R-α-lipoic acid does not reverse hepatic inflammation of aging, but lowers lipid anabolism, while accentuating circadian rhythm transcript profiles

Abstract

To determine the effects of age and lipoic acid supplementation on hepatic gene expression, we fed young (3 mo) and old (24 mo) male Fischer 344 rats a diet with or without 0.2% (wt/wt) R-α-lipoic acid (LA) for 2 wk. Total RNA isolated from liver tissue was analyzed by Affymetrix microarray to examine changes in transcriptional profiles. Results showed elevated proinflammatory gene expression in the aging liver and evidence for increased immune cell activation and tissue remodeling, together representing 45% of the age-related transcriptome changes. In addition, age-related increases in transcripts of genes related to fatty acid, triglyceride, and cholesterol synthesis, including acetyl-CoA carboxylase-β (Acacb) and fatty acid synthase (Fasn), were observed. Supplementation of old animals with LA did not reverse the necroinflammatory phenotype but, intriguingly, altered the expression of genes governing circadian rhythm. Most notably, Arntl, Npas2, and Per changed in a coordinated manner with respect to rhythmic transcription. LA further caused a decrease in transcripts of several bile acid and lipid synthesis genes, including Acacb and Fasn, which are regulated by first-order clock transcription factors. Similar effects of LA supplementation on bile acid and lipid synthesis genes were observed in young animals. Transcript changes of lipid metabolism genes were corroborated by a decrease in FASN and ACC protein levels. We conclude that advanced age is associated with a necroinflammatory phenotype and increased lipid synthesis, while chronic LA supplementation influences hepatic genes associated with lipid and energy metabolism and circadian rhythm, regardless of age.

genome-wide transcription changes are evident in older animals from many species, resulting in shifts in metabolic function, loss of stress response, and increased inflammation (24–26, 39, 40). Oxidative stress also plays a role in the aging process, both as a result of alterations in gene expression and functional declines in mitochondria (11, 34). Thus, there has been research focus on finding dietary interventions or supplements that may address these functional changes associated with advanced age.

R-α-lipoic acid (LA) is a cofactor for mitochondrial α-ketoacid dehydrogenases and thus serves a vital role in cellular bioenergetics. Nonprotein-bound LA, which transiently accumulates in cells and tissues after oral administration, initiates a number of effects that are surprisingly varied in scope (reviewed in Ref. 46). In vitro, supplemental LA exerts potent antioxidant effects that may limit free-radical-induced oxidative damage (52) and maintain other endogenous antioxidants in their reduced, active state (10, 37, 52). LA may also affect the activity of a variety of kinases and phosphatases via cysteine modification (46). In addition, our laboratory has identified LA as an age-essential micronutrient that acutely reverses the age-related decline in nuclear factor erythroid 2-related factor 2 (Nrf2)-dependent cellular antioxidant and detoxification defenses (50, 51). We also showed that LA reduces chronic inflammation associated with aging, most likely by inhibiting NF-κB activation (27, 61). Acute stimulation of antioxidant gene expression was only seen in aged but not young animals, suggesting that LA may exert both age-specific effects on cell and organ function and initiate common signal transduction pathways regardless of age.

Because of the particular cellular pathways influenced by orally administered LA, this compound is now under active consideration as a clinical adjunct for the treatment of diverse pathologies such as hypertension (31), atherosclerosis (59), proinflammatory conditions (23, 47), hypertriglyceridemia (3), and diabetes-induced polyneuropathies (43, 62). However, it is still unknown whether LA supplementation given over days or weeks at doses appropriate for pharmacological intervention would instigate similar or altogether distinct effects compared with those achieved by acute LA treatment. Thus, this study aimed to determine hepatic transcript changes in young and old rats with or without 2-wk LA supplementation. Herein we present the hepatic gene expression changes seen in the context of aging and the effects of LA supplementation in both old and young animals. We found that, in the liver, LA mainly influences two groups of potentially interconnected gene clusters regardless of age, namely, central circadian clock genes and genes for fatty acid metabolism and bioenergetics.

MATERIALS AND METHODS

Animals.

Young (3 mo) and old (24 mo) Fischer 344 male rats were fed an AIN-93M diet (Dyets) without or with 0.2% (wt/wt) LA (Mak Wood) for 2 wk prior to death. For the LA-supplemented diet, LA was uniformly incorporated into the pelleted diet during formulation. A strict feeding protocol was followed in which animals were fed between 8:00 and 9:00 AM each day. To minimize the contribution of differences in food intake between animals fed LA-containing and control diets, a staggered pair-feeding regimen was employed: the LA-fed rats were allowed to eat ad libitum, and the amount of food consumed was recorded. The same amount of food was then provided to its paired rat on the standard diet the following day. Each morning (∼8:00 AM), the remaining food was removed and weighed. The amount of food consumed (and therefore the amount of LA consumed) was determined by subtraction. Despite declines in food consumption, no significant decreases in body weight were observed (Fig. 1). During the 2-wk supplementation period, two old rats, both from the LA-supplemented group, died of natural causes and thus were not available for gene expression analysis. No abnormalities in liver morphology were observed for these particular animals other than the usual age-related changes. Thus, death appeared not to be attributable to an LA-specific effect.

Fig. 1.

Pair feeding normalizes food consumption and body weight changes of rats within respective age groups. Food consumption in young (A) and old (B) animals decreased by ∼30% over the course of the 2-wk feeding period. Data for lipoic acid (LA) animals was shifted by 1 day to better visualize the results of pair feeding. Total body weight (C) increased slightly, but not significantly (P > 0.05) throughout the study. Differences in food consumption normalized to body weight (D) reflected the age of the animals. However, there was no significant effect of LA on body weight, food consumption, or normalized food consumption within age groups. Statistical analysis was carried out by 2-way ANOVA within age groups.

The remaining animals were fasted for ∼13 h before death. Rats were anesthetized using diethyl ether, and 0.2% (wt/vol) heparin injected into the iliac vein to prevent blood coagulation. Following death, the livers were perfused with ice-cold PBS to remove blood and immediately excised. Lateral slices of the livers' median lobe were stored in RNAlater (Ambion) as well as snap frozen in liquid nitrogen. To ensure that there was no bias in death time, animals were killed between 8:00 AM and 12:00 PM over four consecutive days, with the order of treatment groups varied each day. All animal experimentation described in this study was carried out in accordance with National Institutes of Health guidelines and was reviewed and approved by the Institutional Animal Care and Use Committee of Oregon State University (approval no. 3751).

Microarray analysis of RNA.

Total RNA was extracted from liver slices stored in RNAlater using the RNeasy minikit (Qiagen), according to manufacturer's instructions. Quality was confirmed through the comparison of amounts of intact ribosomal RNA subunits using the Bioanalyzer 2100 instrument (Agilent). For microarray analysis, Affymetrix Rat Genome 2.0 GeneChips were used. Microarray experimentation and analysis was performed in accordance with guidelines for Minimal Information About Microarray Experiments (MIAME), and raw data was deposited in the Gene Expression Omnibus (GEO) public database (accession no. GSE27625). Creation of cDNA templates, labeling, hybridization, and scanning were performed in accordance with manufacturer's protocols as described in the GeneChip Expression Analysis Technical Manual (no. 701021, Rev. 5).

Microarray data analysis.

Microarray data analysis was carried out using GeneSifter software (Geospiza). Normalization was performed using GCRMA, the probe sequence-sensitive version of the Robust Multi-array Average algorithm (RMA) (57). Probe sets were identified as statistically significant using two-tailed t-tests with Benjamini and Hochberg multiple testing correction (2) to obtain a false discovery rate of 5% or better. Probe sets were considered differentially expressed if there was a greater than twofold change between experimental groups. Lists of differentially expressed probe sets were then reduced to lists of corresponding genes by removing probe sets without annotation and redundancies corresponding to the same gene. In all tables, the accession number given under the heading “Gene ID” were provided by Affymetrix and correspond to the transcript record used to design the probe set.

Biological relevance of whole genome expression profiles.

Lists of differentially expressed genes were analyzed using the Database for Annotation, Visualization and Integrated Discovery (DAVID; http://david.abcc.ncifcrf.gov) (12, 13). Using this suite of statistical tools, enrichment analysis was performed to identify overrepresented Gene Ontologies (GO; www.geneontology.org) and pathways. Benjamini and Hochberg multiple testing correction (2) was performed, and adjusted P values were generated to assess relevance of gene lists to processes and pathways in the GO database. Additionally, functional annotation clustering was used to organize related ontologies into clusters. Gene ontology terms that were not statistically over-represented (P > 0.05) were removed from analysis. Lists of gene transcripts with similar functional designations, as determined by GO terms and literature review, were assembled and provided in the text or supplemental information.

PCR gene expression confirmation.

RNA used for microarray analysis was also analyzed for individual mRNA changes by TaqMan Gene Expression Array (Applied Biosystems) and real-time quantitative PCR (qPCR). Briefly, total RNA was reverse transcribed using the Retroscript kit (Ambion) and employing random decamers and heat denaturing conditions. Selected TaqMan primers (Applied Biosystems) were used according to the manufacturer's instructions in a total reaction volume of 20 μl, and ^ΔΔC_T relative quantification was used to determine differential expression of the gene of interest relative to β-actin between control and LA supplemented animals in each age group.

Western blot analysis.

Protein analysis of selected genes was carried out using Western blot analysis of snap-frozen liver tissue homogenized in a buffer composed of 20 mM HEPES pH 7.4, 50 mM glycerol 2-phosphate, 2 mM DTT, 1 mM Na₃VO₄, 2 mM EDTA, 1% Triton X-100, 10% glycerol, 1% protease inhibitor cocktail (cat. no. P8340; Sigma), and 1% each of phosphatase inhibitors 1 and 2 (cat. nos. P2850 and P5726; Sigma). The homogenate was centrifuged at 13,100 g for 10 min, whereupon the supernatant was saved, and protein concentration was determined by the Bradford protein assay (Bio-Rad). Samples were then mixed 1:1 with sample buffer (100 mM Tris·HCl pH 6.8, 4% SDS, 20% glycerol, 0.001% bromophenol blue, 100 mM DTT). Samples were loaded on a 4–10% linear gradient Tris·HCl gel (Bio-Rad) and separated by SDS-PAGE. Membranes were blotted with antibodies for acetyl-CoA carboxylase (ACC), phosphorylated ACC (phospho-ACC; cat. nos. 3662, 3661; Cell Signaling), and fatty acid synthase (FASN; cat. no. 610962; BD Biosciences). Appropriate horseradish peroxidase-linked secondary antibodies were used for detection by chemiluminescence reagents (cat. nos. 1859674 and 1859675; Thermo Scientific). Membranes were additionally blotted with an antibody for β-Actin (cat. no. A5441; Sigma) for use as an internal control. All protein bands were quantified by densitometry using Image J (National Institutes of Health).

Statistics.

Except where noted, statistical significance for pairwise comparisons was determined using two-tailed Student's t-test, with P < 0.05 considered statistically significant. For comparisons involving all four experimental groups, two-way ANOVA was applied with P < 0.05 once again considered significant if no interaction was observed. Data in graphs are presented as means ± SE.

RESULTS

Average linkage clustering analysis of the normalized, unfiltered, microarray data revealed that, as expected, age is a fundamental factor determining whole genome expression profiles (Fig. 2). LA supplementation provided a secondary level of segregation of the treatment groups, with the exception of one old control animal (OC1), which had an expression profile similar to old animals on the LA-supplemented diet. Thus, clear demarcations are present between the four experimental groups, implying that gene expression changes induced by LA are consistent within treatment groups and are not an artifact of random changes.

Fig. 2.

Hierarchical clustering of gene expression profiles. Whole genome expression data from the 30 surviving study animals are represented in a hierarchical clustering dendogram produced using Genesifter software. *Statistical outliers from treatment group clusters. YC, young control; YLA, young LA-supplemented; OC, old control; OLA, old LA-supplemented.

The gene expression profiles of two old rats, OC5 and OLA2 (Fig. 2), were found to be divergent from the rest of the animals. Upon further analysis, these two animals yielded data that were two orders of magnitude removed from the means of their respective groups. Necropsy records and general blood chemistries revealed gross undefined liver and kidney pathologies (data not shown). Therefore, data derived from these animals were excluded from subsequent analysis. Removal of OC5 and OLA2 resulted in eight rats each in the young control (YC) group and the young LA-supplemented (YLA) group, seven animals in the old control (OC) group, and five animals in the old LA-supplemented (OLA) group.

Differential gene expression with age.

Analysis of the microarray data showed that there were 953 probe sets or 668 distinct genes whose expression changed at least twofold with age. Overall, transcripts of 509 genes were significantly increased and 159 transcripts decreased in old vs. young rats (Table S1; Supplemental data can be found with this article in the online Am J Physiol Regul Integr Comp Physiol). Functional annotation clustering analysis, carried out using DAVID (Tables 1 and S2), showed a significant over-representation of inflammatory processes, immune cell activation and infiltration, and attendant tissue remodeling. Further review showed that 44% of gene changes were related to a necroinflammatory phenotype. Specifically, there were notable increases in transcripts of proinflammatory, cell death, cell cycle progression, adhesion molecules, fibrosis, and tissue remodeling genes (Table S3). Additionally, a number of genes for signaling molecules, protein kinases and phosphatases, G-coupled proteins, and phosphoinositide signaling-related proteins, exhibited age-dependent changes in hepatic expression profiles (Table S1).

Table 1.

Aging results in significant modulation of hepatic gene expression: ontology clustering profiles

Cluster	Enrichment Score	Gene Ontology Term	Number of Genes	Fold Enrichment*	P Values†
Cluster 1	10.80	Leukocyte activation	43	4.98	6.64E-15
		T cell activation	29	6.39	2.72E-12
		Lymphocyte activation	35	5.14	2.24E-12
		Leukocyte differentiation	26	5.17	7.28E-09
		GT cell differentiation	19	7.10	2.22E-08
		Lymphocyte differentiation	22	5.71	3.58E-08
		Positive regulation of lymphocyte activation	20	5.09	9.18E-07
		Immune system development	33	3.11	1.90E-06
		Positive regulation of T cell activation	17	5.75	2.12E-06
		Regulation of lymphocyte differentiation	13	5.63	1.14E-04
Cluster 2	6.90	Response to wounding	43	2.50	5.44E-06
		Inflammatory response	25	2.84	3.12E-04
Cluster 3	6.11	Positive regulation of cell death	41	3.02	1.18E-07
		Positive regulation of apoptosis	40	2.99	2.36E-07
		Regulation of apoptosis	61	2.22	7.31E-07
		Regulation of cell death	61	2.18	1.13E-06
		Induction of apoptosis	22	2.83	1.05E-03
Cluster 4	5.84	Response to endogenous stimulus	52	2.24	6.43E-06
		Response to hormone stimulus	46	2.23	3.89E-05
		Response to steroid hormone stimulus	30	2.54	3.00E-04
		Response to peptide hormone stimulus	23	2.56	2.65E-03
		Response to corticosteroid stimulus	16	3.01	6.27E-03
Cluster 5	5.77	Chemotaxis	21	6.24	2.06E-08
		Leukocyte migration	16	6.93	5.65E-07
		Leukocyte chemotaxis	12	8.00	9.77E-06
		Cell migration	29	2.76	1.07E-04
		Cell motion	35	2.08	2.14E-03

^*Fold increase above the number of genes expected at random for a given Gene Ontology term.

^†Adjusted P value obtained from Benjamini and Hochberg multiple testing correction (2).

Despite the concerted induction of proinflammatory and immune response genes, it was equally notable that there was no coincident activation of anti-inflammatory or antioxidant defense genes. While message levels of certain stress-response genes increased, such as Nfe2l2 (2.11-fold), Gpx2 (10.37-fold), Gstp1 (2.38-fold), and Hmox1 (3.41-fold), there was an equal number of antioxidant and detoxification genes whose message levels declined (Table S1).

In concert with the above gene expression profiles, there was a profound shift in gender-related gene transcripts. Consistently, male-specific gene transcripts declined with age, while female-specific genes showed a corresponding increase in expression (Table S3). Additionally, a marked number of transcripts involved in steroid hormone metabolism were downregulated in aging, such as Hsd3b5 (−1598.42-fold), Sult1e1 (−546.7-fold), and Cyp3a2 (−204.87-fold). Overall, this feminization of old male rats represented nearly 17% of all annotated transcripts that changed with age.

Finally, nearly 9% of the 668 age-associated transcript changes occurred in genes involved in energy metabolism (Table S3). Overall, there was a trend toward decreased fatty acid catabolism and an increase in fatty acid, terpenoid, and cholesterol synthesis. These transcript profiles indicate the presence of a mild dyslipidemia in aged rats, akin to nonalcoholic fatty liver disease (58). Additionally, we observed alterations in amino acid, ketone body, and carbohydrate metabolism, which are consistent with a mild dyslipidemic phenotype (Table S3).

Effect of LA supplementation on age-dependent transcriptional changes.

To address our hypothesis that LA supplementation mitigates age-related differences in hepatic gene transcript levels, we compared the genes whose expression significantly changed with age to those that exhibited differences under LA-supplementation in old animals. As illustrated in Fig. 3A, LA affected transcripts corresponding to 62 genes (104 probe sets, Table S1) in old rats. However, only 25 of these genes were also influenced by age, representing < 2% of the overall age-affected genes. The levels of only 14 of these gene transcripts, many of which are associated with lipid metabolism (e.g., Fasn, acetyl-CoA carboxylase-β, Scd2) were reversed by LA-supplementation (Table 2). This small number of genes suggests that LA did not coherently reverse the necroinflammatory phenotype in the aging rat liver.

Fig. 3.

Venn diagrams of gene transcript profiles. Circles represent the number of gene transcripts reaching statistical significance and a greater than 2-fold change in expression between the denoted animal groups; they do not indicate the direction of transcript changes. A: 668 genes were differentially regulated in old compared with young animals. However, only 62 genes were affected by LA supplementation in old animals vs. their respective controls. Overlapping 25 genes (intersection) were affected by both age and LA supplementation in old animals. B: LA supplementation changed transcript levels of 166 and 62 genes in the livers of young and old animals, respectively. The intersection represents the 39 genes affected by LA in both age groups.

Table 2.

Genes with altered gene age-related expression, but whose levels reverted following a R-α-lipoic acid (LA) supplementation

Gene Name	Gene Symbol	Aging FC (OC vs. YC)	LA FC (OLA vs. OC)	Net FC (OLA vs. YC)	Gene ID, Accession No.
ATP-binding cassette, sub-family G (WHITE), member 5	Abcg5	−19.25	3.27	−5.89	NM_053754
Acetyl-Coenzyme A carboxylase beta	Acacb	3.70	−2.86	1.29	NM_053922
Basic helix-loop-helix domain containing, class B3	Bhlhb3	2.06	−2.06	1.00	NM_133303
Carboxylesterase 3	Ces3	−3.02	2.54	−1.19	L46791
Cytochrome P450, family 3, subfamily a, polypeptide 9	Cyp3a9	−2.09	4.36	2.09	U46118
Fatty acid synthase	Fasn	4.41	−4.98	−1.13	NM_017332
Kyphoscoliosis	Ky	3.62	−3.09	1.17	BE111310
Melanoma cell adhesion molecule	Mcam	2.76	−2.16	1.28	BI277043
Monocyte to macrophage differentiation-associated	Mmd	2.75	−2.51	1.10	BG376037
Similar to cysteine-rich glycoprotein	RGD1566394	5.45	−3.67	1.49	AA946147
Regulator of G protein signaling 1	Rgs1	2.42	−2.30	1.05	BM386789
Stearoyl-Coenzyme A desaturase 2	Scd2	10.14	−3.95	2.57	NM_031841
Tissue inhibitor of metalloproteinase 3	Timp3	5.16	−2.14	2.41	AI599265
Testis specific X-linked gene	Tsx	34.22	−12.49	2.74	NM_019203

FC, fold change; OC, old control group; YC, young control group; OLA, old LA-supplemented group.

In addition, LA further accentuated the age-induced changes in 11 genes (Table 3). The genes identified in this group did not appear to represent any particular metabolic pathway or influence regulatory enzymes. Overall, these results indicate that LA does not play a major role in mitigating age-induced changes at the transcriptional level.

Table 3.

Age-related gene changes that are accentuated by LA supplementation

Gene Name	Gene Symbol	Aging FC (OC vs. YC)	LA FC (OLA vs. OC)	Net FC (OLA vs. YC)	Gene ID, Accession No.
Cyclin-dependent kinase inhibitor 1A	Cdkn1a	4.3	2.37	10.19	U24174
Interferon, alpha-inducible protein (clone IFI-15K)	G1p2	2.93	2.5	7.33	BE096523
Gremlin 2 homolog	Grem2	−4.71	−2.01	−9.47	AA817956
LIM domain only protein 7	LMO7	−2.53	−2.07	−5.24	BI284480
6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 1	Pfkfb1	−2.92	−2.45	−7.15	NM_012621
Rhomboid, veinlet-like 7	Rhbdl7	−2.26	−2.06	−4.66	BI294778
Solute carrier family 2, member 5	Slc2a5	−2.42	−3.45	−8.35	NM_031741
Solute carrier family 6 (neurotransmitter transporter, taurine), member 6	Slc6a6	2.46	3.8	9.35	NM_017206
Thyrotroph embryonic factor	Tef	−2.05	−2.31	−4.74	NM_019194
Thymidylate kinase family LPS-inducible member	Tyki	2.33	2.23	5.2	BI276216
Ubiquitin specific peptidase 2	Usp2	−2.14	−3.66	−7.83	AF106659

Differential gene expression with LA supplementation.

LA supplementation affected 39 annotated genes regardless of age (Fig. 3B and Table 4). We observed that LA regulated each of these genes in a consistent manner in young and old animals. However, old animals generally showed a diminished response to LA in terms of the magnitude of gene transcript changes. Functional annotation clustering using DAVID revealed that genes related to circadian rhythm also were surprisingly affected, which has not been observed in other studies where LA and gene expression were monitored. Npas2, Arntl, Per2, Nr1d2, all genes that constitute the core regulatory machinery of the circadian system (20, 44), were influenced by LA supplementation (Fig. 4). These results thus indicate that chronic LA supplementation may ultimately alter circadian oscillation and its downstream metabolic output.

Table 4.

Genes affected by LA regardless of age

Gene Name	Gene Symbol	Young (FC)*	Old (FC)†	Gene ID
aryl hydrocarbon receptor nuclear translocator- like	Arntl	21.95	5.09	AB012600
cyclin-dependent kinase inhibitor 1A (p21, Cip1)	Cdkn1a	11.01	2.37	U24174
carboxylesterase 3	Ces3	2.37	2.54	L46791
choline kinase alpha	Chka	3.09	2.89	AW140851
claudin 1	Cldn1	3.11	2.56	NM_031699
cytochrome P450, family 3, subfamily a, polypeptide 9	Cyp3a9	3.88	4.36	U46118
ets variant 6	Etv6	2.37	2.26	BM387000
fibroblast growth factor 21	Fgf21	5.49	2.5	NM_130752
insulin induced gene 2	Insig2	2.43	2.18	AA818627
kynureninase (L-kynurenine hydrolase)	Kynu	2.7	2.03	NM_053902
neuronal PAS domain protein 2	Npas2	25.5	6.14	BI278550
similar to ENSANGP00000020885	RGD1563825	44.18	4.83	AI180253
solute carrier family 6 (neurotransmitter transporter, taurine), member 6	Slc6a6	15.36	3.8	NM_017206
torsin family 3, member A	Tor3a	2.38	5.5	BF284295
tribbles homolog 3	Trib3	5.89	2.82	AB020967
tubulin, beta 2a	Tubb2a	5.48	2.88	BM384071
acetyl-coenzyme A carboxylase alpha	Acaca	−2.9	−4.56	NM_022193
ATP citrate lyase	Acly	−2.28	−2.77	NM_016987
basic helix-loop-helix family, member e40	Bhlhe40	−6.16	−2.2	NM_053328
cytochrome P450, family 7, subfamily a, polypeptide 1	Cyp7a1	−2.48	−5.2	NM_012942
D site of albumin promoter binding protein	Dbp	−44.85	−10	AI230048
ELOVL family member 6, elongation of long chain fatty acids	Elovl6	−10.52	−5.32	BE116152
fatty acid synthase	Fasn	−7.8	−4.98	NM_017332
G protein-coupled receptor 146	Gpr146	−3.01	−2.08	AI170446
gremlin 2 homolog	Grem2	−4.16	−2.01	AA817956
laminin, alpha 3	Lama3	−3.89	−3.86	U61261
LIM domain 7	Lmo7	−2.77	−2.07	BI284480
hypothetical protein LOC685203	LOC685203	−2.81	−2.67	BI281851
nuclear receptor subfamily 1, group D, member 2	Nr1d2	−3.06	−2.54	U20796
period homolog 2	Per2	−11.31	−9.55	NM_031678
period homolog 3	Per3	−5.77	−3.58	BG374483
pyruvate kinase, liver and RBC	Pklr	−4.81	−2.13	NM_012624
patatin-like phospholipase domain containing 3	Pnpla3	−4.99	−9.44	BI273908
solute carrier family 2 (facilitated glucose/fructose transporter), member 5	Slc2a5	−12.56	−3.45	NM_031741
thyrotrophic embryonic factor	Tef	−3.83	−2.31	NM_019194
thyroid hormone responsive	Thrsp	−10.97	−4.35	NM_012703
testis specific X-linked gene	Tsx	−2.12	−12.49	NM_019203
ubiquitin specific peptidase 2	Usp2	−2.86	−3.66	AF106659
wee 1 homolog	Wee1	−4.72	−2.39	BE113999

^*Fold change in gene expression induced by LA in young animals (YLA vs. YC).

^†Fold change in gene expression induced by LA in old animals (OLA vs. OC).

Fig. 4.

LA modulates circadian rhythm transcripts. Shown is a schematic representation of the intertwined feedback loops that regulate circadian rhythm [negative arms: chryptochrome (Cry), casein kinase 1 (Ck1), Per, and the Rev-erb class of orphan receptors of which Nr1d2 is a member; positive arm: Npas2 (a functional homologue of the Clock gene) and Arntl (also known as Bmal1)]. Below are first-order clock transcription factors [Wee1, Tef, Dbp, Bhlhe40 (also known as Dec1), and Bhlhe41 (also known as Dec2)]. At the time of death and tissue isolation, LA increased transcript levels of genes in the positive arm, while simultaneously decreasing transcript levels of some genes in the negative arms of the circadian oscillators. Transcript level changes were mirrored in decreased levels of first-order clock transcription factors. Numbers represent fold change differences induced by LA supplementation in old animals vs. corresponding controls. *Gene transcript changes were only evident in livers of old animals.

Further analysis of the hepatic circadian rhythm genes showed that LA had a coordinated effect: LA-fed young and old rats exhibited enhanced transcript levels of Npas2 and Arntl and coincident declines in Per2 and 3, and Nr1d2, relative to unsupplemented controls. Considering that the Per and Nr1d2 are reciprocally regulated by Npas2 and Arntl in the core machinery of the circadian clock (Fig. 4), these results suggest that LA supplementation significantly alters the rhythmicity of circadian transcripts relative to controls. In concert with these findings, several so-called “first order clock transcription factors,” Bhlhe40, Wee1, Dbp, and Tef were all significantly suppressed (Fig. 4). Thus, the age-independent action of LA cohesively affects hepatic circadian clock genes.

To further define whether LA-mediated changes to the core clock gene network ultimately influenced genes involved in metabolic pathways, we analyzed the dataset for genes either directly or indirectly influenced by circadian cycles, and vice versa (16, 22, 42). Based on the data shown in Table 2, it was not surprising to observe that LA supplementation markedly affected transcript levels of key fatty acid synthesis genes and metabolism genes, many of which were for regulatory enzymes for their given pathways. For example, important genes involved in fatty acid synthesis (e.g., Fasn, Acaca, Elovl-6) and bile acid metabolism (Cyp7a1) were significantly downregulated in LA supplemented rats, which was in keeping with both their general regulation by first-order transcription factors of the circadian clock and what would be expected from the time of day when the animals were killed.

Validation of LA-induced gene expression.

As a validation of the microarray data, 12 genes differentially expressed under LA supplementation were selected for relative mRNA quantification by real-time qPCR. Expression levels of 10 of these 12 genes were altered by LA regardless of age, and two were affected only in one age group. Results of the qPCR analysis showed that changes in message levels for all 12 genes increased or decreased in accordance with their microarray profiles (Table 5). However, for two of the transcripts in old animals (acetyl-CoA carboxylase-β and Nr1d2) and one transcript in young animals (G6pd), qPCR analysis did not confirm that expression changes met the minimum of twofold threshold and P value. Nevertheless, expression of the selected genes as measured by qPCR corroborated the LA-induced changes in fatty acid synthesis and circadian rhythm gene expression observed in the gene array analysis (Table 5).

Table 5.

Confirmation of microarray data by real-time qPCR

	(YLA vs. YC)		(OLA vs. OC)
Gene	Array*	PCR†	Array*	PCR†
Npas2	25.50	18.79	6.14	4.98
Arntl	21.95	14.71	5.09	3.50
Dbp	−44.85	−56.18	−10.00	−10.34
Bhlhe40	−6.16	−4.47	−2.20	−1.88
Nr1d2	−3.06	−2.57	−2.54	−1.83§
Per2	−11.31	−6.49	−9.55	−5.12
Per3	−5.77	−29.78	−3.58	−6.49
Fasn	−7.80	−4.10	−4.98	−2.83
Acaca	−2.90	−3.00	−4.56	−2.30
Acacb	NC‡	NC	−2.86	−1.60§
Thrsp	−10.97	−14.01	−4.35	−4.05
G6pd	−2.18	−1.29§	NC	NC

^*Fold change according to Affymetrix microarray.

^†Fold change according to real-time qPCR.

^‡NC denotes either no change in microarray data vs. non-LA-supplemented controls or, for qPCR data, a corroboration that there was no change.

^§Data did not meet either two-fold change vs. control, or a P value <0.05.

To confirm that the LA-induced gene expression changes correlated with corresponding enzyme levels, we determined the hepatic protein amounts of FASN and acetyl-CoA carboxylase by Western blot analysis. The protein levels for both enzymes were significantly lower in young and old rats supplemented with LA compared with pair-fed controls (Fig. 5). In addition, the phosphorylation status of acetyl-CoA carboxylase, the rate-controlling enzyme in fatty acid synthesis, was also suppressed by LA, suggesting a markedly lower overall activity of this enzyme in liver of LA-fed rats. Collectively, these results corroborate the transcript data and show that LA attenuates hepatic fatty acid synthesis and lipid metabolism.

Fig. 5.

LA induces a decline in fatty acid anabolic enzyme levels and/or activity. LA supplementation resulted in statistically significant (P < 0.001) declines in protein amounts of fatty acid synthase (Fasn) (A), and acetyl-CoA carboxylase (ACC) (B). Phosphorylation status of ACC (denoted as phospho-ACC) as a marker of enzyme activity (C), revealed that LA induced a marked decline in ACC activity (P < 0.001). Throughout, statistical significance was assessed using two-way ANOVA. D: representative Western blot analysis for the data shown in (A–C).

DISCUSSION

In the present study, we found that aging is associated with profound changes in hepatic gene expression. Our data fully reflect other published reports for aged rodents in which stress- and immune-response genes were upregulated, while gene expression associated with general metabolism was diminished (1, 26, 35, 45, 53). A majority of the transcripts that increased with age were those involved in immune response, immune cell infiltration, proinflammatory processes, and tissue remodeling, without a compensatory induction of phase II or antioxidant stress-response genes. Overall, this suggests that aging tips the balance to a pro-oxidant and proinflammatory state that also has been called “inflamm-aging” (8). Associated with this necroinflammatory phenotype, ontological enrichment analysis revealed that genes involved in fatty acid metabolism, ketone body production, carbohydrate metabolism, and cholesterol biosynthesis increased significantly with age. This dyslipidemia may be a consequence or an effector of the aforementioned chronic inflammation, as numerous studies have identified that these processes are reciprocally linked (7, 48, 55).

LA-induced changes in the rat liver transcriptome.

We initially hypothesized that LA would predominantly attenuate expression of genes associated with inflammation in aged rats. However, declines in inflammation-associated transcripts were conspicuously absent in the old LA-supplemented animals. Alternatively, LA supplementation downregulates transcripts for fatty acid metabolism (Fasn, Acaca, Thrsp), which has been observed by others (15). This result is buttressed by our previous work showing that LA attenuates hypertriglyceridemia in ZDF diabetic rats (3) and lowers atherosclerotic plaque formation in LDL receptor/apoE knockout mice (59). Thus, a potential clinical benefit of long-term LA supplementation worthy of further exploration is as an antidyslipidemic agent to limit pathophysiologies, such as metabolic syndrome, type II diabetes, and cardiovascular disease.

LA-sensitive transcription.

Feeding rats with LA for 2 wk is a relatively short time period compared with the animals' full life span. However, this period certainly is more of a chronic form of supplementation relative to an acute pharmacological bolus of LA, such as a single intraperitoneal injection or gavage. Providing LA for 2 wk also is a relatively long time period to assess changes in mRNA levels, which are often affected quite rapidly. For example, when provided directly to cells in culture, LA immediately stimulates a number of signal transduction pathways, including the insulin receptor PI3-kinase/Akt and MAP kinases [as reviewed by Shay et al. (46)]. LA also interacts with redox-sensitive signaling molecules [e.g., thioredoxin (17, 37)], and transcription factors [e.g., Nrf2/Keap1, AP-1, SP-1, and NF-κB (6, 19, 28, 30, 33, 49, 60)]. Based on this evidence, LA would be expected to induce phase II detoxification, increase the expression of cellular antioxidant genes, increase glucose sensitivity, and act to limit proinflammatory processes in the liver. In contrast, the data from this study suggest that chronic, longer-term LA supplementation, as opposed to acute administration, influences genes regulated by PPAR, PGC-1α, HNF4α, and SREBP-1 (5, 18, 41). The reason for this discrepancy is not entirely clear, but may be due to the length of LA treatment, the species or strain of animals, or the method of LA administration. It is also possible that LA rapidly induces anti-inflammatory genes, but transcript levels return to baseline prior to animal death at the end of 2 wk of LA supplementation. More work will be necessary to fully examine the effects of both short- and longer-term LA supplementation on hepatic transcript levels. However, the present profile supports the view that LA attenuates lipid metabolism and potentially promotes glucose utilization in the long term, but does not affect pathways overtly governed by redox-sensitive molecules, as is seen in acute LA supplementation studies. Thus, it is intriguing to note that LA may chronically influence gene networks based more on its lipid nature rather than being a potent redox effector.

LA and circadian rhythm.

The finding that LA supplementation affects expression of central circadian rhythm genes (Npas2, Arntl, Nr1d2, and Per2 and 3), along with a host of downstream circadian clock-controlled genes, represents one of the most novel results of the present work. To our knowledge, LA has not been shown to influence this system previously.

As cues affecting circadian rhythm are designed to regulate and adapt to daily events for efficient energy utilization (20, 44), it may not be surprising that LA modulates these genes considering its influence on food intake and lipid metabolism. Because this study was not designed to analyze circadian rhythm, it is premature to suggest that LA directly mediates clock-controlled gene expression rather than merely promoting a cellular environment that influences these oscillators.

Nevertheless, it is enticing to speculate about the mechanism(s) and ramifications of LA as a potential circadian effector. Circadian rhythms are regulated by a set of oscillators governed by a cadre of genes whose expression is organized into transcriptional feedback loops (Fig. 4) (20, 44). In turn, these genes initiate expression of numerous downstream clock-controlled transcription factors. Our data show that at the time of animal death, LA upregulates genes in the positive arm (Arntl, Npas2) and downregulates genes in the negative arms (Per2, Per3, Nr1d2) of the circadian core oscillators (Fig. 4). Thus, LA may alter the rhythmicity of the central hepatic clock genes and attenuate expression of first-order clock transcription factors (Tef, Dec1, Dbp, and Wee1). Since numerous reports now show that central circadian clock gene expression is strongly influenced by downstream metabolic stimuli (21, 22, 56), LA may also target PPARα- and PGC-1α-mediated genes, which in turn form a feedback loop to the core circadian proteins Arntl and Nr1d1 (4, 9, 29). Furthermore, it has been shown that circadian regulatory genes and first-order transcription factors play a role in governing lipid metabolism-related genes and transcription factors such as Thrsp, PPARα, PPARγ, and Cyp7a1 (4, 16, 36, 54). It is reasonable to hypothesize that any LA-induced changes in circadian regulation are manifested in downstream changes of lipid synthesis and bioenergetics.

Despite the growing list of metabolic and age-dependent effects attributed to nonprotein-bound LA, it is notable that dietary supplementation with this compound influenced the expression of a relatively small number of genes. In fact, our data showed that LA modulated the expression of only 166 genes in young animals and 62 genes in old animals, with 39 genes commonly regulated in both age groups. Reports indicate that ∼10% of hepatic gene expression follows circadian rhythm, and pathways associated with circadian cycles include energy metabolism, cell growth and proliferation, host defenses, and hormone production (14, 32, 38). Regardless of the mechanisms involved, the effect of LA on transcription via circadian cycles may explain why this dithiol compound elicits such diverse effects in animals fed pharmacological doses of LA, despite modulating the expression of relatively few genes.

This study was undertaken as an unbiased screening of gene expression in aging and under LA supplementation, and therefore was discovery based rather than hypothesis driven. As such, there are certain advantages and disadvantages inherent to this approach. The unbiased screening allowed for the discovery of the previously unseen action of LA on circadian regulatory gene expression. However, deeper inquiry into biochemical mechanisms from this study is difficult as procedural refinement and end point determination are arrived at retrospectively. Thus, the interactions between supplemental LA, lipid metabolism, and circadian rhythm are speculative, but open new avenues for future experimentation and study.

Perspectives and Significance

This study provides an unbiased view of hepatic transcriptional changes in aging and after 2-wk supplementation with LA in both young and old animals. LA supplementation showed potential to mitigate age-related dyslipidemia, indicating that it may have clinical value as an intervention strategy in these cases. Furthermore, the effect of LA on transcription of circadian genes has a myriad of physiological implications. Circadian dysfunction is known to occur in aging, and disruptions to normal timing of circadian genes have been linked to a wide variety of insults and disorders, including altered sleep patterns, infertility, abnormal lipogenesis and gluconeogenesis (Arntl), impaired memory (Npas2), improper cell division and cancer development (Per2), and locomoter difficulty and retinal degeneration (Nr1d2) (20). Meanwhile, downregulation of fatty acid synthesis may indicate a shift in bioenergetics from energy storage to energy utilization, mirroring a change from a fed state to a fasted state. Considering the well-documented links between circadian rhythm and metabolism, we posit that LA, a compound usually considered important for its anti-oxidant properties, may also be acting through the circadian system, affecting metabolism as a downstream consequence. Thus LA may strongly influence general metabolic function, even in aged animals where circadian rhythm is adversely affected. Future studies will focus on whether and how LA directly affects cycling of circadian genes.

GRANTS

This study was supported by National Institute on Aging Grant RO1-AG-17141A and National Center for Complementary and Alternative Medicine Grant PO1-AT-002034, and by the National Institute of Environmental Health Sciences-funded Grant P30-ES-00210 from the Environmental Health Science Center at Oregon State University.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

B.F. and T.M.H. conception and design of research; L.A.F., A.J.M., J.A.B., E.J.S., J.S.M., R.F.M., K.P.S., and T.M.H. performed experiments; L.A.F. and A.J.M. analyzed data; L.A.F. and T.M.H. interpreted results of experiments; L.A.F. and A.J.M. prepared figures; L.A.F. drafted manuscript; L.A.F., A.J.M., J.A.B., R.F.M., B.F., and T.M.H. edited and revised manuscript; L.A.F., A.J.M., B.F., and T.M.H. approved final version of manuscript.