Conceived and designed the experiments: NA AM JS LD. Performed the experiments: NA SA DR IG. Analyzed the data: NA SA DR SS IG. Contributed reagents/materials/analysis tools: AM. Wrote the paper: NA SS DR LD.
Current address: Dow AgroSciences, Indianapolis, Indiana, United States of America
The authors have declared that no competing interests exist.
Opiates produce significant and persistent changes in synaptic transmission; knowledge of the proteins involved in these changes may help to understand the molecular mechanisms underlying opiate dependence. Using an integrated quantitative proteomics and systems biology approach, we explored changes in the presynaptic protein profile following a paradigm of chronic morphine administration that leads to the development of dependence. For this, we isolated presynaptic fractions from the striata of rats treated with saline or escalating doses of morphine, and analyzed the proteins in these fractions using differential isotopic labeling. We identified 30 proteins that were significantly altered by morphine and integrated them into a protein-protein interaction (PPI) network representing potential morphine-regulated protein complexes. Graph theory-based analysis of this network revealed clusters of densely connected and functionally related morphine-regulated clusters of proteins. One of the clusters contained molecular chaperones thought to be involved in regulation of neurotransmission. Within this cluster, cysteine-string protein (CSP) and the heat shock protein Hsc70 were downregulated by morphine. Interestingly, Hsp90, a heat shock protein that normally interacts with CSP and Hsc70, was upregulated by morphine. Moreover, treatment with the selective Hsp90 inhibitor, geldanamycin, decreased the somatic signs of naloxone-precipitated morphine withdrawal, suggesting that Hsp90 upregulation at the presynapse plays a role in the expression of morphine dependence. Thus, integration of proteomics, network analysis, and behavioral studies has provided a greater understanding of morphine-induced alterations in synaptic composition, and identified a potential novel therapeutic target for opiate dependence.
Repeated exposure to opiates, such as morphine, produces significant and persistent changes in synaptic transmission and plasticity that may contribute to altered behaviors associated with addiction, dependence and withdrawal. While the molecular and cellular mechanisms underlying these long-lasting changes are not fully understood, substantial evidence shows that opiates play a critical role in the modulation of neurotransmitter release, particularly in the mesolimbic dopaminergic system. Chronic morphine exposure increases dopamine signaling in structures of this system
Given the importance of presynaptic neurotransmitter release in drug addiction, we undertook a quantitative subcellular proteomic analysis to investigate the effects of morphine on striatal presynaptic protein levels. Proteomics serves as a powerful tool to reveal changes in protein abundance in response to drug administration
Here we used an integrated proteomics, graph theory-inspired network analysis, and behavioral approach to elucidate the presynaptic molecular events induced by repeated morphine administration. This has enabled a greater understanding of morphine-induced alterations in synaptic composition, and has allowed the identification of potential therapeutic targets for opiate dependence and addiction.
To identify and quantify proteins regulated by morphine, presynaptic (PRE) proteins from saline- and morphine-treated rats were subjected to differential isotopic labeling and LC-MS/MS analysis. Five experiments were performed, using forward (saline = light, morphine = heavy) and reverse (saline = heavy, morphine = light) labeling (
Protein extracts from striatal PRE fractions of saline- and morphine-treated rats were labeled either with succinic anhydride (light) or deuterated succinic anhydride (heavy) and analyzed by LC-MS/MS. (
The levels of NCAM and contactin 1 were decreased in the striatal PRE fraction of morphine-treated animals, but not in the total homogenate. The decrease in presynaptic NCAM was observed for the 180-, 140-, and 120-kDa isoforms of the protein.
Gene Name | Protein Name | UniProt Acc. # |
|
||
|
||
NSF | N-ethylmaleimide sensitive factor | Q9QUL6 |
AP2A2 | AP-2 complex subunit alpha-2 | P18484 |
STXBP1 | Syntaxin-binding protein 1 (Unc-18 homolog) | P61765 |
|
||
GNAL | GTP-binding protein Golf alpha subunit | Q80WZ0 |
GNB1 | Guanine nucleotide-binding protein G(I)/G(S)/G(T) subunit beta 1 | P54311 |
GNB2 | Guanine nucleotide-binding protein G(I)/G(S)/G(T) subunit beta 2 | P54313 |
GNB3 | Guanine nucleotide-binding protein G(I)/G(S)/G(T) subunit beta 3 | P52287 |
YWHAZ | 14-3-3 protein zeta/delta | P63102 |
|
||
SEPT3 | G-septin gamma | Q9R245 |
SEPT7 | Septin 7 | Q9WVC0 |
TUBB2B | Tubulin beta chain 15 | Q3KRE8 |
|
||
CNTN1 | Contactin 1 | Q63198 |
NCAM1 | Neural cell adhesion molecule 1, 140 kDa isoform | P13596 |
OPCML | Opioid-binding protein/cell adhesion molecule precursor (OBCAM) | P32736 |
|
||
HSPA5 | 78 kDa glucose-regulated protein precursor (GRP 78) | P06761 |
HSPA8 | Heat shock cognate 71 kDa protein | P63018 |
TCP1 | T-complex protein 1 subunit alpha | P28480 |
|
||
CS | Citrate synthase, mitochondrial precursor | Q8VHF5 |
DLD | Dihydrolipoyl dehydrogenase | Q6P6R2 |
HK1 | Chain A, Rat Brain Hexokinase Type I Complex With Glucose And Inhibitor Glucose-6-Phosphate | P05708 |
HK2 | Hexokinase 2 | P27881 |
VDAC2 | Voltage-dependent anion-selective channel protein 2 | P81155 |
VDAC3 | Voltage-dependent anion-selective channel protein 3 | Q9R1Z0 |
|
||
|
||
ATP5C1 | ATP synthase gamma chain, mitochondrial | P35435 |
ATP5B | ATP synthase subunit beta, mitochondrial precursor | P10719 |
COX4I1 | Cytochrome c oxidase subunit 4 isoform 1, mitochondrial precursor | P10888 |
COX5A | Cytochrome c oxidase subunit Va | P11240 |
COX5B | Cytochrome c oxidase subunit Vb | P12075 |
|
||
CDC42BPA | Serine/threonine-protein kinase MRCK alpha | O54874 |
|
||
HBB | Beta-globin | Q6PDU6 |
Proteins with morphine/saline ratios of at least 0.5 standard deviations from the mean and that showed consistent changes in at least 2 of the experiments were selected. The numbers in parentheses indicate the number of proteins that were downregulated or upregulated.
To enrich the list and identify a network of proteins downregulated by morphine, we used the Genes2Networks
Proteins from the seed list (yellow) were connected via intermediates from the background dataset. Significant intermediates are shown in red (score>3) or orange (score between 2–3). The two clusters that were used to make predictions of morphine-regulated proteins are outlined.
Since functionally related nodes are likely to interact with each other while being more separate from the rest of the network
Some of the predictions generated by the cluster analysis were verified using Western blot analysis (
Using graph theory-based methods, L1CAM, neurocan (CSPG3), and CSP were predicted to be decreased at the presynapse by morphine treatment. Western blot analysis showed a decrease in (
In the smaller cluster, morphine administration led to a decrease in the levels of L1CAM in the PRE fraction, but not in the total homogenate (
In the largest cluster, we identified CSP as a significant intermediate (with the highest score of 6.45). Western blot analysis showed a significant decrease in the levels of this protein in the PRE fraction, but not in the total homogenate, after chronic morphine administration (
Western blot analysis showed an increase in the levels of Hsp90 in the PRE fraction after chronic morphine administration. No changes were observed in total homogenates suggesting redistribution of the protein rather than increases in gene expression changes. Protein levels were normalized to actin. **p<0.01 compared to saline treated (n = 6). A representative blot of 6 is shown.
To assess the behavioral implications of the observed increase in Hsp90 levels, we examined the effect of Hsp90 inhibition on naloxone-precipitated morphine withdrawal (
(
Withdrawal Signs | Vehicle(n = 12) | GA (5 mg/kg)(n = 4) | GA (20 mg/kg)(n = 7) |
Rearing | 55.33±7.87 | 32.86±13.45 | 22.33±9.38 |
Jumping | 37.18±6.49 | 19.5±14.66 | 2.86±1.72 |
Forepaw Tremors | 54.42±9.43 | 58.00±19.84 | 21.23±7.56 |
Teeth chattering | 5.17±0.45 | 2.75±0.63 |
1.57±0.48 |
Ptosis | 4.42±0.31 | 2.25±0.75 |
3.14±0.51 |
Diarrhea | 2.58±0.52 | 0.25±0.25 |
0.42±0.30 |
*p<0.05,
**p<0.01,
***p<0.001 Vehicle vs GA.
In this study, we undertook a quantitative subcellular proteomic analysis to study the effects of morphine on striatal presynaptic protein levels. We used a five-day paradigm of chronic intermittent escalating morphine administration that results in the development of significant opiate dependence
Our analysis showed morphine-induced downregulation of most of the proteins identified. Some of these are involved in various steps of synaptic vesicle trafficking, including vesicle priming (NSF), docking (syntaxin-binding protein 1 or Unc-18 homolog), and endocytosis (AP-2 alpha2), supporting previous reports of downregulation of SNARE complex formation
Our proteomics data showed morphine-induced decreases in several cytoskeleton-associated and cell adhesion molecules, including tubulin beta chain 15, septin 3 and 7, contactin 1, NCAM1, and opioid-binding cell adhesion molecule (OPCML or OBCAM) (
Similar results were observed for proteins predicted by our network analysis to be modulated by morphine. For instance, a decrease in processed neurocan was mostly seen in the PRE fraction. This, taken with the finding that decreases in L1CAM levels were seen only in PRE fraction and not in homogenate, further supports the novel concept of regulation of protein levels by redistribution of proteins from PRE to extra-synaptic areas (as opposed to changes in gene expression).
Previously, it was shown that long-term exposure to morphine leads to a decrease in the levels of several G protein subunits (αi2, αi3, β1, β2) in human neuroblastoma SH-SY5Y cells stably expressing the μ opioid receptor
Another group of proteins, shown by our proteomics analysis to be downregulated by chronic morphine administration, includes molecular chaperones such as GRP 78, (Hsc70), and TCP1. Hsc70, a constitutively expressed protein, is a member of the 70 kDA heat shock protein family (Hsp70). It is enriched in the mammalian nervous system, particularly at synapses, where it plays a role in the folding of denatured proteins
Having found significant morphine-induced changes in the levels of CSP and Hsc70, we sought to determine whether chronic morphine administration would also affect the levels of Hsp90, which is known to interact with these proteins under normal conditions
To further assess the functional implications of the observed increase in Hsp90 levels, we used a morphine withdrawal paradigm to determine whether this increase plays a role in morphine dependence. Our results showed that inhibition of Hsp90 by geldanamycin dose-dependently decreases somatic signs of morphine withdrawal, suggesting that Hsp90 may play an important role in dependence-associated behaviors and that its inhibition may alleviate symptoms of withdrawal in opiate-dependent subjects. Moreover, Hsp90 inhibitors may represent potential therapeutics to prevent the cellular adaptations to chronic morphine administration, since it was recently shown that inhibition of Hsp90 partially inhibits the adenylyl cyclase superactivation observed after chronic morphine administration
Morphine and other addictive drugs produce significant and persistent adaptations at the synaptic level that may underlie their long-lasting addictive potential
Protocols involving animals were conducted in accordance with the recommendations set forth in the
After chronic morphine treatment, animals were sacrificed by decapitation and the brains rapidly removed. Isolation of a presynaptic (PRE) fraction was performed as previously described
Labeling experiments were performed as described elsewhere
Peptide separation and MS analysis were performed using a capLC™ (Micromass, UK) system coupled to an HCTUltra-PTM Discovery system ion-trap mass spectrometer (Bruker Daltonics, Billerica, MA, USA) equipped with an electrospray ionization source. The sample was injected using a manual injector (Valco Instruments Co, Inc., TX, USA) and loaded onto a trap column (PepMap™, C18, 5 µm, 100 Å, LC Packings) using solvent A and washed for 5 min. The trapped peptides were then eluted in the reverse direction onto a reverse phased capillary column (LC Packings™ 300 µm i.d.×15 cm, C18 PepMap100, 100 Å) using a solvent gradient at a flow rate of 2 µL/min. The solvent gradient was generated using solvent A and solvent B (95% aqueous acetonitrile with 0.1% FA and 0.01% TFA in water). The 70 min gradient run for LC separation included 3 steps: 5–80% solvent B in 15–55 min (linear); 80% solvent B for 55–60 min (isocratic); 80–5% solvent B in 60–65 min (linear). MS data acquisition and the subsequent MS/MS of selected peaks were performed in a data-dependent manner using the Esquire software (Bruker Daltonics). For each MS scan, three peptides were selected to be fragmented, for 300–500 ms, based on their charge (preferably +2) and intensity. Dynamic exclusion of previously fragmented precursor ions was set to 2 spectra for a period of 60 s. The MS and MS/MS scan were performed in the range of m/z 300–1500 and 50–2000 respectively.
The data were processed using the Data analysis software (Bruker Daltonics). MS data obtained between 20 and 45 min of the LC run was searched for compounds using an automated search option. The short-listed compounds with their respective MS/MS scans were directly exported to Biotools software (Bruker) for database searching using an in-house Mascot database search engine (Matrix Science). The mass tolerance of this study was set at 0.1% for MS and 0.5 Da for MS/MS. The search parameters included fixed modifications for cysteine (carbamidomethyl) and variable modifications for methionine (Met-oxidized), succinic anhydride or succinic [2H4] anhydride modified lysine and N-terminal amines. With Mascot, every tandem mass spectrum was assigned a list of matching database peptide sequences accompanied by a score representing the quality of each sequence identification. A Mascot score of 50 is commonly used as a cut-off for 95% confident identification. Only proteins identified with a Mascot score ≥60 for every peptide were considered for further analysis (for further experimental details on these measurements see references
The proteins altered by morphine treatment were placed in the context of signaling pathways and protein complexes using the software tool Genes2Networks (
A binomial proportions test was used to evaluate the significance of interactions between proteins from the background dataset with the seed list. The z-sore (referred to as “score”) for each protein from the background dataset was computed as described previously
10 µg of protein from each fraction were resolved in 7.5% SDS-PAGE and analyzed by Western blots with the following antibodies: CSP (1∶3000, Stressgen, Victoria, BC), α-contactin (1∶1000, gift from J Salzer, NYU), neurocan (1∶2000, gift from R. Margolis, NYU) L1CAM (1∶5,000, gift from D. Felsenfeld, MSSM), NCAM1 (gift from G. Phillips, MSSM), Hsp90 (1∶10,000, Stressgen), Hsp70 (1∶10,000, Stressgen), actin (1∶10,000, Sigma).
C57BL/6 adult male mice (20–25 g) were injected with chronic morphine or saline (i.p.) as described above (n = 4–12 animals per group). Two hours after the last injection, animals were injected with geldanamycin (5 or 20 mg/kg i.p.) or vehicle (20% DMSO in 0.9% saline), followed by naloxone (1 mg/kg s.c.). After naloxone injection, six somatic signs of withdrawal were evaluated for a period of 30 min. Three signs (jumping, rearing and forepaw tremors) were counted and three signs (teeth chattering, ptosis and diarrhea) received a score of 1 for every 5-min interval in which it was present. After behavioral tests were complete, animals were sacrificed, brains were extracted and used to prepare homogenate and PRE fractions as described above. Hsp90 levels in these fractions were determined by ELISA as described previously
Clusters identified in the network of proteins altered by morphine treatment. Clusters were identified and visualized using CFinder, which uses the clique percolation method to identify overlapping clusters. A total of 13 overlapping clusters were identified in the network: 3 clusters with k = 4 and 10 clusters with k = 3.
(TIF)
A flow chart summarizing the process of proteomic data analysis and computational predictions. Simplified schematic of the approaches used to identify morphine-regulated presynaptic proteins by quantitative proteomics, and to map potential presynaptic signaling pathways and protein complexes by graph theory. These were then used to predict novel morphine-regulated proteins. F = forward labeling, R = reverse labeling. The number of proteins quantified is indicated in parentheses.
(TIF)
Each sample represents a pool of 3 striatal PRE fractions from saline- and morphine-treated rats. In total, 175 unique proteins were identified, and 143 of these were quantified. Proteins identified were those with Mascot scores ≥60.
(DOC)
Analysis of MS/MS spectra led to the identification of 175 proteins, 143 of which were quantified by determining the peak intensity of the labeled peptides. Only 30 of these proteins were robustly and consistently altered by morphine treatment.
(DOC)
We thank I. Bushlin for critical reading of the manuscript and helpful suggestions.