INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

DYSTROPHIN SERVES AS A membrane-associated protein that interfaces with cytoskeletal actin networks, signaling, and transmembrane proteins (14, 22). Inherited abnormalities of the dystrophin protein have been associated with different phenotypes. For example, patients with Duchenne muscular dystrophy (DMD) (complete loss of function with dystrophin abnormality) lose the ability to walk by the age of 12 yr and eventually succumb to respiratory failure by the second or third decade of life, whereas Becker muscular dystrophy patients (partial loss of function with dystrophin abnormality) are still able to walk at the age of 15 yr and typically do not undergo respiratory failure until after the fourth decade (6). In contrast, the muscles of DMD patients appear to undergo continuous cycles of degeneration and regeneration, with a gradual failure of regeneration. Histological examination of DMD muscle fibers has created the concept that the manifestation of this gene defect triggers a pathological cascade, including the following: 1) membrane fragility, 2) aberrant calcium homeostasis, 3) mechanical susceptibility to injury, 4) activated degradative mechanisms (e.g., calpain), 5) fibrofatty replacement, 6) failure of regenerative and/or repair systems, and 7) others, including vascular ischemia (6, 14, 22). However, the absence of dystrophin is not always this devastating, especially in other mammalian species.

A genetic homolog model of DMD is the mdx mouse, first identified as dystrophin deficient in 1989 (21). The causative mechanism is a point mutation in exon 23, resulting in a premature stop codon (15). The mdx mouse has the critical hallmarks of DMD, including loss of dystrophin-associated proteins, a susceptibility to contractile-induced damage, elevated serum creatine kinase, and muscle-fiber degeneration (34). However, whereas the mdx mouse shows a complete loss of function of the dystrophin protein, it has a mild clinical course compared with DMD, with an early episode of widespread skeletal muscle necrosis at 3-4 wk of age. The mdx muscle then shows subsequent regeneration followed by a relative resistance to further degeneration of skeletal muscle for a period of some months, with only minor physical impairments (decreased voluntary wheel-running) (9, 47), although increased disabilities do appear after 11 mo of age. The progression of the murine disease appears to be slower, with very delayed fibrofatty infiltration (13, 31). It is our hypothesis, previously stated by Infante and Huszagh (22), that the mdx mouse staves off severe disability with differential gene effector(s) and/or responder(s) that protect 16-wk-old mdx muscle from damage, and that this differential expression will show discordant expression profiles with DMD. The goal of this report was to use a global mRNA expression profiling to compare and contrast the response of human and mouse skeletal muscle to the same biochemical defect (dystrophin deficiency). We focused on mdx muscle that had successfully regenerated (e.g., static stable phase of the disease) and compared these findings with a previous report of 6- to 9-yr-old boys with active DMD disease (10). The differentially expressed transcripts are potential candidates for conferring protection to murine dystrophin-deficient muscle and are, therefore, possible therapeutic targets for modulation in the progressive skeletal muscle dystrophinopathies in humans.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Normal [C57Bl10 (black 10)] and mdx [C57Bl10 (black 10) mdx/mdx] mice were bred by Dr. Joseph A. Granchelli (University of New York at Buffalo, New York) from the original breeding pairs supplied by Jackson Laboratories (Bar Harbor, MA).

RNA processing. Sixteen-week-old males were killed by cervical dislocation. Sixteen weeks of age was selected to target mdx skeletal muscle in its relative "benign" steady-state condition, after the acute degeneration-regeneration phase at ~3-4 wk of age. Gastrocnemius muscles were excised and flash-frozen in isopentane cooled with liquid nitrogen. Both gastrocnemius muscles from a single mouse formed one observation, with n = 4 for control and n = 4 for mdx groups. All muscle samples were processed in parallel and hybridized such that muscle mRNA from each individual animal was applied to an individual array. Muscles were powdered in liquid nitrogen with mortar and pestle and then in TRIzol (GIBCO BRL, Gaithersburg, MD) by using a Polytron (Kinematica, Lucerne, Switzerland) on setting 7 for three pulses of 15 s. Total RNA was extracted according to the guanidine thiocyanate method of Chomczynski and Sacchi (11). Poly(A)⁺ mRNA was isolated from total RNA by using OligoTex columns (Qiagen, Valencia, CA). One microgram of mRNA was allowed to hybridize with an oligo T7-(dT)24 primer for cDNA synthesis (Genset Oligos, Huntsville, AL) followed by first- and second-strand synthesis with Superscript Choice (GIBCO BRL). The resulting cDNA was transcribed in vitro with biotinylated nucleotides from a BioArray high-yield kit (Enzo Diagnostics, Farmingdale, NY). A final cleanup of the cRNA was performed with an RNeasy kit (Qiagen). Biotinylated cRNA samples were hybridized to Affymetrix murine genome U74Arev2 arrays and analyzed by fluorescent intensity scanning according to Affymetrix protocols (Affymetrix Expression Analysis Technical Manual). The hybridization and scanning of the arrays was performed in the laboratory of Dr. Eric P. Hoffman at Children's National Medical Center (Washington, DC).

GeneChip array analysis. Of the 12,488 gene sequences offered on the Affymetrix murine genome U74Arev2 GeneChip array, ~6,000 have been functionally characterized in the mouse UniGene database (build 74). Additionally, ~6,000 expressed sequence tag (EST) clusters were also analyzed on these arrays. The probe set for each transcript consists of 16 different, perfectly matched (complementary) 25-base segments corresponding to different regions along the length of a transcript. Similarly, 16 mismatched pairs, which do not complement perfectly the transcript's sequence, containing one incorrect base, are used as a measure of nonspecific binding. The mismatched probes' fluorescence intensity is subsequently subtracted (like background) from the perfect-match intensity to yield a more accurate reading of a transcript's relative expression. All preliminary analyses of each array were carried out with Microarray suite 4.01 (Affymetrix). The average difference intensity information for each gene was exported into Excel or Access (Microsoft, Redmond, CA). Statistical analyses were done in Excel. Some genes did not show measurable expression, and their average-difference intensity values were near or below zero. For those that were less than zero, average intensity values had to be reset to a small, positive value to perform statistics. We chose to reset such values to 20 average difference-intensity units. Others have performed similar corrections (33, 46).

Statistical methods. First, an unequal variance, two-tailed t-test was used to compare transcript expression intensities between control and mdx groups (values are expressed as means ± SE; n = 4) for 12,488 mRNAs. Second, the FDR procedure (3) was used as a criterion for deciding which t-test results should be called significant. By setting the FDR criterion at 0.01, 1% (on the average) of the results called significant by the t-test may not be true rejections of the null hypothesis.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The wet weight of the gastrocnemius in mdx mice (158 ± 9 mg) did not differ (21% greater, P = 0.13) from that in control mice (130 ± 13 mg). Sixteen-week-old mdx mice have previously been reported to have 17, 21, and 0% larger extensor digitorum longus, soleus, and plantaris muscles, respectively, than age-matched mice (20).

Extracted RNA per milligram of muscle wet weight was twice as high (P = 0.0004) in the mdx (2.4 ± 0.12 µg RNA/mg muscle) than in the control (1.2 ± 0.11 µg RNA/mg muscle) muscle, verifying an earlier report (29). Equal quantities of mRNA from each mouse's gastrocnemius muscles were applied to each of the GeneChip microarrays. Results in Table 1 are thus the relative amount of mRNA per microgram of RNA extracted from each mouse's muscle and hybridized on an array. Because extracted RNA per whole gastrocnemius muscle was more than twice (P = 0.0006) in mdx (371 ± 15 µg RNA/whole muscle, n = 4) than in the control (155 ± 7 µg RNA/whole muscle, n = 4) group, the fold increase of mRNAs for the entire mdx muscle would actually be greater than fold increases reported in Table 1. Thus the underestimated fold changes in Table 1 are due to their differences in RNA abundance, which differed between control and mdx muscles. In four control and four mdx muscles, 5,304 ± 111 and 5,977 ± 430 genes, respectively, were detected as "present" above background with the use of Affymetrix Microsuite 4.01 software on the U74Arev2 GeneChip arrays. One muscle sample in each group was tested in duplicate to verify chip-to-chip reproducibility. All raw data and interpretation files are available on the Children's National Medical Center Microarray website (http://microarray.cnmcresearch.org).

With the use of a FDR of 0.01, 137 transcripts had a P < 0.0002 in the unequal-variance t-test and a twofold change, which were considered significant (Table 1). Eighty-six percent (i.e., 124) of the significant differences represented in Table 1 were upregulated transcripts in mdx muscle, compared with control, and 14% (i.e., 23) were downregulated. With the FDR set at 1%, 137 results were found significant. Therefore, we can expect approximately 1 of these 137 results (on the average) to be wrongly rejected, a true null hypothesis. One hundred fifteen of these mRNAs were associated with known descriptions or attributes indicating cell-type specificity and/or function, whereas 22 were ESTs. Of these ESTs, 14 were homologous or orthologous to known genes in other species, and 8 had no known similarities. No mRNAs in Table 1 were found to have a discordant directional change compared with 13 published analyses for mRNAs and proteins from mdx muscles. In agreement with previous reports, the results show increases in mRNAs from mdx muscles for myogenin (44), ₂-tubulin (44), H19 (44), lysozyme M (16, 44), ₁(III) procollagen (18, 44), and cathepsin B (16), whereas decreases in mRNAs for S-adenosylmethionine decarboxylase (44) and myostatin (44) have also been reported. The results in Table 1 also agree with the immunohistochemical level increase reported for tenascin C (41), fibronectin (26), myogenin (23), cathepsin B (40), cathepsin H (40), and cathepsin L (40). Comparisons to published human DMD mRNA analyses identified novel discordant directional changes in mRNA for -thymosin, calponin, follistatin-like, myogenin, guanidinoacetate methyltransferase, and mast cell chymase (Table 2). Fold changes for many genes for metabolic proteins were less in 16-wk-old mdx than control muscles (Table 3).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Despite thousands of publications on dystrophin, the reasons for the progression of the human disease, differential muscle involvement, and different phenotypes in various species are not fully understood. Sander et al. (39) described this dystrophin mechanistic mystery as follows: "Despite a wealth of recent information about the molecular basis of DMD, effective treatment for this disease does not exist because the mechanism by which dystrophin deficiency produces the clinical phenotype is unknown." The strategy employed in the present study was to identify those mRNAs that were differentially expressed in a mouse model (mdx) without full-length dystrophin protein, but whose skeletal muscle at 16 wk of age shows successful regeneration. One hundred thirty-seven mRNAs were found to be different from age-matched normal mice of the same strain. A second strategy was then to compare the 137 differentially expressed mdx transcripts with those previously found discordant in DMD muscles (10) (i.e., having altered expression in mdx, but not in DMD, muscle at a time when a relative rescue from further degeneration was occurring in the mdx muscle). The goal was to identify candidate genes that may confer protection against dystrophin-deficiency-induced myofiber damage. Our initial effort may be limited by the comparison of a single-time-point mouse microarray data against human DMD microarray data because of differences in the following: species, age, temporal disease course, muscle specificity, posture, diet, nocturnal habit, and statistical methods.

Dystrophin and its associated proteins function to link the intracellular actin cytoskeleton of muscle to laminins in the extracellular matrix. The actin-filament binding activity of dystrophin has been well characterized, where multiple actin binding sites cause a side-by-side alignment of actin filaments along dystrophin and protect actin filaments from depolymerization in vitro (35, 36). This interaction leads to a strong association of -actin filaments with the plasma membrane, but this association is completely lost with dystrophin deficiency (37). Thus our observations of the differential upregulation in the more benign mdx than in the more devastating DMD of some actin-associated mRNAs whose proteins regulate actin polymerization suggest a new hypothesis of a potential rescuing role by the cytoplasmic actin remodeling, which can be subsequently tested. Transcripts for actin-related protein 2/3 (subunit 4), -thymosin, and calponin were all increased in the mdx muscles but, although present on the human microarray, were not increased in the DMD muscle (Ref. 10; Table 2). The actin-related protein 2/3 complex is the cellular factor that generates new actin filaments (branching) in a site-directed, signal-controlled fashion at the leading edge of motile cells (4, 28) and forms identical branches in vitro (32). Alteration of the bimodal spatial stability by cytoskeletal actin network remodeling with branching processes near the cell membrane has been proposed by Sambeth and Baumgaertner (38) to be essential for the induction of a spontaneous breaking of isotropic cell motion observed in processes such as the amoeboid crawling of animal cells in advancing neural growth cones. Supporting this actin remodeling postulate is the 2.3-fold increase in -thymosin mRNA, as -thymosin binds to actin monomers, facilitating their polymerization into filaments (14). Future experiments at the protein level would be required to test this hypothesis of whether these mRNA differences are reflections of the differences in the severity of the phenotype in the absence of intact dystrophin protein or due to differences between species.

Myostatin mRNA in mdx muscle was only 25% of the level found in controls (Table 1). Tkatchenko et al. (44) previously detected myostatin mRNA downregulation in the mdx mouse using suppression subtractive hybridization but made no mention of its possible significance. We have also observed decreased myostatin mRNA in DMD skeletal muscle (unpublished observations). Myostatin is a transforming growth factor- family member that acts as a negative regulator of skeletal muscle mass, because mice without this gene exhibit hypertrophy (27). This adaptation might play some role in the sporadic vs. widespread fiber hypertrophy and/or maintenance of muscle mass and functional rescue of mdx muscle, which is known to be a factor in the compensatory strength of mdx mice. The enhanced regenerative capacity of mdx muscle is in concordance with the upregulation of myogenin mRNA, a key myogenic differentiation gene for skeletal muscle fiber development (23). Future experiments should test the candidate genes identified here at the protein level and should functionally test their relative impact on dystrophic muscle.

Other mRNAs were differentially expressed between mdx and DMD muscles. For example, mast cell chymase mRNA was increased fivefold in mdx muscle (Table 2) but was unchanged in DMD muscle (10). At the 16-wk-old age selected, mdx muscle essentially has no fibrosis (7) compared with the 6- to 9-yr-old subjects with DMD. Mast cell chymase activates matrix metalloproteinases (which degrade the extracellular matrix) and processes precollagenases, whose product degrades collagen and cleaves fibronectin (43). Mast cell inhibitors resulting in extracellular matrix degradation provide a potential mechanism for improving mdx muscle strength (19). Whereas guanidinoacetate methyltransferase mRNA increased 3.1-fold in mdx muscle, it was decreased fourfold in DMD muscle (10). As guanidinoacetate methyltransferase catalyzes the last step in creatine biosynthesis, we speculate that its increase in mdx muscle could reflect a crucial cellular response that increases at an mRNA level (and potentially at the protein level) in an attempt to compensate for the leak and loss of creatine kinase (30). Markers of apoptosis (increased caspase, decreased 70-kDa heat shock protein) and increased protein degradation (cathepsins, lysosomal proteins) occurred in the 16-wk-old mdx muscles, suggesting continuing remodeling.

In muscle biopsies from male 6- to 9-yr-old Duchenne patients, Chen et al. (10) observed a greater than twofold downregulation of 26 mRNAs for proteins that are involved in mitochondrial function and energy metabolism, which they suggested indicated a generalized mitochondrial dysfunction and "metabolic crisis." Mitochondrial dysfunction has been reported previously by using a variety of assays in both human dystrophy patients and animal models (1, 17, 25). Another difference between mdx and DMD muscles is the amplitude of decrease of those transcripts for mitochondrial and metabolic enzymes changed in the 16-wk-old mdx muscle (Table 3). Previously in DMD muscle, mitochondrial and metabolic transcripts have been reported to decrease two- to sixfold (10), whereas many of the same mRNAs showed less than a twofold decrease in mdx muscle (Table 3 and Ref. 17). However, because RNA per gram of 16-wk-old mdx muscle was twice that of age-matched controls, the estimated concentration of mRNA for mitochondrial transcripts per gram of 16-wk-old mdx muscle is essentially unchanged (1/2 mRNA/RNA times 2× RNA/g = unchanged mRNA/g), in contrast to the decrease found in DMD muscle (10).

The extensive signaling and cell receptor mRNAs (29 different transcripts) altered in mdx muscle call attention to far more complex signaling changes to produce these differences in mRNA responses due to the loss of functional dystrophin expression than heretofore appreciated. A number of these mRNAs showed largefold changes. For example, the mRNA of the cell-surface glycoprotein CD53 increased 71-fold in mdx muscle. CD53 is a transmembrane-4 superfamily (TM4SF) protein (see Ref. 48 for references). TM4SF proteins can regulate cell signaling, motility, and tumor cell metastasis. TM4SF proteins also tend to assemble into protein complexes at the plasma membrane, where they may recruit growth factor ligands and phosphatidylinositol 4-kinase into proximity with integrins. The Src-associated adaptor protein (RA70) mRNA increased 41-fold in mdx muscle. RA70 is highly homologous to human Src kinase-associated phosphoprotein (SKAP55) and, according to Kouroku et al. (24), may play an essential role in the Src signaling pathway in various cells. The p67^phox mRNA increased 25-fold in mdx muscle. The assembly of a membrane-associated flavocytochrome b₅₅₉ with the cytosolic proteins p47^phox and p67^phox and the small GTPase Rac (1 or 2) activate the superoxide (superoxide anion)-generating NADPH oxidase of phagocytes (1). These changes in signaling transcripts may provide new directions for investigation.

The present approach identified mRNAs differentially expressed in only mdx or DMD muscles. Because both lack appropriate expression of dystrophin protein with different phenotypes, mRNA differences between mdx and DMD muscles provide the basis for testable hypotheses as to how mdx muscle is salvaged from the early deleterious fate of the DMD muscle. The number and varied gene function of the identified mRNAs differentially expressed in mdx muscle suggest that there may be a complex interplay of groups of genes that may provide key insights elucidating the more benign and less devastating pathological mechanisms involved with mouse mdx, unlike human dystrophin-deficient muscular dystrophy.