Sarcolipin and ubiquitin carboxy-terminal hydrolase 1 mRNAs are over-expressed in skeletal muscles of α-tocopherol deficient mice

Abstract

The transcriptome of ataxic muscles from α-tocopherol transfer protein deficient (ATTP-KO), 23-month old, mice was compared with that of their normal littermates. Genes encoding sarcolipin (sln) and ubiquitin carboxyl-terminal hydrolase (uchl1) were over-expressed (≥ 10-fold) in ataxic muscles. SLN is a 3.2 kDa membrane protein that binds to sarcoplasmic reticulum calcium ATPase, regulates Ca⁺⁺ transport and muscle relaxation–contraction cycles. UCHL1 is a 24.8 kDa member of proteosome proteins; it is over-expressed in myofibrillar myopathy and is associated with neurodegenerative diseases. Furthermore, six additional transcripts, three encoding thin-filament proteins and three encoding Ca⁺⁺ sensing proteins that participate in contraction-relaxation cycle, and eight transcripts that encode members of lysosomal proteins were also over-expressed in ataxic muscles. These observations suggest that chronic α-tocopherol (AT) deficiency activates critical genes of muscle contractility and protein degradation pathways, simultaneously. The magnitude of induction of sln and uchl1 was lower in asymptomatic, 8-month old, ATTP-KO mice and in 8-month old mice fed an AT-depleted diet. These studies suggest sln and uchl1 genes as novel targets of AT deficiency and may offer molecular correlates of well documented descriptions of neuromuscular dysfunctions in AT-deficient rodents. Since the neuromuscular deficits of ATTP-KO mice appear to be similar to those of patients with ATTP mutations, it is suggested that over-expression of sln and uchl1 may also contribute to AT-sensitive ataxia in humans.

Introduction

α-Tocopherol transfer protein (ATTP) is a ∼32 kDa protein that selects α-tocopherol (AT) from seven other members of the dietary vitamin E family and preferentially transfers AT to lipoproteins secreted by the liver [1]. ATTP is abundantly expressed in liver [2–4], but its expression is low or undetectable in peripheral organs and various regions of the central nervous system [5]. Therefore, hepatic ATTP is a major determinant of blood and tissue concentrations of AT [6,7]. The physiological importance of ATTP and AT is underscored by the discovery of several mutations in attp gene in patients displaying ataxia with vitamin E deficiency, AVED [8,9]. Although progressive neuromuscular disease associated with vitamin E deficiency in patients with AVED or abetalipoproteinaemia or chronic cholestasis [10,11] is well documented, the identity of the genes that may contribute to the failure of the neuromuscular system remain to be clarified.

The development of transgenic mice that recapitulate the systemic AT-deficiency and neuromuscular phenotypes of AVED patients with attp mutations offer a useful model to investigate mechanisms of actions of AT in vivo [12–17]. Most studies of the genetic model of AT-deficiency (ATTP-KO mice) have focused on defining AT-dependent changes in the central nervous system, which is severely depleted in AT [5,12] in spite of the mice consuming a diet containing sufficient AT to maintain normal fertility. For example, somatosensory evoked potentials in the cerebral cortex of 18–20 month old ATTP-KO mice were shown to be markedly attenuated when compared to those recorded from WT mice [14]. This observation may be attributed to low expression of genes encoding synaptic and calcium regulatory proteins in the cerebral cortex of ATTP-KO mice [12,18].

Muscles and various regions of the brains of ATTP-KO mice are severely AT-depleted, possibly from birth, compared to their WT littermates, in spite of consuming a diet that contains sufficient AT for fertility [5,19]. An earlier study described atrophic muscle fibres in 20-month old ATTP-KO, but not in the muscles of 12-month old ATTP-KO, ataxic mice [14]. These observations suggest that histological changes appear after the initiation of ataxia. More comprehensive descriptions of muscle fibre necrosis, decrease in cytochrome oxidase activity and ‘a waddling ataxic gait’ has also been described in ∼1-year old rats that consumed an AT-depleted diet, post-weaning [20]. However, temporal changes in the activity of muscle genome imposed by chronic AT-deficiency are less well characterized. A previous study addressed temporal changes in muscle gene expression during ∼1-year of AT-deficiency, imposed by an AT-deficient diet, that did not produce obvious neuromuscular deficits [21].

We hypothesized that simultaneous changes in the transcription of distinct sets of multiple, functionally-related, genes are likely to play an important role in the emergence and persistence of the ataxic phenotype that occurs late during chronic AT-deficiency. The present study aimed at identifying the possible contribution of ∼8500 muscle genes in the pathogenesis of ataxia, defined by shorter stride lengths of ‘old’, 23-month, ATTP-KO mice [14,18] compared to their normal littermates. In addition, some of the most responsive AT-sensitive genes identified in ‘old’ ataxic mice were also assayed in asymptomatic, ‘adult’, 8-month mice to define separately the contribution of age and AT-deficiency obtained by either the deletion of the ATTP-gene or by feeding an AT-depleted diet.

Materials and methods

Diets and mice

The AT-depleted and AT-sufficient (35 IU tocopheryl acetate/Kg diet) diets used in this study were prepared as previously described [22,23]. Mice were allowed to access food and water ad libitum.

All the experiments in mice were humanely conducted and approved by Institutional Animal Care and Use Committee, University of California, Davis. Male mice from our colony of WT and ATTP-KO mice were studied [5]. Eight-month old ‘adult’ mice (six WT mice and four ATTP-KO mice) and 22–23 month old, ‘old’, mice (five WT mice and four ATTP-KO) were used for these studies. The group of mice assigned for 8 month study included six WT mice which were randomly sub-grouped to be fed either the AT-depleted (n=3) or the AT-sufficient (n=3) diets.

Skeletal muscle collection and processing

Mice were anaesthetized with intraperitoneal injection of pentobarbital (120 mg/Kg). After collection of blood by cardiac puncture, quadriceps and gastrocnemius muscles were exposed, dissected and immediately frozen in dry-ice. The frozen samples were stored at −80°C.

RNA extraction and sample preparation for GeneChip and qRT-PCR analysis

Frozen quadriceps and gastrocnemius muscles from each mouse were pulverized in pre-frozen porcelain mortar and pestle (in dry ice) to obtain finely powdered frozen muscle which was transferred to a 2.0 ml conical glass-teflon homogenizer. RNA was extracted with PARIS™ protein and RNA isolation kit essentially as described by the manufacturer (Applied Biosystems, Foster City, CA). Total RNA was quantified spectrophotometrically.

GeneChip (Mouse Genome 430 A 2.0) analysis was performed as previously described [24], using four GeneChips. A 5 μg aliquot of each of the four total RNA extracts (two extracts from WT-muscles and two extracts from ATTP-KO muscles) were processed. The scanned GeneChip images were processed to obtain hybridization signal intensities as described in Affymetrix GeneChip Operating System, V 1.4 (Affymetrix, Santa Clara, CA). To obtain the list of genes that were differentially expressed between the ATTP-KO and the WT mice, the ‘.cel’ files were imported into dChip software (1.0.01). The data were normalized and processed to obtain differentially expressed genes with default settings [25]. The data from WT-mice were used as baseline and those from the ATTP-KO mice were used as experimental to obtain fold-changes. The list of genes with fold-changes was then edited to exclude all the genes that showed less than 2-fold change and those that did not have annotations. The data were tabulated to show gene names, their fold-change attributed to ATTP-KO genotype, Affymetrix probe set identifiers and gene number in mouse genome database. The Affymetrix probe set identifiers offer an opportunity to identify, from the Affymetrix Database, the precise deoxynucleotide sequences of probes used in the Mouse genome 430 A 2.0 GeneChips and for designing oligonucleotide primers for quantitative PCR verifications.

Quantitative real-time PCR (qRT-PCR) was performed as previously described [22]. Nineteen aliquots of cDNAs synthesized from muscle RNA extracts from 19 different mice were used for validation of the selected genes identified by the GeneChip assay. The selection criterion was based of their relevance to the novel findings of this study. The oligonucleotide sequences of primer pairs were obtained using Primer Express software (Applied Biosystems) and their sequences are shown in Table I.

Table I.

List of genes selected from GeneChip data and subjected to qRT-PCR validation. The specific primer pairs [deoxynucleotide sequences (5′ to 3′) are shown for each primer] for each gene. The oligonucleotide sequences from Affymetrix probeset identifiers were used to design primers of the respective genes using the primer design software. The primers were custom prepared and used as described in the Materials and methods section.

Gene name	Forward primer	Reverse primer
ATPase, Ca++ transporting, cardiac muscle, slow twitch 2	GTGGCCAGATTGCTCTACAGTG	AGGGCCCATTAGAAAGCATGT
B cell leukaemia/lymphoma 6	GAAGTTTTCAATGATGGACGGG	GCCTACACTTCAAAAAGGGATGG
CD36 antigen	GAAAATCAAGCTCCTTGGCATG	ACTCCAATCCCAAGTAAGGCC
Cathepsin B	GAACCACTGTGGCATTGAATCA	TCAGTGCGTGGGATTCCAG
Cathepsin L	ACCACTGTGGACTTGCCACC	CCCATCAATTCACGACAGGAT
Cathepsin S	GACGACCCCTCCTGTACGG	GCCAACCACAAGAACACCATG
Glyceraldehyde-3-phosphate dehydrogenase	TTGTGGAAGGGCTCATGACC	TCTTCTGGGTGGCAGTGATG
Sarcolipin	TGTGCCCCTGCTCCTCTTC	TGATTGCACACCAAGGCTTG
Troponin C, cardiac/slow skeletal	GAAGGACGACAGCAAAGGGA	CGGAAGAGATCCGACAGCTC
Troponin I, skeletal, slow 1	CAGCCTATGCGCACACCTTT	TCCCCTTTGTGTGCCATTTC
Ubiquitin carboxy-terminal hydrolase L1	ACGGCCATCTGTACGAGCTC	CATGGTTCACTGGAAAGGGC
α-tocopherol transfer protein	TGGAAACTCAACGCAATGGAG	GCCAGCCTTCCAGGTCAAA

Statistics

The means, standard error of means (SEMs) and significance of difference (p-values) in fold-change determined by qRT-PCR were calculated by t-test using GraphPad Prizm software 4.0 (La Jolla, CA).

Results

The analytical strategy included confirmation of genotype of the mice (Figure 1), obtaining the primary list of AT-sensitive genes (and their functional assignments) from muscles of 23-month old ataxic mice (Table II), confirmation of the selected genes by qRT-PCR (Table III) and assaying the AT-sensitive, qRT-PCR confirmed genes in the AT-deficient muscles of 8-month old mice (Table IV).

Figure 1.

α-TTP mRNA (ATTP) expression in muscles and livers of 2-year old wild type (WT) and α-tocopherol transfer protein deficient (KO) mice. ATTP and GAPDH mRNA expressions were assayed by GeneChips and qRT-PCR. The GeneChip data are from muscles of WT (n=2) and ATTP-KO (n=2) mice. The qRT-PCR data were normalized to GAPDH expression in each sample. ATTP expression in KO-liver was very low (p <0.00001) and it was undetectable in muscles. The qRT-PCR data are mean±SEM, for five WTand four KO tissues.

Table II.

α-tocopherol sensitive genes in muscles of 23-month old ataxic, male mice. Gene expression data from WT muscles (n=2 mice) were used as baseline and those from ATTP-KO muscles (n=2 mice) were used as experimental to calculate fold-change by dChip data analysis software for Affymetrix arrays. The data were grouped according to function from the list of 57 genes that were differentially expressed.

Functional group and gene name	Fold change	probe set ID	Accession #
Cytoskeletal and Ca sensitive
sarcolipin	14.7	1420884_at	AK008863
periostin, osteoblast-specific factor	6.5	1423606_at	BI110565
PDZ and LIM domain 5	3.7	1421413_a_at	NM_022554
troponin I, skeletal, slow 1	3.6	1450813_a_at	NM_021467
ATPase, Ca++ transporting, cardiac muscle, slow twitch 2	3.5	1452363_a_at	AA245637
tropomyosin 3, gamma	3.4	1449996_a_at	NM_022314
troponin C, cardiac/slow skeletal	2.8	1418370_at	NM_009393
myosin, heavy polypeptide 7, cardiac muscle, beta	2.7	1448553_at	NM_080728
capping protein (actin filament) muscle Z-line, beta	2.6	1453960_a_at	AK007209
vimentin	2.3	1450641_at	M24849
matrix Gla protein	2.2	1448416_at	NM_008597
calsequestrin 2	2.0	1422529_s_at	NM_009814
Protein degradation
ubiquitin carboxy-terminal hydrolase L1	8.4	1448260_at	NM_011670
cathepsin S	6.2	1448591_at	NM_021281
lysozyme	4.1	1423547_at	AW208566
P lysozyme structural	3.6	1436996_x_at	AV066625
cathepsin B	3.3	1448732_at	M14222
proteasome (prosome, macropain) 26S sub-unit, non-ATPase	2.4	1451056_at	BB034143
ubiquitin-conjugating enzyme E2D 3 (UBC4/5 homologue, yeast)	2.3	1455480_s_at	BG070759
cathepsin L	2.1	1451310_a_at	J02583
lysosomal-associated protein transmembrane 4A	2.0	1423368_at	BI695636
Immune-inflammation
immunoglobulin kappa chain, constant region	5.7	1452417_x_at	AV057155
lectin, galactose binding, soluble 3	4.4	1426808_at	X16834
apolipoprotein E	2.9	1432466_a_at	AK019319
beta-2 microglobulin	2.7	1452428_a_at	AI099111
CD36 antigen	2.3	1450883_a_at	BB534670
apolipoprotein D	2.3	1416371_at	NM_007470
histocompatibility 2, D region	2.2	1451784_x_at	L36068
histocompatibility 2, class II antigen A, alpha	2.1	1435290_x_at	BE688749
Thiol oxidation and metabolic stress
metallothionein 2	3.3	1428942_at	AA796766
cytochrome P450, family 2, sub-family e, polypeptide 1	3.2	1415994_at	NM_021282
thioredoxin interacting protein	3.2	1415997_at	AF173681
heat shock protein family, member 7 (cardiovascular)	2.8	1421290_at	BG968304
esterase D/formylglutathione hydrolase	2.1	1417825_at	NM_016903
cysteine and glycine-rich protein 3	2.1	1460318_at	NM_013808
spermine oxidase	−2.1	1424268_at	BC004831
Vesicular and secretory
synaptophysin-like protein	3.0	1422879_at	BE333485
synaptopodin 2	2.9	1450828_at	NM_080451
reticulon 4	2.4	1421116_a_at	NM_024226
vesicle-associated membrane protein 3	−2.8	1437708_x_at	BB552111
Mitochondrial
diaphorase 1 (NADH)	3.2	1422185_a_at	NM_029787
pyruvate dehydrogenase E1 alpha 1	2.6	1449137_at	NM_008810
indolethylamine N-methyltransferase	2.4	1418697_at	NM_009349
glutamate oxaloacetate transaminase 2, mitochondrial	2.3	1417716_at	U82470
Miscellaneous
carbonic anhydrase 3	3.4	1460256_at	NM_007606
tumour differentially expressed 2	3.2	1454811_a_at	AV026664
ADP-ribosylation factor-like 6 interacting protein 2	3.2	1416794_at	NM_019717
neural precursor cell expressed, developmentally down-regulated gene 4	2.9	1451109_a_at	BG073415
glucan (1,4-alpha-), branching enzyme 1	2.2	1420654_a_at	NM_028803
protein phosphatase 3, catalytic sub-unit, beta isoform	2.2	1427468_at	M81483
integral membrane protein 2A	2.1	1451047_at	BI966443
musculoskeletal, embryonic nuclear protein 1	2.1	1427201_at	AJ277212
cell division cycle 42 homologue (S. cerevisiae)	2.1	1449574_a_at	BF143638
HESB like domain containing 2	2.1	1423652_at	AV209097
protein kinase inhibitor, alpha	2.0	1420858_at	AK010212
DEAD (Asp-Glu-Ala-Asp) box polypeptide 51	−2.0	1428728_at	AK010015
B-cell leukaemia/lymphoma 6	−2.3	1421818_at	U41465

Table III.

Validation of GeneChip data by qRT-PCR assay. The genes selected from the list of 57 differentially expressed genes were analysed by qRT-PCR. Muscle RNA extracts from ‘old’, 23-month mice were used for validation. Four samples (two from WT mice and two from ATTP-KO mice) used for GeneChip assay and additional muscle RNA extracts from five mice (three WT mice and two ATTP-KO mice) were processed. The expression data from WT muscles were used as baseline and those from ATTP-KO muscles were used as experimental to calculate fold-changes as described in the Materials and methods section.

		Fold change (ATTP-KO/WT)
Gene name	symbol	GeneChip (n=2)	qRT-PCR Mean±SEM (n=4-5)
sarcolipin	Sln	14.7	33.0±1.1
troponin C1	Tnnc1	2.8	4.8±0.4
troponin I	Tnni	3.6	2.3±0.4
calcium transporting ATPase	Atp2a2	2.0	3.4±0.5
protein kinase inhibitor, alpha	Pkia	2.0	1.7±0.4
ubiquitin caroxy-terminal hydroxylase L1	Uchl1	8.4	10.2±1.5
CD 36	CD36	2.3	3.3±0.8
cathepsin S	Ctss	6.2	3.0±0.3
cathepsin B	Ctsb	3.3	4.3±1.6
cathepsin L	Ctsl	2.1	1.4±0.4
B-cell leukemia/lymphoma 6	Bcl6	−2.3	−1.9 ± 0.3

Table IV.

Changes in the expression of the sarcolipin and ubiquitin carboxy-terminal hydrolase 1 gene clusters in skeletal muscles from ‘adult’, 8-month old, pre-ataxic and ‘old’, 23-month, ataxic mice. The expression of the seven genes of sln gene cluster and four genes of uchl1 gene cluster was assayed by qRT-PCR of cDNA samples from skeletal muscles of ‘adult’, 8-month, mice fed either an AT-sufficient or an AT-depleted diet. Fold-change in the expression of the genes in the AT-depleted muscles was obtained by comparing its expression with that in the muscles of mice fed the AT-sufficient diet. Similarly, the effect of AT-deficiency imposed by the deletion of ATTP gene were obtained by comparing the expression of the genes in ‘adult’, 8-month, AT-depleted muscles of ATTP-KO mice with their WT-littermates.

		Fold change determined by qRT-PCR
	Genotype	WT	KO	KO
	Age	8-month	8-month	23-month
	Diet	DEFICIENT	NORMAL	NORMAL
Gene name	symbol	Mean±SEM (n=3-5)	Mean±SEM (n=3-5)	Mean±SEM (n=4-5)
sarcolipin	Sln	11.3±2.3	6.7±1.8	33.0±1.1
troponin C1	Tnnc1	1.3±0.3	1.4±0.4	4.8±0.4
troponin I1	Tnni1	3.6±0.6	1.5±0.2	2.3±0.4
calcium transporting ATPase	Atp2a2	3.1±0.5	−1.2±0.7	3.4±0.5
calsequestrin 2	Casq2	1.1±0.2	1.5±0.3	1.7±0.4
tropomyosin 3, gamma	Tpm3	1.1±0.2	1.8±0.4	3.3±0.9
myosin, heavy polypeptide 7	Myh7	−1.5 ± 0.6	1.1±0.2	1.5±0.2
ubiquitin carboxy-terminal hydroxylase L1	Uchl1	4.2±1.0	7.6±0.5	10.2±1.5
cathepsin S	Ctss	4.1±0.8	1.9±0.2	3.0±0.3
cathepsin B	Ctsb	1.6±0.2	1.3±0.1	4.3±1.6
cathepsin L	Ctsl	1.0±0.2	−1.5±0.8	1.4±0.4
		pre-ataxic		ataxic

Confirmations of ATTP-KO and WT genotypes

Figure 1 shows that the expression of ATTP mRNA in muscles of 2-year old ataxic mice, assayed by either GeneChips or qRT-PCR, is below the limits of detection. In contrast, ATTP mRNA expression is robust (similar to that of GAPDH as indicated by the ratio of 1) in WT, 2-year old, male mice and is very low in ATTP-KO mice, as expected. A previous study has shown that the muscles and other tissues of ATTP-KO mice are deficient in AT [19].

AT-sensitive muscle transcriptomes

To determine the genome-wide signature of mRNA changes associated with chronic AT-deficiency and the resulting ataxic phenotype, muscle RNA extracts from ‘old’ WT and ATTP-KO mice were processed. The GeneChip assay detected ∼12 000 mRNAs in muscles. Comparison of ATTP-KO- with that of WT-muscle transcriptomes identified 57 genes that were differentially expressed (≥ 2-fold and the difference in signal intensity for each gene between the two data sets was ≥ 100 units) (Table II). Fifty-three genes were over-expressed in ATTP-KO compared with WT-muscles. These genes were then clustered according to their functions and literature descriptions. Six groups of genes each including 4–7 genes were obtained (Table II). Thirteen genes could not be assigned to any of the six clusters and are listed as miscellaneous in Table II. The functional groups included genes that encode: cytoskeletal and calcium sensing-regulated proteins, those that participate in protein degradation, immune-inflammation, thiol-oxidation and metabolic stress, vesicular and secretory functions and those associated with mitochondria.

Over-expression of genes that affect Ca⁺⁺ mediated muscle contraction-relaxation cycles

A cluster of four-genes (sarcolipin (sln), troponin I (tnni), troponin C (tnnc) and calsequestrin 2 (casq2)), whose encoded proteins are known to participate in Ca⁺⁺ dependent muscle contraction and relaxation cycles, were over-expressed in muscles of ATTP-KO compared with WT littermate mice (Table II). There was a 14-fold induction of sln, the gene encoding SLN, a 3.6 kDa protein [26] that inhibits re-uptake of sarcoplasmic calcium by inhibiting the activity of ATPase, Ca⁺⁺ transporting 2, also known as SER-CA2A2 [27]. Gene encoding a target of SLN, SERCA2A2 was also induced. Table II also shows a 2-fold induction of the gene that encodes calsequestrin 2, the main Ca⁺⁺ buffer of sarcoplasmic reticulum [28]. At least three additional genes that regulate calcium-dependent contraction-relaxation cycles of skeletal muscles were also co-activated with sln (Table II). Coordinated inductions of tnni, tnnci and tpm3 are particularly remarkable because the encoded proteins form the thin filaments of striated muscles and they have been suggested to form an inhibitory unit that prevents interactions between actin and myosin [29] and, thereby, may modulate muscle relaxation and contraction cycles. This cluster of seven genes is designated hereafter as ‘the sarcolipin cluster’.

Activation of proteolytic and inflammatory responses by chronic AT-deficiency

Genes associated with proteolysis were over-expressed in ataxic, AT-deficient muscles (Table II). Many of the genes in this cluster (e.g. cathepsins B, L and S) are associated with lysosomes and phagosomes [30]. Three genes encoding members of the ubiquitin-dependent proteosomal pathway (some of which are associated with processing by phagosomes [31]) are also included in this cluster.

Ataxic muscles of ATTP-KO mice also displayed markers of overactive inflammatory response (Table II). Inductions of genes encoding members of major histocompatibilty class II and beta-2 microglobulin are included in one group for their roles in immune functions that depend on ubiquitin-proteosomal pathways [31,32]. Co-inductions of CD36 and apolipoproteins D and E genes are noteworthy because they participate in lipoprotein metabolism and are frequently implicated in inflammatory responses [33,34]; CD36 is a receptor for oxidized low density lipoproteins and apolipoproteins are integral components of these macromolecular complexes. A 3-fold induction of gene encoding cytochrome P4502e1 in the ATTP-KO muscles is further suggestive of increased oxidation of lipids and of proteins which may be targeted to proteosome complexes [35,36] in AT-deficient muscles.

QRT-PCR assay of selected genes validates GeneChip data

To address the possible relation between the ataxic phenotype and mRNA expression, we focused on genes that have previously been associated with muscle contraction-relaxation cycles and muscle protein degradation (Table III). The induction and repression of all the genes selected from the list of 57 differentially expressed genes, identified by the GeneChip assay, were confirmed by qRT-PCR assay of the same RNA samples used for the GeneChip assay (Table III). The precise magnitude of the change obtained by the two methods differed; for example, the GeneChip assay under estimated sln induction, compared to that obtained by qRT-PCR assay (Table III). Furthermore, muscle RNA extracts from additional ‘old’ mice (n=5, three WT mice and two ATTP-KO mice) also showed the predicted change in the expression of the same genes (Table III). Hence, qRT-PCR validation assays for the expression of genes identified by the GeneChip assay attest to the reliability and reproducibility of the analytical procedures used for the genome-wide analysis of muscle mRNAs and the initial analytical step in the search for AT-sensitive genes in ataxic muscles.

Effect of age and dietary AT-deficiency on the sarcolipin gene cluster

The contribution of age on the expression of the sarcolipin gene cluster, implicated in the regulation of muscle contraction-relaxation cycle, was assayed in muscles from ‘adult’ WTand ATTP-KO mice and in adult mice fed the AT-depleted diet. A 6-fold-induction of sln was detected in the ATTP-KO mice when compared to the expression of sln in muscles from WT mice (Table IV). Furthermore, when WT mice were fed the AT-depleted diet, muscle sln was also induced (Table IV). Thus, data from three different groups of mice of two different genotypes demonstrate that muscle sln expression is a sensitive indicator of AT-deficiency. Moreover, sln is the most responsive AT-sensitive gene within the sln cluster, in muscles of both ‘adult’ and ‘old’ mice (Table IV).

Two genes, one encoding TROPONIN I and the other encoding calcium transporting ATPase2a2 (also known as SERCA2a2), appear to be more sensitive to diet-induced AT-deficiency compared to deficiency resulting from the deletion of the ATTP-gene. Data in Table IV also suggest differential sensitivity of each gene within the sln cluster to AT-deficiency. For example, the large induction of troponin C1 (tnnc1) and that of tropomyosion 3, gamma (tpm3) is seen only in the muscles of ‘old’ ATTP-KO mice.

Effect of age and dietary AT-deficiency on the uchl1-gene cluster

Four of the nine genes in uchl1 gene cluster were selected for qRT-PCR verification in ‘adult’ and ‘old’muscles (Table IV). Expression of three of the four genes in this cluster was lower in muscles of ‘adult’ ATTP-KO compared with ‘old’ ATTP-KO mice. The expression of uchl1 was activated, 4.2- and 7.6-fold, by AT deficiency obtained by AT-deficient diet and by the deletion of ATTP gene, respectively; both these inductions were lower than that detected in muscles from ‘old’ ATTP-KO mice. Similarly, the fold-inductions of two of the three cathepsins were also lower in ‘adult’ AT-deficient mice, as indicated by ≤ 1.6-fold change, when compared to their inductions in ‘old’ mice (Table IV).

Collectively, these results show that AT-deficiency caused by either dietary depletion or by the deletion of ATTP-gene activates sln- and uchl1-gene clusters. The magnitude of these inductions was higher in ‘old’ mice compared to that in ‘adult’ mice.

Discussion

Transcriptomic programme for ataxic phenotype in skeletal muscles of ATTP-KO mice

The most significant and novel result of this study is the identification, by two independent analytical procedures, a robust induction of seven muscle genes, the sln cluster (Table IV), whose simultaneous activation may contribute to age-related ataxia, defined by shorter stride-lengths of ATTP-KO mice compared to their WT-littermates [14,18]. The sln cluster of seven genes consisted of sln, tnni, tnnc, tpm3, ATP2a2 (also known as serca2a2), myh7 and casq2. Remarkably, five of these seven genes, sln, tnni, tnnc, serca2a2 and casq2, encode proteins that are known to play an essential role in calcium mediated contraction–relaxation cycles of muscles [26,29,37]. The lack of co-ordinated induction of all these genes in younger, AT-deficient, mice (Table IV) which were asymptomatic prompt the interpretation that the sarcolipin cluster of ‘old’ ATTP-KO mice may be associated with the neuromuscular deficits observed in mice with severe AT-deficiency [14,18].

Of the seven genes in the sarcolipin gene cluster, sln showed a > 14-fold induction; the highest induction in this group, and it was independently confirmed by qRT-PCR of RNA extracts from nine different muscle samples (Table IV). Additional confirmation of sln induction by AT-deficiency was also obtained by qRT-PCR data from muscles of younger ATTP-KO mice and from muscles of WT mice fed an AT-depleted diet (Table IV). Previously mice displaying myopathy showed a > 70-fold induction of sln in response to the deletion of nebulin-gene [38]. sln encodes a 3.1 kDa protein that inhibits the activity of SERCA2, a Ca⁺⁺ pump that actively removes Ca⁺⁺ from sarcoplasm and participates in the recovery of muscle from contraction [27]. Hence, over-expression of sln detected in muscles of AT-deficient mice may delay muscle recovery from contraction. The muscles of ∼10-month old, AT-deficient rats also displayed abnormal electromyographic properties [39], possibly due to over-expression of sln. This postulate is further reinforced by the observations that the forced over-expression of sln in rat soleus muscle inhibits SERCA, decreases luminal Ca⁺⁺ and impairs contractility [40]. Our preliminary electrophysiological experiments that assayed isometric contractions of gastrocnemius muscle group via sciatic nerve stimulation showed that muscles from a 1-year old ATTP-KO mouse failed to maintain tetanic force and failed to recover after titanic stimulation when compared to the muscle of a 1-year old WT mouse (Carlsen and Bodine, unpublished). Further electrophysiological experiments are necessary to validate these preliminary observations and to further distinguish between the contribution of motor units and muscle fibre functions.

Two previous studies have suggested co-regulation of SLN and SERCA [26,41]. The data in Tables II–IV suggest that the expressions of the two mRNAs are also co-regulated. Hence, the robust induction of sln and other members of sarcolipin gene cluster is a novel and unanticipated action of AT-deficiency in skeletal muscle and may contribute to the ataxic phenotype of the ATTP-KO mice. Further studies are necessary to assay the expression of SLN protein and its fibre-type distribution in ataxic muscles of ATTP-KO mice. Our preliminary studies with commercially available anti-SLN polyclonal antibody have been unsuccessful both in cardiac atria (positive control where SLN is abundantly expressed [26]) and in skeletal muscles of ATTP-KO mice. We note with interest that SLN was first detected as a protein that co-purified with SERCA1 from fast twitch skeletal muscle [42] and was identified as a proteolipid protein and designated ‘sarcolipin’ [43]. A lack of SLN protein induction in the presence of a large induction of mRNA would implicate a role for micro RNA(s) which are known to block protein synthesis [44,45]. Experimental evaluations of these possibilities remain major challenges in addressing the molecular basis of AT-sensitive neuromuscular functions.

The large induction of sln was accompanied by the co-inductions of tnni, tnnc, tpm3 and casq2 (Tables II, III – IV), all of which participate in regulating Ca⁺⁺ dependent muscle contraction and relaxation cycles (Figure 2). Co-inductions of tnni, tnnc and tpm3 may be consistent with the observation that TNNI, TNNC and TPM3 form an inhibitory unit that prevents the interactions between actin and myosin [29]. Tables II–IV also show induction of the gene that encodes casq2, the main Ca⁺⁺ buffer of sarcoplasmin reticulum [28]. A previous study has shown that over-expression of casq2 in cardiac muscle decreases contractility [46]. Therefore, over-expression of these four genes may further add to the inhibitory actions of sln over-expression on skeletal muscle contractility of ATTP-KO mice. Future studies are necessary to determine if the changes in mRNAs detected here are translated into the respective proteins and if all the members of the sarcolipin gene cluster are necessary to display short-stride lengths of the old, ATTP-KO mice. Our data (Table IV) from ‘adult’, 8-month old, asymptomatic, AT-deficient mice suggest that the co-inductions of all the genes within the sarcolipin gene cluster may be required for the display of shorter stride lengths of ATTP-KO mice compared to those of their WT-littermates.

Figure 2.

Schematic representation of proteins that participate in calcium-dependent regulation of muscle contraction-relaxation cycles. Calcium is stored, bound to calsequestrin, in the lumen of sarcoplasmic reticulum. It is released, through ryanodine receptors, into the sarcoplasm in response to a nerve stimulus, causing muscle contraction. To initiate relaxation, Ca⁺⁺ is pumped out by SERCA2. Activity of SERCA2 is inhibited by SLN and PLN. Genes encoding proteins (shades of green or grey, darker=higher expression) were over-expressed in ataxic muscles of 2-year old male, ATTP-KO mice compared to those in age-matched, male WT-mice. The numbers indicate fold-induction of the mRNA, obtained by qRT-PCR, in the ATTP-KO muscles. sarcolipin (SLN), troponin I and troponin c, tropomyosin, phospholamban (PLN), sarco-endoplasmic reticulum calcium ATPase2 (SERCA2), calsequestrin, myosin of thick filament, ryanodine receptor.

The only previous skeletal muscle study of chronic (up to ∼13 months) vitamin E deficiency in rats that used GeneChips identified 56 AT-sensitive genes, all of which were induced by dietary vitamin E deficiency [21]. Hence, two studies that used different in vivo models to produce AT-deficient muscles and different data analysis protocols arrived at very similar numbers of AT-sensitive genes. There is also a considerable overlap in the functional groups of genes that are modulated by severe AT-deficiency. The induction of the cluster of genes that encode vesicular and secretory function (Table II) is also noteworthy because two previous studies have documented a role for AT in the modulation of genes that regulate vesicle assembly and function [12,47]. Comparison of the muscle AT-sensitive genes in the two studies gives important clues about the genes that may contribute to the ataxic phenotype of vitamin E deficient rodents and, possibly, humans. The data from dietary AT-deficient rats [21] were remarkable for the lack of change in the sarcolipin gene cluster described in this study (Tables II, III and IV). It is also noted that the rats were asymptomatic. This observation underscores the possible relevance of the sarcolipin gene cluster in the delayed display of ataxia in the mouse model of AVED [14,18].

AT-sensitive changes in the cerebral cortex may also contribute to the ataxia phenotype of AT-deficient mice [12,14] and neurological deficits of AT-deficient rats [48]. Previous studies in rodents [49] and humans [50] have suggested that chronic AT-deficiency results in initial loss of neural functions. Hence, the transcriptomic responses of skeletal muscles of ATTP-KO mice described here may be in addition to neural deficits. Further studies that address the changes in gene expression programmes of the various regions of the central nervous system, the peripheral nervous system and the muscular system are needed to characterize the chronology of molecular changes that are associated with the emergence of neuromuscular phenotypes associated with chronic AT-deficiency.

AT-deficiency activates protein degradation and inflammation pathways

Muscles of ATTP-KO mice had increased expression of genes that encode proteolytic enzymes associated with lysosomes and phagosomes and members of proteosomal pathway (Tables II, III and IV). Presence of increased levels of lysosomes in muscles of an AT-deficient patient have also been described [11]. The induction of cathepsins (Tables II, III and IV) was also reported in the previous study [21]. Unlike the previous study, the present study detected over-expression of CD36 gene (Tables II, III and III), which encodes the receptor for oxidized lipoproteins [51,52], in ‘old’ but not in ‘adult’ AT-deficient mice. Since the induction of CD36 was detected only in the ‘old’ ataxic mice we infer that the inflammatory process is also a late event which may be triggered by increased accumulation of oxidized phospholipids [53] in AT-deficient muscles. Genes encoding apolipoproteins D and E genes were also activated by AT-deficiency (Table II). Products of these genes are often used as markers of increased inflammation [33,34] and have been reported to be induced in myopathies [54,55] that are not associated with AT-deficient muscles. Comparative analysis of muscle inflammatory markers associated with various myopathies may enable identification genes that are specifically associated with chronic AT-deficiency.

Temporal disconnect between the onset of AT-deficiency and the appearance of neuromuscular phenotypes

The molecular mechanisms that account for the long lag-phase (months) between the initiation of AT-deficiency and the display of neuromuscular phenotype remain unexplained. The phenomenon is extensively documented. For example, the maximum length of deficiency in the asymptomatic rats was 1-year post-weaning [21], which contrasts with almost 2 years of AT-deficiency, including the pre-weaning period, in ataxic, ATTP-KO mice.

It can be hypothesized that ‘pro-ataxic’ metabolites, yet to be identified, are generated and accumulate in terminally differentiated muscle fibres during chronic AT-deficiency. Since tocopherols are primarily sequestered in biological membranes [56], it is reasonable to propose that AT-depleted biological membranes, including nuclear membranes, are a significant source of these ‘pro-ataxic’ metabolites. Furthermore, phospholipids, a ubiquitous class of membrane lipids, are also present in the nucleoplasm [57] and have been suggested to be associated with, and modulate the activity of, chromatin [58]. Hence, AT-depleted muscle membranes may have altered phospholipid metabolism. The present GeneChip data may be interpreted to support this hypothesis. The assay identified a 2-fold induction of cyp2e1 (Table II, Thiol oxidation and metabolic stress group). A previous study has suggested that AT inhibits CYP2E1 activity [59]. CYP2E1 has been suggested to increase reactive oxygen species [35,36] which may affect activities of redox-sensitive transcription factors such as nuclear factor (erythroid-derived 2)-like 2 (NRF2), nuclear factor of kappa light polypeptide gene enhancer in B-cells, (NF-κB) and hypoxia-inducible factor [60–62]. A previous transcriptomic analysis of lungs from ATTP-KO mice suggested modulation of AhR-NRF2-NF-κB transcription factor networks [22]. A search of the muscle transcriptomic data (Table II) for genes that may be induced by the activation of these transcription factors identified genes such as diaphorase 1 (also known as NAD(P)H:menadione oxidoreductase), metallothionein 2, thioredoxin interacting protein and heat shock protein family, member 7 (Table II). The previous GeneChip analysis of AT-deficient muscles from rats fed AT-deficient diet for ∼1-year also suggested increased expression of seven genes related to oxidative stress [21]. Collectively, these data offer mRNA correlates of metabolic pathways which may contribute to temporal changes in metabolites, possibly derived from lipids of membranes and nucleoplasm, which may result in an altered transcriptomic programme, precipitating failure of the neuromuscular functions with chronic AT-deficiency.

Conclusions

Comparative analysis of skeletal muscle transcriptomes from ‘old’, 23-month, ataxic, ATTP-KO mice and their WT-littermates identified two genes, sln and uchl1, as sensitive responders of AT-deficiency. Sln encodes a ∼3.2 kDa proteolipid protein that is an inhibitor of sarcoplasmic reticulum calcium transporting ATPase and it regulates relaxation-contraction cycles. Furthermore, six additional genes whose encoded proteins participate in relaxation-contraction cycle were also over-expressed in ataxic muscles. Hence, these seven co-regulated genes appear to form a functional cluster, ‘the sarcolipin gene cluster’, which may contribute to delayed ataxia in α-tocopherol-transfer protein deficient mice and possibly in AVED patients. These transcriptomic data suggest a role for α-tocopherol in the transcriptional regulation of genes that affect calcium dynamics across sarcoplasmic reticulum.