Spatial multiomic insights into acute cocaine exposure

Abstract

Recent studies provide compelling evidence that cocaine-induced neurotoxicity begins within hours of a single acute cocaine exposure. Despite this, a comprehensive understanding of the molecular alterations occurring in vivo within the reward system following such an exposure has been lacking. In this study, we developed an analytical workflow that combines desorption electrospray ionization mass spectrometry imaging of metabolites at different temperatures with microscale proteomics of brain regions. We present a multiomic perspective on the molecular consequences of acute cocaine exposure on the principal areas of the reward system and the hippocampus. Our findings include distinct region-specific alterations in the tricarboxylic acid cycle and lipid synthesis within the reward circuitry highlighting a significant energy depletion in mice 24 hours post-cocaine injections. Additionally, we linked widespread reductions in key neurotransmitters across the reward system and calcium-level alterations, to changes in synaptic plasticity and mitochondrial dysfunction. Mitochondrial dysfunction and energy metabolism disruption were evident through imbalances in the mitochondrial adenosine triphosphate production and electron transport chain components, increased susceptibility to oxidative stress, disturbances in mitochondrial transport proteins, and fluctuations in creatine and taurine levels. Among the brain regions within the reward circuitry, the prefrontal cortex exhibited the most pronounced effects. This study provides a holistic overview of the intricate interplay between proteins and metabolites, unveiling molecular mechanisms within the reward circuitry regions affected during the onset of cocaine-induced neurotoxicity.

Introduction

Cocaine is a central nervous system stimulant known for its psychostimulant and euphoric effects. Cocaine use has been shown to lead to neurotoxicity, which is associated with high morbidity and mortality. Cocaine is rapidly metabolized after reaching the bloodstream (half-life of ∼1 h), and produces several metabolites, some of which are toxic (1). Even a single low dose of this substance leads to oxidative stress (2) and causes substantial structural changes in the brain (3).

From a pharmacological point of view, cocaine acts by binding to monoamine transporters due to its structural resemblance to monoamines and blocks presynaptic reuptake of the monoamine neurotransmitters, such as serotonin, dopamine, and norepinephrine. This prolongs synaptic transmission, and the majority of the neurostimulatory effects mediated by cocaine can be attributed specifically to the potentiation of dopaminergic signaling by blocking the dopamine transporter (4–7). This corrupts the normal circuitry of rewarding and adaptive behaviors, which gradually leads to addiction (8). The reward circuitry, which is composed of the several brain regions including the ventral tegmental area (VTA), nucleus accumbens (NAc), and prefrontal cortex (PFC), is primarily impacted by the neuronal adaptations that underlie addiction (9). Further, cocaine can directly permeate the neurons and accumulate inside negatively charged mitochondria, which leads to their dysfunction (9–11). In general, oxidative stress, mitochondrial dysfunction, and cell death are the three major mechanisms attributed to cocaine-induced neurotoxicity (12).

These effects can be detected several minutes to several hours after acute cocaine exposure (13). Findings of elevated oxidative stress markers, with a decrease in the total antioxidant capacity in rat PFC and NAc in less than 1 hour following cocaine administration have been reported (2). In addition, cocaine also upregulates extracellular glutamate concentrations in the reward system (14), which is associated with the increased activation of N-methyl-D-aspartate (NMDA) glutamate receptors, and subsequent increase in intracellular $Ca2+$ concentration. In fact, even a single dose of cocaine leads to long-term potentiation of AMPA (alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid) glutamate receptors, which may result in drug addiction (15). Furthermore, the increase of $Ca2+$-concentration inside cells caused by cocaine also generates reactive nitrogen species and hydroxyl radicals (16).

In terms of mitochondrial impact, a single-dose administration of cocaine has been found to affect mitochondrial gene transcription in the frontal cortex of rats 15 hours post-injection (17). Intracellular cocaine directly mediates toxicity on a properly functioning electron transport chain (ETC) and impairs oxidative phosphorylation. This leads to a reduction in the production of adenosine triphosphate (ATP) and an increase in the division of mitochondria (a process known as fission) (9). In addition, in vitro exposure of mouse primary microglial cells to cocaine resulted in the loss of mitochondrial membrane electrical potential 6 hours post-exposure (18). Hence, oxidative stress and the hallmarks of mitochondrial dysfunction were observed within 24 hours after cocaine administration.

It is noteworthy that mass spectrometry, and specifically mass spectrometry imaging, has been increasingly employed to study molecular changes in cocaine abuse (19–22). Most of these studies, however, focus on detecting cocaine in tissues and fingerprints for analytical and forensic purposes. Nevertheless, several recent reports have characterized spatial molecular changes in the brain following cocaine exposure (23–25). For instance, hippocampal metabolite modifications were imaged using matrix-assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) 1 hour after cocaine administration and showed upregulation of taurine, adenosine monophosphate and phosphatidylinositol PI (38:4) (23). Dramatic changes in phospholipid and lipid profiles were detected by time-of-flight secondary ion mass spectrometry imaging in drosophila fly brain orally fed with cocaine (25). Lipid remodeling were reported in rodents brain using shotgun lipidomics and imaged in rat hippocampii following repeated cocaine exposure using MALDI-MSI (24).

In this study, we pursued the idea of systematically characterizing early molecular changes, which are specific to the reward system and occur within the first 24 hours following a single cocaine exposure in mice. Acute cocaine exposure typically involves a single administration of 15–20 mg/kg, which has been shown to elicit significant behavioral and neurochemical changes (17). These changes are most likely to take place in response to energy depletion that follows the initially increased energy demand upon cocaine exposure, oxidative stress and early mitochondrial damage. Genomic and transcriptomic analysis of the brain, and especially central nervous system-derived cell lines, after cocaine administration are relatively well documented (26, 27). However, these studies have often failed to reflect the minuscule molecular modifications that occur within the reward system in vivo following acute cocaine exposure. Here we developed a new analytical workflow that combines mass spectrometry imaging with 50 μm spatial resolution of the brain regions with region-specific microscale proteomics. In our study, we utilized desorption electrospray ionization mass spectrometry imaging (DESI-MSI), an ambient imaging technique that in contrast to MALDI-MSI does not require sample preparation with matrix coating. We also utilized DESI-MSI at different temperatures, enhancing the detection of various metabolites and lipids. This workflow allowed us to detect minuscule molecular alterations between four brain regions (three of them are part of the reward circuitry): NAc, PFC, VTA, and hippocampus (Hip) and compare them between each other and to the rest of the brain in a single run. We uniquely complemented DESI-MSI analysis with proteomics data of the same brain regions for the first time. We used microscale proteomic analysis allowing sub-100 μg brain regions to be examined, which is below the typical tissue sample amount for proteomics (28, 29). Our results provide a thorough insight into the regulation pathways of metabolites, lipids, and neurotransmitters within various brain regions during the initiation of neurotoxicity, induced by acute cocaine exposure.

Results and discussion

Workflow overview

To study the effects of cocaine in the mice brain, we developed a workflow that combines DESI-MSI at various temperatures with region-specific microscale proteomics (Figure 1). This multimodal analysis aimed to acquire complementary data of proteomics, metabolomics and lipidomics to spatially correlate changes in the brain. We have reached not only coverage of a broad spectrum of molecules but also worked at a detailed spatial resolution, which allowed the extraction of chemical information from small brain regions. This analysis was followed by proteomic interrogation of the same regions. Figure 2 summarizes this combination of metabolic and proteomic workflows that was applied to study the brains of mice treated with either cocaine or saline and investigated 24 hours post-injection. To address the possible sampling heterogeneity and to allow maximum data overlay, we tightly controlled the thickness and the location of brain tissue extracted for metabolomics and the size and location of biopsy for proteomics.

The proteomic data obtained by liquid chromatography–mass spectrometry (LC–MS) was assessed with GO:enrichment analysis and also parsed through the Reactome database with gene set enrichment analysis using Pathway Analysis with Down-weighting of Overlapping Genes (PADOG). As a result, we revealed numerous regulations in protein composition per region, as shown in Figure 2A and B. Proteins with post-translational modifications (including phosphorylation and ubiquitination) and significantly altered abundance, were also interrogated across the brain regions (see Supplementary material file Limma_output_Peptides [XLSX]).

The majority of the regulated proteins were associated with neurotransmitter transport and mitochondrial metabolism (Figure 2C). Pathway analysis of the proteomic data with the Reactome showed a general trend for downregulation of pathways in the PFC, with slight downregulation trends in both the VTA and NAc (Figures S1–S3). Interestingly, the Hip displayed different regulatory patterns compared to those observed in the reward system regions.

Below, we present a detailed account of the molecular alterations discovered by this novel workflow 24 hours following the acute cocaine treatment. Figure 3 demonstrates the main molecular changes we have uncovered.

Tricarboxylic acid cycle disruption

Exposure to cocaine is known to inhibit glycolysis and impede the tricarboxylic acid (TCA) cycle (9), which is a closed-loop metabolic chain of reactions taking place in mitochondria (Figure 3A). Decreased TCA activity may lead to fewer electron carriers, such as nicotinamide adenine dinucleotide-hydrogen (NADH) and $FADH2$⁠, energy imbalances, and elevation of reactive oxygen species (ROS) production.

In our in-depth molecular view into the TCA cycle of the reward circuitry regions (Figure 4, Table S1), we observed that aconitate (the conversion intermediate of citrate to isocitrate) is differentially regulated in the reward system regions. In particular, it is significantly upregulated in the PFC (⁠$logFC=5.702$⁠, $P=0.017$⁠), slightly elevated in NAc and suppressed in the VTA of cocaine vs saline-treated mice (Figure 3A, Table S1).

Further, α-ketoglutarate is prominently depleted in the NAc (⁠$logFC=−1.606$⁠, $P=0.008$⁠), VTA (⁠$logFC=−1.508$⁠, $P=0.0015$⁠), and reduced in the PFC. Similar trends of depletion are observed with succinate. Oxaloacetate, the intermediate that completes the cycle, exhibits a tendency for depletion in PFC and NAc but elevation in the VTA. This indicates that for the PFC and the NAc regions, the TCA cycle is impaired somewhere in the steps of transition of aconitate into α-ketoglutarate, while for the VTA, the impairment occurs in the transition of oxaloacetate to citrate and aconitate. Conversely, lactate shows a tendency towards upregulation in all regions of reward circuitry (PFC, NAc, and VTA).

A closer look into region-specific microscale proteomics data reveals an upregulation in mitochondrial isocitrate dehydrogenase (IDH2), particularly in the PFC (⁠$logFC=0.59$⁠, $P=0.016$⁠). This enzyme is responsible for the conversion of isocitrate to α-ketoglutarate through the production of NAD(P)H from NAD(P)⁺, indicating that (1) the α-ketoglutarate pool is initially increased in the PFC but is likely to be competitively consumed by other pathways; and (2) the rising ratio of NAD(P)H to NAD(P)⁺ favors the transamination of α-ketoglutarate to excitatory transmitter glutamate (30), which can be further transformed to glutamine and to the inhibitory transmitter GABA. Our MSI data show that the levels of GABA and glutamine are depleted in mice 24 h post-cocaine injection throughout the entire reward system. (For GABA $logFC=−3.108$⁠, $P=0.008$ in NAc, see Table S1). These findings suggest a possible consumption of these molecules by competing pathways.

Thus, glutamate can also be converted to excitatory neurotransmitter aspartate, and further to the urea cycle intermediate, ornithine. While aspartate is downregulated across all studied brain regions with statistically significant changes in the Hip (⁠$logFC=−0.837$⁠, $P=0.084$⁠) and the VTA (⁠$logFC=−1.534$⁠, $P<0.001$⁠), ornithine exhibit a trend towards upregulation, particularly in the PFC, as shown in Figure 3A. Ornithine is further transformed to spermidine, an aliphatic polyamine, responsible for maintaining membrane potential and inhibiting neuronal nitric oxide synthase (31). Spermidine also shows a tendency towards upregulation in both the NAc and the VTA. Hence, our spatial metabolomics data provide some clues for the consumption of the neurotransmitters by the alternative pathways, such as their utilization in the urea cycle.

In the Hip, different metabolite and protein alterations were observed, which sets it apart from the regions of the reward system. Trending elevation of citrate (or isocitrate) and aconitate with downregulation of α-ketoglutarate and trends of upregulation of glutamate were detected in this region, indicating that the TCA cycle may be impaired by consuming α-ketoglutarate toward glutamate synthesis. The proteomics data show glutamine synthetase being significantly downregulated in this region (⁠$logFC=−0.978$⁠, $P=0.019$⁠), further supporting the trending elevation of glutamate. Upregulation of calcium-binding mitochondrial carrier protein Aralar2 (⁠$logFC=0.443$⁠, $P=0.043$⁠) in combination with downregulation of aspartate (⁠$logFC=−0.837$⁠, $P=0.084$⁠) were also observed in this region, which is likely to affect respiration and increase glycolysis (32). Aralar2 is a malate/aspartate shuttle (MAS) component and is activated as a response to higher levels of calcium in the cytosol (32). On the other hand, trends of glutamate upregulation detected solely in the Hip may suppress glycolysis and alter mitochondrial oxidative metabolism (33). Further, the upregulation of the 2-oxoglutarate/malate carrier protein (SLC25A11) (⁠$logFC=0.885$⁠, $P=0.012$⁠) in the Hip decreases the proton gradient at the inner mitochondrial membrane and disturbs energy metabolism (34). These findings, in combination with the upregulation of Slc-transporter (SLC6A11) (⁠$logFC=1.10$⁠, $P=0.022$⁠) and the downregulation of mitochondrial transport protein (mitochondrial import inner membrane translocase subunit Tim13 [TIM13]) in the Hip (⁠$logFC=−2.03$⁠, $P=0.0002$⁠), indicate a significant dysregulation of the molecular transport through inner mitochondria membrane in the Hip along with the TCA cycle impairment.

Effects on fatty acid and lipid synthesis

One of the possible reasons for TCA cycle impairment could be the conversion of citrate to acetyl-CoA and its transport outside the mitochondria in cocaine-treated mice. In the cytosol, acetyl-CoA serves as a precursor for fatty acid synthesis. Accordingly, by spatially resolved fatty acid analysis we observe an upregulation of arachidonic acid in all reward system regions (Figure 5, and Figure S1). Arachidonic acid is the intermediate of the endocannabinoid system involved in the modulation of dopamine neurotransmission (35). Importantly, microscale proteomics independently indicates that fatty acid binding protein (FABP7) is upregulated in the PFC, while mitochondrial trifunctional enzyme subunit alpha (Hadha), which is responsible for breaking down fatty acids into acetyl-CoA, shows an increase in both PFC and Hip in cocaine-treated mice (Supplementary material file Limma_output_Proteins [XLSX]).

Other molecules, which are strongly and consistently upregulated throughout the reward circuitry of cocaine-treated mice, are phosphatidylethanolamines (Figure 5, Figure S4). The predominant two pathways for PE biosynthesis are (1) the decarboxylase (PSD) pathway that takes place in mitochondria and (2) the cytidine diphosphate (CDP)-ethanolamine (Kennedy) pathway that occurs in the endoplasmic reticulum. The PSD pathway begins with the transport of phosphatidylserine (PS) into the mitochondria, where it is subsequently decarboxylated to form PE (36), whereas the CDP-ethanolamine pathway converts exogenous ethanolamine to PE and is often referred to as the de novo pathway of PE synthesis (36, 37). The other two ER pathways (acylation of lyso-PE and head group base exchange) weakly contribute to the cellular pool of PE (38). While the CDP-ethanolamine pathway produces PE species containing mono- or di-unsaturated fatty acids, the mitochondrial PSD pathway generates PE species enriched with polyunsaturated fatty acids (38).

In the regions of the reward system, and especially in PFC, we observe a strong elevation of PE species containing one or more polyunsaturated fatty acids. For detecting structurally similar species with multiple polyunsaturations in fatty acid chains, mathematical processing was performed for each m/z peak identified as a PE to recognize a possible gain of an even number of protons (Supplementary material Lipid_correlation [XLSX]). For the majority of recognized PE species, multiple hits were detected revealing numerous polyunsaturated PE species upregulated in the regions of the reward system after cocaine treatment (Figure S4, and Supplementary Material Lipid_correlation [XLSX]). Microscale proteomics indicated downregulation of PE synthesis pathway in the regions of the reward system, with the major downregulation in PFC (see Figure S1, Metabolism). It is reasonable to assume that these species are rapidly produced in the mitochondria as the result of cocaine-related damage. It was previously shown that PE species play a pivotal role in mitophagy (39) and inhibit mitochondrial respiratory activity in a dose-dependent manner in vitro (40). There are numerous possible fates of elevated PE species in the cell. They can be integrated into mitochondrial membranes, exported to other cellular compartments, used as precursors for the production of phosphatidylcholines (PC), or utilized as substrates for basic post-translational modifications such as glycosylphosphatidylinositol (GPI) anchors (38, 41, 42).

Neurotransmitter regulation and calcium homeostasis

Combined analysis of proteins and metabolites in the reward circuitry post-cocaine injection revealed significant changes in neurotransmitter regulation and associated pathways critical for synaptic plasticity. Specifically, the excitatory neurotransmitter, glutamate, and inhibitory neurotransmitter, GABA were downregulated in all regions of the reward system (Figure 3A). Related processes, including the activation of NMDA and AMPA ionotropic glutamate receptors and intracellular calcium uptake, were affected in all regions of the reward circuitry (Figure 3B, Figures S1–S3). These processes can also reflect molecular changes of long-term potentiation (LTP) of excitatory synapses that leads to synaptic plasticity (15, 43, 44). It was previously shown that a single cocaine exposure induces LTP in the VTA measured 24 hours later (15, 45, 46). The decreased levels of glutamate and its metabolic consumption found in this study and discussed above (section 2.1) may partially explain the increase in the activation of glutamate receptors. GABA-level downregulation is likely to be compensatory for glutamate decrease. Downregulations in GABA-evoked currents and GABA levels were previously reported in acute and chronic cocaine consumption (47, 48).

These changes not only show the profound impact of cocaine on general neurotransmitter homeostasis but also underpin underlying conditions for these alterations: within each region, we observed an abundance of changes in proteins directly related to synaptic plasticity and neuronal processes. Downregulated NrCam observed in all regions of the reward system has implications on glutamatergic molecules and matrix adhesion, impacting synaptic stability and communication (Supplementary material file Limma_output_Proteins(XLSX)). Previous studies have shown that low expression of NrCam and glutamate downregulation coincide with the development of drug addiction (49, 50). This downregulation may be substantiated by the suppressed NFκB signaling, as indicated by our Reactome pathway analysis. Furthermore, in all regions of the reward circuitry, a downregulation of the GluR2-containing AMPA receptor trafficking pathway is detected, which results in an increase in calcium levels, presumably via an increase in GluR2-lacking AMPA receptors (51), underscoring the potential interplay between neurotransmission and calcium dysregulation (52, 53). G-protein signaling regulation of DARPP-32 (cAMP-regulated phosphoprotein 32kDa) is another hallmark of cocaine neurotoxicity, which is common for all reward system regions. In Figure 3B, we show that DARPP-32-mediated events are dependent on cAMP-dependent protein kinase (PKA) activity and calcium levels.

In the PFC, there is a slight increase in gamma-aminobutyric acid receptor subunit alpha-2 (GABRA2), which may be implicated in addiction and directly related to the development of behavioral sensitization. Also in the PFC, marked downregulations of A-kinase anchor protein 5 (AKAP5) and Homer protein homolog 1 (HOMER1) are observed (Supplementary material file Limma_output_Proteins [XLSX]). Both proteins play critical roles in LTP and the regulation of protein trafficking in excitatory glutamatergic synapses. Further in the PFC, there is an increased expression of EF-hand domain-containing protein D1 (EFHD1) and a decrease in plasma membrane calcium-transporting ATPase 4 (AT2B4). EFHD1 is linked to calcium ion binding, while AT2B4 is vital for calcium extrusion from the cell. These changes result in shifts in calcium-dependent neuronal signaling and in increased intracellular calcium concentrations, which might affect neurotransmission and neuron excitability. Decreased levels of complexin-3 (CPLX3), a modulator of neurotransmitter release, further support the changes in synaptic vesicle exocytosis in the PFC.

In the NAc, the downregulation of AT2B4 implies disturbances in calcium homeostasis, potentially resulting in an elevated intracellular calcium concentration. This may be countered by the upregulation of calcium/calmodulin-dependent 3’,5’-cyclic nucleotide phosphodiesterase 1B (PDE1B), which modulates cAMP and cGMP levels, thereby influencing neuronal signaling pathways. Additionally, the sodium-driven chloride bicarbonate exchanger in the NAc shows a significant upregulation. As a bicarbonate cotransporter, it plays a pivotal role in regulating intracellular pH, impacting neuronal excitability, signaling, and amino acid transport, specifically for proline, glycine, leucine, and alanine. This uptick suggests modifications in ionic equilibrium that might affect neurotransmission, especially considering the diminished levels of glutamate and GABA in this region (Figure 4, Table S1). The upregulation of leucine observed by spatial metabolomics in the NAc among other regions of the reward system (Figure 4, Table S1), could be tied to these transporter shifts, highlighting potential disruptions in the amino acid balance within this brain area. Furthermore, the decreased phosphorylation of liprin α3 in the NAc could impact the organization and function of the presynaptic active zone, potentially altering neurotransmitter release (Supplementary material file Limma_output_Peptides [XLSX]).

In the Hip, upregulation of proteins such as vesicle-associated membrane protein 3 (VAMP3), synaptic vesicle glycoprotein 2A (SV2A), solute carrier family 6 member 11 (S6A11), and neuronal calcium sensor 1 (NCS1) increases the excitatory neurotransmission (Supplementary material file Limma_output_Proteins [XLSX]). This is further corroborated by the increased glutamate levels and reduced GABA in this region, which coincide with the downregulation of calcium-dependent secretion activator 2 (CAPS2) (54).

Regarding calcium regulation in the Hip, the upregulation of calcium-binding proteins hippocalcin (HPCA) and neurocalcin delta (NCALD) suggests enhanced calcium levels and $Ca2+$-associated neurotransmitter regulation in the Hip.

Mitochondrial function

Our spatial metabolomic analysis revealed that creatine, an important metabolite associated with mitochondrial function and energy regulation, was downregulated in all regions of the reward circuitry (Figures 3A and 4, Table S1). Creatine is partially metabolized in mitochondria by mitochondrial creatine kinase C and acts as an energy buffer to maintain cellular homeostasis. Its downregulation is one of the hallmarks of mitochondrial dysfunction (55). Another important metabolite for mitochondrial function, taurine, was upregulated throughout the reward system and downregulated in the Hip (Figures 3A and 4, Table S1). Upregulation of taurine levels is a cellular response to neurotoxicity damage, which aims to increase ATP levels, prevent $Ca2+$ -induced mitochondrial permeabilization, it suppresses the damage from pathological mitochondrial ROS, and act as an alkaline pH buffer for mitochondrial matrix enzymes (56–58).

Further, we see distinct mitochondrial function-related changes in different brain regions as a response to cocaine exposure within the region-specific-microscale proteomics data (Supplementary material file Limma_output_Proteins [XLSX]). For example, cytochrome B-c1 complex subunit 6 (UQCRH), the essential component of the mitochondrial electron transport chain responsible for oxidative phosphorylation, shows an upregulation in both the PFC and the NAc. This upregulation leads to an increase in the rate of electron transfer from ubiquinol to cytochrome C, and results in enhanced proton gradient, which promotes ATP synthesis through oxidative phosphorylation. However, its isoform, COX6B1, is upregulated and ADP/ATP translocase 4 (SLC25A31) is downregulated only in the PFC, which could be the result of disruptions in oxidative phosphorylation and ATP synthesis in this region. We also find that in the PFC there is evidence for mitochondrial function disruption as attested by changes in the levels of proteins, such as methylmalonate-semialdehyde dehydrogenase (ALDH6A1). ALDH6A1 is involved in amino acid metabolism and acyl-CoA binding and is overexpressed in the PFC in cocaine-treated animals. Further, mitochondrial thioredoxin-dependent peroxide reductase (PRDX3) and catalase (CAT) are upregulated in the PFC, while oxidation resistance protein 1 (OXR1) is downregulated. This suggests an increased susceptibility of the PFC to oxidative neuronal damage.

In the Hip, mitochondrial ATP synthase subunit B1 (ATP5PB) is upregulated. ATP5PB produces ATP from ADP using the proton gradient from the respiratory chain, which is also found to be an affected pathway by GO enrichment analysis (Figure 2D). This may be a response to higher energy needs. Moreover, electron transfer flavoprotein dehydrogenase and succinate dehydrogenase complex iron sulfur subunit B, are found to be downregulated in the Hip. These alterations align well with the TCA cycle depletion discussed above. Similarly, the mitochondrial enzyme superoxide dismutase [Mn], responsible for combating oxidative stress and prohibitin-2 , regulating cytochrome C oxidase assembly and mitochondrial respiration, exhibits upregulation specifically in the hippocampal region. Prohibitin also stabilizes the dynamin-like GTPase OPA1 (optic atrophy 1), which mediates mitochondrial inner membrane fusion (59), facilitating ATP production and allowing dilution of ROS.

In the NAc, a notable upregulation of pyridoxine-5’-phosphate oxidase (PNPO), which catalyzes oxidation to pyridoxal 5’-phosphate (PLP), is observed. This suggests a region-specific response to $Ca2+$-level increase (60). PNPO is involved in vitamin B6 metabolism (also an upregulated pathway in Reactome analysis).

In the VTA, the downregulation of the mitochondrial protein PDHB, a crucial component of the pyruvate dehydrogenase complex, is observed and suggests a reduced conversion of pyruvate to acetyl-CoA, which can explain the unique impairment of the TCA cycle in this region that occurs in the transition of oxaloacetate to citrate and aconitate.

Hence, we suggest that even acute cocaine exposure distinctly modulates the oxidative phosphorylation pathway in these brain regions and can lead to alterations in ATP production. Eventually, these disruptions may have an impact on the respiratory ETC and create a chemical gradient across the inner membrane that drives transmembrane transport.

Conclusions

We have presented a data-driven molecular assessment of acute cocaine exposure on the interplay between metabolites and proteins in the brain using an integrated analytical workflow. This multiomic approach combines untargeted metabolic brain imaging by DESI-MSI with region-specific microscale proteomics, which enables cross-validation of results and uncovering physiological pathways involved. The limitations of our study were using rodent models and inherent heterogeneity of measured samples. Yet, the results unveiled a broad molecular landscape of region-dependent physiological processes related to neurotoxicity and mitochondrial dysfunction, which are potentially relevant to human physiology and provide a new angle on addiction development. Understanding these mechanisms is crucial for developing targeted strategies to mitigate the detrimental effects of cocaine on brain function and metabolism. We believe that the outcomes of this study are foundational, establishing a baseline for future addiction studies utilizing a multi-omics approach.

Materials and methods

Chemicals and reagents

For lysis buffer preparation, 20 mM Tris and 137 mM NaCl were purchased from Thermo Fisher Scientific (Waltham, MA), 20 mM HEPES was obtained from Sigma Aldrich Chemie (Steinheim, Germany), and 6 M urea was acquired from Acros Organics (Geel, Belgium). A concentration of $1μg/mL$ RapiGest SF Surfactant from Waters Corporation (Milford, MA) was also used.

For peptide reconstitution and DESI-MSI, a 3% acetonitrile and 0.1% formic acid solution was prepared in-house. The 95:5 $MeOH: H2O$ solution with $500pg/μL$ Leucine Enkephalin was also formulated in-house. Commercial HeLa digest used for system validation was purchased from Thermo Fisher Scientific (Waltham, MA).

For LC–MS analysis, ultra performance liquid chromatography (UPLC)–MS grade water, acetonitrile, and formic acid were purchased from Fisher Scientific (Geel, Belgium). Mobile phases A and B were prepared using 0.1% formic acid in water and 3% dimethyl sulfoxide in acetonitrile, both obtained from Thermo Fisher Scientific (Waltham, MA).

Animal license and handling

All experimental protocols involving animal subjects were conducted in strict accordance with ethical standards approved by the Institutional Animal Care and Use Committee of the Hebrew University of Jerusalem. Mice (C57BL/6) aged 5 to 6 weeks were administered acute intraperitoneal injections of cocaine (or saline, as a control) at a dosage of 20 mg/kg and saline 1 ml/kg, and were subsequently euthanized 24 hours post-injection. For the purposes of MSI, brain tissues were promptly excised and immediately frozen in isopentane cooled with dry ice, then stored at a temperature of $−80∘$C. For proteomic analyses, biopsies were obtained from 100 μm slices of brain tissue on ice, specifically targeting four regions: PFC, VTA, NAc, and Hip. These biopsy samples were immediately flash-frozen in liquid nitrogen and preserved at $−80∘$C until further processing. The time between mouse sacrifice and freezing of the samples was no longer than 90 seconds for mass spectrometry imaging and 5 minutes for proteomics samples.

Proteomic analysis

Lysis and digestion

Biopsy tissue samples were homogenized using a pestle, then homogenates resuspended in a lysis buffer comprised of 20 mM Tris, 137 mM NaCl, 20 mM HEPES, 6 M urea, and 1 μg/mL RapiGest SF Surfactant. Ultrasonic lysis was executed utilizing a Covaris M220 Focused-ultrasonicator in microTUBE-50 (55 μL) tubes, under the following settings—70 Peak Intensity Power/20 Duty Factor/200 Cycles per Burst/5 min at $4∘$C. This was followed by centrifugation at 14,000 RPM for 90 minutes to eliminate cellular debris. The final protein concentration was quantified with NanoDrop. Peptide digestion was carried out with Filter Aided Sample Preparation (FASP) protocol, as detailed in (28, 29, 61, 62). Peptide samples were subsequently reconstituted in a solution of 3% acetonitrile and 0.1% formic acid, achieving a final protein concentration of 150 ng/μL.

Liquid chromatography–mass spectrometry

Post-digestion, tryptic peptides were subjected to analysis using a Synapt G2-Si HDMS mass spectrometer connected to a nanoAcquity UPLC system featuring a nanoelectrospray ionization source (Waters Corporation, UK). The UPLC configuration involved a C18 trap column (⁠$5μm,180μm×20mm$⁠) and an HSS-T3 C18 analytical column (⁠$1.8μm,75μm×250mm$⁠), set in trapping mode. Each injection contained 300 ng of the sample and was subjected to a 120-minute gradient from $3%$ to $40%$ of mobile phase B, maintained at a constant flow rate of $0.3μL/min$⁠. Mobile phases A and B comprised $0.1%$ formic acid and $3%$ dimethyl sulfoxide in water and acetonitrile, respectively. Lock-mass correction was performed using intermittent 60-second sprays of a reference solution containing [Glu1]-fibrinopeptide B (⁠$0.1μM$⁠) and leu-enkephalin (⁠$1μM$⁠). Data was acquired through a data-independent acquisition (DIA) workflow in positive ion mode using a $UDMSE$ methodology. The system’s performance was validated by injecting a commercial HeLa digest (Thermo Scientific, MA) after every ninth sample (28, 29, 61, 62).

Protein identification and label-free quantification

Data processing employed ProteinLynx Global Server (PLGS) 3.0.3 with the SWISSPROT Mouse database, using predefined analytical settings: an FDR of 0.01, Trypsin digestion, and allowance for a single missed peptide cleavage. The minimum ion matches were set at 1 for peptides and 3 for proteins, and at least two peptide matches per protein were required. Fixed and variable modifications were specified, including carbamidomethyl cysteine and variable modifications such as lysine acetylation (28, 61). Phosphorylation and ubiquitination were considered in additional data search. Workflow details are in Tables S2 and S3. ISOQuant 1.8 was used for relative protein quantification via the TOP3 method and subsequent analyses. Settings for these analyses are detailed in the Supplementary material ISOQuant report files provided in Supplementary material Protein_ISOQuant (XLSX) and Peptide_ISOQuant (XLSX).

Mass spectrometry imaging

Sample preparation for DESI-MSI

Brain tissues were sectioned using a Leica CM1950 cryostat with chamber temperature set to $−$20^∘C. Sections with a thickness of 20 μm were prepared at two different bregma levels—1.18 mm and $−$2.80 mm incorporating regions of interest PFC and NAc, and Hip and VTA and subsequently collected on glass slides (SuperFrost Plus, Thermo Scientific).

Desorption electrospray ionization mass spectrometry imaging

DESI-MSI analyses were carried out on a Xevo G2-XS QTof mass spectrometer equipped with a DESI-XS ion source (Waters, Wilmslow, UK). The spectrometer was operated in sensitivity mode with a capillary voltage of 0.7 kV for both positive and negative ion modes, and a cone voltage set at 40 V. MSI was performed with a heated ion transfer tube to enhance the signal from metabolites and fatty acids at 60^∘C in positive and 100^∘C in negative ionization mode, and at 450^∘C in the negative mode for the identification of lipids. Mass-to-charge (⁠$m/z$⁠) acquisition ranges were set as follows: 50–600 $m/z$ for metabolites and 500–1,600 $m/z$ for lipids. Gas pressure was held at approximately 7 psi (48 kPa). The solvent system was 95:5 $MeOH: H2O$ with 500 pg/μL Leucine Enkephalin for lock mass, at a flow rate of 2 μL/min. The spatial resolution was 50 μm. Optical images of H&E stained tissue sections were obtained with brightfield microscopy on Zeiss SlideScanner Axioscan 7.

Imaging data analysis

Raw data files were uploaded to Lipostar MSI 1.3 software for peak picking, using a set of predefined parameters: Savitzky-Golay smoothing and baseline correction were activated, with parameters such as window size, degree, and iterations set to 7, 2, and 1 respectively. Minimum signal-to-noise ratio, noise window size were set to 1.0, and resolving power was 20,000 at 200 $m/z$⁠. The lock mass calibration was set to 554.2615 for negative mode (LeuEnk-H) and 556.2765 (LeuEnk+H) for positive mode with minimum intensity—100 units. Anatomical regions of interest (ROIs) were identified based on H&E staining. All resulting images were root mean square (RMS) normalized and displayed using the viridis color scale. The RMS normalized intensities for ROIs were then exported to a CSV file as shown in the SI file (see Supplementary material Limma_input_metabolites [XLSX]) for subsequent statistical analysis. Lipid desaturation analysis was performed using mass lists extracted from Lipostar MSI version 1.3 following data acquisition at 450^∘C in negative mode with DESI MSI. For each m/z peak identified, we checked the potential increase in mass corresponding to the gain of an even number of protons (⁠$i=2,4,…,100$⁠). This entailed searching for an addition of mass/charge increments of $i×X×1.0072766u$ within a mass tolerance of 0.02 ppm. Peaks that formed possible pairs were subsequently assessed for spatial correlation, applying a threshold of greater than 0.65 for significance. To assist with interpretation, additional data were compiled, such as the number of protons added (i), the m/z of the protonated species (iX), the m/z for ¹³C isotopes (iX(C13)), and the mean intensities of the paired peaks, see Supplementary material Lipid_correlation (XLSX).

Metabolite identification confirmation with LC–MS

Identification of detected metabolites was performed against metabolic standards in in-house MS library by LC–MS using a Dionex Ultimate 3,000 high-performance liquid chromatography (UPLC) system coupled to an Orbitrap Q-Exactive Plus Mass Spectrometer (Thermo Fisher Scientific) with a resolution of 70,000 at 200 mass/charge ratio (⁠$m/z$⁠), electrospray ionization in the HESI source, and polarity switching mode to enable both positive and negative ions across a mass range of 70 to 1,000 $m/z$⁠, was used. The UPLC setup included a ZIC-pHILIC column (SeQuant; $150mm×2.1mm$⁠, 5 μm; Merck) with a Sure-Guard filter (SS frit 0.5 μm). Five μL of brain region extracts in methanol were injected and the compounds were separated with a mobile phase gradient of 15 min, starting at 20% aqueous (20 mM ammonium carbonate adjusted to pH 9.2 with 0.1% of 25% ammonium hydroxide) and 80% organic (acetonitrile) and terminated with 20% acetonitrile. The flow rate and column temperature were maintained at 0.2 mL/min and 45 ^∘C, respectively, for a total run time of 26 min. Fragmentation patterns in MS/MS of the brain extracts were compared with the fragmentation patterns of analytical standards to infer metabolite identities, subsequently used in the interpretation of DESI-MSI data. All metabolites were detected using mass accuracy below 5 ppm. Thermo Xcalibur was used for the data acquisition.

Lipid structure identification with MS/MS

Phospholipid ions identities were verified by tandem MS in the negative ion mode using direct injection of brain region extracts in methanol by electrospray ionization in the HESI source into the Orbitrap Q-Exactive Plus Mass Spectrometer (Thermo Fisher Scientific) at a higher energy collisional dissociation (HCD) mode. Selected ion monitoring (SIM) analysis was chosen and allowed for optimization of signal intensity for each peak of interest.

Statistics and reproducibility

Data from ISOQuant and exported ROI intensities were subjected to further statistical scrutiny using R version 3.6.3. Differential protein levels were evaluated using the empirical Bayes approach via the limma package (28, 61–63). To control for false discovery rate (FDR), the q-value package was used, setting an FDR threshold of less than 0.05. Only log2-fold changes exhibiting an adjusted P-value below 0.05 were deemed statistically significant. Proteomics investigations was conducted using both biological and technical replicates. Metabolomics analysis was done on three biological replicates.

Databases used

The databases, including UniProtKB and SWISSPROT, were employed for protein identification and other related analyses. The label-free quantitation data were used for multidata set PADOG in the Reactome GSA software. Altered protein levels were processed for GO enrichment analysis with Panther 15.0 using all identified proteins as reference. The multiple data sets in this case were replicates of the results obtained for each condition in relation to the results obtained for control samples (saline treatment). Metascape was used to generate the pathways overview.

Database resources

For protein identification and protein function mapping, UniProtKB database was utilized (64). Label-free quantitative data were also analyzed with PADOG using the Reactome GSA software (65). The Panther 15.0 tool was used for Gene Ontology (GO) enrichment analysis. Pathway overviews were generated using Metascape (66). The data sets for these analyses were replicates for each condition.

Supplementary Material

Supplementary material is available at PNAS Nexus online.

Funding

The work in this paper was supported by research grants from the Israel Science Foundation (1840/20, K.M.), Ministry of Science and Technology of Israel (1001578342, K.M.), United States-Israel Binational Science Foundation (2019237, K.M.), Israel Cancer Research Fund (20-204-RCDA, K.M.), National Institute of Psychobiology in Israel (K.M.), The Israel Science Foundation (R.Y., 1283/16), and by ICARE (IMRIC center for addiction research, R.Y. and K.M.). Swedish Research Council (2018-03988, E.T.J.; 2021-03293 and 2022-04198, P.E.A.), Åke Wiberg Foundation (E.T.J.), Carl Trygger Foundation for Scientific Research (E.T.J.). This research was conducted using the Spatial Mass Spectrometry unit supported by the Science for Life Laboratory (P.E.A.) and the Spatial Metabolomics Unit founded with the help of the Wolfson Foundation and the Wolfson Foundation Charity Trust (PR/oys/jw/md/eh/22747/22641, K.M.) and the Metabolomics Core Unit, Shared Research Facility, Faculty of Medicine-Hebrew University (K.M., O.S.). The research exchange of Mariya Nezhyva was made possible owing to the Israel Council for Higher Education PhD Sandwich Scholarship.

Author Contributions

M.N.: conceptualization; investigation; writing-original draft; project administration; writing-review and editing; S.S.-Z.: formal analysis; investigation; M.K.: formal analysis; investigation; E.B.-A.: formal analysis; investigation; O.S.: data curation; formal analysis; investigation; P.A.: resources; supervision; funding acquisition; C.T.: formal analysis; supervision; investigation; R.Y.: conceptualization; resources; supervision; funding acquisition; K.M.: conceptualization; supervision; funding acquisition; writing-review and editing; E.J.: conceptualization; supervision; funding acquisition; writing-review and editing.

Data Availability

The metabolomic data are available in the Biostudies repository with identifier S-BSST1255. The proteomic data is available at the PRIDE repository with the DOI 10.6019/PXD048649.

References

Author notes