Astrocyte-specific overexpressed gene signatures in response to methamphetamine exposure in vitro

Astrogliosis is among the earliest consequences of Meth use to the CNS. We investigated the early effects of Meth exposure on primary cortical astrocytes, by focusing on genes that were upregulated to above a conservative 4-fold threshold, which corresponded to more than 62% of the total changes in gene expression caused by the drug at any concentration over 24 h of exposure, as determined by gene array. Importantly, over the course of the experiment, we did not observe significant decrease of cell viability due to the exposure to different concentrations of Meth, as assessed by TdT detection by in situ hybridization (not shown).

We found that 411 genes were increased when the astrocytic cultures were exposed to Meth at concentrations of 10 or 100 μM, in comparison to controls, while in treatments with 1 μM of Meth, 180 genes were significantly increased. All together, 179 genes were significantly upregulated to above 4-fold by all the three doses of Meth compared to vehicle-treated astrocytes (Table 1). The correlation coefficient confirms that for the majority of the genes, Meth induced a dose-dependent response pattern (Table 1), which is visualized in Fig. 1. Figure 1a shows fold change of all the upregulated genes at the different doses of Meth. All genes that increased to above 4-fold at 10 μM of Meth compared to control were also significantly increased with 100 μM, but only 32% of those were also increased by the 1-μM condition. The calculation of the average fold change showed a significant overall dose-response effect (Fig. 1b), with a Pearson coefficient equal to 0.89 and a p value ≤0.0001. However, the examination of a correlation coefficient in all the individual upregulated genes showed that 28% of all genes exhibited a flat response that was equal in all 3 doses, while 22% of the genes showed a negative correlation coefficient, suggesting an inverse dose-response effect.

We examined the top 30 most upregulated genes in each of the three doses of Meth used to stimulate astrocytes (Table 2). Surprisingly, the majority of the genes differed between the three doses. However, a few genes appeared among the 30 most upregulated genes by all conditions, compared to controls. These genes, which appear in Table 1 as italics, were the mitogen-activated protein kinase kinase 5 (MAP2K5), the G protein-coupled receptor 65 (GPR65), ectodysplasin A (ED1), the neuron navigator 3 (NAV3), and CXCL5, which were chosen for a deeper analysis of the behavior of genes in network with them and pathway changes resulting from Meth exposure.

MAP2K5, GPR65, and CXCL5 were further validated by PCR and also at the protein levels. The qPCR validation results confirmed the potential importance of these three molecules in the direct response of astrocytes to Meth (Fig. 2a). The PCRs were performed in the same samples used for gene array and also in two independent experiments. MAP2K5 and GRP65 were significantly upregulated by all three doses, and CXCL5 was significantly upregulated in 10 and 100 μM treatments (Fig. 2a). We confirmed the relevance of these findings at the protein level, using specific antibodies against MEK5, the protein transcribed by the MAP2K5 gene, and GRP65, by western blot (Fig. 2b, c). The enrichment of these proteins was also confirmed by imaging, using specific antibodies against MEK5, GRP65, and CXCL5 (Fig. 2c, d). The increase in these markers first identified using systems biology tools suggests the value of the approach and its power for the identification of changes in genes that depend on or are associated with MAP2K5, GRP65, and CXCL5. We again utilized systems biology to search for gene networks associated to these genes, as well as to ED1 and NAV3, exhibiting synchronic behaviors in response to Meth in astrocytes.

For the examination of the behavior of genes in network with MAP2K5, GPR65, CXCL5, ED1, and NAV3, we used GeneMania in Cytoscape, and for that, we focused on the 10 μM dose, which represents levels of drug reaching the brain in Meth users [23]. The analysis of the changes was conducted using a protocol utilized in our lab, to determine the behavior of genes associated to the ones we chose to prioritize, based on pathway, physical and genetic interactions, shared protein domains, or coexpression, in order to predict molecular networks with which astrocytes might respond to acute drug abuse. Using GeneMania and JActiveModules in Cytoscape [24‐27], we identified such gene node clusters. The highest score node contained 141 genes, which were all upregulated, and which clustered with a coefficient of 0.174, suggesting that Meth has a strong effect on astrocytic gene networks. Ninety of those genes (63.8%) showed multi-edged node pairs, suggesting a strong interaction between molecular changes and processes triggered by Meth in astrocytes (Additional file 1: Figure S1). MAP2K5, GPR65, and ED1 (EDA or the rat homolog of CD68), which were consistently among the 30 genes most upregulated by all three doses of Meth (Table 2—italic letters), were also represented in this large gene cluster. A literature examination suggests that these genes could be a link between the acute response of astrocytes to Meth and the potential development of CNS alterations. For instance, ED1 is a microglial marker but it can be found expressed on tumoral astrocytes [28]. Interestingly, MAP2K5 is also a characteristic of tumorigenesis [29]. GPR65, on the other hand, is a proton- and acid-sensing G-protein receptor that plays an important role in cell survival and is also known as TDAG8 [30, 31].

We examined subfamilies of genes assembled as child nodules, by connecting first neighbors of MAP2K5 (Fig. 3a) and GPR65 (Fig. 3b), which led to subnetworks respectively assigned to neuronal support and inflammation. For instance, the genes that appeared in connection with MAP2K5 (Fig. 3a) were annotated to MAPK signaling (p = 0.0022), gap junction (p = 0.0051), the GnRH signaling pathway (p = 0.0062), and also neuroactive ligand-receptor interactions (p = 0.008), suggesting association to neurological outcomes. The genes connected to GPR65 (Fig. 3b) were associated to pathways involving cytokine-cytokine interaction (p = 0.00027), chemokine signaling pathway (p = 0.04), B cell receptor signaling pathway (p = 0.003), Fc-gamma R-mediated phagocytosis (p = 0.0048), and systemic lupus erythematous (p = 0.0052), suggesting a role in potential inflammatory outcomes.

ED1 was not represented in either one of these subnetworks, but a cluster analysis centered on first neighbors of this gene (Fig. 3c), resulted in a group of genes functionally assigned as glycoproteins (p = 0.0018), and functionally annotated to immune response (p = 0.029) and cell adhesion (p = 0.03).

Other genes among the 30 most upregulated ones by all three doses were NAV3 and CXCL5 (Table 2). NAV3 was represented within subnetworks associated to GPR65 (Fig. 3b), as well as to ED1 (Fig. 3c). CXCL5, on the other hand, segregated on a network (Fig. 4) that was heavily associated to chemokine signaling pathway (p = 8.7E−7), cytokine-cytokine receptor interaction (p = 6E−5), neuroactive ligand-receptor interaction (p = 0.0037), and calcium signaling pathways (p = 0.01).

We further dissected the highest score node, to produce subnetworks that were derived by the introduction of a display restriction connecting genes only through pathway and physical associations. This restrictive approach generated one network (Fig. 5), in which MAP2K5 appeared as the strongest upregulated gene, in correlation with other genes described in neurological processes, metabolism, and inflammation (Fig. 5). Importantly, as a control, the cortical astrocyte marker ErbB3 [32] was represented in this network.

We performed pathway enrichment analysis on the set of genes that were upregulated by Meth in astrocytes, using the DAVID Bioinformatics Database (KEGG_PATHWAY) (Table 3) and the iPathwayGuide (Fig. 6). We found that Meth treatment on astrocytes caused an important enrichment of genes that are relevant in neuroactive ligand-receptor interactions, immunity, and metabolic outcomes, as shown in Table 3.

Further analysis was conducted to examine the number of gene perturbation accumulation (pACC) versus overrepresentation p value (pORA) within pathways (Fig. 6) and that also indicated that CNS- and pathology-relevant genes were strongly represented. For instance, genes involved in neuroactive ligand-receptor interaction (p = 0.001) and circadian rhythm (p = 0.012) led to an important contribution to changes induced by Meth in astrocytes. A substantial, but not statistically significant, representation of genes involved in chemokine signaling pathways was also observed (p = 0.056). Interestingly, Meth also triggered genes that are involved in the resistance to infections (Fig. 6). The strong representation of pathways that may be involved in neurological outcomes prompted an analysis of individual genes.

Of the genes annotated to the neuroactive ligand-receptor interaction pathway (Fig. 7), a few were interesting for being associated to neurological disorders in other systems, for instance, the Period Circadian Clock 2 (PER2) [33, 34]. Others have been also described to have essential inflammatory roles, for instance, prostanoid receptors, such as the prostaglandin E receptor 3 (PTGER3), the thromboxane A2 receptor (TBXA2R), and the prostaglandin D2 receptor (PTGDR), which play important roles in inflammatory reactions and hyperalgesia [35].

The other genes upregulated by Meth treatment of astrocytes that could be mentioned for their potential connection with CNS-related syndromes were the following: the inositol 1,4,5-triphosphate receptor, type 1 (ITPR1, 9.24-fold upregulation) is involved in spinocerebellar ataxia [36]; the cold autoinflammatory syndrome 1 (CIAS1, 10.94-fold) in cryopirin-associated periodic syndrome [37]; the RET proto-oncogene (7.4-fold) in congenital central hypoventilation syndrome [38, 39]; PER2 (4-fold) in familial advanced sleep phase syndrome [40]; nucleoporin 62 kDa (NUP62, 4.72-fold) in infantile bilateral striatal necrosis [41]; ubiquitin-activating E1 (UBE1,10–45-fold) [42] and ankyrin repeat domain 2 (stretch responsive muscle, ANKRD2, 9.06-fold) in paralytic syndromes and cerebral palsy; interleukin 1 receptor antagonist (IL1RN, 6.43-fold), the interleukin 2 receptor common gamma chain (IL2RG, 10.02-fold), the ATP-transporter 2 (TAP2, 20.47-fold), PTGDR (11.97-fold), PTGER3 (27.56-fold), TBXA2R (6.3-fold), and the Complement Component 2 (C2, 5.8-fold) in CNS inflammation; the AarF domain containing kinase 4 (ADCK4, 11.04-fold) in coenzyme Q deficiency [43]; the zinc finger protein 41 (ZNF41, 5.39-fold) and the angiotensin II receptor, type 2 (AGTR2, 10.44-fold) in fragile X syndrome, cerebral ataxia, mental retardation, and disequilibrium syndrome [44, 45]; laminin, alpha 2 (LAMA2, 4.31-fold) in congenital muscular dystrophies and control of the blood-brain barrier [46]; and the cholinergic receptor, nicotinic, alpha 1 (CHRNA1, 7.73-fold) involved in congenital myasthenic syndrome and considered as an important potential drug target [47].

Using qRT-PCR, we confirmed the transcriptional upregulation of several of these genes, prioritized by their presence in the two most strongly represented pathways, the neuroactive-ligand and the cytokine-cytokine receptor interaction pathways (Table 4), in all three doses of Meth in vitro. These genes were the IL1RN, IL2RG, as well as the prostanoid receptors PTGDR, PTGER3, and TBXA2R (Fig. 8 and Table 4). These genes play important functions in inflammatory processes in the brain and elsewhere. We also examined other upregulated genes that have been described as molecules that could potentially influence the immune environment in the CNS, TAP2, and C2 (Fig. 8). All these validated genes have been suggested to play important roles in the CNS and neurological disorders [23, 48‐53]. Our validation by PCR confirmed that the Meth treatment has the capacity to directly stimulate the upregulation of these genes, which have an involvement in inflammatory processes in the brain. Among them, TAP2 and PTGDR were upregulated by 10 and 100 μM Meth but not by 1 μM, while the other genes were transcriptionally increased by all the doses, validating the gene array data.

Although our analysis of changes in astrocytic gene expression was focused on enrichments, a few of the genes highlighted in our study have showed network connections with genes downregulated by the Meth treatment. The increase in MAP2K5, for instance, was associated with the decrease on serine/threonine protein kinase 2 (PAK2) (0.6-fold, p = 0.041) (Fig. 5). Other molecule that showed several downregulated connectors was IL2RG (Fig. 9). A gene cluster centered on IL2RG was associated with several components of the Jak-Stat signaling pathway (p = 0.0002, Benjamini = 0.002), primary immunodeficiency (p = 0.008, Benjamini = 0.03), and cytokine-cytokine receptor interaction (p = 0.006, Benjamini = 0.06) pathways. Of the genes in the IL2RG interactive subnetwork, several were significantly downregulated, such as the inducible T cell costimulator ligand (ICOSLG) (0.3-fold, p = 0.006) or the cytochrome P450 family member CYP2B6 (0.1-fold, p = 0.0012), suggesting that the upregulated genes may interfere with, or become affected by, transcriptional suppressions. The role of the downregulated genes in the development of astrocytic changes caused by Meth must be examined in the future.