Supplementary Online Material
Supplementary Table 1. Control analysis of agonal-pH states shows the number of sensitive genes in each pathway. Using a loose criterion of ±1.2 fold change there were a total of 1,609 genes dysregulated in Anterior Cingulate Cortex (AnCg) and 1,778 genes dysregulated in DLPFC. Restricting the fold change to ± 1.25 and the top 5% ranked T values for differential expression, the number of dysregulated genes was reduced in AnCg to 356 and in the DLPFC to 341. A high concordance of dysregulated genes between cortical regions was also shown for apoptosis, chaperone, proteasome, and reactive oxygen stress classifications.
Pathway / Region / Pathway Total / ± 1.2 Fold Change Number of Genes / Common / Average % / Direction of Change In Gene Expression (Low pH Compared to High pH Control) / Pathway Induction In Relation To Decrease pH / Repressed (R)Induced (I)
Down / Up
Mitochondria / AnCg / 321 / 108 / 88 / 32.7 / 91 / 17 / -5.3 / R
DLPFC / 103 / 87 / 16 / -5.4 / R
Apoptosis / AnCg / 293 / 59 / 47 / 20.3 / 23 / 36 / 1.5 / I
DLPFC / 60 / 26 / 34 / 1.3 / I
Chaperone / AnCg / 100 / 17 / 12 / 16.5 / 11 / 6 / -1.8 / R
DLPFC / 16 / 12 / 4 / -3.0 / R
Proteasome / AnCg / 32 / 13 / 9 / 42.2 / 12 / 1 / -12.0 / R
DLPFC / 14 / 12 / 2 / -6.0 / R
Reactive Oxygen Stress / AnCg / 25 / 8 / 7 / 30.0 / 3 / 5 / 1.6 / I
DLPFC / 7 / 3 / 4 / 1.3 / I
Total / 771 / 28.3
# Final percentage is an average of the (AnCg and DLPFC) changed number of genes divided by the total number in the pathway.
Supplementary Table 2. The agonal-pH control genes that are dysregulated in one or more brain regions (AnCg, DLPFC, Cerebellum) are shown. The fold change and T-value ranking for each brain region is shown. Considering all agonal-pH sensitive genes in control group analysis only, i.e. there were 570 genes that were dysregulated across two or more brain regions (DLPFC, Anterior Cingulate, or cerebellum) meeting a fold change criteria of ± 1.25 and in the top 5% ranked T values. The complete Excel table can be obtained from our website
http://www.pritzkerneuropsych.org/data/data/File022206.aspx
Supplementary Table 3A. The 225 agonal-pH genes that were differentially expressed in controls for 3 brain regions were analysed by an over-representation algorithm with EASE software (Hosack et al., 2003). Significant p-values following Bonferroni correction for multiple comparisons are shown. The over-representation threshold for differentially expressed genes was: significant in 3 brain regions, a fold change ±1.25, and in the top 5% T-values.
hydrogen ion transporter activity / 25 / 3.68E-16
monovalent inorganic cation transporter activity / 25 / 3.26E-15
cation transporter activity / 28 / 8.90E-14
ribonucleotide biosynthesis / 13 / 6.91E-07
nucleoside phosphate metabolism / 10 / 1.41E-06
ATP biosynthesis / 10 / 1.41E-06
proton transport / 13 / 1.51E-06
ribonucleotide metabolism / 13 / 1.93E-06
primary active transporter activity / 16 / 3.35E-06
purine ribonucleotide biosynthesis / 12 / 3.57E-06
ATP metabolism / 10 / 5.08E-06
hydrogen transport / 13 / 7.57E-06
purine ribonucleotide metabolism / 12 / 7.77E-06
purine nucleotide biosynthesis / 12 / 9.93E-06
purine nucleotide metabolism / 12 / 2.00E-05
group transfer coenzyme metabolism / 10 / 2.14E-05
ion transport / 30 / 2.33E-05
monovalent inorganic cation transport / 20 / 6.04E-05
energy pathways / 20 / 7.87E-05
transporter activity / 55 / 1.43E-04
coenzyme biosynthesis / 10 / 4.17E-04
cation transport / 23 / 4.28E-04
nucleotide biosynthesis / 13 / 6.85E-04
cofactor biosynthesis / 11 / 4.00E-03
Mitochondrion / 29 / 5.45E-03
Supplementary Table 3B. The 570 agonal-pH genes that were differentially expressed in controls for 2 or more brain regions were analysed by an over-representation algorithm with EASE software (Hosack et al., 2003). Significant p-values following Bonferroni correction for multiple comparisons are shown. The over-representation threshold for differentially expressed genes was:
significant in 2 or more brain regions, a fold change ±1.25, and in the top 5% T-values. This list contains the same genes as Table 3A and additional genes that were significant in only 2 brain regions.
Supplementary Table 4. The number of mitochondrial Unigene clusters that were significantly correlated with pH and age. The numbers of significant correlations are shown for each brain region using 30 subjects from microarray analysis #3.
Number of Significant Correlationsa With nDNA Mitochondrial Gene ExpressionbBrain Region / pHc / Agec
With Agonal-pH subjects / Without Agonal-pH subjects / With Agonal-pH subjects / Without Agonal-pH subjects
DLPFC / 189 / 6 / 66 / 104
Anterior Cingulate / 201 / 4 / 56 / 24
Cerebellum / 171 / 4 / 29 / 94
a Number of correlations significant at p < 0.05 level.
b A total of 321 non-redundant transcripts.
c.Age and pH were not significantly correlated (r = -0.17 , p = 0.35)
Supplementary Table 5. Significant correlations of mitochondrial related gene expression in the DLPFC with pH and age. Shown are examples with significant negative correlation, no correlation, and significant positive correlation with gene expression and pH or age.
Gene Symbol / Correlation(s) of Gene Expression With / RefSeq SummarypH / Age
ATP6V0E
/ -0.64 / 0.45 / Component of vacuolar ATPase (V-ATPase) that mediates acidification of eukaryotic intracellular organelles.ACADM
/ 0.036 / 0.48 / Medium-chain specific (C4 to C12 straight chain) acyl-Coenzyme A dehydrogenase. The homotetramer enzyme catalyzes the initial step of the mitochondrial fatty acid beta-oxidation pathway.LRPPRC
/ 0.70 / -0.15 / A single candidate gene, LRPPRC, was identified and shown to be the causative gene underlying Leigh syndrome, French-Canadian type (LSFC). This syndrome has fatal metabolic acidosis as one consequence and localizes to mitochondria.COQ7
/ -0.01 / -0.66 / Similar to a mitochondrial di-iron containing hydroxylase in Saccharomyces cerevisiae that is involved with ubiquinone biosynthesis. Mutations in the yeast gene lead to slower development and longer life span.BZRP
/ -0.46 / 0.55 / Present mainly in the mitochondrial compartment of peripheral tissues, interacts with some benzodiazepines and appears to be a key factor in the flow of cholesterol into mitochondria to permit the initiation of steroid hormone synthesis.MUT
/ 0.06 / 0.52 / MUT is a vitamin B12-dependent mitochondrial enzyme which catalyzes the isomerization of methylmalonyl-CoA to succinyl-CoA.
Supplementary Table 6. Mitochondrial genes from microarray analysis #3. Mitochondrial genes that are ranked in the top 5% T values of all genes for BPD, MDD, or control agonal-pH are shaded gray. The top 5% T values in BPD DLPFC are sorted by ascending rank (1, 2, 3, … 8846). The control analysis ranks are shown to the right for three brain regions.
Supplementary Table 7.
Proteasome genes from microarray analysis #3 that ranked in the top 5% T values for BPD, MDD, or control agonal-pH are shaded gray. The control analysis ranks are shown to the right for three brain regions. From a total of 32 proteasome genes, nine were found in the top 5% dysregulated genes in BPD in the Anterior Cingulate cortex (Fishers exact test, p = 0.00005, corrected for multiple testing) and predominantly were upregulated. However, when a more stringent correction was used to remove the top 10% agonal-pH genes (rank < 881), the proteasome classification did not remain significant.
Supplementary Table 8.
Chaperone gene expression in mood disorder showing fold change and ranks from microarray analysis #3. Chaperone genes that ranked in the top 5% T values for BPD, MDD, or control agonal-pH are shaded gray. The control analysis ranks are shown to the right for three brain regions.
There were 100 probesets represented on the HGU95Av2 Affymetrix chip in the chaperone classification. In BPD AnCg there were 12 of 100 genes in the Top 5% (p = 0.02 corrected for multiple testing). However, when a more stringent correction was used to remove the top 10% agonal-pH genes (rank < 881), the chaperone classification did not remain significant.
Supplementary Table 9. Apoptotic gene expression in mood disorder showing fold change and ranks from microarray analysis #3. Apoptotic genes that ranked in the top 5% T values for BPD, MDD, or control agonal-pH are shaded gray. The control analysis ranks are shown to the right for three brain regions. There was not an over-representation of genes related to apoptosis in BPD or MDD. In the AnCg for BPD, 14 of 273 apoptotic and anti-apoptotic genes were in the top 5% of differentially expressed genes and 12 were up regulated. For MDD in the AnCg, nearly the opposite was found, 11 of 15 gene comparisons were down regulated in the top 5%.
Supplementary Table 10.
Response to oxidative stress gene expression in mood disorder showing fold change and ranks from microarray analysis #3. Response to oxidative stress genes that ranked in the top 5% T values for BPD, MDD, or control agonal-pH are shaded gray. The control analysis ranks are shown to the right for three brain regions. Response to oxidative stress was not significantly over-represented in the top 5% genes for BPD and MDD AnCg (2 genes out of 19 genes total in the category).
Supplementary Table 11. The genes from category c microarray analysis #3 are combined for mitochondrial, proteasome, chaperone, and reactive oxygen classifications that are displayed in Supplementary Tables 3-7. The top 5% T values for genes in BPD DLPFC are sorted by ascending rank (1, 2, 3, … 8846). The control analysis ranks are shown to the right for three brain regions. The category c genes shown in this table share codirectional fold changes with agonal-pH control comparisons.
Supplementary Table 12. Lithium prescription at the time of death and BPD subject’s gene expression for mitochondrial-related genes. The dysregulated genes shown in Table 6 were screened for change in BPD subjects without lithium prescription and subjects with lithium prescription. It should be noted that all BPD brains were negative for therapeutic levels of lithium. Thus, the pattern of dysregulation shows a possible gene expression alteration following recent use of lithium. HSPA5 appeared to be changed in both BPD groups, indicating a gene without agonal effect that remains dysregulated irrespective of lithium prescription.
Supplementary Table 13. SSRI prescription at the time of death and MDD subject’s gene expression for mitochondrial-related genes. The dysregulated genes shown in Table 6 were screened for change in MDD subjects without SSRI prescription and subjects with SSRI prescription. Thus, the pattern of dysregulation shows gene expression alteration following possible recent SSRI usage. HSPA2 was the most differentially expressed gene in both MDD groups (SSRI+ and SSRI-), and a gene without agonal effect.
Supplementary Table 14. The number of genes in category a and b by functional classification. Of 771 genes surveyed, about 4% are in category a and b.
Functional Classification
/ Number of Genes Dysregulated / Number of Genes In ClassificationMitochondria / 12 / 321
Apoptosis / 12 / 293
Chaperone / 5 / 100
Proteasome / 1 / 32
Reactive Oxygen Stress / 2 / 25
Total / 32 / 771
Supplementary Figure 1. The scatterplot of LRPPRC (leucine-rich PPR-motif containing) with pH (upper left) and age (upper right) shows a positive relationship between gene expression and pH but no relationship to age. There was no significant relationship for COQ7 with pH (lower left), however age and COQ7 (coenzyme Q7 homolog, ubiquinone (yeast)) were found to be negatively correlated (lower right).
Supplementary Figure 2. Mitochondrial related gene expression example (ABCB7) shows a significant correlation with pH even after removal of subjects with agonal factors and low pH, however there is no significant group difference between BPD or MDD and Controls. The mitochondria related gene HSPA2 shows no correlation with pH after removal of subjects with agonal factors and low pH, and shows significant differences between MDD and Control. Both examples are from Anterior Cingulate Cortex.
Supplementary Figure 3. Mitochondrial copy number was significantly altered in prolonged agonal duration cases compared to rapid death (n = 40). Mitochondrial copy numbers (MtDNA / nDNA copy number ratios) were compared among all subjects with rapid death to prolonged agonal duration cases classified according to the Westerly rating scale. The median mitochondrial DNA copy number increased in prolonged agonal duration compared to rapid death cases (p = 0.01) as shown in the Box plot.