Negative symptoms encompass diminution in emotional expression and motivation, some of which relate to human attributes that may not be accessible readily in animals. Additionally, their refractoriness to treatment precludes therapeutic validation of putative models. This review considers critically the application of mutant mouse models to the study of the pathobiology of negative symptoms. It focuses on 4 main approaches: genes related to the pathobiology of schizophrenia, genes associated with risk for schizophrenia, neurodevelopmental-synaptic genes, and variant approaches from other areas of neurobiology. Despite rapid advances over the past several years, it is clear that we continue to face substantive challenges in applying mutant models to better understand the pathobiology of negative symptoms: the majority of evidence relates to impairments in social behavior, with only limited data relating to anhedonia and negligible data concerning avolition and other features; even for the most widely examined feature, social behavior, studies have used diverse assessments thereof; modelling must proceed in cognizance of increasing evidence that genes and pathobiologies implicated in schizophrenia overlap with other psychotic disorders, particularly bipolar disorder. Despite the caveats and challenges, several mutant lines evidence a phenotype for at least one index of social behavior. Though this may suggest superficially some shared relationship to negative symptoms, it is not yet possible to specify either the scope or the pathobiology of that relationship for any given gene. The breadth and depth of ongoing studies in mutants hold the prospect of addressing these shortcomings.
While it is widely accepted that negative symptoms in schizophrenia constitute a major, pernicious cause of functional debility, and impaired quality of life, there is considerably less agreement on a number of related challenges that impact directly on attempts to model such psychopathology in rodents in general and in genetically modified mice in particular; eg, clinical debates endure as to the nature of “primary” vs “secondary” negative symptoms and the relationship between negative symptoms and a putative “deficit syndrome.”1–3 Until such clinical debates are resolved, it will not be possible to seek fully homologous or isomorphic models of these or, indeed, any other domains of psychopathology in schizophrenia.
In general terms, negative symptoms encompass diminution in emotional expression and motivation, some of which relate to human attributes that may not be accessible readily in animals; this has long been recognized as highly problematic.4,5 Additionally, uncertainty as to the pathophysiological basis of negative symptoms, together with their essential refractoriness to any treatment modality,2,6 impedes “proxy” approaches and precludes therapeutic validation of putative models. The difficulties for mutant mouse studies created by such general issues are exacerbated on considering more specific challenges.
The Negative Symptom Challenge
Scope of Negative Symptoms
The domain of negative symptoms is widely held to encompass features such as anhedonia, avolition, blunted affect, poverty of speech [alogia], and social withdrawal [asociality] and to be distinguishable both phenomenologically and psychometrically from their positive symptom counterparts.2,3 However, while factor analytic studies consistently resolve such negative symptoms into a domain of psychomotor poverty that is distinct from the positive symptom domains of reality distortion and disorganization, there is less clarity as to whether psychomotor poverty is itself a unitary or polydimensional domain; there is some evidence to suggest at least 2 negative symptom domains: diminished expression (blunted affect and poverty of speech) and anhedonia-asociality.1,7 Thus, the challenge posed is whether mutant studies are seeking to illuminate the basis of a single construct or the bases of diverse constructs.
Relationship to Cognitive Dysfunction
An associated challenge is the relationship of negative symptoms to cognitive dysfunction. While evidence indicates that both constructs contribute importantly to functional impairment and may bear some psychometric relationship to each other, this relationship is weak and varies with the domain of cognition at issue.1,8 Thus, the challenge posed is the extent to which mutant studies relating to cognition (see Arguello and Gogos, this issue) inform on processes bearing some relationship to negative symptoms and their putative pathophysiology, perhaps, in terms of some shared involvement of cortico-striato-pallido-thalamo-cortical network dysfunction/dysconnectivity9–12 or on an independent process in schizophrenia that is unrelated to negative symptoms.
Specificity of Negative Symptoms
Another fundamental challenge is whether the concept of negative symptoms, however defined, is specific to schizophrenia or applies also to other neuropsychiatric disorders. There is evidence for the identification of negative symptoms, or at least negative symptom-like features, also in depression and Parkinson disease.13 Thus, as above, the challenge posed is the extent to which mutant studies relating to disorders such as depression and Parkinson disease may inform on processes bearing some relationship to negative symptoms in schizophrenia and their putative pathophysiology, perhaps, in terms of some shared involvement of cortico-striato-pallido-thalamo-cortical network dysfunction11,13,14 or on independent processes unrelated to negative symptoms in schizophrenia.
It is on this complex and uncertain clinical background that molecular genetics, neurobiology, and behavioral neuroscience converge. Their conjoint purpose is the phenotypic study of mice mutant for genes associated with aspects of the putative pathophysiology of or risk for schizophrenia that may inform on the basis of negative symptoms and indicate novel therapeutic targets.
Modelling Negative Symptoms in Animals
Certain negative symptoms, such as poverty of speech, are extremely difficult to model in animals; indeed, they may be uniquely human conditions.15,16 In contrast, anhedonia, asociality, and avolition represent constructs that, at least theoretically, apply to and are accessible in both humans and animals. However, while many such behaviors in rodents may possess superficial similarity to those observed in patients, whether a given model system is homologous to or isomorphic with the human condition is dependent primarily on our understanding of (1) the underlying taxonomy of “core emotional tendencies,” (2) their molecular/cellular bases, and (3) the extent to which these processes are conserved across species and then expressed across a diversity of species-specific behaviors.17,18
Deficits in social functioning represent a core negative symptom in schizophrenia2,3 and constitute perhaps primary focus, as disturbances in social behavior, particularly social withdrawal, provide a quantifiable “negative symptom” readily amenable to modelling in animals. However, where a given animal model indicates impairment in social interaction, this may confer the model with face validity only for this symptom type because these deficits may alternatively reflect changes across several emotional and cognitive domains in both human and rodents. The latter consideration may be addressed, at least partly, by employing a comprehensive phenotyping strategy capable of capturing and assessing multiple domains and several aspects within each domain, eg, social approach behavior, aggression, and social cognition.19
Social approach-avoidance behaviors of putative relevance to schizophrenia are typically measured in rodents by distance between 2 unfamiliar animals placed in a novel environment or the time a pair spend engaged in a defined species-specific element of “active” social interaction. Such assessments of social interaction in a novel environment have generally been conducted across studies using established protocols20,21; these typically involve use of automated analysis with appropriate object tracking software to provide indices such as inter-animal distance and contact time, with complementary analysis using a time-sampling procedure to score social behaviors according to the presence or absence of a set of species-typical affiliative (eg, investigative sniffing) or agonistic (eg, biting, pinning) behaviors.
Analysis of free social interaction in a novel environment is subject to certain caveats and methodological considerations. First, in a dyadic paradigm, the social encounter can be initiated by either mouse, while in a social choice paradigm (see “Social Choice” section) the experimental mouse initiates the social encounter. Second, when social interaction tasks are conducted in a novel environment an effect of treatment or genotype on response to novelty may modulate social behavior. Third, it has been argued that impairment in social functioning in schizophrenia may reflect several other factors, including anhedonia, anxiety, or deficits in social cognition.19 Finally, as many rodent models of social withdrawal were developed as screens for anxiogenic/anxiolytic drug activity,22 genotype- or treatment-related effects on social behavior may also reflect a change in anxiety, emphasizing a requirement for multiple construct measures and/or manipulation of experimental parameters known to alter the anxiety component in such tasks.23
Choice paradigms for affiliative behaviors are now commonly used to test interest to engage in social interaction in mouse mutant models related to schizophrenia and other psychiatric disorders that are characterized by profound impairment in social interaction.24,25 Social choice tasks have the advantage that they rely upon spontaneous behaviors, thereby requiring no previous training. Social affiliative behavior is typically assessed in an apparatus with 3 interconnected chambers, with 2 dividing walls containing doors allowing access to each of the side chambers. Sequentially, the test mouse is allowed to freely explore (1) a chamber containing an unfamiliar conspecific vs an empty chamber (ie, to study sociability), then (2) a chamber containing an unfamiliar conspecific vs a chamber containing a familiar conspecific (ie, to study preference for social novelty). The sociability phase reflects social approach-avoidance behavior, while the social novelty phase assesses social recognition memory and the ability to discriminate and respond appropriately to a socially novel stimulus.
This task has now been well characterized in terms of mouse strain differences.26,27 It has also been shown that sociability in this task correlates well with frequency of social investigative behaviors in free social interaction assays.24 A number of factors have been identified which may influence the behavior of the test mouse in a social choice paradigm. As social recognition in mice is highly dependent upon olfactory sensory control, it is important to control for phenotypic or treatment effects on olfaction. Social approach behavior in these paradigms may also be influenced by the test animal's appraisal of each conspecific, eg, in terms of social status or aggression.
It has been suggested that impaired social functioning in schizophrenia involves impaired interplay between different dysfunctional cognitive domains relating to processing and interpreting social cues, ie, social cognition.28 Social memory or social recognition has also been typically assessed in a 2-stage procedure: the test animal is first introduced to an unfamiliar (usually juvenile) conspecific for a brief period, during which social behaviors are scored, followed 30 min later by a second stage, during which both animals are reintroduced and social behaviors again recorded; a reduction in social exploration following the interval reflects integrity of social memory.29,30 Assessment of recognition memory using social recognition-discrimination paradigms provides a parsimonious index of memory because the task relies on spontaneous exploratory behavior and does not require additional stimuli; this avoids the complication of interpreting data involving conditional and unconditional stimuli.
When aggression is present in schizophrenia, the nature of its relationship to psychopathology and cognitive dysfunction is unclear.31,32 In rodents, 5 varieties of behavior have been studied under the rubric of aggression: (a) play fighting, (b) offensive aggression, (c) defensive aggression, (d) maternal aggression, and (e) predatory aggression.33 Although numerous procedures have been offered for assessing offensive and defensive aggression in rodents,33–35 few of these paradigms or the investigators employing them distinguish between the varieties of aggressive behavior outlined above.
Typically, aggressive behavior in rodents is assessed via dyadic interaction where the test animal is confronted with an unfamiliar conspecific. Two situations commonly used involve a neutral setting (ie, a clean, unfamiliar cage) or the home cage (ie, a “resident-intruder” procedure). Factors which influence the display of offensive or defensive aggression include strain of the test subject, size of the area used in the encounter, duration of isolation of test subject and rearing conditions,36 social status of conspecific, and age and sex of both parties.37 Assessment of aggressive behavior is now commonly employed as part of a central nervous system (CNS) phenotyping screen for mutant mice, although some have questioned the extent to which differing studies purporting to measure aggressivity are in fact examining the same construct.36
Long-term exposure to “social defeat” has been proposed as an environmental factor relevant to the development of schizophrenia.38,39 In this context, dominance status and complexity of social structure have been shown to modify behavior in rodents across a variety of domains.40 Social dominance is usually assessed in the tube test,41 whereby 2 chambers each containing an unfamiliar mouse are connected by a narrow cylindrical tube which does not allow mice to pass within the tube. A subject is considered dominant when it remains in the tube while its opponent has retreated.
Other indices of social behavior in rodents include social play, which involves patterns relevant to the development of agonistic, sexual, and social behavior in adulthood.42 Play behavior in mice includes play soliciting behavior (push under, crawl below, push past between cage wall, and cage mate) and social grooming; it occurs mainly between weaning and puberty.43,44
On moving beyond social behavior to other domains, mouse models for negative symptoms of schizophrenia enter yet more difficult terrain. In relation to anhedonia, decrease in sucrose consumption has been commonly interpreted as evidence of reduction in reward function in rodents.45,46 However, using sucrose volume intake as an index of anhedonia is problematic, given alternative explanations for changes in this measure; in particular, during long-term consumption analysis of intake may be confounded by extraneous factors such as conditioned taste aversion, presence of competing behaviors such as locomotion or stereotypies, or visceral malaise.47 A further level of complication is that, as for social behavior, voluntary sucrose consumption has been used also to model anhedonia in relation to depression13,45; indeed, stress-induced anhedonia in this task has been shown to be sensitive to antidepressant treatment.48,49
A large psychological and neurobiological literature on motivation has yet to inform substantively on models of avolition in schizophrenia.
It has been proposed that progressive ratio schedule procedures, ie, operant task variants whereby response demands for reward increase across a series of trials, may provide a useful model of reduced motivation in schizophrenia.50 However, when employing a progressive ratio schedule, it is important to distinguish phenotypically between a high “breaking point,” which may be attributable to the level of motivation the animal is willing to transfer to work for reward, and “perseveration,” which may be attributable to enhanced impulsivity or disinhibition of a conditioned response.51,52 Others have considered assessment of motivation using operant paradigms where rats are offered a choice between lever pressing for a preferred reward food or ad libitum access to a less-preferred food.53 However, it should be noted that adapting specific, often complex operant paradigms established in rats to measure motivational and effort-based processes in mutant mice can prove difficult because stable performance in these types of tasks is generally more difficult to achieve in mice.
Rodent paradigms used to assess antidepressant drug action have also been applied to assess avolition and anhedonia in experimental models of schizophrenia, in terms of behavioral features commonly interpreted as relating to depression; these include tests such as the forced swim task and tail suspension test, which purport to assess “behavioral despair” in rats and mice. However, there endure the conceptual challenges of (1) the extent to which negative symptom–like features in depression might be related psychopathologically and pathophysiologically to negative symptoms in schizophrenia13 and (2) the lack of sufficient sensitivity of behavioral measures in small rodents to effect the necessary distinctions between features related to clinically similar symptoms in schizophrenia and depression.54
Modelling restriction in range of affect in schizophrenia is predicated on having some rodent index of affect. This has long-challenged research into affective disorders, from which there has been little cross-fertilization to research into schizophrenia: models of depressed mood are often validated in terms of antidepressant response, when antidepressants are without material effect on negative symptoms in schizophrenia; conversely, models of elevated mood are few, with antipsychotics being more effective in treating manic symptoms in bipolar disorder than negative symptoms in schizophrenia. As modelling reduced emotional expression in rodents clearly represents a general challenge, some investigators have interpreted decreases in tests of anxiety, such as the elevated plus maze and the open field test, as a measure of blunted affect.55 However, such interpretations remain conjectural and have yet to be substantiated.
Criteria for Validating Rodent Models of Negative Symptoms
The paucity of preclinical assays that provide rodent analogues of the negative symptom domain has disrupted progress in establishing criteria for their validation. Aside from face validity, rodent models of negative symptoms fare even less well with respect to construct and predictive validity.
In contrast to their positive counterparts, uncertainty as to the pathophysiological basis of negative symptoms and their lack of response to treatment with antipsychotic drugs impedes both “proxy” approaches and psychopharmacological validation. Negative symptoms respond poorly, if at all, to essentially all first- and second-generation antipsychotic drugs, with even clozapine exerting at best modest therapeutic efficacy2,6; thus, it is not clear whether, in addition to nonresponsivity to other antipsychotics, responsivity or nonresponsivity to clozapine should be considered a validating criterion for rodent models. It has been suggested that, when of any effectiveness, a longer duration of antipsychotic treatment may necessary to see significant reduction in negative relative to positive symptoms.56,57 However, this lacks the substance for even pragmatic model validation. Furthermore, because D2 dopamine (DA) receptor antagonism endures as the primary mechanism of antipsychotic activity, and because the dopaminergic (DAergic) system plays an important role in motivation and emotion, antagonism of D2-mediated reward and reinforcement might be expected to induce or exacerbate anhedonia and avolition.58
It is on this chastening background that we review phenotypic studies relating to negative symptoms in mice mutant for genes associated with aspects of the putative pathophysiology of schizophrenia or with risk for schizophrenia (see table 1 for summary of evidence for negative symptom phenotypes in mutant models).
|Candidate gene||Negative symptom|
|Altered DA neurotransmission||D2 over expression||?||?||+|
|Altered glutamatergic neurotransmission||NMDAR dysregulation||+||?||?|
|Schizophrenia risk genes||DISC1||+||+||?|
|Neurodevelopmental-synaptic genes||Complexin 1||+||?||?|
|Candidate gene||Negative symptom|
|Altered DA neurotransmission||D2 over expression||?||?||+|
|Altered glutamatergic neurotransmission||NMDAR dysregulation||+||?||?|
|Schizophrenia risk genes||DISC1||+||+||?|
|Neurodevelopmental-synaptic genes||Complexin 1||+||?||?|
Genes Related to DAergic Neurotransmission
Over recent years, the long-standing DAergic hyperfunction hypothesis of schizophrenia has been subjected to a series of elaborations: while positive symptoms appear to be related to increased release of DA onto subcortical D2 receptors that may be attenuated by D2 antagonist antipsychotics, negative symptoms may reflect associated reduction in cortical release of DA, particularly onto D1 receptors in prefrontal cortex.9,10,59
In contrast to positive symptoms, few studies have explicitly applied mutant mice approach to understand putative DAergic underpinnings to negative symptomatology in schizophrenia. One of the limitations to the application of constitutive gene deletion studies to this symptom domain is regional selectivity; in prevailing constitutive mutants, DA receptor subtypes and associated entities are deleted over the entire brain, when the prevailing hypothesis posits the differential involvement of cortical as opposed to subcortical brain regions. While the necessary studies with conditional mutants are awaited, there are to date a range of constitutive mutant studies that have sought to understand the independent roles of DA receptor subtypes and associated entities in processes of putative relationship to negative symptoms.
DA Receptor Subtypes
While polymorphisms in D1, D2, and D4 receptor genes may be associated with risk for schizophrenia,60 reports of associations with domains of psychopathology are limited; eg, variants in the D1 gene have been associated with responsivity to clozapine61 and variants in the D2 gene have been associated with negative symptoms62 and their limited responsivity to antipsychotics.63 In parallel, extensive phenotypic studies in mutants with knockout (KO) of each of the 5 DA receptor subtypes64,65 include aspects of behavior such as emotionality, reward, and social interaction that, in broad terms, may relate to negative symptomatology.
Evidence from studies in D1 and D2 KOs indicate that D1 and particularly D2 receptors play roles in diverse aspects of emotional behavior such as novelty seeking/detection, emotional arousal, retrieval of fear memory, limbic aspects of behavioral responses leading to the drive of action, and reward.65,66 More specifically, D2 KOs display a marked reduction in responding for rewarding lateral hypothalamic stimulation,67 suggesting a role for the D2 receptor in hedonic responses and, by inference, in anhedonia.
Mutants with selective overexpression of subcortical D2 receptors evidence deficits that include reduced incentive motivation, as indexed by reduced lever pressing for food reward in both an operant timing task and under a progressive ratio schedule of reinforcement.68 It remains to be determined whether this constitutes a model of anhedonia or alternatively involves the interplay of learning processes and cognitive mechanisms.
Evidence that variation in the dopamine transporter (DAT) gene may be associated with negative symptoms69 is complemented by the finding that DAT KOs with heightened DAergic function70 show impairments in social interaction,71,72 including disruption to social hierarchies under conditions where wild types (WTs) showed stable hierarchies. Under both group- and isolation-housed conditions, DAT KOs exhibited increased reactivity and aggression in the course of social contact, while during isolation, exposure to a novel environment exacerbated these social deficits. Stereotyped and perseverative patterns of social responses were a common feature of the DAT KO repertoire and abnormal social behavior coincided with the emergence and predominance of these inflexible behaviors.72 Importantly, these data suggest that social interaction may be disrupted under conditions of chronic DAergic hyperfunction. It should be noted that DAT KOs evidence impaired olfactory discrimination in the relative absence of impairment in odor detection.73 Thus, as noted previously, olfactory deficits might contribute to and confound the interpretation of changes in social functioning.
In contrast, tests assessing the rewarding values of tastants or food indicate DAT KOs to develop a more positive bias toward a hedonically positive tastant74 and enhanced resistance to extinction of food-reinforced operant behavior75; this would reflect the role of DA in updating rewarding values, habit learning, and memory. Increased sucrose consumption in DAT KOs would be further consistent with disruption to hedonic processes.76 In a sucrose-motivated runway task, mutants with DAT knockdown, to 10% of the complement in WT, showed greater motivation for the task (wanting) without influencing responsivity for sucrose reward (liking). These findings differ from those in DAT KOs; while they indicate that chronic DAergic hyperfunction produces changes in incentive motivation, they are in the opposite direction to what might be expected as part of a negative symptom profile77 and must be juxtaposed with the above findings on social behavior.
The enzyme catechol-O-methyltransferase (COMT) is involved in the catabolism of DA, with functional polymorphisms in the COMT gene indicated to exert differential regulation of DA metabolism in the prefrontal cortex and related cognitive processes, particularly working memory.78 While the COMT gene lies within a chromosomal region (22q11) of interest for psychosis, associations with risk for schizophrenia60,79 and dysfunction in cognitive processes mediated by the prefrontal cortex78,80 remain uncertain; COMT genotype has been associated with aggression in schizophrenia.81,82 While sociability and social novelty preference are unaltered in both heterozygous and homozygous COMT Kos,83 heterozygous COMT mutants evidence increased aggression in the resident-intruder test.84
The Chakragati Mouse
The Chakragati mouse is a serendipitously discovered, insertional transgenic mutant characterized by DAergic dysfunction, including increased D2 receptor density; such mutants display reduced social interaction, including decreased proximity during a dyadic test and reduced social approach behaviors.85,86
Mutants for components of several related aspects of DAergic neuronal development, morphology, and signal transduction have been constructed (eg, AKT1, FGFR1, GSK3β, Nurr1). However, their phenotypic evaluation in the context of schizophrenia does not yet extend systematically to models of negative symptoms.87,88
Renewed interest in the DA hypothesis of schizophrenia points evidentially to subcortical D2 hyperfunction in relation to positive symptoms and their attenuation by antipsychotic drugs; this is supported by some mutant studies. A postulated role also for cortical D1 hypofunction in negative symptoms is less well supported but remains heuristic. Mutant studies paint a complex picture where it proves difficult to specify the relative roles of D1 and D2 receptors in relation to individual domains of behavior that might relate to clinical psychopathology. For example, there endures the paradox that in animals antipsychotic (D2 antagonist) drugs acutely attenuate the effects of reward normally associated with “pleasure,” whereas in patients, antipsychotics act incrementally against positive symptoms with negligible effect on negative symptoms such as anhedonia.89
Genes Related to Glutamatergic Neurotransmission
Glutamate receptors have been suggested to play an important role in the pathogenesis of schizophrenia. In addition to subcortical D2-mediated hyperfunction and putative cortical D1-mediated hypofunction in schizophrenia (see “Genes Related to DAergic Neurotransmission” section), there is evidence for glutamatergic hypofunction. Alongside the well-characterized psychotomimetic properties of phencyclidine (PCP) and other N-methyl-D-aspartate (NMDA) glutamate receptor antagonists, NMDA deficits in the brain have been described in schizophrenia90–92 and antipsychotic activity has been reported for a metabotropic glutamate receptor agonist.93 While much clinical genetic data have focused on genes encoding the NMDA receptor and interacting signaling components as susceptibility candidates, there is also a growing body of evidence linking schizophrenia susceptibility with genetic variance in other glutamate receptor classes, including metabotropic receptor subtypes as well as non-NMDA ionotropic receptors, namely x-amino-hydroxy-5-methyl-4-isoxazole-propionic acid (AMPA) and, to a lesser extent, kainate receptors.94
Mice expressing reduced levels of the NR1 subunit of the NMDA receptor display abnormalities across several negative symptom–related domains, including social behavior. These include greater distance from unfamiliar mice during free social interaction in a novel environment, together with reduced social investigative and aggressive behaviors when acting as the resident in the resident-intruder paradigm; interestingly, these deficits showed little sensitivity to amelioration by clozapine.95,96 Decreased sociability, as assessed in the sociability and preference for social novelty test, has also been observed in NR1 hypomorphs.25,96,97 However, modest impairment in olfactory function may contribute to these social deficits.25,96
Grin1 (D481N) mutants, having reduced NMDA glycine site occupancy, display a decrease in sociability but not in social novelty preference; this deficit in sociability showed limited sensitivity to amelioration by clozapine.98 Interestingly, treatment with the selective glycine transporter 1 inhibitor SSR103800 attenuated deficits in social recognition in adult rats induced by neonatal injections of PCP.99 Additionally, agonists at the glycine site of the NMDA receptor may have some efficacy as adjunctive therapies for the negative symptoms of schizophrenia.100–102 Overall, studies in both rodents and humans would indicate therapeutic potential for glycine agonism in the treatment of negative symptoms.
NMDA Receptor–Related Processes
Abnormalities in various components of the NMDA receptor signaling complex have been implicated in schizophrenia.87,103 In particular, preclinical studies have implicated several such regulatory components, including the glial glutamate and aspartate transporter (GLAST) and the postsynaptic density-enriched scaffold and signaling molecule SynGAP.104,105 A rare genetic variant in the human gene encoding GLAST has been reported in schizophrenia,106 and postmortem brain studies have demonstrated altered GLAST expression in the dorsolateral prefrontal cortex, anterior cingulate cortex, and thalamus in schizophrenia.107,108
Heterozygous SynGAP mutant mice evidence intact sociability but impaired social novelty preference.104 Conversely, when assessed in the same paradigm GLAST KOs evidence a marked reduction in sociability, with intact social novelty preference and dyadic social interaction in a novel environment; there was no effect on sucrose preference as a putative index of anhedonia.105
Mutants with heterozygous deletion of glutamate carboxypeptidase II, a signaling component implicated in NMDAR activation, evidenced reduced sociability in a social choice paradigm.109
Non-NMDA Ionotropic Glutamate Receptors
Mutants with KO of the AMPA GluR1 receptor subunit fail to show the increase in aggression toward a conspecific that normally follows social isolation, in a manner similar to the effects of treatment with an AMPA/kainate antagonist,110 and display reduced social behavior as measured by anogenital-directed social investigation.111
Metabotropic Glutamate Receptors
Recent studies have suggested alleles of several metabotropic receptor subtypes to be associated with increased risk for schizophrenia.112,113 Pharmacological modulation of activity at mGluR3 or mGluR5 receptor subtypes may ameliorate social interaction deficits in pharmacological or environmentally based models for negative symptoms in mice (PCP treatment and isolation rearing, respectively).114,115 However, metabotropic receptor mutants have yet to receive systematic investigation in relation to social or other behaviors relevant to negative symptoms.
Other Glutamate-Related Processes
Vesicular glutamate transporters (VGluTs) 1 and 2 are recognized markers of glutamatergic neurons that are responsible for the vesicular packaging of glutamate in the presynaptic axon terminal.116–118 Abnormal VGluT1 expression in schizophrenia has been reported in the striatum and hippocampus119 and in the anterior cingulated.120
Mutants with heterozygous deletion of VGluT1 exhibited reduced sucrose consumption consequent to chronic mild stress.121 Mutants with conditional, heterozygous deletion of VGluT2 in the cortex, hippocampus, and amygdala during the third postnatal week evidence reduced social dominance in the tube test and spend more time interacting with unfamiliar conspecifics in a novel environment.55 These data would suggest that reduced expression of VGluTs is associated with an array of social and anhedonic phenotypes.
Mutants for components of several related aspects of glutamatergic transmission have been constructed (eg, D-serine, mGluR1-8, NR2A [GluRϵ1]). However, their phenotypic evaluation in the context of schizophrenia does not yet extend systematically to models of negative symptoms.87,88
Enduring interest in glutamatergic hypotheses of schizophrenia points evidentially to NMDA hypofunction in relation to both positive and negative symptoms. This is supported and elaborated by mutant studies that indicate, with some consistency, disruption to a number of social and hedonic processes. Therapeutically, studies in mutants have contributed to interest in glycine transporter inhibitors, as indirect facilitators of glutamatergic transmission, for the treatment of negative symptoms. The incisiveness and specificity of mutants have the potential to illuminate the development of glutamatergic neuronal (dys)function because it might relate to the pathobiology of schizophrenia and, particularly, to delineate more optimal therapeutic targets in the glutamatergic transmission-signaling cascade.
Genes Associated With Risk for Schizophrenia
Over the past several years, molecular genetics has identified a number of candidate risk genes, using both association and linkage studies, as documented and synthesized in recent systematic reviews and meta-analyses.60,122–126 Inconsistency between studies and a continually evolving tableau in ongoing, “real-time” meta-analyses60,127 may reflect: (a) a putative polygenic basis to schizophrenia, with several genes of small effect contributing to overall liability; (b) that implicated genes confer risk not for schizophrenia per se but, rather, for psychosis as a dimensional construct that transcends any unitary diagnostic category; (c) a diversity of genetic loci associated with different domains of psychopathology; (d) as a variant of the above, that individual genes or combinations of genes are associated with endophenotypes within the overall schizophrenia syndrome; and (e) that genetic risk may depend upon interactions between individual susceptibility genes (epistasis) and/or interaction between susceptibility genes and exposure to one or more environmental adversities.123,128 Most recently, there has been intense interest in multiple copy number variations each conferring risk for schizophrenia in relatively small numbers of cases.106,129 It remains to be determined whether a plethora of genome-wide association studies will clarify or further confound these issues.
Although relatively few studies have sought to delineate the relationship between schizophrenia risk genes and domains of psychopathology, this approach has the potential to provide an important conceptual link toward understanding the genetics of schizophrenia. The construction of mice mutant for genes either implicated in CNS processes relevant to putative pathophysiologies of the disorder or associated directly with risk for schizophrenia has provided an important translational stimulus to addressing these questions.
A study in a Scottish pedigree demonstrated that a familial mutation in the disrupted-in-schizophrenia-1 (DISC1) gene, due to a balanced chromosomal translocation at 1q42.1–1q42.3, segregated with several psychiatric disorders, including schizophrenia; this association between DISC1 and schizophrenia has been replicated across diverse populations.124,125 During embryonic development, DISC1 appears to play an important role in neurodevelopment and structural plasticity via interaction with several proteins, including phosphodiesterase-4B, Fez1, NudEL, and LIS1.130 While there is little clinical evidence for any specific relationship between DISC1 and negative symptoms, a relationship with social anhedonia in a large population cohort has been reported.131
Among several mutant lines with disruption to DISC1,132 a DISC1 mutation (Q31L) generated using chemical mutagenesis demonstrated disruption to both sociability and preference for social novelty; additionally, this line evidenced decreased sucrose consumption.133 In a conditional transgenic line with inducible expression of a DISC1 C-terminal fragment, early postnatal (day 7) induction was associated with reduced sociability.134 Conversely, expression of a dominant-negative truncated form of DISC1 under the CaMKII promoter did not disrupt social interaction.135 A conditional transgenic line with forebrain-specific expression of mutant human DISC1 was associated with a sex-specific decrease in social investigation in males, with increased aggressivity in a dyadic test of social interaction but no effect on sociability or social novelty preference.136
Dystrobrevin-binding protein 1 (DTNBP1; dysbindin) was initially identified as a schizophrenia susceptibility gene after fine mapping of a linkage region on chromosome 6p22 in Irish multiplex families; this has since been replicated across diverse populations.60 DTNBP1 expression is decreased in schizophrenia in the dorsolateral prefrontal cortex and hippocampus.137,138 Clinical genetic studies in schizophrenia have indicated associations between DTNBP1 and negative symptoms.139
The sdy mouse, a spontaneous mutation constituting a murine model of Hermansky-Pudlak syndrome,140 is characterized by a large deletion encompassing 2 exons of the DTNBP1 gene and shows no expression of dysbindin protein. In a test of dyadic social interactions, dysbindin (sdy) mutants evidence a reduction in social contact time.141
Following an initial report in 2 independent samples, the G72/G30 gene complex has been associated with risk for schizophrenia across numerous populations60,142; this gene regulates the activity of D-amino acid oxidase (DAO); hence, the alternative nomenclature D-amino acid oxidase activator.
In transgenic mutants carrying the human G72/G30 genomic region, nonaggressive social interaction is intact, while male mutants show a reduction in aggressive behaviors; there were also deficits in olfactory function.143
Following an initial report in an Icelandic sample, the identification of neuregulin-1 (NRG1) as a putative risk gene for schizophrenia has been replicated across many populations127,144; furthermore, studies in postmortem brain tissue support a role for NRG1 and associated signaling through ErbB receptors in the pathobiology of schizophrenia.145–147 Distinct targeted mutations of various NRG1 isoforms have made it possible to delineate some of their specific functions, including some that relate to negative symptoms.
Mutants with heterozygous deletion of transmembrane (TM) domain (pan-isoform) NRG1 display selective impairment in response to social novelty, as demonstrated by intact sociability but absence of preference to investigate a novel over a familiar conspecific.148 In contrast, heterozygous epidermal growth factor (EGF)-like domain (pan-isoform) NRG1 KO mice display reduced sociability as measured in a social choice paradigm,149 the differences between these findings and those reported in TM domain NRG1 mutants may relate to the mutation or several important procedural differences (type of social stimulus used, lighting conditions). Mutants with loss of ErbB signaling in oligodendrocytes also show impaired social interaction in a dyadic paradigm.150 TM-NRG1 mutants,148,151 but not EGF-like domain149 or type III isoform-specific NRG1 mutants,152 also display enhanced aggression in social encounters, while mutants with conditional KO of the ErbB2/B4 receptor show increased aggression in the resident intruder paradigm; this deficit was reversible by treatment with clozapine.153
B-site amyloid precursor protein-cleaving enzyme 1 (BACE1) has been implicated in NRG1 signaling.154 Its function has been studied using a social choice task variant, the social habituation-dishabituation paradigm, to assess response to social novelty. In this task, the test mouse is repeatedly exposed to a juvenile conspecific and social behaviors are then recorded across sessions (habituation); a novel social stimulus is then added and response to the new stimulus is examined (dishabituation); BACE1 KOs evidenced reduced dishabituation, suggesting decreased behavioral response to social novelty.155
Calcineurin is a calcium- and calmodulin-dependent protein phosphatase composed of 2 subunits, a regulatory subunit of calcineurin B and a catalytic subunit of calcineurin A (CNA) that has been implicated in downstream regulation of DAergic signal transduction and in NMDA receptor–dependent synaptic plasticity; PPP3CC is the gamma isoform of CNA. Variation in the PPP3CC gene has been associated with risk for schizophrenia.122,126
Mutants with conditional, forebrain-specific calcineurin KO display a sustained decrease in social contacts with an unfamiliar mouse in a home cage environment.156 KO of ryanodine receptor 3, an interacting partner alongside calcineurin, results in a decrease in social contacts in both home cage and novel environments; there were no effects in the test of sociability and preference for social novelty.157
Regulator of G-Protein Signaling-4
Regulator of G-protein signaling-4 (RGS4) was initially identified as a putative risk gene for schizophrenia in a multinational sample and reported to show reduced expression in postmortem brain; however, subsequent meta-analyses across numerous samples and further studies in postmortem brain have indicated these issues to be less clear.60,158,RGS4 polymorphisms have been associated with poorer social function in schizophrenia and greater amelioration of that dysfunction by risperidone.159
Mutants with RGS4 KO (cre-deleted RGS4lacZ/lacZ) have yet to be examined for behaviors related to negative symptoms.160 However, mutants deficient in phospholipase C-β1, a signaling molecule that mediates activity within several neurotransmitter pathways and with which RGS4 interacts,161 display reduced social dominance, as measured in the tube test; they also evidence reduced whisker trimming, a form of mutual grooming related to social dominance.162
Mutants for additional genes, either implicated in risk for schizophrenia or interacting with those above, have been constructed (eg, DAO, FEZ1, Nogo receptor 1 [RTN4R], PRODH, ZDHHC8). However, their phenotypic evaluation in the context of schizophrenia does not yet extend systematically to models of negative symptoms.87,88
Advances in the molecular genetics of schizophrenia can, in a “top-down” manner, prompt construction of a line of mutants for each risk gene as it is identified; the purpose is then to investigate phenotypically the functional role of that gene because it might relate, in the present context, to negative symptoms. However, studies can also proceed also in a “bottom-up” manner; eg, is intact sociability but impaired social novelty preference in NRG1 mutants18,148 related to a particular pattern of social deficit in patients carrying a given NRG1 risk polymorphism?
A particular complication is that for any given risk gene of interest, several mutants may be available. Diverse mutants for DISC1 and NRG1 illustrate the dilemma in determining which may be the most informative on the psychopathology and pathobiology of schizophrenia. These decisions will only be clarified by greater understanding of the neurobiology of such entities in the context of the neurobiology of schizophrenia itself.
It must be considered also whether negative symptoms can be modeled, in any simple way, by a single-gene manipulation. If schizophrenia reflects the operation of several risk genes of small effect that act in a complex environmental milieu, negative (and indeed other) domains of psychopathology may involve gene × gene interactions (epistasis) and gene × environment interactions.123,163 To the extent that this is sustained, progress may require generation of mutants with concurrent disruption to 2 or more risk genes of interest and assessment of mutant phenotypes in relation to external biological and psychosocial adversities.87,88
In addition to the above genes associated with specific DAergic and glutamatergic pathophysiologies or with risk for schizophrenia, other genes regulate more general synaptic processes implicated in schizophrenia, particularly in the context of developmental disruption to neuronal connectivity.11,12,123,164
Complexins are small presynaptic proteins that bind to the soluble N-ethylmaleimide–sensitive factor attachment protein receptor (SNARE) assembly and stabilize the SNARE complex for fast calcium-mediated exocytosis165; complexin 1 (Cplx1) expression is decreased in the postmortem brain in schizophrenia.166 Mutants with Cplx1 KO show disruption in preference for social novelty in the absence of any effect on sociability or olfactory function; in a resident-intruder paradigm, male Cplx1 KO showed reduced aggression toward an unfamiliar intruder.165
Reelin is a neuronal glycoprotein involved in CNS development and expressed in γ-aminobutyric acid (GABA)–containing cells of the cortex, hippocampus, and cerebellum. In schizophrenia, reelin and the GABA synthesizing enzyme glutamic acid decarboxylase (GAD67) are downregulated in cortical GABAergic interneurons, such that partial deletion of reelin has been offered as a pathophysiological model of schizophrenia.167 Mutation of one allele encoding the reelin protein (the heterozygous reeler mouse) results in higher levels of social dominance in the tube test167 but does not otherwise disrupt social interaction.169
Mutation in the neuronal PAS domain protein 1/3 (NPAS1/NPAS3) transcription factors resulted in reduced expression of reelin in various brain areas that was accompanied by impaired social recognition.170 Methionine-induced epigenetic reelin promoter hypermethylation in mice resulted in deficits in aggression and habituation in the resident intrude test and reduced social interaction in a novel environment.167,171
Stable Tubule–Only Polypeptide
The stable tubule–only polypeptide (STOP) proteins are involved in the cold stability of microtubules, brain development and connectivity, synaptic plasticity, and neurotransmission.
Mutants with STOP KO evidence reduced sniffing of a conspecific introduced into the home cage, together with reduction in aggressive responses, in the absence of any substantive disruption to olfaction; these social deficits were poorly sensitive to amelioration by chlorpromazine and haloperidol.172 Provocatively, while other behavioral, synaptic vesicular, and electrophysiological abnormalities described also in STOP mutants appear to be ameliorated by treatment with the microtubule stabilizer epothilone D,173 any effects on these recently reported social deficits are yet to be reported.
Synapsins are a family of neuron-specific, vesicle-associated phosphoproteins involved in the regulation of neural development and transmitter release; synapsin II mRNA is reduced in the medial prefrontal cortex in schizophrenia, while chronic treatment with haloperidol increases synapsin II mRNA in rats.174 Mutants with synapsin II KO demonstrate a marked reduction in social interaction.175
Rather than deriving in a “top-down” manner from clinical molecular genetic studies, these mutants are of considerable value via their relationship to mechanisms of synapse formation, plasticity, and connectivity that are posited to be disrupted in schizophrenia. Importantly, they can be related phenotypically, here in the context of negative symptoms, to pathobiology via psychopathological neuroimaging and postmortem studies in patients. Thus, on a long-term basis, such mutant studies may contribute importantly to clarifying the pathobiology of negative symptoms, at a more fundamental level than is apparent for approaches based on current neurochemically based hypotheses or individual risk genes.
A variant approach is to consider the neurobiology of behavioral processes that could relate to negative symptoms, with a view to their study in schizophrenia-related mutants with putative negative symptom phenotypes. Several molecules have been show to play a critical role in such behaviors. Several examples, involving mutant studies of social behavior, are outlined in the subsequent paragraphs.
The neuropeptide oxytocin is a modulator of animal176,177 and human178,179 social functioning. Central administration of oxytocin to rodents improves social interaction, 180,181 and exogenous oxytocin reverses deficits in social behavior following prenatal exposure to a stressor.182 Conditional oxytocin KOs show deficits in social recognition183 and in intrastrain but not interstrain social recognition,184 indicating a more specific role for oxytocin in social discrimination. Interestingly, clozapine but not haloperidol has been shown to increase plasma concentrations of oxytocin.185
The antidiuretic hormone arginine-vasopressin (AVP) is known to play an important role in social and emotional behavior176: AVP-V1aR KOs display impaired social recognition memory and social interaction that can be rescued by reexpression of AVP-V1aR in the lateral septum186–188; AVP-V1bR KOs display impaired social recognition and conspecific aggression,189,190 with disruption to sociability and preference for social novelty.191
Mutants with KO of neuronal nitric oxide synthase show impaired social recognition.192 Pretreatment with an NOS inhibitor reverses deficits in social interaction induced by PCP,193 a treatment that increases NO in prefrontal cortex.194
A related variant approach involves inbred strains of mice with neurodevelopmental phenotypes that may inform on schizophrenia. For example, the BTBR T+tf/J inbred strain displays impairment in dyadic social interactions, reduced social transmission of food preference, disrupted sociability, and reduced social play.195
Accessing “Inaccessible” Negative Symptom Constructs?
In contrast to asociality, anhedonia and to some extent avolition, in animals blunted affect is confounded with our concepts and measures of either polarity of “affect,” while poverty of speech may be uniquely human. However, some researchers have sought to meet the challenge of these “inaccessible” constructs and have offered novel behavioral indices and end points for their assessment.
In relation to impaired processing of emotions in humans, a recent animal model has been offered196; using fear processing and ketamine-induced glutamatergic hypofunction, impaired amygdala-based fear processing was reversed by clozapine but not by haloperidol. This paradigm has yet to be applied to mutant models relating to schizophrenia.
In relation to poverty of speech, reduction in stress-induced vocalization has been offered as an animal model. Specifically, isolation-induced ultrasonic vocalizations in neonates, which are produced to elicit maternal approach and/or retrieval, are increasingly used in the phenotypic study of mice mutant for genes associated with neurodevelopmental disorders, in particular those characterized by communicative/social deficits.197 Reductions in ultrasonic vocalizations in separated pups have been observed in several mutant models of schizophrenia, including the reeler mouse198 and DISC1 mutants.199 While investigation of adult mouse vocalizations has proved more difficult, abnormalities in vocalizations signaling male-female recognition have been observed in D2 KOs.200 However, just as olfactory deficits may confound the investigation of social behavior in mutants, it is important to assess potentially confounding factors such as lung function or larynx morphology on vocalization.201
An ethological approach affirms that characterisation of the species-specific behavioral repertoire takes precedence in any analysis of the clinical relevance of behavioral changes encountered in experimental models. This approach has been used extensively in systematic investigation of the phenotype of mutants with KO of each of the 5 individual DA receptor subtypes.64 Among other naturalistic behaviors, disturbance in nest building has been offered as a murine measure of the negative symptom of self-neglect in Dvl1 KOs41 and NMDA NR1 hypomorphic mutants.97 However, disruption of nest building is likely to be multifactorial and may be subject to other interpretations.
Despite rapid advances over the past several years,18,19,87,88,123,202,203 it is clear that we continue to face substantive challenges in applying mutant models to better understand the pathobiology of negative symptoms (and other domains of psychopathology) in schizophrenia.
First, the majority of evidence relates to impairments in social behavior, with only limited data relating to anhedonia and negligible data concerning avolition and other aspects of negative symptoms.
Second, even for the most widely examined behavior, studies in the various mutant lines have used diverse tests of sociability and aggression. While the test of sociability and social novelty preference is, perhaps, emerging as the most widely applied paradigm, this currently allows meaningful comparisons between only a minority of studies. In this regard, there endures also the problem that the “same” test applied in different laboratories to the “same” subjects may, for poorly understood reasons,204 generate different results.
Third, in the absence of validating pharmacology, other than perhaps nonresponse, to what extent should the dearth of systematic psychopharmacological studies be understood as rational conservation of resources or negligence in not confirming such nonresponse. There endures the challenge of how to interpret the few but potentially important findings with clozapine vis-à-vis the clinical debate as to its clinical efficacy for negative symptoms.
Fourth, modelling must proceed in cognizance of increasing evidence that genes and pathobiologies implicated in schizophrenia overlap with other psychotic disorders, particularly bipolar disorder, in which negative symptoms may be less evident.
Despite the caveats and challenges considered above, it should not be overlooked that several mutant lines evidence a phenotype for at least one index of social behavior, independent of whether the gene at issue relates to a putative pathophysiological processes or to risk for schizophrenia. Though this may suggest superficially some shared relationship to negative symptoms, it is not yet possible to specify either the scope or the pathobiology of that relationship for a given gene. Furthermore, whether each mutant line indicates the same or a different phenotypic relationship to the individual components of negative symptoms is poorly understood. Conditional mutants, where expression of a gene at issue can be controlled in space (ie, differentially across brain regions) and/or temporally (ie, differentially over stages of development), have the potential to markedly increase the yield from mutant studies.
As an essential context, it must be emphasized that (1) our knowledge of the psychopathological boundaries and pathophysiology of negative symptoms in patients is also far from clear, and (2) these uncertainties derive, at least in part, from the diversity of clinical psychopathology, treatment response, and outcome. Thus, it could be argued that the diversity of findings from putative mutant models is actually reflective of clinical reality.
Science Foundation Ireland (07/IN.1/B960) and Health Research Board (PD/2007/20).