Differential gene expression in the contralateral hemisphere of the rat brain after focal ischemia

Ischemic stroke is one of the most severe polygenic brain diseases. Here, we performed further functional genetic analysis of the processes occurring in the contralateral hemisphere (CH) after ischemia–reperfusion injury in rat brain. Comparison of RNA sequencing data for subcortical samples from the ipsilateral hemisphere (IH) and CH after 90 min of transient middle cerebral artery occlusion (tMCAO) and corresponding sham-operated (SO) controls showed four groups of genes that were associated with ischemic processes in rat brain at 24 h after tMCAO. Among them, 2672 genes were differentially expressed genes (DEGs) for IH but non-DEGs for CH, 34 genes were DEGs for CH but non-DEGs for IH, and 114 genes had codirected changes in expression in both hemispheres. The remaining 16 genes exhibited opposite changes at the mRNA level in the two brain hemispheres after tMCAO. These findings suggest that the ischemic process caused by a focal ischemia induces complex bilateral reactions at the transcriptome level in the rat brain. We believe that specific genome responses in the CH and IH may provide a useful model for the study of the potential for brain repair after stroke.

Ischemic stroke is a multifactorial disease with a complex etiology and global consequences 1,2 . Cerebral ischemia causes a cascade of biochemical changes in brain tissues 3 . During cerebral ischemia, the induction of glutamatemediated pathways triggers cell death by massive calcium influx following vessel occlusion 4,5 , and the lack of oxygen causes failure of the respiratory chain and mitochondrial function 6,7 . These changes lead to an inflammatory reaction, which is accompanied by an acute imbalance of matrix metallopeptidases 8 . Violation of the function of the blood-brain barrier (BBB) allows the invasion of immune cells and pathogenic factors into brain cells 9,10 . Similar effects caused by disturbance of the BBB, such as accumulation of excess oxygen radicals and activation of apoptosis, can also be caused by reperfusion after ischemia 11,12 . The involvement of these processes in the development of ischemic injury can be observed at the transcriptome and proteome levels using stroke models in small laboratory animals 3,[13][14][15][16][17][18][19][20][21] .
Ischemia-related metabolic changes occur in both the damaged ipsilateral hemisphere (IH) as well as in the opposite contralateral hemisphere (CH) of the brain 22,23 . The processes of transhemispheric diaschisis and spreading depression/depolarization after focal ischemic damage are associated with massive transmembrane ion shifts distant from the original site of injury [23][24][25][26] . The gene expression patterns can change in the CH after focal cerebral ischemia 9,[27][28][29][30] . In experiments involving the CH and sham-operated (SO) controls, we recently found traces of a bilateral genetic response after focal stroke induced by transient middle cerebral artery occlusion (tMCAO) in the rat brain 31 . Comparisons with control conditions identified overlapping differentially expressed genes (DEGs) that reflected the general transcriptome response of IH subcortical cells at 24 h after tMCAO. We also identified sets of nonoverlapping genes that are unique in the SO and CH controls. These DEGs may indicate that transcriptome changes in IH subcortical cells are associated with transhemispheric differences in models of ischemia. The functional annotation of gene sets reveals an association with inflammatory and immune reactions occurring in the CH zone in response to focal ischemia-reperfusion (IR) injury 31 . However, the role of genomic parameters in the CH in the mechanisms of ischemic injury and neuroprotection remains underestimated and requires further study.
In the present study, we performed further functional genetic analysis of the processes occurring in the CH after IR injury. Comparison of RNA sequencing (RNA-Seq) data for rat brain subcortical samples from the IH

Results
RNA-Seq analysis of the effects of IR on the mRNA level in subcortical structures of the CH relative to the SO controls. We previously used RNA-Seq to evaluate the transcriptional activity of the mRNAs for 17,367 genes in the subcortical structures of the rat brain at 24 h after tMCAO. A pairwise comparison for the primary analysis of the RNA-Seq data for IR-c vs. SO-l revealed significant changes at the mRNA level for 164 genes at 24 h after tMCAO. Here, we analyzed these DEGs further and found that the expression levels of most of these DEGs were higher in the IR-c than in the SO-l group (96 upregulated vs. 68 downregulated DEGs) (Fig. 1a, see Supplementary Table S1). The volcano plot in Fig. 1b shows the differences in mRNA expression levels between the IR-c and SO-l groups. We note that upregulated DEGs included Cyr61, Atf3, Socs3, Apold1, and Inmt, which were upregulated by ≥ 5 times. The top five downregulated DEGs in IR-c vs. SO-l were P2rx2, Th, Hcrt, Pmch, and Parpbp, in which the mRNA level was reduced by ≥ 4 times (Fig. 1c).
RT-PCR analysis of the expression of four upregulated (Pla2g3, Cd14, Hspb1, Socs3), two downregulated (Pmch, Adcyap1), and two nonsignificantly (Bcl2l2, Mmp9) regulated genes was used to verify the RNA-Seq www.nature.com/scientificreports/ results in IR-c vs. SO-l. The characterization of the primers is shown in Supplementary Table S2. The real-time RT-PCR results confirmed the RNA-Seq data (Fig. 1d).

Comparison of RNA-Seq results and identification of specific genome responses in the CH and IH.
Using RNA-Seq, we previously identified 2802 DEGs with 1390 upregulated and 1412 downregulated mRNAs in the rat ipsilateral subcortex relative to the corresponding brain samples from SO rats (IR-i vs. SO-r) 31 .
Here, we performed a meta-analysis of the mRNA sequencing results. We analyzed the effects of focal IR on the mRNA levels in each of the hemispheres of the brain and compared these to the levels in the corresponding SO control (Fig. 2a). The results of the pairwise comparisons of the data for IR-i vs. SO-r and IR-c vs. SO-l are presented in the Venn diagram in Fig. 2b. The number of DEGs was 10 times lower in the CH (IR-c vs. SO-l) than in the IH (IR-i vs. SO-r) groups. Simultaneously, we found 130 overlapping genes whose expression changed in both hemispheres at 24 h after tMCAO (Fig. 2b). The Venn diagrams shown in Fig. 2c include only the upregulated DEGs and those shown in Fig. 2d show only the downregulated DEGs for both comparisons.
We identified 67 DEGs that were upregulated in both the comparisons for IR-i vs. SO-r and IR-c vs. SO-l ( Fig. 2c) and 47 DEGs whose mRNA was downregulated in both pairwise comparisons (Fig. 2d). Thus, the expression level of 114 of 130 overlapping DEGs between IR-i vs. SO-r and IR-c vs. SO-l changed codirectionally in both hemispheres at 24 h after focal IR. The top 10 overlapping genes with the greatest fold change between the IR-c vs. SO-l groups are presented in Fig. 2e. Among them were genes encoding proteins with transcription factor (Atf3, Lmx1a), catalytic (Socs3, Inmt), lipid and protein binding (Apold1, Sostdc1), and transporter (Slc17a6) activities based on the PANTHER data. A full list of these DEGs is included in Supplementary Table S3.
The Venn diagram in Fig. 2e shows 2672 and 34 DEGs that were unique for the effects of IR in the IH (IR-i vs. SO-r) and CH (IR-c vs. SO-l) transcriptomes, respectively. Thus, among the 2672 genes unique to the comparison between IR-i and SO-r were genes encoding proteins of the neurotransmission system (Gria3, Adcy1, Grm5, Neurod6, Kcna2, and Kcnh5); components of the immune system (Ccl3, Ccl7, and Ccl22); the 70-kD heat shock protein (Hspa1a), nitric oxide synthase (Nos3); and others (see Supplementary Table S4). Among the 34 DEGs unique to the comparison between IR-c and SO-l were Pla2g3, Htra1, Tnfrsf11b, Iqgap3, S100a9, Avp, Nr4a3, Folr1, Pmch, and other genes encoding transcription factors, hormones, and immune response proteins (see Supplementary Table S5). The differential expression data for the top five unique genes for the comparisons IR-i vs. SO-r and IR-c vs. SO-l flanks (relative complements) of the Venn diagram ( Fig. 2a) are presented in Fig. 2f and g, respectively.
Hierarchical cluster analysis of all DEGs for the comparisons IR-i vs. SO-r and IR-c vs. SO-c is illustrated in Fig. 2h. The common features of the differential expression profiles of these comparison groups reflect the effects of IR in both rat brain hemispheres. At the same time, individual differences between the groups seem to characterize the specific genome responses in the CH and IH at the transcriptome level.  Fig. 2d): Hrh3, Chst15, Drd2, Rasd2, Drd1, Hpca, Lrrc10b, Slc24a4, Scn4b, Neu2, Gng7, Adora2a, Asic4, Syndig1l, Rgs9, and Gpr6 were identified. Each of these was downregulated for IR-i vs. SO-r and, conversely, upregulated for IR-c vs. SO-l. These results suggest an opposite directionality of the changes in expression for these DEGs in the two rat brain hemispheres. The differential expression data for these DEGs for IR-i vs. SO-r and IR-c vs. SO-l are shown in Fig. 3c. Using the PANTHER functional annotation tool, we found that these genes are associated predominantly with the neurosignaling system.
Signaling pathways associated with DEGs in the CH of the rat brain at 24 h after tMCAO. Using DAVID, we found only one signaling pathway (rno05133:Pertussis) that was associated with DEGs for the comparison IR-c vs. SO-l. The GSEA tool, which provides information about enrichment using a more gentle statistical approach, revealed > 100 signaling pathways (FDR q-value < 0.05) associated with DEGs for IR-c vs. SO-l. The top five included Orexin receptor, AP1, IL18, IL6, and IL7 pathways, and signaling by receptor tyrosine kinases (see Supplementary Table S6). Most of the signaling pathways, including those listed, were associated predominantly with upregulated DEGs for the comparison IR-c vs. SO-l. Some overlapped between the IR-c vs. SO-l and IR-i vs. SO-r enrichment results. Concomitantly, four signaling pathways unique for the CH were identified that were associated with DEGs (cutoff > 3.5-fold) for IR-c vs. SO-l but not for IR-i vs. SO-r (see Supplementary  Table S6). These included p38 signaling mediated by MAPKAP kinases, signaling pathways for the regulation of TLR by endogenous ligand, hypertrophy model, and extracellular vesicles in the crosstalk of cardiac cells. We note that upregulated DEGs were associated predominantly with each of the signaling pathways identified in the pairwise comparisons (see Supplementary Table S6).
Analysis of the involvement of the DEGs that had codirectionally changed mRNA level in two brain hemispheres after tMCAO in the signaling pathways that were modulated during ischemia. We analyzed 114 DEGs (67 upregulated and 47 downregulated) that had codirectionally changed at the mRNA level in the two brain hemispheres at 24 h after tMCAO (Fig. 2c,d). Using DAVID, the signaling pathway annotations (KEGG PATHWAY (KP), REACTOME PATHWAY (RP), and WIKIPATHWAYS (WP)), were downloaded for these DEGs. Based on previous data 31 , we selected only those annotations with a Padj < 0.05 for each set of DEGs for the comparison IR-i vs. SO-r. This produced a list of 133 KP, 21 RP, and 6 WP annota-  4A).
In the network presented in Fig. 4, nodes are designated as DEGs, and each line connecting the nodes indicates an involvement of the protein product of the corresponding gene in signaling pathway functioning. All genes except for Cryap and Kcnc2 had connections with the other genes on the list analyzed (Fig. 4a). The lines   Fig. 4b. Predominantly, upregulated DEGs were associated with such pathway. Among them were Cd14, Flt1, Fos, Gadd45g, Jun, Myc, Nr4a1, Hspb1, Dusp1, and Dusp6. Only one gene (Taok1) that is associated with the MAPK signaling pathway was downregulated in both the CH and IH relative to the SO controls at 24 h after tMCAO (Fig. 4b).
The networks shown in Fig. 4 show the spectrum of genes and their functional connections that reflect changes in the activities between different hemispheres of the brain at 24 after tMCAO. Such genes are associated predominantly with modulation of the immune and inflammatory responses in rat brain cells under IR conditions.
Analysis of the involvement of the DEGs that exhibited opposite changes in the mRNA level in two brain hemispheres after tMCAO in the signaling pathways modulated during ischemia. Next, we analyzed 16 DEGs (Hrh3, Chst15, Drd2, Rasd2, Drd1, Hpca, Lrrc10b, Slc24a4, Scn4b, Neu2, Gng7, Adora2a, Asic4, Syndig1l, Rgs9, and Gpr6) that were downregulated in IR-i vs. SO-r but upregulated in IR-c vs. SO-l (Fig. 3c). Similarly, based on previous data 31 , only those annotations (KP, RP, WP) with a Padj < 0.05 for the set of DEGs in the comparison IR-i vs. SO-r were selected for analysis. We formed a list of 27 KP, 12 RP, and 2 WP annotations that were associated with eight of the 16 DEGs analyzed. The remaining four DEGs analyzed did not have significant annotations identified using DAVID. The association of DEGs with functional annotations suggest the participation of Hrh3, Drd2, Drd1, Scn4b, Gng7, Adora2a, Asic4, and Rgs9 in the signaling pathways that exhibited changes in their activities under IR conditions in the subcortical structures of the brain at 24 h after tMCAO (Fig. 5).
In the network shown in Fig. 5, nodes are designated as DEGs or signaling pathways. Each line connecting the nodes indicates the involvement of the protein product of the corresponding gene in signaling pathway functioning. All signaling pathways were clustered and annotated using the KP, RP, and WP databases. In the network shown in Fig. 5, the top five genes involved in the most signaling pathways were Gng7 (29 pathways), Drd1 (eight pathways), Drd2 (eight pathways), Adora2a (seven pathways), and Rgs9 (six pathways). Two or more of the genes analyzed are involved in the implementation of 13 signaling pathways. The signaling pathways "Signal Transduction" (RP) and "Signaling by GPCR" (RP) were associated with the most DEGs (five) on our list; Hrh3, Drd2, Gng7, Adora2a, and Rgs9 genes were associated with each of these signaling pathways. Additionally, we identified signaling pathways, "Neuroactive ligand-receptor interaction" (KP), "Cocaine addiction" (KP), "cAMP signaling pathway" (KP), and "Dopaminergic synapse" (KP) that were associated with four, three, three, and three DEGs from our list, respectively (Fig. 4). Interestingly, the signaling pathways were associated predominantly with the functioning of the neurosignaling system.
We note that all DEGs analyzed had opposite directionality of expression changes in the two rat brain hemispheres relative to the corresponding SO controls. Thus, this network shows the spectrum of genes and functional-related systems whose activities may change in different hemispheres of the brain at 24 after tMCAO.

Discussion
In present study, we performed functional genetic analysis of the processes occurring in the CH after a focal stroke. We have previously used the tMCAO (90 min) rat model under MRI monitoring 31 . This model mimics the effects of ischemic stroke events in humans 32,33 . All rats in the IR group had a focal hemispheric lesion lying in the IH region and that spread to the adjacent cortex 31 . We found no pathological changes in the CH of ischemic rats in the MRI data 31 . Based on the RNA-Seq data for this series of rat brain samples from the IH and CH and corresponding SO samples 31  www.nature.com/scientificreports/ in the CH. The obtained data support the previously established immunomodulatory effects after ischemia in brain areas away from the focal injury 31,34,35 .
Using RNA-Seq, we previously compared hundreds of DEGs in the rat IH subcortex relative to the corresponding brain samples from SO rats 31 . Here, our comparison of the sets of DEGs in the IH and CH relative to the left and right hemispheres in SO animals, respectively, identified four groups of genes (see Supplementary  Fig. S1 online). First, we found genes that overlapped between the pairwise comparisons and that had a codirectional change in expression in both the IH and CH. These genes reflect the general effects of ischemia on the genome in brain cells located at different distances from the stroke site. These genes are associated predominantly with modulation of immune and inflammatory responses (e.g. immune system, MAPK, TNF signaling pathway) in rat brain cells under IR conditions. Thirteen of these DEGs were also identified by Fury et al. using a mouse tMCAO model with 30 min filament period at 1 day after occlusion (brain area + 3.1 to − 4.1 mm from bregma) 30 . Socs3, Inmt, Lbp, Maff, Serpine1, Angptl4, Gbp2, Ifitm3, Gpnmb, Gfap, Emp1, Upp1, and Acer2К found in our study and that by Fury et al. were upregulated in both studies and belong to the first group of genes that exhibited a codirectional change in expression in both the IH and CH. Analysis of the results obtained in two pairwise comparisons of IR-c vs. SO-l between Fury et al. and our present studies is shown in Supplementary  Fig. S2. Fifty overlapped genes included 13 aforementioned genes. Meanwhile, most of the genes (533) were only among Fury et al. results, whereas 114 genes were only among our results. A small number of overlapping DEGs can be explained by objective variations in the experimental conditions of different studies (type of animals, occlusion time, brain area, bionformatics). But also it can indicate an active spatial-temporal regulation of the gene expression in the brain under conditions of cerebral ischemia. Perhaps, the differences can accumulated as the analyzed area of the brain moves away from the focus of ischemic damage.
The second group of genes was the most numerous (2,672 genes) and included those genes whose expression level changed significantly in the IH but not in the CH after tMCAO. The pattern of transcriptional activity largely reproduced the inflammatory and neurotransmitter-related effects of ischemia in the IH 15,30 . However, these genes reflect the unique effects of ischemia in the brain region containing the area of the ischemic focus rather than far from the focus of damage.  www.nature.com/scientificreports/ The third group included 34 genes (e.g. Pla2g3, Htra1, Tnfrsf11b, Iqgap3, S100a9, Avp, Nr4a3, Folr1, and Pmch) that are unique to the CH transcriptome reaction. The mRNA levels for these genes changed significantly only in the CH, a finding that reflects effects remote from the area of ischemia. Six of these genes, namely Parpbp, Pmch, Th, Hcrt, P2rx2, and S100a9, exhibited cutoff > 3.5-fold and were identified by GSEA as associated with the p38 signaling pathway mediated by MAPKAP kinases and the signaling pathway for the regulation of TLR by endogenous ligand, which are unique to the CH. This rat tMCAO model reflects ischemia and recovery after stroke in humans. Our study is the first to identify this set of genes in the response to IR injury, and these findings may provide a basis for further biomedical research in this field.
The fourth group of genes is of particular interest. We identified 16 genes (Hrh3, Chst15, Drd2, Rasd2, Drd1,  Hpca, Lrrc10b, Slc24a4, Scn4b, Neu2, Gng7, Adora2a, Asic4, Syndig1l, Rgs9, and Gpr6) that exhibited opposite changes in expression levels in the two rat brain hemispheres. Using the Functional Annotation Tools PANTHER and DAVID, we found that these genes are associated predominantly with the neurosignaling systems and may be nodes for the regulation or brain neurotransmission after stroke. To examine this regulatory network, we used the signaling pathway depository from the three most representative databases, KP, RP, and WP. Each of these databases has its own specific features and each contains information about the functional annotation of protein gene products and their biological functions [36][37][38] . However, databases differ in the number of pathways they contain, average number of proteins per pathway, types of biochemical interactions included, and subcategories of pathways they provide. Pathways are also often described at different levels of detail, with different types of data, and with vaguely defined boundaries [39][40][41] . Therefore, we used a few databases to overcome the subjective features of each. We found that the "Signal transduction" (RP), "Signaling by GPCR" (RP), "Neuroactive ligandreceptor interaction" (KP), "Cocaine addiction" (KP), cAMP signaling pathway (KP), and Dopaminergic synapse (KP) were top neurosignaling pathways that were simultaneously associated with a full set of DEGs in the IH and with the most DEGs from fourth group.
Concomitantly, the top five genes involved in the most signaling pathways were identified as Gng7 (29 pathways), Drd1 (eight pathways), Drd2 (eight pathways), Adora2a (seven pathways), and Rgs9 (six pathways). The role of these genes in ischemia is only partially understood and continues to be actively studied. We have reported downregulation of Gng7 expression in the rat brain at 24 h after tMCAO 15 . Gng7 encodes subunit gamma 7 of guanine nucleotide-binding proteins (Gγ7, GNG7). The role of Gγ7 function in dopamine D1 and D2 receptor neurosignaling has been reported 42,43 . Interestingly, D1 receptor-mediated endogenous tissue plasminogen activator upregulation contributes to BBB injury after acute ischemic stroke 44 . Postischemic administration of a D2 receptor agonist reduces cell death by activating mitochondrial pathway following ischemic stroke 45 . D2 and Drd2 are associated with the regulatory action of neuroactive drugs after stroke 46,47 . Additionally, the role of adenosine A2A receptors, which are encoded by Adora2a, under cerebral IR conditions has been reported 48 .
Recently, Ito et al. directly compared the molecular responses between mice that had recovered naturally (spontaneously) from stroke and those that had not recovered, and reported novel molecular signatures that have been masked by studies using external restorative treatments 49 . The RNA-Seq data of Ito et al. revealed a panel of recovery-related genes in the motor cortex of spontaneously recovered mice and highlighted the involvement of the contralesional cortex, particularly the Adora2a, Drd2, and Pde10a-mediated cAMP signaling pathway, in spontaneous recovery 49 . Interestingly, we found overlapping between the gene lists of Ito et al. and our fourth group of genes mentioned above. This overlapping is probably not a coincidence and suggests the functional significance of genetic clustering in the processes of postischemic brain recovery, including using the potential of the CH cells.
A limitation of our study is the lack of different timepoints to provide the dynamic changes of RNA expressions, as well as functional evidence supporting the changes in the mRNA expression of these genes. By establishing the functional interactions between the DEGs, we believe that we have identified changes at the RNA level in both the CH and IH that provide clues about the effects of stroke on the brain.
In conclusion, we comprehensively analyzed changes in the transcriptome in the IH and CH after a focal stroke in a tMCAO rat model. We identified evidence of a bilateral genetic response triggered by ischemic stroke in the cerebral hemispheres at the transcriptome level on a genome-wide scale. The specific genome responses in the CH and IH may be useful for the study of the potential for regeneration of brain cells after stroke.

Methods
Animals. White 2-month-old male rats of the Wistar line (weight, 200-250 g) were obtained from the AlCondi, Ltd. animal breeding house (Moscow, Russia) as previously described 31 . All protocols adhered to the Guide for Care and Use of Laboratory Animals to minimize pain and suffering. All methods involving animals are reported in accordance with ARRIVE (Animal Research: Reporting of in Vivo Experiments) guidelines (https:// arriv eguid elines. org).
tMCAO model in rats. The tMCAO rat model was applied with magnetic resonance imaging (MRI) as previously described 31 . The rats were decapitated at 24 h after tMCAO (IR group). Rats in the SO group were subjected to a similar surgical procedure under anesthesia comprising neck incision and separation of the bifurcation but without tMCAO. Each experimental group comprised at least five animals. MRI confirmed that all rats in the IR group had a focal hemispheric lesion lying in the subcortex region of the right (ipsilateral) brain hemisphere and that this had spread to the adjacent cortex. MRI showed no pathological changes in the CH of ischemic rats 31 . MRI of ischemic foci after tMCAO is shown in Supplementary Fig. S3. Also, stroke score data (MRI) are shown in Supplementary Table S8.
The subcortical structures of the brain in the IH group (IR-i group) and CH group (IR-c group) of ischemic rats, and in the right (SO-r) and left (SO-l) hemisphere of SO rats were obtained previously 31  www.nature.com/scientificreports/ subcortical structures were taken in the range from + 2 to − 2 mm from the bregma. The samples included the striatum. All IR-i, IR-c, SO-r, and SO-l samples were placed in RNAlater (Ambion, Austin, TX, USA) solution for 24 h at 4 °C, stored at − 70 °C and then used for RNA isolation. RNA integrity was checked using capillary electrophoresis (Experion, BioRad, Hercules, CA, USA). RNA integrity number (RIN) was at least 9.0 31 .