Supplementary Information

Table S1: Primers for real-time PCR

Sequence
Mapk6 / Fw: TATCGATGAGGTGCAGCTTG
Rev: GTTCTCGTGGTGATCTGGGT
Cox5a / Fw: TGCCTGGGAATTGCGTAAAGGGATG
Rev: TCAAGGCCCAGCTCCTCTGGA
Ndufab1 / Fw: GGCCATGGAAGACGAATTTGGGTTT
Rev: TCCAGTCAGCAGCCACATCCA
Ndufs7 / Fw: GCGTGCTGTGCCGTGGAGAT
Rev: CGTACACCTTTCGGAGCGCGG

Table S2: Genes with differential expression corroborating findings from previous microarray studies

Gene symbol / Entrez Gene Name / SMIT KO / SMIT_Li / IMPA KO / IMPA_Li / Reference
Hcn1 / Hyperpolarization-activated, cyclic nucleotide-gated K+ 1 / p<0.033, +1.11 / p<0.063, +1.08 / NS / NS / (Brandish et al. 2005)
Gpr88 / G-protein coupled receptor 88 / p<0.065, -1.18 / p<0.004, -1.35 / NS / NS / (Brandish et al. 2005)
Atp5g1 / ATP synthase, H+ transporting, mitochondrial F0 complex / NS / NS / NS, +1.08 / p<0.03, +1.13 / (McQuillin et al. 2007)
Ptprk / Protein tyrosine phosphatase, receptor type, K / p<0.05, -1.1 / p<0.005, -1.17 / NS / NS / (Brandish et al. 2005; McQuillin et al. 2007; Seelan et al. 2008)
Gpld1 / Glycosylphosphatidylinositol specific phospholipase D1 / p<0.03, -1.18 / p<0.012, -1.2 / NS / NS / (McQuillin et al. 2007)
Pik3r3 / Phosphatidylinositol 3-kinase, regulatory subunit / p<0.07,
+1.12 / p<0.02, +1.17
/ p<0.07,
+1.12 / p<0.02, +1.17
/ -1.08 / -1.1 / McQuillin et al. 2007))
Bai3 / Brain-specific angiogenesis inhibitor 3 / NS
(-7%) / p<0.013
(-15%) / p<0.015
(-8%) / p<0.025
(-7%) / McQuillin et al. 2007))
Adcyap1 / Adenylate cyclase activating polypeptide 1 / NS, 1.35 / p<0.08, 1.49 / NS / NS / (Brandish et al. 2005)
Mt3 / Metallothionein 3 / p<0.05
-1.26 / p< 0.04
-1.26 / NS / NS / (Chetcuti et al. 2008)
Actb / Actin, beta / p<0.009
-1.1 / p<0.008
-1.1 / NS / NS / (Chetcuti et al. 2008)
Alpl / Alkaline phosphatase, liver/bone/kidney / p<0.011
-1.4 / p<0.045
-1.2 / NS / NS / (Seelan et al. 2008)

Table S3: No effect of rotenone treatment on general motor function assessed by the hanging wire and rotarod tests

Hanging wire (sec)

/

Rotaroad (sec)

Regular food+ vehicle (n=20) /

28±4

/

41±3

Lithium + vehicle (n=20)

/

36±5

/

43±2

Regular food + rotenone (n=18)

/

38±5

/

42±3

Lithium + rotenone (n=17) /

29±4

/

45±2

Two-way ANOVA analysis was used to test the effect of lithium and rotenone in the hanging wire and the rotarod tests. No significant effect of lithium or rotenone and no interaction was found in either of the tests. Hanging wire: p>0.90, p>0.68, p>0.07, respectively. Rotarod: p>0.77, p>.06, p>0.9, respectively

Table S4. Previous microarray studies of lithium treatment

Tissue / Treatment period
(dose) / Fold of change cut-off / Main finding / Reference
Rat brain / 7 days/42 days
(0.17% in chow) / 2 / Most of the genes affected following 42 days of treatment were not affected after 7 days of treatment. The affected genes were functionally related to cell signaling, inositol cycle, energy production and transcription. / (Bosetti et al. 2002)
Rat cortex slices / 24h, Li/Li+carbachol/ Li+carbachol+inositol
(Li – 2 mM,
carb – 50 µM,
inositol - 10 mM) / 1.3 both Li vs. Li+Carb.
and
Li+Carb. vs. Li+Carb.+Ino / Inositol depletion induced by carbachol and Li treatment increased expression of 29 genes. Out of these, three genes are related to increased dopamine signaling – Adcyap (encoding PACAP which is involved in regulation of biological clock), Gch, Pam. / (Brandish et al. 2005)
Mouse brain / 2 weeks
(0.4 % in chow) / No fold of change cut-off / 13 out of the differentially expressed genes were related to inositol metabolism. Expression changes were also found in genes related to thyroxin metabolism, myelin-related genes and genes coding for protein phosphatases. / (McQuillin et al. 2007)
Mouse brain / 7 days
(1.16 mg/kg i.p.) / 4 / 19 genes were differentially expressed following lithium treatment. The genes were functionally related to processes such as maintaining metal ion homeostasis and chemical/electrical gradients across membranes, regulation of RNA polymerase II (transcription), protein degradation, and G-protein-coupled signal transduction. / (Chetcuti et al. 2008)
Human neuronal cells / 33 days
(1.5 mM) / 1.1 / Functional annotation of differentially expressed genes highlighted MAPK pathway, regulation of cytoskeleton and calcium signaling. Genes most highly modulated by lithium, included those related to neuronal function, oxidative metabolism and cell cycle. / (Seelan et al. 2008)
Rat frontal cortex / 3 weeks
(50 mg/kg i.p.) / 1.5 / The differentially expressed genes were related to numerous pathways, mainly signal transduction and energy and metabolism. Expression of four genes previously associated with bipolar disorder was altered: Comt, Vapa, Pdk1, Dtnb / Fatemi et al, 2009)

None of the studies used statistical analysis of functional enrichment of the differentially expressed genes.

ber of genes included in a particular gene-set.

** FDR q-value – false discovery rate, adjusted to the gene-set size and multiple hypotheses testing

Fig. S1: Schematic representation of the study design(see legend in the next page)

A. The microarray part of the study. In green – animals, data and analyses related to the SMIT1 colony; in red – animals, data and analyses related to the IMPA1 colony; in black – analyses performed on the combined data obtained from both colonies.

B. The behavioral part of the study. Black arrows denote the beginning of rotenone (purple) and Li (light blue) treatment. Colored arrows denote the behavioral experiments; red – rotarod and hanging wire tests, blue – FST, orange – amphetamine-induced hyperlocomotion.

Fig. S2: Statistical analysisof the raw data of the microarrays

Three-way mixed model ANOVA carried out to analyze the combined effect of lithium in both colonies revealed that the major source of variation between the groups was the colony. Type=treatment (Li or regular food), batch=random effect determined by principal component analysis (PCA), colony=IMPA1 or SMIT1.

Fig. S3: Hierarchal clustering of all genes differentially expressed in Li-treated mice from either of the colonies or in one of the KOs

Two-way ANOVA was carried out separately in each colony to reveal genes the expression of which was affected by Li treatment or KO of either IMPA1 or Slc5a3 as compared to WT-untreated mice. 4,643 genes differentially expressed (p<0.05) by either Li treatment or gene KO in either of the two colonies were used for the analysis. Signals from both colonies were used for the calculation of Z-scores. The mice created two major clusters corresponding to the two colonies (IMPA1 and SMIT1). Highlighted in turquoise is the only gene cluster presenting similar expression pattern in the two colonies. Among the two sets of colored bars below the figure the upper represents the colony (blue – SMIT1, orange – IMPA1) and the lower - the group of mice (yellow – WT-untreated, pink – KO, grey – WT-Li treated).

Fig. S4: Functional annotation of the gene cluster showing similar expression pattern between the IMPA1 and SMIT1 colonies

Functional analysis of the gene cluster with similar expression in the two colonies (highlighted in turquoise in Fig. S2A) using DAVID software showed that most of the functional annotations of the genes composing the cluster are related to mitochondrial respiration.

Fig. S5: DAVID Functional annotation tool revealed enrichment in mitochondria-related genes among genes with differential expression in KO mice and in Li-treated mice of both colonies

Significantly enriched terms (p<0.05, FDR<0.05) within the genes with similar differential expression (p<0.0.5) in KO mice and in Li-treated WT mice compared to WT-untreated mice.

A. Annotation clustering of the 654 genes meeting the criteria in the SMIT1 colony.

B. Annotation clustering of the 159 genes meeting the criteria in the IMPA1 colony.


Fig. S6: Localization of the commonly differentially expressed genes among the different complexes of the respiratory chain

The oxidative phosphorylation pathway is depicted. Proteins encoded by nuclear genes in each complex are listed below the depiction.. Green boxes represent proteins existing in mice (similar to humans). Asterisks (green - SMIT1 colony, red - IMPA1 colony) denote the location of the proteins and the names of proteins encoded by genes with similar differential expression in KO mice and lithium-treated WT mice compared to their WT-untreated littermates. In both colonies, the commonly differentially expressed genes were located in all the respiratory complexes except for complex II.

Fig. S7: DAVID Functional annotation tool shows enrichment in mitochondrial terms among theupregulated genes in Li-treated mice and in KO mice in both colonies

Significantly enriched terms (p<0.05, FDR<0.05) within the genes significantly upregulated (p<0.0.5) in KO mice and in Li-treated WT mice compared to WT-untreated mice in (A) IMPA1 colony and (B) SMIT1 colony. No significantly (FDR<0.05) enriched genes were found among the downregulated genes in either of the colonies.

Fig. S9: Venn diagrams of genes with significantly altered expression

A. Upregulated genes. B. Downregulated genes

Fig. S10: Real-time PCR verification of selected mitochondria-related genes in the frontal cortex of IMPA1 KO, SMIT1 KO and Li-treated micein each of the colonies

The three genes studied showedmodest upregulation in the KO mice and in Li-treated mice, concordantwith their expression change observed in the microarray analysis.Yet, possibly due to the relatively small expression changes and the greater variability in the results obtained using real-time PCR, these differences reached statistical significance only for some of the groups and of the genes studied.

Results are means +/- S.E.M.

* p<0.05 compared to WT-untreated mice.

** p<0.01 compared to WT-untreated mice.

A.IMPA1 colony. One-way ANOVA: Cox5a, F2,28 =3.391, p<0.05; Ndufab1, F2,28=4.625, p<0.02; Ndufs7, F2,28=4.861, p<0.015.Post-hoc Fisher's LSD: Cox5a, Li-treated vs. WT-untreated, p<0.015; IMPA1 KO vs. WT-untreated, p<0.22. Ndufab1, Li-treated vs. WT-untreated, p<0.006; IMPA1 KO vs. WT-untreated, p<0.11. Ndufs7, Li-treated vs. WT-untreated, p<0.005; IMPA1 KO vs. WT-untreated, p<0.2).B.SMIT1 colony.Welch’s t-test: Cox5a, Li-treated vs. WT-untreated, p<0.12; SMIT1 KO vs. WT-untreated, p<0.14. Ndufab1, Li-treated vs. WT-untreated, p<0.49; SMIT1 KO vs. WT-untreated, p<0.023. Ndufs7, Li-treated vs. WT-untreated, p<0.38; SMIT1 KO vs. WT-untreated, p<0.156).

In both coloniesno significant differenceswere found between Li-treated and KO mice in the expression of neither of the genes.