The postsynaptic density (PSD) is an electron-dense region of excitatory glutamatergic synapses located just beneath the postsynaptic membrane. The PSD was first discovered by electron microscopy in 1956 [1
] and was later found to consist of 30–50 nm-thick, disc-shaped protein structures [2
]. Within these protein structures are several classes of protein families, many of which are involved in processes such as scaffolding and signal transduction. Each of these families are organized in two different structural layers of the PSD: the core and the pallium [4
]. The core is the structural layer located near the postsynaptic membrane, while the pallium is positioned beneath the core and is thought to be more labile.
One group of proteins that has previously been identified in the PSD core is the membrane-associated guanylate kinases (MAGUKs) [5
] (Figure 1
A). These proteins are comprised of three main domains including the PDZ, SH3, and guanylate kinase (GK) domains [5
]. One of the most abundant proteins within the MAGUK family is PSD-95 (also known as DLG4 or SAP90) [7
], which is involved in structural maintenance and signaling through interactions with integral membrane proteins and receptors, protein complexes, and other structural proteins within the PSD [10
]. In addition to PSD-95, the MAGUK family includes PSD-93 (DLG2), SAP-102 (DLG3), and SAP-97 (DLG1).
Guanylate kinase-associated proteins (GKAPs) are another class of proteins found in the PSD core. This family was first isolated by Kim et al. [13
] and found to directly bind to the GK domains of MAGUKs through co-immunoprecipitation assays and immunohistochemistry. GKAPs are often referred to as disks large-associated proteins (DLGAPs, also referred to as SAPAPs), which include four different isoforms designated DLGAP1,2,3, and 4 (Figure 1
A). These isoforms enable the formation of protein complexes with MAGUKs and proteins found in the pallial layer of the PSD.
An additional protein family present in the PSD pallium and associated with these complexes is the SH3 and multiple ankyrin repeat domain protein (Shank) family. As their name suggests, these proteins contain an SH3 domain, as well as ankyrin repeats, a PDZ domain, a proline-rich domain, and a SAM domain. These proteins were first identified by Naisbitt et al. [14
], who demonstrated that the C-terminal region of GKAP binds to the PDZ domain of Shank. There are three Shank isoforms (Shank1,2,3) that are capable of binding to both MAGUKs and GKAPs and have been shown to form a PSD-95–SAPAP–SHANK complex [9
]. This complex has been implicated in scaffolding and organization of signaling complexes at glutamatergic synapses [15
The Homer family of proteins is also found in the PSD pallium. The two characteristic structural regions of Homer proteins include an Enabled/vasodilator-stimulated phosphoprotein homology 1 (EVH1) domain [17
] and a carboxyl-terminal coiled-coil domain [17
]. Homer proteins self-polymerize and interact with Shank proteins, creating a matrix-like structure [4
]. This scaffolding structure is involved in excitatory signal transduction as well as in receptor plasticity [27
]. There are three different Homer
genes (1–3) that are differentially expressed throughout the brain [27
In addition to structural proteins, protein kinases are an important component of the signaling pathways within the PSD pallium. Ca2+
/calmodulin-dependent protein kinase II (CaMKII) is a serine-threonine kinase that comprises approximately 1–2% of the total proteome in the cerebral cortex and hippocampus [28
]. Studies have shown a marked accumulation of CaMKII at the PSD with increasing levels of neuronal excitation [4
]. CaMKII has also been implicated in NMDA-dependent long-term potentiation (LTP) [4
] through regulation of its activity by Ca2+
/calmodulin and autophosphorylation.
There is increasing interest in understanding the functions of proteins in the intricate PSD network because of their potential involvement in a wide variety of neuropsychiatric disorders. For instance, several reports have linked Shank3 to autism spectrum disorder (ASD) [15
]. Specifically, deletion of the Shank3B isoform in mice resulted in an ASD-like, compulsive grooming phenotype, which was more prominent than the other Shank3-associated phenotypes investigated in this report [15
]. Another study observed a similar excessive grooming phenotype of a different genetic Shank3B knockout mouse, further implicating Shank3B in ASD-like behavioral disorders [35
]. Similarly, mutations in either PSD-95 or SynGAP have been shown to be associated with intellectual disorders and autism [37
]. Furthermore, PSD proteins such as DLG isoforms, DLGAP1, Gria2/3, Grin2a/b, CaMKII, and Homer isoforms have all been implicated in schizophrenia, among many other disorders [40
]. It is apparent from this growing list that studying the organization and function of proteins within the PSD has become an important focus in neuroscience research. Fractionation methods for enriching PSD proteins were developed decades ago [2
]; however, because the PSD is not enclosed in a bilayer, it can be challenging to minimize contamination of the PSD fraction with other subcellular proteins [46
]. Apart from the enrichment method, the structure of synapses themselves can make analyses difficult. Synapses differ significantly from one another and can change their composition rapidly, making reproducibility and accuracy of the analysis important [47
]. Despite these challenges, researchers have made significant efforts to study the proteome of the PSD, particularly through the use of mass spectrometry [27
]. Note, however, that many of the PSD fractionation methods use Triton-X100, a detergent that is not compatible with mass spectrometry analysis. Therefore, care must be taken when preparing a PSD fraction to minimize detergent interference.
Mass spectrometry analysis of PSD fractions from mouse and human cortical tissue identified 1556 and 1461 proteins, respectively [53
]. Interestingly, there was a 70% overlap of proteins in the mouse and human PSDs. A later study identified 2876 PSD-associated proteins from mouse brain tissue using immunopurification prior to mass spectrometry analysis [57
]. Recently, label-free quantitation was performed on 48 PSD samples from 12 human neocortical brain regions, identifying 1213 proteins in total [51
]. While these discovery studies have made significant progress identifying PSD proteins, targeted mass spectrometry-based assays are needed to provide the highest possible sensitivity, quantification precision, and accuracy [59
]. Our group previously used multiple reaction monitoring (MRM) coupled with stable-isotope peptide standards (SIS) to quantify 112 rat synaptic proteins [58
]. Though this assay has made significant improvements in the quantitation of PSD proteins, it lacks high mass accuracy.
Currently, there are two major approaches for targeted, high-mass-accuracy quantitative mass spectrometry. The first method is data-independent analysis (DIA), which was first proposed by Venable et al. [60
]. DIA uses sequential window acquisition to fragment and quantify all precursor and product ions within a sample [61
]. Unlike data-dependent methods, DIA offers high reproducibility and quantitation, while maintaining sensitivity at higher levels of multiplexing [65
]. One study has already demonstrated the use of DIA analysis on fractionated PSD samples from mouse hippocampal tissue, which resulted in the identification of 2102 protein groups in the PSD fractions [66
]. The second approach is a more targeted method called parallel reaction monitoring (PRM). Like MRM/selected reaction monitoring (SRM) methods, PRM offers similar accuracy and reproducibility; however, it provides a wider dynamic range and improved selectivity [67
Given the advantages of these targeted methods, we developed new DIA and PRM assays to quantify PSD proteins. For the PRM assay, heavy labeled peptides were synthesized and used as internal standards for accurate protein quantitation. Two different mouse datasets were used to evaluate the performance of these methods: PSD-enriched fractions versus pre-fractionation, and wild-type (WT) versus Shank3B knockout (KO) PSD fractions. These assays enabled accurate quantitation of PSD proteins and provide promising tools for future PSD proteomics studies.
Collectively, these assays demonstrated the power and selectivity of targeted mass spectrometry for quantitation of PSD proteins. Performing PRM and DIA assays in parallel enabled the identification and quantitation of over 2000 proteins before and after enrichment of the PSD from mouse cortical tissue. Many of these proteins displayed similar trends in both assays, including the scaffolding protein Shank3, which had significantly decreased expression in Shank3B knockout PSD samples compared to wild-type samples. Furthermore, proteins that have routinely been identified in PSD fractions in other proteomics studies, such as PSD-95, DLGAPs, and glutamate receptors (Gria), displayed significantly increased expression in PSD fractions compared to pre-enrichment samples in both PRM and DIA assays.
Though many of the proteins displayed similar expression profiles in both assays, there were also some discrepancies which can be attributed to differences in the number and specific peptides identified and quantified in each protein. For instance, CaMKIIa was significantly increased (p = 0.015) in Shank3B KO PSD samples compared to WT after DIA analysis, which identified 14 total peptides for CaMKIIa. However, PRM analysis of three peptides corresponding to CaMKIIa in the same samples resulted in a quantitative profile that was trending in a similar direction, but the expression difference was not significant with a t-test. Conversely, Csnk2a1 displayed a significant decrease (p = 0.017) in expression in KO versus WT samples after PRM analysis, while there was no significant difference in expression after DIA analysis of the same samples. However, the DIA analysis used five total peptides to quantify Csnk2a1, and only two out of the three PRM target peptides were identified and included in the DIA quantitation. These differences illustrate the importance of careful design, optimization, and validation of targeted assays for quantitative proteomics. In addition to mass spectrometry method development, sample selection also becomes important to determine the utility of the assays. This is one reason why two different sample sets were used for initial validation of the PSD targeted assays.
The quantification of proteins from pre-fractionated samples and PSD-enriched samples of mouse cortical tissues was initially performed to demonstrate the selectivity and utility of these assays for different sample types. Determining the limit of detection and quantitation of these proteins allows one to assess the degree of PSD enrichment and the level of contaminating proteins, which is commonly performed using methods such as immunoblot analysis. The second comparison of PSD proteins from Shank3B KO and WT mice was selected on the basis of prior Shank3-related proteomic analyses [15
]. The Shank3B knockout line used in our study was originally generated by homologous recombination that resulted in the disruption of the PDZ domain of Shank3B (exon 13-16) [15
]. Initial proteomic characterization of this knockout line was performed in striatal synapses using immunoblot analysis, which revealed a significant decrease in protein expression of many characteristic PSD proteins in KO versus WT, including PSD-95, glutamate receptors, and CaMKIIa [15
]. Interestingly, DIA analysis of mouse cortical tissue revealed inverse results to those seen in the Peça et al. study; however, this difference may be attributed to the brain region analyzed, as Shank3, but not Shank1 or Shank 2, is highly expressed in the striatum of mouse brain [15
]. Another study used ion-mobility-enhanced DIA analysis to assess changes in the striatal and hippocampal proteomes of Shank3Δ11-/-
knockout mice, revealing significant decreases in the expression of glutamate receptors, including Grin1, Grin2B, Gria1, and Gria2, compared to wild-type [74
]. Both PRM and DIA analysis of mouse cortical tissues did not show significant differences in glutamate receptors between WT and Shank3B KO animals, which again could be due to differences in the brain regions analyzed and to differences in the knockout mouse lines used.
In conclusion, we report on the validation and utilization of both PRM and DIA assays for quantitation of PSD proteins, which have now been demonstrated on two different sample sets. These assays provide a high-mass-accuracy, reproducible method for quantitation of PSD proteins that can be used as tools for a variety of applications in mouse or rat brain tissue. Together, the results from these analyses show promise for future studies of PSD proteomics and neurological disorders.