|

 

Allmanlab.caltech.edu

J Autism Dev Disord (2012) 42:2569–2584DOI 10.1007/s10803-012-1513-0 Microglia in the Cerebral Cortex in Autism Nicole A. Tetreault • Atiya Y. Hakeem •Sue Jiang • Brian A. Williams • Elizabeth Allman •Barbara J. Wold • John M. Allman Published online: 31 March 2012 ! Springer Science+Business Media, LLC 2012 We immunocytochemically identified microg- capillaries into brain tissue. The brain has its own immune lia in fronto-insular (FI) and visual cortex (VC) in autopsy system based on microglia, which are derived from the brains of well-phenotyped subjects with autism and mat- macrophage lineage and reside throughout the brain, where ched controls, and stereologically quantified the microglial they mount defenses against invading microorganisms and densities. Densities were determined blind to phenotype clear damaged tissue and metabolic waste (Graeber and using an optical fractionator probe. In FI, individuals with Streit ). This is achieved through phagocytosis, autism had significantly more microglia compared to con- in which the microglia ingest these substances.
trols (p = 0.02). One such subject had a microglial density Nimmerjahn et al. (Davalos et al. ) and in FI within the control range and was also an outlier Wake et al. directly observed the activity of behaviorally with respect to other subjects with autism. In microglia in intact living mouse brains using two-photon VC, microglial densities were also significantly greater in microscopy in animals that express green fluorescent pro- individuals with autism versus controls (p = 0.0002).
tein specifically in microglia. Their experiments showed Since we observed increased densities of microglia in two that microglial cell bodies are relatively stationary, but functionally and anatomically disparate cortical areas, we their fine processes are in constant motion on a minute-to- suggest that these immune cells are probably denser minute basis. They observed that the microglial processes throughout cerebral cortex in brains of people with autism.
continually probe the immediate area, so that the popula-tion conducts a complete surveillance coverage of brain Microglia ! Autism ! Fronto-insular cortex ! tissue every few hours. When the microglial processes encounter damaged tissue, metabolic byproducts such asoxidized lipoproteins, or invading microorganisms, theyrespond by expanding and engulfing these substances and transporting them back to the microglial cell body wherethey are stored for an indeterminate period of time. The The brain is substantially isolated from the body's immune microglia contact other types of glia and neurons as part of system by the blood–brain barrier, which restricts the their constant surveillance, but when they encounter other passage of most immune cell types and proteins from microglia there is mutual repulsion of their processes,which may account for their relatively uniform spacing.
There is also evidence that microglial processes can strip N. A. Tetreault (&) ! A. Y. Hakeem ! S. Jiang ! synapses away from their dendrites, suggesting that B. A. Williams ! B. J. Wold ! J. M. Allman microglia may have another role in modifying neuronal Division of Biology, California Institute of Technology, 1200East California Blvd., MC216-76, Pasadena, CA 91125, USA connections in development and plasticity (Blinzinger and Kreutzberg ; Graeber et al. ; Kreutzberg ;Paolicelli et al. Wake et al. ) reported through in vivo imaging that microglia make transient direct con- Comprehensive Autism Center, 2604 B El Camino Real #214,Carlsbad, CA 92009, USA tact with synapses, and that the frequency of this contact is J Autism Dev Disord (2012) 42:2569–2584 dependent on neural activity. In experiments in visual Baltimore, as shown in Table FI was identified based on cortex, microglia contact frequency was decreased by criteria such as the presence of the Von Economo neurons silencing the visual input by injecting tetrodotoxin into the and sulcal location (Allman et al. and corresponds to eyes (Wake et al. ). Inducing neural degeneration by the posterior part of Brodmann's area 47. VC was identi- transient ischemia increased the duration of microglial fied by using the calcarine sulcus as a landmark; the contact with synapses which was followed by synapse dissections involved the sulcal lip corresponding to Brod- elimination (Wake et al. mann's areas 17 and 18. The NICHD Brain and Tissue Microglia are closely related developmentally and Bank for Developmental Disorders provided detailed functionally to macrophages. Both originate from the clinical records, with personal identification removed, for monocyte lineage in the bone marrow. Microglia first each individual with autism whose brain we studied, as appear in small numbers in the brain during embryogene- summarized in the phenotypic descriptions in Table To sis, but they emerge prominently during the early postnatal confirm the diagnosis of autism, the medical records of period when they enter the brain from the bloodstream to each person with autism were reviewed in depth by a form what has been called the fountain of microglia, in clinical psychologist (EA) who specializes in autism. In which they migrate along the course of the fibers of the each case we have at least one thorough clinical description corpus callosum to all parts of the brain (Imamoto and of the subject by either a psychologist or psychiatrist. Ten Leblond The initial population of microglia can be of our eleven subjects with autism had the autistic diag- augmented by subsequent invasion into the brain of cir- nostic interview-revised (ADI-R), which is the result of a culating macrophages, which apparently assume the mi- structured interview with a parent of the individual with croglial phenotype after entering neural tissue (Schmid autism. Three of the individuals with autism had ADIR et al. ). In a preliminary study of gene expression, we records, but the actual scores were not in the file. One observed in some of our autistic cases increased expression individual with autism had a childhood autism rating scale of a network of genes centered on interleukin-6 (Tetreault and met the criteria for an autism diagnosis. The records et al. Interleukin-6, together with several other additionally include measures of behavioral development genes in the network, is characteristic of activated versus such as the Bayley tests, as well as a history of medications quiescent microglia (Thomas et al. ). People with and other health issues reported by physicians and clinical autism have significantly increased cytokines in frontal psychologists, described in Table .
cortex and elevated levels of cytokines in the cerebro-spinal fluid compared to control subjects (Li et al. ; Sectioning and Immunocytochemistry Zimmerman et al. and there is evidence for immunesystem dysfunction in the development of autistic children Samples were sectioned in the coronal plane at 50 lm on a (Ashwood et al. ; Chez and Guido-Estrada microtome with a vibrating blade (Microm HM 650 V) in These observations motivated us to conduct a quantitative 0.1 M phosphate buffer solution (PBS) and stored in well study of the density of microglia in brains of individuals dishes with PBS and sodium azide. The microglia were with autism compared to controls. Our goal in this work is immunocytochemically stained with an antibody to IBA1 to quantify microglial differences between subjects with (ionizing calcium adaptor molecule-1), the gene product of autism and age-matched controls in two cortical areas, the Aif1 gene (allograft inflammatory factor 1), raised fronto-insular cortex (FI) and primary visual cortex (VC).
against the C-terminus of IBA1, which labels microglia and Multiple lines of evidence have previously implicated FI in monocytes. We used the IBA1 antibody because it yields autism (Allman et al. Di Martino et al. ; Santos excellent and selective staining of microglia in formalde- et al. ); VC was selected because of its functional hyde-fixed human archival brain tissue (Streit et al. difference and anatomical distance from FI, in an effort to The utility of IBA1 for the study of microglia has also been span the diversity within neocortex.
shown through expression of the IBA1 gene coupled withenhanced employing 2-photon microscopy to image the development and motility of this class of cells in the brains of livingmice (Hirasawa et al. Wake et al. ). We used a concentration of 1:1,000 of IBA1 antibody (Wako, CodeNo. 019-19741). Four batches of immunostaining were Formaldehyde-fixed (8 % solution) human right FI and performed including duplicate sections from both FI and right VC tissue from subjects with autism and controls was VC of each of the subjects, and each of the staining pro- obtained from the NICHD Brain and Tissue Bank for cedures showed consistent and robust immunostaining Developmental Disorders at the University of Maryland- across the sections. Free-floating sections were rinsed with J Autism Dev Disord (2012) 42:2569–2584 Table 1 Autistic and neurotypical control subjects used for microglial density measurements Fall from 9th story Serotonin syndrome Drowning with seizure Cardiac arrhythmia Subdural hemorrhage Accident, multiple injuries Multiple injuries Multiple injuries Multiple injuries Multiple injuries Head and neck injuries The tissue source is NICHD Brain and Tissue Bank for Developmental Disorders at the University of Maryland, Baltimore, MD. The letters inthe age column are for the purpose of differentiating subjects of the same age in the graphs in Figs. and PMI post-mortem interval,X microglia density measurements were made for this structure PBS and then incubated with 1 % citrate buffer (Chemicon, without primary antibody and incubated with goat IgG at cat # 21545) for 30 min at 37 "C for antigen retrieval.
the same concentration as the primary antibody. No Sections were rinsed with PBS, treated to remove endog- immunostaining was observed in these control sections.
enous peroxidase activity with 0.75 % hydrogen peroxideand methanol for 20 min, and then rinsed with PBS. The Quantification of Microglial Densities blocking step, to eliminate random antibody binding, used0.1 % Triton X-100, 4 % normal goat serum (NGS), 1 % Microglial density in FI and VC was measured blind to BSA, and 3 % dry milk in PBS for 3 h. Primary antibody phenotype and quantified using the program Stereo Inves- was incubated for 38 h at 4 "C in a PBS solution that tigator (MBF Bioscience, Williston, VT) with a Reichert included 0.1 % Triton X-100, 2 % NGS and 1:1000 anti- Polyvar microscope equipped with a motorized stage and a IBA1. Sections were then rinsed with PBS, incubated with camera for visualization. All sections were quantified in at biotinylated anti-rabbit (BA-1000,Vector Laboratories) at least two separate replications with different regions of 1:200 for 2 h, and then rinsed again with PBS. A Vecta- interest, and some sections were quantified up to four times stain Elite ABC kit (pk-6100, Vector Laboratories) was with both different and identical regions of interest. For all used for the avidin–biotin-peroxidase method, then sec- of the samples, duplicate sections of FI and VC were tions were incubated for 30 min. After sections were once classified and quantified for reproducibility. Independent again rinsed with PBS, immunoreactivity was visualized by raters quantified and classified blind random sections to using a chromagen, 30-diaminobenzidine and nickel (SK- replicate the method. The represented density measure is 4100, Vector Kit). Null control sections were incubated an average of the blind replicated runs. Quantification was J Autism Dev Disord (2012) 42:2569–2584 J Autism Dev Disord (2012) 42:2569–2584

J Autism Dev Disord (2012) 42:2569–2584 performed within a region of interest that spanned layerstwo to six; microglial distribution appeared relativelyhomogenous throughout the layers in our samples. Esti-mated cell counts were performed using the optical dis-sector probe at 409 magnification (oil immersionNA = 1.0) with a dissector height of 16 lm (flanked by2.0 lm guard zones), a counting frame of 260 lm 9160 lm and a grid size of 425 lm 9 425 lm. To avoidoversampling, we used the Gunderson counting rule suchthat cells intersecting only 3 of the 6 surfaces of the dis-sector cube were counted. Microglial density per mm3 wascalculated by dividing the optical fractionator estimate ofthe number of cells present in the full thickness of thesection within the region of interest by the area of the Fig. 1 Stereological procedure for quantifying and identifying region of interest and the thickness at which the section microglia in control and the brains of individuals with autism; thered and green frame defines the borders of the region of interest for was cut to account for any tissue shrinkage.
counting microglial cells according to the Gundersen et al. ()procedure. A microglial cell was included if it was in the counting Statistical Analysis frame or if the soma crossed the green line and was excluded from thecounting when the cell soma crossed the red line to avoidoversampling. We used an optical dissector height 16 lm (flanked Densities for the subjects with autism and control popula- by 2.0 lm guard zones) and dissector probe at 409 magnification (oil tions were compared using the Mann–Whitney test with immersion NA = 1.0). Some of the cells are out of focus in the two-tailed p value. Correlation levels between replications photomicrograph, which is caused by the high numerical aperture of were measured using Pearson's r-squared. Possible con- the lens which creates many depth planes through the tissue which isnecessary for quantifying cells in three dimensions. Immunocyto- founds in the subjects with autism that could alter mi- chemical labeling with Iba1 (1:1000, Wako), a specific marker for croglial densities were examined. Binary confounds, microglia and macrophages (Sasaki et al. in FI of the 14 year including whether death was by drowning and whether old male with autism (UMB4315) (Color figure online) seizures were present, were tested using the Mann–Whitney test; a possible confounding correlation with post- with autism (n = 10) had significantly higher microglial mortem interval was tested using Pearson's r-squared.
density (p = 0.02060) than control subjects (n = 12)(Mann–Whitney test with two-tailed p value).
The 12-year-old male UMB4305 was a unique case in this group of people with autism because there was noincrease in microglial density compared with controls.
Figure depicts the stereological method and photomi- Although the ADI-R scores for this case are in the autistic crograph of the microglial quantification method in a brain range, he was diagnosed as having pervasive develop- of an individual with autism. We found significantly higher mental disorder not otherwise specified (PDD-NOS), and, density in the individuals with autism than in the controls in addition, with psychosis NOS, and ADHD. UMB4305 in both FI (p = 0.0206, see Fig. ) and VC (p = 0.0002, was the only one among all subjects tested who was treated see Fig. ). The numbers are represented as the average of for psychosis, including administration of the drugs que- the microglial densities for the multiple replications per- tiapine, olzapine, and risperdal (Table ). For these rea- formed in each individual. Comparisons were made using sons, we think this individual may have suffered from a Mann–Whitney tests with two-tailed p values. The repeated condition distinct from the other individuals who had the quantifications in the same structure are highly signifi- autism diagnosis. According to the neuropathology report cantly correlated: for FI, and VC, r2 = 0.6480, p 0.0001 for UMB4305, ‘‘there were three small foci of yellow (Fig. when the blind replications are from the exact ROI discoloration noted in the leptomeninges overlying the the correlation is r2 = 0. 9780, p 0.0001 for the intra- right antero-inferior frontal pole, right gyrus rectus and left rater reliability. Notably, the individuals with autism gyrus rectus which measured 0.2 9 0.2 9 0.2 cm. Well- cluster together in FI and VC, except for a single outlier circumscribed regions of shrinkage and slight yellow dis- subject with autism, while the controls all cluster together coloration were present in the cortical ribbon underlying in both FI and VC.
the discolored leptomeninges. … There was necrosis Figure shows the microglial cell densities in FI of around the small area of the contusions that included the autistic subjects and controls for the combined and aver- entire cortical ribbon through layer one. The small frontal aged data for both microglial quantifications. Individuals lobe contusions had visible macrophages surrounded by J Autism Dev Disord (2012) 42:2569–2584 FI Microglia / mm
CN_2_M_MB5282 CN_4_M_MB4670
Fig. 2 Microglial densities in FI in subjects with autism and D: cause of death was drowning. BD: brain damage, brain contusion, neurotypical brains are represented as the average for the replicated hemorhage, or edema. TD: traumatic death (MVA, fall) with possible runs. Individuals with autism (n = 10) have a significantly greater head injury, not explicitly mentioned. OD: drug overdose (not density of microglia, the key cellular participants in the inflammatory necessarily cause of death). S: seizures (not necessarily cause of response in the brain, compared to controls (n = 12) p = 0.0206 death). Numbers in black are post-mortem interval in hours (Mann–Whitney). LS: known to have spent time on life support.
reactive astrocytes observable with a hematoylin-eosin Using the estimated value for human neocortical grey stained sections.'' The report noted that beyond these local volume from Frahm et al. ), which is 584,706 mm3, contusions, the cortical layers were normal and the neurons one can then estimate the density of microglia in the in the cerebral cortex of the fronto-parietal lobe, hippo- neurotypical human cortex by dividing by the total number campus, basal ganglia, and cerebellum were unremarkable.
of microglia, which is approximately 5,951 (CD45 positive Figure presents similar data for primary visual cortex cells) per mm3 in the total human neocortex (Fig. b).
(VC). Total microglial densities were significantly greater This is close to our estimated microglial densities for in VC for the individuals with autism (n = 9) versus the control (n = 11) subjects (p = 0.0002 Mann–Whitney test In Table 5 of Lyck et al. (the column headed ‘‘total with two-tailed p value). The increase in microglial density neocortex'' refers to the neocortical gray matter only. In their methodsSection 2.2.7, ‘‘Estimation of Cell Numbers,'' they describe their is present throughout almost our entire sample of subjects selection of the region of interest, saying, ‘‘… followed by delineation with autism, with ages ranging from 3 years of age to 22.
the border between white matter and neocortex at 2109 magnification We address the two exceptions to this broad finding, UMB (10 9 lens) marking the white matter as ‘exclusive region','' 1185 and UMB 1713, below.
indicating that their cell number estimates were made from a regionthat excluded white matter. Further, Fig. 2b from this paper indicates After measuring microglial densities, we consulted Lyck that the brain slices were segmented into ‘‘frontal neocortex,'' et al. (in which the number of microglia in the cortex ‘‘temporal neocortex,'' ‘‘parietal neocortex,'' ‘‘occipital neocortex,'' of three well-documented neurotypical brains was carefully and ‘‘white matter,'' implying that the various neocortex segments do and comprehensively quantified using a CD45 antibody not include white matter. Thus, in Table 5 the column heads ‘‘frontalcortex,'' ‘‘temporal cortex,'' etc. presumably refer specifically to the with unbiased stereology. They reported an average of 3.48 gray matter portions of those regions, and ‘‘total neocortex'' (which is billion CD45 positive cells in the entire human neocortex.
a sum of the other four columns) also includes only gray matter.
J Autism Dev Disord (2012) 42:2569–2584 oglia / mm
CN_2_M_MB5282 CN_4_M_MB4670
Fig. 3 Microglial densities in visual cortex in autistic and neurotyp- total microglia, the key cellular participants in the inflammatory ical brains are represented for the average of the replicated runs.
response in the brain, compared to controls (n = 11) p = 0.0002 Individuals with autism (n = 9) have a significantly greater density of control FI (6,479 microglia per mm3) and control VC autism and found no statistically significant relationship (6,048 microglial cells per mm3). In FI, individuals with between microglial density and drowning versus other autism had an 18 % higher microglial density compared to causes of death; traumatic versus other causes of death; our neurotypical cases, and in VC 21 % higher microglial having been on life support or not; having a recorded drug density compared to our neurotypical cases.
overdose or not; or having had seizures or not (Table These findings demonstrate that, at the time of death, There was no significant difference between the subjects there were significantly higher microglial densities in the with autism and controls with respect to age of the subjects subjects with autism compared to the control subjects, and or post-mortem interval (PMI). However, the controls had that this change in microglial density is widespread significantly greater (p = 0.0328) brain weight (1,501 g) throughout the cerebral cortex in autism. The microglial versus the subjects with autism (1,374 g) (Mann–Whitney densities in FI and VC in the same subject were signifi- test). This difference was driven mainly by one control cantly correlated (both measures were available in 10 subject (M5387) with very high brain weight (1,750 g).
controls and 8 autistic subjects for a total of 18 subjects) This is 310 g greater than the average brain weight with Pearson's r2 = 0.4285, p = 0.0024 (Fig. This (1,440 g) for a 12 year old male (Dekaban and when indicates that the elevation in density is consistent between the outlier is removed there is no significant difference in these areas, and probably throughout the cortex, in both brain weight between the subjects with autism and the subjects with autism and controls.
control subjects. The differences in microglial density We tested several confounding variables that could alter between individuals with autism and controls remain sig- microglial densities in FI and VC of the subjects with nificant when the one outlier was removed for density J Autism Dev Disord (2012) 42:2569–2584 oglia / mm
oglia / mm
Control FI count 1 vs. count 2
Averaged Counts (micr
Autistic FI count 1 vs. count 2
Control VC count 1 vs. count 2
Autistic VC count 1 vs. count 2
Microglia / mm3
FI Averaged Counts (microglia / mm3)
Fig. 4 The repeated blind density measurements are strongly corre- Fig. 6 Microglial densities in FI and VC are significantly correlated.
lated. Density measurement count one in FI versus density measure- Pearson's r2 = 0.4285, p = 0.0024, for the sample of 10 controls and ment count two in FI and density measurement count one in VC 8 individuals with autism in which measurements were available for versus density measurement count two in VC (Pearson's r2 = 0.6480, both structures. Note that for both structures the individuals with p 0.0001) for two different regions of interest (ROI). When the autism (red) cluster, as do the controls (blue) (Color figure online) blind replications are from the exact ROI the correlation is r2 = 0.
9780, p 0.0001. Notably, the subjects with autism (FI solid red Microglial densities were negatively correlated with age circles and VC outlined red circles) and controls (FI solid blue circles in VC in our controls (Pearson's r2 = 0.6833, p = 0.0017) and VC outlined blue circles) cluster in FI and VC, except for one and barely missed statistical significance in FI (Table autistic outlier in FI (Color figure online) Microglial densities thus tend to decrease with age incontrols, while in people with autism the microglial den- measurements (FI, p = 0.0257 and VC, p = 0.0001, sities remain relatively high and constant with age in both Mann–Whitney tests). In addition, brain weight and mi- FI and in VC. Finally, microglial densities in VC in con- croglial density were not significantly correlated in indi- trols were negatively correlated with PMI (Pearson's viduals with autism compared to control cases for FI and r2 = 0.3952, p = 0.0383) but there was no significant VC (Table Morgan et al. ) found brain weight was correlation in VC for individuals with autism, or in FI for negatively correlated with microglial density in the grey either group (Table ). Morgan et al. (found that matter, but that the microglial differences between subjects microglial densities were negatively correlated with PMI with autism and control subjects persisted when they sta- across their subjects as a whole population but not for tistically controlled for brain weight.
controls or people with autism as subgroups.
Fig. 5 a Average microglial densities for subjects with autism (red) and controlsubjects (blue) in FI incomparison to total microglial density (black) estimated fromdata in Frahm et al. andLyck et al. b Average microglial densities in VC.
oglia / mm
oglia / mm
Error bars represent the standard deviation (Color figureonline) J Autism Dev Disord (2012) 42:2569–2584 Table 3 Confound statistics for the autistic cases FI autistics (6 drowning, 5 non-drowning): total density VC autistics (5 drowning, 4 non-drowning): total density drowning versus other COD, p = 0.7619 (ns) drowning versus other COD, p = 0.2857 (ns) FI autistics (4 seizures, 6 no seizures): total density seizures VC autistics (3 seizures, 6 no seizures): total density seizures versus no seizures, p = 0.2571 (ns) versus no seizures, p = 0.7143 (ns) FI autistics, total density versus PMI, N = 10, Pearson's VC autistics, total density versus PMI, N = 9, Pearson's r2 = 0.0658, p = 0.4743 (ns) r2 = 0.00159, p = 0.9189 (ns) FI controls, total density versus PMI, N = 12, Pearson's VC controls, total density versus PMI, N = 11, Pearson's r2 = 0.2628, p = 0.0883 (ns) r2 = 0.3952, p = 0.0383 (significant) FI autistics, total density versus brain weight, N = 10, Pearson's VC autistics, total density versus brain weight, N = 9, Pearson's r2 = 0.0077, p = 0.8095 (ns) r2 = 0.1311, p = 0.3384 (ns) FI controls, total density versus brain weight, N = 11, Pearson's VC controls, total density versus brain weight, N = 10, r2 = 0.00296, p = 0.6126 (ns) Pearson's r2 = 0.0295, p = 0.6348 (ns) FI autistics, total density versus age, N = 10, Pearson's VC autistics, total density versus age, N = 9, Pearson's r2 = 0.0080, p = 0.8054 (ns) r2 = 0.3477, p = 0.0947 (ns) FI controls, total density versus age, N = 12, Pearson's VC controls, total density versus age, N = 11, Pearson's r2 = 0.3159, p = 0.0572 (ns) r2 = 0.6833, p = 0.0017 (significant) Drowning, seizures, PMI age and brain weight do not account for the increase in microglial density for autistics compared to the controls. Thecontrols had significantly greater (p = 0.0302) brain weight (1,501.4 g) verses the autistics (1,356.7 g). This difference was driven mainly byone control subject (M5387) with very high brain weight (1,750 g) which is 310 g greater than the average brain weight (1,440 g) for a 12 yearold male (Dekaban ) and when the outlier is removed there is no significant difference (ns) for brain weight comparing the autistic andcontrol cases. The controls have a significant correlation for microglial density with age in VC (r2 = 0.6833 and p = 0.0017), where over timethe microglia decrease with age and a similar trend occurs in FI but does not reach statistical significance We found that FI of two control subjects had unusually with autism in grey and white matter. Five of Morgan high microglial densities compared to the other controls.
et al.'s cases with autism overlap with those used in our They were UMB1185, the 4-year-old control case, and the study (Table ). We found that the subjects with autism we 23-year-old control UMB1713, who had suffered from had in common with Morgan et al. showed an increase in head and neck injuries. The injuries sustained by microglial density in both FI (five subjects in common) and UMB1713 are such that could cause an increase in mi- VC (four subjects in common), which is consistent with croglial density if death was not immediate (Engel et al.
Morgan et al.'s findings in dlPFC. In addition, Morgan , Loane and Byrnes Both of these individuals found five of the thirteen individuals with autism had an showed increases in microglial densities in FI, but not in increase in microglial activation (Morgan et al. VC. By contrast, our subjects with autism had global Precedent for Morgan's and our microglial observations increases in microglial densities, shown both in FI and in comes from Vargas et al. (who found significantly VC. This regional difference suggests the possibility of more microglial activation in the cerebellum of autistic injury-related pathology in these two controls.
brains versus controls and a trend toward more microglialactivation in the middle frontal and anterior cingulatecortices, although the cortical results were not statistically significant. One of our individuals with autism was used inthe Vargas study (Table ) (Vargas et al. We observed increased densities of microglia in two dis- Our methodologies differed, however, in several parate cortical areas. One possibility is that these increased respects from those of Morgan et al. ). We quantified densities reflect abnormalities specific to these particular microglia in two cortical regions, FI and VC, consistently cortical areas, since there is evidence that each is involved in the right hemisphere, whereas Morgan quantified a sin- in autism, or alternatively these results may reflect a gle region, dlPFC, using either the right or left hemisphere.
widespread difference that occurs throughout the cortex or The reports of increased microglial densities are consistent, even much of the brain. Consistent with the possibility that but there are differences in density measurements in the effect is pan-cortical, Morgan et al. ) reported an Morgan's and our studies. The differences in density increase in microglia in subjects with autism in dorsal measurements for the individuals with autism and controls lateral prefrontal cortex (dlPFC) compared to controls, and can be attributed to our differing calculations and consid- found an increase in somal size in microglia in subjects eration of the shrinkage factor within the tissue. To account J Autism Dev Disord (2012) 42:2569–2584 Table 4 Autistic cases used in the Vargas, Morgan and our study for microglial densities Tetreault et al. (this study) For this study we quantified two regions in cortex, FI and VC which have not previously been quantified and showed that six additional autisticcases have increased microglial density measures. An X indicates that the subject was evaluated in the study for shrinkage, we calculated the microglial density per control FI and VC that are near the expected densities mm3 by dividing the optical fractionator estimate of the calculated from Lyck et al. ) and Frahm et al. ( number of cells present in the full thickness of the section On average the individuals with autism had 18–21 % within the region of interest by the area of the region of higher microglial density in FI and VC compared to neu- interest and the thickness at which the section was cut. Our rotypical subjects. How and when does the increased results for control samples are very close to values calcu- density of autistic microglial arrays arise, and how is it lated for microglia based on the total number of microglia maintained? Of course we have no data prior to the time of in the entire neurotypical cortex determined through ster- death, but the consistency of results among 10 subjects eology (Lyck et al. and cortical volumes (Frahm with autism of differing ages argues that people with aut- et al. ) (see Fig. a, b).
ism have developed a remarkably stable steady-state The differences between our study and Vargas et al.
microglial density. Given the age range, this is probably () are that they stained microglia with an antibody to established before age three. It is not clear how long the HLA-DR and used an area fraction quantification method increase in microglia persisted in each of the subjects with based on the Delesse sampling procedure (Gundersen et al.
autism, but our results show that control subjects have a ). That method gives an estimate of the fractional area significant negative correlation between microglial density of the region of interest covered by the cell type being in VC and age, indicating that microglial densities nor- measured. The Delesse method does not, however, produce mally decrease throughout childhood and early adulthood cell numbers or three-dimensional densities. By contrast, in neurotypical subjects. However, in people with autism, we stained with an antibody to IBA1 and measured there is a relatively steady condition of increased microg- microglial density in our tissue. Though the specifics of lial density from childhood into adulthood. It seems pos- antibody and methods differed, our data taken together sible that some persistent stimulus is the cause of this with Vargas et al. (and Morgan et al. (point to sustained higher level of microglial density in the subjects elevated microglial density in autism, possibly involving with autism. Imaging experiments of quiescent microglia in the entire cerebral and cerebellar cortices. This argues that intact living cortex suggest that they conduct a complete further investigation of microglial abnormalities and the surveillance of the cortex every few hours (Davalos et al.
microglial pathways in people with autism may be ; Nimmerjahn et al. ). The greater density, and important for understanding the cellular basis of the autism thus closer spacing of the microglia, in brains of individ- uals with autism compared with control brains, suggests There are also some caveats. We cannot be sure that that this surveillance is more intense in autism.
IBA1 stains all microglia, and there is evidence for Sickness behavior results from systemic infection and/or microglial heterogeneity (Carson et al. Mittelbronn inflammation, driving an increase in signals to the brain et al. ; Schmid et al. However, the spacing of that cause changes in metabolism, social withdrawal, the stained microglia we have observed is consistent with appetite suppression and a general ill feeling (Exton ; complete coverage of a relatively regular array of microglia Hart ; Perry ). Sickness behavior is another in the cortex. In addition, we found microglial densities in example of how a systemic infection or its related J Autism Dev Disord (2012) 42:2569–2584 inflammation can alter both behavior and the inflammatory among the originators of the pathologic processes in aut- response in the brain. There is evidence that maternal viral ism, or are they a response (perhaps even a protective one) infection in the first trimester and bacterial infection in the to some other aspects of this condition? Microglia have second trimester are correlated with an increase in off- neuroprotective functions including the phagocytosis of spring reported to have autism (Atlado´ttir et al. ). In a invading microorganisms and metabolic waste. The recent microarray analysis of gene expression in brains of increase of microglial densities in individuals with autism individuals with autism compared to controls, Voineagu could be a function of neuroprotection in response to et al. (found a module of enriched immune and microglial genes, although these genes have not been found In contrast, microglia can also phagocytize synapses and in genome wide association studies that have sought to whole neurons, thus disrupting neural circuits. For exam- identify genes that predispose to autism. Voineagu et al.
ple, when the axons of motor neurons are cut, the microglia () conclude that the enriched gene expression of strip them of their synapses (Blinzinger and Kreutzberg immune and microglial genes observed in their study has a ; Cullheim and Thams Graeber et al. non-genetic etiology and may reflect internal or external Another example of the disruption of circuitry arises from environmental influences, which suggests the possibility the direct phagocytosis of neurons. Neurons communicate that the sustained higher levels of microglia density in with microglia by emitting fractalkine, which appears to people with autism may also be environmentally mediated.
inhibit their phagocytosis by microglia. Deleting the gene Chez and Guido-Estrada ) report that a subset of for the microglial fractalkine receptor (Cx3cr1) in a mouse people with autism have a consistent pro-inflammatory model of Alzheimer's disease has the effect of preventing condition of the brain and cerebral spinal fluid and pro- the microglial destruction and phagocytosis of layer 3 posed that a systemic infection of the mother may lead to neurons that was observed in these mice in vivo with inflammation in the brain and autism. A recent report from 2-photon microscopy (Furhmann et al. In particular, Wei et al. ) found an increase of IL6 in cerebellar Cx3cr1 knockout mice have greater numbers of dendritic cortex in subjects with autism, which could alter cell spines in CA1 neurons, have decreased frequency sEPSCs migration and disrupt imperative circuits for normal and had seizure patterns which indicate that deficient development (Wei et al. In a mouse model of fractalkine signaling reduces microglia-mediated synaptic maternal infection for offspring brain development, it has pruning, leading to abnormal brain development, immature been reported that offspring from a mother having a single connectivity, and a delay in brain circuitry in the hippo- injection of IL6 during pregnancy alters fetal brain devel- campus (Paolicelli et al. ). In summary, the increased opment (Smith et al. ) which indicates that a maternal density of microglia in people with autism could be pro- infection can impact brain development and may play a tective against other aspects of this condition, and that a critical role in autism. Girard et al. ), using a lipo- possible side-effect of this protective response might polysaccharide (LPS) mouse model of maternal inflam- involve alterations in neuronal circuitry.
mation, found that a single treatment of an IL-1 receptorantagonist, concurrent with the LPS injection, had the Microglial Defects as Causes of Disease result that the IL-1 receptor antagonist protected againstmaternal placental inflammation and the offspring had By contrast, there are diseases that arise from intrinsic normal brain development. Furthermore, it is well docu- defects in the microglia themselves which can cause ster- mented that peripheral infection can dysregulate inflam- eotypic behavioral dysfunctions. A naturally occurring mation in the brain and increase monocyte infiltration into genetic defect in human microglia is the cause of a the cerebral cortex (D'Mello et al. ); it is also reported remarkable neuropsychiatric disease that was first observed that people with autism have elevated levels of cytokines in Japan and Finland, but has subsequently been found (Chez and Guido-Estrada which may disrupt the throughout the world. Nasu-Hakola disease is caused by a homeostatic balance in the cortex resulting in a greater defect in the gene TREM2 or DAP12 which together form density of microglia.
a receptor complex which is strongly expressed inmicroglia but not in astroglia or oligodendroglia (Paloneva Are Microglia Predators or Protectors? et al. Sessa et al. ). In the Allen Brain Atlas,DAP12 is preferentially expressed in olfactory, anterior The increased microglial densities we observed in the cingulate, and insular cortices in the mouse cortices of our subjects with autism appear to be a robust discriminator between the brains of people with autism in TREM2 or DAP12 impair the capacity of the microglia versus neurotypical brains, and these findings raise a major to phagocytose damaged tissue and increase the secretion question. Are markedly increased numbers of microglia of inflammatory cytokines in the olfactory, insular and J Autism Dev Disord (2012) 42:2569–2584 cingulate cortices resulting in microglia-mediated dementia these structures to microglial disruption. This vulnerability specific to these structures (Bianchin et al. ; Neumann might also be related to the preferential expression DAP12 and Takahashi ). Bianchin et al. report that at in the anterior cingulate and insular cortices. Area FI around age 35 in affected patients there are: ‘‘[i]ncipient investigated in our study corresponds to the ventral part of personality changes that can only be noticed by relatives anterior insular cortex. Thus, while changes in microglial and close friends. The behavioral alterations then become density appear to be widespread in brains of autistic indi- progressively more evident during the next months. The viduals, some areas may be more vulnerable than others to patients start to present silly and facetious behavior, lack of its effects.
insight, social inhibition, and other unrestrained behavior.
Sometimes they seem to have a euphoric attitude and are Visual Abnormalities in Autism easily distractible, seemingly lacking adequate associatedemotional components. As the disease progresses, the When we began this investigation we anticipated microglial patients evolve to a state of profound dementia.'' The abnormalities in the frontal cortices because many lines of remarkable behavioral specificity of the microglial defect evidence suggest that these structures are involved in autism in Nasu-Hakola disease shows that the microglia can (Allman et al. ; Courchesne and Pierce ; Di Mar- influence social behavior in a highly specific manner.
tino et al. ). We included visual cortex based first on its Another stereotypic behavioral defect arising from lack of involvement in prominent social and homeostatic abnormal microglia is obsessive grooming in mice with a functions and its physical distance from FI. Yet, abnormal- mutation of the gene Hoxb8 (Chen et al. Hoxb8 is ities in visual behavior are among the first signs of autism in expressed only in the microglia in the adult mouse brain, infancy. Beginning at the end of the first year, the earliest and these cells originate in spinal bone marrow (Chen et al.
signs of autism include atypical eye contact and visual ). When adult mice with the Hoxb8 mutation were tracking, and prolonged fixation, a tendency to perseverate irradiated so as to kill the bone marrow and then received visual attention on an original stimulus when presented with bone marrow transplants with the intact gene, the mice a competing stimulus (Zwaigenbaum et al. recovered from their excessive grooming pathology, their The increased microglial densities in visual cortex may skin lesions healed, and their fur grew back to normal.
be representative of a pan-cortical microglial phenotype When normal mice were irradiated and received bone related to the autistic phenotype associated with perceptual marrow from donors with the mutated Hoxb8 gene, they integration. In Kanner's original description of autism he developed the excessive grooming pathology. With these emphasized his patients' intense fixation on detail and experiments and a variety of other elegantly executed ‘‘inability to experience wholes without full attention to the controls, Chen et al. ) demonstrated that the Hoxb8 constituent parts'' as a characteristic feature of the disorder mutation with expression restricted to the microglia caused (Kanner ). Frith (drew attention to the tendency for typically developing children and adults to process resembles obsessive–compulsive disorder in humans, information for meaning and gestalt (global) form, often at which involves abnormalities in orbito-frontal and anterior the expense of attention to or memory for details and cingulate cortices (Graybiel and Rauch These surface structure. Happe and Frith (proposed that structures are also implicated in autism (Allman et al.
autistic subjects show ‘‘weak central coherence,'' a pro- ; Di Martino et al. Santos et al. Simms cessing bias favoring local information, and a relative et al. The excessive grooming in the Hoxb8 mice is failure to extract the gist or ‘‘see the big picture'' in also reminiscent of the stereotypical behaviors that are everyday life. The tendency of individuals with autism to commonly found in a subset of individuals diagnosed with focus on detail at the expense of global perceptions has autism (Goldman et al. been experimentally verified in many studies and may Together with the striking changes in social behavior account in part for impairments in the recognition of faces present in Nasu-Hakola disease, these data suggest that the (Behrmann et al. ; Happe and Frith ).
circuitry of anterior cingulate and orbito-frontal cortices This difficulty perceiving the gist or global features of a may be particularly sensitive to the disruptive effects of stimulus configuration by subjects with autism may be abnormal microglia. A strong association between reduced analogous to the difficulties experienced by subjects with activity in the anterior cingulate and anterior insular cor- autism in making rapid intuitive decisions (Allman et al.
tices (adjacent to orbito-frontal cortex) in social tasks in A variant of the ‘‘weak coherence'' theory applied to subjects with autism versus controls was revealed in a frontal lobe function and specifically linked to activated meta-analysis of 24 functional imaging studies (Di Martino microglia and their possible role in altering the development et al. and the reduced activity in these structures in of this structure was proposed by Courchesne and Pierce autism may also be related to the apparent vulnerability of ). Happe and Frith propose that ‘‘weak J Autism Dev Disord (2012) 42:2569–2584 coherence'' in autism is due to reduced connectivity leukoencephalopathy—PLOSL): A dementia associated with throughout the brain due to lack of synchronization of neural bone cystic lesions. From clinical to genetic and molecularaspects. Cellular and Molecular Neurobiology, 24, 1–24.
activity (Brock et al. ) or lack of connecting fibers (Just Blinzinger, K., & Kreutzberg, G. (1968). Displacement of synaptic et al. or lack of top-down connections (Frith ). A terminals from regenerating motoneurons by Microglial cells.
similar disruption of global connectivity might disrupt the Zeitschrift fu¨r Zellforschung und Mikroscopische Anatomie, 85, integrative functions of FI in social behaviors as suggested Brock, J., Brown, C. C., Boucher, J., & Rippon, G. (2002). The by the reduced activity in this area in subjects with autism temporal binding deficit hypothesis of autism. Development and versus controls (Allman et al. Di Martino et al. ).
Psychopathology, 4, 209–224.
An increased density of microglia throughout the cortex in Carson, M. J., Bilousova, T. V., Puntambekar, S. S., Melchior, B., people with autism might contribute to these phenomena Doose, J. M., & Ethell, I. M. (2007). A rose by any other name?The potential consequences of microglial heterogeneity during through alterations in the neuronal circuitry.
CNS health and disease. Neurotherapeutics, 4, 571–579.
Chen, S. K., Tvrdik, P., Peden, E., Cho, S., Wu, S., Spangrude, G., et al. (2010). Hematopoietic origin of pathological grooming inHoxb8 mice. Cell, 141, 775–785.
Chez, M. G., & Guido-Estrada, N. (2010). Immune therapy in autism: We found significantly increased microglial densities in historical experience and future directions with immunomodu- individuals with autism in two widely separated and latory therapy. Neurotherapeutics, 7, 293–301.
functionally different cortical areas, FI and VC. In light of Courchesne, E., & Pierce, K. (2005). Why the frontal cortex in autism our findings, as well as increased densities in the cerebel- might be talking only to itself: local over-connectivity but long-distance disconnection. Current Opinion in Neurobiology, 15, lum (Vargas et al., and dorsal lateral prefrontal cortex (Morgan et al. we propose that microglial Cullheim, S., & Thams, S. (2007). The microglial networks of the densities are elevated throughout the cerebral and cere- brain and their role in neuronal network plasticity after lesion.
bellar cortices in individuals with autism. Future studies are Brain Research Reviews, 55, 89–96.
Davalos, D., Grutzendler, J., Yang, G., Kim, J. V., Zuo, Y., Jung, S., needed to explain the mechanisms responsible for the et al. (2005). ATP mediates rapid microglial response to local increased densities and the relationship between this phe- brain injury in vivo. Nature Neuroscience, 8, 752–758.
nomenon and the behavioral manifestations of autism.
Dekaban, A. S. (1978). Changes in brain weights during the span of human life: Relation of brain weights to body heights and body This work was supported by grants from the weights. Annals of Neurology, 4, 345–356.
Simons Foundation (SFARI #137661), the James S. McDonnell Di Martino, A., Ross, K., Uddin, L., Sklar, A., Castellanos, F., & Foundation, and by NIH grant MH089406. The brain tissue and related Milham, M. (2009). Processes in autism spectrum disorders: An anonymous phenotypic information was obtained from the NICHD activation likelihood estimation meta-analysis. Biological Psy- Brain and Tissue Bank for Developmental Disorders. We especially chiatry, 65, 63–74.
thank Dr. Ronald Zielke, Robert Johnson and Melissa Davis for pro- D'Mello, C., Le, T., & Swain, M. G. (2009). Cerebral microglia viding the brain tissue and anonymous clinical records; our study recruit monocytes into the brain in response to tumor necrosis would not have been possible without their dedicated service. We thank factor alpha signaling during peripheral organ inflammation.
the anonymous reviewers for their helpful comments and criticisms.
Journal of Neuroscience, 29, 2089–2102.
Engel, S., Schluesener, H., Mittelbronn, M., Seid, K., Adjodah, D., Wehner, H. D., et al. (2000). Dynamics of microglial activationafter human traumatic brain injury are revealed by delayed expression of macrophage-related proteins MRP8 and MRP14.
Acta Neuropathologica, 100, 313–322.
Allman, J. M., Tetreault, N. A., Hakeem, A. Y., Manaye, K. F., Exton, M. S. (1997). Infection-induced anorexia: Active host defense Semendeferi, K., Erwin, J. M., et al. (2010). The von Economo strategy. Appetite, 29, 369–383.
neurons in frontoinsular and anterior cingulate cortex in great Frahm, H. D., Stephan, H., & Stephan, M. (1982). Comparison of apes and humans. Brain Structure and Function, 214, 495–517.
brain structure volumes in Insectivora and Primates: I, neocor- Allman, J., Watson, K., Tetreault, N., & Hakeem, A. (2005). Intuition tex. Journal fu¨r Hirnforschung, 23, 375–389.
and autism: A possible role for von Economo neurons. Trends in Frith, U. (2004). Is autism a disconnection disorder? Lancet Cognitive Science, 9, 367–373.
Neurology, 3, 577.
Ashwood, P., Wills, S., & Van der Water, J. (2006). The immune Furhmann, M., Bittner, T., Jung, C., Burgold, S., Ochs, S. M., response in autism: A new frontier for autism research. Journal Hoffman, N., et al. (2010). Microglial Cx3cr1 knockout prevents of Leukocyte Biology, 80, 1–15.
neuron loss in a mouse model of Alzheimer's disease. Nature Atlado´ttir, H. O., Thorsen, P., Østergaard, L., Schendel, D. E., Lemcke, Neuroscience, 13, 411–413.
S., Abdallah, M., et al. (2010). Maternal infection requiring Girard, S., Tremblay, L., Lepage, M., & Se´bire, G. (2010). IL-1 hospitalization during pregnancy and autism spectrum disorders.
receptor antagonist protects against placental and neurodevel- Journal of Autism and Developmental Disorders, 40, 1423–1430.
opmental defects induced by maternal inflammation. Journal of Behrmann, M., Thomas, C., & Humphreys, K. (2006). Seeing it Immunology, 184, 3997–4005.
differently: Visual processing in autism. Trends in Cognitive Goldberg, W. A., Osann, K., Filipek, P. A., et al. (2003). Language Science, 10, 258–264.
and other regression: Assessment and timing. Journal of Autism Bianchin, M. M., Capella, H. M., Chaves, D. L., Steindel, M., Grisard, and Developmental Disorders, 33, 607–616.
E. C., Ganev, G. G., et al. (2004). Nasu-Hakola disease Goldman, S., Wang, C., Salgado, M. W., Greene, P. E., Kim, M., & (polycystic lipomembranous osteodysplasia with sclerosing Rapin, I. (2009). Motor stereotypies in children with autism and J Autism Dev Disord (2012) 42:2569–2584 other developmental disorders. Developmental Medicine and (TREM2) for central nervous tissue immune homeostasis.
Child Neurology, 51, 30–38.
Journal of Neuroimmunology, 184, 92–99.
Graeber, M. B., Bise, K., & Mehraein, P. (1993). Synaptic stripping in Nimmerjahn, A., Kirchhoff, F., & Helmchen, F. (2005). Resting the human facial nucleus. Acta Neuropathologica, 86, 179–181.
microglial cells are highly dynamic surveillants of brain Graeber, M. B., & Streit, W. J. (1990). Microglia: Immune network in parenchyma in vivo. Science, 308, 1314–1318.
the CNS. Brain Pathology, 1, 2–5.
Paloneva, J., Manninen, T., Christman, G., Hovanes, K., Mandelin, J., Graeber, M. B., & Streit, W. J. (2010). Microglia: Biology and Adolfsson, R., et al. (2002). Mutations in two genes encoding neuropathology. Acta Neuropathologica, 119, 89–105.
different subunits of a receptor signaling complex result in an Graybiel, A. M., & Rauch, S. L. (2000). Toward a neurobiology of identical disease phenotype. American Journal of Human obsessive-compulsive disorder. Neuron, 28, 343–347.
Genetics, 71, 656–662.
Gundersen, H. J., Bendtsen, T. F., Korbo, L., Marcussen, N., Møller, Paolicelli R. C., Bolasco G., Pagani F., Maggi L., Scianni M., A., Nielsen, K., et al. (1988). Some new, simple and efficient Panzanelli P., et al. (2011) Synaptic pruning by microglia is stereological methods and their use in pathological research and diagnosis. Acta Pathologica, Microbiologica, et Immunologica 1456–1458. Epub 2011 Jul 21.
Scandinavica, 96, 379–394.
Perry, V. H. (2010). Contribution of systemic inflammation to chronic Happe, F., & Frith, U. (2006). The weak coherence account: detail- neurodegeneration. Acta Neuropathologica, 120, 277–286.
focused cognitive style in autism spectrum disorders. Journal of Santos, M., Uppal, N., Butti, C., Wicinski, B., Schmeidler, J., Autism and Developmental Disorders, 36, 5–25.
Giannakopolous, P., et al. (2011). Von Economo neurons in Hart, B. L. (1998). Biological basis of the behavior of sick animals.
autism: a stereological study of frontoinsular cortex in children.
Neuroscience and Biobehavioral Reviews, 12, 123–137.
Brain Research, 1380, 206–217.
Hirasawa, T., Ohsawa, K., Imai, Y., Ondo, Y., Akazawa, C., Uchino, Sasaki, Y., Ohsawa, K., Kanazawa, H., Kohsaka, S., & Imai, Y.
S., et al. (2005). Visualization of microglia in living tissues using (2001). Iba1 is an actin-cross-linking protein in macrophages/ Iba1-EGFP transgenic mice. Journal of Neuroscience Research, microglia. Biochemical and Biophysical Research Communica- 81, 357–362.
tions, 286, 292–297.
Imamoto, K., & Leblond, C. P. (1978). Radioautographic investiga- Schmid, C. D., Melchior, B., Masek, K., Puntambekar, S. S., Danielson, tion of gliogenesis in the corpus callosum of young rats. II.
P. E., Lo, D. D., et al. (2009). Differential gene expression LPS/ Origin of microglial cells. Journal of Comparative Neurology, IFNc activated microglia and macrophages: In vitro versus in vivo.
180, 139–163.
Journal of Neurochemistry, 109, 117–125.
Just, M. A., Cherkassky, V. L., Keller, T. A., & Minshew, N. J.
Sessa, G., Podini, P., Mariani, M., Meroni, A., Spreafico, R., (2004). Cortical activation and synchronization during sentence Sinigaglia, S., et al. (2004). Distribution and signaling of comprehension in high-functioning autism: Evidence of under TREM2/DAP12, the receptor system mutated in human poly- connectivity. Brain, 127, 1811–1821.
cystic lipomembraneous osteodysplasia with sclerosing leukoen- Kanner, L. (1968). Autistic disturbances of affective contact. Acta cephalopathy dementia. The European Journal of Neuroscience, Paedopsychiatrica, 35, 100–136.
20, 2617–2628.
Kreutzberg, G. W. (1996). Microglia: A sensor for pathological Simms, M. L., Kemper, T. L., Timbie, C. M., Bauman, M. L., & Blatt, events in the CNS. Trends in Neurosciences, 19, 312–318.
G. J. (2009). The anterior cingulate cortex in autism: Hetero- Li, X., Chauhan, A., Sheikh, A. M., Patil, S., Chauhan, V., Li, X. M., geneity of qualitative and quantitative cytoarchitectonic features et al. (2009). Elevated immune response in the brain of autistic suggests possible subgroups. Acta Neuropathologica, 118, patients. Journal of Neuroimmunology, 207, 111–116.
Loane, D. J., & Byrnes, K. R. (2010). Role of microglia in Smith, S. E., Li, J., Garbett, K., Mirnics, K., & Patterson, P. H.
neurotrauma. Neurotherapeutics, 7, 366–377.
(2007). Maternal immune activation alters fetal brain develop- Lyck, L., Santamaria, I. D., Pakkenberg, B., Chemnitz, J., Schrøder, ment through interleukin-6. The Journal of Neuroscience, 27, H. D., Finsen, B., et al. (2009). An empirical analysis of the precision of estimating the numbers of neurons and glia in Streit, W. J., Braak, H., Xue, Q.-S., & Bechmann, I. (2009).
human neocortex using a fractionator-design with sub-sampling.
Dystrophic (senescent) rather than activated microglial cells are Journal of Neuroscience Methods, 182, 143–156.
associated with tau pathology and likely precede neurodegener- MacDonald, R., Green, G., Mansfield, R., Geckeler, A., Gardenier, ation in Alzheimer's disease. Acta Neuropathologica, 118, N., Anderson, J., et al. (2007). Stereotypy in young children with autism and typically developing children. Research in Develop- Tetreault, N. A., Williams, B. A., Hasenstaub, A., Hakeem, A. Y., mental Disabilities, 28, 266–277.
Liu, M., Abelin, A. C. T., et al. (2009) RNA-Seq studies of gene Matson, J. L., & Lovullo, S. V. (2008). A review of behavioral expression in fronto-insular (FI) cortex in autistic and control treatments for self-injurious behaviors of persons with autism stuides reveal gene networks related to inflammation and spectrum disorders. Behavior Modification, 32, 61–76.
synaptic function. Program No. 437.3. 2009 Neuroscience Minio-Paluello, I., Baron-Cohen, S., Avenanti, A., Walsh, V., & Aglioti, Meeting Planner. Chicago, IL: Society for Neuroscience, 2009.
S. M. (2009). Absence of embodied empathy during pain observa- tion in Asperger syndrome. Biological Psychiatry, 65, 55–62.
Thomas, D. M., Francescutti-Verbeem, D. M., & Kuhn, D. M. (2006).
Mittelbronn, M., Dietz, K., Schluesener, H. J., & Meyeremann, R.
Gene expression profile of activated microglia under conditions (2001). Local distribution of microglia in the normal adult associated with dopamine neuronal damage. The FASEB Jour- human central nervous system differs by up to one order of nal, 20, 515–517.
magnitude. Acta Neuropathologica, 101, 249–255.
Vargas, D. L., Nascimbene, C., Krishnan, C., Zimmermann, A. W., & Morgan, J. T., Chana, G., Pardo, C. A., Achim, C., Semendeferi, K., Pardo, C. A. (2005). Neuroglial activtion and neuroinflammation Buckwalter, J., et al. (2010). Microglial activation and increased in the brains of patients with autism. Annals of Neurology, 57, microglial density observed in the dorsolateral prefrontal cortex in autism. Biological Psychiatry, 68, 368–376.
Voineagu, I., Wang, X., Johnston, P., Lowe, J. K., Tian, Y., Horvath, Neumann, H., & Takahashi, K. (2007). Essential role of the S., et al. (2011). Transcriptomic analysis of autistic brain reveals microglial triggering receptor expressed on myeloid cells-2 convergent molecular pathology. Nature, 474, 380–384.
J Autism Dev Disord (2012) 42:2569–2584 Wake, H., Moorhouse, A. J., Jinno, S., Kohsaka, S., & Nabekura, J.
brain and alters neural cell adhesion, migration and synaptic (2009). Resting microglia directly monitor the functional state of formation. Journal of Neuroinflammation, 19(8), 52.
synapses in vivo and determine the fate of ischemic terminals.
Zimmerman, A., Jyonouchi, H., Comi, A., Connors, S., Milstien, S., The Journal of Neuroscience, 29, 3974–3980.
Varsou, A., et al. (2005). Cerebrospinal fluid and serum markers Walters, A. S., Barrett, R. P., Feinstein, C., Mercurio, A., & Hole, W. T.
of inflammation in autism. Pediatric Neurology, 35, 195–201.
(1990). A case report of naltrexone treatment of self-injury and Zwaigenbaum, L., Bryson, S., Rogers, T., Roberts, W., Brian, J., & social withdrawal in autism. Journal of Autism and Developmen- Szatmari, P. (2005). Behavioral manifestations of autism in the tal Disorders, 20, 169–176.
first year of life. International Journal of Developmental Wei, H., Zou, H., Sheikh, A. M., Malik, M., Dobkin, C., Brown, W.
Neuroscience, 23, 143–152.
T., et al. (2011). IL-6 is increased in the cerebellum of autistic

Source: http://www.allmanlab.caltech.edu/pdfs/Tetreault2012.pdf