Animals

Virgin female (7-8 weeks old, N=3 per diet group) C57BL/6 mice (Taconic, Ejby, Denmark) were mated with C57BL/6 male mice. Following observation of a vaginal plug (gestation day 0), the mice were randomized into four different diet groups: Normal protein (20% casein; NP; Hope Farms 4400.00, Woerden, NL) or low protein (8% casein; LP; Hope Farms 4400.01), both with no detectable taurine, with or without 1% (w/v) taurine in the drinking water (tau, i.e. NP+tau or LP+tau) (synthetic taurine; Sigma-Aldrich, St. Louis, MO, USA).

The four different diet groups were considered largely isocaloric since taurine supplementation only marginally contributes to the calorie intake as taurine is not metabolized and cannot make up for the decreased intake of essential amino acids caused by the LP diet. Mice were kept in a 12-hour light/dark cycle. At day 19 the mice gave birth and newborn pups were weighed, killed by decapitation and liver and hind leg skeletal muscle quickly and quantitatively removed, weighed, quick frozen in liquid nitrogen and stored at -80 °C for further analysis. For the newborn mice, a total of 25 pups from 12 different dams were analyzed, with 5≤n≤7 pups per group, all groups being a mixture of male and female pups. For examination of an effect of taurine upon NP offspring, 3 animals from NP and NP+tau were sacrificed at 4 weeks of age, followed by decapitation, quick removal of liver and quadriceps muscle. Tissue was quick frozen in liquid nitrogen and stored at -80 °C until further analysis. All experimental procedures were approved by The National Committee on Animal Experimentation, Denmark and by the local animal facility at the University of Copenhagen, Denmark.

RNA preparation

RNA was extracted from muscle and liver tissue using Trizol (Invitrogen, Carlsbad, CA, USA) as described by the manufacturer. In brief, tissue was homogenized in Trizol with a 5 mm steel bead, using a Qiagen Tissuelyzer (Qiagen). Upon lysis, RNA was extracted with chloroform and precipitated using isopropanol. Finally, RNA was subjected to cleanup using RNeasy columns (Qiagen). RNA quality was assured using a Bioanalyzer (Agilent, Santa Clara, CA, USA).

Microarray experiment

Three randomly selected samples of RNA from each of the four dietary groups as well as from the 4-week-old NP and NP+tau groups were analyzed at the RH Microarray Center (Rigshospitalet, Copenhagen, Denmark). RNA labelling, Affymetrix Mouse 430 2.0 array (Affymetrix Inc., Santa Clara, CA, USA) hybridization, and scanning were performed according to the manufacturer’s instructions. Each pup used was from a different dam, with liver and muscle being analyzed from the same animal. It was ensured that all groups constituted either one male and two females or one female and two male pups. Arrays from liver and muscle were processed and analyzed separately as described below. Samples from the 4-week-old animals were also analyzed separately.

Raw CEL files from all groups were normalized using quantile normalization, and log2 probe expression values were obtained using RMA in one step using justRMA (R 2.7.0 and Bioconductor 2.2 [21]). Control probes and probes not present on at least one microarray (according to MAS5 absent present calls) were removed before further analysis. For statistical analysis of the remaining probes a strategy of pre-selecting probes for subsequent 2-way analysis of variance (ANOVA) was adopted. Using significance analysis of microarrays (SAM) [22], we selected genes that were different between the following pairs of groups with a false discovery rate (FDR) of less than 10%: NP vs NP+tau, NP vs. LP, NP+tau vs. LP+tau, and LP vs. LP+tau. The resulting list of significantly changed probes were then analyzed using a 2-way ANOVA model: log2(probe expression level) = protein taurine protein*taurine, with p<0.05 considered significant. Due to the previous SAM pre-selection, no adjustment for multiple testing was done. Finally, this list of probes was subjected to a student’s t-test examining whether or not there was a significant (p<0.05) difference between NP and LP. In case of multiple probes for the same gene, the probe with the best rescue effect (see below) was chosen, and in the case of multiple probes having the same rescue effect the probe with the highest average expression level in NP was chosen. The resulting list was annotated using Bioconductor 2.2 and manually curated, and genes considered unknown (expressed sequence tags, putative genes, RIKEN cDNAs etc.) were removed.

Changes in gene expression levels caused by the maternal LP diet were considered fully rescued by taurine when the following criteria assessed by students’ t-tests were met: 1) The gene expression level of LP and LP+tau must be significantly different (p<0.05), and 2) the gene expression level between NP and LP+tau must not be significantly different (p>0.05). Likewise, a gene was considered partially rescued if: 1) The gene expression level between LP and LP+tau was significantly different (p<0.05) and 2) LP+tau confer a change in the opposite direction of LP compared to NP.

No differences in gene expression were found when directly comparing NP or NP+tau diet offspring with LP+tau offspring using SAM (FDR<10%, data not shown) in both liver and skeletal muscle. Others have used this as proof of a total phenotype reversal [23], however the above more fine-grained analysis show only a partial rescue effect of taurine.

Cluster analyses were carried out on log2 expression values using dChip [24], clustering both samples and genes, standardized to row means, with precalculated correlation distances using average linkage. Over-representation of specific gene sets among the genes significantly changed by an LP diet, were estimated using DAVID 2008 [25] with probe ids for the unique genes as input compared against a list of probe ids for known unique genes without control genes and probes without a valid EntrezGene id as background. Fishers exact test was used to examine if taurine selectively rescued specific gene sets rather than having a general rescue effect on all genes. Gene annotations were carried out using information from DAVID 2008 and Genecards (http://www.genecards.org) [26]. Mitochondrial genes were assessed as genes with a GO cellular compartment term containing either mitochondrial or mitochondrion. All statistical analyses on microarray data were performed using R 2.7.0.

In order to detect changes in gene expression caused by gestational taurine supplementation in 4-week-old offspring (NP+tau vs. NP), 4-week samples were normalized by tissue as above and probes not present on at least one microarray (according to MAS5 absent present calls) were removed. Following this, a gene set enrichment analysis (GSEA) [27] was performed with the GO cellular compartments (CC) and KEGG genesets using standard parameters. Furthermore, a subset probeset consisting of mitochondrial genes (GO CC containing mitochon*) were subjected to a SAM analysis examining differences between NP and NP+tau in both 0- and 4- week animals, with an FDR of less than 10% considered significant. The data discussed in this publication have been deposited in NCBIs Gene Expression Omnibus(GEO, http://www.ncbi.nlm.nih.gov/geo) [28] and are accessible through GEO series accession number GSE12730 (newborn samples) and GSE20577 (4 week samples).

Biochemical measurements

A 5% (wt/vol) homogenate of skeletal muscle tissue was prepared in a 2-ml Potter-Elvehjem glass-teflon homogenizer for 1 min at 0 °C in a buffer containing 25 mM glycyl-glycin, 150 mM KCl, 5 mM MgSO4, 5 mM EDTA, 1mM DTT, 0.02% BSA, and 0.1% Triton X-100, pH 7.5. The liver homogenate was prepared in the same buffer applying a Qiagen Tissuelyzer (Qiagen, Valencia, CA, USA) at 0 °C employing ice-cold racks, homogenizing for 3 times 1 min at 30 Hz. The homogenate was frozen in liquid nitrogen, thawed on ice, whirl mixed, and centrifuged at 22,000 g for 2 min at 4 °C. The supernatant was stored at -80 °C for determination of enzyme activities and protein concentration. Protein content was measured using BSA (fraction V) as standard [29]. Citrate synthase (CS, EC 4.1.3.7) activity was determined as described previously [30] and expressed relative to the total protein content as units per mg protein.

Quantitative real-time RT-PCR

Reverse transcription (RT) reactions were performed using random hexamers on 2 µg RNA using an RT kit (Applied Biosystems, Foster City, CA) in a reaction volume of 100 µl. The resulting cDNA product was stored at -20 °C until further analysis. The primers and probes for all genes and the 18S rRNA endogenous control were pre-developed TaqMan probes and primer sets from Applied Biosystems: ACSS2 (acyl-CoA synthetase short-chain family member 2; Mm00480101_m1), COX7A1 (cytochrome c oxidase, subunit VIIa 1; Mm00438296_m1), COX7C (cytochrome c oxidase, subunit VIIc; Mm01340476_m1), CS (citrate synthase; Mm00466043_m1), EGFR (epidermal growth factor receptor; Mm00433023_m1), FST (Follistatin; Mm00514982_m1), GLDC (glycine decarboxylase; Mm00506891_m1), GPD2 (glycerol phosphate dehydrogenase 2, mitochondrial; Mm00439082_m1), MH1 (myosin, heavy polypeptide 1, skeletal muscle, adult; Mm01332489_m1), MYBPC2 (myosin binding protein C, fast-type; Mm00525419_m1), PDK4 (pyruvate dehydrogenase kinase, isoenzyme 4; Mm00443325_m1), SPP1 (secreted phosphoprotein 1; Mm00436767_m1), UCP3 (uncoupling protein 3 (mitochondrial, proton carrier); Mm00494074_m1) PGC-1α (peroxisome proliferator receptor γ coactivator 1; Mm00447183_m1). All assay reagents were from Applied Biosystems. The mRNA levels of all genes and the endogenous control, 18S rRNA, were determined by real time RT-PCR using an ABI PRISM 7900 sequence detector (Applied Biosystems). The amplification mixtures were amplified according to standard conditions using 50 cycles in a 10 µl volume in triplicate. The relative mRNA content of both the target and the endogenous control gene was calculated from the cycle threshold values by using a standard curve constructed from a serial dilution of aliquots of cDNA pooled from all the samples. The relative expression levels of all genes were determined by normalization to the endogenous control, 18S rRNA, which was found not to differ between diet groups.

Statistics

Birthweight, liver and muscle weight as well as all enzyme activity data exhibited a normal distribution, and data are presented as means ± standard error of the mean (SEM). All microarray and mRNA data were log-transformed in order to obtain a normal distribution, and data are presented as geometric means with 95% confidence intervals. Statistical analyses were carried out using a 2-way ANOVA with Bonferroni corrected students’ t-tests as post hoc tests. All statistical analyses were performed using SAS 9.1.2 (The SAS Institute), except for the microarray analysis which was performed as described above. A p value < 0.05 was considered significant.

Birthweight and citrate synthase enzyme activity in liver and muscle of newborn mice

As a measure of the mitochondrial fraction of tissue mass, citrate synthase (CS) specific enzyme activity and mRNA level [31] was determined in liver and skeletal muscle of the newborn pups. There was no difference between diet groups in CS enzyme specific activities (Table 1) or mRNA levels (see Additional file 1).

Taurine partially rescues the effects of gestational protein restriction on gene expression levels in liver and skeletal muscle of newborn mice

In liver, the expression of a total of 2012 non-redundant transcripts, all encoding known genes, were significantly changed by the maternal LP diet, while a somewhat smaller number (967) of non-redundant transcripts, all encoding known genes, were changed in skeletal muscle (Table 2). Of the 2012 genes affected in the liver, 667 transcripts were up- and 1345 down-regulated. Similarly, in skeletal muscle; of the 967 genes affected, 312 were up- and 655 transcripts were down-regulated compared with a maternal NP diet (Table 2).

A number of significantly changed transcripts and one not significantly changed (CS) were randomly selected for RT-PCR validation of the microarray data. Positive validation was obtained for 7 out of 8 and 6 out of 9 transcripts encoding known genes in liver and skeletal muscle, respectively (see Additional file 1).

Taurine supplementation prevented a major part of the gene expression changes seen with the maternal LP diet (Table 2). Taurine had a rescuing effect on 30% (600 out of 2012) of the changed transcripts of known genes in liver and on 46% (444 out of 967) in skeletal muscle (Table 2), rendering the taurine effect more pronounced on muscle than on liver (p<10^-16, Fishers exact test).

When genes were grouped according to function (figure 1A) (see Additional file 2), a large difference was seen between up and down-regulated genes involved in energy metabolism (figure 1A) in both liver and skeletal muscle., albeit in a different direction in the two tissues. Likewise, when mitochondrial genes were grouped according to function, a clear difference between liver and skeletal muscle could be seen (figure 1B) (see Additional file 3 and 4). The rescuing effect of taurine was significantly higher on genes involved in fatty acid metabolism in liver compared to other gene sets (Table 3), whereas in skeletal muscle this rescuing effect was present only for genes involved in oxidative phosphorylation and the TCA cycle (Table 4).

Gestational taurine supplementation has a lasting effect

Birthweight

The 40% decrease in birthweight of LP offspring observed in this study is somewhat larger than previously reported in the same mouse strain (7% and 28%) [32]. We observed no difference in the ratio of organ weight:total body weight, which seems to disagree with a previous study in rats [33], and we saw a large rescue effect of taurine upon birthweight, in disagreement with a similar study in rats [19]. However, these disagreements likely reflect species or diet composition differences.

Changes in global gene expression patterns in liver and skeletal muscle

Genes involved in oxidative phosphorylation and other mitochondrial genes were the most over-represented in the genes changed in the LP offspring (Table 3 and 4). This suggests that dysregulation of energy metabolism is a key component in the development of the low birthweight phenotype. Furthermore, as CS activity may be considered a measure of mitochondrial mass, the lack of a change in CS activity suggests this dysregulation may happen without a change in mitochondrial mass. However, this seem to be at variance with Park et al. [34], who reported a decreased mtDNA:gDNA ratio in both liver and skeletal muscle in 5-week-old offspring of maternal low protein rats.

In liver, a large fraction of genes involved in amino acid metabolism, protein synthesis, TCA cycle and energy metabolism genes involved in both Complex I-IV and ATP synthesis showed an increased expression, whereas in skeletal muscle the opposite effect was observed. The concentration of taurine in skeletal muscle is ~6 fold higher than in the liver in mice [35], suggesting a different taurine requirement of the two tissues. It should also be noted that a decrease in taurine availability is known to occur under LP conditions [36], and that taurine is a constituent of mitochondrial tRNA [13] and may thus be required for normal mitochondrial function.

Downregulation of oxidative phosphorylation in skeletal muscle has in several cases been linked to development of type 2 diabetes and insulin resistance [37, 38]. Although the exact mechanism is unknown, a decrease in the number of skeletal muscle mitochondria and/or in the activity of oxidative phosphorylation as well as mitochondrial dysfunction at birth may, if it is permanent, be associated with a decreased capacity for β-oxidation capacity, which may lead to increased intracellular lipid content and subsequent disruption of insulin signaling [39]. However, this hypothesis has recently been questioned [40].

Interestingly, in humans birthweight is positively associated with the expression of PGC-1α in skeletal muscle in adults [6], but despite PGC-1α controlling mitochondrial biogenesis [41], no concomitant difference in mitochondrial gene expression and oxidative phosphorylation was observed in adults with a low birthweight [6]. However, the decrease in mitochondrial gene expression seen in this study and not in adult humans may reflect the difference between newborns and adults or species difference.

Partial phenotype rescue by taurine supplementation

Taurine has previously been shown to rescue the adverse effects of a maternal LP diet on pancreatic function in the offspring [19, 20, 23, 42]. We find a large rescuing effect on gene expression profiles in both liver and muscle by taurine supplementation, with significantly more mitochondrial genes rescued in skeletal muscle than in liver, suggesting taurine as an important factor in the control of mitochondrial gene expression. Furthermore, we observed that the rescuing effect was tissue specific, as genes of fatty acid metabolism in the liver and genes involved in oxidative phosphorylation and TCA cycle in skeletal muscle were preferentially rescued compared with other genes (Table 3 and 4). A similar study examining the effect of maternal LP diet and taurine on islet gene expression also found decreased expression of respiration and TCA cycle genes, changes which were rescued fully by taurine supplementation [23]. Combined with our results, this strongly suggests that the effect exerted by taurine is mediated by a mitochondrial mechanism.

In this context it should be noted that plasma taurine concentration has been suggested to be a marker of fetal well being and a prerequisite for normal fetal development [43]. This is corroborated by the observation that lack of the taurine transporter (TAUT) confers a large decrease in exercise capacity [14], which may also be related to the observation that TAUT expression increases during myogenesis and that taurine is able to protect against dexamethasone induced atrophy [44]. Thus, a picture of taurine as a necessary expression factor in myogenesis, and perhaps in mitochondrial biogenesis as well, emerges.

In liver a deficiency of taurine, as seen in the TAUT knockout mouse, causes an increase in hepatic apoptosis and inflammation, as well as a decrease in mitochondrial coupling of oxidative phosphorylation [45]. Also, taurine has been shown to have an anti-apoptotic effect upon the liver and to normalize tamoxifen induced mitochondrial dysfunction in rats [46]. There are several reports on the effect of a maternal low protein diet on offspring DNA methylation patterns [47, 48], but there is no evidence of taurine involvement in this process. Interestingly, taurine has been shown to decrease N-methylation in rat heart [49].

Taurine supplementation during gestation alone, with no other dietary manipulations of the maternal diet, resulted in minute changes to mitochondrial gene expression in both skeletal muscle and liver, primarily in lipid metabolism and protein translation. Others have shown both beneficial [50] and detrimental effects [51] of gestational taurine supplementation upon the adult metabolic phenotype, however no mechanism for these changes have been identified. The observation that gestational taurine supplementation also in the control group has an effect on mitochondrial gene expression and lipid metabolism corroborates the notion of taurine as being important for development of normal mitochondrial function, a field of research which needs more attention.