|Home | About | Journals | Submit | Contact Us | Français|
Maternal intrauterine inflammation is implicated in neurodevelopmental disorders in the offspring. Serotonin is crucial for regulating maturation in the developing brain, and maternal inflammation may result in disruption of the serotonergic system in the perinatal period. Saline or endotoxin was injected intrauterine in pregnant rabbits term. Newborn rabbits underwent positron emission tomography (PET) imaging with α[11C]methyl--tryptophan (AMT) to evaluate tryptophan metabolism in vivo. Decrease in standard uptake value for AMT and decrease in serotonin concentration was noted in the frontal and parietal cortices of endotoxin kits when compared with controls. In addition, a significant decrease in serotonin-immunoreactive fibers and decreased expression of serotonin transporter (5HTT) was measured in the somatosensory cortex. There was a three-fold increase in the number of apoptotic cells in the ventrobasal (VB) thalamus without loss of raphe serotonergic cell bodies in endotoxin kits when compared with controls. Glutamateric VB neurons projecting to somatosensory cortex transiently express 5HTT and store serotonin, regulating development of the somatosensory cortex. Intrauterine inflammation results in alterations in cortical serotonin and disruption of serotonin-regulated thalamocortical development in the newborn brain. This may be a common link in neurodevelopmental disorders resulting in impairment of the somatosensory system, such as cerebral palsy and autism.
Maternal intrauterine infection and inflammation is a risk factor for the development of cerebral palsy in the neonate (Leviton and Gilles, 1979; Yoon et al, 2000). The term cerebral palsy encompasses a group of motor-impairment syndromes that occurs due to injury to the developing brain, which may be associated with deficits in somatosensory perception and/or alterations in the normal development of the somatosensory system (Bax et al, 2005; Wingert et al, 2008). Infection-induced maternal immune activation leads to a fetal inflammatory response mediated by cytokines that has been implicated in the development of not only periventricular leukomalacia and cerebral palsy, but also neurodevelopmental disorders such as autism and schizophrenia (Patterson, 2009; Vargas et al, 2005; Yoon et al, 2000). The timing of the immune challenge with respect to the gestational age and neurologic development of the fetus may be crucial in the response elicited in the neonate (Meyer et al, 2006).
Serotonin has a critical role in the development of the fetus and the neonate influencing neurogenesis, neuronal migration, and synaptogenesis in the brain (Lauder and Krebs, 1978; Mazer et al, 1997). Disruption of the serotonergic system by either pharmacological or genetic manipulations in the neonatal period in rodents results in delayed development of cortical layers, disruption of thalamocortical afferents patterns in the barrel fields, and a decrease or disorganization of the barrel fields (Blue et al, 1991; Bennett-Clarke et al, 1994; Cases et al, 1996; Persico et al, 2001).
Serotonin is synthesized from the precursor tryptophan, which is very tightly regulated at the fetal/maternal interface in the placenta. Tryptophan is transported competitively at the placenta via the large neutral amino-acid carrier (LAT1) (Kudo and Boyd, 2002) and is catalyzed along the kynurenine pathway by tryptophan 2,3-dioxygenase and indoleamine 2,3-dioxygenase (IDO), both of which are expressed in the placenta and have a vital role in the prevention of allogeneic rejection of the fetus (Munn et al, 1998) and regulation of fetoplacental blood flow in late gestation (Ligam et al, 2005). As IDO is induced by interferon-γ, it is upregulated in the placenta with maternal infection (Mackler et al, 2003). Intrauterine infections in women are associated with upregulation of kynurenine pathway enzyme mRNA expression in the placenta (Manuelpillai et al, 2005). Increased metabolism of tryptophan by the kynurenine pathway at the placenta may result in abnormalities in fetal brain development because of decreased tryptophan available for serotonin synthesis and the neurotoxic effects of kynurenine metabolites.
We have previously demonstrated that intrauterine endotoxin administration in the rabbit results in motor deficits resembling cerebral palsy and microglial activation in the newborn (Saadani-Makki et al, 2008). In this study, we evaluated the effect of intrauterine endotoxin administration on postnatal brain tryptophan metabolism in vivo using positron emission tomography (PET), with the tracer α[11C]methyl--tryptophan (AMT) (Chugani et al, 1997, 1999), and in vitro by measuring serotonin content, serotonin immunocytochemistry, tryptophan hydroxylase (TPH2) immunohistochemistry, serotonin transporter (5HTT) expression, and terminal deoxynucleotidyl transferase-mediated 2'-deoxyuridine 5'-triphosphate-biotin nick end labeling staining. We hypothesized that exposure to inflammation in utero would result in alterations in tryptophan metabolism and development of the serotonergic system in the somatosensory cortex of the postnatal brain.
All animal procedures were approved by the institutional animal care and use committee of Wayne State University. New Zealand white rabbits with timed pregnancies were obtained from Covance Inc (Covance Research Products, Inc, Kalamazoo, MI, USA). The pregnant rabbits were divided into three groups: (1) control-saline (n=4), (2) endotoxin (n=5), and (3) control-no intervention (n=4). Pregnant rabbits in the control saline and endotoxin groups underwent laparotomy at gestational day 28 (term pregnancy 31 to 32 days) and were injected with 1mL of saline or 1mL saline containing 20μg/kg of Escherichia coli endotoxin (E. coli serotype O127:B8; Sigma Aldrich, St Louis, MO, USA) along the length of the uterus as previously described (Kannan et al, 2007). The control-no intervention group included pregnant rabbits that had no surgery or intervention and was added to determine the effect of the stress of surgery or exposure to anesthetic agents in utero. All kits were born spontaneously on gestational day 31 and the litter size ranged from 7 to 11 kits. Kits born to dams exposed to endotoxin in utero were born with motor deficits suggestive of cerebral palsy as previously described (Saadani-Makki et al, 2008).
Newborn rabbits underwent PET scanning on day 1 of life (within the first 24hours after birth). The PET scans were performed using the microPET R4 tomograph (Siemens Preclinical Solutions, Knoxville, TN, USA). The rabbit kits were positioned in an in-house developed head and body holder as previously described (Kannan et al, 2007). In short, three fixed spheres attached to the holder filled with fluid that was visible both on PET (radioactivity) and MR (water) images were used for coregistration of the two modalities. Following anesthesia with 0.1% to 0.2% isoflurane, the rabbit kits were placed in prone position on the microPET bed with the head in the center of the field of view. After an initial 17minutes transmission scan with a rotating Ge-68 point source to correct for attenuation of the 511keV photons, the kits were injected with 10 to 20MBq of the tracer [11C]AMT (specific activity at the time of injection was 11 to 15GBq/μmol, as described previously by Chakraborty et al (1996)) intravenously in a peripheral vein, and a 60-minute list mode data acquisition in 3D mode was initiated. All scans were started at ~5minutes after tracer injection. The list mode data were subsequently rebinned into discrete time frames (6 × 10minutes) and attenuation corrected sinograms were reconstructed using the ordered subset maximization expectation (OSEM) iterative algorithm, yielding an isotropic image resolution of about 2mm full width at half maximum. After completion of the PET data acquisition, each animal underwent magnetic resonance imaging scanning for coregistration and high-resolution anatomical image volumes acquired with a T1-weighted Fast Gradient Echo sequence using a 4.7-T Brucker magnet and a head coil with the animal in the head holder as previously described (Kannan et al, 2007).
The images were subsequently processed using the AMIDE software (A Medical Image Data Examiner, version 0.9.2). The MR and microPET image volumes were coregistered by manually matching the position of the three fiducial markers in both data sets. Following coregistration, 3D regions of interest (ROIs), the first involving the periventricular region (PVR), including the area around the ventricles, hippocampus, corpus callosum, and thalamus, and the second involving the cortex (frontal cortex and parietal cortex) were defined in the MR image volumes and copied to the dynamic image sequences, yielding dynamic time activity curves for both the control groups and the endotoxin group. The ROIs were drawn on 11 planes of 0.8mm each, starting from the beginning of the lateral ventricle to the posterior-most part of the dorsoventral hippocampus (three planes anterior to bregma and eight planes from the level of the bregma; Figure 1A). The standard uptake value (SUV) was calculated for each time point by dividing the mean tracer concentration (MBq/cm3) by the injected activity (MBq) per weight (g), and the resulting mean SUVs were plotted over time for all ROIs for each group. As the activity appeared to plateau between 20 and 40minutes of the scan, this time point was taken as representative of the maximum tryptophan metabolism and SUVs characterizing the normalized tracer uptake during this period were compared among the groups. A third ROI comprising the whole brain in these planes was also drawn, and the ratio of the SUV in the PVR to the SUV in the whole brain was compared for all three groups.
All rabbit kits were euthanized on day 1 of life following the PET scan and perfused with 4% paraformaldehyde. The brains were postfixed, cryoprotected and 30μm sections were cut and mounted onto poly--lysine-coated slides (Sigma Aldrich, St Louis, MO, USA). The sections were blocked with 5% goat serum followed by incubation with the primary antibodies rat anti-serotonin (1:50; Millipore, Billerica, MA, USA) for 48hours or rabbit polyclonal anti-TPH2 (1:50; Thermo Scientific, Rockford, IL, USA) for 24hours at 4°C. This was followed by incubation with biotinylated secondary goat anti-rat IgG or anti-rabbit IgG (1:200; Vector, Burlingame, CA, USA) for 1hour at room temperature and Avidin biotin complex for 1hour. Color was developed using 3,3′-diaminobenzidine (Sigma Aldrich).
A commercially available competitive ELISA Kit from Immuno-Biological Laboratories, Inc (IBL-America, Minneapolis, MN, USA) was used to determine the levels of serotonin in (1) the cortex and (2) the hippocampus, of day 1 rabbit kits. The frontal and parietal cortices, and hippocampus were dissected from frozen newborn brains, homogenized, and 40mg protein was extracted from each of the lysates. Extracted protein was dissolved in diluent containing 0.1% (w/v) ascorbic acid, and normalized to 200μg/mL for each sample. The assays were performed as per the manufacturer's instructions. The quantity of serotonin in brain homogenates was determined using a standard curve containing known amounts of serotonin (N=5 kits for control-no intervention group; 6 kits for control-saline group, and 7 kits for the endotoxin group).
To obtain an unbiased estimate of the total number of serotonergic neurons in the raphe nucleus, an optical fractionator probe (Stereo Investigator, MicroBrightfield Inc, Williston, VT, USA) was used (Gundersen et al, 1988). The cells stained with anti-serotonin antibody were visualized under BH2 Olympus light microscope (Olympus Optical Co. GmbH, Hamburg, Germany) attached to a motorized stage and a stage encoder. The boundary of the dorsal raphe in each section was defined at × 4 magnification by drawing a contour. The contour included the majority of serotonin-stained neurons, whereas diffusely stained neurons outside the boundary were excluded from the count. The sections were further visualized under × 40 with a sampling grid (area sampling fraction (asf)=150 × 150μm2), and systematic uniform random sampling (X=315.2 and Y=353.9). The average section thickness ranged between 18 and 20μm, and the guard zone was calculated as 2μm with dissector height=14 or 16μm. The total number of cells was determined by the formula
where ΣQ− is the total number of neurons counted, t is the mean section thickness, h is the height of the optical dissector, asf is the area sampling fraction, and ssf is the section sampling fraction. The coefficient of error for each animal ranged from 0.07 to 0.1.
Spherical probes were used to estimate the total length of serotonin-immunoreactive fibers in the somatosensory cortex (Spherical probe, Stereo Investigator, MBF Bioscience, Williston, VT, USA). The somatosensory cortex including the cortical plate and layers V and VI in the parietal area 1 was traced under × 4 magnification. The lateral edge of the region was defined by the lateral-most border of the CA3 region of the hippocampus, and the medial edge was at the level of the mid-hippocampus (as illustrated in Figure 3A). Three 30μm coronal sections that were 240μm apart were chosen for each kit at the level of dorsal hippocampus (n=4 rabbit kits from four litters for the control-saline and endotoxin groups and from three litters for the control-no intervention group). The minimum average thickness of the sections postprocessing was measured to be 20μm and an XY grid of 250 × 250μm2 was used. For all the samples, the virtual sphere was maintained at a radius of 10μm. The software automatically generates virtual spheres within each specified grid, which could be visualized as a series of circles of changing circumferences on focusing through the tissue (in the z direction). Every fiber intersecting the virtual sphere was marked when the tissue was moved in z direction. Following stereological principles of systematic random sampling, the number of intersections of serotonin-immunoreactive fibers with each circumference, indicating the intersection of the focal plane and the virtual sphere, was counted. The total length of fibers (L) was calculated according to the equation
where ΣQi is the number of intersections counted, ν is the volume of the sampling box (1,250,000μm3), a is the surface area of the probe (628.319μm2), and ssf is the section sampling fraction (1/8). The length of the serotonin-immunoreactive fibers along an area of 1μm2 in the somatosensory cortex was calculated by dividing the total length (L) by the total area measured.
Total RNA from brain tissue (frontal and parietal cortex) was purified by using AllPrep DNA/RNA/Protein Mini Kit (Qiagen Inc, Valencia, CA, USA) as per the manufacturer's instructions (n=4 kits each for control-no intervention and control-saline groups and 5 kits for endotoxin group). The RNA samples were quantified using the Nanodrop ND-1000 Spectrophotometer and integrity of RNA was verified using the Agilent 2100 Bioanalyzer (Foster City, CA, USA) with the Eukaryote Total RNA Nano assay. Real-time reverse transcription polymerase chain reactions (RT-PCR) were conducted using Applied Biosystems (ABS, Carlsbad, CA, USA) Sequence Detection System 7000 (Applied Biosystems). Primer Express software (ABS) was used to design specific rabbit primers for 5HTT, and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) that served as control. The primer sequences are: for 5HTT, forward primer: 5′-TTTCCGGATTCGTCATCTTC-3′, and reverse primer, 5′-CAAGCCCAGTGTGATTAGCA-3′ GAPDH, forward primer, 5′-CCTACCCCAATGTATCCGTTGTG-3′, and reverse primer, 5′-GGAGGAATGGGAGTTGCTGTTGAA-3′. A two-step RT-PCR reaction was performed using SYBR Green PCR Master Mix (ABS). Single-stranded complementary DNA was reverse transcribed from the total RNA samples by using the high-capacity complementary DNA Reverse Transcription Kit with RNase inhibitor (Applied Biosystems) followed by the PCR amplification with the complementary DNA as template, forward and reverse primers of the genes, and the SYBR green master PCR mix. Amplification conditions included 30minutes at 48°C, 10minutes at 95°C, 40 cycles at 95°C for 15seconds, and 60°C for 1minute. Samples were quantified using the δ CT (threshold cycle, amount of target=2−ΔΔCT) method.
To determine whether there was an increase in the number of apoptotic cells in the ventrobasal (VB) thalamus, coronal sections at the level of the dorsoventral hippocampus from the same brains that were stained for serotonin immunoreactivity in the control-saline and endotoxin groups, were processed according to the manufacturer's instruction in the terminal deoxynucleotidyl transferase-mediated 2′-deoxyuridine 5′-triphosphate-biotin nick end labeling-based detection kit (NeuroTACS In situ Apotosis Detection kit staining; R&D Systems, Minneapolis, MN, USA).
For each kit, three coronal sections, each section 90μm apart at the level of nucleus ventral basalis thalami (as indicated in Figure 7A) was quantified for apoptotic cells in the control saline and endotoxin group. For each section, four clearly separated fields at × 40 magnification of the nucleus ventral basalis thalami were examined, and the number of apoptotic cells and the total number of cells were counted by a blinded investigator. Apoptotic cells were defined based on morphology and the presence of nuclear fragmentation.
Before the analyses, the distributions of all outcomes were examined, and all data were checked for accuracy, and for outliers. Because of the nesting of kits within litters and the repeated measurements for each kit, generalized estimating equation was used to compare the SUV for AMT in the PVR and the cortex, ratio of SUV in PVR to whole brain serotonin content in the cortex and hippocampus, the ratio of apoptotic cells to the total number of cells counted, and the fiber length between the groups. When reported, the means and 95% confidence intervals (CIs) or s.d. for each were obtained from the generalized estimating equation analysis. T-tests were used to compare the means for the serotonergic cell counts in the raphe nucleus between the control-no intervention and the endotoxin groups, and a one-way analysis of variance was used to compare 5HTT expression between the groups as there was a single kit from each of the litters (no nesting of kits within litters) for these two measures.
A total of 14 kits were scanned with AMT on day 1 of life. There were five kits from five litters in the endotoxin group, five kits from four litters in the control-PBS group, and four kits from three litters in the control-no intervention group. The SUV of AMT, which is the activity in the ROI normalized for the weight of the animal and dose injected, was calculated for the PVR and for a region including frontal and parietal cortices. Overall, there was a significant difference in cortical SUVs between the three groups (χ2=45.93, df=2, P<0.001) (Figure 1B). Cortical SUVs for the endotoxin group (mean=0.35; 95% CI=0.29 to 0.42) were found to be significantly lower than both the control-saline group (0.56; 95% CI=0.52 to 0.60) and the control-no intervention group (0.65; 95% CI=0.59 to 0.71) (χ2=28.44, df=1, P<0.001 and χ2=43.24, df=1, P<0.001 for endotoxin versus both control groups, respectively). Interestingly, we also determined a significant difference between SUVs derived from the two control groups (χ2=5.79, df=1, P=0.016), which may be due to stress of surgery and anesthesia or may be the result of a low-level inflammatory response caused by injection of saline into the uterus in the control-saline group (Figure 1B). Although there was a significant decrease in AMT uptake in the cortex in the endotoxin kits, no significant differences in the SUVs in PVR were found between the endotoxin group (0.59; 95% CI=0.45 to 0.74), control-saline group (0.66; 95% CI=0.61 to 0.72), and control-no intervention group (0. 71; 95% CI=0.63 to 0.78) (χ2=2.05, df=2, P=0.40, for test of overall group differences) (Figure 1B). When comparing the ratio of AMT uptake in the PVR to the whole brain, a significant increase was noted for the endotoxin kits when compared with the controls (mean (95% CI) ratio of SUV in PVR to that of the whole brain was 1.02 (0.91 to 1.12) for control-no intervention; 1.11 (1.08 to 1.15) for control-saline; and 1.46 (1.27 to 1.65) for endotoxin group, P<0.001 for endotoxin versus each of the control groups and P=0.08 for both control groups; (χ2=16.13, df=2, P<0.001 for overall group differences)). These results indicate that although there was no difference between the groups in the absolute value of AMT uptake in the PVR between the groups, there was a relative increase in tryptophan metabolism in the PVR of endotoxin kits when compared with that in the whole brain.
Measurement of cortical tissue homogenates for serotonin concentration by enzyme-linked immunosorbent assay demonstrated that endotoxin kits have significantly lower levels of serotonin (mean±s.d.=0.188±0.042ng/mL) in the cortex compared with the control-no intervention (0.546±0.049ng/mL) and control saline groups (0.586±0.002ng/mL) (χ2=89.91, df=2, P<0.001 for overall group differences; P<0.001 for endotoxin versus both controls and P=0.4 for control-no intervention versus control-saline; Figure 2). Similarly, the serotonin levels in the hippocampus were significantly lower in the endotoxin kits when compared with the controls (mean±s.d.=0.55±0.15ng/mL for control-no intervention, 0.61±0.17ng/mL for control saline group, and 0.18±0.10ng/mL for the endotoxin group; P<0.001 for endotoxin versus both control groups and P=0.5 for control-no intervention versus control saline group; χ2=94.5, df=2, P<0.001 for overall group differences) (Figure 2). This indicates that the decreased AMT SUV seen in the cortex of endotoxin kits reflects decreased tryptophan metabolism to serotonin, whereas the relatively higher AMT uptake in the PVR (that includes the hippocampus) would indicate increased tryptophan metabolism along a nonserotonin pathway in kits exposed to maternal inflammation in utero.
On immunohistochemical staining for serotonin, a decrease in serotonin-immunoreactive terminals in the somatosensory parietal cortex was observed in the endotoxin animals when compared with the controls (Figures 3B and 3C). A significant difference in serotonin fiber length was noted between all groups (χ2 =48.5, df=2, P<0.001), with a greater decrease in the total length of the fibers in the endotoxin kits (mean±s.d.=0.123±0.040μm/μm2 of area measured) when compared with both control groups (0.206±0.028μm/μm2 for control-no intervention, and 0.173±0.038μm/μm2 for control saline group, P<0.001, χ2=44.16, df=1 for endotoxin versus control-no intervention; and P=0.002, χ2=9.78, df=1, for endotoxin versus control-saline) (Figure 3C). There was a decrease in the serotonin-immunoreactive fibers in the control-saline group when compared with the controls that had no maternal intervention (P=0.004, χ2=8.1, df=1) similar to that noted with the PET imaging, which may be related to some degree of inflammation and stress associated with the surgery.
To determine whether this decrease in serotonin-staining terminals was due to injury to the thalamocortical axons that transiently take up serotonin during development or due to decrease in the number of serotonergic neurons that project to the cortex from the brainstem, the number of serotonergic cells identified by serotonin staining in the dorsal raphe nucleus of the brain stem was counted in the endotoxin and control-no intervention kits. Serotonergic cells from the raphe nucleus project throughout the brain and these cells express TPH2, the rate-limiting enzyme for serotonin synthesis. The cortex was stained for TPH2 to differentiate between fibers from the raphe nucleus and that from the thalamus (which does not express TPH2). There was no significant difference in the number of cells in the raphe nuclei (mean±s.d. was 113,036.63±20,422.52 in the endotoxin group versus 146,873.21±25,591.58 in the control group, P=0.342 n=4 kits/group) (Figure 4). The fibers staining for TPH2 in the cortex was qualitatively observed to be similar in both the control and endotoxin groups (Figure 5). Although raphe nuclei neuron projections are found throughout the brain, they are very sparse in the cortex (Rebsam et al, 2002) and the terminals stain very faintly for TPH2 in the cortex (Weissmann et al, 1987). Hence, the intensity of staining was insufficient to evaluate fibers using the spherical probe technique.
Intrauterine endotoxin administration results in reduced 5HTT mRNA expression in the cortex and decreased number of thalamic neurons in the VB thalamus of endotoxin kits. Thalamocortical afferents project from the VB thalamus to the parietal somatosensory cortex and transiently express 5HTT and vesicular monoamine transporter during brain development (Lebrand et al, 1996; Persico et al, 2001). There was a significant reduction in the mRNA expression of 5HTT in the cortex of endotoxin kits when compared with the control groups (F(2,10)=56.07, P<0.001) (relative ratio of 5HTT expression to GAPDH expression, mean (95% CI) values were 1.11 (0.98 to 1.24) for control-no intervention, 0.88 (0.75 to 1.01) for control-saline, and 0.32 (0.20 to 0.44) for endotoxin; P<0.001 for endotoxin versus both controls; Figure 6). A decrease in 5HTT expression in the cortex of the control-saline group was observed when compared with control-no intervention (P=0.02), which is similar to the difference seen in PET imaging and in serotonin immunoreactivity in the cortex between the two control groups. There was also a significant increase in the number of apoptotic cells in the VB thalamus of kits that were exposed to intrauterine endotoxin injection when compared with those exposed to saline in utero (Figure 7) (mean±s.d. of the ratio of apoptotic cells to the total number of cells counted was 0.36±0.08 for the endotoxin kits versus 0.11±0.07 for the control-saline, difference between groups=0.28±0.13, χ2=22.71, df=1, P<0.001).
In this study, we have demonstrated that maternal intrauterine endotoxin administration leads to a significant decrease in multiple serotonergic markers in the offspring. Tryptophan metabolism to serotonin as measured by AMT SUV and serotonin levels in the frontal and parietal cortices of the neonatal rabbit brain were significantly decreased following intrauterine endotoxin exposure. In addition to the decreased serotonin content, there was a loss of serotonin-immunoreactive terminals and decreased expression of 5HTT measured in the parietal sensory cortex of endotoxin-exposed kits. Although serotonergic raphe nuclei cell bodies and TPH2-positive cortical fibers were intact in the endotoxin-treated kits, there was increased apoptosis of VB thalamic cells. These results indicate that the loss of serotonin-immunoreactive fibers in the cortex is most likely due to loss of thalamic neurons and thalamocortical afferents that transiently express 5HTT to uptake and store serotonin during development. Serotonin signaling has been shown to influence axonal outgrowths and thalamocortical pathfinding (Bonnin et al, 2007). Diminished serotonin signaling during development may result in defects in thalamocortical circuit formation. These results are significant because they demonstrate that maternal intrauterine inflammation can disrupt the serotonergic developmental regulation of thalamocortical innervation in somatosensory cortex of newborn kits.
Tryptophan is an essential amino acid that is metabolized in several pathways of which the most common one is the serotonergic pathway. Serotonin is a monoamine neurotransmitter that is known to have a vital role in regulating brain development (Lauder and Krebs, 1978). Decreased tryptophan metabolism resulting in a decrease in serotonin content in the cortex may be secondary to induction of IDO in the placenta and in activated microglia in the PVR of the newborn brain following intrauterine endotoxin exposure. Maternal inflammation results in induction of IDO in the placenta that would limit the tryptophan available for the fetal brain for serotonin synthesis. Similarly, microglia when activated, overexpress IDO that would shunt tryptophan away from serotonin formation along the kynurenine pathway resulting in serotonin depletion in the developing brain.
During development, serotonin-immunoreactive fibers in the cortex primarily consist of two populations, afferents of raphe serotonergic neurons and thalamocortical glutamatergic neurons transiently expressing 5HTT and vesicular monoamine transporter leading to uptake and storage of serotonin (D'Amato et al, 1987; Lebrand et al, 1996; Bennett-Clarke et al, 1997; Persico et al, 2001). Evidence from both pharmacological and gene knockout experiments demonstrates that serotonin has a role in thalamocortical development. Immunocytochemistry for serotonin and [3H]citalopram binding to serotonin uptake sites both have demonstrated a transient serotonergic innervation of primary sensory cortex between postnatal days 2 and 14 during the period of synaptogenesis in rat cortex (D'Amato et al, 1987). This actually represents transient expression of the high-affinity 5HTT (Lebrand et al, 1996) and vesicular monoamine transporter (Persico et al, 2001) by glutamatergic thalamocortical neurons. During the first 2 postnatal weeks in rats, these thalamocortical neurons take up and store serotonin, although they do not synthesize serotonin. Depletion of serotonin during this crucial period delays the development of the barrel fields of the rat somatosensory cortex, decreases the size of the barrel fields, and disrupts the synaptic connectivity in sensory cortices (Blue et al, 1991; Bennett-Clarke et al, 1994). In rabbits, the somatosensory barrels are not as clearly defined and distinct as in rats and mice, and the time course for the development of the somatosensory cortex differs. The cortical barrels appear earlier in the rabbit, with appearance at postnatal day 1 (31 to 32 days gestation), becoming indistinct and similar to that in adults by postnatal day 5 (PND5) (Rice et al, 1985). As the time course for the development of somatosensory cortex occurs earlier in the rabbit, it is likely that the critical period for the trophic role of serotonin on thalamocortical development occurs in the perinatal period in rabbits instead of the first and second postnatal week as in rats and mice.
Transient expression of 5HTT during postnatal development was noted in 70% to 80% of the neurons of the VB thalamic nucleus projecting to the sensory cortex (Lebrand et al, 1996). In our model, there was a significant decrease in the expression of 5HTT in the parietal cortex along with apoptosis of the neurons in the VB thalamus in the endotoxin-treated animals. Moreover, there was no difference in the number of serotonergic neurons in the raphe nucleus or in the presence of TPH2-staining fibers in the cortex between the groups. As serotonergic neurons in the raphe nucleus and their projections develop very early in gestation (Rubenstein, 1998), an inflammatory insult closer to term may not influence these cells, although an arrest in the development or maturation of the fibers cannot be ruled out. Taken together, our data suggest that the decrease in serotonin-immunoreactive fibers in the cortex may primarily be due to a loss of thalamocortical afferents. This is in accordance with studies that have shown that experimental thalamic lesions involving the VB thalamus in postnatal rats resulted in the loss of 5HT-immunoreactive fibers in the cortex (Lebrand et al, 1996). We hypothesize that maternal inflammation may either lead to a direct cytokine-mediated injury to the thalamocortical afferents with retrograde involvement of the VB neurons, or may result in direct injury to the VB neurons leading to loss or impaired development of thalamocortical afferents. Future studies directed at evaluating the development of the serotonergic system at different time points following a maternal inflammatory exposure may help elucidate this further.
We propose that one link between maternal intrauterine inflammation and fetal brain development may be the transport and metabolism of serotonin and its precursor tryptophan in the placenta. Studies in mice deficient in peripheral serotonin synthesis have shown that maternal serotonin production is crucial for normal fetal neurogenesis and development (Cote et al, 2007). The placenta also expresses TPH2 (Correa et al, 2009), resulting in serotonin production that would be available for transport to the fetus. Induction of IDO in inflammatory conditions with increased tryptophan metabolism along the kynurenine pathway can lead to decreased tryptophan availability for serotonin synthesis in the placenta and in the fetus. Thus, by the regulation of tryptophan levels, IDO activity may influence availability of serotonin in the fetus. Furthermore, several kynurenine pathway metabolites (e.g., quinolinic acid, kynurenine, 3-hydroxykynurenine) are neurotoxic (Stone, 2001). These tryptophan metabolites might also have a role in neurodegeneration of thalamocortical neurons in this model.
Maternal intrauterine inflammation resulting in immune dysfunction during development has been implicated in the development of neurodevelopmental disorders such as autism, schizophrenia, and cerebral palsy (Fatemi et al, 2008). A common link among these disorders appears to be the presence of activated microglia and evidence for immune dysregulation in the developing brain (Patterson, 2009). Brain tissues of autistic patients were found to have increased number of activated microglia and astrocytes along with an increase in the levels of proinflammatory cytokines (Vargas et al, 2005). Animal models of maternal inflammation have been shown to result in motor deficits, behavioral abnormalities, and neuroglial activation in the offspring (Meyer et al, 2006; Saadani-Makki et al, 2008). Viral infections in the prenatal period have been associated with alterations in serotonin synthesis in the cerebellum of P14 mice (Winter et al, 2008). We have previously shown that maternal intrauterine endotoxin administration results in motor deficits suggestive of cerebral palsy in the newborn rabbit along with microglial activation in the PVR (Saadani-Makki et al, 2008). Using the same model, we have shown in this study that intrauterine endotoxin administration results in decreased tryptophan metabolism to serotonin and loss of serotonin containing thalamocortical terminals in the parietal cortex of the neonatal rabbit.
The decreases in serotonin synthesis in this rabbit model of maternal inflammation induced perinatal brain injury show some similarity to those found in children with autism. In nonautistic children, serotonin synthesis capacity was high with values >200% of that in adults until the age of 5 years, subsequently declining toward adult levels. In contrast, serotonin synthesis capacity in autistic children increased gradually between the ages of 2 and 15 years to values 1.5 times normal adult values (Chugani et al, 1999). These data suggest that humans undergo a period of high brain serotonin synthesis capacity during early childhood, and that this developmental process is disrupted in autistic children.
Impairment in sensory perception and integration are not only documented in patients with autism and autism spectrum disorders (Molloy et al, 2003) but are also known to be associated with motor deficits in patients with cerebral palsy (Wingert et al, 2008). Indeed, numerous reports have associated the degree of motor impairment to the extent of somatosensory deficits in cerebral palsy (Gordon and Duff, 1999; Wingert et al, 2008). Diffusion tensor imaging studies have shown that patients with cerebral palsy have severe disruption of the thalamocortical connections to the somatosensory cortex, along with injury to the descending corticospinal pathways leading to the hypothesis that the injury to the ascending somatosensory pathways would define the extent of clinical impairment (Hoon et al, 2009). We have demonstrated that maternal intrauterine exposure to endotoxin results in decreased serotonin synthesis along with the loss of thalamocortical afferents in somatosensory cortex of the newborn rabbit. This may disrupt the normal development of neocortical circuitry in the somatosensory cortex, resulting in altered sensory processing and sensorimotor integration adding to the motor deficits in patients with cerebral palsy. This may also provide a link between prenatal exposure to inflammation and the development of autism.
The authors thank Dr S DiCarlo for the use of his StereoInvestigator, Mr Xin-Lu for technical help with PET scanning, Mr Y Shen for technical help with magnetic resonance imaging, and Dr A Abbas for his help with immunostaining.
The authors declare no conflict of interest.
This study was supported in part by 1K08HD050652, NICHD, NIH, and the Perinatology Research Branch, Eunice Kennedy Shriver NICHD, NIH, DHHS.