Genomic trade-offs: are autism and schizophrenia the steep price of the human brain?

Evolution often deals in genomic trade-offs: changes in the genome that are beneficial overall persist even though they also produce disease in a subset of individuals. Here, we explore the possibility that such trade-offs have occurred as part of the evolution of the human brain. Specifically, we provide support for the possibility that the same key genes that have been major contributors to the rapid evolutionary expansion of the human brain and its exceptional cognitive capacity also, in different combinations, are significant contributors to autism and schizophrenia. Furthermore, the model proposes that one of the primary genes behind this trade-off may not technically be "a gene" or "genes" but rather are the highly duplicated sequences that encode the Olduvai protein domain family (formerly called DUF1220). This is not an entirely new idea. Others have proposed that the same genes involved in schizophrenia were also critical to the rapid expansion of the human brain, a view that has been expressed as "the same 'genes' that drive us mad have made us human". What is new is that a "gene", or more precisely a protein domain family, has been found that may satisfy these requirements.

DUF1220 copy number is associated with schizophrenia risk and severity: implications for understanding autism and schizophrenia as related diseases.

The copy number of DUF1220, a protein domain implicated in human brain evolution, has been linearly associated with autism severity. Given the possibility that autism and schizophrenia are related disorders, the present study examined DUF1220 copy number variation in schizophrenia severity. There are notable similarities between autism symptoms and schizophrenia negative symptoms, and divergence between autism symptoms and schizophrenia positive symptoms. We therefore also examined DUF1220 copy number in schizophrenia subgroups defined by negative and positive symptom features, versus autistic individuals and controls. In the schizophrenic population (N=609), decreased DUF1220 copy number was linearly associated with increasing positive symptom severity (CON1 P=0.013, HLS1 P=0.0227), an association greatest in adult-onset schizophrenia (CON1 P=0.00155, HLS1 P=0.00361). In schizophrenic males, DUF1220 CON1 subtype copy number increase was associated with increased negative symptom severity (P=0.0327), a finding similar to that seen in autistic populations. Subgroup analyses demonstrated that schizophrenic individuals with predominantly positive symptoms exhibited reduced CON1 copy number compared with both controls (P=0.0237) and schizophrenic individuals with predominantly negative symptoms (P=0.0068). These findings support the view that (1) autism and schizophrenia exhibit both opposing and partially overlapping phenotypes and may represent a disease continuum, (2) variation in DUF1220 copy number contributes to schizophrenia disease risk and to the severity of both disorders, and (3) schizophrenia and autism may be, in part, a harmful by-product of the rapid and extreme evolutionary increase in DUF1220 copy number in the human species.

Replicated linear association between DUF1220 copy number and severity of social impairment in autism.

Sequences encoding DUF1220 protein domains exhibit an exceptional human-specific increase in copy number and have been associated with several phenotypes related to brain size. Autism is a highly heritable and heterogeneous condition characterized behaviorally by social and communicative impairments, and increased repetitive and stereotyped behavior. Given the accelerated brain growth pattern observed in many individuals with autism, and the association between DUF1220 subtype CON1 copy number and brain size, we previously investigated associations between CON1 copy number and autism-related symptoms. We determined that CON1 copy number increase is associated with increasing severity of all three behavioral features of autism. The present study sought to replicate these findings in an independent population (N = 166). Our results demonstrate a replication of the linear relationship between CON1 copy number and the severity of social impairment in individuals with autism as measured by Autism Diagnostic Interview-Revised Social Diagnostic Score, such that with each additional copy of CON1 Social Diagnostic Score increased 0.24 points (SE = 0.11, p = 0.036). We also identified an analogous trend between CON1 copy number and Communicative Diagnostic Score, but did not replicate the relationship between CON1 copy number and Repetitive Behavior Diagnostic Score. Interestingly, these associations appear to be most pronounced in multiplex children. These results, representing the first replication of a gene dosage relationship with the severity of a primary symptom of autism, lend further support to the possibility that the same protein domain family implicated in the evolutionary expansion of the human brain may also be involved in autism severity.

Generation of mice lacking DUF1220 protein domains: effects on fecundity and hyperactivity.

Sequences encoding DUF1220 protein domains show the most extreme human lineage-specific copy number increase of any coding region in the genome and have been linked to human brain evolution. In addition, DUF1220 copy number (dosage) has been implicated in influencing brain size within the human species, both in normal populations and in individuals associated with brain size pathologies (1q21-associated microcephaly and macrocephaly). More recently, increasing dosage of a subtype of DUF1220 has been linked with increasing severity of the primary symptoms of autism. Despite these intriguing associations, a function for these domains has not been described. As a first step in addressing this question, we have developed the first transgenic model of DUF1220 function by removing the single DUF1220 domain (the ancestral form) encoded in the mouse genome. In a hypothesis generating exercise, these mice were evaluated by 197 different phenotype measurements. While resulting DUF1220-minus (KO) mice show no obvious anatomical peculiarities, they exhibit a significantly reduced fecundity (χ(2) = 19.1, df = 2, p = 7.0 × 10(-5)). Further extensive phenotypic analyses suggest hyperactivity (p < 0.05) of DUF1220 mice and changes in gene expression levels of brain associated with distinct neurological functions and disease. Other changes that met statistical significance include an increase in plasma glucose concentration (as measured by area under the curve, AUC 0-30 and AUC 30-120) in male mutants, fasting glucose levels, reduce sodium levels in male mutants, increased levels of the liver functional indicator ALAT/GPT in males, levels of alkaline phosphatase (also an indicator of liver function), mean R and SR amplitude by electrocardiography, elevated IgG3 levels, a reduced ratio of CD4:CD8 cells, and a reduced frequency of T cells; though it should be noted that many of these differences are quite small and require further examination. The linking of DUF1220 loss to a hyperactive phenotype is consistent with separate findings in which DUF1220 over expression results in a down-regulation of mitochondrial function, and potentially suggests a role in developmental metabolism. Finally, the substantially reduced fecundity we observe associated with KO mice argues that the ancestral DUF1220 domain provides an important biological functionthat is critical to survivability and reproductive success.

DUF1220 protein domains drive proliferation in human neural stem cells and are associated with increased cortical volume in anthropoid primates.

Genome sequences encoding DUF1220 protein domains show a burst in copy number among anthropoid species and especially humans, where they have undergone the greatest human lineage-specific copy number expansion of any protein coding sequence in the genome. While DUF1220 copy number shows a dosage-related association with brain size in both normal populations and in 1q21.1-associated microcephaly and macrocephaly, a function for these domains has not yet been described. Here we provide multiple lines of evidence supporting the view that DUF1220 domains function as drivers of neural stem cell proliferation among anthropoid species including humans. First, we show that brain MRI data from 131 individuals across 7 anthropoid species shows a strong correlation between DUF1220 copy number and multiple brain size-related measures. Using in situ hybridization analyses of human fetal brain, we also show that DUF1220 domains are expressed in the ventricular zone and primarily during human cortical neurogenesis, and are therefore expressed at the right time and place to be affecting cortical brain development. Finally, we demonstrate that in vitro expression of DUF1220 sequences in neural stem cells strongly promotes proliferation. Taken together, these data provide the strongest evidence so far reported implicating DUF1220 dosage in anthropoid and human brain expansion through mechanisms involving increasing neural stem cell proliferation.

DUF1220 domains, cognitive disease, and human brain evolution.

We have established that human genome sequences encoding a novel protein domain, DUF1220, show a dramatically elevated copy number in the human lineage (>200 copies in humans vs. 1 in mouse/rat) and may be important to human evolutionary adaptation. Copy-number variations (CNVs) in the 1q21.1 region, where most DUF1220 sequences map, have now been implicated in numerous diseases associated with cognitive dysfunction, including autism, autism spectrum disorder, mental retardation, schizophrenia, microcephaly, and macrocephaly. We report here that these disease-related 1q21.1 CNVs either encompass or are directly flanked by DUF1220 sequences and exhibit a dosage-related correlation with human brain size. Microcephaly-producing 1q21.1 CNVs are deletions, whereas macrocephaly-producing 1q21.1 CNVs are duplications. Similarly, 1q21.1 deletions and smaller brain size are linked with schizophrenia, whereas 1q21.1 duplications and larger brain size are associated with autism. Interestingly, these two diseases are thought to be phenotypic opposites. These data suggest a model which proposes that (1) DUF1220 domain copy number may be involved in influencing human brain size and (2) the evolutionary advantage of rapidly increasing DUF1220 copy number in the human lineage has resulted in favoring retention of the high genomic instability of the 1q21.1 region, which, in turn, has precipitated a spectrum of recurrent human brain and developmental disorders.

The human homologue of the RNA polymerase II-associated factor 1 (hPaf1), localized on the 19q13 amplicon, is associated with tumorigenesis.

The 19q13 amplicon in pancreatic cancer cells contains a novel pancreatic differentiation 2 (PD2) gene (accession number AJ401156), which was identified by differential screening analysis. PD2 is the human homologue of the RNA polymerase II-associated factor 1 (hPaf1). In yeast, Paf1 is part of the transcription machinery, acting as a docking protein in between the complexes Rad6-Bre1, COMPASS-Dot1p, and the phosphorylated carboxyl terminal domain of the RNA polymerase II. As such, Paf1 is directly involved in transcription elongation via histone H2B ubiquitination and histone H3 methylation. The PD2 sequence is highly conserved from Drosophila to humans with up to 98% identity between rodent and human, suggesting the functional importance of PD2/hPaf1 to maintain cellular homeostasis. PD2 is a modular protein composed of RNA recognition motif, DEAD-boxes, an aspartic/serine (DS)-domain, a regulator of the chromosome condensation domain and myc-type helix-loop-helix domains. Our results further showed that PD2 is a nuclear 80 kDa protein, which interacts with RNA polymerase II. In addition, we have demonstrated that the overexpression of PD2 in the NIH 3T3 cells result in enhanced growth rates in vitro and tumor formation in vivo. Altogether, this paper presents strong evidence that the overexpression of PD2/hPaf1 is involved in cancer development.

High-throughput sequence identification of gene coding variants within alcohol-related QTLs.

Low initial response to alcohol has been shown to be among the best predictors of development of alcoholism. A similar phenotypic measure, difference in initial sensitivity to ethanol, has been used for the genetic selection of two mouse strains, the Inbred Long-Sleep (ILS) and Inbred Short-Sleep (ISS) mice, and for the subsequent identification of four quantitative trait loci (QTLs) for alcohol sensitivity. We now report the application of high throughput comparative gene sequencing in the search for genes underlying these four QTLs. To carry out this search, over 1.7 million bases of comparative DNA sequence were generated from 68 candidate genes within the QTL intervals, corresponding to a survey of over 36,000 amino acids. Eight central nervous system genes, located within these QTLs, were identified that contain a total of 36 changes in protein coding sequence. Some of these coding variants are likely to contribute to the phenotypic variation between ILS/ISS animals, including sensitivity to alcohol, providing specific new genetic targets potentially important to the neuronal actions of alcohol.

Comparison of global brain gene expression profiles between inbred long-sleep and inbred short-sleep mice by high-density gene array hybridization.

BACKGROUND: Inbred long-sleep (ILS) and short-sleep (ISS) mice show significant central nervous system-mediated differences in sleep time for sedative dose of ethanol and are frequently used as a rodent model for ethanol sensitivity. In this study, we have used complementary DNA (cDNA) array hybridization methodology to identify genes that are differentially expressed between the brains of ILS and ISS mice. METHODS: To carry out this analysis, we used both the gene discovery array (GDA) and the Mouse GEM 1 Microarray. GDA consists of 18,378 nonredundant mouse cDNA clones on a single nylon filter. Complex probes were prepared from total brain mRNA of ILS or ISS mice by using reverse transcription and 33P labeling. The labeled probes were hybridized in parallel to the gene array filters. Data from GDA experiments were analyzed with SQL-Plus and Oracle 8. The GEM microarray includes 8,730 sequence-verified clones on a glass chip. Two fluorescently labeled probes were used to hybridize a microarray simultaneously. Data from GEM experiments were analyzed by using the GEMTools software package (Incyte). Differentially expressed genes identified from each method were confirmed by relative quantitative reverse transcription-polymerase chain reaction (RT-PCR). RESULTS: A total of 41 genes or expressed sequence tags (ESTs) display significant expression level differences between brains of ILS and ISS mice after GDA, GEM1 hybridization, and quantitative RT-PCR confirmation. Among them, 18 clones were expressed higher in ILS mice, and 23 clones were expressed higher in ISS mice. The individual gene or EST's function and mapping information have been analyzed. CONCLUSIONS: This study identified 41 genes that are differentially expressed between brains of ILS and ISS mice. Some of them may have biological relevance in mediation of phenotypic variation between ILS and ISS mice for ethanol sensitivity. This study also demonstrates that parallel gene expression comparison with high-density cDNA arrays is a rapid and efficient way to discover potential genes and pathways involved in alcoholism and alcohol-related physiologic processes.

Discovery of a new Pomc1 allele in the LS x SS recombinant inbred strains: relationship to locomotor behavior.

Various lines of evidence suggest that a polymorphism in the gene for proopiomelanocortin, Pomc1, might account for some portion of the genetic variance for open-field activity in the LS x SS RI strains. To test this hypothesis, approximately 1600 bp of Pomc1 was sequenced from genomic DNA of seven of the LS x SS strains. Two distinct alleles containing a total of five single-base pair differences were detected. A substitution was found in the coding region causing a Pro-to-Ser conversion, two substitutions occurred in the 3' untranslated region of the mRNA, and a substitution and a deletion were detected in the 3' untranscribed flanking region. The fragment containing the coding region substitution was sequenced in an additional 15 of the LS x SS strains. A total of 12 strains contained one form of the allele, while 10 had the other. The strain distribution pattern of open-field activity scores between the two allels suggests that these alleles do not contribute to the genetic variation of open-field activity in the LS x SS RI strains.

Genomic structure and chromosomal localization of a human myo-inositol monophosphatase gene (IMPA).

Manic-depressive illness is a serious psychiatric disorder that in many, but far from all, patients can be treated with lithium. The main causes for discontinuation of lithium therapy are unpleasant or serious side effects and lack of response. The reason for the striking variation in clinical efficacy of lithium treatment among bipolar patients is not known. The enzyme myo-inositol monophosphatase (IMPase) has been postulated as a target for the mood-stabilizing effects of lithium, but variation in the coding region of the human IMPA gene encoding IMPase activity has not been observed in manic-depressive patients (Steen et al., Pharmacogenetics, 1996, 6, 113-116). It is nevertheless conceivable that polymorphisms or mutations in the noncoding regions of this gene could influence the lithium response in psychiatric patients. As a first step in investigating this possibility, we here report the genomic structure of the human IMPA gene. The gene is composed of at least nine exons and covers more than 20 kb of sequence on chromosome 8q21.13-q21.3. In the 3'-untranslated part of the gene, we observed a polymorphism (a G to A transition) and also two short sequences similar to the inositol/cholin-responsive element consensus. Finally, we postulate that two additional IMPA-like transcripts originate from the human genome, one from a position close to IMPA itself on chromosome 8 and the other from chromosome 18p. Our data may contribute to the identification of genetic factors involved in the pathogenesis and determination of treatment response in manic-depressive illness.

Human chromosome 1 localization of the gene for a prostaglandin F2alpha receptor negative regulatory protein.

A protein that copurifies with the bovine prostaglandin F2alpha (FP) receptor has been isolated and the corresponding rat cDNA has been cloned. Transfection experiments suggest that this protein inhibits binding of [3H]prostaglandin F2alpha ([3H]PGF2alpha) to FP. Histologically, this protein (FP regulatory protein or FPRP) shows a distribution coinciding well with those cells and tissues that respond to PGF2alpha. A portion of the 3' untranslated region of the human homolog to fprp was subcloned, sequenced, and oligonucleotide primers chosen that allow polymerase chain reaction (PCR) amplification specifically of the human fprp sequence. These primers were then used in a PCR-based mapping-protocol. The human fprp gene was first socalized through human/rodent somatic cell hybrids to human chromosome 1 (100% concordance), and further through yeast artificial chromosome (YAC) pools to region 1p13.1-q21.3 (level 1 mapping). In view of the specific histologic localization of this negative regulator, possible pathological conditions are mentioned that may cosegrepate with this chromosomal region.

A strategy for the identification of candidate genes for alcohol-related phenotypes and other human disorders using rapid polymerase chain reaction mapping of gene-based sequence-tagged sites.

We describe a method for the rapid identification and mapping of human genes, including those possibly contributing to disease and alcohol-related phenotypes. New human genes are identified from cDNA libraries through single-pass sequencing into the 3' untranslated (3'UT) regions of human brain cDNAs. Primers derived from the 3'UT region sequences [representing gene-based, sequence-tagged sites (STSs)] are used for polymerase chain reaction (PCR) analyses of the CEPH megabase insert yeast artificial chromosome (YAC) DNA pools. With this approach, approximately 18,000 megabase YACs can be screened and a single YAC identified using only 52 PCR reactions. The YAC localization in conjunction with other mapping techniques, such as PCR mapping to human chromosomes using somatic cell hybrids, allows identification of chromosomal band locations. In this manner, each gene can be associated with its own STS, which in turn specifies both a corresponding genomic clone and specific location in the genome. These locations can be compared with the purported locations of disease genes. The locations of the STSs can also be compared with those of Quantitative Trait Loci implicated for quantitative traits (e.g., alcohol-related phenotypes) on the basis of synteny between the mouse and human genes. Using this strategy, we found candidates for 78 human disease/syndrome genes among the first 220 genes mapped.

Gene-based sequence-tagged-sites (STSs) as the basis for a human gene map.

Using our data set of 3,143 single pass sequences from human brain cDNA libraries, we have developed a strategy in which gene-based sequence-tagged-sites (STSs), derived from 3'untranslated regions of human cDNAs, are rapidly assigned to megabase-insert yeast artificial chromosomes and somatic cell hybrids to generate regional gene mapping data. Employing this approach, we have mapped 318 cDNAs, representing 308 human genes. Ninety-two of these mapped to regions implicated in human genetic diseases, identifying them as candidate genes. Extension of this strategy has the potential to result in virtually every human gene having, at its 3' end, its own associated STS, with each STS in turn specifying both a corresponding genomic clone and a specific regional location in the genome.

The gamma-aminobutyric acid receptor gamma 3 subunit gene (GABRG3) is tightly linked to the alpha 5 subunit gene (GABRA5) on human chromosome 15q11-q13 and is transcribed in the same orientation.

GABAA receptors are heterooligomeric ligand-gated ion channels that mediate the effect of the inhibitory neurotransmitter gamma-aminobutyric acid. The GABAA receptors consist of at least 15 different receptor subunits that can be classified into 5 subfamilies (alpha, beta, gamma, delta, rho) on the basis of sequence similarity. Chromosomal mapping studies have revealed that several of the GABAA receptor subunit genes appear to be organized as clusters. One such cluster, which consists of the GABAA receptor beta 3 (GABRB3) and alpha 5 (GABRA5) subunit genes, is located in chromosome 15q11-q13. It is shown here that the GABAA receptor gamma 3 subunit gene (GABRG3) also maps to this region. Lambda and P1 phage clones surrounding both ends of GABRG3 were isolated; the clones derived from the 5' end of GABRG3 were linked to an existing phage contig spanning the 3' end of GABRA5. The two genes are located within 35 kb of each other and are transcribed in the same orientation.

Localization of the gene for a novel human adenylyl cyclase (ADCY7) to chromosome 16.

A novel form of human adenylyl cyclase (ADCY7) has been discovered in the human erythroleukemia cell line (HEL). This cell line has been widely used as a model for studies of the characteristics of human platelets. Data from HEL cells suggests that ADCY7 may be the major AC form in human platelets. In the current study polymerase chain reaction (PCR) techniques coupled with use of human/rodent somatic hybrid panels and a yeast artificial chromosome (YAC) library were used to determine the chromosomal localization of the gene (adcy7) for ADCY7 enzyme. A 251-bp product from the 3' untranslated region of human adcy7 was amplified for PCR mapping and the results localize the adcy7 gene to region 16q12-16q13 of the human genome. The AC enzyme family is characterized by the presence of 12 membrane-spanning domains in its sequences, and this chromosomal region is known to contain other genes coding for proteins characterized by 12 membrane-spanning domains.

Cloning, functional coexpression, and pharmacological characterisation of human cDNAs encoding NMDA receptor NR1 and NR2A subunits.

Using expression cloning, and more recently using polymerase chain reaction cloning approaches, a family of rat N-methyl-D-aspartate (NMDA) receptor subunit cDNAs has been described (NR1, NR2A, NR2B, NR2C, and NR2D). Here we report cloning and sequencing of cDNAs encoding isoforms of the human NR1 subunit (NR1a, NR1d, and NR1e) that differ at their C-terminal end as a result of alternative splicing and also of a cDNA encoding the human NR2A subunit. The deduced amino acid sequences of the human NR1 subunit isoforms differed from the published rat NR1 subunit sequences at only eight positions, all of which were N-terminal to the alternatively spliced domains. The human NR2A subunit deduced amino acid sequence differed from the published rat NR2A subunit sequence at 81 of the 1,464 amino acids, with most of the substitutions being located in the C-terminal half of the subunit. The gene for NR2A has been localised to human chromosome 16. We also report the expression and pharmacological characterisation of recombinant human NR1a/NR2A heteromeric receptors in Xenopus oocytes. These receptors had EC50 values of 2.14 and 2.05 microM for glutamate and glycine, respectively, and an IC50 of 46.8 microM for Mg2+. Responses were antagonised by D-2-amino-5-phosphonovalerate, L-689,560, pH 6.3, zinc, and MK-801. No modulatory effect was observed on application of ifenprodil, confirming previous observations with rat NR1 + NR2A recombinant receptors.

Role of the beta subunit in determining the pharmacology of human gamma-aminobutyric acid type A receptors.

A cDNA encoding the human gamma-aminobutyric acid (GABA)A receptor beta 2 subunit has been cloned and sequenced. The deduced amino acid sequence of this cDNA shows only a single amino acid change from the rat sequence (Asn-347 in rat, serine in human). Using polymerase chain reaction amplification of human-specific products from human x rodent somatic cell hybrid DNAs, the gene has been assigned to human chromosome 6. By expressing recombinant human GABAA receptors containing different beta subunits (beta 1, beta 2 or beta 3) in both transfected cells and Xenopus oocytes, we have been able to determine the influence of the beta subunit on the pharmacology of the receptor. For a number of benzodiazepine binding site compounds, a barbiturate, and several neurosteroids, neither the affinity nor the efficacy of the compounds is influenced by the type of beta subunit present in the receptor molecule. These data suggest that the beta subunit does not significantly influence the benzodiazepine, barbiturate, or steriod site pharmacologies of human GABAA receptor subtypes.