Do Complement factors “connect the dots” in Schizophrenia?

“…everything connects to everything else…”

Leonardo da Vinci

Schizophrenia is proposed to be the result of complex interaction between multiple genetic and environmental factors that alter healthy brain development (Insel, 2010). Typical onset of schizophrenia is during adolescence and young adulthood, and the presence of neurocognitive deficits and social impairments before the onset of schizophrenia suggest pronounced premorbid neurodevelopmental abnormalities (Millan et al., 2016) characterized by decreased neuronal volume, dendritic spine loss and increased packing density. The precise etiology of these deficits is unknown, although genetic factors have been associated with the risk for this illness. The contribution of inflammation has recently gained significant traction.

Over the decades, several inflammatory mediators were examined for their association with schizophrenia including complement proteins. Complement proteins are part of the innate immunity and potentially were important. Initial studies focused on these proteins because of their role in inflammation, thus connecting the dots between inflammation and schizophrenia. However, levels of complement proteins in peripheral blood in schizophrenia patients have not shown a consistent pattern of alteration. In that light, recent genome-wide association studies (GWAS) in large samples providing adequate power to detect variants with genome-wide significance, revealed 108 significantly associated genetic loci with the risk of schizophrenia (Schizophrenia Working Group of the Psychiatric Genomics Consortium, 2014). The strongest risk was repeatedly identified on the Human Leukocyte Antigen (HLA) region, which is a gene-rich region with complex linkage disequilibrium patterns. After painstaking experiments, Sekar et al reported that a copy number variant consisting of four genes (denoted as R-C-C-X for STK19 (RP1), C4 (C4A or C4B), CYP21A1 or CYP21A2, and TNXB) accounted for a portion of the risk associated with the HLA region (Sekar et al., 2016). It is important to note that C4A variants associated with schizophrenia and synaptic pruning are not single nucleotide polymorphisms on a candidate gene, but they are gene copy numbers that result from a recombination site at CYP21A2 that leads to mono-, bi-, or trimodular cassettes. These cassettes generate multiple functional copies of C4A/C4B. Intronic insertion of human endogenous retroviral (HERV) elements affect the transcription of C4A/C4B. Thus, Sekar et al reported that both C4A and C4B protein expression was increased proportional to copy numbers in key regions of postmortem brain tissues of schizophrenia patients; however, C4A but not C4B copy numbers correlated with schizophrenia risk. Further, C4-deficient mice showed decreased synaptic pruning (Sekar et al., 2016). Complement proteins have in recent years been implicated as serving a role in the phagocytosis of redundant (or ineffective) synapses during adolescence, a critical period of risk for schizophrenia onset (Stephan et al 2012). Together, these important observations implicated genetically mediated alterations in complement activity as underlying alterations in synaptic pruning and schizophrenia risk (Figure 1, also on Front Cover).

Figure 1.

The synaptic pruning hypothesis of schizophrenia. The figure on the left illustrates a healthy neuron (with a robust action potential) and microglia; the figure on the right shows a weak action potential and an activated microglia "eating up" a dysfunctional dendritic process via complement proteins.

Based on multiple supportive evidence from postmortem, sleep and EEG studies, Feinberg originally proposed that aberrant peri-adolescent pruning of synapses (resulting in either too much or too little pruning) as underlying the pathogenesis of schizophrenia (Feinberg, 1982). Subsequently, it was postulated that the neurobiological abnormalities observed in schizophrenia suggest an exaggerated pruning of synapses during adolescence/young adulthood which could explain the adolescence onset (Keshavan et al., 1994). While no direct evidence of excessive pruning is yet to emerge in schizophrenia, several “dots” are getting connected to yield a coherent view. First, gray matter loss in first episode schizophrenia and among at-risk individuals for psychosis evidenced by imaging (Cannon et al., 2015) may reflect synaptic pruning, but gray matter metrics are composite physical measures of synapses and other neural elements, glia, interneuronal space, and microvasculature. Second, phosphorus magnetic resonance spectroscopy (³¹P MRS) studies on first episode schizophrenia noted greater neuropil contraction compared to controls (Pettegrew et al, 1991), and more recently on young adult-onset, adolescent-onset schizophrenia patients and controls showed that increasing C4A and also C4B copy numbers were associated with greater neuropil contraction, suggesting a gene dosage effect on neuropil contraction (Prasad et al., 2018). Although ³¹P MRS does not directly assess the synapses, relative signals from variations in membrane phospholipid metabolites are comparatively more specific and sensitive to changes in the neuropil, of which synapses are integral parts. Finally, there is support for decreased dendritic spine density in post-mortem brains of schizophrenia patients (Garey et al., 1998; Glantz and Lewis, 2000) compared to healthy subjects and patients with major depression. It is notable that decreased dendritic spine density was primarily confined to basilar dendrites in deep layer 3 of the frontal and temporal cortex. However, decreased dendritic spine density may not be only due to excessive synaptic pruning, but related to decreased formation or maintenance of dendritic spines.

Given these challenges, the publication of Sekar et al study renewed interest in examining complement proteins to elucidate factors associated with synaptic pruning. In recent publications in this Journal, Laskaris et al (Laskaris et al., 2018), examined alterations in peripheral blood of C1q, C3, and C4 in ultra-high-risk, first episode psychosis and chronic schizophrenia patients. They found increased levels of C3 and C4 among ultra-high-risk subjects, no change in first-episode patients, and elevated C4 levels in chronic schizophrenia patients. While this is not a longitudinal study, these findings suggest elevated complement levels before the onset of schizophrenia with possible “normalization” at first-break, followed by elevation during a prolonged course of schizophrenia. Pathophysiological implications of these findings are, thus, unclear. This finding, if replicated in longitudinal follow-up studies, suggests that elevations in selected complement proteins may tip the balance toward conversion to psychosis and thus be a useful predictive biomarker.

Kopczynska et al showed no differences in C3 or C4 among first episode schizophrenia patients (Kopczynska et al., 2017). This finding is similar to Laskaris et al study for first episode patients. Although it is tempting to propose an association of complement protein variations with increased synaptic pruning in ultra-high-risk subjects, it is important to note that peripheral complement proteins do not cross intact blood-brain barrier, and that complement proteins in the brain are synthesized locally in the microglia and neurons. Therefore, peripheral levels of C4 are not a proxy for complement activity in the brain, but a study by Yang et al (Yang et al., 2003) demonstrated that serum C4 protein concentrations correlate with C4 gene size, which would suggest that an increase in C4 copy number could predict increased levels of C4 both centrally and in the periphery. Finally, the authors did not evaluate the possibility that specific isoforms within this complement cascade may be altered at various stages of a psychotic illness, such as C4A or C4B (Sekar et al., 2016).

In summary, while the idea that complement alterations may prove to be an etiologically meaningful and therapeutically actionable biomarker in schizophrenia, many questions remain. First, it is important to note that there are other gene variants associated with schizophrenia risk, implicating other synaptic proteins and the development of new tools are making it more accessible to better understand the role of the complement system on synaptic pruning. For example, integrating genetics with in vivo brain imaging (Prasad et al., 2018) is a promising approach. Gene-environment interactions could also impact pathogenic complement protein processes in schizophrenia. Second, the relation between serum and CNS complement proteins is unclear, and further studies need to examine cerebrospinal fluid samples as well, across prodromal, early and chronic psychotic states; CSF and serum biomarker profiling has been found to characterize disease state in immune disorders such as multiple sclerosis (Ingram et al 2012) and Systemic lupus erythematosus (Jongen et al 2000). Lastly, innovative studies are needed to directly demonstrate increased pruning in schizophrenia. Clearly, more needs to be learnt to “connect the dots” between these tantalizing observations to move us toward a more complete understanding of at least one pathophysiological pathway to schizophrenia.