|Home | About | Journals | Submit | Contact Us | Français|
DiGeorge, or 22q11 Deletion Syndrome (22q11DS), the most common survivable human genetic deletion disorder, is caused by deletion of a minimum of 32 contiguous genes on human chromosome 22, and presumably results from diminished dosage of one, some, or all of these genes—particularly during development. Nevertheless, the normal functions of 22q11 genes in the embryo or neonate, and their contribution to developmental pathogenesis that must underlie 22q11DS are not well understood. Our data suggests that a substantial number of 22q11 genes act specifically and in concert to mediate early morphogenetic interactions and subsequent cellular differentiation at phenotypically compromised sites—the limbs, heart, face and forebrain. When dosage of a broad set of these genes is diminished, early morphogenesis is altered, and initial 22q11DS phenotypes are established. Thereafter, functionally similar subsets of 22q11 genes—especially those that influence the cell cycle or mitochondrial function—remain expressed, particularly in the developing cerebral cortex, to regulate neurogenesis and synaptic development. When dosage of these genes is diminished, numbers, placement and connectivity of neurons and circuits essential for normal behavior may be disrupted. Such disruptions likely contribute to vulnerability for schizophrenia, autism, or attention deficit/ hyperactivity disorder seen in most 22q11DS patients.
DiGeorge, or 22q11 Deletion Syndrome (22q11DS) is one of the most common “copy number variant” (CNV) genetic disorders currently known, occurring in an estimated 1/ 3000 live births (Shprintzen et al., 2005). The dysmorphology common to many 22q11DS phenotypes—cardiovascular, craniofacial, limb malformations as well as thymic dysplasia—suggests that heterozygous deletion of the minimal 1.5MB of chromosome 22 (or the larger, more typical 3MB deletion that includes this region; (Carlson et al., 1997), and resulting diminished dosage of 22q11 genes, disrupts early development. Similarly, the significantly elevated susceptibility of 22q11DS patients for schizophrenia, autism, attention deficit/ hyperactivity disorder (ADHD), language delay, and other behavioral disorders (Murphy et al., 1999, Niklasson et al., 2008), all of which are thought to be “neurodevelopmental disorders” that compromise cerebral cortical circuitry (Geschwind and Levitt, 2007, Weinberger, 1987, 1996), suggests that 22q11DS disrupts brain development in parallel with peripheral and visceral morphogenesis.
Despite these obvious signs that diminished 22q11 gene dosage disrupts critical developmental mechanisms, little is known about how such changes in 22q11 gene function result in the constellation of 22q11DS phenotypes. Over the past 10 years, work from this laboratory and several others has defined some functions of individual 22q11 genes, and identified some consequences of diminished 22q11 gene dosage for cardiovascular, limb, craniofacial, and brain development (Merscher et al., 2001, Maynard et al., 2003, Meechan et al., 2009, Sigurdsson et al., 2010). Our work suggests that 22q11 genes operate as functionally related “sets” at distinct times during development. Diminished dosage of a substantial set of 22q11 genes expressed at sites of mesenchymal/ epithelial interaction—which later become the 22q11DS phenotypic sites—likely compromises early morphogenesis in the brain, face, heart and limbs. Subsequently, diminished dosage of a subset of these 22q11 genes—particularly those that regulate the cell cycle—disrupts neurogenesis in the cerebral cortex. Finally, a distinct subset of mitochondrial 22q11 genes may compromise postnatal synaptogenesis. Such combined, recurrent function of 22q11 genes could disrupt differentiation from early embryonic through late postnatal development, especially in the brain. This concatenation of dosage-sensitive changes in development, especially if modified by additional genetic or environmental factors, could explain the variable behavioral pathology, as well as peripheral dysmorphology, associated with 22q11DS.
Most 22q11DS cases reflect spontaneous deletion of the 1.5 or 3MB region of human chromosome 22 that gives the disorder its name. There are, however, cases in which 22q11-deleted individuals (some lacking clinically obvious phenotypes) pass the deletion to their offspring with consequences from mild to severe (Leana-Cox et al., 1996). This provides additional evidence that genetic modifiers beyond the 22q11 genes themselves play a key role in determining 22q11DS phenotypic severity. The majority of deletions are most likely generated maternally during oogenic meiosis; however, paternal origin is also seen (Leana-Cox et al., 1996, Ryan et al., 1997). Despite speculation on potential maternal effects that modify 22q11DS phenotypes, especially in the brain (Eliez et al., 2001), there is currently no report of weak alleles generating effective nulls at specific loci, nor any evidence for paternal imprinting or allelic biasing that would result in loss of function of single 22q11 genes. Our analysis of parent of origin effects in the 26 contiguous murine orthologues of human 22q11 genes found on mouse chromosome 16 (Figure 1A;(Maynard et al., 2006)) found neither paternal nor maternal imprinting (often conserved between species; (Morison et al., 2005, Paulsen et al., 2001). Similarly, there are no substantial allelic biases that might render expression levels of 22q11 genes lower than the 50% expected for heterozygous deletion. Thus, the available evidence suggests that the primary molecular consequence of 22q11 deletion is diminished dosage of approximately 50% ((Meechan et al., 2006); Gopalikrishna et al., unpublished observations)) of multiple genes in the region.
The brain, along with other tissues, has diminished expression of most if not all 22q11 genes. We confirmed diminished expression of deleted 22q11 genes at the mRNA or protein level in the brains of mice carrying a deletion that parallels the minimal critical deletion in 22q11DS (Lgdel mice; Figure 1B,C, D; (Meechan et al., 2006)). mRNA levels for at least 10 22q11 genes with potentially important developmental functions are reduced by approximately 50% in the brains of late embryonic through adult Lgdel mice (Figure 1C). Histological analysis using in situ labeled preparations indicates that this change is more likely due to “per cell” alterations than changes in expression patterns (Figure 1B). Finally, diminished protein levels accompany diminished message levels for 4/ 5 genes for which we were able to quantify protein levels. The fifth gene, Ufd1l, shows clear translational dosage compensation, with Ufd1l protein levels reaching 100% of wild type in the Lgdel embryonic, postnatal and adult brain (Figure 1D). This unusual case of translational dosage compensation is consisten t with observations for Ufd1l or its orthologues in a broad range of species where 100% dosage is critical, and partial loss of function can be lethal (Johnson et al., 1995). Thus, the apparent immediate consequence of heterozygous 22q11 deletion in mouse models of the human genomic lesion is 50% diminished expression of most if not all 22q11 genes at critical phenotypic sites including limb buds, aortic arches and hearts, craniofacial primordia (branchial arches/ maxillary process) and the frontonasal mass/ forebrain (see below for details), as well as the developing and adult brain thereafter.
The constellation of 22q11DS phenotypes: dysmorphology of the limbs, aortic arches and heart, face and palate, and altered cognitive function—which implies altered forebrain, particularly cerebral cortical, development—suggests that a shared early developmental mechanism might be compromised by diminished dosage of one, some or all of the deleted 22q11 genes at each site. We hypothesized that this mechanism was most likely the interactions between mesenchyme (M) and adjacent epithelia (E) that drive morphogenesis at each site (Creazzo et al., 1998, LaMantia, 1999, Tickle and Eichele, 1994). Accordingly, we first evaluated whether a significant proportion of 22q11 genes from the minimal critical deleted region were expressed selectively or specifically in normal mouse embryos at established sites of M/ E interaction including the developing limb buds, aortic arches (cardiac primordia), branchial arches (facial primordia) and forebrain. A PCR-based screen of the entire set of murine 22q11 orthologues in micro-dissected limb buds, aortic arches/ hearts, branchial arches, and frontonasal mass/ forebrain from embryonic day 10 (E10.5) mouse embryos showed that a large subset of the murine 22q11 genes—22/ 28 orthologues—was expressed at phenotypically compromised sites of M/ E interaction (Figure 2A; (Maynard et al., 2003)). Further analysis using immunohistochemical or in situ hybridization in whole embryos showed that most of these genes are specifically or selectively expressed at sites of M/ E interaction compromised in 22q11DS (Figure 2B;(Maynard et al., 2002, Maynard et al., 2003, Meechan et al., 2006)). These observations provide a foundation for our hypothesis that a broad set of 22q11 genes normally act at sites of M/ E interaction to influence initial patterning and morphogenesis, and that diminished dosage of one, some or all of these genes during this epoch of development compromises initial development of each phenotypic site in 22q11DS.
To begin to test this hypothesis we first asked whether dosage of a subset of these genes—distributed throughout the region, including those whose function seemed to be likely to contribute significantly to early development—was actually diminished in a mouse model of 22q11DS. In Lgdel model embryos, we showed by quantitative PCR in microdissected limb buds, aortic arches and hearts, branchial arches, and frontonasal mass/ forebrains that 22q11 gene expression is indeed diminished by 50% in parallel at each site (Figure 2C; (Meechan et al., 2006)). In addition, we showed that this change was unlikely to reflect morphogenetic alteration in which tissue was lost since expression of the Snail transcription factors, which are associated with the neural crest component of the mesenchyme at each site compromised by 22q11DS (Sakai et al., 2006, Sefton et al., 1998) are not changed in expression patterns or levels, despite diminished levels—but not altered patterns—of 22q11 genes (Figure 2D). Even though there are conserved binding sites for Snail transcription factors in many upstream regulatory regions of the 22q11 genes (Meechan et al., 2006), there is little indication that Snail or any other local transcriptional regulator mediates dosage compensation at the mRNA expression level. In fact, we have only found one example thus far of dosage compensation for 22q11 genes: Ufd1l (see Figure 1;(Meechan et al., 2006)). Overall, in Lgdel embryos, message levels appear consistently diminished to approximately 50% of wild type at morphogenetic sites compromised in 22q11DS.
Together, these findings suggest, with few exceptions, diminished dosage of multiple 22q11 genes at sites of M/ E interaction can compromise local morphogenesis. Indeed, there is evidence of disrupted morphogenesis in the Lgdel mouse aortic arches ((Merscher et al., 2001), as well as skeletal elements in the limb, where we found that proximal and distal bones are modestly, but significantly shorter (p-value is between less than 0.01 and 0.0005) in deleted versus wild type littermates (Figure 3A,B). Nevertheless, there are no significant changes in the overall weight of Lgdel adult mice, their brain weights (Figure 3C,D), or gross morphological appearance of the mature Lgdel brain. There may be, however, subtle disruptions of patterning in the developing forebrain that parallel dysmorphology seen at other M/ E sites in the Lgdel mouse model of 22q11DS, as well as changes in cerebral cortical structure similar to those in 22q11DS patients (see below). Several reports in the literature compare disruptions of development at each 22q11DS phenotypic site with more extreme phenotypes seen in mutants where signaling pathways (Shh, Fgfs, Bmps, and retinoic acid-RA) that mediate M/ E interactions in the limbs, heart, face and forebrain are disrupted (Bachiller et al., 2003, Frank et al., 2002, Garg et al., 2001, Vermot et al., 2003, Washington Smoak et al., 2005). Thus, 22q11 genes might influence normal development either because they are targets of M/ E inductive signals—“downstream” of local signals and transcriptional regulators—or they themselves are “upstream” factors necessary to maintain normal mechanisms of local tissue-tissue interactions that guide morphogenesis and differentiation.
Autism, attention deficit/ hyperactivity disorder, and schizophrenia—the major behavioral disorders currently associated with 22q11 deletion—are believed to reflect altered cortical development, regardless of underlying genetic or etiologic diversity (Arnsten, 2006, McGlashan and Hoffman, 2000, Minshew and Williams, 2007). Imaging studies as well as limited post-mortem analysis of the cerebral cortex in 22q11Ds patients suggests that anatomical and cytological changes seen in the cerebral cortex in individuals with a variety of developmental disorders—including areal differences in cortical thickness polymicrogyria, and periventricular heterotopias—are also seen in 22q11DS patients (Bearden et al., 2007, 2009, Kiehl et al., 2009, Schaer et al., 2006, van Amelsvoort et al., 2004). These anomalies are thought to impact normal brain function, and in 22q11DS patients they may reflect sensitivity of cortical development to diminished 22q11 gene dosage. The mechanisms that normally insure appropriate numbers and locations of cortical neurons, including neurogenesis and migration, are now considered likely targets for pathogenesis in neurodevelopmental disorders (Bearden et al., 2009, Mao et al., 2009, Meechan et al., 2009). Thus, we used the Lgdel mouse model of 22q11DS to assess consequences of diminished 22q11 gene dosage for cortical neurogenesis and migration. The value of studies of cortical development in this mouse model goes beyond its direct relevance to 22q11DS—the Lgdel and 2 other 22q11DS models (Gogos et al., 1999, Long et al., 2006) represent rare opportunities to assess cortical development in model organisms with genomic lesions that directly recapitulate those seen in patients with major neurodevelopmental disorders including schizophrenia, ADHD, and autism.
We began our analysis of potential defects in cortical development by asking whether 22q11 genes continue to be expressed in the cortical rudiment during the last half of gestation in the mouse, when cortical neurogenesis and migration occur. Most 22q11 orthologues in the 1.5MB minimal critical deleted region are expressed in the developing brain from E12 through birth, and many are localized to the cerebral cortex (Maynard et al., 2003, Meechan et al., 2006, Meechan et al., 2009). Thus we evaluated functions of these 22q11 genes, first in silico, and subsequently in vitro and in vivo, to identify those that might influence neurogenesis or migration for further analysis of expression and function. We found a subset that apparently modulate the cell cycle: Ranbp1, Htf9C, Cdc45l, Hira, and Ufd1l. We recognized that Ufd1l was unlikely to make a significant contribution due to translational dosage comp ensation (see Figure 1D, and (Meechan et al., 2006)); however, the remaining cell cycle genes are all expressed selectively or preferentially in the ventricular and subventricular zone, where cortical neural stem cells are found (Figure 4A). Furthermore, most of these genes are expressed maximally during mid to late gestation—the peak period of cortical neurogenesis—and decline substantially thereafter (Meechan et al., 2006, 2009). The apparent function, localization, and expression dynamics of these genes suggests they might influence cortical neural progenitor proliferation. Thus, we asked whether proliferative characteristics of any of the sub-classes of progenitors in the cortical ventricular zone are disrupted when 22q11 gene dosage is altered.
We found that basal progenitors, a specific class of cortical precursor cells, are compromised by diminished dosage of 22q11 genes during mid-gestation (Figure 4B,C). Basal progenitors are thought to be the transit amplifying cells of the developing cortex (Noctor et al., 2004). They divide rapidly and asymmetrically, and their offspring are primarily projection neurons of cortical layers 2, 3 and 4 (Noctor et al., 2004). Basal progenitor proliferation is diminished in distinct regions of the embryonic Lgdel cortex, particularly anterior frontal regions; fewer appear to be in either S or M phase than in wild type littermates (Meechan et al., 2009). Similar changes are not seen in Prodh or Tbx1 mutant mice, despite the suggestion that these two individual 22q11 genes contribute significantly to 22q11DS-related pathology (Gogos et al., 1999, Merscher et al., 2001, Paylor et al., 2006). A consequence of this proliferative deficit in Lgdel embryos is, perhaps not surprisingly, diminished frequency of projection neurons in layers 2 through 4 (Figure 5C). Despite this deficit in projection neuron production, neurogenesis ceases at the same time in Lgdel and wild type cortex. Thus, diminished 22q11 dosage slows proliferation of basal progenitors initially; these cells never recover their full neurogenic capacity, and they yield fewer cortical projection neurons. We did not detect proliferative changes in the other main class of cortical progenitors, apical progenitors or radial glia, thought to be cortical neural stem cells. This suggests that the basal progenitor pool is generated appropriately; however, once generated these cells fail to progress normally through the series of rapid asymmetric cell divisions that result in appropriate numbers of projection neurons in the supragranular layers of mouse frontal association cortices.
As basal progenitors are failing to produce normal numbers of cortical projection neurons in discrete neocortical areas in the Lgdel mouse, tangential migration of embryonic interneurons from the basal forebrain into the same cortical areas is also disrupted (Figure 5A,B). As with embryonic basal progenitors, this developmental disruption results in a specific defect in the postnatal brain: in this case, altered distribution of parvalbumin immunoreactive interneurons between cortical layers (Figure 5D), although total numbers of parvalbumin-labeled cells were not diminished. Parvalbumin cells make up a large cohort of molecularly distinct interneuron sub-types in the adult rodent and primate cortex (Conde et al., 1994, Tamamaki et al., 2003). Post mortem studies of schizophrenic brains have indicated reductions in parvalbumin positive cortical cell numbers (Beasley et al., 2002) or diminished parvalbumin gene expression in the cortex (Hashimoto et al., 2003) indicating that the integrity of this class of interneuron may be compromised as part of schizophrenia pathogenesis (Benes, 2000, Lewis et al., 2005). Thus, in the Lgdel 22q11DS model, the initial migration and subsequent cortical distribution of a class of interneurons implicated in schizophrenic pathology is altered. Remarkably, the most substantial change in distribution of parvalbumin interneurons is a reduction in their frequency in cortical layer V (and apparent increase in layer II-IV). Layer V parvalbumin corresponds to a population of basket interneurons that heavily innervate layer 2-4 pyramidal cells (Kisvarday, 1992). The coincident reduction of layer 2-4 projection neurons as well as altered placement of interneurons in the Lgdel mouse model indicates that cortical circuit development is disrupted in multiple locations by diminished 22q11 gene dosage. Such disruptions may contribute to behavioral phenotypes reported in the Lgdel mouse model (Long et al., 2006; and see below), and suggest that developmental changes may underlie at least some of the vulnerability for psychiatric and behavioral disorders in 22q11DS.
The basal progenitor phenotype in Lgdel mice, the accompanying interneuron migratory deficit, and their potential consequences for cortical circuit development and function, provide new opportunities to further define potential developmental pathogenic mechanisms that contribute to diseases of cortical connectivity associated with 22q11DS. 22q11 gene dosage most likely influences a network of cell cycle regulators and other factors that insure appropriate numbers of cortical projection neurons as well as migratory targeting of GABAergic interneurons. Neither Tbx1 nor Prodh, both suggested as causal genes for cardiovascular or neural phenotypes in 22q11DS (Gogos et al., 1999, Merscher et al., 2001, Lindsay et al., 2001, Paylor et al., 2006) contribute to the basal progenitor proliferative defect nor its sequellae seen in the Lgdel mouse (Meechan et al., 2009). Among the remaining 22q11 genes, Cdc45l and Ranbp1 are attractive candidates for further study: both are robustly and specifically expressed in proliferative zones and their levels decline by 50% in the developing cortex of the Lgdel mouse. Furthermore, they have defined roles in proliferation: Cdc45l protein is a critical component of the Go Ichi Ni San-mini chromosome maintenance complex (GINS-MCM) that facilitates chromatin unwinding and polymerase activity during DNA replication, and Ranbp1 protein plays an important role in centrosomal cohesion and chromosomal alignment/ segregation during mitosis (Guarguaglini et al., 1997, Moyer et al., 2006, Tedeschi et al., 2007). Analysis of proliferative and migratory phenotypes in mice heterozygously deleted for Cdc45l, Ranbp1, or both, will clarify their contributions to disrupted cortical neurogenesis in the Lgdel 22q11DS mouse model.
Signaling molecules available in the developing cortex are also likely to contribute to the basal progenitor regulatory network that includes 22q11 genes. The generation, proliferation and maintenance of basal progenitors most likely rely on the integrity of Fgf signaling pathways (Kang et al., 2009, Sahara and O'Leary, 2009). Deletions distal to the 3MB 22q11 interval have been associated with basal progenitor defects via ERK/ FGF signaling (Samuels et al., 2008). Thus there may be a cluster of genes in the 22q11 region important for—and sensitive to—signaling pathways that influence basal progenitors. Fgf signaling molecules and receptors are selectively expressed in regions of the developing cortex. Accordingly, inappropriate regional expression or activity of these molecules due to diminished 22q11 gene dosage may disrupt basal progenitor proliferation. Such disruptions may actually begin during patterning of the anterior/ ventral forebrain—the source of cortical GABAergic interneurons—which relies upon local Fgf8, Fgf15 and Fgf17 signaling and the integrity of Fgf receptors (Borello et al., 2008, Cholfin and Rubenstein, 2007, Faedo et al., 2010, Storm et al., 2006). We have shown that the expression and patterning of Fgf8, as well as Fgf signaling activity, depends upon M/ E interactions between the frontonasal mesenchyme and the ventral forebrain (Tucker et al., 2008). Thus, it is possible that the altered cortical GABAergic migration in the Lgdel model of 22q11DS reflects not only disregulation of cortical development during late gestation, but anomalous patterning and inductive signaling at earlier stages of embryonic development.
Our current data provides a new outline of how a genomic lesion associated with vulnerability for autism, ADHD, schizophrenia, and other affective disorders might disrupt the initial generation and placement of key elements of cortical circuits during development. Nevertheless, several questions remain: We do not yet know whether the basal progenitor proliferative defect or interneuron migratory defect is cell autonomous. We do not know whether interneurons beyond the parvalbumin subpopulation have similarly altered migration and position, or perhaps are also modified in frequency. Finally, we do not know the functional consequences of the proliferative and migratory changes we have found. These cell biological changes might contribute to some of the subtle behavioral, synaptic and network changes reported recently in several 22q11DS mouse models (Mukai et al., 2004, 2008, Stark et al., 2008; Sigurdsson et al, 2010). Additional electrophysiological analysis in distinct regions of the postnatal cortex, paired with critical behavioral analyses that probe relevant mouse behaviors will be needed to determine whether localized abnormalities in neurogenesis and migration contribute to functional deficits relevant for the behavioral disorders that accompany 22q11DS.
The last phase of cortical circuit development—following neurogenesis and migration—encompasses dendritic and axonal growth, trophic-mediated competitive interactions that regulate cell survival, as well as establishment and activity-dependent stabilization of synaptic connections. The 22q11 minimal critical region lacks genes generally thought of as mediators of axonal growth, development, or survival; there are no ligands or receptors for neurotrophic factors, neural cell adhesions molecules, nor genes directly associated with process growth or apoptosis. A few 22q11 genes have indirectly been associated with these processes. Septin5 interacts with the SNARE complex at neuronal synapses (Beites et al., 2005), and over-expression apparently disrupts synaptic communication (Son et al., 2005). Initial reports, however, suggest that null mutation of Septin5 in mouse models has no identifiable consequences in the nervous system (Peng et al., 2002), although more recent reports suggest subtle, strain specific anomalies (Suzuki et al., 2009). Thus, if 22q11 genes influence synapse formation and processing capacity in cortical circuits, they most likely act indirectly, perhaps by regulating essential cellular mechanisms that impact synaptogenesis rather than adult synaptic function.
Our data indicates that 22q11 genes may influence the ubiquitous but essential role of mitochondrial function during cortical circuit development and maintenance, especially during activity dependent synaptogenesis. Synaptogenesis requires substantial metabolic support (Kazama et al., 2008), as does the ongoing maintenance of synaptic function. Mitochondrial activity in the rodent brain peaks at birth (Nakai et al., 2000), corresponding with a peak period in synapse formation. There is a similar peak of synapse formation in the early postnatal primate brain, including humans, with maximal synaptic density occurring just after birth, followed by a progressive refinement (and net loss) of synapses during adolescence (Huttenlocher, 1979, Rakic et al., 1986). Thus, our identification and characterization of 6 mitochondrial 22q11 genes—nearly 1/ 3 of the full set of 22q11 genes known to be expressed in the developing or mature brain—suggests that heterozygous 22q11 deletion might negatively influence mitochondrial function necessary for synaptogenesis. This hypothesis is further supported by the expression dynamics of the 22q11 mitochondrial genes: most are maximally expressed in the brain at or around birth, and several decline thereafter (Meechan et al., 2006). Accordingly, disruption of bioenergetic homeostasis during synapse formation and refinement due to diminished dosage of mitochondrial 22q11 genes might provide a final insult to an already disrupted history of neural development in 22q11DS.
To investigate the possible role of mitochondrial 22q11 genes in synapse development or other neuronal functions, we confirmed or established that protein products of 22q11 transcripts predicted to be mitochondrial by bioinformatic-based analysis were actually localized to mitochondria (Figure 6A). We next asked where these genes are expressed in the developing and mature brain–particularly in cortical neurons (Maynard et al., 2008). We found that each of this distinct subset of six mitochondrial 22q11 genes–Mrpl40, encoding a mitochondrial ribosomal subunit protein; Prodh, a key enzyme for amino acid catabolism as well as metabolism; Slc25a1, a mitochondrial citrate transporter; Txnrd2, a mitochondrial thioredoxin reductase; T10 which encodes a protein of unknown function, and Zdhhc8, whose localization and function includes mitochondrial regulation of cell survival as well as potential palmitoyl transferase activity—is robustly expressed in the developing and mature cortex, some with distinctive laminar distributions (see Figure 1). Moreover, many are detected biochemically in mitochondria from brain synaptosome fractions (Figure 6B). Zdhhc8 protein, in particular, is present in synaptic mitochondria, particularly those in glutamatergic synaptic terminals (Figure 6C). All 6 mitochondrial genes reach or maintain maximal expression around birth (P0)—concurrent with the period of peak synaptogenesis in the mouse—and most decline thereafter (Meechan et al., 2006). Reduced expression of 22q11 genes in the Lgdel mouse appears to influence ongoing function of cortical mitochondria as well: during early postnatal life (when the energetic demand is highest, as is the number of synapses), there is a dysregulation of normal mitochondrial gene expression, with several mitochondrial genes expressed at significantly higher levels in the cerebral cortex of the mouse model of 22q11DS (Figure 7). This dysregulation may represent compensation for reduced expression of 22q11 mitochondrial genes. The increased expression of other redox complex subunits may help to increase the overall metabolic output of mitochondria compromised by the genetic lesion that defines 22q11DS.
Of these six mitochondrial 22q11 genes, Prodh and Zdhhc8 have been most frequently implicated in 22q11-related behavioral phenotypes based on human genome-wide association studies (Chen et al., 2004, Kempf et al., 2008, Liu et al., 2002, Mukai et al., 2004) as well as analysis of mature mutant mice (Gogos et al., 1999, Mukai et al., 2004, 2008). There is little definitive data on the cellular function of Prodh protein, beyond an apparent role in proline metabolism and mitochondrial redox regulation (Gogos et al., 1999, Phang et al., 1980). Zdhhc8, in contrast, may have multiple functions including palmitoyl transferase activity, cell cycle regulatory capacity, and dosage dependent roles in mitochondrial-regulated cell survival (Maynard et al., 2008, Mukai et al., 2004, 2008). Zdhhc8 contains a short “DHHC” protein domain that is strongly conserved among a large (>20 member) protein family. Some DHHC domain-proteins are involved in palmitoylation: catalytic addition of palmitic acid residues that can modify the activity or localization of substrate proteins. Based upon homology to the first identified palmitoyl transferase, yeast ERF2 (Lobo et al., 2002), the DHHC domain has been confirmed as a catalytic site for mammalian palmitoyl transeferases (Huang et al., 2009, Iwanaga et al., 2009). Thus, Zdhhc8, which has palmitoyl transferase activity in vitro, is assumed to have similar activity in vivo. Nevertheless, Zdhhc8 has also been identified as a modulator of cell cycle progression at the G2/ M checkpoint in response to DNA damage (Sudo et al., 2007). Furthermore, we identified a potential biochemical interaction between Zdhhc8 and Uqcrc1, a mitochondrial core complex 1 subunit, and found that over expression of GFP- or HA-tagged Zdhhc8 in multiple cell lines rapidly leads to Bcl2-regulated apoptosis mediated by the intact DHHC domain (Figure 7) (Maynard et al., 2008). Finally, Zdhhc8 null mice have altered behavior and synaptic communication (Mukai et al., 2008), including decreased open field behavior, sensory gating deficits—surprisingly in females only—and decreased sensitivity to the NMDA/ glutamate receptor antagonist MK801 (Mukai et al., 2004, 2008). These disruptions have been associated with decreased ability of Zdhhc8 mutant neurons to form dendrites and spines in vitro. It is not yet clear, however, how Zdhhc8 might influence dendritic differentiation, or whether Zdhhc8 is required for maintenance of postsynaptic dendritic domains.
These diverse and uncertain functional associations suggest that Zdhhc8 protein may have distinct roles in neuronal development and subsequent homeostasis under different cellular conditions. Zdhhc8 may interact with electron transport machinery in mitochondria to influence bioenergetic capacity, cell growth or survival. The nature of the putative interaction between Zdhhc8 and Uqcrc1 has yet to be confirmed in neuronal cells. It is possible, however, that this interaction influences electron transport efficiency by modifying Uqcrc1 function and may thus modulate metabolic abilities of neurons to cope with energy demands of activity dependent synapse formation and stabilization. The role of palmitoylation in brain development and function has only begun to be explored. In addition to a presumed role in subcellular targeting, palmitoylation has been implicated in apoptosis (Guo et al., 2007, Peterson et al., 2008), apparently by directly modulating the generation of ceramides (Turpin et al., 2006). One possible explanation for multiple functions of Zdhhc8 is suggested by the nature of the palmitoyl transferase activity of ERF2, the DHHC orthologue in yeast. ERF2 apparently regulates palmitoylation of the yeast RAS homologue, resulting in translocation of RAS from the endoplasmic reticulum to the plasma membrane. In the absence of a third protein, however, ERF4, which directly interacts with ERF2, this palmitoylation does not occur (Zhao et al., 2002). Zdhhc8's cellular function may similarly depend upon its interactions with other cellular proteins. Zdhhc8 may have different functions when it interacts with GCP16 (a presumptive functional ERF4 homologue; Swarthout et al., 2005) in the endoplasmic reticulum than when it interacts with mitochondrial proteins like Uqcrc1. Changes in expression levels of interacting proteins in distinct cell classes, therefore, could mediate the apparent functional distinctions observed for Zdhhc8.
Aside from Zdhhc8, there are few specific indications of potential mechanistic roles for the remaining 22q11 mitochondrial genes during synaptic development. Indeed, one of these genes, T10, encodes a protein with no apparent homology to other known proteins. Nevertheless, the developmental modulation of expression levels, the localization of the 22q11 mitochondrial proteins to synapses, and the transient altered expression of additional mitochondrial genes in the Lgdel model of 22q11DS all indicate that 22q11 gene dosage—especially that of the mitochondrial 22q11 genes—can modulate mitochondrial function and metabolic capacity during early postnatal synaptogenesis.
It is not clear if disrupted cortical development in the Lgdel mouse results in specific behavioral deficits. Nevertheless, aberrant behaviors commonly associated with altered forebrain circuitry are seen in the Lgdel as well as other 22q11DSmouse models. The Lgdel as well as the Df(16) mouse, another 22q11DS genomic model (Stark et al., 2008), have deficits in pre-pulse inhibition (PPI), which reflects disrupted “sensory gating” (Long et al., 2008, Stark et al., 2008). Similar inabilities to “gate” or filter unnecessary sensory information are a hallmark of psychiatric illnesses including schizophrenia (Braff and Geyer, 1990). Accordingly, PPI deficits in Lgdel and Df(16) mice are consistent with the hypothesis that diminished 22q11 gene dosage results in altered cortical circuitry associated with behavioral pathology seen in schizophrenia. The Df(16) mouse also has deficits in conditioned fear responses, open field exploration, and t-maze tasks (Stark et al., 2008; Sigurdsson et al., 2010); however, conditioned fear and open field deficits are not seen in the Lgdel mouse (Long et al., 2008). Conditioned fear and spatial working memory tasks such as the t-maze rely upon coordinated activity of hippocampal and cortical circuits, and reduced hippocampal/ cortical oscillatory synchrony in Df(16) mice (Sigurdsson et al., 2010) suggests parallel disrupted synaptic or circuit function; nevertheless, the precise relationship between the behavioral deficits and the physiological disruption has not yet been defined. Variable deficits in Lgdel and Df(16) lines, however, highlight challenges in correlating genomic lesions, physiological changes, and behavior in mice which modest genetic and environmental variability influences performance. It is possible that more precise behavioral tasks that assess the murine equivalent of “executive function” (Bussey et al., 2001), which is thought to be a primary behavioral target of schizophrenia pathogenesis (Greene et al., 2008), may provide more stable, reproducible insight into specific deficits caused by diminished 22q11 gene dosage. Finally, caution must be applied when comparing behavioral deficits present in mice and disruptions in working memory, executive function, affect, and social interaction seen in patients with schizophrenia.
The correlation of 22q11DS with schizophrenia vulnerability, first reported in the early 1990's, provided one of the earliest associations between a consistent genomic lesion, and a behavioral disorder believed to reflect disordered cortical circuitry. In parallel with this genetic discovery—based upon completely independent arguments—several investigators elaborated the “neurodevelopmental hypothesis” of schizophrenia pathogenesis (Weinberger, 1987, 1996), and this hypothesis has since been broadened to include additional behavioral disorders—including autistic spectrum and ADHD (Geschwind and Levitt, 2007)—seen in 22q11DS patients. Thus, when considering the relationship between 22q11DS and behavioral disorders, one must ask whether schizophrenia, autism, or ADHD-like pathology in 22q11DS patients truly reflects aberrant brain development. As summarized above, a small number of studies of complete loss of function, or heterozygous deletion of individual genes from the 22q11 minimal (1.5MB) or typical (3MB) deleted region have shown anomalies in mouse behavior, circuit activity or synaptic function (Mukai et al., 2004, 2008, Stark et al., 2008). It is not clear from these studies, however, whether the anomalies are due to developmental changes, or acute change in neuronal function that impacts circuitry and behavior in the adult brain. The data from our laboratory summarized in this review indicates that diminished dosage of subsets of 22q11 genes can modify initial forebrain patterning, subsequent cortical neurogenesis and migration, and finally, mitochondrial support of activity-dependent synapse formation and elimination in the early postnatal brain. Each successive assault on normal developmental mechanisms might compound damage done previously. Thus, the cortex of 22q11DS patients may have neurons with altered identity due to aberrant specification of early precursors, altered frequency due to deficits in neurogenesis, altered positions due to disrupted migration, and altered connections due to dysregulation of mitochondrial function. Whether these departures from development underlie the behavioral disorders found in 22q11DS patients remains to be determined. Nevertheless, the coincidence of disrupted forebrain development, altered cortical circuitry and elevated incidence of schizophrenia, autism and ADHD— all considered to be “neurodevelopmental disorders”— argues strongly that 22q11DS is indeed a genetic disorder of neural development.
We thank Amanda Peters for assistance with various phases of the work done in our laboratory. DWM and TMM are recipients of NARSAD Young Investigator Awards. NICHD (HD42182 to A-S.L), NIMH (Silvio M. Conte Grant (MH64065), and NARSAD Independent Investigator Awards (ASL) have provided support. Confocal microscopy and RNA expression analysis utilized UNC Neuroscience Center core facilities (NS031768).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.