Physiol Genomics 18: 184–195, 2004.
First published May 11, 2004; 10.1152/physiolgenomics.00222.2003.
Conserved and muscle-group-specific gene expression patterns shape postnatal
development of the novel extraocular muscle phenotype
Georgiana Cheng,1,3 Anita P. Merriam,1 Bendi Gong,3
Patrick Leahy,4 Sangeeta Khanna,1 and John D. Porter1–3
Departments of 1Neurology and 2Neurosciences, 3Visual Sciences Research Center, and the 4Comprehensive
Cancer Center, Case Western Reserve University and University Hospitals of Cleveland, Cleveland, Ohio 44106
Submitted 22 December 2003; accepted in final form 7 May 2004
skeletal muscle; hindlimb; microarray; allotype
PATTERNED COVARIATION of contractile protein isoforms and
energetic mechanisms defines four muscle fiber types (I, IIA,
IIX/D, and IIB) that are highly conserved among mammalian
skeletal muscles (4, 8, 52, 77). Fiber type content is, in turn, a
determinant of muscle group heterogeneity in contraction
speed and fatigue resistance. Despite the value and durability
of the fiber type concept, muscles still exhibit inherent differences not easily explained by variations in relative content of
stereotypical fiber types. An integrated model of skeletal mus-
Article published online before print. See web site for date of publication
(http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: J. D. Porter, Dept. of
Neurology, Case Western Reserve Univ., 11100 Euclid Ave., Cleveland, OH
44106-5040 (E-mail: [email protected]).
184
cle biology requires an understanding of the breadth, causes,
and consequences of phenotypic variation among skeletal muscle groups.
The muscle allotype concept arose as a framework to account for the breadth of phenotypic diversity available to
skeletal muscle (23, 24). Allotype identities appear to be
resident in muscle precursor cell populations, but the expression of allotype-specific traits apparently is dependent upon
interactions between precursor cells of specific lineage with
motoneuron pools that provide appropriate activation patterns.
Three allotypes were defined on the basis of their potential to
express specialized myosins: masticatory (super fast myosin),
extraocular muscle (EOM; Myh13), and limb (no allotypespecific myosins). Allotype heterogeneity is, however, not
limited to myosin heavy chain content. EOM, for example, is
further adapted due to its eye movement role. Some central
concepts in muscle biology, such as the established fiber type
classification schemes and M-line/creatine kinase system, do
not apply to EOM (1, 53, 63). Instead, EOM comprises six
allotype-specific fiber types, including two non-twitch, multiply innervated types, and expresses embryonic, neonatal, cardiac, and tissue-specific protein isoforms that are atypical of
adult skeletal muscle (27, 30, 44, 53, 55, 71, 75, 81). Recent
expression profiling studies have further shown that adult EOM
is fundamentally distinct from the limb and masticatory muscle
allotypes (12, 18, 36, 60); these data suggest that allotype
specificity may be defined by more than simple differences in
myosin heavy chain expression patterns.
Myogenic mechanisms underlying allotype specificity are
poorly understood. In limb, distinct myoblast populations and
regulatory pathways give rise to type I and II myofibers (9, 45,
73), and hypaxial and epaxial muscle precursors differentially
activate regulatory genes (16, 17). Since the allotype concept
highlights differences between craniofacial and spinal muscles,
rostrocaudally distributed regulatory cascades may be mechanistic in muscle divergence (46, 51, 76). As yet, few transcription factors have been linked to muscle group identities (e.g.,
Lbx1-forelimb extensors, Mox2-appendicular muscle, En2 and
MyoR/Tcf21-masticatory muscle, and Pitx2-EOM) (14, 19, 39,
40, 42, 72). Other than these data, there has been little research
geared toward understanding the full scope of developmental
processes behind skeletal muscle allotypes.
Allotype heterogeneity has significant consequences. Differential responses to inherited metabolic and neuromuscular
diseases are seen both between and within muscle allotypes.
Most myopathies target the limb allotype but have an unexplained predilection for proximal muscles, and the rarer distal
myopathies exhibit their own distinctive patterns of muscle
targeting. Likewise, muscular dystrophies produce allotype-
1094-8341/04 $5.00 Copyright © 2004 the American Physiological Society
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
Cheng, Georgiana, Anita P. Merriam, Bendi Gong, Patrick
Leahy, Sangeeta Khanna, and John D. Porter. Conserved and
muscle-group-specific gene expression patterns shape postnatal
development of the novel extraocular muscle phenotype. Physiol
Genomics 18: 184–195, 2004.First published May 11, 2004; 10.1152/
physiolgenomics.00222.2003.—Current models in skeletal muscle
biology do not fully account for the breadth, causes, and consequences
of phenotypic variation among skeletal muscle groups. The muscle
allotype concept arose to explain frank differences between limb,
masticatory, and extraocular (EOM) muscles, but there is little understanding of the developmental regulation of the skeletal muscle
phenotypic range. Here, we used morphological and DNA microarray
analyses to generate a comprehensive temporal profile for rat EOM
development. Based upon coordinate regulation of morphologic/gene
expression traits with key events in visual, vestibular, and oculomotor
system development, we propose a model that the EOM phenotype is
a consequence of extrinsic factors that are unique to its local environment and sensory-motor control system, acting upon a novel
myoblast lineage. We identified a broad spectrum of differences
between the postnatal transcriptional patterns of EOM and limb
muscle allotypes, including numerous transcripts not traditionally
associated with muscle fiber/group differences. Several transcription
factors were differentially regulated and may be responsible for
signaling muscle allotype specificity. Significant differences in cellular energetic mechanisms defined the EOM and limb allotypes. The
allotypes were divergent in many other functional transcript classes
that remain to be further explored. Taken together, we suggest that the
EOM allotype is the consequence of tissue-specific mechanisms that
direct expression of a limited number of EOM-specific transcripts and
broader, incremental differences in transcripts that are conserved by
the two allotypes. This represents an important first step in dissecting
allotype-specific regulatory mechanisms that may, in turn, explain
differential muscle group sensitivity to a variety of metabolic and
neuromuscular diseases.
EXTRAOCULAR MUSCLE DEVELOPMENT
MATERIALS AND METHODS
Animals and tissues. Muscles were from Sprague-Dawley rats
(Harlan, Indianapolis, IN) euthanized with CO2 at postnatal days 0
(P0), P7, P14, P21, P28, and P45. Animal procedures received prior
Institutional Animal Care and Use Committee approval.
Electron microscopy. Rats were perfused with physiological saline
followed by 1% paraformaldehyde/2% glutaraldehyde fixative solution in 0.1 M phosphate buffer. EOMs were removed, postfixed in 4%
glutaraldehyde fixative solution, followed by 1% osmium tetroxide in
0.1 M phosphate buffer, and processed into plastic resin following
standard procedures (6). Some muscles were processed for visualization of neuromuscular junctions using acetylcholinesterase histochemistry. Sections were examined and photographed using a Zeiss model
10C electron microscope.
DNA microarray. To minimize inter-litter/animal variability, muscles were pooled from multiple rats for each of three independent
replicates/age/muscle group. EOM samples included all rectus and
oblique muscles, while gastrocnemius and soleus muscles were
pooled as representative of hindlimb muscle. Tissues were snap frozen
in liquid N2 and stored at ⫺80°C. cRNA was prepared for use on
Affymetrix (Santa Clara, CA) RG-U34A arrays, as described (36,
60–62, 65, 66). Briefly, RNA was extracted using TRIzol reagent
(GIBCO-BRL; Invitrogen, Rockville, MD). Pellets were resuspended
at 1 ␮g RNA/␮l DEPC-treated water, and 8 ␮g was used in a reverse
transcription reaction (SuperScript II; Life Technologies, Rockville,
MD) to generate first-strand cDNA. Double-strand cDNA was synthesized and used in an in vitro transcription (IVT) reaction to
generate biotinylated cRNA. Fragmented cRNA (15 ␮g) was used in
a 300-␮l hybridization cocktail containing herring sperm DNA and
BSA as carrier molecules, spiked IVT controls, and buffering agents.
A 200-␮l aliquot of cocktail was hybridized to microarrays for 16 h at
45°C. For manufacturer’s standard posthybridization wash, doublestain, and scanning protocols, we used an Affymetrix GeneChip
Fluidics Station 400 and Gene Array scanner.
Microarray data analysis. Raw data from microarray scans were
initially normalized and analyzed with Affymetrix Microarray Suite
(MAS) 5.0. MAS evaluates sets of perfect match (PM) and mismatch
(MM) probe sequences to obtain both hybridization signal values and
Physiol Genomics • VOL
18 •
present/absent calls for each transcript. The raw data series are housed
in the National Center for Biotechnology Information (NCBI) Gene
Expression Omnibus (GEO) database under series record accession
number GSE903 (GSM13668 through GSM13712). We used the
MAS filter only to exclude transcripts that were absent from all
samples from further analysis [present calls for EOM and hindlimb
were 42.6% (SD 3.7) and 39.6% (SD 4.3), respectively, of the 8,799
transcripts on U34A microarrays]. Microarray data from EOM and
hindlimb then was normalized using the “robust multichip average”
(RMA) algorithm (25) in ArrayAssist 2.0 (Iobion Informatics, La
Jolla, CA). RMA executes a background adjustment, a quantile
normalization, and a summation of individual probe set intensities
using a log scale linear additive model for the log transform of
background corrected/normalized PM probe intensities. RMA reportedly has higher sensitivity and specificity than either MAS 5.0 or
dChip (25).
Cluster analysis of the EOM temporal series was performed with
CAGED 1.1 (http://genomethods.org/caged/) (69). CAGED uses
Bayesian “clustering by dynamics” to identify a statistical model of
the most probable set of clusters of transcripts in a time series, without
relying upon any predefined similarity threshold. Background corrected/normalized probe signal output from RMA was averaged for each
transcript/age and then used as input to CAGED. To filter for transcripts differentially expressed in the EOM temporal series, fold
difference threshold was set at greater than or equal to twofold. We
then coded and combined the EOM and hindlimb data sets and
repeated the CAGED analysis. This approach allowed determination
of differentially expressed EOM transcripts that either 1) did not
undergo a similar twofold change in hindlimb or 2) exhibited different
temporal patterns of change in EOM and hindlimb muscle (i.e., fell
into different CAGED clusters).
Self-organizing map (SOM) analysis was implemented in GeneSpring 5.0 (Silicon Genetics, Redwood City, CA), using the RMA
transcript signal data from EOM and hindlimb muscle as input. For
statistical comparisons of expression patterns, we used parametric
testing (Welch t-test/Welch ANOVA, P ⱕ 0.001), applying the
Benjamini and Hochberg false discovery rate algorithm for multiple
testing correction. SOM identified transcripts differing in overall
temporal pattern between the EOM and hindlimb muscle series and
arranged them so that similar patterns appeared in nearest neighbor
clusters.
Affymetrix transcript annotations were replaced with official gene
nomenclature using NCBI databases, and gene functions were assigned based upon gene ontology and other data in NCBI LocusLink,
UniGene, and PubMed and Weizmann Institute of Science GeneCards
(http://bioinfo.weizmann.ac.il/cards/).
Promoter analysis. Here, we used the significance analysis of
microarrays (SAM) algorithm to filter EOM and hindlimb microarray
data for those transcripts that most clearly defined the EOM allotype.
Of 117 transcripts identified by SAM, 57 were known genes (52
unique) with 3⬘ sequence available from NCBI. RefSeq identifiers
obtained for each gene from LocusLink were used to extract 500 bases
of rat genomic DNA sequence upstream of transcription start sites.
Match software then was used in conjunction with the TRANSFAC
database (http://www.gene-regulation.com/) to locate putative transcription factor binding sites in genomic sequence of each of the 52
differentially expressed genes.
Real-time quantitative PCR. The same samples used for microarray
were used for real-time quantitative PCR (qPCR). One microgram of
total RNA was reverse transcribed using oligo-dT primer (Invitrogen).
One microliter of cDNA was diluted (1:6 to 1:10) and then used for
qPCR. Primers used for qPCR were Myh3 (embryonic)
(NM_012604), forward 5⬘ GATGGTGGTCCATGAAAGTGA 3⬘, reverse 5⬘ AGGGGTTACGTGGAAATTAAGC 3⬘; Myh4 (IIB)
(L13606), forward 5⬘ TAAGTGAAGAGTAAGGCAGCTCTGA 3⬘,
reverse 5⬘ GGATTAAATAGAATCACATGGGGAC 3⬘; Myh6 (␣cardiac) (X15938), forward 5⬘ ACACGAAGCGTGTCATCCAG 3⬘,
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
related phenotypes not predictable by current knowledge of the
localization and functions of disease gene products. Since
many neuromuscular diseases are not fully penetrant, and
targeted muscle groups vary widely, it may be impossible to
fully understand disease mechanisms without in-depth knowledge of muscle group diversity and its impact on disease.
Because of its exceptional phenotype and disease responsiveness, EOM may provide insights into the breadth, causes,
and consequences of muscle-group-specific identities. EOM
resistance to muscular dystrophy and particular sensitivity to
myasthenia gravis have been ascribed to constitutive, rather
than adaptive, differences from the limb allotype (31, 53, 60,
65). Determination of the precise mechanisms underlying such
exceptional patterns of sparing or involvement in neuromuscular disease may provide important clues to pathogenesis and
identify new treatment strategies.
We have proposed that the highly specialized EOM allotype
is a consequence of a novel myoblast lineage interacting with
extrinsic factors during a postnatal critical period of development (5, 7, 12, 57). Here, DNA microarray was used to identify
conserved and muscle-group-specific transcriptional patterns
during the critical period of EOM development. These data
represent an important first step in determining how genetic
and epigenetic factors shape the differentiated muscle allotypes.
185
186
EXTRAOCULAR MUSCLE DEVELOPMENT
reverse 5⬘ GGTCCCCTATGGCTGCAAT 3⬘; Myh7 (I or ␤-cardiac)
(X15939), forward 5⬘ GAGTTAAATGCACTCAACGCCA 3⬘, reverse 5⬘ CCTGAAGCTCTTTGAGCTTCTT 3⬘; Myh8 (perinatal)
(K02111), forward 5⬘ CAAGTGGCTGAAGGAAAGGCA 3⬘, reverse
5⬘ AGTGGGAGAAAAGTAAACACGAGAG 3⬘; Myh13 (extraocular) (AF075250), forward 5⬘ ATGTGGGAGGCCAGAAGAT 3⬘, reverse 5⬘ AGTCTCCCTCTGCTCTCCTGGA 3⬘. qPCR used the
Roche LightCycler (Mannheim, Germany) with the LightCycle-FastStart DNA Master SYBR Green I kit, following the manufacturer’s
protocol.
RESULTS AND DISCUSSION
Physiol Genomics • VOL
18 •
1
The Supplementary Material for this article (Supplemental Tables S1–S6
and Supplemental Figs. S1–S5) is available online at http://physiolgenomics.
physiology.org/cgi/content/full/00222.2003/DC1.
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
Adult EOM is divergent from other skeletal muscles (12, 18,
36, 60). We addressed the mechanisms responsible for EOM
divergence by first characterizing morphological and transcriptional profiles of postnatal EOM and then contrasting these
with hindlimb muscle data. Specifically, we asked 1) How do
transcriptional patterns relate to the postnatal morphogenesis of
EOM? 2) What dynamic temporal patterns are evident in gene
expression data from postnatal EOM? and 3) How do the
temporal patterns of conserved and muscle-group-specific transcripts differ between EOM and limb allotypes?
Postnatal morphogenesis of EOM. Postnatal EOM morphogenesis was first characterized by electron microscopy to
determine critical stages for transcriptional analysis. Unlike
most neonatal skeletal muscles (33, 34, 41, 70), P0 EOMs were
still composed of centrally nucleated myotubes with secondary
myotubes tightly apposed to many primary myotubes (Fig. 1, A
and G). Intramuscular axons were unmyelinated, and primitive
neuromuscular junctions contacted only primary myotubes,
activating secondary myotubes only indirectly through primary-secondary tight junctions (Fig. 1, D and G). By P7, the
central nuclei characteristic of myotubes had migrated peripherally, separately innervated primary and secondary myofibers
were evident, and intramuscular axons were thinly myelinated
(Fig. 1, B and E). Subsequent maturation of sarcoplasmic
reticulum and T-tubule networks demarcated the myofibrils,
allowing recognition of the basic EOM singly (small myofibrils) and multiply innervated (large myofibrils) fiber morphologies by P14 (Fig. 1C). Intramuscular axons were more heavily
myelinated by this stage, although neuromuscular junctions
remained primitive in capping small myofibers until after P21
(Fig. 1, F and H). Microvascular content increased rapidly after
P21. The two muscle layers and six myofiber types characteristic of adult EOM (see Ref. 53, 75) were recognizable by P28
(Fig. 1, I and J). Myofiber development between days P28 and
P45 was largely restricted to increases in diameter and mitochondrial content.
Using the ultrastructural data collected here, as well as
information from other sources (6, 20, 43, 48, 55, 56, 58, 75,
79), we constructed an integrated developmental profile for
EOM (Fig. 2A). Overall, the primary and secondary waves of
myogenesis that typify skeletal muscle are conserved in EOM
(6, 48, 54, 56). The distinctive EOM multiply innervated fibers
emerged early (days P7 to P14), in the absence of any extrinsic
cues from visually driven eye movements, but maturation of all
fiber types continued through the visual critical period. Temporal patterning of EOM myogenesis is, in part, directed by
visual- and vestibular-driven eye movements, since maldevelopment of either sensory system impairs EOM development (5,
7). Overall, postnatal EOM morphogenesis temporally lags
behind that of most other muscles (34, 41), most likely due to
the delay in onset of purposeful eye movements until after
eyelid opening (⬃P12). An integrated developmental profile
(Fig. 2) provides a framework for interpretation of postnatal
gene expression dynamics.
Dynamic gene expression profile of postnatal EOM. A
temporal expression series was obtained for EOM at six stages
between birth (P0) and adult (P45). Time points were based
upon major cellular, anatomic, and physiological events in
EOM and oculomotor motoneuron development (Fig. 2A). We
used a novel strategy to identify transcripts exhibiting dynamic
postnatal changes (ⱖ2-fold compared with a P0 baseline).
Averaged normalized hybridization signal data from RMA
were used as input for CAGED, an algorithm designed to
extract and cluster genes with similar temporal patterns (69).
CAGED generated 21 distinct clusters, containing 754 differentially regulated transcripts, in postnatal EOM (Fig. 3 and
Supplemental Table S1). (The Supplementary Material for this
paper is available at the Physiological Genomics web site.)1
The EOM series exhibited three major temporal patterns
(Fig. 3A). Group I comprised 10 clusters (n ⫽ 302 transcripts)
that were upregulated from birth. P14 was an inflection point,
between a rapid rise and plateau in expression level, for the
majority of group I clusters (clusters I.1, I.2, I.3, I.5, I.6, and
I.7). The emergence of oculomotor motoneuron discharge
patterns that resembled those of adult (79) may be, in part,
responsible for the observed stabilization of transcriptional
changes after P14. The eight clusters in group II (n ⫽ 355
transcripts) showed downregulation between days P0 and P45.
Although some clusters had a rapid decrease in expression after
birth and a plateau after P14 (clusters II.1, II.2, and II.3),
similar to the behavior of many group I clusters, more often the
group II clusters exhibited gradual and nearly linear decreases
in expression level from birth to P45. Group III clusters (n ⫽
97 transcripts) had a generally flat expression pattern, although
transient expression peaks (at P14 for cluster III.3 and between
P7 and P21 for cluster III.2) were seen in two of the three
unique clusters. The majority of transcripts that CAGED identified as dynamically regulated in the EOM time series are
known genes, and 57% could be grouped into functional
categories with significance to postnatal muscle development
(Supplemental Table S1 and Fig. 3B). We did, however,
identify 241 transcripts that were expressed sequence tags
(ESTs) (7 with at least ⫾8-fold changes during the developmental series); this subset is of interest for future studies since
it is likely to contain unidentified genes that may be novel
to EOM.
Genes associated with transcription, cell signaling, cell cycle/cell death, cell surface/cell adhesion, cytoskeleton, extracellular matrix, and protein/nucleic acid metabolism were
found at considerably higher frequency in group II (Fig. 3B),
indicating progressive downregulation of these functions as
myogenesis is completed. Only cluster II.1 contained transcripts with a precipitous decline after birth and plateau after
P14 (e.g., Tacstd1, Lgals7, and Cd24, which function in cell
surface/cell adhesion), although some other clusters showed
EXTRAOCULAR MUSCLE DEVELOPMENT
187
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 1. Electron microscopic analysis of extraocular muscle (EOM) development between birth (day P0) and day P28. A: P0 EOM
is composed of centrally nucleated tubes. Primary myotubes (1⬘) are still associated with secondary myotubes (2⬘). B: by P7,
secondary myogenesis is complete, and peripherally nucleated myofibers predominate. C: singly (SIF) and multiply (MIF)
innervated myofiber classes are evident by P14. Note myelinated axon leading to neuromuscular junction capping a SIF. D–F:
intramuscular nerves gradually acquire myelin, P0 (D), P7 (E), and P14 (F). G: at P2, nerve terminal (nt) lies on the surface
of a primary myofiber, with overlying Schwann cell (Sch). Little postsynaptic specialization is evident. Secondary myotube
(2⬘) is closely related to primary (arrow denotes interdigitation). H: by day P21, nerve terminal (nt) is opposed by some
postjunctional folding of SIF type, but still lies on the myofiber surface rather than being embedded in a sarcolemmal
depression. I: orbital layer maturation at P28, with recognizable orbital SIF and MIF fiber types. J: global layer maturation
at P28, with global MIF and the red (rSIF), intermediate (iSIF), and pale (pSIF) singly innervated fiber types. Bars ⫽ 5 ␮m
(A–F and I and J) and 2 ␮m (G and H).
Physiol Genomics • VOL
18 •
www.physiolgenomics.org
188
EXTRAOCULAR MUSCLE DEVELOPMENT
this trend to a more modest extent (e.g., clusters II.2, II.3, and
II.5). Of 28 cell cycle/death transcripts dynamically regulated
in postnatal EOM (e.g., Ccnd2, Cdk4, and Ccnl), 21 were in
clusters II.4, II.6, II.7, and II.8, which have very similar
temporal patterns of gradual downregulation, and all but 3 of
the genes in this functional class were in group II (only Ccng1
and two probes for Gadd45a, a positive regulator of apoptosis,
were in group I). Correspondingly, these same four group II
clusters also contained the majority (50 of 68 or ⬃74%) of
transcripts functioning in nucleic acid/protein structure and
metabolism. Representative transcripts in this class play roles
in DNA metabolism or packaging (e.g., Hmgb2, Top2a,
Nap1l1, Prim1, and Apex), ribosomal structure (e.g., Rpl18 and
Rps10), or RNA processing (e.g., Ptb and Sfrs10) that would be
expected to parallel cell cycle/death functions. By contrast,
protein metabolism transcripts were more likely to be upregulated or to exhibit little postnatal change (e.g., Stnl, Pcmt1,
Eef2, and Eif4ebp1 in groups I or III). Finally, cytoskeletal
(e.g., Tuba1, Actg2, and Krt1–18) and extracellular matrix
(e.g., fibril forming, Col5a1, and nonfibril forming, Col12a1,
collagens) genes were primarily distributed in group II clusters
(95% and 80%, respectively), with the majority showing very
gradual decreases in expression (55% in clusters II.4, II.5, and
II.6). The principal fibril-forming collagens (types I and III) did
Physiol Genomics • VOL
18 •
not meet the criteria for differential regulation from days P0
to P45.
Energy metabolism transcripts were more frequently found
in group I clusters (84% of category) (Fig. 3B), indicating
induction as myotubes transition to functional myofibers. Clusters I.1, I.2, and I.3 had the most rapid increases in expression
and leveled off after P14. The 15 known genes in these clusters
include adult contractile protein isoforms (Myh4 and Myh13),
channels/transporters (Pva, Atp1a2, Atp1b1, Slc2a4, and
Fabp3), and genes encoding energy metabolism enzymes
(Ckmt2, Acadm, and Got1). At least three of these are either
EOM specific (Myh13) or expressed at higher levels in EOM
than other skeletal muscles (Pva and Ckmt2) (10, 63). By
contrast, a significant fraction (66%) of all dynamically regulated metabolism transcripts were distributed across a restricted
set of clusters (clusters I.6, I.7, and I.8) with very gradual
increases in expression level. The majority of enzymes in the
glycolytic, tricarboxylic acid, respiratory, and ␤-oxidation of
fatty acid pathways were in these three clusters, suggesting
tight coregulation of energetic mechanisms in developing
EOM. Taken together, microarray findings correlated with
morphological indicators of energetic status (mitochondria/
microvasculature) that mature after P14 (see Figs. 1 and 2A).
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 2. Compilation of morphophysiological and microarray data in EOM development. A: schematic relating morphogenesis of
EOM to key events in visual, vestibular, and oculomotor system development. Myogenic events are indicated by solid black bars;
neurogenic and behavioral events are indicated by open bars. Data are from the present study and prior literature (see text for
references). B: temporal gene expression profiles by functional transcript categories for all non-EST transcripts detected in postnatal
EOM by CAGED. To weight the relative contributions of transcripts at each age, fold change values (absolute value) were summed
for each age and then expressed as a percentage of the P45 total. Note that the majority of transcriptional changes were by P21–P28.
Transcription for some functional categories (e.g., muscle development/structure, energy metabolism, and immune/cell defense)
peaked and then declined before P45. ECM, extracellular matrix.
EXTRAOCULAR MUSCLE DEVELOPMENT
189
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 3. Analysis of the postnatal expression series for EOM using the CAGED algorithm. A: clustering of transcripts with similar
expression patterns. CAGED identified transcripts that were dynamically regulated (i.e., changed by ⱖ2-fold during time series)
in postnatal EOM. Fold changes (Ln) were calculated against P0 values. Group I clusters were upregulated from birth (note
inflection point at P14 for many clusters), group II clusters were downregulated, and group III clusters exhibited relatively flat
expression patterns. B: histogram showing gene distribution among functional classes for each of the three groups.
Ion channel and transporter genes required by excitable
tissues also were more likely to be found in group I (58% of
category) (Fig. 3B). Most members of this class were not
among the CAGED clusters that were rapidly up- or downregulated in postnatal EOM. Instead, muscle ion channels and
binding proteins were modestly upregulated (e.g., two probes
for Cacna1s in clusters I.6 and I.8 and S100a1 in cluster I.8) or
showed transient spikes in expression (Kcnj11 and Vdac1 in
cluster III.2). The expression patterns of these transcripts are
Physiol Genomics • VOL
18 •
consistent with the temporal appearance and elaboration of the
T-tubule and sarcoplasmic reticulum networks and with reported postnatal changes in EOM calcium content and calciumATPase content and activity (58).
Few muscle-specific transcriptional regulators met the criteria for dynamic regulation in postnatal EOM, and all were
downregulated (Myog in cluster II.6, Gata6 in cluster II.5, and
Csrp3 in cluster III.1). Conservative microarray analytic tools,
such as the RMA algorithm used here, tend to exclude genes
www.physiolgenomics.org
190
EXTRAOCULAR MUSCLE DEVELOPMENT
Physiol Genomics • VOL
18 •
⬃P7 and recognition of the major two classes of differentiated
EOM fiber types, singly and multiply innervated, by P14; and
3) the induction of energy metabolism transcripts and ion
channels and various transporters accompanying the initiation
of myofiber function. The dynamic changes in gene expression
patterns were nearly complete by P14 to P21, indicating that
eyelid opening and the onset of purposeful fixation and targeting eye movements represent important landmarks in EOM
maturation (see Fig. 2).
Overview of expression differences between postnatal EOM
and hindlimb muscle. To identify developmental features
unique to EOM, we generated an expression profile for the
limb allotype and compared these profiles using two strategies.
Combined gastrocnemius and soleus muscles were selected for
this comparison. They represent all traditional skeletal muscle
fiber types, like EOM with an overall bias toward fast-twitch
fibers [an estimate of the mass of muscle from each fiber type
for gastrocnemius is 1% type I, 3% type IIA, 28% type IIX/D,
and 68% type IIB; soleus is 86% type I, 6% type IIA, 8% type
IIX/D, and 0% type IIB (15)]. First, we repeated the CAGED
analysis with the combined EOM and hindlimb data to identify
transcripts that either were dynamically regulated only in the
EOM series or exhibited temporal pattern differences between
the two muscle groups (i.e., appeared in different CAGED
clusters). Second, SOM was used to identify patterned differences in gene expression levels between the EOM and hindlimb profiles, without the CAGED restriction to dynamically
regulated transcripts only.
CAGED identified 314 transcripts that were dynamically
regulated (mean value for three replicates/age/muscle ⱖ2-fold)
in EOM, but not in hindlimb, 130 that were dynamically
regulated in hindlimb, but not in EOM, and 440 that were
dynamically regulated in both muscles, between days P0 and
P45 (Supplemental Fig. S1). Here, we focused upon the 314
transcripts dynamically regulated only in EOM (Supplemental
Table S2) and 291 transcripts that were dynamically regulated
in EOM and hindlimb, but fell into different CAGED clusters
(Supplemental Table S3 and Supplemental Fig. S1), indicating
that their postnatal expression patterns were dissimilar. The
transcripts identified here represent potential causes or consequences of EOM divergence from the traditional skeletal muscle allotype. Transcripts were assigned to functional categories
(Supplemental Tables S2 and S3 and Fig. 4). Other than ESTs
and unclassified (other) transcripts, the energy metabolism and
cell signaling categories best distinguished the EOM and limb
allotypes.
SOM then was used to directly compare expression level
data between muscle groups (P ⱕ 0.001, Welch t-test/Welch
ANOVA). This approach clustered transcripts using a similarity measure, yielding 15 distinct gene clusters (n ⫽ 837
transcripts) meeting criteria for muscle-group-specific expression profiles (Supplemental Table S4). Four SOM clusters
contained transcripts with expression levels for EOM ⬎ hindlimb (n ⫽ 129 transcripts; Supplemental Fig. S2, A, E, H, and
L), while expression levels were higher for hindlimb in the
remaining 11 clusters. Transcript distribution across functional
categories was assessed (Supplemental Fig. S3). Transcripts
with expression levels in EOM ⬎ hindlimb were more frequent
among the transcription, cell surface/adhesion, extracellular
matrix, channels/transporters, and immune/cell defense categories.
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
expressed at very low levels. Two transcripts that are signal
transducers for myoblast-to-myotube differentiation were induced (Mapk12 in cluster I.8 and Vamp5 in cluster I.6). Muscle
structure and development transcripts were broadly distributed
among groups I (n ⫽ 9), II (n ⫽ 5), and III (n ⫽ 6).
EOM-specific (Myh13) and type IIB (Myh4) myosins were in
cluster I.2, with rapid postnatal increases and plateau after P14.
Transcripts representing the ␣-cardiac (Myh6) and ␤-cardiac
(Myh7) myosins and sarcomeric structural components, Mybph
and Myl3, were in clusters I.8 and III.3, with little postnatal
increase in expression level. Contractile regulatory proteins,
Tpm1, Tnnt2, and Tpm3, showed changes opposite to myosin
heavy chain transcripts; these were found in group II clusters
with slow repression after birth. Since the majority of muscle
developmental regulators and contractile proteins did not meet
the ⱖ2-fold threshold for dynamically regulated transcripts,
postnatal elaboration of distinctive muscle traits must be considered as a gradual process with few sudden developmental
shifts. Rapid upregulation of an EOM-specific trait, Myh13,
contrasts with this pattern and supports the appearance of a
strong inductive signal in EOM by P7. This concurs with other
data on Myh13 expression (6) and may explain this transcript’s
sensitivity to postnatal visual/vestibular deprivation (5, 7).
EOM is uniquely spared in the dystrophin-glycoprotein
complex-based muscular dystrophies (29, 32, 38, 59, 64, 67,
68). Although normal EOM expresses all known components
of this complex (2, 37, 65), only two muscular dystrophyrelated transcripts, Capn3 (3 different probes in clusters I.6 and
I.8) and Cav3 (cluster III.1), met criteria for developmental
regulation in postnatal EOM. EOM status in Capn3 and Cav3
knockout mice has not yet been assessed. Likewise, only two
of many transcripts known to participate in neuromuscular
junction formation and maintenance, Agrn (cluster II.7) and
Chrne (cluster I.7), were dynamically regulated in postnatal
EOM, despite concurrent changes in neuromuscular junction
morphology. Transient expression behavior of myelin protein
transcripts (Mbp and Mpz in cluster III.2) or myelin signaling/
processing transcripts (Mapk12 and Ugt8 in clusters I.8 and
III.2) coincided with intramuscular nerve development.
EOM exhibits higher capillary content than most skeletal
muscles due to its highly oxidative energetics (75). Vascular
signaling transcripts (Vegf in cluster I.6 and Egln3 in cluster
I.9) closely paralleled the morphogenesis of EOM fiber types
and postnatal increase in energy metabolism transcripts (which
predominate in cluster I.6). By contrast, markers of differentiated endothelial cell and smooth muscle phenotypes exhibited
an inverse relationship (Acta2, Actg2, and Tagln in cluster II.2,
Csrp2 in cluster II.4, and Edg2 in cluster II.6), falling at
approximately the same rate as the rise in metabolism transcripts.
The transcriptional profile identified here for postnatal EOM
using the CAGED algorithm includes genes that met an induction/suppression threshold (ⱖ2-fold) and therefore are most
likely to contribute toward emergence of the novel phenotype.
To compare morphophysiological and microarray data, gene
expression profiles are shown graphically by functional category in Fig. 2B. Data are consistent with 1) the slow, progressive downturn of cellular processes associated with myoblast
proliferation, cell-cell contact and fusion, and general cell
growth that would be expected following completion of myofiber formation; 2) the completion of secondary myogenesis by
EXTRAOCULAR MUSCLE DEVELOPMENT
Taken together, CAGED and SOM analyses suggest that the
number of transcripts truly unique to the EOM allotype is
relatively small. However, nearly one-third of transcripts either
detected as differentially regulated by CAGED only in the
EOM series or with significantly higher expression in EOM by
SOM were ESTs. This finding suggests that there may be more
genes with expression confined to EOM than are currently
known. Based upon the known genes, data thus far are consistent with the hypothesis that a limited range of tissuespecific mechanisms interact with more subtle differences in
generalized skeletal muscle mechanisms to determine the novel
EOM phenotype.
Patterned differences between EOM and hindlimb muscles
from CAGED and SOM. We focused here upon identities and
functions of those transcripts with EOM-biased expression
patterns from CAGED, SOM, or both analyses. Transcription
factors are essential to the emergence of tissue-specific identities. For the developmental stages examined here, several
transcriptional regulators were dynamically regulated only in
EOM (Gata6, Hif1a, Jun, Neurod1, Nifb, Nr2f1, Sox10, Thrsp,
Tieg, and Zpf36l1) or were expressed at constitutively higher
levels in EOM (Gtf3c1 and Pitx2). While any of these may play
a key role in regulation of the EOM allotype, only one has been
tested to date. Pitx2 deletions cause EOM agenesis, showing
that it plays a critical role in early muscle development (19,
39). The maintained expression of Pitx2 throughout EOM
maturation further suggests that it may be essential in the
emergence and maintenance of EOM-specific properties.
The energy metabolism category contained the largest number of transcripts identified by CAGED as dynamically regulated only in EOM (n ⫽ 45; Fig. 4). SOM identified additional
metabolism transcripts with constitutive expression levels for
Physiol Genomics • VOL
18 •
EOM ⬎ hindlimb. The transcriptional regulator, Hif1a, is
essential for coordinating adaptive responses to hypoxia/ischemia, by activating genes for glycolytic enzymes, glucose transport, and angiogenesis. A prior study indicating that Hif1a
expression is essential for embryogenesis of cephalic mesenchyme and vasculature (26), coupled with our gene expression
data, suggests that it may be a determinant of energetic mechanisms in the EOMs. Moreover, our observation of the coordinate, dynamic upregulation of Hif1a and Egln3, which stabilizes and thereby prolongs Hif1a activity, and the EOM ⬎
hindlimb expression of Hif1a downstream targets/effectors
(Vegfb, Hmox2, Orp150, and Tf) collectively support an important role for Hif1a signaling in EOM development.
Since prior studies also identified substantial adult EOMhindlimb differences in energy metabolism (18, 36, 60), we
mapped the normalized microarray signal levels for all enzymes onto glycogen and glycolytic pathways (Fig. 5). Developmental patterns in rate-limiting enzyme transcripts in glycogen anabolic (glycogen synthase) and catabolic (phosphorylase) pathways were consistent with our failure to detect
glycogen deposits in postnatal EOM and with prior data showing that EOM does not use glycogen as a key energy store (18,
36, 60, 75). Nonmuscle isoforms of phosphorylase (Pygl and
Pygb) were, however, transiently high in neonatal EOM, a
finding that correlates with high glycogen levels in prenatal
EOM (unpublished data). Likewise, several glycolytic enzymes and regulators of glycolysis exhibited EOM-hindlimb
differences, including EOM utilization of non-skeletal muscle
enzyme isoforms for lactic dehydrogenase (Ldhb), enolase
(Eno2), and aldolase (Aldoc) and its low expression of 3-phosphoglycerate kinase and the skeletal muscle isoform of a key
glycolytic regulator, 6-phosphofructo-2-kinase/fructose 2,6bisphosphatase. The adaptive value of this divergence for
EOM, including any efficiency gain from non-muscle enzyme
isoforms, is unknown. Finally, genes encoding many lipid
transporters and fatty acid ␤-oxidation enzymes were either
dynamically upregulated only in EOM (e.g., Cd36, Fabp4,
Facl2, Cpt1b, and Acadl) or exhibited more dramatic induction
in postnatal EOM vs. hindlimb (e.g., Hadhb, and Acdml). Two
additional fatty acid transport/metabolism transcripts (Decr1
and Apoe) were detected at higher constitutive levels in EOM
by SOM. These data suggest that fatty acids may provide an
alternative energy source to glycogen in EOM, much like they
do in other highly active muscles such as heart and insect flight
muscle (21).
Our expression data also provide evidence for EOM-hindlimb divergence in oxidative energetics, commensurate with
the high mitochondrial content of some EOM fiber types that
emerges during the time course studied here. In particular, we
identified 23 mitochondrial transcripts that either were dynamically regulated in EOM only by CAGED analysis or were
expressed at constitutively higher levels in EOM by SOM
analysis (Supplemental Table S2). Dynamic regulation of lipoprotein lipase (Lpl) in EOM, but not hindlimb, may be
mechanistic in the high mitochondrial content and lipid metabolism of this muscle group, since Lpl overexpression is
known to increase both parameters in skeletal muscle (22).
Consistent with the oxidative capacity of the EOM allotype,
the higher vascular development in EOM vs. hindlimb was
supported by higher constitutive levels and/or dynamic reguwww.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 4. Functional distribution of differences in EOM and limb allotypes from
the CAGED algorithm. CAGED was repeated after coding and combining the
EOM and hindlimb transcriptional series. Histogram illustrates distribution of
EOM-hindlimb differences across functional categories for the 605 transcripts
that met the CAGED criterion of ⱖ2-fold change only in the EOM series (n ⫽
314) or presence of EOM and hindlimb transcripts in different clusters (n ⫽
291), indicative of differing postnatal regulatory patterns.
191
192
EXTRAOCULAR MUSCLE DEVELOPMENT
lation of several vasculature-related transcripts only in EOM
(e.g., Cldn3, Fbln5, Grn, Hif1a, Klf4, Plcg1, Tf, and Vegfb).
The divergence of EOM from the more traditional skeletal
muscles with respect to energy metabolism mechanisms may
be directly responsible for its novel response to metabolic
myopathies. For example, low reliance upon specific transcripts correlates with EOM mild or absent responses to glycogen storage disease type 1 (G6pt1), glycogenosis type IX
myopathy (Pgk1), and various hereditary and acquired myopathies due to purine nucleotide cycle defects (Ampd1). By
contrast, the high mitochondrial content of EOM is associated
with substantial allotype sensitivity to accumulation of mitochondrial DNA mutations and the resultant mitochondrial
myopathies (28, 47). Collectively, these correlations of gene
expression patterns with disease sensitivity support the hypothesis that allotype-specific traits are determinants of the severity
of response to a variety of muscle diseases.
Physiol Genomics • VOL
18 •
Differences in myosin heavy chain isoform expression were
first used to define skeletal muscle allotypes. We used qPCR to
validate myosin heavy chain isoform expression patterns detected in postnatal EOM and hindlimb muscle using RMA
(Supplemental Figs. S4 and S5). qPCR data were highly
correlated with expression patterns obtained from microarray
and both were in agreement with prior findings of the developmental regulation of Myh3, Myh8, and Myh13 in adult EOM
(6, 81).
Intrinsic and extrinsic mechanisms in EOM divergence.
Given a subset of genes that are preferentially expressed in
EOM, promoter analysis can be used to obtain a better understanding of upstream regulatory mechanisms for the novel
EOM allotype. We filtered the DNA microarray temporal
series for those genes that most clearly distinguished EOM,
using the SAM algorithm (117 transcripts; Supplemental Table
S5). We then analyzed cis-regulatory genomic sequence (500
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
Fig. 5. Expression differences between EOM and limb allotypes for two major carbohydrate metabolism pathways. Rate-limiting
enzymes are shown in red. For each enzyme, the postnatal expression pattern for the transcript is shown by a color bar; EOM is
indicated on the left and hindlimb is on the right. Transcript signal intensity is indicated as shown in the inset. Multiple isoforms
are shown for some genes, and, when known, skeletal muscle-specific isoforms are underlined. A: glycolytic pathway. B: glycogen
metabolism pathway.
EXTRAOCULAR MUSCLE DEVELOPMENT
Physiol Genomics • VOL
18 •
of traits. Our promoter analysis of a subset of genes with
EOM-enriched expression helps narrow the field of candidates
that may be involved in allotype regulation.
Conclusions
We have used morphological and DNA microarray analyses
to establish an integrated postnatal profile for EOM development. Based upon the coordinate regulation of these traits with
key events in visual, vestibular, and oculomotor system development, we propose a model that the EOM phenotype is a
consequence of extrinsic factors that are unique to its local
environment and sensory-motor control system, acting upon a
novel myoblast lineage. Using the CAGED and SOM algorithms, we identified a broad spectrum of differences in postnatal transcriptional patterns of the EOM and limb muscle
allotypes along with several differentially regulated transcription factors that may signal allotype specificity. Here, we have
highlighted allotype energetics differences but identified divergence in many other properties that remain to be further
explored.
Collectively, this analysis shows that emergence of the EOM
allotype is the consequence of tissue-specific mechanisms that
direct expression of a limited number of EOM-specific transcripts and broader, incremental differences in transcripts that
are conserved across the two allotypes. We propose that
transcriptional mechanisms that are shared by EOM and cardiac muscle (e.g., Pitx2 and Nkx2.5) or EOM and the eye
proper (e.g., Pitx2, Pax2, and Tbx15) represent the best candidates for regulation of the EOM allotype. Comparing expression profiles of entire skeletal muscle groups may partially
mask the divergence of the EOM and hindlimb by failing to
reveal unique combinations of transcripts expressed in specific
fiber types. We recently have shown that even the two distinct
EOM layers differ in gene expression profiles (35). Single fiber
type expression analysis then is likely to be critical in understanding the heterogeneity of skeletal muscle allotypes. Taken
together, these data represent an important first step in dissecting allotype-specific regulatory mechanisms that subsequently
may explain the differential sensitivity of the skeletal muscle
allotypes to metabolic and neuromuscular diseases.
ACKNOWLEDGMENTS
We thank Marty Veigl, Scott Shnider, and Dan Rischar for assistance with
the conduct and analysis of microarray studies, Mary Gail Engle for electron
microscopy, and Francisco Andrade for assistance with tissue dissections.
Kinga Tomczak, Marco Ramoni, and Zak Kohane, of The Children’s Hospital
(Boston) and The Children’s Hospital Informatics Program, provided access
to, and invaluable advice on the use of, the CAGED algorithm.
GRANTS
This work was supported by National Institutes of Health (NIH) Grants
R01-EY-015306, R01-EY-09834, and R01-EY-12779 to J. D. Porter. DNA
microarray and bioinformatics core module support were provided by NIH
Grants P30-CA-43703 and P30-EY-11371, respectively.
REFERENCES
1. Andrade FH, Merriam AP, Guo W, Cheng G, McMullen CA, Hayess
K, Van Der Ven PF, and Porter JD. Paradoxical absence of M lines and
downregulation of creatine kinase in mouse extraocular muscle. J Appl
Physiol 95: 692–699, 2003. First published April 25, 2003; 10.1152/
japplphysiol.00358.2003.
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
bases upstream of transcription start sites) that was available
for 52 known genes in this group. Conserved transcription
factor binding motifs among these genes represent putative
regulatory mechanisms. Binding sites for 86 different transcription factors were found in 3⬘ sequence of EOM-enriched
genes (Supplemental Tables S5 and S6). Binding sites for
ubiquitously expressed factors, such as heat shock transcription
factor (found upstream of transcription start sites for 96% of
the 52 genes) and C/EBP (upstream of 44% of genes), were
commonly found, whereas binding motifs for transcription
factors known to be active in skeletal muscle were prominent,
but encountered less frequently (e.g., Sp1, 31%; Msx1, 23%;
Pou2f1, 23%; Usf, 8%, Myog, 6%, and Myod1, 2%).
Consistent with EOM expression of multiple cardiac muscle
traits (12, 36, 60), binding sites for transcription factors known
to be active in heart were detected among EOM-enriched genes
(e.g., Pax4, 62%; Pitx2, 15%; Nkx2.5, 13.5%; Supplemental
Table S6). Binding sites for transcription factors that are
involved in eye development and/or are responsive to retinoic
acid (e.g., Pax2, 25%; Msx1, 23%; Hoxa3, 21%; Znf42, 19%;
Pitx2, 15%; Tfap2a, 10%; Pbx1, 8%; Hnf3b, 6%) also were
detected at moderate frequency among these EOM-enriched
genes. This finding is consistent with both present data and
prior reports (12, 35, 36, 60) of enhanced expression of retinoic
acid-sensitive and ocular development-related genes in EOM.
Prior studies also have shown that Pitx2 is essential for heart
and eye development and EOM myogenesis (19, 39) and may
regulate the EOM-specific Myh13 (3). Here, we provide additional evidence of a role for Pitx2 in EOM myogenesis by
showing its maintained expression into adult EOM and identifying several additional EOM-enriched transcripts with upstream Pitx2 motifs (Ace, Atp1a1, Cd9, Cds1, Dcn, Gpx3,
LOC64312, and Scya11).
Ultimately, the EOM allotype is shaped by myoblast lineage-specific traits interacting with epigenetic mechanisms,
such as motoneuron activity patterns, diffusible trophic factors,
circulating hormones, and other features of the local environment. Studies using recently derived immortalized cell lines
suggest that EOM myoblasts themselves are unique (unpublished data), while in vivo studies point to novel patterns of
regulation of myogenic transcripts in EOM (46, 50, 80).
Disruption of oculomotor motoneuron maturation alters the
molecular profile of EOM, including suppression of Myh13 (5,
7, 11). EOM perinatal expression patterns of Pitx2, a potential
regulator of Myh13, then may also be activity dependent.
Specific neurotrophic mechanisms do function in EOM development, as spinal motoneurons cannot substitute for oculomotor motoneurons in supporting EOM maturation in organotypic
nerve-muscle coculture (57). Finally, local interactions, including those between neural crest and EOM precursor cells, likely
influence gene expression patterns in developing EOM (13,
49, 78).
In typical skeletal muscle, there is no apparent molecular
“master regulator” of specific fiber types (74). Instead, regulatory mechanisms are multifactorial with contractile and fatigue
properties independently determined by and tailored to the
functional demands placed upon specific muscle fiber types
and/or groups. Based upon our present data, it is probable that
the EOM allotype also is the consequence of combinatorial
activity of a variety of transcriptional mechanisms, with individual signaling pathways responsible for only a narrow range
193
194
EXTRAOCULAR MUSCLE DEVELOPMENT
Physiol Genomics • VOL
18 •
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
Cellular and developmental control of O2 homeostasis by hypoxia-inducible factor 1 alpha. Genes Dev 12: 149–162, 1998.
Jacoby J, Ko K, Weiss C, and Rushbrook JI. Systematic variation in
myosin expression along extraocular muscle fibres of the adult rat. J Muscle Res Cell Motil 11: 25–40, 1990.
Kadenbach B, Munscher C, Frank V, Muller-Hocker J, and Napiwotzki J. Human aging is associated with stochastic somatic mutations of
mitochondrial DNA. Mutat Res 338: 161–172, 1995.
Kaminski HJ, al-Hakim M, Leigh RJ, Katirji MB, and Ruff RL.
Extraocular muscles are spared in advanced Duchenne dystrophy. Ann
Neurol 32: 586–588, 1992.
Kaminski HJ, Kusner LL, and Block CH. Expression of acetylcholine
receptor isoforms at extraocular muscle endplates. Invest Ophthalmol Vis
Sci 37: 345–351, 1996.
Kaminski HJ, Li Z, Richmonds C, Ruff RL, and Kusner L. Susceptibility of ocular tissues to autoimmune diseases. Ann NY Acad Sci 998:
362–374, 2003.
Karpati G and Carpenter S. Small-caliber skeletal muscle fibers do not
suffer deleterious consequences of dystrophic gene expression. Am J Med
Genet 25: 653–658, 1986.
Kelly AM. Satellite cells and myofiber growth in the rat soleus and
extensor digitorum longus muscles. Dev Biol 65: 1–10, 1978.
Kelly AM and Zacks SI. The histogenesis of rat intercostal muscle. J Cell
Biol 42: 135–153, 1969.
Khanna S, Cheng G, Gong B, Mustari MJ, and Porter JD. Genomewide transcriptional profiles are consistent with functional specialization
of the extraocular muscle layers. Invest Ophthalmol Vis Sci In press.
Khanna S, Merriam AP, Gong B, Leahy P, and Porter JD. Comprehensive expression profiling by muscle tissue class and identification of
the molecular niche of extraocular muscle. FASEB J 17: 1370–1372, 2003.
Khanna S, Richmonds CR, Kaminski HJ, and Porter JD. Molecular
organization of the extraocular muscle neuromuscular junction: partial
conservation of and divergence from the skeletal muscle prototype. Invest
Ophthalmol Vis Sci 44: 1918–1926, 2003.
Khurana TS, Prendergast RA, Alameddine HS, Tome FM, Fardeau
M, Arahata K, Sugita H, and Kunkel LM. Absence of extraocular
muscle pathology in Duchenne’s muscular dystrophy: role for calcium
homeostasis in extraocular muscle sparing. J Exp Med 182: 467–475,
1995.
Kitamura K, Miura H, Miyagawa-Tomita S, Yanazawa M, KatohFukui Y, Suzuki R, Ohuchi H, Suehiro A, Motegi Y, Nakahara Y,
Kondo S, and Yokoyama M. Mouse Pitx2 deficiency leads to anomalies
of the ventral body wall, heart, extra- and periocular mesoderm and right
pulmonary isomerism. Development 126: 5749–5758, 1999.
Lu JR, Bassel-Duby R, Hawkins A, Chang P, Valdez R, Wu H, Gan
L, Shelton JM, Richardson JA, and Olson EN. Control of facial muscle
development by MyoR and capsulin. Science 298: 2378–2381, 2002.
Luff AR and Atwood HL. Changes in the sarcoplasmic reticulum and
transverse tubular system of fast and slow skeletal muscles of the mouse
during postnatal development. J Cell Biol 51: 369–383, 1971.
Mankoo BS, Collins NS, Ashby P, Grigorieva E, Pevny LH, Candia A,
Wright CV, Rigby PW, and Pachnis V. Mox2 is a component of the
genetic hierarchy controlling limb muscle development. Nature 400:
69–73, 1999.
Mascarello F and Rowlerson AM. Myosin isoform transitions during
development of extra-ocular and masticatory muscles in the fetal rat. Anat
Embryol (Berl) 185: 143–153, 1992.
McLoon LK and Wirtschafter JD. N-CAM is expressed in mature
extraocular muscles in a pattern conserved among three species. Invest
Ophthalmol Vis Sci 37: 318–327, 1996.
Mitchell PO and Pavlath GK. Multiple roles of calcineurin in skeletal
muscle growth. Clin Orthop 403, Suppl: S197–S202, 2002.
Mootoosamy RC and Dietrich S. Distinct regulatory cascades for head
and trunk myogenesis. Development 129: 573–583, 2002.
Muller-Hocker J, Seibel P, Schneiderbanger K, and Kadenbach B.
Different in situ hybridization patterns of mitochondrial DNA in cytochrome c oxidase-deficient extraocular muscle fibres in the elderly. Virchows Arch A Pathol Anat Histopathol 422: 7–15, 1993.
Nag AC and Cheng M. Differentiation of fibre types in an extraocular
muscle of the rat. J Embryol Exp Morphol 71: 171–191, 1982.
Noden DM. Patterning of avian craniofacial muscles. Dev Biol 116:
347–356, 1986.
Noden DM, Marcucio R, Borycki AG, and Emerson CP Jr. Differentiation of avian craniofacial muscles. I. Patterns of early regulatory gene
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
2. Andrade FH, Porter JD, and Kaminski HJ. Eye muscle sparing by the
muscular dystrophies: lessons to be learned? Microsc Res Tech 48:
192–203, 2000.
3. Briggs MM and Schachat F. Early specialization of the superfast myosin
in extraocular and laryngeal muscles. J Exp Biol 203: 2485–2494, 2000.
4. Brooke MH and Kaiser KK. Muscle fiber types: how many and what
kind? Arch Neurol 23: 369–379, 1970.
5. Brueckner JK, Ashby LP, Prichard JR, and Porter JD. Vestibuloocular pathways modulate extraocular muscle myosin expression patterns.
Cell Tissue Res 295: 477–484, 1999.
6. Brueckner JK, Itkis O, and Porter JD. Spatial and temporal patterns of
myosin heavy chain expression in developing rat extraocular muscle.
J Muscle Res Cell Motil 17: 297–312, 1996.
7. Brueckner JK and Porter JD. Visual system maldevelopment disrupts
extraocular muscle-specific myosin expression. J Appl Physiol 85: 584–
592, 1998.
8. Burke RE, Levine DN, and Zajac FE III. Mammalian motor units:
physiological-histochemical correlation in three types in cat gastrocnemius. Science 174: 709–712, 1971.
9. Calvo S, Vullhorst D, Venepally P, Cheng J, Karavanova I, and
Buonanno A. Molecular dissection of DNA sequences and factors involved in slow muscle-specific transcription. Mol Cell Biol 21: 8490–
8503, 2001.
10. Celio MR and Heizmann CW. Calcium-binding protein parvalbumin is
associated with fast contracting muscle fibres. Nature 297: 504–506, 1982.
11. Cheng G, Mustari MJ, Khanna S, and Porter JD. Comprehensive
evaluation of the extraocular muscle critical period by expression profiling
in the dark-reared rat and monocularly deprived monkey. Invest Ophthalmol Vis Sci 44: 3842–3855, 2003.
12. Cheng G and Porter JD. Transcriptional profile of rat extraocular muscle
by serial analysis of gene expression. Invest Ophthalmol Vis Sci 43:
1048–1058, 2002.
13. Couly GF, Coltey PM, and Le Douarin NM. The developmental fate of
the cephalic mesoderm in quail-chick chimeras. Development 114: 1–15,
1992.
14. Degenhardt K and Sassoon DA. A role for Engrailed-2 in determination
of skeletal muscle physiologic properties. Dev Biol 231: 175–189, 2001.
15. Delp MD and Duan C. Composition and size of type I, IIA, IID/X, and
IIB fibers and citrate synthase activity of rat muscle. J Appl Physiol 80:
261–270, 1996.
16. Dietrich S. Regulation of hypaxial muscle development. Cell Tissue Res
296: 175–182, 1999.
17. Dietrich S, Schubert FR, Healy C, Sharpe PT, and Lumsden A.
Specification of the hypaxial musculature. Development 125: 2235–2249,
1998.
18. Fischer MD, Gorospe JR, Felder E, Bogdanovich S, Pedrosa-Domellof
F, Ahima RS, Rubinstein NA, Hoffman EP, and Khurana TS. Expression profiling reveals metabolic and structural components of extraocular
muscles. Physiol Genomics 9: 71–84, 2002. First published March 5,
2002; 10.1152/physiolgenomics.00115.2001.
19. Gage PJ, Suh H, and Camper SA. Dosage requirement of Pitx2 for
development of multiple organs. Development 126: 4643–4651, 1999.
20. Hanson J, Lennerstrand G, and Nichols KC. The postnatal development of the inferior oblique muscle of the cat. III. Fiber sizes and
histochemical properties. Acta Physiol Scand 108: 61–71, 1980.
21. Haunerland NH. Fatty acid binding protein in locust and mammalian
muscle. Comparison of structure, function and regulation. Comp Biochem
Physiol B Biochem Mol Biol 109: 199–208, 1994.
22. Hoefler G, Noehammer C, Levak-Frank S, el-Shabrawi Y, Schauer S,
Zechner R, and Radner H. Muscle-specific overexpression of human
lipoprotein lipase in mice causes increased intracellular free fatty acids and
induction of peroxisomal enzymes. Biochimie 79: 163–168, 1997.
23. Hoh JF, Hughes S, and Hoy JF. Myogenic and neurogenic regulation of
myosin gene expression in cat jaw-closing muscles regenerating in fast
and slow limb muscle beds. J Muscle Res Cell Motil 9: 59–72, 1988.
24. Hoh JFY, Hughes S, Hugh G, and Pozgaj I. Three hierarchies in skeletal
muscle fibre classification: allotype, isotype, and phenotype. In: Cellular
and Molecular Biology of Muscle Development, edited by Kedes LH and
Stockdale FE. New York: Liss, 1989, p. 15–26.
25. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, and Speed
TP. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids
Res 31: e15, 2003.
26. Iyer NV, Kotch LE, Agani F, Leung SW, Laughner E, Wenger RH,
Gassmann M, Gearhart JD, Lawler AM, Yu AY, and Semenza GL.
EXTRAOCULAR MUSCLE DEVELOPMENT
51.
52.
53.
54.
55.
57.
58.
59.
60.
61.
62.
63.
64.
65.
Physiol Genomics • VOL
18 •
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
Willem M, Mayer U, and Kaminski HJ. Constitutive properties, not
molecular adaptations, mediate extraocular muscle sparing in dystrophic
mdx mice. FASEB J 17: 893–895, 2003.
Porter JD, Merriam AP, Leahy P, Gong B, and Khanna S. Dissection
of temporal gene expression signatures of affected and spared muscle
groups in dystrophin-deficient (mdx) mice. Hum Mol Genet 12: 1813–
1821, 2003.
Porter JD, Rafael JA, Ragusa RJ, Brueckner JK, Trickett JI, and
Davies KE. The sparing of extraocular muscle in dystrophinopathy is lost
in mice lacking utrophin and dystrophin. J Cell Sci 111: 1801–1811, 1998.
Ragusa RJ, Chow CK, St Clair DK, and Porter JD. Extraocular, limb
and diaphragm muscle group-specific antioxidant enzyme activity patterns
in control and mdx mice. J Neurol Sci 139: 180–186, 1996.
Ramoni MF, Sebastiani P, and Kohane IS. Cluster analysis of gene
expression dynamics. Proc Natl Acad Sci USA 99: 9121–9126, 2002.
Rubinstein NA and Kelly AM. Development of muscle fiber specialization in the rat hindlimb. J Cell Biol 90: 128–144, 1981.
Rushbrook JI, Weiss C, Ko K, Feuerman MH, Carleton S, Ing A, and
Jacoby J. Identification of alpha-cardiac myosin heavy chain mRNA and
protein in extraocular muscle of the adult rabbit. J Muscle Res Cell Motil
15: 505–515, 1994.
Schafer K and Braun T. Early specification of limb muscle precursor
cells by the homeobox gene Lbx1h. Nat Genet 23: 213–216, 1999.
Schiaffino S and Serrano A. Calcineurin signaling and neural control of
skeletal muscle fiber type and size. Trends Pharmacol Sci 23: 569–575,
2002.
Spangenburg EE and Booth FW. Molecular regulation of individual
skeletal muscle fibre types. Acta Physiol Scand 178: 413–424, 2003.
Spencer RF and Porter JD. Structural organization of the extraocular
muscles. Rev Oculomot Res 2: 33–79, 1988.
Summerbell D, Ashby PR, Coutelle O, Cox D, Yee S, and Rigby PW.
The expression of Myf5 in the developing mouse embryo is controlled by
discrete and dispersed enhancers specific for particular populations of
skeletal muscle precursors. Development 127: 3745–3757, 2000.
Termin A, Staron RS, and Pette D. Myosin heavy chain isoforms in
histochemically defined fiber types of rat muscle. Histochemistry 92:
453–457, 1989.
Trainor PA and Tam PP. Cranial paraxial mesoderm and neural crest
cells of the mouse embryo: co-distribution in the craniofacial mesenchyme
but distinct segregation in branchial arches. Development 121: 2569–2582,
1995.
Tsuzuki S, Yoshida S, Yamamoto T, and Oka H. Developmental
changes in the electrophysiological properties of neonatal rat oculomotor
neurons studied in vitro. Neurosci Res 23: 389–397, 1995.
Tzahor E, Kempf H, Mootoosamy RC, Poon AC, Abzhanov A, Tabin
CJ, Dietrich S, and Lassar AB. Antagonists of Wnt and BMP signaling
promote the formation of vertebrate head muscle. Genes Dev 17: 3087–
3099, 2003.
Wieczorek DF, Periasamy M, Butler-Browne GS, Whalen RG, and
Nadal-Ginard B. Co-expression of multiple myosin heavy chain genes, in
addition to a tissue-specific one, in extraocular musculature. J Cell Biol
101: 618–629, 1985.
www.physiolgenomics.org
Downloaded from http://physiolgenomics.physiology.org/ by 10.220.33.5 on June 18, 2017
56.
expression and myosin heavy chain synthesis. Dev Dyn 216: 96–112,
1999.
Patapoutian A, Miner JH, Lyons GE, and Wold B. Isolated sequences
from the linked Myf-5 and MRF4 genes drive distinct patterns of musclespecific expression in transgenic mice. Development 118: 61–69, 1993.
Peter JB, Barnard RJ, Edgerton VR, Gillespie CA, and Stempel KE.
Metabolic profiles of three fiber types of skeletal muscle in guinea pigs and
rabbits. Biochemistry 11: 2627–2633, 1972.
Porter JD and Baker RS. Muscles of a different “color”: the unusual
properties of the extraocular muscles may predispose or protect them in
neurogenic and myogenic disease. Neurology 46: 30–37, 1996.
Porter JD and Baker RS. Prenatal morphogenesis of primate extraocular
muscle: neuromuscular junction formation and fiber type differentiation.
Invest Ophthalmol Vis Sci 33: 657–670, 1992.
Porter JD, Baker RS, Ragusa RJ, and Brueckner JK. Extraocular
muscles: basic and clinical aspects of structure and function. Surv Ophthalmol 39: 451–484, 1995.
Porter JD and Hauser KF. Diversity and developmental regulation of
extraocular muscle: progress and prospects. Acta Anat (Basel) 147: 197–
206, 1993.
Porter JD and Hauser KF. Survival of extraocular muscle in long-term
organotypic culture: differential influence of appropriate and inappropriate
motoneurons. Dev Biol 160: 39–50, 1993.
Porter JD and Karathanasis P. The development of extraocular muscle
calcium homeostasis parallels visuomotor system maturation. Biochem
Biophys Res Commun 257: 678–683, 1999.
Porter JD and Karathanasis P. Extraocular muscle in merosin-deficient
muscular dystrophy: cation homeostasis is maintained but is not mechanistic in muscle sparing. Cell Tissue Res 292: 495–501, 1998.
Porter JD, Khanna S, Kaminski HJ, Rao JS, Merriam AP, Richmonds
CR, Leahy P, Li J, and Andrade FH. Extraocular muscle is defined by
a fundamentally distinct gene expression profile. Proc Natl Acad Sci USA
98: 12062–12067, 2001.
Porter JD, Khanna S, Kaminski HJ, Rao JS, Merriam AP, Richmonds
CR, Leahy P, Li J, Guo W, and Andrade FH. A chronic inflammatory
response dominates the skeletal muscle molecular signature in dystrophindeficient mdx mice. Hum Mol Genet 11: 263–272, 2002.
Porter JD, Leahy P, Khanna S, Cheng G, and Veigl ML. Applicability
and practice of genome-wide expression profiling: insights from studies of
skeletal muscle diversity and disease mechanisms. In: Perspectives in
Gene Expression, edited by Appasani K. Westboro, MA: Eaton, 2003, p.
221–244.
Porter JD, Merriam AP, Gong B, Kasturi S, Zhou X, Hauser KF,
Andrade FH, and Cheng G. Postnatal suppression of myomesin, muscle
creatine kinase, and the M-line in rat extraocular muscle. J Exp Biol 206:
3101–3112, 2003.
Porter JD, Merriam AP, Hack AA, Andrade FH, and McNally EM.
Extraocular muscle is spared despite the absence of an intact sarcoglycan
complex in gamma- or delta-sarcoglycan-deficient mice. Neuromuscul
Disord 11: 197–207., 2001.
Porter JD, Merriam AP, Khanna S, Andrade FH, Richmonds CR,
Leahy P, Cheng G, Karathanasis P, Zhou X, Kusner LL, Adams ME,
195