A perspective on the potential involvement of impaired proteostasis in neuropsychiatric disorders

Abstract

Recent genetic approaches have demonstrated that genetic factors contribute to the pathologic origins of neuropsychiatric disorders. Nevertheless, the exact pathophysiological mechanism for most cases remains unclear. Recent studies have demonstrated alterations in pathways of protein homeostasis, proteostasis, and identified several proteins which are misfolded and/or aggregated in the brains of patients with neuropsychiatric disorders, thus providing early evidence that disrupted proteostasis may be a contributing factor to their pathophysiology. Unlike neurodegenerative disorders in which massive neuronal and synaptic losses are observed, proteostasis impairments in neuropsychiatric disorders do not lead to robust neuronal death but likely act via loss and gain of function effects to disrupt neuronal and synaptic functions. Furthermore, abnormal activation of or overwhelmed ER and mitochondrial quality control pathways may exacerbate the pathophysiological changes initiated by impaired proteostasis, as these organelles are critical for proper neuronal functions and involved in the maintenance of proteostasis. This perspective article reviews recent findings implicating proteostasis impairments in the pathophysiology of neuropsychiatric disorders, and explores how neuronal and synaptic functions may be impacted by disruptions in protein homeostasis. A greater understanding of the contributions by proteostasis impairment in neuropsychiatric disorders will help guide future studies to identify additional candidate proteins and new targets for therapeutic development.

Introduction

Neuropsychiatric disorders including autism spectrum disorder (ASD), schizophrenia (SZ), bipolar disorder (BD), and major depressive disorder (MDD) affect a substantial population worldwide and altogether represent a significant healthcare issue with great socioeconomic impact. Whilst approximately 1-2% of the general population is estimated to be affected by ASD (1.85%) (1), SZ (<1%) (2–6), and BD (2.6%) (7), recent data from the United States revealed that about 7% of the adult population and approximately 14% of adolescents experienced at least one major depressive episode in a year (8). Through traditional linkage analyses of families with a medical history of neuropsychiatric disorders, genetic factors have been known for decades to play a significant role in the pathophysiological process. With the advent of whole genome and exome sequencing studies, our understanding of the genetic factors causing these disorders have charted significant progress. Notably, de novo (absent from both parents and siblings) point mutations and copy number variations (CNVs) are frequently identified from patients with neuropsychiatric disorders, thus helping to explain many sporadic cases in which there are no family history of the disorders (9). Furthermore, whereas rare de novo CNVs and point mutations disrupting protein structure/function often have larger effect sizes and therefore may contribute directly to pathophysiology, such massive sequencing studies have also identified many common variants that on their own may not exert a heavy pathogenic burden but collectively confer enhanced risk. Therefore, such observations suggest that these disorders are polygenic and highly complex in nature, and that genetic factors alone may not fully account for all cases. As a result, recent studies have begun to explore the impairment of protein homeostasis, or proteostasis, as a putative mechanism which may contribute to the pathophysiology of neuropsychiatric disorders.

Disruption of protein homeostasis in neuropsychiatric disorders

In broad terms, proteostasis is maintained at multiple levels during the lifetime of a protein. Given that most cellular functions are mediated by proteins, dysregulation of proteostasis is known to have detrimental consequences. The correct folding and assembly of proteins into monomers and multimeric complexes are central to their normal functions. During and following protein synthesis, various intra- and inter-molecular interactions dictate how the polypeptide chain will be organized into its unique three-dimensional structure (10). However, due to many interconvertible conformational states, proteins can be entrapped into non-native conformations by energetic barriers. In addition to influence by such intrinsic properties, environmental factors including temperature, pH, and salt concentration can also alter protein conformations and influence their misfolding and aggregation potentials. Therefore, a number of cellular quality control mechanisms including co- and post-translational processes assist polypeptide chains to adopt their native structures (Figure 1). For instance, the unfolded protein response (UPR) and integrated stress response (ISR) pathways are primarily activated via stress signals triggered by aberrant protein folding inside the endoplasmic reticulum (ER), and in turn act to slow down de novo protein synthesis and enhance protein folding capacity (Figure 2). Aside from such homeostatic measures, degradative pathways such as autophagy and ubiquitin proteasomal system (UPS) can be used for the removal of unnecessary or damaged proteins (11). Specially, if protein misfolding and/or unfolding occurs inside the ER, ER-associated degradation (ERAD) is used for the retro-translocation of the polypeptide chain out of the ER and sent to the UPS for degradation (12). In addition, proper proteostasis also includes the regulation of post-translational modifications, such as proteolytic cleavage, phosphorylation, and acetylation, that can affect protein-protein interactions and enzymatic activities involved in various cellular signaling pathways. Notably, these post-translation modifications also have been demonstrated to have profound influence on the conformational dynamics and stability of proteins, thereby further disturbing proteostasis (13–15). Impaired proteostasis may occur when the aforementioned homeostatic mechanisms are impaired by genetic and non-genetic factors, potentially allowing aberrant protein conformational states to progress into soluble oligomers and insoluble aggregates. Altogether, it is clear that maintenance of protein homeostasis is critically important to proper cellular functions.

Figure 1. Homeostatic pathways against protein misfolding and aggregation.

Various mechanisms have been developed by the cell to protect against protein misfolding and aggregation. During and following mRNA translation, chaperones assist the nascent polypeptide chain via co-translational and post-translational mechanisms to fold properly into the correct “native” three-dimensional conformations. Once correctly folded, proteins can carry out their functions as monomeric proteins or multimeric protein complexes. However, they may become unfolded due to cellular stresses and require chaperones to return to their native conformations or shuttle for removal by proteasomal or autophagic degradation. If such misfolded proteins are not recognized by chaperones for refolding or degradation in a timely manner or the aforementioned cytoprotective mechanisms have been overwhelmed, these misfolded proteins may assume non-native conformations. Such non-native conformations remain as non-functional monomers or form soluble oligomers which could have functional properties different from their native monomeric counterparts. Furthermore, additional interacting proteins may be sequestered into these conformationally-altered species via heterotypic interactions and result in network-wide proteostatic disruptions.

Figure 2. ER stress and aberrant overactivation of ER quality control pathways.

Cell surface and secreted proteins are typically translated by ribosomes and imported into the ER in a co-translational manner. Misfolded proteins, due to pathologic mutations or reduced protein folding capacity in the ER, may disrupt ER-Golgi trafficking, which results in the ER accumulation of affected proteins, and increase ER-associated degradation (ERAD). Both of these could have negative effects as the expression and subcellular localization required for proper cell functions may be greatly impacted. Furthermore, the resulting ER stress consequently initiates the unfolded protein response (UPR), which triggers PERK, ATF6, and/or IRE1α-dependent pathways to enhance the ER protein folding capacity. Meanwhile, the overload in this protein folding capacity is reduced by suppressing global protein synthesis through the phosphorylation of eIF2α. Chronic eIF2α phosphorylation due to persistent ER stress may have detrimental effects as de novo protein synthesis is required for various types of synaptic plasticity and thereby disrupt the proper functioning of affected neurons.

Much of our current knowledge concerning proteostasis impairments in the brain derives from extensive work focussed on neurodegenerative disorders. Decades-old observations from post-mortem human brain samples of patients suffering from Alzheimer’s Disease (AD), Parkinson’s Disease (PD), Huntington’s Disease (HD), amyotrophic lateral sclerosis (ALS), and frontotemporal lobar degeneration (FTLD) have revealed protein misfolding and aggregation to be the principal culprits (16). Each of these diseases is characterized by the visible formation of insoluble proteinaceous deposits of a disease-associated protein such as amyloid beta (Aβ), α-synuclein, huntingtin, and TDP-43, respectively, in affected neurons. The formation of such insoluble aggregates is the result of an imbalance between protein misfolding/aggregation and cellular cytoprotective mechanisms, which is frequently caused by either disease-associated mutations (e.g., α-synuclein A53T) (17) or alterations in their processing pathways (e.g., γ-secretase cleavage of APP by presenilin 1/2) (18–21). The aggregated protein in turn accumulates in a time-dependent manner and disrupts normal cellular functions. With time, the affected neurons ultimately succumb to apoptotic or other forms of cell death and manifest as the corresponding neurodegenerative disorder following substantial neuronal loss (22).

Interestingly, a growing list of studies using post-mortem brain samples or blood collected from patients with neuropsychiatric disorders have identified changes indicative of alterations in proteostasis. Specifically, protein expression changes in regulators of autophagy (23–25), UPS (26–34), and ER quality control (ERQC) pathways (35–42) have been associated with patients diagnosed with ASD, schizophrenia, bipolar, and major depressive disorders (Table 1). In fact, protein levels of critical regulators of autophagy such as Atg7, Beclin 1, LC3, and p62/SQSTM1 have been shown to be reduced in BD and SZ patients; whereas increases in mRNA transcripts of ATG12, BECN1, and MAP1LC3A/B/C were observed in MDD (23–25). Notably, previous studies have revealed that many components of the UPS, ranging from ubiquitin itself, E2 and E3 enzymes, to subunits of the proteasome, are downregulated in SZ patients (27–29,31–34). Furthermore, it should be noted that reduced proteasomal degradative activities were observed in post-mortem brain samples from SZ patients (26), and several recent studies have observed either increased levels of ubiquitinated (43,44) or insoluble proteins (44) from a subset of SZ brain samples; further implicating proteasomal dysfunction in the disorder. Importantly, the study that identified increases of both ubiquitinated and insoluble proteins in SZ brains was able to replicate the finding in three distinct sample sets from independent brain banks (44). Furthermore, mass spectrometric analyses revealed that insoluble proteins were enriched in neurons and may exhibit cell type specificity. In addition, multiple studies have revealed alteration of UPR and ERAD pathways for ASD, BD, MDD, and SZ. Gene expression analyses have found upregulation of various components of the UPR and ERAD pathways in patient-derived tissue samples (35–40,42). In contrast, two studies demonstrated reduced UPR activation in response to tunicamycin-induced ER stress using lymphocytes collected from BD patients (41,45). The discrepancy between these studies may reflect a difference between the response to acute and chronic ER stress, but suggest that ER quality control is dysregulated in patients with neuropsychiatric disorders. Altogether, these findings from clinical samples indicate that disrupted proteostasis is observed in neuropsychiatric disorders, but precisely how proteostasis impairments contribute to their pathophysiology remains elusive.

Table 1.

Alterations in proteostasis pathways observed in brain or blood samples from patients with neuropsychiatric disorders

Disorder	Pathway affected	Genes/proteins affected	Citation
Autism spectrum disorder	ERQC	ATF4, ATF6, CHOP, IRE1, PERK, total and spliced XBP1 mRNA ↑	(40)
Bipolar disorder	Autophagy	MAP1LC3A, SQSTM1 mRNA ↓ mRNA ↓ LC3, p62/SQSTM1 proteins ↓	(25)
	ERQC	phospho-EIF2α, CHOP, GRP78 proteins ↓ (lymphocytes subjected to tunicamycin-induced ER stress)	(45)
		CHOP, XBP1 mRNA ↓ (lymphocytes subjected thapsigarcin- or tunicamycin-minduced ER stress)	(41)
		BIP mRNA ↑ Total and unspliced XBP1 mRNA ↓	(39)
Major depressive disorder	Autophagy	ATG12, BECN1, LC3 mRNA (isoform not specified) ↑	(24)
	ERQC	BIP, CHOP, EDEM1, XBP1 mRNA ↑	(35)
		Calreticulin, GRP78, GRP94 protein ↑ (suicide subjects only)	(38)
		ATF4C, GRP78, GRP94 mRNA ↑ (suicide subjects only)	(42)
Schizophrenia	Autophagy	BECN1 mRNA ↓ BCL2 mRNA ↑	(23)
		ADNP mRNA ↑ ADNP2 mRNA ↑
		ATG7 protein ↓	(33)
	ERQC	BIP protein ↑ PERK protein ↓ phopho-IRE1α ↓ XBP1 mRNA splicing ↑	(37)
		EDEM2, HRD1, UGGT2 protein ↑	(36)
		ATF4, PERK protein ↓	(88)
	UPS	Trypsin-like activity (nucleus) ↓ Chymotrypsin-like activity (cytosol) ↓	(26)
		19S regulatory particle subunits Rpt1, Rpt3, Rpt6 protein ↓ 11S regulatory particle α subunit protein ↓	(27)
		HERPUD1, PSMA1, PSMB6, PSMC6, PSMD8, PSMD9, PSME1, UBB, UBE2D1, UBE2D3, UCHL1 mRNA ↓ UBL4A mRNA ↑	(28)
		PSMA1, UCHL1, USP9X mRNA ↓	(34)
		UBE2N mRNA ↓	(29)
		UBE2K mRNA and protein ↑	(30)
		UBE3B mRNA and protein ↓	(31)
		UCHL1, USP14 mRNA ↓	(32)
		NEDD4, PIAS3 RNF7, UBA3, UBA6, UFL1, protein ↓	(33)

Notably, unlike in neurodegenerative disorders, there have yet to be reports of visible aggregates in post-mortem brain samples of patients with neuropsychiatric disorders. Instead, proteins are described to be conformationally-altered on a biochemical basis in that they are reported to be increased in detergent-insoluble fractions of patient-derived samples (46–49) or display aggresome-like punctate subcellular localizations upon overexpression in immortalized cell lines or primary neurons (47,50–53). It remains to be seen whether this trend will continue to hold true while additional conformationally-altered proteins are identified and linked to these disorders in future studies. Importantly, unlike neurodegenerative disorders, neuropsychiatric disorders are not typically degenerative in nature, as robust neuronal cell death is not typically detected. Furthermore, in vivo prion-like spread of visible protein aggregates is not generally observed in neuropsychiatric disorders. These observations likely reflects a difference in the biochemical and biophysical properties of the pathogenic proteins involved. Such differences may be the critical features that distinguish between aggregation-prone proteins in these disorders and the proteins responsible for neurodegenerative disorders. The substantial loss of specific neurons in neurodegenerative disorders suggests that treatment may require therapeutic strategies such as stem cell replacement (54) or trans-differentiation (55) for full recovery. In contrast, the absence of extensive neuronal cell death in neuropsychiatric disorders provides a glimpse of hope that these conditions are potentially reversable if impaired proteostasis can be restored.

Loss and gain of function effects by protein misfolding and aggregation in preclinical studies

Whilst the abovementioned studies have observed changes in protein degradation or stress pathways in patient-derived samples, there have also been direct identification of proteins that are increased in the detergent-insoluble fractions generated from subsets of post-mortem human brain samples of patients with neuropsychiatric disorders compared to healthy controls. This list of conformational-altered proteins includes DISC1 (47,49,53), dysbindin-1 (47), collapsin response mediator protein 1 (CRMP1) (48), Trio-binding protein 1 (TRIOBP-1) (50), and GABA_A receptor-associated proteins (GABARAPs) (46). In addition, other studies have also demonstrated disease-associated mutations in DISC1 (53), neuronal PAS domain protein 3 (NPAS3) (52), and DEAD-box helicase 3 X-linked (DDX3X) (51) to enhance their aggregation potentials. The elucidation of how conformational alterations of these proteins impair neuronal and synaptic functions will provide important mechanistic insight into how disrupted proteostasis may ultimately cause neuropsychiatric disorders.

Through previous studies based on proteins associated with neurodegenerative disorders and the abovementioned studies on proteins which are misfolded or aggregated in neuropsychiatric disorders, the common molecular mechanism underlying their pathophysiology may be the loss and gain of functions of the protein of interest. The increase in insoluble protein due to misfolding effectively reduces the functional protein pool, as they can no longer assume the correct three-dimensional structures to properly perform their functions. In addition, reduced solubility may also affect their subcellular localization such that they are no longer in the proper locations to carry out their normal functions. For example, the misfolding of DISC1 was found to disrupt the interaction with nuclear distribution protein nudE-like 1 (NUDEL/NDEL1), which depends on an oligomeric state of DISC1 (49,56). Functionally, this interaction has been shown to be critical to DISC1’s function in regulating microtubular dynamics (57) and resulted in schizophrenia-like behavioural deficits when it was disrupted in mouse models (58–60). In addition, the overexpression of full-length wild-type DISC1 in a transgenic rat model was also shown to result in aggregation and mislocalization of DISC1. Such changes may have limited the functional pool available to mediate its functions in regulating dopamine homeostasis (61), the disruption of which contributed to anxiogenic-like behaviour (62) and deficits in learning and memory (63,64).

Alternatively, proteostasis impairments causing protein misfolding and aggregation have also been shown to have gain of function effects that disrupt neuronal functions. As the term implies, such effects cannot be recapitulated simply by the reduction or deletion of the aggregation-prone protein of interest but are caused by the enhanced recruitment and sequestration of interacting proteins into the misfolded and aggregated assemblies formed by the conformationally-altered proteins. An example of this observation is the abnormal enhancement in interaction between p62 and GABARAPs in autophagy-deficient neurons (46,65). p62 belongs to a group of proteins which act as autophagic cargo receptors that recognize ubiquitinated proteins and target them to autophagosomes for degradation through interactions with LC3 (66). In these studies, autophagy was impaired in Atg7 knockout (KO) or Ulk2 heterozygous mice, which consequently led to the accumulation and formation of p62⁺ aggregates and thus they, along with its cargo, no longer can be degraded by autophagy. The aggregates in turn sequester and effectively reduce the functional pool of GABARAPs required for the proper trafficking of GABA_A receptors to the cell surface. The resulting reduction of surface GABA_A receptors decreased inhibitory GABAergic neurotransmission, thereby disrupting excitatory-inhibitory (E-I) balance in affected brains and causing ASD and SZ-like behavioural deficits. A similar gain of function effect was observed for purified DISC1⁺ aggresomes (47), as they could recruit and sequester heterologous dysbindin, a protein previously linked to SZ (67). Although the functional consequences of this aberrant sequestration was not determined, the authors speculated that the mislocalization and reduction of a significant soluble pool of dysbindin would likely have compromised its functions.

As demonstrated by these studies, the interplay or co-aggregation between distinct proteins is of interest, since they could result in much broader effects. Notably, in addition to being sequestered into aggregates formed by proteins associated with neurodegenerative disorders (e.g., huntingtin and TDP-43) (68,69), DISC1 itself has also been demonstrated to enhance the aggregation propensity of other proteins such as dysbindin and CRMP1 (47,48). Given the intrinsic ability of DISC1 to form aggregates spontaneously (68) and interact with various proteins (70), it will be of interest to examine which interacting partners are also sequestered and differentially aggregated in the brains of patients with neuropsychiatric disorders. For instance, studies have shown how alterations in cellular environment can influence the propensity of DISC1 to undergo misfolding and aggregation. It was previously demonstrated that the level of dopamine can enhance DISC1 aggregation (61), and a recent study revealed that viral infection increases the expression of α-synuclein and DISC1 by disrupting their clearance via autophagy, thereby tipping the balance towards their aggregation (71). Furthermore, functional studies to examine how the conformationally-altered proteins and other proteins which were secondarily recruited into them impact neuronal functions will provide novel insights into how impaired proteostasis ultimately leads to pathophysiological changes.

Finally, it should be noted that there have also been demonstrations of “functional” aggregates of proteins such as CPEB and Orb2 (72–74). Whereas the discussion presented here on protein misfolding and aggregation primarily focused on their detrimental effects on neuronal and synaptic functions relating to neuropsychiatric disorders, some regulated conformational changes to functional prion-like aggregates or amyloids are required for proper brain functions. Many important questions regarding the distinguishing features between functional and disease-causing protein aggregates and the molecular mechanisms underlying their conformational switches remain to be addressed (72,73). New insights into these functional aggregates will provide a better understanding of the misfolding and aggregation of proteins associated with neuropsychiatric disorders, and perhaps lead to novel strategies to prevent or reverse the aberrant conformational changes of the disease-causing proteins.

Potential involvement of ER and mitochondria quality control pathways

Aside from direct effects caused by misfolded and aggregated proteins due to the aforementioned loss and gain of function mechanisms, disturbances in proteostasis may further disrupt neuronal and synaptic functions by overwhelming cellular cytoprotective pathways. We will next discuss about two signalling pathways which normally provide the cell with additional means to maintain proteostasis but, once overwhelmed, may become vulnerable and consequently exacerbate the pathology of neuropsychiatric disorders.

As described previously, alterations in ER quality control pathways have been observed in both post-mortem brain samples or leukocytes collected from SZ (36,37), BD (39,41,45), and MDD (35,38,42) patients. These pathways, including ERAD and UPR, are activated due to ER stress caused by protein misfolding within the ER (Figure 2). ETPR consists of three major pathways involving activating transcription factor (ATF6), inositol-requiring transmembrane kinase/endoribonuclease 1α (IRE1α), and PKR-like ER kinase (PERK) (74). Whereas the ATF6 and IRE1α pathways mainly act to enhance the ER protein folding capacity via transcriptional mechanisms, PERK induces an ATF4-dependent transcriptional response but also activates ISR, which suppresses global protein synthesis in a eIF2α-dependent manner (75). Notably, by using patient-derived samples from various neurodegenerative disorders, these pathways were previously shown to be activated in response to the increased ER stress caused by proteostasis impairments and contribute to the disorders (76–80). Their aberrant activation may therefore contribute to neuropsychiatric disorders in a similar manner.

Although it remains unclear precisely how abnormal activation of these ER quality control pathways disrupts neuronal functions and contribute to pathophysiology, there is evidence that hints at their involvements in neuropsychiatric disorders. Several ASD-associated mutations in synaptic cell adhesion molecules such as neuroligin-3 and −4, CNTNAP2, and CADM1 were found to impair their trafficking to the cell surface and cause them to accumulate in the ER, thereby activating UPR (81–86). Notably, the synaptic deficits of NLGN3 R451C mutant mice were rescued by suppressing one of the downstream UPR pathways (86), thus highlighting its potential involvement. Furthermore, ER stress induced by tunicamycin treatment alone was found to elicit social behavioural deficits and abnormal hyperactive brain connectivity in an UPR-dependent manner (87). By contrast, protein levels of PERK and ATF4 were observed to be reduced in SZ patients and forebrain-specific PERK deletion resulted in SZ-associated behavioural deficits (88). Together, these studies indicate that disturbances in ER-mediated protein folding and trafficking caused by abnormal ER stress response may contribute to the development of neuropsychiatric disorders, and therefore warrants further investigations.

In addition, the global shutdown of protein synthesis induced by phosphorylated eIF2α can lead to deficits in synaptic plasticity (89–91) and in turn contribute to the pathophysiological changes that occur in neuropsychiatric disorders. Notably, aberrant UPR/ISR activation leading to increased eIF2α phosphorylation has been shown to contribute to the pathophysiology in neurodegenerative disorders, as genetic and pharmacologic inhibitions of eIF2α were previously demonstrated to reverse learning and memory deficits in animal models of Alzheimer’s disease (92–96). To date, two studies have demonstrated the beneficial effects of pharmacological inhibition of eIF2α phosphorylation in mouse models of neuropsychiatric disorders (97,98), thereby revealing a detrimental role of eIF2α phosphorylation in their pathophysiology. Further work to examine whether inhibition of eIF2α phosphorylation provides rescue effects in additional animal models or iPSC-derived cell models will be crucial in understanding how the aberrant eIF2α phosphorylation by UPR/ISR overactivation contributes to the pathophysiology of these disorders.

Aside from the endoplasmic reticulum, the mitochondrion is another organelle of interest that may be affected by disrupted proteostasis and contribute to neuropsychiatric disorders. Recent studies have revealed the mitochondrion to be involved in proteostasis in a direct manner (Figure 3). Specifically, a recent study identified ATP as a hydrotrope and demonstrated its ability to directly prevent liquid-liquid phase separation and protein aggregation (99). As such, mitochondrial ATP production may be crucial for protein homeostasis. In addition, a process termed mitochondria as guardian in cytosol (MAGIC) was recently described in yeast that can import a suite of aggregation-prone cytosolic proteins into the mitochondria for their degradation (100). Because most mitochondrial proteins are encoded by the nuclear genome, they are typically translated in the cytosol and imported into the mitochondria via the translocases of the outer membrane (TOM) and the inner membrane (TIM) complexes (101). An elaborate network of proteases and chaperones then ensures proper folding and targeting of these imported proteins to the different mitochondrial compartments. In particular, MAGIC utilizes the TOM (Tom70 and Tom40) and TIM (Tim23) complexes to import cytosolic protein aggregates and degrades them using the mitochondrial protease Pim1. In human cells, a similar pathway involving the FUNDC1-HSC70 interaction and LONP1 mitochondrial protease was found to be crucial for the translocation and degradation of mitochondrion-associated protein aggregates (MAPAs) (102). Hence, there appears to be an evolutionarily-conserved pathway that allows the mitochondrion to provide an additional means to deal with cytosolic protein aggregates when cytosolic proteostasis is impaired. Notably, the import of cytosolic protein aggregates into the mitochondria also allows for their temporary storage and subsequent bulk removal via mitophagy, the process through which whole mitochondria are degraded by autophagy (103).

Figure 3. Involvement and vulnerability of mitochondria.

Nuclear-encoded mitochondrial proteins are normally imported by the TOM/TIM complexes and processed into mature proteins by mitochondrial proteases. The mitochondrial unfolded protein response (UPR^mt) is triggered by a decline in mitochondrial protein import to compensate and correct for these changes. Recently, a process called MAGIC was shown to utilize the mitochondrial protein import machinery to transport misfolded and aggregated proteins into the mitochondria for degradation. This was further hypothesized as a means to ensure proper removal of misfolded and aggregated proteins as the mitochondria can be selectively removed via mitophagy. However, the import of problematic proteins into the mitochondria may also lead to mitochondrial dysfunctions if they are not properly dealt with in a timely manner. Critical mitochondrial functions such as Ca²⁺ homeostasis and ATP production may be disrupted and further disturb neuronal functions such as vesicular release and local translation, and could in turn greatly impact neurotransmission and synaptic plasticity, respectively.

However, the mitochondrion’s ability to handle misfolded and aggregated proteins, either formed internally or imported from the cytosol, is not unlimited. Hence, a mitochondrial unfolded protein response (UPR^mt), which involves mitochondria-to-nucleus communication to initiate a transcriptional program to restore mitochondrial proteostasis, is triggered upon mitochondrial dysfunction (104–108). Notably, UPR^mt and mitochondrial functions were dysregulated in response to Aβ-induced proteotoxic stress and UPR^mt induction provided protective effects (109), thus implicating mitochondria in coping with impaired proteostasis. It will be of interest to examine in patient samples for abnormal UPR^mt activation or alterations to the pathway as an additional marker of proteostasis impairment since it has yet to be examined in the context of neuropsychiatric disorders.

Given the mitochodrion’s direct role in proteostasis via MAGIC and UPR^mt, disruptions of proteostasis in affected neurons could potentially overwhelm such pathways and in turn affect mitochondrial functions. While mitochondrial defects have traditionally been associated with various neuropathies and myopathies, recent studies have identified mitochondrial dysfunctions in patient samples and animal models of neuropsychiatric disorders, most notably for BD and ASD (110–119). Since the mitochondrion has been shown to directly dysregulate neurotransmission and synaptic plasticity through its involvement in ATP production and calcium homeostasis (120–127), mitochondrial dysfunction may further exacerbate the pathophysiology of neuropsychiatric disorders by impairing such functions.

Conclusion

Despite the complexity of neuropsychiatric disorders, genetic studies during the past decades have clarified a genome-wide landscape of genetic risk for these disorders. Nonetheless, genetic factors still cannot explain the full pathology of many sporadic cases. Accordingly, it is important to consider how cellular environments impact protein structure, and in turn the propensity for protein misfolding and aggregation. Although these cellular environmental factors alone unlikely trigger proteostasis impairments and result in pathophysiology, they may be disease modifying factors that enhances susceptibility or influence severity of the disorders. Further understanding of how genetic and cellular environmental factors conceitedly impact proteostasis will provide new insights into how they ultimately contribute to the observed pathophysiologic changes.

The involvement of impaired proteostasis in neuropsychiatric disorders reviewed here helps to explain, at least in part, the complex nature of these disorders. The additive effects of multiple subtle perturbations in protostasis combine together to result in system-wide deficits through various loss and gain of function mechanisms which are inherent to protein misfolding and aggregation. Recent studies have implicated impaired proteostasis to be a key underlying mechanisms for these disorders and illustrated how such disruptions may contribute to the pathophysiological process. Historically, the study of various neurodegenerative disorders began with traditional microscopic observations of insoluble proteinaceous deposits in affected neuronal subpopulations using biopsy and post-mortem brain samples from patients, which was followed by molecular dissection of such insoluble materials. The molecules identified in this process are now regarded as key leads for mechanistic understanding and biomarkers for neurodegenerative disorders. We may optimistically hope for similar success in research on neuropsychiatric disorders in the future. In order to achieve this goal, a continued search for additional disease-associated aggregation-prone proteins may be fruitful. Since neuropsychiatric disorders are not characterized by visible protein aggregates via histological analyses, future efforts to identify new aggregation-prone proteins involved in pathophysiology will need to be made primarily through biochemical approaches. By utilizing unbiased approaches such as epitope discovery and quantitative mass spectrometry like those employed to identify CRMP1 and GABARAPL2 (46,48), novel aggregation-prone candidate proteins and their interactomes can be identified using brain samples from patients and iPSC-derived neurons or brain organoids (128).

In addition to identifying new pathogenic aggregation-prone proteins, it will be critical to determine precisely how the changes in protein abundance, conformation, and/or aggregation status identified in patient samples (Table 1) ultimately impact protein homeostasis, and in turn disturb neuronal and synaptic functions. Once additional proteins have been validated for their associations with the disorders, they can in turn be used as a biochemical endophenotype through which patients may be classified into subgroups for mechanistic studies to better understand the causes and consequences of the underlying disturbances in proteostasis. Such an experimental strategy may provide a possible means of establishing these candidate aggregation-prone proteins as new markers for research and possibly for diagnosis and treatment.