Neuregulin-1 signaling in schizophrenia: ‘Jack of all trades’ or master of some?

Application of the ‘retrospectoscope’ reveals two publications in 2002 to have played a seminal role in one of the major contemporary fronts in schizophrenia research: Moises and colleagues [1] suggested a role for neuregulin in a glial growth factor deficiency and synaptic destabilization hypothesis of schizophrenia; and Steffansson and colleagues [2] indicated neuregulin 1 (NRG1) to be a candidate gene for the disorder. Since then, there have been 179 publications on neuregulin and schizophrenia, with no less than 56 appearing in the year to date. Schizophrenia is a chronic, debilitating psychotic illness characterized by serious disturbances of perception, cognition and social functioning; it affects almost one in every 100 people worldwide. The neuregulins are a family of growth factors that are encoded by four genes (NRG1–4) and share a common EGF-like domain; interaction of these EGF-like domains with membrane-associated tyrosine kinases (ErbBs) activates intracellular signaling pathways that are held to play a prominent role in neurodevelopment, including neuronal migration, axonal patterning and synaptic morphogenesis and function. Why, then, should neuregulin have attracted such breadth and depth of attention in relation to the pathobiology of schizophrenia?

Currently, developmental models of schizophrenia hold center stage. They propose that genes regulating neurodevelopment and/or early environmental insults result in disruption to brain development and increase risk for the diagnostic symptoms of schizophrenia later in life; dysfunction in neuregulin signaling therefore has prima facie credentials for a role in schizophrenia. Additionally, NRG1 is one of several genes that have now been associated replicably with risk for schizophrenia. Recent reviews have documented the still expanding literature on the putative role of neuregulin in schizophrenia [3–5]; this includes functional genomic studies that implicate NRG1 genotype in structural brain pathology, psychopathology, cognitive impairment and transition from an ‘at risk mental state’, together with evidence for abnormalities in neuregulin signaling in human post-mortem brain tissue. These approaches are complemented by a similarly expansive literature on phenotypic studies in mutant mice (knockouts and transgenics) that allow the functional role of NRG1 to be further explored [3,6].

However, such studies appear now to have reached a point where an amber light is flashing. Specifically, NRG1 is only one of several genes that have now been associated replicably with risk for schizophrenia; does each gene contribute to overall risk for diagnosis of schizophrenia and its underlying pathobiology, perhaps via expression in pathways that ultimately converge on a common process such as glutamate-mediated neurotransmission, or do specific genes influence risk for distinct aspects (endophenotypes) of the overall schizophrenia syndrome via independent pathobiological processes [3]?

Interpretation of emergent studies can present a number of challenges. For example, post-mortem brain studies in schizophrenia have reported diverse alterations in NRG1, and associated ErbB4 expression and protein levels in the prefrontal cortex and hippocampus [4,5,7]; where increases in NRG1 indices have been reported, does this reflect an increase in neuregulin signaling or a compensatory response to impaired signaling therein? Furthermore, chronic treatment with the antipsychotic haloperidol, a well-recognized potential confound in human post-mortem brain studies in schizophrenia, has recently been reported to increase expression of NRG1β (EGF domain) brain-active isoforms and ErbB4 in the rat prefrontal cortex and hippocampus [8].

Indeed, multiple NRG1 isoforms have been described, the diversity of which has been attributed to alternative splicing and the existence of multiple 5´ flanking regulatory elements: NRG1 I–IV share the EGF-like signaling domain, are defined by different amino acid terminals, and differ in concentration as well as regional distribution in the rodent and human CNS [4,5]. Although schizophrenia-related single nucleotide polymorphisms (SNPs) are located primarily in noncoding regions of the NRG1 gene, regulatory processes, such as the binding of transcription factors, splicing, mRNA degradation and translation, may be affected, thereby altering the quantity, ratio of expression and distribution of NRG1 isoforms [4,5]. Therefore, depending on the balance of expression of the different NRG1 gene products, as well as possible epigenetic modifications resulting from potentially fundamental gene–environment interactions [3], a variety of different disease-relevant phenotypic profiles are possible for NRG1. Owing to the pleiotropic nature and complexity of NRG1, together with the likelihood of critical epistasis with other risk genes (gene–gene interactions) [3,9], careful attention is required when distinguishing between possible contributions of individual NRG1 isoforms to schizophrenia endophenotypes.

Generation of mice with targeted mutations via gene knockout or overexpression now allows identification of the functional significance of these various NRG1 isoforms. Owing to the crucial role played by NRG1 in cardiac development, the null mutation is not viable. For this reason, the vast majority of these studies are conducted using heterozygous knockout mice. Most NRG1 proteins contain a transmembrane (TM) domain. Heterozygous knockout of the NRG1 TM domain produces several schizophrenia-related behavioral abnormalities, including disturbed interaction with and subsequent adaptation to a novel environment, sensorimotor gating deficits and, in particular, disruption to social novelty preference but with intact sociability and spatial working memory [6,10].

Deletion of TM domain NRG1 is likely to affect multiple isoforms containing this domain; therefore, it is not surprising that some overlap exists between the phenotype of this particular mutant and more isoform-specific mutants. For example, mutants with deletion of type I/type II NRG1 evidence impaired latent inhibition (a measure of attention thought to be disrupted in schizophrenia) with intact sensorimotor gating and normal levels of activity [6]; recently, a transgenic mutant overexpressing type I NRG1 has also been described [11], although interpretation of the phenotype is complicated by the presence of tremor that is not recognized as a clinical feature of schizophrenia [12]. Conversely, type III NRG1 mutants evidence impairments in sensorimotor gating and working memory performance, as well as morphological changes including reduced dendritic spine density, enlarged lateral ventricles, and hypofunctionality of medial prefrontal cortex and CA1 region of the hippocampus [13]. It remains possible that some phenotypic effects of specific mutations of NRG1 isoforms are masked by compensatory activity of other genes or by environmental factors.

In summary, although concern endures as to the extent to which NRG1 genotype and dysfunction in neuregulin signaling can account for all of the clinical and pathobiological features of schizophrenia, in the manner of a ‘Jack of all trades’, provocative findings continue to emerge. For example, type III NRG1 has recently been implicated in the regulation of α7 nicotinic acetylcholine receptor (AChR) expression on the surface of axons [14]; this is of considerable interest, as there is convergent evidence from genetic and neurobiological studies to implicate the α7AChR as a possible pathological mechanism in schizophrenia, particularly in relation to cognitive impairment and putative treatment modalities [15]. Additionally, NRG1 mutant mice are hypersensitive to the acute effects of Δ⁹-tetrahydrocannabinol [16], the active ingredient of cannabis, consumption of which is associated with an approximate doubling of risk for psychosis [17]; this suggests an important gene–environment interaction in regulating the pathobiology of schizophrenia. Finally, the notion of NRG1 as a risk factor for schizophrenia may require revision; there is an emergent but compelling body of evidence for NRG1 as a risk factor for both schizophrenia and bipolar disorder [3,18,19].

The existing evidence would suggest that the contribution of NRG1 to etiopathogenesis is dependent not only on the location and strength of neuregulin signaling in the brain but also on the pattern of expression of the various NRG1 isoforms and ErbB receptors, as well as involvement of interacting molecules (e.g., B-site APP-cleaving enzyme 1). Improved characterization of the functional properties of these NRG1 isoforms will be necessary to elucidate the relative contribution of genetic variation at the NRG1 locus to the development of psychotic illness and to reveal novel antipsychotic drug targets, most probably on a dimensional rather than a diagnostic basis.

Financial & competing interests disclosure

Support for this work was provided by a Health Research Board postdoctoral fellowship to C O’Tuathaigh and a Science Foundation Ireland Principal Investigator grant to J Waddington. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.