Once established, maintaining drug abstinence is a primary goal of substance abuse treatment [1
]. With this goal in mind, we have used a behavioral model of drug relapse liability to examine proteomic changes in the mPFC, a key activation site in drug craving and relapse [14
]. Previously, we have demonstrated mRNA expression and epigenetic changes in the mPFC with this model that persist even with prolonged abstinence from cocaine self-administration [9
]. Alterations in brain mRNA [9
] and protein levels [27
] have been reported in many model systems and may be regulated, in part, through epigenetic changes [9
]. However, long-lasting protein changes have not been examined in this self-administration model of relapse liability.
In studies of cocaine self-administration and abstinence, persistent neuronal protein changes are of such specific interest because they may underlie the potential to relapse even after prolonged abstinence. Grimm and colleagues have shown increases in brain-derived neurotrophic factor (BDNF) protein in the VTA, NAc, and amygdala following 30 and 90 days of withdrawal from cocaine self-administration [29
]. Conversely, increases in the levels of glutamate receptor subunits (GluR1 and NR1) have been observed in these brain areas and in the striatum after similar periods of abstinence [31
]. Of specific interest, changes in glutamate receptor subunit expression have been linked behaviorally to cocaine craving and relapse. Changes in GluR1 and GluR2 in the nucleus accumbens have been independently identified to be associated with increased extinction responding and drug craving [23
Although the aforementioned proteins were not observed in this proteomic screen, a number of additional proteins have been identified that may contribute to the development and/or expression of the increased drug-seeking and drug-taking behavior that occurs during abstinence. While many protein changes that occur during cocaine self-administration return to normal levels sometime between 1 and 100 days of enforced abstinence, many proteins levels remain changed even after 100 days. Others are altered during the abstinence period, after remaining unchanged during cocaine administration. These may be of importance to the incubation of cocaine seeking/taking. The proteins, identified in this study to be changed, create a physiological profile that, when visualized by PCA, allows for an accurate grouping of animals based on the biochemical consistency of the molecular fingerprint of cocaine withdrawal. The result of this is a separation of naïve, non-incubated (1 day), and incubated (100 days) groups strictly by proteomic profile. In a general context, while not producing a predictive model, this visualization tool demonstrates that the mPFC neuroproteome does not return to a naïve state even after extended abstinence from cocaine self-administration. This clustering is an effect that is seen when observing the profiles of changed proteins, but disappears when examining the expression profiles of all protein spots, indicating the importance of this specific subset of proteins.
Of the identified proteins, a number of metabolic proteins were persistently changed, indicating a possible role for altered metabolism associated with withdrawal from cocaine administration, an observation made previously in non-human primates [32
]. In addition, PANTHER analysis of all the identified proteins revealed an over-representation of cell structure proteins among the changes. This list, including dynamin-1, neurofilament medium, and alpha-internexin, highlights the important role that these proteins may play in neuronal restructuring that occurs following cocaine abuse and abstinence (for a review, see [33
]). A number of changed proteins have also been previously identified to play a role in substance abuse, including neurofilament medium, SNAP-25, and dynamin-1. HSP73 is also highlighted as a protein of interest based on its cellular role.
Neurofilament medium is a member of the neural intermediate filament family that also includes neurofilaments heavy and light. Present in the axons of neurons, neurofilaments play a role in axon growth. Although alterations in the levels of neurofilament medium do not persist into periods of abstinence, an increase in levels directly following self-administration could indicate a remodeling of the neuronal architecture. Additionally, a small-magnitude increase in alpha-internexin, a protein known to associate with neurofilaments, reinforces the possibility of increases in axonal growth that could be capable of producing long-lasting changes in neuronal structure. Previous studies have shown decreases in the levels of these neurofilaments in the VTA following repeated cocaine, morphine, or alcohol administration [34
] as well as following chronic nicotine administration [36
Changes in synaptic proteins are well represented in these findings, and demonstrate the importance of the role of synaptic communication in withdrawal-associated behaviors. Dynamin-1, a key player in endocytosis of synaptic vesicles, has also been implicated in substance abuse, and has been studied during periods of drug withdrawal. Garcia-Fuster and colleagues have reported increases in dynamin-1 levels in the cerebral cortex with chronic heroin and morphine administration, as well as heroin withdrawal [37
]. Results from other groups have confirmed this finding following chronic morphine [38
] and have shown that the localization of dynamin-1 at the post-synaptic density is also altered by morphine [39
]. Changes in dynamin-1 levels associated with cocaine administration and withdrawal, however, have not been reported previously.
Another synapse-associated protein, SNAP-25, associated with the soluble NSF attachment receptor (SNARE) complex involved in vesicle exocytosis, is also altered by chronic morphine. Alterations in phosphorylation states of SNAP-25 have been associated with decreases in SNARE complex formation following chronic morphine [40
]; however, this is the first demonstration of changes following cocaine administration. The changes observed in this study provide evidence of alterations in a post-translationally modified form of SNAP-25, which could possibly represent a phosphorylated state. Both dynamin-1 and SNAP-25 are unaltered after 1 day of enforced abstinence; however, both are decreased following 100 days of abstinence, possibly indicating a decrease in the ability for and/or efficiency of synaptic communication that develops during abstinence periods.
HSP73 belongs to the family of heat shock proteins that is involved in response to stressors within cells. Unlike many other heat shock proteins, HSP73 is constitutively expressed. Although changes in levels of HSP73 have been linked to cell death in Parkinson’s disease [41
], HSP73 levels have not been linked to substance abuse. However, levels of HSP73 have been shown to decrease in the embryonic mouse brain upon exposure to ethanol [42
]. Our observed increases in levels of HSP73 indicate a cocaine-induced stress upon cells in the mPFC following cocaine self-administration and acute abstinence.
Much as in the microarray field, there is an ongoing debate over the best statistical methods to minimize both Type I (false positive) and Type II (false negative) errors in discovery proteomic experiments. Power calculations for 2D-DIGE have been proposed that suggest the need for high samples numbers (N) to detect smaller magnitude changes [43
]. The proposed power calculations reduce the likelihood of Type I errors, but retain the possibility of increased Type II error. Additionally, power calculations will be dependent on the species, organ system, and treatment studied. Previously, 2D-DIGE has identified consistent changes of less than 1.4-fold in the brain [44
]. Our statistical analysis was based on our previous experience with functional genomic [9
] and proteomic [18
] studies of drug abuse as well as previous reports of behaviorally-relevant magnitudes of protein expression changes in the brain with cocaine-abuse [23
] and using the standards in the microarray field as guidance [46
]. Even with this rationale for our statistical analyses, we performed a confirmation experiment on dynamin-1 and validated that the 34% decrease observed by 2D-DIGE was recapitulated as a 27% decrease by immunoblotting. As the proteomic field continues to search for high-throughput confirmatory methods, similar to qPCR confirmation of microarray data, verification of proteomic discovery data currently presents a challenge due to the variety of post-translational modifications and isoforms identified, which are not quantified by standard western blotting. New methodologies, such as multiple reaction monitoring, offer potential solutions to this difficulty, but overcoming this challenge is a crucial next step in proteomics to compliment the advances in discovery approaches.
The limitations involved in this type of study revolve around the collective expression of several different cell types (neurons, astrocytes, microglia, oligodendrocytes), each with different gene and protein expression patterns [47
]. Addiction studies are generally focused on the neuronal population, which comprises only 20–30% of the cellular population of the brain [48
]. This mixture of cellular populations has a direct effect on quantitation resulting in a muting of expression differences that occur in a subset of cells. Improvements in laser-capture microdissection may allow for future proteomic studies to examine specific cell types [49
Another challenge to neuroproteomic studies is that proteomic methods are limited in the number of proteins they can examine. Unlike gene expression studies which can examine tens of thousands of genes, proteomic studies, regardless of the specific techniques, can only examine a few thousand proteins at a time [12
]. Without the amplification techniques of molecular biology, proteomic studies often examine only the most abundant proteins.
Unchanged proteins were identified in this study for the purpose of investigating the classes of proteins represented by whole tissue proteomics (Supplementary Table 2
). Indeed, high abundance proteins, such as actin, tubulin, enolases, heat shock proteins, and mitochondrial respiration proteins compromised a large fraction of the over 150 proteins identified. Fractionation techniques, such as synaptosomal or post-synaptic density preparations, among others, may provide an approach to examine less abundant proteins related to cellular regions of interest [39
The strength of this study lies in the extension of proteomic changes that occur with drug administration into the realm of withdrawal periods using a large-scale discovery proteomic platform. Many of the studies referenced above refer to protein levels while drug (cocaine, heroin, and morphine) is still present within the body (or immediately after cessation of drug administration), while few extend analysis into periods of abstinence. This study demonstrates that many protein changes do not persist into periods of abstinence; however, there are many changes that do persist, leaving cocaine-abstinent animals with an altered proteomic profile. Those changes that persist may contribute to specific functions related to relapse liability. To further understand the implications of the changes identified, future studies will focus on localization and confirmation of these changes to specific cellular populations. Furthermore, the need has been highlighted for technological advancements in neuroproteomic studies of drug abuse and abstinence-associated physiology. Additional methodologies will be needed to increase the coverage and sensitivity of neuroproteomic studies to identify proteins involved in psychiatric disorders, including drug abuse.