Epitranscriptome of the ventral tegmental area in a deep brain-stimulated chronic unpredictable mild stress mouse model

2.1 Deep brain stimulation

Before surgery, 6-week-old C57BL/6J mice were handled for 7 days to habituate them to the experimenter and the stimulation procedure. Subsequently, we anesthetized the mice intraperitoneally (pentobarbital sodium, 0.7%, 10 µL/µg) and, using a stereotaxic frame, implanted electrodes (bipolar, two parallel tungsten wires twisted, 0.22 mm in diameter) reaching the bilateral NAc (coordinates: 1.18 mm anterior, 1.35 mm lateral, and 4.50 mm ventral to bregma), according to the Mouse Brain in Stereotaxic Coordinates (Paxinos and Franklin, 2001; Paxinos and Watson, 2005). Following 5–7 days of recovery and then 2 weeks of a 4-week CUMS regime, we subjected the mice to high-frequency stimulations (100 Hz, 100 µA, 100 µs pulse width) daily for 1 h per day (30 min/unilateral, bilateral stimulation) (i.e., NAc-DBS). Thus, stimulations took place from the second week of CUMS, and they continued for another 2 weeks (i.e., the end of CUMS). We controlled current intensity by a constant-current isolated stimulator and we synchronized our stimulation protocols using Master-8 (A.M.P.I., Israel). We also placed control animals (CUMS-sham) into stimulation chambers daily for 1 h per day and connected them to the stimulator without applying any current. At the end of the experiments, we confirmed the localization of electrodes in the NAc on bright-field photomicrographs of coronal sections. Consequently, mice with misplaced electrodes were excluded from all data analysis. VTA tissue was collected after the 2-week period of stimulation or sham treatment.

1. Ethical approval: The research has been complied with all the relevant national regulations and institutional policies for the care and use of animals. All experiments with animals were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee of the Academy of Military Medical Sciences.

2.2 CUMS model

We used a CUMS regimen that followed the procedure originally described by Willner et al. [23] and subsequently adapted for mice [9,24]. This stress model involves the application of repeated, mild physical and psychological stressors. The CUMS paradigm provides a model of a chronic depressive-like state that develops gradually over time in response to stress; hence, CUMS is considered to be a more naturalistic in its induction [20,25]. In our study, we subjected the mice to different kinds of stressors several times a day for 4 weeks in a chronic, inevitable, and unpredictable way. For ethical reasons, the stress procedure did not involve food and water deprivation or immobilization. The stressors used were as follows: damp sawdust; changing the sawdust; placement in an empty cage or an empty cage with water covering the floor; placement in a soiled cage with an aversive odor; cage tilting (45° for 12 h); noise stress (15 min); inversion of the 12:12 h light:dark cycle; lights on or off for a short time during the dark phase or light phase, respectively; foot shock (150 mA, 30 min); and ice-water swimming (15°C, 5 min). We administered these stressors in a pseudo-random manner so that they could occur at any time of day or night and we changed the stressor sequence weekly to ensure the stress procedure was unpredictable. During the behavioral tests, the stress procedure was slightly modified: we reduced the number of stressors applied during the light period so as not to interfere with the tests. In addition, we did not subject test mice to any stressors 12 h before the behavioral tests. Nonstressed mice were left undisturbed in their home cages. In all the experiments, the first 2 weeks of CUMS were DBS-free, whereas the second 2 weeks of CUMS also included administration of DBS. To determine the behavioral effects of the CUMS regimen and DBS treatment and to confirm induction of depression-like behaviors [26,27,28,29,30], we examined sucrose consumption in a sucrose preference test (SPT) and total immobility in a tail suspension test (TST), and forced swimming test (FST) was carried out the next day. We also measured locomotor activity using an open field test (OFT), and no stress was applied the day before the test. We isolated non-stressed animals for 1 day before the OFT in order to match the conditions used for CUMS mice. The methods and results of the above-mentioned behavioral tests are shown in the Supplementary Materials and Supplementary Figures S1 and S2.

2.3 m⁶A-mRNA epitranscriptomic microarray detection

We collected the VTA of mice for assessment of m⁶A at designated time points by manual dissection of fresh brains on ice. For each sample, we randomly selected and pooled 6–8 animals from the same group. Brains were immediately flash-frozen after dissection and we then collected defined tissue punches of VTA using a 1 mm round tissue punch while sectioning brains on a cryostat. Total RNA from each sample was quantified using the NanoDrop ND-1000. We performed sample preparation and microarray hybridization based on Arraystar’s standard protocols. Briefly, the total RNA was immunoprecipitated with anti-N⁶-methyadenosine (m⁶A) antibody. We then labeled the modified and unmodified RNAs with Cy5 and Cy3, respectively, as cRNAs in separate reactions using an Arraystar Super RNA Labeling Kit. We combined the cRNAs and hybridized them onto an Arraystar Mouse mRNA Epitranscriptomic Microarray (8 × 60 K, Arraystar; 48,161 mRNAs). After washing the slides, we scanned the arrays in two-color channels using an Agilent Scanner G2505C. We then used Agilent Feature Extraction software (version 11.0.1.1) to analyze the acquired array images. First, we normalized the raw intensities of modified (Cy5-labeled) and unmodified (Cy3-labeled) RNAs using the average log2-scaled spike-in RNA intensities. We then calculated the m⁶A methylation level for the percentage of modification based on the Cy5-labeled and Cy3-labeled normalized intensities. We also calculated expression levels based on the total Cy5-labeled and Cy3-labeled normalized intensities. Subsequently, we identified differentially m⁶A-methylated or expressed mRNAs between the two comparison groups by filtering with a fold change (FC) ≥1.5 and without having replicated samples. In additional analysis, we performed hierarchical clustering using R, gene ontology (GO) analysis using the topGO package (also in R), and pathway analysis using Fisher’s exact test. Details of the experimental process are given in the Supplementary Materials.

3.1 CUMS downregulates gene expression and hypermethylated m⁶A-mRNA modification in the VTA of depression mouse model

We examined gene expression and m⁶A-mRNA modification in the VTA, an important nucleus in depression (Figure 1a and Supplementary Figure S3) [31]. We found thousands of gene expression differences between wild-type (WT) and sham-treated CUMS mice (i.e., those that received DBS electrode implants but not electrical stimulation). Compared with WT mice, CUMS induced at least a 1.5-fold increase or decrease in 9,324 genes, with downregulated genes being three times more prevalent than upregulated genes (Figure 1b, c and Supplementary Figure S4a, b and Supplementary Tables S1, S2). By a process of comparison and screening, we found that CUMS-induced downregulation of genes related to gamma-aminobutyric acid (GABA)/dopamine/glutamate/serotonin receptors and transporters (i.e., Gabrb2, Drd2, Grin1, Htr3a, Slc6a18, Slc6a2, Slc6a20a, Slc6a4, Slc6a5, Slc6a7, Slc6a8, and Slc6a9), synaptic and post-synaptic function (i.e., Snap25, Syn1, Syn2, Syn3, Nxph4, Ppp1r1a, and Dlg4), and transcription factors (i.e., Tcf7l2). In addition, the expression of gene transcripts related to the dopamine signaling pathway (i.e., Pde1b and Syn1) was also downregulated [17,18,32]. Furthermore, the Fto (EntrezID: 26383) gene, which encodes a nucleic acid demethylase, was downregulated. Given that the inactivation of Fto impairs dopamine receptor type 2 (D2R)-dependent control of neuronal activity and behavioral responses [32], the Drd2 gene, which encodes D2R, was also downregulated in our study. GO analysis of the genes downregulated during CUMS revealed enrichment of the GO terms related to cellular localization, transport, assembly, and biosynthesis (Figure 1d, e and Supplementary Table S3). These data suggest that CUMS, which induced the depressive phenotype in mice, acts through pathophysiological pathways and changes neuroexcitability by altering the expression of key neuronal genes involved in all aspects of neurobiology.

Figure 1

Chronic unpredictable mild stress (CUMS) downregulates gene expression in the ventral tegmental area (VTA) of the depression mouse model. (a) Timeline of the experimental process used in this study. Red lines represent CUMS period. (b) Heatmap showing protein-coding genes for which expression levels changed by at least 1.5-fold in CUMS mice compared with wild-type (WT) control mice. (c) Pie chart representing overall changes in gene expression in CUMS mice compared with WT (upregulation: FC > 1.5; downregulation: FC < 0.67). (d) Gene ontology (GO) analysis of differentially expressed mRNAs in CUMS mice compared with WT (upregulation: FC > 1.5; downregulation: FC < 0.67; BP: biological process; CC: cellular component; MF: molecular function). (e) GO analysis of genes downregulated by at most 0.67-fold in CUMS mice compared with WT.

We studied m⁶A-mRNA changes in the VTA of CUMS mice using epitranscriptomic microarray. Compared with WT results, CUMS resulted in 17.8% of genes undergoing m⁶A-mRNA modification (7,605 of 42,655 genes), with ∼98.8% of these genes being hypermethylated (Figure 2a, b and Supplementary Figure S4c, d and Tables S4, S5). These hypermethylation changes induced by CUMS likely lead to changes in the molecular functions of neurons; GO analysis revealed that CUMS-induced m⁶A hypermethylation of genes was associated with the GO terms related to molecular binding function, intracellular organelle/cytoplasmic part, and biosynthesis and metabolic process. In addition, hypomethylated genes were associated with the GO terms related to transcription and translation function, chromatin/chromosome, morphogenesis and development process, and receptor activity (Figure 2c–e and Supplementary Table S3). These results suggest that m⁶A-mRNA methylation in the VTA is relatively high and increases during stress responses.

Figure 2

Chronic unpredictable mild stress (CUMS) hypermethylated m⁶A modification in the ventral tegmental area (VTA) of the depression mouse model. (a) Heatmap showing protein-coding genes for which m⁶A-mRNA modification changed by at least 1.5-fold in CUMS mice compared with wild type (WT) control mice. (b) Pie chart representing the overall changes in m⁶A-mRNA modification in CUMS mice compared with WT (hypermethylation: FC > 1.5; hypomethylation: FC < 0.67). (c) Gene ontology (GO) analysis of differentially m⁶A-methylated mRNAs in CUMS mice compared with WT (hypermethylation: FC > 1.5; hypomethylation: FC < 0.67; BP: biological process; CC: cellular component; MF: molecular function). (d) and (e) GO analysis of m⁶A modifications that involved (d) hypermethylation or (e) hypomethylation in CUMS mice compared with WT (hypermethylation: FC > 1.5; hypomethylation: FC < 0.67).

3.2 NAc-DBS upregulates gene expression and hypermethylated m⁶A-mRNA modification in the VTA of CUMS mice

With the aim of uncovering the molecular mechanisms underlying the corrective effects of NAc-DBS on depressive behaviors in mice subjected to CUMS, we examined gene expression and m⁶A-mRNA modification profiles in the VTA. Two weeks of high frequency (100 Hz)-dependent bilateral NAc-DBS (30 min/unilateral) caused robust activity-dependent gene expression in the VTA (Figure 3a). Indeed, we found thousands of gene expression differences between CUMS mice that had and had not undergone DBS treatment. Compared with CUMS mice that were not treated with DBS, 7,157 genes were either upregulated or downregulated by at least 1.5-fold following DBS treatment, with upregulated genes three times more prevalent than downregulated genes (Figure 3b, c and Supplementary Tables S1, S2). Comparison and screening showed that DBS induced upregulation of genes related to neuronal activity and synaptic function (i.e., Adora2a, Dnah1, Dnah3, Dnah14, Dnah11, and Nxph2), neurotransmitter receptors and transporters (i.e., Slc6a5, Slc6a7, Slc6a8, Scl6a18, Gabra3, Gabra4, Gabbr2, Gabbr1, and Gabrb2), transcription factors (i.e., Tcf7l2), and mRNA methylation modification (i.e., Mettl22, Fto, and Alkbh2) [17,18,32], which had high frequency-dependent changes in expression in VTA neurons (Supplementary Table S1). Furthermore, DBS upregulated transcripts that encode proteins related to glutamate/neuronal signaling pathways, such as Grin3b, as well as those implicated in dopaminergic signaling specifically, including Pde1b, Kcnj6, Syn (synapsin)1, Syn2, and Syn3 (Supplementary Table S1). GO analysis of genes upregulated by DBS revealed enrichment in GO terms, relative to CUMS mice untreated by DBS, such as cellular component biosynthetic process, transcriptional regulation process, and gene expression regulation. For molecular function especially, the activities of mitogen-activated protein kinase kinase (MAPKK) binding and mitogen-activated protein kinase (MAPK) binding, which belong to the MAPK signaling pathway and are closely related to neuroplasticity [33], were upregulated (Figure 3d, e and Supplementary Figure S5 and Table S3).

Figure 3

Nucleus accumbens-deep brain stimulation (NAc-DBS) upregulates gene expression in the ventral tegmental area (VTA) of chronic unpredictable mild stress (CUMS) mice. (a) Schematic diagram of electrode placement for DBS and the timeline for implantation and tissue collection (NAc: nucleus accumbens; NAcLatS: lateral shell area of the nucleus accumbens; NAcMedS: medial shell area of the nucleus accumbens; scale bar: 500 µm). (b) Heatmap showing protein-coding genes for which gene expression changed by at least 1.5-fold in CUMS mice subjected to DBS (CUMS_DBS) compared with control CUMS mice that did not receive DBS (CUMS_Sham). (c) Pie chart representing overall changes in gene expression in CUMS_DBS mice compared with CUMS_Sham mice (upregulation: FC > 1.5; downregulation: FC < 0.67). (d) Gene ontology (GO) analysis of differentially expressed mRNAs in CUMS_DBS mice compared with CUMS_Sham mice (upregulation: FC > 1.5; downregulation: FC < 0.67; BP: biological process; CC: cellular component; MF: molecular function). (e) GO analysis of genes that were upregulated by at least 1.5-fold in CUMS_DBS mice compared with CUMS_Sham mice.

We also profiled m⁶A-mRNA changes in the VTA of CUMS mice with DBS treatment. Compared with CUMS only induced methylation changes, DBS treatment caused methylation changes in 1,061 genes, 62.9% and 37.1% of which were hypermethylated and hypomethylated, respectively (Figure 4a, b and Supplementary Tables S4, S5). GO analysis revealed that genes with m⁶A hypermethylation induced by DBS were enriched in GO terms related to biological development and differentiation, ion transport and ion channel activity, and protein receptor activity; the hypomethylated genes induced by DBS were related to GO terms such as protein binding/activity, cytoskeleton, and metabolic process (Figure 4c–e and Supplementary Table S3). These findings indicate that m⁶A-mRNA methylation is relatively high in the VTA and increases during DBS treatment.

Figure 4

Nucleus accumbens-deep brain stimulation (NAc-DBS) induces hypermethylated m⁶A modifications in the ventral tegmental area (VTA) of chronic unpredictable mild stress (CUMS) mice. (a) Heatmap showing the protein-coding genes for which m⁶A-mRNA modification was changed by at least 1.5-fold in CUMS mice subjected to DBS (CUMS_DBS) compared with control CUMS mice not treated with DBS (CUMS_Sham). (b) Pie chart representing overall changes in m⁶A-mRNA modifications in CUMS_DBS mice compared with CUMS_Sham mice (hypermethylation: FC > 1.5; hypomethylation: FC < 0.67). (c) Gene ontology (GO) analysis of differentially m⁶A-methylated mRNAs in CUMS_DBS mice compared with CUMS_Sham mice (hypermethylation: FC > 1.5; hypomethylation: FC < 0.67; BP: biological process; CC: cellular component; MF: molecular function). (d) and (e) GO analysis of m⁶A modifications involving (d) hypermethylation or (e) hypomethylation in CUMS_DBS mice compared with CUMS_Sham mice (hypermethylation: FC > 1.5; hypomethylation: FC < 0.67).

3.3 NAc-DBS dynamically reverses the gene expression and m⁶A-mRNA modification induced by CUMS

To understand the potential for DBS to reverse the effects of CUMS, it is necessary to screen for genes that are concurrently modulated by CUMS and DBS at the gene expression and m⁶A modification levels: screening for these genes will further our understanding of CUMS pathogenesis and the mechanism of DBS. At the gene expression level, 1,118 genes that were upregulated by CUMS were downregulated by DBS, while 3,702 genes that were downregulated by CUMS were upregulated by DBS. At the level of m⁶A modification, 254 genes that were hypermethylated during CUMS showed the reverse effect with DBS, while 75 genes that were hypomethylated during CUMS were hypermethylated by DBS. GO analysis of genes that were dynamically regulated by both CUMS and DBS revealed enrichment for the GO terms related to molecular binding function, intracellular part, organism development and biological regulation, biosynthesis and metabolic processes, and response to stimulus. In relation to m⁶A modification, dynamically changed genes were enriched with the GO terms MAPKK binding, transcription factor complex, biological development, and morphogenesis (Figure 5a–d and Supplementary Tables S6, S7).

Figure 5

Nucleus accumbens-deep brain stimulation (NAc-DBS) dynamically reverses gene expression and m⁶A-mRNA modification induced by chronic unpredictable mild stress (CUMS). (a–d) Left column: genes can be modulated by CUMS and reversed by DBS at the gene expression and m⁶A modification levels, respectively (up/hyper: FC > 1.5; down/hypo: FC < 0.67); right column: corresponding Go analysis related to data in the left column. (e) Genes with expression levels that do not change substantially in each group (GE: 0.67 < FC < 1.5; GE: gene expression) but for which m⁶A modification is dynamically regulated by DBS (m⁶A-hyper: FC > 1.5; m⁶A-hypo: FC < 0.67).

Interestingly, some genes for which gene expression levels did not change significantly between the sham- and DBS-treated CUMS mice still showed dynamic m⁶A modification regulation with DBS. Specifically, 150 of these genes with m⁶A modification changes were first regulated by CUMS (131 hypermethylated genes and 19 hypomethylated genes) and then regulated in a reverse manner by DBS (Figure 5e and Supplementary Table S6). Given the dynamic genes changes elicited by DBS, the underlying molecular mechanism of the treatment could be explored further at the RNA transcriptomic and epigenomic levels.

3.4 Correlation analysis of gene methylation and gene expression in the VTA

It has been reported that epigenetic changes, such as m⁶A-mRNA modifications, greatly affect gene expression [34,35,36]. Thus, we performed correlation analysis of gene methylation and gene expression in CUMS mice with or without DBS treatment. We found that among the m⁶A modifications that led to upregulation of gene expression, 34.8 and 35.1% of the genes were modified by hypermethylation and hypomethylation, respectively; conversely, the number of genes downregulated by hypermethylation and hypomethylation was 22.7 and 7.4%, respectively (Figure 6a). In comparison to WT results, CUMS mice that did not receive DBS treatment showed 85.7 and 12.2% downregulation and upregulation of hypermethylated genes, respectively; among the remaining 2% of total genes that were hypomethylated, 0.6 and 1.4% were downregulated and upregulated, respectively (Figure 6b). These results indicate that a linear relationship did not exist between RNA methylation and gene expression in the VTA, and this correlation may depend on different forms of external stimuli (Supplementary Tables S2, S5, and S8).

Figure 6

Correlation analysis of gene methylation and gene expression in the ventral tegmental area (VTA). (a and b) Methylation associated with gene expression in (a) chronic unpredictable mild stress (CUMS) mice treated with deep brain stimulation (DBS) and (b) CUMS mice not treated with DBS. Green dots represent hypermethylation genes associated with downregulated gene expression, red dots represent hypermethylation genes associated with upregulated gene expression, blue dots represent hypomethylation genes associated with downregulated gene expression, and purple dots represent hypomethylation genes associated with upregulated gene expression (m⁶A-hyper/GE-up: FC > 1.5; m⁶A-hypo/GE-down: FC < 0.67; GE: gene expression). (c and d) Genes for which the overall expression remained the same but m⁶A changed in (c) CUMS mice treated with DBS and (d) CUMS mice not treated with DBS. The blue column represents the proportion of genes up/downregulated that were hypermethylated (CUMS_DBS vs CUMS_Sham: 228; CUMS_Sham vs WT: 2535), the red column represents the proportion of genes for which overall expression remained the same but that were hypermethylated (CUMS_DBS vs CUMS_Sham: 439; CUMS_Sham vs WT: 4,982), the yellow column represents the proportion of genes up/downregulated that were hypomethylated (CUMS_DBS vs CUMS_Sham: 190; CUMS_Sham vs WT: 56), and the green column represents the proportion of genes for which overall expression remained the same but that were hypomethylated (CUMS_DBS vs CUMS_Sham: 204; CUMS_Sham vs WT: 32) (GE: 0.67 < FC < 1.5; m⁶A-hyper/GE-up: FC > 1.5; m⁶A-hypo/GE-down: FC < 0.67; GE: gene expression).

Gene screening also revealed some genes for which overall expression remained the same but m⁶A changed. Compared with CUMS mice that did not receive DBS, ∼1.0% (439/42,655) and ∼0.48% (204/42,655) of changes involved hypermethylation and hypomethylation in DBS-treated mice (Figure 6c). The hypermethylated genes were related to transcription factors (i.e., Tcfl5), neuronal activity, and post-synaptic function (i.e., Dnah6 and Fgf23), neurotransmitter receptors and transporters (i.e., Gabra5 and Slc7a9), and m⁶A methyltransferase and readers (i.e., Mettl18 and Hnrnph1); the hypomethylated genes were related to transcription factors (i.e., Tfe3), synaptic function (i.e., Syt8), cholinergic receptor (i.e., Chrna1), and transporters (i.e., Slc7a2) (Figure 6c). In comparison to WT mice, CUMS mice showed ∼11.7% (4,982/42,655) changes involving hypermethylation, with genes that were related to transcription factors (i.e., Tcf4), signal transduction (i.e., Stap2), neuronal activity and synaptic function (i.e., Adora1, Dnah3, Ppp1r8, and Snap23), neurotransmitter receptors and transporters (i.e., Gabra3, Gabrr2, Grin2c, Grin3b, Grm1, Htr4, Htr7, Drd4, and Slc6a8), methyltransferase and demethylases complexes (i.e., Mettl3, Wtap, Rbm5, and Alkbh4), and m⁶A readers (i.e., Hnrnpc); in contrast, only ∼0.08% (32/42,655) of changes involving hypomethylation was detected in CUMS mice (Figure 6d and Supplementary Table S8). These data suggest that the m⁶A methylation of some genes does not affect their gene expression in the VTA, and this process is also apparently independent of external stimuli.

3.5 NAc-DBS alters neural excitability by modulating neuronal signaling pathways

Given the numerous changes to gene expression and m⁶A modification reported earlier, we also determined the related signal pathways that potentially contribute to the long-term effects of DBS. Our results suggested that in the dopaminergic VTA/substantia nigra/caudate putamen/NAc circuitry [37,38,39,40], genes such as Drd2, Drd4, Pde1b, Kcnj6, Syn1, Syn2, and Syn3 have been implicated in dopaminergic signaling [32], and some of these genes were affected by CUMS at the gene expression level, with these effects being reversed by DBS (Supplementary Tables S1, S4, and S9). However, gene expression in the cocaine addiction-related pathway (i.e., expression of CREB3L4, DLG4, GNAI2, GPSM1, GRIN2C, GRM2, NFKB1, and PPP1R1B in mmu05030, pathway identifier used in KEGG) was downregulated in CUMS mice treated with DBS (Supplementary Figure S6). Other neuronal signaling pathways, including glutamate signaling (i.e., Grin1, Grin2c, and Grin3b), serotonin signaling (i.e., Htr4 and Htr7), and GABA signaling (i.e., Gabra3, Gabra4, Gabbr1, Gabbr2, and Gabrb2), were affected by DBS at the gene expression level; all these are involved in the regulation of the excitability of neural circuits. However, at the methylation level, only GABA signaling-related genes were affected (i.e., Gabra4 and Gabra5) (Supplementary Tables S1 and S4). At the gene expression level, GABA signaling was positively correlated with the morphine addiction-related pathway (i.e., expression of ADCY4, ARRB1, GABBR1, GABBR2, GABRA3, GABRA4, GABRB2, GABRG2, GNAS, GNB3, GNG10, GNG11, GNG13, GNG2, GNG5, GNG7, GNG8, GRK, GRK5, KCNJ6, KCNJ9, PDE10A, PDE1A, PDE1B, PDE4B, and PDE7B in mmu05032, the pathway identifier used in KEGG), whose genes were upregulated by DBS in CUMS mice (Figure 7 and Supplementary Tables S1 and S9).

Figure 7

Nucleus accumbens-deep brain stimulation (NAc-DBS) alters neural excitability by modulating neuronal signaling pathways. Screened differentially expressed genes were enriched in the morphine addiction-reference pathway. Orange marked nodes were associated with upregulated genes (FC > 1.5), whereas green nodes had no significance (0.67 < FC < 1.5) (http://www.genome.jp/kegg-bin/show_pathway?mmu05032).