 4)
. As a constituent of the neurotransmitter acetylcholine (ACh) and of signaling lipids (such as lysophosphatidylcholine, platelet-activating factor, and sphingosylphosphorylcholine), choline is necessary for intercellular communication (2
, 4)
. In addition, after its conversion to betaine, choline becomes a source of methyl groups for enzymatic methylations as the latter compound can be used to synthesize methionine and subsequently S-adenosylmethionine, a methyl group donor for most biological methylation reactions including the methylation of cytidines in CpG dinucleotide sequences of DNA (2
, 4)
. The latter process constitutes the major epigenetic mechanism regulating gene expression (5)
.
The availability of choline for normal development of cognitive function in rodent models is critical. Prenatal choline deficiency during the second half of pregnancy in rats [embryonic days (E) 11 to 17] produced offspring with diminished memory function in adulthood (6)
, whereas prenatal choline supplementation, during the same period of pregnancy was sufficient to improve memory performance throughout life and remarkably prevented age-related memory decline in animals as old as 24–26 mo of age (7)
.
The early events of development such as cell proliferation, migration, and apoptosis ultimately determine the structure and function of the brain. The memory facilitation associated with prenatal choline supplementation was correlated with altered distribution and morphology of septal neurons, suggesting that choline availability is involved in the organizational changes in the basal forebrain during development (8
, 9)
. Choline deficiency has been shown to significantly decrease the rate of mitosis in the progenitor neuroepithelium adjacent to the septum and increase the number of apoptotic cells in the septum (9
, 10)
. Progenitor cell proliferation and apoptosis in the hippocampus of both the mouse and rat were also influenced in a similar manner by dietary availability of choline in utero (9
, 11)
. Thus, the difference in memory due to prenatal choline availability is likely to be mediated by alterations in the birth, migration, and death of cells in the brain during critical periods during development.
The observed structural changes in the brain then translate into functional changes, specifically neurochemical, biochemical, and electrophysiological differences. Prenatal choline intake modulates ACh synthesis and release in basal forebrain cholinergic neurons (12)
, which are known to participate in memory processes (13)
. In addition, prenatal choline availability altered the activation levels of essential molecular components of long-term potentiation (LTP), such that phosphorylation of hippocampal mitogen-activated protein kinase (MAPK) and cAMP response element (CRE) binding protein (CREB) in response to stimulation by glutamate, N-methyl-D-aspartate (NMDA), or depolarizing concentrations of potassium were increased by prenatal choline supplementation and reduced by prenatal choline deficiency (14)
. Concordant changes in LTP induction were also observed. Prenatal choline supplementation enhanced hippocampal LTP in the CA1 region by decreasing the stimulus intensity required for LTP induction (15
, 16)
, possibly as a result of an augmented NMDA receptor-mediated neurotransmission (17)
.
To better understand the mechanisms by which prenatal choline availability affects brain development at a molecular level, we analyzed mRNA expression of samples obtained from prenatally choline-deficient, choline-supplemented, and control rats of various ages during development (P1 to P34 for hippocampus and E16 to P34 for frontal cortex) using the Affymetrix GeneChip arrays. This method revealed a set of genes the expression pattern of which is indeed modulated by prenatal choline intake. Of the 530 hippocampal and 815 cerebral cortical mRNA species that were sensitive to prenatal choline status, we selected a subset of genes that are known to participate in signaling pathways that may mediate the observed changes in LTP and behavior. This group of genes included a growth factor insulin-like growth factor (IGF) II (Igf2); protein kinases PKCß2, CaMKI, and CaMKIIß, a neurotransmitter receptor GABABR1; and a transcription factor Zif268 (EGR1). We verified the results from the microarray analysis using reverse transcriptase-polymerase chain reaction (RT-PCR) and Western blot analysis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 19
, with 7.9 mmol/kg choline chloride substituted for choline bitartrate; Dyets Inc., Bethlehem, PA, USA). From E11 to E17, some dams received a version of the diet that contained 35.6 mmol/kg of choline chloride (supplemented) or a version of the diet completely deficient of choline. After E17, the control diet was given to all dams and offspring. Offspring were weaned on P25.
On E14, 16, and 19, three pregnant rats per group were anesthetized and the uteri were removed. The brain of each fetus was removed rapidly, dissected in ice-cold LI5 medium, and the frontal cortex was taken (without the olfactory bulbs). Three fetuses were taken from each mother per embryonic day for a total of nine fetuses per experimental group. On postnatal days, three individual offspring were used per group (one offspring per litter). On P1, 8, 15, 18, 20, 22, 25, and 34, the offspring were anesthetized and the frontal cortex and hippocampus were removed as above. For E14 through P1 animals, a dissecting microscope was used to aid in the preparation of brain regions. The brain tissues were immediately homogenized in cold 4 M guanidine isothiocyanate solution pH 7.0 (containing 100 mM ß-mercaptoethanol and 25 mM Na-citrate) and placed on dry ice. First, the RNA was extracted using the phenol/chloroform method (20)
and then precipitated once with isopropanol and then with ethanol. The RNA pellet was resuspended in diethyl pyrocarbonate (DEPC)-treated water. The quality of RNA was evaluated with an Agilent Bioanalyzer. A virtual gel was generated using 1 µl of RNA per sample according to the manufacturer’s instructions (Agilent, Santa Clara, CA, USA). RNA concentration was measured with SYBR Green II (Molecular Probes, Eugene, OR, USA) using a FluorImager (Molecular Dynamics, Sunnyvale, CA, USA).
Microarray analysis
For the study of embryonic time points, the cortical RNAs from three embryos from the same mother were pooled. All other RNAs were studied individually. Total RNA (20 µg) was reverse transcribed with Superscript II to generate first strand cDNA using the Superscript Kit (Invitrogen, Carlsbad, CA, USA) substituting the T7 RNA polymerase promoter-containing poly-T primer (T7T24) as described previously (21)
, with the exception that first strand synthesis was carried out at 50°C to prevent mispriming from rRNA. cDNA was purified subsequently using BioMag Carboxyterminated beads (Polysciences, Inc., Warrington, PA, USA). In vitro T7 polymerase transcription reactions for synthesis and biotin labeling of complementary RNA (cRNA), Qiagen RNeasy spin column purification, and cRNA fragmentation were carried out as described previously (21)
. GeneChip hybridization mixtures contained 9 µg fragmented cRNA, 0.5 mg/ml acetylated BSA, and 0.1 mg/ml herring sperm DNA in a total volume of 200 µl 1 X MES buffer. Additionally, 11 similarly labeled bacterial and bacteriophage cRNA sequences were synthesized as described previously (21)
and included in each hybridization mixture at concentrations ranging from 0.5–150 pM to provide a standard curve. The standard curve measurements enabled normalization of oligonucleotide array data and conversion of fluorescent intensity difference averages into mRNA frequency in parts per million (ppm) as described previously (22)
. Reaction mixtures were hybridized for 18 h at 45°C to Affymetrix RG_U34A oligonucleotide arrays. The hybridization mixtures were removed and the arrays were washed, stained with Streptavidin R-phycoerythrin (Molecular Probes) using the GeneChip Fluidics Station 400, and scanned with a Hewlett Packard GeneArray Scanner following the manufacturer instructions. Fluorescence data were collected and converted to gene specific difference averages using MicroArray Suite 4.0 software. Technical quality control variables monitored for these microarrays were (mean±SD): raw Q (measure of background) 1.5 ± 0.35; sensitivity to spiked standard curve: (2.5 ppm±0.6); scaling factor (3.9±4.1); mean signal (174.3±23.0). Quality of mRNA and overall labeling and amplification was balanced across the arrays as assessed by consistent GAPDH 5'/3' ratios of 0.84 ± 0.18.
Data analysis
Probesets were filtered down to those that were called Present by the GCOS MicroArray software in at least one entire experimental cohort (1 dietary treatment on any developmental day n=3 arrays). The signal data were log2-transformed and were fitted as a function of Diet and Time (i.e., age) by a linear two-factor ANOVA model, including a Diet*Time interaction term. With the use of the residual mean square from the ANOVA model, multiple pairwise contrasts were calculated for each probeset, comparing each diet within each time point. The P values for each pairwise contrast of diets were adjusted to false discovery rates (FDRs) using the Benjamini-Hochberg method (23)
. Only probesets with one or more contrasts with P < 0.002 were used for subsequent analysis. This P value cutoff generated 2024 probesets with an FDR = 29% in the hippocampus, and 2779 probesets with an FDR = 18% for the frontal cortex. This set of genes the expression of which was sensitive to the prenatal availability choline was further filtered down by considering only those probesets the mean values of which in any contrast differed by a factor of 1.5 (fold change
1.5). This procedure generated 530 in the hippocampus and 815 in the frontal cortex. Data from each experiment were analyzed with Excel 2004 (Microsoft Corporation, Redmond, WA, USA) and the statistical software packages SYSTAT (SYSTAT Software, Inc., San Jose, CA, USA) and R (www.r-project.org). Eisen cluster analysis (24)
was performed using Cluster 3.0 (Human Genome Center, University of Tokyo) and visualized by Java TreeView Version 1.0.4.
RT-PCR
Hippocampal RNAs that were previously analyzed with microarrays were used for reverse transcriptase PCR using Superscript One-Step RT-PCR with Platinum Taq (Invitrogen, Life Technologies). First strand cDNA synthesis was performed with 50 ng of total RNA (except 100 ng for Igf2), oligo dT primer, and reverse transcriptase at 48°C (45 min). Primers used for PCR include GAPDH (BD Biosciences, San Jose, CA, USA), GABABR1c (forward: AGTGGAGGAAGACCCTAGAG; reverse: ATCATGGTCACAGGAG CAGT), GABABR1d (forward: TGAAACGCAGGACACCATGA; reverse: TCACTTGTAAAGCAAATGTAC), CAMKIIß (forward:AGGCTGTTCTCCATTGT CACCA; reverse:ATACAGGATCACCCCACATGCC), Igf2 (forward:CCAGGT GACAGGACTGGCAC; reverse:CCTGAAAACACCCATCCCAC), PKCß2 (forward: GTTGTGGGCCTGAAGGGGAACG; reverse: TGCCTGGTGAACTCTTTGTCG), Zif268 (forward: GAGCCAAGTCCTTCTAGTCAGTAG; reverse: TGTGAGAGTTAC AGTCGAGCAGTA) and CAMKI (forward: TGGACTGCTGGTCCATAGGAG; reverse: TCCTCATGTGCCGAACCAC). PCR was performed using Platinum TaqDNA polymerase with a denaturing step for 2 min at 94°C followed by 35 cycles (except 40 cycles for Igf2) of 1 min at 94°C, 1 min at 55°C and 2 min at 72°C and terminated by an elongation step at 72°C for 7 min. Under these conditions, the amount of PCR product obtained using brain-derived samples was within the linear range of the assay. PCR products were size fractionated on a 10% polyacrylamide gel and stained with ethidium bromide. PCR products were visualized with a Kodak Image Station 440, and product intensities were quantified using Kodak 10 software. Data from each experiment were analyzed by ANOVA and Tukey test with the statistical software SYSTAT.
Immunoblot analysis
For immunoblot analysis, additional animals were generated as described above. On P18, 20, 25, and 34 the rats were anesthetized and the frontal cortex and hippocampus were removed rapidly as above. Whole tissue extracts were prepared by adding lysis buffer [50 mM Tris pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 10% glycerol, 2 mM 4-(2-aminoethyl)-benzenesulfonyl fluoride, 1 µg/ml leupeptin, 2 µg/ml aprotinin, 2 µg/ml pepstatin] to tissue, followed by gentle sonication, incubation on ice for 15 min, and a brief centrifugation to clear. The extracts were normalized for total protein and subjected to SDS-PAGE. After being transferred to an Immobilon P membrane (Millipore, Billerica, MA, USA), the membrane was blocked with 5% nonfat dry milk in 1x TBS containing 0.1% Tween 20 for 1 h and then was probed with primary antibody overnight. The antibodies used included a monoclonal anti-ß-actin antibody (Sigma, St. Louis, MO, USA), and polyclonal antibodies against Igf2 (Upstate Biotechnology, Lake Placid, NY, USA), CaMKI, CaMKIIß, PKCßII, GABABR1, and Zif268/EGR1 (Santa Cruz Biotechnology, Santa Cruz, CA, USA).
The antibody/antigen complexes on the blots were detected with either anti-rabbit or anti-mouse IgG peroxidase conjugates and were visualized using the chemiluminescence method (Western Lightning, Perkin Elmer, Wellesley, MA, USA) and a Kodak Image Station 440. Band intensities were quantified using Kodak 10 software. The membranes were stripped in a solution containing 62.5 mM Tris-HCl pH 6.7, 2% SDS, and 100 mM ß-mercaptoethanol for 30 min. After incubation with 5% nonfat dry milk in 1 X TBS containing 0.1% Tween 20 for 1 h at room temperature, membranes were reprobed with a primary antibody as described previously. Protein levels were normalized with ß-actin values. Data from each experiment were analyzed by ANOVA and Tuke test with the statistical software SYSTAT.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
on the entire data set consisting of 9 arrays (3 animals per group) per each day tested was performed and an example for P18 is shown in Fig. 1
A. The arrays from the prenatally choline-deficient animals were clearly differentiated from the choline-supplemented and control animals. Cluster analysis on the entire P18 data set did not distinguish the control and the prenatally choline-supplemented groups from each other. However, when the subset of 303 hippocampal genes (Fig. 1B
) that were sensitive to prenatal choline status were subjected to Eisen cluster analysis, the distinction between the control and choline-supplemented arrays was evident (Fig. 1B
).
|
Although the Eisen cluster analysis was sensitive enough to distinguish the three dietary treatment groups from one another, the levels of mRNAs for the overwhelming majority of expressed genes were not affected by prenatal choline status. To illustrate the consistency of the data acquired from the microarray experiments, we selected six genes as representatives of the majority of those that were unaffected by prenatal choline intake (Supplemental Fig. 1). This group of genes included aldolase A, ribosomal protein L13, heat shock protein 60 (hsp60), laminin receptor 1, growth-associated protein (GAP) 43 (GAP43), and ubiquitin C. The expression of these genes was consistent among the three groups of animals, even when the expression was developmentally regulated. The fact that these genes encode proteins of different structures and function and that their expression is distinctly regulated during development, yet unaltered by choline availability, supports the validity of any choline-evoked changes observed in the expression of other genes. Although the expression of most genes was unaffected, the expression of 
530 genes was sensitive to prenatal choline deficiency, or supplementation, or both. The alterations in gene expression levels evoked by prenatal choline intake were not drastic and, in most cases, transient. The greatest differences were observed between the prenatally choline-deficient and choline-supplemented rats. The control animals tended to have gene expression levels that either matched that of one of the other two groups or that fell in between the levels of the other two groups, confirming that choline deficiency and supplementation are, in fact, the extremes. In no case was the gene expression measured from choline-deficient and choline-supplemented groups at a similar level while the control group differed. We were interested in the role of signaling molecules that may mediate the observed changes in behavior; therefore, the genes that are known to be involved in signaling pathways associated with learning and memory were selected for further validation and study. These genes included CAMKIIß, PKCß2, and GABABR1. The period from P15 to P20 appears to be a key point in hippocampal development when the effects of prenatal choline availability can clearly be observed. This period correlates with the earliest time at which behavioral differences have been found (14)
.
CaMKII is a multifunctional kinase that phosphorylates a diverse group of proteins such as ion channels, transcription factors, synapsin, tubulin, and mictrotubule associated proteins (25
26
27
28)
. CAMKII
and CAMKIIß are the two predominant isoforms found in the central nervous system (CNS) and are primarily located in neurons (29)
. CAMKII has been shown to mediate LTP and facilitate spatial learning and memory (30
31
32
33
34
35
36
37)
, whereas CAMKIIß plays a role in synaptic vesicle binding and neurotransmitter release (38)
. Overexpression of CAMKIIß led to an increase in dendritic arborization, whereas reduced CAMKIIß levels [evoked by RNA interference (RNAi)] decreased dendritic arborization and synapse formation (39)
. CaMKIIß gene expression was fourfold higher in the prenatally choline-deficient rats in comparison to the choline-supplemented group on P18 as determined by microarray analysis (Fig. 2
A). RT-PCR produced similar results for CaMKIIß mRNA levels on P18 (Fig. 2B
). The difference in mRNA abundance between the groups of rats correlated to a similar but less drastic (1.7-fold), difference in protein levels of CAMKIIß (Fig. 2C
).
|
The protein kinase C family of enzymes has been implicated in synaptic plasticity and memory in a wide range of animal species. The activation PKCß2, an isoform that is translated from an alternatively spliced variant of the PKCß gene (40)
, was induced by oxidative stress (41)
and ischemia (42)
. PKCß knockout mice exhibited a loss of learning, specifically deficits in both cued and contextual fear conditioning (43)
. On P18, prenatal choline-deficient animals had the highest level of PKCß2 mRNA product by RT-PCR and protein by Western blot analysis, confirming the results obtained by the microarray analysis (Fig. 3
). Increased levels of PKCß2 in prenatally choline-deficient rats may help to compensate for deficits in other aspects of signaling that are required for learning.
|
GABA is the main inhibitory neurotransmitter in the mammalian CNS, where it exerts its effects through ionotropic (GABAA/C) receptors to produce fast synaptic inhibition, and metabotropic (GABAB) receptors to produce slow, prolonged inhibitory signals. Activation of GABAB receptors by the agonist baclofen inhibits the release of ACh from cholinergic neurons in the striatum and hippocampus, thus implicating the GABAB receptor in regulating ACh release (44
, 45)
. GABABR1 has several isoforms that are generated by alternative splicing of its mRNA, and the U34A microarray has probes specific to the c and d isoforms. The expression of GABABR1c and GABABR1d showed a similar pattern of expression with a transient spike seen only in prenatally choline-deficient rats on P18 and P20 (Fig. 4
A, B). Therefore, we designed primers for RT-PCR analysis that could distinguish the c and d isoforms from the b isoform to verify the microarray results (Fig. 4C
). RT-PCR revealed that levels of the c and d isoforms of GABABR1, but not the b isoform, were 2-fold higher in the prenatally choline-deficient animals at P18 (Fig. 4D, E, F
). Not surprisingly, an increase in GABABR1 protein level was not observed (Fig. 4G
), since the antibody used for the immunoblots was not isoform specific and our RT-PCR analysis indicated that the b isoform was by far the most abundant (data not shown) and its mRNA levels were not affected by prenatal choline availability.
|
Frontal cortex
We similarly analyzed mRNA expression of cerebral cortical samples obtained from rats at E16, E19, P1, P8, P15, P18, P22, and P34. Again, we selected four genes to be representative of the majority of those not affected by prenatal choline availability. This set (see Supplemental Fig. 2) included aldolase A, ATPase Na+/K+ ß1, heat shock protein 60 (hsp60), and ubiquitin carboxy-terminal hydrolase L1. Some of these genes also appeared in the hippocampus among those that were not influenced by choline availability. Most of these genes were developmentally regulated, but no differences in expression levels were seen between the prenatally choline-deficient, choline-supplemented and control rats. The majority of changes in gene expression due to prenatal choline status was transient and occurred during the period immediately following birth to 2 wk of age. As described previously for the hippocampal genes, the expression of 880 genes in the frontal cortex was determined to be responsive to prenatal choline availability, and again, the signaling molecules, which may play a role in learning and memory, were selected to study further. These genes included CAMKI, Zif268, and Igf2.
Although CaMKII has been implicated in learning and memory, the biological role of CaMKI has been primarily associated with promotion of neurite extension and growth cone mobility (46)
. More recently, CAMKI was shown to be essential for full expression of LTP and MAPK-dependent translation activation in the hippocampus (47)
. Interestingly, CaMKI mRNA level was transiently higher in the prenatally choline-supplemented rats compared to the prenatally choline-deficient rats over the period of P8-P18 (Fig. 5
A). RT-PCR and immunoblot analysis performed on RNA and protein lysates, respectively, from P18 animals confirmed the microarray results (Fig. 5B, C
). Increased levels of CAMKI during this period in development may lead to enhanced neurite extension in the prenatally choline-supplemented animals.
|
Induction of LTP in the dentate gyrus (48
, 49)
causes a transient activation of immediate early genes that are a critical step in the subsequent cascade of up-regulation of expression of multiple genes. Zif268 is an immediate early gene, encoding a zinc finger transcription factor (50
51
52)
. Mice with a targeted inactivation of the Zif268 gene are deficient in the expression of long-term synaptic plasticity and of long-term memory (53)
. Zif268 mRNA is up-regulated in the hippocampus of the behaving rat after exposure to a novel stimulus (54)
and in the inferior temporal gyrus of the monkey during associative learning (55)
. Zif268 is also believed to play a critical role in memory consolidation (for a review see ref 56
). Prenatally choline-supplemented rats had a higher level of Zif268 mRNA in the frontal cortex than choline-deficient and control rats on both P18 and P22 (Fig. 6
A). Consistent with the microarray analysis, the level of Zif268 product from RT-PCR was 2.3-fold higher in the cortex of prenatally choline-supplemented animals as compared to choline-deficient animals on P18 (Fig. 6B
). Immunoblot analysis of P18 cortical lysates supported these data and showed that prenatally choline-supplemented animals had an 4-fold higher level of Zif268 protein than did choline-deficient animals (Fig. 6C
). Although the level of Zif268 mRNA expression in choline-supplemented rats fell to the level of choline-deficient rats on P25, the Zif268 protein level remained 1.4-fold higher in choline-supplemented animals on that day, suggesting that the change may not be transient (data not shown). Moreover, in the P18 hippocampus, Zif268 mRNA abundance and protein level were also significantly higher in prenatally choline-supplemented rats compared to prenatally choline-deficient animals and Zif268 protein level was significantly higher in prenatally choline-supplemented animals as compared to control animals (Fig. 6D, E
). The increased level of Zif268 mRNA and protein in the cortex and hippocampus of prenatally choline-supplemented animals is consistent with the improved memory capability of these animals.
|
Igf2 belongs to a family of growth factors that also includes insulin and IGF I (Igf1). Igf2 mediates growth in the earliest stages of embryonic development (57
, 58)
and may participate in regulating the formation of the anterior pituitary as its expression is highly regulated during the pre- (59)
and postnatal (60)
development of the gland. Igf2 also promotes the differentiation of the basal forebrain cholinergic neurons in culture (61)
and increases ACh release from the hippocampus (62)
. Cortical expression of Igf2 mRNA was increased by >2-fold on P34 in the prenatally choline-supplemented rats as compared to the two other groups (Fig. 7
A). Consistent with the microarray analysis, the level of Igf2 product from RT-PCR was 2.3-fold higher in the cortex of prenatally choline-supplemented animals as compared to choline-deficient rats (Fig. 7B
). The differences in Igf2 mRNA expression among the groups of rats were also observed at the protein level on P25 (Fig. 7C
). The Igf2 protein level was 2.5-fold higher in prenatally choline-supplemented rats than in control rats and 9-fold higher than in choline-deficient rats.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, and by our neurochemical studies showing that phosphorylation, and therefore activation, of hippocampal MAPK was similarly enhanced in these animals at P18 but not at P25 – an age at which the controls caught up (14)
.
Up-regulation of expression of Zif268, CAMKI, and Igf2 in brains in prenatally choline-supplemented rats is consistent with the known functions of these proteins in LTP, in modulation of ACh release and in the processes of learning and memory (see above), as well as with improved memory performance observed in these animals. In contrast, reduced expression of CaMKIIß and PKCß2 in the hippocampus of prenatally choline-supplemented rats and concomitant up-regulation of expression of these enzymes in prenatally choline-deficient subjects are at odds with the current understanding of the role that they play in learning and memory. These data are reminiscent, however, of the previously reported up-regulation, in the hippocampus of prenatally choline-deficient rats, of high-affinity choline transport (14)
— a process that provides the choline precursor for ACh synthesis. It is also worth noting that, while prenatal choline deficiency is associated with impaired LTP and memory deficits in relatively difficult tasks, prenatally choline-deficient animals also show somewhat improved performance in simple memory tasks relative to controls (63)
. Thus, both prenatal choline-deficiency and supplementation cause complex molecular reorganization of the brain that correlates with some aspects of altered cognitive function.
In a study designed to determine the effects of choline on global gene expression in mouse neural cortical precursor cells in vitro, Niculescu et al. (64)
found that among 1000 genes that were sensitive to choline 80% were overexpressed when the cells were grown in a medium containing low (5 µM) as compared to high (70 µM) concentrations of choline after a 2 day period. It is noteworthy that in our study performed in vivo a majority (70%) of the hippocampal genes affected by prenatal choline availability showed the highest levels of expression in prenatally choline-deficient rats even weeks after the period of altered availability of choline had ended (data not shown). One possible mechanism for these effects may be related to the function of choline as a donor of methyl groups ultimately used for DNA methylation. Each time cells undergo DNA replication, the daughter strand of the newly formed DNA needs to be methylated so that the methylation patterns of the parent strand are copied. In this fashion the changes in DNA methylation patterns established when choline availability is altered may persist through multiple cell cycles (65)
. In fact, animals fed diets deficient in methyl donors (choline and methionine) had hypomethylated DNA (66
67
68
69)
. In general, DNA methylation within the regulatory regions of many genes causes transcriptional repression (70
71
72
73
74)
. Indeed, in choline-deficient mouse fetuses, global and gene-specific DNA methylation was decreased in the ventricular and subventricular zones of hippocampal Ammon’s horn and this correlated with the induction of expression of Cdkn3, a gene known to be silenced by DNA methylation (67)
. However, DNA methylation may also up-regulate transcription by preventing the binding of a transcriptional repressor to a genomic silencer element, as in the case of the Igf2 gene (75
, 76)
. Our data are consistent with this model and show that on P34 cortical expression of Igf2 mRNA is increased by >2-fold in the prenatally choline-supplemented rats as compared to the two other groups. The increased expression of Igf2 by prenatal choline supplementation is particularly intriguing because the Igf2 protein has been shown to potentiate ACh release in rat hippocampal slices (62)
, and previous studies in this laboratory demonstrated that depolarization-evoked ACh release was significantly increased in hippocampal slices from prenatally choline-supplemented rats as compared to controls and prenatally choline-deficient animals (12)
. It is tempting to propose that the up-regulated Igf2 expression in prenatally choline-supplemented rats may prove to be the mediator of the enhanced ACh release in these animals.
Previous studies suggested that the GABAergic system is modified by dietary choline intake, both prenatally (77)
and postnatally (78)
. For example, increased levels of the calcium binding protein calretinin, which can be used as a marker of the onset of GABAergic neuronal differentiation in developing fetal brain, were found in the primordial dentate gyrus of prenatally choline-deficient fetuses on E17 as compared to control fetuses. A reduced level of calretinin protein was detected in prenatally choline-supplemented fetuses on E17 as well as at 24 mo of age (79)
. Microarray analysis revealed differences in the mRNA abundance of GABABR1 between the three groups of rats. Two probe sets on the Affymetrix U34A chip that were specific for the GABABR1 c and d isoforms showed a significant increase in the mRNA abundance in the prenatally choline-deficient animals as compared to choline-supplemented animals. Using primers that produced isoform-specific PCR products that varied in size (Fig. 5)
, we were able to verify that prenatally choline-deficient rats had higher levels of mRNA for the c and d isoforms of GABABR1 but not the b isoform. There are six known isoforms of GABABR1, each of which is generated by alternative splicing (80
81
82
83
84
85
86)
. The a and b GABABR1 isoforms are the two most abundant and are differentially expressed in various brain regions during development (82
, 87
88
89)
. GABABR1c has a 93-bp insertion (exon 18) that generates an additional 31-amino acid sequence in the fifth transmembrane region of GABABR1b, which may introduce structural variation in the second extracellular loop and fifth transmembrane domain (80
; Fig. 5C
). GABABR1d also has an amino acid sequence initially identical to GABABR1b but with an additional insertion of 566 bp between exon 21 and exon 22 that generates a divergent amino acid sequence in the carboxyl-terminal end (80
; Fig. 5C
). GABABR1c and d are expressed in the CNS as well as various peripheral tissues. As of yet, a functional analysis has not been performed on GABABR1c or d. Our results suggest that prenatal choline availability may differentially influence the expression of the GABABR1 gene. Due to the lack of an antibody with isoform specificity, we were unable to measure differences in protein levels. Although this is the first time to our knowledge that choline has been shown to affect alternative splicing, other studies have determined that either modulation of diet or exposure to ethanol may influence splicing of various genes. For example, alterations in nutritional status have been reported to cause differential regulation of Igf1 mRNAs by a post-transcriptional mechanism, such as nuclear splicing and/or RNA degradation, which attenuates translation of Igf1 mRNAs (90)
. Starvation and refeeding have been shown to regulate splicing of glucose-6-phosphate dehydrogenase (91)
. Ethanol and/or estradiol treatment of primary cultures of pituitary cells regulates alternative splicing of the dopamine D2 receptor (92)
. When measured at P21, expression levels of splice variants of NMDA receptor subunit 1 (NR1) and subunit 2 (NR2) were modified by prenatal ethanol exposure (93)
. Chronic ethanol treatment has been shown to reduce the amount of the exon-5 containing splice variants of NR1 mRNA in mouse fetal cortical neurons (94)
. In rats, chronic intermittent ethanol exposure altered GABAA receptor subunit and splice variant expression (95)
. In general, alternative splicing of genes either serves as a regulatory device to modulate levels of gene expression or as a way to produce variant protein isoforms with differing functions. Because no information is available on the specific functions of the b and c isoforms of GABABR1, we do not know how changes in the levels of these proteins evoked by prenatal availability of choline may modulate GABAergic neurotransmission.
Brain tissue is highly heterogeneous, composed of many cell types, and thus it remains to be determined if the changes in mRNA and protein levels reported here are due to alterations in the abundance of these species per cell (i.e., true changes in gene expression) or to differences in the tissue abundance of cell types that express the transcripts and their protein products. In either case, our data show that the supply of choline during the prenatal period modulates the molecular organization of the hippocampus and cerebral cortex and, moreover, that the pattern of expression of key genes encoding proteins known to participate in signal transduction pathways involved in learning and memory are subject to long-lasting modulation by the availability of a single nutrient during critical periods in development.
Previous studies in rats showed that pregnancy reduces hepatic choline pools (96)
and that choline supplementation prevents these losses (97)
. Moreover, supplemental choline is readily available to the fetuses, raising concentrations of choline-containing compounds in their brain (97)
. These data are consistent with the notion that the behavioral, cellular, and molecular modifications observed in the offspring of rat dams consuming diets with varying choline content are due to the changes in the fetal brain choline levels. However, it is also possible that some of these effects are mediated indirectly via changes in maternal metabolism and/or health. Importantly, the current dietary guidelines in the United States call for increased daily intake of choline during pregnancy (475 mg) and nursing (550 mg), as compared to other times in women’s life (425 mg) (98)
, to ensure the maintenance of good health of both the mother and baby.
|
ACKNOWLEDGMENTS |
|---|
Received for publication July 6, 2006. Accepted for publication December 6, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
