Regulation of myosin heavy chain antisense long noncoding RNA in human vastus lateralis in response to exercise training

Abstract

Alterations to muscle activity or loading state can induce changes in expression of myosin heavy chain (MHC). For example, sedentary individuals that initiate exercise training can induce a pronounced shift from IIx to IIa MHC. We sought to examine the regulatory response of MHC RNA in human subjects in response to exercise training. In particular, we examined how natural antisense RNA transcripts (NATs) are regulated throughout the MHC gene locus that includes MYH2 (IIa), MYH1 (IIx), MYH4 (IIb), and MYH8 (Neonatal) in vastus lateralis before and after a 5-wk training regime that consisted of a combination of aerobic and resistance types of exercise. The exercise program induced a IIx to IIa MHC shift that was associated with a corresponding increase in transcription on the antisense strand of the IIx MHC gene and a decrease in antisense transcription of the IIa MHC gene, suggesting an inhibitory mechanism mediated by NATs. We also report that the absence of expression of IIb MHC in human limb muscle is associated with the abundant expression of antisense transcript overlapping the IIb MHC coding gene, which is the opposite expression pattern as compared with that previously observed in rats. The NAT provides a possible regulatory mechanism for the suppressed expression of IIb MHC in humans. These data indicate that NATs may play a regulatory role with regard to the coordinated shifts in MHC gene expression that occur in human muscle in response to exercise training.

INTRODUCTION

Skeletal muscle is highly sensitive to increases in aerobic and resistance-loading activity. A stimulus such as this triggers adaptive responses that alter the functional characteristics of the myofiber. This may manifest by alteration of the size and contractile kinetics of the myofiber, resulting in increased strength and endurance, and a shift to a slower contractile phenotype. Of particular consequence is modification to protein levels of myosin heavy chain (MHC), the motor protein of striated muscle. Physical exercise or inactivity can alter the quantity and quality of MHC. Changes in the quantity of MHC is notable because it makes up ~40% of myofibrillar protein (63). The type, or quality, of MHC also has significant impacts on the function of muscle. There are three protein isoforms of MHC expressed in human skeletal muscle (I, IIa, IIx). Another isoform, IIb MHC, is typically only expressed in skeletal muscle of small mammals and not normally expressed at the protein level in human limb muscle (48, 58). The intrinsic contractile speeds of a given myofiber are a consequence of the type of MHC expressed, with I-MHC generating the slowest, and IIa, IIx, IIb MHC generating successively faster velocities of contraction (47).

With inactivity/unloading stimuli, muscles atrophy and fatigue-resistant motor units shift their phenotype from energy efficient, oxidative, and slow-contracting type to energetically inefficient, glycolytic, fatigable, and faster fiber type (6, 17, 18, 61). Sedentary humans appear to have a baseline fiber type consistent with greater content of IIx MHC, and lesser of I-MHC, as compared with their more active counterparts (4, 9, 31, 32, 49, 61a). Conversely, increased physical activity can induce the opposing response; the muscle will hypertrophy and become more fatigue resistant in response to exercise with a shift in expression to slower MHC isoforms (3, 4, 20, 26, 46, 54, 56). The modifications to MHC expression occur primarily via alterations to transcription levels of the corresponding genes (1, 19, 43). Understanding how the genes encoding MHC are transcriptionally regulated will help in understanding the mechanisms that underlie how exercise and inactivity impact muscle phenotype and function.

The chromosomal organization, gene order, and orientation of the MHC genes are conserved across mammalian species. The MHC genes encoding IIa (MYH2), IIx (MYH1), IIb (MYH4) and the developmentally regulated Neonatal (MYH8) and Embryonic (MYH3) are aligned in tandem and separated by relatively short intergenic distances (55, 65). We have previously shown that this organization is an important feature in the strategy for the coordinated regulation of these genes (43, 52). As shown in rats, shifts in MHC appear to involve long noncoding natural antisense RNA transcripts (NAT) complementary to the IIa, Iix, and IIb MHC genes (43, 44, 52). A NAT antithetical to IIa MHC, referred to as aII NAT, is transcribed from a transcription start site (TSS) proximal to the IIx MHC TSS and transcribed across the entire length of the genomic region encoding IIa MHC (43). When IIx MHC is transcriptionally upregulated with muscle inactivity, the aII NAT is also upregulated (43). This is associated with downregulation of IIa MHC transcription, leading to the conclusion that antisense transcription represses sense transcription of IIa MHC. A similar sense-antisense relationship has also been demonstrated in the regulation of IIx, IIb, Neonatal, and Embryonic MHC genes in response to spinal cord transection, muscle unloading and reloading, thyroid hormone treatment, resistance exercise, and postnatal development (21, 43–45, 52). A NAT to I/β MHC (MYH7) has been reported only in cardiac muscle and is inversely expressed with I/β MHC in rodents in response to hypo- and hyperthyroid states, diabetes, pressure overload, and postnatal development (21–23, 25, 40). The I/β MHC NAT, subsequently named Mhrt, was shown to be critical for heart function and found to be cardioprotective (25). Mechanistically, Mhrt was determined to inhibit the function of Brg1, a chromatin remodeler, which triggers pathological gene expression and cardiac myopathy (25).

The MHC NATs are a class of long noncoding (lnc) RNAs. NATs are transcribed from the opposite DNA strand of an overlapping gene (32a, 34, 38). Sense/antisense pairs are highly prevalent in mammalian genomes and serve as regulators of gene expression and chromatin remodeling (7, 8, 32a, 62, 66). Transcriptome analyses of porcine muscle determined that the expression pattern of ~40% of NATs was significantly correlated with their sense genes during fetal and postnatal development (70). Mechanisms of cis-acting NAT-mediated transcriptional repression have been well described in other cell types (12, 28, 35, 41, 69). While the involvement of NATs appears to be a regulatory strategy of the MHC genes in rats, it is unknown if humans possess a similar mechanism. Although it remains unknown how or even if the MHC NATs are spliced, an uncharacterized human gene called myosin heavy chain gene cluster antisense RNA (MYHAS; NR_125367.1) may be an additional antisense (AS) lncRNA in the cluster. MYHAS is annotated with 11 exons and spliced from a 241 kb transcript that overlaps the Neonatal, IIb, IIx, and IIa MHC genes on the antisense strand. The current annotation contains a single large intron that overlaps the three type II MHC genes.

The goal of the current study was to characterize the presence and regulation of MHC NATs in human muscle in response to altered loading conditions, specifically when sedentary young adult subjects are exposed to chronic concurrent resistance and high-intensity aerobic exercise. We also sought to determine if an AS lncRNA is spliced in accordance with the annotation for MYHAS. We report that NATs are transcribed on the antisense strand of the IIa, IIx, and IIb MHC genes in humans, and are responsive to exercise in a manner consistent with their hypothesized inhibitory influence on MHC expression. The full MYHAS transcript as currently annotated was not detected.

METHODS

Subjects.

Twenty-one subjects (12 men, 9 women), aged 22.0 ± 0.4 yr, completed the study. Anthropometric characteristics of the subjects are presented in Table 1. Additional details on a subset of these study participants (8 men, 9 women) are reported elsewhere (42). Subjects who self-reported sedentary behaviors, i.e., had not participated in structured exercise in the previous 6 mo, were chosen for inclusion in the study. The study protocol was approved by the Institutional Review Board at University of California, Irvine. All participants provided written informed consent.

Table 1.

Anthropometric characteristics

Variable	Men		Women		Training Effect	Sex Effect	Interaction Effect
	Pre	Post	Pre	Post
Body mass, kg	79.4 ± 4.3	79.0 ± 4.5	61.3 ± 3.1	61.0 ± 3.1	F1,19 = 4.002 n.s.	F1,19 = 10.794 P = 0.0039	F1,19 = 0.129 n.s.
BMI, kg/m2	26.0 ± 1.5	25.9 ± 1.5	23.6 ± 0.9	23.5 ± 0.9	F1,19 = 4.183 n.s.	F1,19 = 1.647 n.s.	F1,19 = 0.0435 n.s.
DXA lean body mass, g	57,202 ± 2,085	57,921 ± 2,223	39,471 ± 1,569	40,167 ± 1,609	F1,19 = 11.732 P = 0.0028	F1,19 = 42.756 P < 0.0001	F1,19 = 0.003 n.s.
DXA lean leg mass, g	19,314 ± 747	19,573 ± 770	12,907 ± 571	13,233 ± 535	F1,19 = 13.302 P = 0.0017	F1,19 = 44.855 P < 0.0001	F1,19 = 0.130 n.s.

Data are means ± SE. DXA, dual-energy X-ray absorptiometry; BMI, body mass index. Statistical significance was determined by repeated measures ANOVA and is reported for P < 0.05; n.s., not significant.

Exercise modality.

Exercise training was performed on a multimode exercise device (M-MED; YoYo Technology AB, Stockholm, Sweden) (10). The M-MED, which utilizes a flywheel of variable inertia, was configured for subjects to perform high intensity interval training combining both resistance exercise, in the form of horizontal squats (4 sets of 7 repetitions), and aerobic exercise in the form of rowing (4 × 4 min intervals at 90% V̇o_2max). Movements with the M-MED provide both concentric (muscles shorten against resistance) and eccentric (muscles resist lengthening due to an external load) resistance. Subjects trained for 5 wk. On alternating days, they performed 11 bouts of resistance exercise and 15 bouts of rowing.

Muscle biopsy.

Percutaneous muscle biopsy samples were obtained from the vastus lateralis (VL) muscle essentially following the technique of Edwards et al. (13, 14). The skin and subcutaneous tissues at the incision site were treated with an injection of local anesthetic (2% lidocaine). A ~6-mm longitudinal incision was made before insertion of a 6 G × 4¾ in. UCH biopsy needle into the muscle (Cadence Science, Lake Success, NY). Tissue samples were quickly blotted and dissected free of any visible adipose or connective tissue before flash freezing in liquid nitrogen, then stored at −80°C until further analysis.

RNA analysis.

Total RNA was extracted from frozen biopsy samples of VL muscle in TRI Reagent (Molecular Research Center). Extracted RNA was DNase-treated using 1 unit of RQ1 RNase-free DNase (Promega) per microgram of total RNA by incubating at 37°C for 10 min, followed by a second RNA extraction using Tri Reagent LS (Molecular Research Center).

For mRNA analysis, reverse transcription (RT) was performed on 1 µg RNA with SuperScript II (ThermoFisher Scientific) using a mix of oligo-d(T) and random primers, followed by polymerase chain reaction (PCR) with Biolase DNA polymerase (Bioline) in 25-µl reaction volumes on a Robocycler (Stratagene). Nonspliced RNA (pre-mRNA and NAT) levels were assessed with a one-step RT-PCR kit (Qiagen). The manufacturer’s protocol was followed with some modifications as described previously to avoid amplification of nonspecific transcripts (24, 43). One-step RT-PCR was performed using 100 ng total RNA and 15 pmol of specific primers in 25 µl total volume and were carried out on a Robocycler (Stratagene). RT reactions were performed at 50°C for 30 min followed by 15 min of heating at 95°C followed by PCR cycling for a varied number of cycles (19–33 cycles). Reactions were performed with a negative control condition in which the RT was carried out with a nonspecific primer followed by PCR with specific primers to ensure against generation of nonspecific amplicons of the same molecular weight as the target amplicon, as described previously (24). The annealing temperature was based on the PCR primers optimal annealing temperature. Primers used to quantify the exercise training effect of the RNA transcripts of MHC genes are listed in Table 2. A gene target description is provided with the forward (F) and its corresponding reverse (R) primer sequences. Additional RT-PCR primers used to assess sense and antisense RNA expression in the type II MHC locus are listed in Table 3 (see also Fig. 6 for depiction of primer locations). The RNA analyses performed with the primers to target both sense and antisense RNA were all performed with 34 PCR amplification cycles to more closely compare abundance levels of RNA transcribed from each DNA strand at a particular site.

Table 2.

RT-PCR primers used to target MHC genes and associated NATs

Target Description	Primer Sequence
I mRNA
Forward	5′-GGTGCGGGAGCTGGAGAATG-3′
Reverse	5′-GGGGCTTTGCTGGCACCTC-3′
I pre-mRNA
Forward	5′-CATCACGCGTTTATCCAGAAGCTCATTAT-3′
Reverse	5′-TGCCAGGTTATCTACACCCAAAATCCAC-3′
IIa mRNA
Forward	5′-GGGTACGGGAGCTGGAAGGAGAGG-3′
Reverse	5′-TTACAGAGGGAAATGACCAAAGATG-3′
IIa pre-mRNA
Forward	5′-GGTTGCCAAAATAGGAATGTGAGATAC-3′
Reverse	5′-ATTTAGCTAGATTGGTGTTGGATTGTTC-3′
aII NAT
Forward	5′-ATGATAATCCCTAGATGTTGCTTCTCCATAGAT-3′
Reverse	5′-CTCTTCACCCGCAGTTTGTTCACC-3′
IIx mRNA
Forward	5′-CAGGACACCAGCGCCCATCT-3′
Reverse	5′-TTTCTTTGGTCACCTTTCAGCAGTT-3′
IIx pre-mRNA
Forward	5′-TATCCTGCCACCATTATAGCTTTATGTTC-3′
Reverse	5′-CCACCTCCCTGGCTTGGTTGAG-3′
xII NAT
Forward	5′-TTCCTCTCCCAGCCAGGGTCCTTA-3′
Reverse	5′-AGGCAGAGTCTAATCAGCTCCAGGTGTTT-3′
IIb mRNA
Forward	5′-AAGAATATTCTCAGGCTGCAGGACTTG-3′
Reverse	5′-ACATTTCGTGCATTTCTTTGGTCACAT-3′
IIb pre-mRNA
Forward	5′-TGTGTTGTTAATTCAAGTTGCCTGGTGTAA-3′
Reverse	5′-TGTCATCAGCATTAACTTTCCCTCAACTTA-3′
bII NAT
Forward	5′-AACATTCTACAACCCCCAGCACTCA-3′
Reverse	5′-AGCAAGGCATCCCACAAACTCG-3′

MHC, myosin heavy chain; NAT, natural antisense RNA transcript.

Table 3.

RT-PCR primers used to target sense and antisense RNA expression in the MHC genes

Target Description	Primer Sequence
Emb-IIa IG1
Forward	5′-AGCTGCTATGGATGCTCGTC-3′
Reverse	5′-CATGCCCAGGAACAGGTGAT-3′
Emb-IIa IG2
Forward	5′-AGGCCTACGTATTGTGCAACT-3′
Reverse	5′-ACGCTGACCTTTTTGCTTGT-3′
Emb-IIa IG3
Forward	5′-TCAGATGCCCTGGGTTTCTTT-3′
Reverse	5′-GCTGGTATGCAGCCAAAAGT-3′
IIa-IIx IG
Forward	5′-GCTCCTGCTTCCCCTAACA-3′
Reverse	5′-TGAGTGGTCCAGATTGCTATGA-3′
IIx-IIb IG1
Forward	5′-TATGCACCTCCACTTATTTTCTATTG-3′
Reverse	5′-TATGCCCATTTATTTATCGTTTTT-3′
IIx-IIb IG2
Forward	5′-GATTTTGTTTTTAACGCCTGATGACCTATT-3′
Reverse	5′-CTCCCTGTGGATTGAAAAGATGGATG-3′
IIb-Neo IG1
Forward	5′-ATCATAGCCATTGCCATAGTGT-3′
Reverse	5′-AATTGCCCAGTGCTTTTGA-3′
IIb-Neo IG2
Forward	5′-TTGTCCTTGCTCTTCTTCGCT-3′
Reverse	5′-AGCAGAAGGGTCATTGGATACG-3′
IIb-Neo IG3
Forward	5′-ACTAACATGGCAAATGATCCAAGAGTGTAG-3′
Reverse	5′-ACCCGTTTAGTTTGTAATATTGCAGACCTT-3′
Neo
Forward	5′-CAGCCTGGATGATCTACGTGAGTGTCT-3′
Reverse	5′-ATAACATAAAGAGGCTCGGAAGAGGTCTG-3′

Fig. 6.

Sense and antisense RNA within the MYH gene cluster. The MYH genes are diagrammed individually and scaled by coding region and intergenic size. The MYH genes are transcribed right to left as depicted. Short black bars (length not to scale) correspond to approximate PCR primer location targeting sense or antisense RNA. Short black open bars correspond to the primers used to quantitate pre- and posttraining (see Figs. 3–5). Unless representative gel images are shown above, these primers were not individually utilized for both sense and antisense RNA analysis. The primers are designated as follows (targets A–S: see methods for sequences): A: Emb-IIa IG1; B: Emb-IIa IG2; C: Emb-IIa IG3; D: IIa mRNA; E: IIa pre-mRNA; F: aII NAT; G: IIa-IIx IG; H: IIx pre-mRNA; I; xII NAT; J: IIx mRNA; K; IIx-IIb IG1; L: IIx-IIb IG2; M: IIb pre-mRNA; N: IIb mRNA; O: bII NAT; P: IIb-Neo IG1; Q: IIb-Neo IG2; R: IIb-Neo IG3; S: Neo. Sense (S) and antisense (AS) RNA was targeted by complementary primer during the RT, while PCR primers are the same for S and AS targets for each target site. Bottom: representative agarose gel images of cDNA amplified at each primer location as indicated by the corresponding letter. The RT-PCR done to produce these gel images were all performed with 100 ng total RNA and PCR was done for 34 cycles.

RT-PCR primers used to validate the exon annotations of MYHAS (RefSeq NR_125367.1) are listed in Table 4 with the targeting exon provided in the target description. RT was performed with the specific reverse primer followed by PCR for 34 cycles as described above. The following forward and reverse primer pairs were utilized to target exons that straddle the large MYHAS intron 2: MYHAS E1/E2 F and MYHAS E4/E5 R, and MYHAS E1/E2 F and MYHAS E5 R. However, neither of these primer sets successfully amplified predictably sized products so the following four additional combinations of primers were tested to confirm these negative results: MYHAS E1 F and MYHAS E4/E5 R, MYHAS E1#2 F and MYHAS E4/E5 R, MYHAS E1#2 F and MYHAS E6 R, MYHAS E1#2 F and MYHAS E8 R. The following primers were employed to target AC005323.1–203 in the MYH2-MYH3 intergenic region: 5323.1–203 E1 F and 5323.1–203 E3 R.

Table 4.

RT-PCR primers used to target MYHAS

Target Description	Primer Sequence
MYHAS E1 forward	5′-AGGAAAGGCCAGAAGTGACTC-3′
MYHAS E2 reverse	5′-TTCTGGGTCTTTGGCCAGGT-3′
MYHAS E4/E5 forward	5′-TGAGGGTGCCAACAGATGTC-3′
MYHAS E6
Reverse	5′-GGTCTTGGGGGCAGTTCTAC-3′
Forward	5′-GTAGAACTGCCCCCAAGACC-3′
MYHAS E8
Reverse	5′-GGAAAGGCTACAGGTCACCC-3′
Forward	5′-GCAGCTTCTTCAGCCCTACA-3′
MYHAS E10 reverse	5′-GAGGCTCCTCTTTCCACGAC-3′
MYHAS E11 reverse	5′-CAACATTGGTTTCCTCCCAGC-3′
MYHAS E1/E2 forward	5′-TGGCCAAAGACCCAGAAGTC-3′
MYHAS E4/E5 reverse	5′-CATCTGTTGGCACCCTCAGT-3′
MYHAS E1/E2 forward	5′-CTGGCCAAAGACCCAGAAGT-3′
MYHAS E5 reverse	5′-GGGTCCCAATGTAGGCAGAG-3′
MYHAS E1#2 forward	5′-AGTGACTCCTGCCGTAGCTG-3′
5323.1-203 E1 forward	5′-GTTGGAGTATTGGGAGGCCG-3′
5323.1-203 E3 reverse	5′-AGAGGCTCCTCTTTCCACGA-3′

The amount of RNA and the number of PCR cycles were adjusted so that the accumulated product was in the linear range of the exponential curve of the PCR amplifications. PCR products were separated by electrophoresis on agarose gels and stained with SYBR Green I. The ultraviolet light-induced fluorescence of stained DNA was captured by a digital camera, and band intensities were quantified by densitometry with ImageQuant software (GE Healthcare) on digitized images. Results are reported as arbitrary units/mg muscle, i.e., the weight of the tissue used in extracting the RNA used for cDNA synthesis. We chose to report the RNA expression per unit muscle weight rather than by RNA concentration, or relative to housekeeping genes, since total RNA is increased in response to exercise training due to increased expression of ribosomal RNA, as previously discussed (29).

MHC mRNA isoform distribution.

The MHC mRNA isoform distribution was assessed by RT with oligo dT/random primers followed by PCR with primers targeting the I, IIa, IIx, and IIb MHC mRNAs. This protocol is similar to that described previously with primers targeting rat MHC (39, 67). In these PCR reactions, each MHC mRNA signal was corrected to an externally added control DNA fragment that was coamplified with the cDNA using the same PCR primer pair. This approach provides a means to correct for any differences in the efficiency and/or pipetting of each PCR reaction. A correction factor was used for each control fragment band on an ethidium bromide-stained gel to account for the staining intensity of the variably sized fragments (222 to 381 bp), as described previously (67). A forward primer designed against a sequence common to each MHC RNA near the stop codon was used in each PCR reaction with the respective reverse primer targeting either the I, IIa, IIx, or IIb MHC mRNAs. PCR primers used for this analysis are listed in Table 5.

Table 5.

PCR primers used for MHC mRNA isoform distribution assay

Target Description	Primer Sequence
MHC common forward	5′-CAGGACACCAGCGCCCA-3′
I MHC mRNA reverse	5′-ATGGGGCTTTGCTGGCACCTC-3′
IIa MHC mRNA reverse	5′-CTTCAGTCATTCCATGGCATCAGGAC-3′
IIx MHC mRNA reverse	5′-TTTCTTTGGTCACCTTTCAGCAGTT-3′
IIb MHC mRNA reverse	5′-ACATTTCGTGCATTTCTTTGGTCACAT-3′

MHC, myosin heavy chain.

Statistical analyses.

Data are reported as means ± SE. Significant differences between pre- and posttraining groups were determined by two-tailed paired t-test (GraphPad Prism 5 Software). Statistical significance was set at P < 0.05. Repeated measures analysis of variance (RM-ANOVA) was used to test for sex differences and interaction effects of sex and training response (JMP Pro 13.2, SAS Institute, Inc.). When raw data did not meet assumptions of the ANOVA (normal distribution, homogeneous variance), analysis was run on log-transformed data.

RESULTS

Myofibrillar cross-sectional area of H&E-stained VL tissue sections were significantly increased by a mean of 13% (Table 6) in response to the 5-wk combined aerobic and resistance exercise training program. Muscle biopsy total RNA concentration was significantly increased by 24% (Table 6). There was also a small, but significant, training-induced increase in lean body mass and lean leg mass (Table 1). There was no sex-specific difference in training response with these measures. Additional details regarding the functional outcomes and changes in biomarkers of a subset of these study participants (8 of the 12 men; same women) are available elsewhere (42).

Table 6.

Functional indicators of exercise training response

Variable	Men		Women		Training Effect	Sex Effect	Interaction Effect
	Pre	Post	Pre	Post
VL myofiber CSA, μm2	6,390 ± 323	7,283 ± 374	4,708 ± 277	5,245 ± 427	F1,18 = 14.699 P = 0.0012	F1,18 = 16.252 P = 0.0008	F1,18 = 0.908 n.s.
RNA, mg/g tissue	0.484 ± 0.012	0.589 ± 0.021	0.441 ± 0.021	0.557 ± 0.017	F1,19 = 57.778 P < 0.0001	F1,19 = 3.779 n.s.	F1,19 = 0.149 n.s.

Data are means ± SE. VL, vastus lateralis; CSA, cross-sectional area. Statistical significance was determined by repeated measures ANOVA and is reported for P < 0.05; n.s., not significant.

Relative MHC mRNA and protein expression.

The exercise training program induced a phenotype shift in the VL muscles of both men and female subjects. These sedentary subjects experienced a muscle loading-induced shift toward a slower contractile phenotype, based on both MHC mRNA and protein isoform analyses. We utilized a RT-PCR-based approach to quantify the proportional expression of types I-, IIa-, and IIx-MHC mRNA levels relative to the total MHC mRNA. As shown in Fig. 1, the exercise training resulted in a small nonsignificant increase in the proportion of I-MHC mRNA levels. A pronounced mRNA distribution shift was observed among the faster type II isoforms, i.e., from IIx MHC in favor of IIa MHC (Fig. 1). There was a 38% increase in the proportion of IIa MHC mRNA and a 51% decrease in the proportion of IIx MHC mRNA.

Fig. 1.

Relative myosin heavy chain (MHC) mRNA. The relative mRNA expression levels of types I, IIa, and IIx MHC are shown. Closed circles, preexercise training. Open circles, postexercise training. *P < 0.05; n = 21.

MHC isoform protein levels were assessed by SDS-PAGE and were reported elsewhere (42). There was no change detected between pre- and posttraining in I-MHC protein levels in either men (pre: 30.7 ± 4.0; post: 30.7 ± 3.6) or women (pre: 33.9 ± 2.9; post: 33.9 ± 3.4). However, there was a statistically significant protein isoform shift from IIx to IIa MHC in response to exercise training. Specifically, both the men and the women had increased relative type IIa protein levels pre to post: 53.2 ± 2.8% to 55.7 ± 3.6% and 48.6 ± 2.2 to 53.7 ± 2.4, respectively (P = 0.04). The relative type IIx protein levels were decreased in both the men and the women pre to post: 16.2 ± 3.2 to 13.6 ± 3.6 and 17.5 ± 3.3 to 12.4 ± 2.2, respectively (P = 0.04). Linear regression analysis showed that relative protein and mRNA levels of each MHC isoform were significantly (P < 0.001) correlated: I-MHC (Pearson’s r = 0.83), IIa MHC (r = 0.83), and IIx MHC (r = 0.77) (data not shown).

Type I MHC.

RT-PCR targeting type I MHC transcripts confirmed that there was not a significant absolute change from pre- to postexercise training in I MHC mRNA (Fig. 2A) or in I MHC pre-mRNA (Fig. 2B).

Fig. 2.

Type I myosin heavy chain (MHC). The expression levels of type I MHC mRNA (A) and pre-mRNA (B) are shown. Closed circles, preexercise training. Open circles, postexercise training. AU, arbitrary units. *P < 0.05; n = 21.

Type IIa MHC.

Individually assayed IIa MHC mRNA levels were significantly (P < 0.05) increased in response to combined aerobic and resistance training (Fig. 3A). Unprocessed RNA transcript, or pre-mRNA, levels were also significantly (P < 0.05) increased (Fig. 3B). We determined that aII NAT is expressed in human skeletal muscle and that it is regulated with exercise. The antisense (AS) RNA was detected overlapping the IIa MHC (MYH2) and levels of aII NAT were significantly decreased in response to the exercise stimulus (Fig. 3C).

Fig. 3.

Type IIa myosin heavy chain (MHC). The expression levels of type IIa MHC mRNA (A), IIa MHC pre-mRNA (B) and aII natural antisense RNA transcript (NAT) (C) are shown. Closed circles, preexercise training. Open circles, postexercise training. AU, arbitrary units. *P < 0.05; n = 21.

Type IIx MHC.

IIx MHC mRNA and pre-mRNA levels were significantly decreased after exercise training as compared with before training (Fig. 4, A and B). There was also AS transcription detected overlapping the IIx MHC coding region (MYH1). Levels of this AS transcript, xII NAT, were significantly increased in post as compared with preexercise training samples (Fig. 4C).

Fig. 4.

Type IIx myosin heavy chain (MHC). The expression levels of type IIx MHC mRNA (A), IIx MHC pre-mRNA (B) and xII natural antisense RNA transcript (NAT) (C) are shown. Closed circles, preexercise training. Open circles, postexercise training. AU, arbitrary units. *P < 0.05; n = 21.

Type IIb MHC.

The IIb MHC protein is not normally expressed in human limb muscle, however we discovered that the human IIb MHC gene (MYH4) is transcribed from both the sense and antisense strands (Figs. 5 and 6). We detected low levels of pre-mRNA and mRNA of human IIb MHC in the VL muscle (Fig. 5, A and B). IIb MHC mRNA levels were altered in a dissonant manner in response to exercise training. Roughly half of the subjects had increased levels while the other half had decreased levels of IIb mRNA. However, the subjects on average demonstrated a large increase in IIb pre-mRNA levels pre- to posttraining (Fig. 5B). This suggests that IIb MHC transcriptional activity is increased. However, a band corresponding to IIb MHC protein could not be detected on SDS-PAGE separation of MHC isoforms (data not shown). This indicates a lack of, or undetectable quantities of, IIb MHC protein.

Fig. 5.

Type IIb myosin heavy chain (MHC). The expression levels of type IIb MHC mRNA (A), IIb MHC pre-mRNA (B), and bII natural antisense RNA transcript (NAT) (C) are shown. Closed circles, preexercise training. Open circles, postexercise training. AU, arbitrary units. *P < 0.05; n = 21.

We also determined that bII NAT is detected by RT-PCR (Figs. 5 and 6). As with IIb mRNA alterations, there was not a consistent response to the exercise stimulus in bII NAT transcription, with approximately half of subjects responding with increased levels, and half with decreased levels of bII NAT (Fig. 5C). There was also no intrasubject consistency regarding directional alterations of bII NAT and either IIb pre-mRNA or mRNA that would indicate a predictable regulatory role played by the NAT in response to exercise.

Sex specific differences in MHC gene regulation.

The MHC RNA data reported above is reported by sex in Tables 7, 8, 9, and 10. RM-ANOVA was used to test for sex effects and interaction effects of sex and training response. There was a significant interaction effect only with IIx MHC pre-mRNA (Table 9). In this case male subjects overall had a 40% decrease while female subjects had a 14% increase on average in IIx MHC pre-mRNA in response to exercise training, and pretraining levels were 49% less in women compared with men, but posttraining levels were similar. There was a statistically significant sex effect only for aII NAT. Women had lower mean levels of aII NAT than men both pre- and posttraining (−59% and −37%, respectively) (Table 10).

Table 7.

Relative MHC mRNA

Variable	Men		Women		Training Effect	Sex Effect	Interaction Effect
	Pre	Post	Pre	Post
%I mRNA	29 ± 3	31 ± 2	32 ± 2	34 ± 2	F1,19 = 1.176 n.s.	F1,19 = 1.028 n.s.	F1,19 = 0.0616 n.s.
%IIa mRNA	49 ± 3	58 ± 2	50 ± 2	58 ± 2	F1,19 = 29.851 P < 0.0001	F1,19 = 0.0571 n.s.	F1,19 = 0.103 n.s.
%IIx mRNA	22 ± 4	11 ± 3	18 ± 4	8 ± 2	F1,19 = 16.737 P = 0.0006	F1,19 = 0.935 n.s.	F1,19 = 0.0771 n.s.

Data are means ± SE in %total myosin heavy chain (MHC) mRNA. Statistical significance was determined by repeated measures ANOVA and is reported for P < 0.05; n.s., not significant.

Table 8.

MHC mRNA

Variable	Men		Women		Training Effect	Sex Effect	Interaction Effect
	Pre	Post	Pre	Post
Type I mRNA	8.9 ± 1.53	9.6 ± 1.98	9.6 ± 1.3	12.3 ± 2.53	F1,19 = 0.710 n.s.	F1,19 = 1.328 n.s.	F1,19 = 0.159 n.s.
IIa mRNA	35.6 ± 3.63	47.7 ± 4.29	28.8 ± 1.2	42.3 ± 4.91	F1,19 = 17.200 P = 0.0005	F1,19 = 1.900 n.s.	F1,19 = 0.0755 n.s.
IIx mRNA	22.3 ± 5.09	10.2 ± 3.07	14.2 ± 4.13	7.9 ± 2.86	F1,19 = 7.894 P = 0.0112	F1,17 = 0.0866 n.s.	F1,19 = 0.0201 n.s.
IIb mRNA	38.1 ± 7.45	35.5 ± 5.61	23.4 ± 3.43	23.4 ± 4.06	F1,19 = 0.0015 n.s.	F1,19 = 0.537 n.s.	F1,19 = 0.227 n.s.

Data are means ± SE in arbitrary units/mg muscle. MHC, myosin heavy chain. Statistical significance was determined by repeated measures ANOVA and is reported for P < 0.05; n.s., not significant.

Table 9.

MHC pre-mRNA

Variable	Men		Women		Training Effect	Sex Effect	Interaction Effect
	Pre	Post	Pre	Post
Type I pre-mRNA	27.8 ± 3.8	32.5 ± 4.8	37.1 ± 5.2	37.3 ± 4.6	F1,19 = 0.433 n.s.	F1,19 = 2.501 n.s.	F1,19 = 0.286 n.s.
IIa pre-mRNA	46.5 ± 4.0	52.5 ± 5.7	42.8 ± 4.4	49.1 ± 2.4	F1,17 = 8.986 P = 0.0081	F1,17 = 0.0874 n.s.	F1,17 = 3.505 n.s.
IIx pre-mRNA	21.5 ± 3.3	12.9 ± 2.0	11.0 ± 1.6	12.6 ± 1.2	F1,19 = 3.224 n.s. (P = 0.088)	F1,19 = 1.271 n.s.	F1,19 = 14.138 P = 0.0013
IIb pre-mRNA	8.3 ± 2.0	14.2 ± 2.8	5.9 ± 1.7	13.8 ± 4.1	F1,19 = 8.931 P = 0.0076	F1,19 = 0.367 n.s.	F1,19 = 0.694 n.s.

Data are means ± SE in arbitrary units/mg muscle. MHC, myosin heavy chain. Statistical significance was determined by repeated measures ANOVA and is reported for P < 0.05; n.s., not significant.

Table 10.

MHC NAT

Variable	Men		Women		Training Effect	Sex Effect	Interaction Effect
	Pre	Post	Pre	Post
aII NAT	26.7 ± 4.7	10.7 ± 1.6	11.1 ± 3.7	6.7 ± 1.3	F1,19 = 4.938 P = 0.0386	F1,19 = 7.385 P = 0.0137	F1,19 = 0.855 n.s.
xII NAT	16.8 ± 1.7	23.7 ± 3.3	15.1 ± 2.6	19.0 ± 2.0	F1,19 = 12.752 P = 0.0020	F1,19 = 1.082 n.s.	F1,19 = 0.0452 n.s.
bII NAT	11.8 ± 0.7	14.3 ± 1.3	13.6 ± 2.3	13.5 ± 2.2	F1,19 = 1.326 n.s.	F1,19 = 0.0599 n.s.	F1,19 = 0.699 n.s.

Data are means ± SE in arbitrary units/mg muscle. MHC, myosin heavy chain; NAT, natural antisense RNA transcript. Statistical significance was determined by repeated measures ANOVA and is reported for P < 0.05; n.s., not significant.

Sense and antisense transcription in the myosin heavy chain gene locus.

We targeted sense and antisense RNA for analysis at different sites throughout the MYH gene cluster on human chromosome 17, using a subset of subjects’ RNA (see Fig. 6). PCR was performed with identical conditions and primer sets on cDNA created from separate RT reactions that reverse transcribed either sense or antisense RNA at various target locations. It is apparent from this analysis that AS RNA is more abundant than sense RNA at target sites overlapping or in intergenic regions of, the MYH4 (IIb) and MYH8 (Neonatal) genes (see Fig. 6, target site L–S).

AS RNA was detected at the 5′- and 3′-ends overlapping the IIb MHC (MYH4) coding region, as well as in the flanking regions (see Fig. 6, target sites L–P). In addition, we detected abundant AS transcript throughout the intergenic region between MYH4 and MYH8. The intergenic DNA sequence of the 21.3 kb region between MYH4 and MYH8 is found to consist of 65% total interspersed repeats (repeatmasker.org), which limits the options for designing specific PCR primers that are unique to this gene locus. However, we did reliably detect specific AS RNA in regions corresponding to ~1 kb and ~14 kb downstream of MYH4 (see Fig. 6, target sites P and Q) and also in the proximal promoter region of MYH8 (~-1.5 kb; see Fig. 6, target R) and overlapping MYH8 (approximately +3 kb; see Fig. 6, target S). Thus antisense transcript was determined to be present in all regions complementary to nonrepetitive sequence of the IIb-Neonatal intergenic region.

AS RNA is also in greater abundance than sense RNA in the region upstream of MYH2 at −5 kb and −9 kb relative to the MYH2 TSS (see Fig. 6, targets B and C). At +4.5 kb from the 3′-end of MYH3, there was AS and sense RNA measured in similar abundance (see Fig. 6, target A). Target site D (Fig. 6) presented with relatively high sense RNA as compared with AS in the MYH2-MYH1 intergenic region.

The prevalence of AS RNA demonstrated throughout the MYH gene cluster lends support to a recently deposited lncRNA annotation (RefSeq NR_125367.1), named myosin heavy chain gene cluster antisense RNA (MYHAS), which spans the entire MYH cluster (see Fig. 7). We sought to validate the exon annotations of MYHAS with RT-PCR primers designed to tile across each of the 11 exons. Four different amplicons of the correct predicted size from four PCR primer pairs targeting exons 4–11 appear to validate the MYHAS annotation in the MYH3-MYH2 intergenic region (see Fig. 7). PCR products from another primer pair also appear to validate the annotation of exon 1 and exon 2 that partially overlaps MYH8. However, six different primer designs targeting various exon combinations (see methods) that straddle the large intron 2, which overlaps MYH4, MYH1, and MYH2 failed to amplify products of the predicted size, indicating that the full-length MYHAS is not expressed in VL muscle. We did successfully amplify a predictably sized PCR product of AC005323.1-203, targeting exon 1 and exon 3 (data not shown), which is in the MYH2-MYH3 intergenic region.

Fig. 7.

Skeletal muscle MYH gene cluster on human chromosome 17. The image shows the MYH genes on the reverse genomic strand (protein coding genes as shown are transcribed from right to left), with MYHAS (RefSeq NR_125367.1) depicted on the forward strand. MYHAS, as annotated, is 3,952 bp and spliced into 11 exons from a 241 kb primary transcript. MYHAS exons are numbered and depicted at an enlarged scale below the MYH genes. MYHAS is shown with the first intron partially overlapping MYH8 (Neonatal) and intron 2 overlaps all of the MYH4 (IIb), MYH1 (IIx), and MYH2 (IIa) genes. Exons 3–11 are transcribed from within the MYH2-MYH3 (Embryonic) intergenic region. Representative RT-PCR gel images are shown with the forward/reverse primers targeting specific MYHAS exons indicated above the gel image (A–E). Gel images of PCR products that were not detected are not shown. Image created in Ensembl with data extracted from release version 91 (2).

DISCUSSION

In the present study we examined VL muscle from sedentary human subjects before and after a 5-wk-long exercise regime of combined aerobic and resistance exercise. This exercise stimulus invoked a shift in the expression profile of MHC isoforms. This shift was primarily confined to the IIa and IIx MHC isoforms, with exercise inducing increased IIa MHC and decreased IIx MHC. We have previously shown evidence in animal models supporting a regulatory mechanism to mediate coordinated shifts between IIa and IIx MHC that involve antisense transcription of the IIa MHC gene (43, 45). We report herein that this regulatory phenomenon is conserved in humans.

Antisense RNA transcription with IIa to IIx MHC shift in human muscle in response to high-intensity exercise.

The IIa MHC antisense transcriptional product, aII NAT, is downregulated with exercise training, while the IIa MHC mRNA is upregulated in a manner consistent with the hypothesized inhibitory effect of the aII NAT such that its downregulation acts to relieve its repressive influence on IIa MHC expression. NATs to the IIx and IIb MHC genes were also discovered. The xII NAT may similarly inhibit IIx MHC expression. We observed that IIx MHC mRNA and pre-mRNA levels were decreased with exercise while xII NAT transcript levels were increased. MHC transitions also followed a reciprocal pattern of expression in rat muscle, with inactivity of rat slow muscle resulting in decreased IIa MHC and increased aII NAT, and exercise in rat fast muscle resulting in increased IIx MHC and decreased xII NAT (43, 52). The role of the bII NAT is not clear, given that there is no role for the IIb MHC isoform in human VL muscle and thus no need for its regulation.

As we have described previously, the function of the MHC NATs is attributed to coordination of the expression of MHC genes that are proximally clustered in a common locus (21, 43, 44, 52). There are two of these loci on separate chromosomes consisting of the β-α-MHCs, the expression of which is coordinated in cardiac muscle, and the Embryonic-IIa-IIx-IIb-Neonatal-Extraocular MHC loci, the expression of which is coordinated in skeletal muscle during postnatal development and in response to altered environmental stimuli, such as muscle loading/activity. We previously presented evidence suggesting that IIa and IIx MHC are coordinated such that IIx MHC upregulation with muscle inactivity and unloading in rats concomitantly activates the aII NAT from a bidirectional promoter (43, 45). The aII NAT is subsequently thought to repress IIa MHC, thus achieving a stoichiometrically balanced shift from IIx to IIa MHC at the RNA level. A similar suggested mechanism of coordination between IIx and IIb MHC was also previously reported in response to resistance exercise training in the rat (52). In the rat the TSS of the aII and xII NATs are both within 1 kb of the TSS of the IIx and IIb MHC genes, respectively, consistent with the structural elements of a typical bidirectional promoter (43, 52, 64).

The aII/IIx bidirectional promoter may also be evident in the sex differences and training responses of expression levels of IIx MHC pre-mRNA and aII NAT (see Tables 9 and 10). Women had significantly lower levels of both IIx MHC pre-mRNA and aII NAT than men in the pretraining muscle samples (−49% and −59%, respectively, P > 0.05). The pre- to post- training responses were such that men had a significant decrease in IIx MHC pre mRNA (−40%), while women did not significantly change (+14%), and men overall had a significant decrease in aII NAT (−60%) while the female-specific decrease did not reach significance (−39%). Thus, as was previously reported in animal models, these data may indicate coregulation of IIx MHC pre-mRNA and aII NAT from a bidirectional promoter. Moreover, the surprising increase reported in IIb MHC pre-mRNA, together with the concurrent increase in xII NAT with exercise training in both sexes could be due to transcriptional activation of the IIb/xII bidirectional promoter. If this bidirectional promoter is functional in humans, this would provide one rationale for the selection pressure that has largely conserved the sequence similarity of the IIb MHC/xII NAT promoter region in humans and other species (e.g., with rat, mouse, monkey) despite the IIb MHC isoform’s essentially nonfunctional status in limb muscle (52).

New insights on IIb MHC expression in human skeletal muscle.

An activating transcriptional response of the gene encoding IIb MHC (i.e., MYH4) in these exercising human subjects was observed. It has been previously documented that IIb MHC mRNA appears to be expressed only in the masseter muscles in the human jaw and minimally in the external oblique (30). IIb MHC mRNA and protein was also reportedly detected in human extraocular muscles (60) and in laryngeal muscles (50, 57). Numerous other studies have failed to detect IIb MHC mRNA or protein in any other human muscles using a variety of methods, although not with RT-PCR (15, 30, 36, 48, 58, 59, 68). Harrison et al. concluded that IIb MHC mRNA was only significantly expressed in human limb muscle under abnormal conditions, such as Duchene muscular dystrophy (27). IIb MHC protein was not, however, detected in these muscles where IIb MHC mRNA was detected, suggesting that there is no functional role for IIb MHC in human limb muscle.

We report that the IIb MHC gene is not transcriptionally silent in human VL muscle; low levels of IIb MHC mRNA and pre-mRNA were detected. Evidence of exercise-induced regulation is also reported with a significant increase in IIb MHC pre-mRNA levels after exercise training. RT-PCR provides an approach to amplify low-abundance transcripts with higher sensitivity than methods utilized in the instances referenced above. Harrison et. al. also detected low levels of IIb mRNA in vastus medialis muscle of normal sedentary humans with RT-PCR (27). Studies of the IIb MHC promoter concluded that the lower promoter activity in human as compared with that of mouse and domesticated pig, which both express IIb MHC in abundance, is largely attributable to subtle (i.e., 1–3) nucleotide differences that effectively prevent binding by transcriptional activators (5, 27). Perhaps, then, transcriptional repressors are unopposed in maintaining an effectively repressed state at the IIb MHC promoter. How is this state maintained? Further research will be required to determine the nature of inhibition and the regulatory mechanism of the bII NAT, but it could, as other NATs have been determined to, act to maintain the repressed state of IIb MHC, thus relegating the fastest MHC isoform to nonfunctional status in humans. Our findings reveal a distinct difference between humans and rats in the architecture of bII NAT transcription. We show that in humans bII NAT is detected at regions overlapping both the 5′- and 3′-ends the IIb MHC gene, including its promoter region, while in the rat bII NAT transcription terminates in the proximity of intron 15, or ~9 kb from the IIb MHC TSS (44). In light of this species difference, it is conceivable that the NAT may recruit repressive chromatin-modifying enzymes to the IIb MHC promoter in human muscle to maintain an effectively repressed transcription state.

Other differences exist between humans and rats regarding bII NAT transcription. In rat muscle we previously mapped the bII NAT TSS to a site ~5 kb downstream from the poly(A) signal site of the IIb MHC gene, which is also proximal to a 700-bp promoter region with high sequence conservation (>70% sequence similarity) across a broad range of mammalian species, including humans (44). However, it is not apparent that this evolutionarily conserved region (ECR) demarcates the promoter of bII NAT in humans, since antisense transcript was detected at several sites to the 5′-side of the ECR corresponding to the rat bII NAT TSS. Indeed, in contrast to the rat, all regions examined with RT-PCR in the human IIb-Neo intergenic region, plus the Neo MHC coding region, were revealed to have antisense RNA. Another species difference is that the bII NAT is not detectable in normal adult rat skeletal muscle (43) and is markedly expressed only during the first few weeks of postnatal development during the developmental shift from neonatal to IIb MHC in fast skeletal muscle (44). Whether these differences may explain why IIb MHC is not normally present in human muscle, or in most other large mammals for that matter, in contrast to rats and other small mammals, remains to be determined. It should also be considered that the sense and antisense RNA transcripts from the IIb MHC gene represent merely transcriptional read-through related to activity elsewhere in the locus.

Validation of spliced antisense overlapping transcripts.

We have previously reported that under certain conditions, in skeletal or cardiac muscle, that any of the MHC genes can be transcribed in the antisense direction (21, 43, 44, 52). With regard to the Embryonic-IIa-IIx-IIb-Neonatal-Extraocular MHC gene-coding cluster, our previous analyses in the rat indicate there are separate and distinct NATs for each MHC that are independently regulated and the aII, xII, and bII NATs each have a TSS distinct from adjoining NATs (43, 44, 52). There may also be AS variants within and overlapping multiple genes of the cluster. MYHAS, as currently annotated based on human cDNA clones, is a 3,952-bp product spliced from a 241 kb transcript that spans the gene cluster (See Fig. 7). Though apparently not further validated after the initial identification from transcriptome-wide screening (33), we were able to validate by RT-PCR that MYHAS exons 1 and 2, and exons 3 through 11, appear to be correctly annotated. However, the 161 kb intron 2, which overlaps the genes encoding IIa, Iix, and IIb MHC, is likely not present in human skeletal muscle, as attempts to amplify a product with exons spanning this intron were not successful (see methods for primers utilized). MYHAS was reportedly measured by RNA-Seq transcript profiling in skeletal muscle and other tissues, e.g., breast, thyroid, esophagus, prostate, and testis (11, 16). Exons 7–10 of MYHAS (as annotated) appear to overlap a previously described AS lncRNA, called linc-MYH, which was implicated in playing a regulatory role in the expression of an array of slow and fast fiber-type genes in the mouse (53). Any function specific to MYHAS, how it may be related to the NATs overlapping the MYH genes, and how linc-MYH fits into the regulatory scheme, remains to be determined.

In conclusion, we report that a regulatory phenomenon involving long noncoding antisense RNA and MHC shifts exists in human skeletal muscle. Exercise training involving a combination of aerobic and progressive resistance training over 5 wk induced a typical IIx to IIa MHC shift. This was associated with a corresponding increase in transcription on the antisense strand of the IIx MHC gene and a decrease in antisense transcription of the IIa MHC gene, suggesting an inhibitory mechanism governed by NATs. The absence of expression of IIb MHC in human muscle is paralleled by the presence of IIb MHC antisense transcript, which is the opposite expression pattern to that of small mammals. Efforts to develop interventions or therapeutic targets aimed at ameliorating loss of muscle function and increased fatigability associated with disuse atrophy should consider this sense-antisense relationship as a potential source to impose regulatory control.

GRANTS

Funding was provided by the National Space Biology Research Institute (NSBRI NCC 9-58-70, MA01601 to G.R.A. and PF02104 to C.E.P.) and the National Institutes of Health (NIH) Grant UCI CTSA UL1 TR000153. Support was also provided by Mercer University School of Medicine to C.E.P.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

C.E.P., F.H., K.M.B., and G.R.A. conceived and designed research; C.E.P., F.H., and L.P.C. performed experiments; C.E.P., F.H., and L.P.C. analyzed data; C.E.P., F.H., T.O., L.P.C., K.M.B., and G.R.A. interpreted results of experiments; C.E.P. prepared figures; C.E.P. drafted manuscript; C.E.P., F.H., T.O., L.P.C., K.M.B., and G.R.A. edited and revised manuscript; C.E.P., F.H., T.O., L.P.C., K.M.B., and G.R.A. approved final version of manuscript.