Genome-Wide Transcriptional Profiles Are Consistent with Functional Specialization of the Extraocular Muscle Layers

Abstract

purpose. Compartmentalization of the extraocular muscles into well-defined orbital and global layers is highly conserved. Recently, the active pulley hypothesis correlated the anatomic properties of orbital–global muscle layers with layer-specific division of labor. Microarray technology was used to identify muscle-layer–specific transcriptional profiles and, thereby, extend understanding of the structure-function characteristics of extraocular muscle layers.

methods. Laser capture microdissection was used to obtain muscle layer samples from monkey medial rectus muscles. RNA was linearly amplified and hybridized to human U133 series microarrays (Affymetrix, Santa Clara, CA), which have sufficient sequence homology for use in subhuman primates. Data was analyzed using Affymetrix and Robust Multichip Average (RMA) algorithms. Select transcripts were verified by quantitative PCR and in situ hybridization.

results. A broad spectrum of transcriptional differences (>181 transcripts) was identified between the two extraocular muscle layers. Patterned differences in the sarcomeric contractile machinery and cytoskeleton were suggestive of key layer differences in contraction speed. Differentially expressed transcript identities, however, extended well beyond those traditionally associated with muscle-fiber–group differences.

conclusions. Muscle layer transcriptional profiles correlated with the different loads and usage patterns of extraocular muscle layers, as proposed in the active pulley hypothesis. The magnitude and breadth of orbital–global layer expression differences strongly suggests that oculomotor control systems may drive two distinct motor output pathways, each comprising separate motoneurons and muscle fibers, with one output path adapted to determining pulley position and the other to movement of the eye.

Extraocular muscles (EOMs) are specialized to perform a diverse repertoire of eye movements, including high-speed saccades, slow pursuit and vergence movements, and maintenance of steady fixation. These functional intricacies are reflected in the complex fiber types seen in EOMs, which differ remarkably from somatic muscle fibers. In addition, the EOMs are normally divided into two distinct compartments or layers. ¹ The peripheral orbital layer lies along the EOM surface, facing the orbital wall. This layer encloses a second portion, the global layer, lying closer to the globe. In some species, the two layers are separated by an internal perimysium. The orbital layer contains small-diameter fibers with numerous mitochondria and abundant blood vessels. The global layer contains relatively large-diameter fibers with variable mitochondrial content and fewer vessels. The distinction between the orbital and global layers in EOM is discernible by histochemistry, particularly in regard to aerobic versus anaerobic metabolism. Fibers in the orbital layer stain intensely for oxidative enzymes, whereas both the staining intensity and proportion of oxidative fibers gradually decrease through the global layer. By contrast, the activity associated with glycolytic enzymes is more intense in the global layer and is weak in the orbital layer. ¹ Some known molecular differences, including myosin heavy chain isoform composition, further characterize the two layers and six fiber types. ^{2 3 4 5 6 7}  

Structural compartmentalization is not unique to EOM, but its significance has been unclear. ^{8 9} However, recent findings suggested a novel mechanism (the active pulley hypothesis) that correlates the distinctive properties of the orbital and global muscle layers with layer-specific division of labor. Each rectus and oblique EOM passes through a collagenous pulley located near the globe equator in Tenon’s fascia. ^{10 11 12} Pulleys are, in turn, coupled to the orbital wall, adjacent EOMs, and equatorial Tenon’s fascia by bands containing collagen, elastin, and smooth muscle. The abundant elastic fibers in and around pulleys provide reversible extensibility to these resilient tissues. Pulleys have important implications for EOM action, redefining the functional origin of an EOM at its pulley. Several lines of evidence, including magnetic resonance imaging, gross examinations, surgical exposures, and histologic studies in humans and monkeys, indicate that the orbital layer of each rectus muscle inserts on its corresponding pulley, whereas only the global layer inserts on the sclera. The orbital layer thus controls linear translation of the pulley, regulating the EOM’s pulling direction in primary and secondary gaze positions, whereas the global layer is responsible for eye rotation. ^{11 13} Consistent with the view of a tight interrelationship between the orbital layer, pulley, and eye movement control, the more restricted oculomotor repertoire of rodents is accompanied by less well-developed orbital layers and muscle pulleys than in primates. ¹⁴  

Our hypothesis was that the substantial anatomic differences in the two EOM layers are consistent with divergent functions in control of pulley and eye movements. Muscle fibers composing the orbital layer must be adapted to the continuous elastic load of the pulley suspension system, whereas those in the global layer require properties to oppose the intermittent, viscous load of the antagonist EOM. ¹¹ For a better understand of the structural–functional characteristics of the two layers, we used contemporary oligonucleotide microarray technology to identify muscle-layer–specific transcriptional profiles. 

Materials and Methods

Animals and Tissue Preparation

Two 4-month-old male macaque monkeys (Macaca mulatta) reared in a normal 12-hour light and dark cycle were euthanatized. These same animals represented control subjects for studies of the peripheral and central consequences of monocular deprivation. ¹⁵ Medial (MR) and lateral (LR) rectus muscle bellies were dissected from origin to myotendinous junction, from both the orbits. Muscles from one orbit of each monkey were embedded in optimal cutting temperature (OCT) compound (Sakura Finetek, Torrance, CA) and snap frozen in liquid nitrogen for cryostat sectioning and laser capture microdissection (LCM) of the individual EOM orbital and global layers, followed by RNA extraction. Muscles from the second orbit of both monkeys were frozen directly in liquid nitrogen for RNA extraction from the whole muscle. Thus, this provided two independent replicates each of the orbital layers O1 and O2, global layers G1 and G2, whole MR muscles MR1 and MR2, and whole LR muscles LR1 and LR2. All animal procedures were approved by the Institutional Animal Care and Use Committees at Case Western Reserve University and Emory University and were in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Laser Capture Microdissection

From each OCT-embedded rectus muscle, we prepared 10-μm-thick frozen sections on membrane (pet foil)-coated slides (Leica, Wetzlar, Germany). The slides were immediately frozen on dry ice and then stored at −70°C for subsequent microdissection. Since longitudinal variations in muscle fiber traits are well described for the orbital layer of the EOM, ¹ we restricted microdissection and collection of layer samples to the endplate region of the muscle. For this purpose, step serial sections were stained with Texas red–conjugated α-bungarotoxin to confirm the presence of en-plaque endplates in the sections of both the orbital and global layers. 

For LCM, rapid toluidine blue staining was used to identify the layers. The sections were immersed in the relevant fixatives or staining solutions (20% alcohol, 0.05% toluidine blue) for 30 seconds each, followed by dehydration with graded alcohol concentrations (75%, 95%, and 100%) for 20 seconds each. An LCM system was used for laser capture of muscle layers (Biosystems AS LCM; Leica). Figure 1A shows an example of a laser-captured orbital layer segment. 

RNA Isolation, Linear Amplification from LCM-Generated Tissue, Target Preparation, and Hybridization to Microarrays

We pooled LCM microdissectates from 15 to 18 sections for both the orbital layer (samples O1 and O2) and global layer (samples G1 and G2) from each MR. Total RNA was isolated from the pooled tissue (PicoPure RNA isolation kit; Arcturus, Mountain View, CA). T7-based amplification then was used to obtain sufficient RNA, since it amplifies without distorting gene-to-gene ratios (i.e., linear amplification). Briefly, we used two rounds of RNA amplification in which 100 to 250 ng of total RNA was used as starting material. First-strand synthesis yielded cDNA incorporating a T7 promoter, and second-strand synthesis yielded double-stranded cDNA (cDNA synthesis kit; Invitrogen, Carlsbad, CA). After cDNA isolation, in vitro transcription (IVT) using T7 RNA polymerase yielded amplified RNA (aRNA; MEGAscript kit; Ambion, Austin, TX). Then, aRNA was used in the second round of amplification for first-strand synthesis with random primer and second-strand synthesis with T7 promoter primer. This yielded double-stranded cDNA that was used in a second IVT reaction to generate biotinylated cRNA using T7 RNA polymerase and biotin-labeled nucleotides (Affymetrix, Santa Clara, CA); 40 to 80 μm of biotin-labeled cRNA was produced per reaction. The biotinylated cRNA was run on a gel to assess the probe size. As seen in Figure 1B , the cRNA size generated for all samples was uniform and in the range of the size of the probes on the Affymetrix microarrays, ∼600 bp. 

Total RNA was extracted from the whole-muscle samples, in accordance with the manufacturer’s instructions (TRIzol; Invitrogen-Gibco, Rockville, MD). RNA pellets were resuspended at 1 μg RNA/μL diethyl pyrocarbonate–treated water. Biotinylated cRNA was generated according to the manufacturer’s protocol (Affymetrix), as described previously. ^{16 17}  

Hybridization of LCM or whole muscle cRNA to human U133A and -B array sets (Affymetrix). After purification and fragmentation, 15 μg of cRNA was used in a 300-μL hybridization mixture containing spiked IVT controls. Approximately 200 μL of mixture was hybridized to DNA microarrays for 16 hours at 45°C. Standard posthybridization wash and double-stain protocols were performed on a fluidics station, and the microarrays were scanned (GeneChip Fluidics Station 400; Affymetrix; and Gene Array Scanner; Hewlett-Packard, Palo Alto, CA). 

Microarray Data Analysis

A computer was used was used for initial data processing and n-fold ratio analyses (Microarray Suite [MAS], software version 5.0; Affymetrix). Pair-wise comparisons were made between each of the different muscle group samples obtained from the same monkey (e.g., LR 1 vs. MR 1, LR 2 vs. MR 2, O1 vs. G1). Transcripts defined as differentially regulated met the criteria of: (1) consistent increased/decreased call versus control in both replicates (Affymetrix computed measure representing confidence in gene expression differences based on Wilcoxon’s signed rank test) and (2) absolute average twofold difference or more. In the HG-U133 set of arrays, 11 probe pairs typically represent each transcript. The MAS 5.0 algorithm assesses probe pair perfect match (PM) and mismatch (MM) saturation differences; calculates a probability; and determines increase, decrease, or no change calls. Any transcripts with expression intensity below 400 for all the samples were excluded since distortion of the n-fold difference results when expression levels are low and may be within the level of background noise. Lists of all differentially expressed genes between the orbital and global layers of MR obtained using MAS are in Supplemental Table 1 at www.iovs.org/cgi/content/full/45/9/3055/DC1. 

Differential gene expression by n-fold difference ratios was also computed from the microarray data using the Robust Mutichip Average (RMA) algorithm, as implemented on computer (ArrayAssist software, version 2.0; Iobion Informatics, La Jolla, CA). This algorithm analyzes the microarray data in three steps: a background adjustment, quantile normalization, and finally summation of the probe intensities for each probe set using a log scale linear additive model for the log transform of (background corrected, normalized) PM intensities. Here, the comparisons were not pair-wise, because the software generates the average of the baseline files to which each experimental file can be compared (i.e., LR1 and LR2 were compared to the average of MR1 and MR2, and O1 and O2 were compared to the average of G1 and G2). ¹⁸ Data from the RMA algorithm are in Supplemental Table 2 at www.iovs.org/cgi/content/full/45/9/3055/DC1. 

Data were visualized as a hierarchical dendrogram generated on computer (GeneSpring software, ver. 4.2.1; Silicon Genetics, Redwood City, CA). Annotations provided by Affymetrix for all probe sets were replaced with the official gene nomenclature using resources provided by the National Center for Biotechnology Information (NCBI). 

Quantitative PCR

Select transcripts were verified in triplicate by quantitative (q)PCR, using the same samples as were used in the microarray studies. Briefly, transcript-specific primers were designed on computer (Primer Express software; Applied Biosystems, Inc. [ABI], Foster City, CA), with the use of NCBI BLAST to ensure specificity of each primer pair. Reverse transcription was performed on 100 ng total RNA with the superscript first-strand synthesis system (Invitrogen). qPCR used a PCR core reagent (SYBR green; ABI) in a 25-μL volume, with a sequence-detection instrument (model 7000; ABI) Glyceraldehyde phosphate dehydrogenase (GAPDH) was used as an internal positive control to ensure that equivalent amounts of RNA were included in each assay. Mean n-fold changes between the orbital and global layers were calculated by averaging of triplicate measurements for each gene. The relative n-fold difference in calculation used the 2-ΔΔCT method. ¹⁹  

In Situ Hybridization

The sequence of the skeletal actinin 3 (ACTN3) and CSRP3-specific oligonucleotide primers was deduced by PCR. PCR product was subcloned into the pCR II vector by using a TA cloning kit (Invitrogen) and fragment sequence was confirmed by sequencing. Digoxigenin (DIG)-labeled cRNA probes (antisense and sense probes) were made by in vitro transcription according to the manufacturers protocol (Roche Applied Science, Indianapolis, IN) and cellular localization of ACTN3 and CSRP3 transcripts was performed by nonisotopic in situ hybridization on 10-μm cryostat sections. Briefly, sections were fixed in 4% paraformaldehyde/phosphate-buffered saline (PBS), washed with PBS, treated with proteinase K, acetylated, and prehybridized at room temperature for 2 hours in hybridization buffer (50% formamide, 5× SSC, 5× Denhardt solution, 250 μg/mL baker’s yeast RNA, and 500 μg/mL salmon sperm DNA). The DIG-cRNA probe hybridization was performed overnight at 65°C. Immunologic detection was by anti-DIG antibody (anti-DIG-AP; Fab fragment; Roche), diluted 1:5000 in buffer (100 mM Tris [pH 7.5], 0.15 M NaCl). Visualization of the signal was by 4-nitro blue tetrazolium chloride/5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP) staining. Sections were dehydrated and mounted in permanent mounting medium. 

Results

Microarray technology (Affymetrix) was used to identify genes that are differentially expressed between the LR and MR EOMs and between the two distinct layers of the MR muscle. Because nonhuman primate arrays are currently unavailable, the HG-U133 set of Affymetrix human arrays was used. Previous studies have validated the use of cross-species microarray hybridization when sequence homology is expected to be close. ^{20 21} The HG-U133A and -B contain 22,283 and 22,645 probe sets, respectively, representative of ∼33,000 well-substantiated human genes. The percentage of transcripts detected as present ranged from 33% to 39% for the U133A array and 20% to 22% for the U133B array for the amplified samples (human tissue samples normally show ∼40% present with these arrays), indicating that a substantial fraction of probe sets were effective at transcript detection in the cross-species hybridization. Likewise, the nonamplified samples showed little variability among themselves in the percentage of transcripts marked as present, although the percentage present was consistently lower than that of the amplified, laser microdissected RNA samples (Supplemental Table 3 at www.iovs.org/cgi/content/full/45/9/3055/DC1). For this reason, all comparisons were restricted to the samples processed similarly. The microarray data for all the samples used in this study have been deposited as series GSE 907 in the NCBI Gene Expression Omnibus Web-based data repository. 

Our objectives were to determine differences in the gene expression patterns of the horizontal rectus muscles and to detect genes that are differentially expressed in the orbital and global muscle layers. Prior studies by us and by others have demonstrated the effectiveness of the Affymetrix MAS algorithm, including the requirement for consistent calls across all replicates. However, RMA methodology has recently been shown to provide higher specificity and sensitivity when using n-fold change analysis to detect differential expression versus MAS 5.0 and dChip. ¹⁸ To assure a minimum number of false positives, only those probe sets that were identified by both methodologies were included into the final list of genes differentially expressed between the MR and LR or the two layers of the MR muscle. The transcripts identified here represent whole tissue differences which could include the muscle fiber, connective tissue, blood, and nerve supply differences. 

Comparison of Medial and Lateral Rectus Muscles

Both the MAS 5.0 and RMA analyses (Supplemental Table 4 at www.iovs.org/cgi/content/full/45/9/3055/DC1) showed that the two horizontal recti have nearly identical transcriptional profiles. Of the more than 39,000 transcripts probed in this study, only 13 were differentially expressed in LR versus MR comparisons (11 higher in LR and 2 higher in MR) when analyzed by MAS alone, and 12 transcripts were differentially expressed (11 higher in LR and 1 higher in MR) when analyzed by RMA alone. Nearly all (10/11) of the probe sets overexpressed in LR, but none of those underexpressed, were common to both analytic methods (Supplemental Table 4). The 10 transcripts with higher expression in LR represented four known genes: hemoglobin alpha 1, hemoglobin alpha 2, hemoglobin beta and calgranulin A, which is highly expressed in the cytosol of neutrophils and monocytes. All the known genes overexpressed in LR were then related to blood components. 

Comparison of Orbital and Global Layers

Using the MAS 5.0 criteria, 337 transcripts exhibited different expression levels (≥2-fold) between the EOM orbital and global layers (Supplemental Table 1); using RMA criteria, 210 transcripts were observed to have a twofold or more difference in expression levels between the two layers (Supplemental Table 2). We focused our attention on the 181 transcripts (representing 166 distinct genes or ESTs) common to both methods (Table 1 ), as these exhibited the most robust expression differences and are likely to best define the two EOM layers. Expression levels of the 181 differentially expressed transcripts are displayed in a hierarchical dendrogram (Fig. 2A) . Of these transcripts, 103 (including 66 known genes) were orbital layer enriched, whereas 78 transcripts (including 47 known genes) were global layer enriched. Transcripts were characterized for known or putative cellular roles and functionally classified using NCBI LocusLink (http://www.ncbi.nlm.nih.gov/LocusLink[b]/) and Weizmann Institute of Science (Rehovot, Israel) GeneCards (http://bioinfo.weizmann.ac.il/cards[b]/) online databases. Of the orbital-layer–enriched genes with known functions, most belonged to three main categories: signaling, nucleic acid, and protein modification, and muscle related. Global-layer–enriched genes with known functions mainly sorted into transcription, signaling, transporter/channel, and muscle-related categories (Fig. 2B) . The subset of differentially expressed muscle-related transcripts was further subclustered to show the expression patterns across the two layers (Fig. 3) . We expected to find transcripts for known orbital/global differences in this list. Embryonic myosin (MYH3), which has been reported to be consistently higher in the orbital layer, was one of the most significant differences between orbital and global layers (∼9–12-fold more in the orbital than in the global layer). Also, the orbital layer showed higher expression of several genes with roles in vascular development, such as Norrie disease (NDP) and endothelin receptor, consistent with the previously shown higher vasculature content of the orbital layer. ¹  

Overall, two interesting patterns of expression could be distinguished from the list of differentially enriched muscle-related genes. First, the orbital layer showed higher expression of muscle genes known to be expressed in slow muscle fiber types and lower expression of fast fiber type genes. The orbital layer showed higher expression of slow/β-cardiac myosin (MYH7) and slow/cardiac troponin (TNNC1), which are present in slow fiber types of limb muscle, and lower expression of myosin binding protein C (MYBC2), tropomodulin 4 (TMOD4), and the actin filament cross-linking protein, alpha 3 actinin (ACTN3), which are specific to fast-twitch fibers of limb muscle. Adenosine monophosphate deaminase 1 (isoform M) (AMPD1), typically expressed at lower levels in slow muscles, also was lower in the orbital layer. By contrast, the expression of two transcripts, sarcolipin (SLN) and carbonic anhydrase 3 (CA3), did not fit the pattern of slow fiber traits in the orbital and fast traits in the global layer. Expression of sarcolipin, a SERCA1 modulator preferentially expressed in limb fast fiber types, was higher in the orbital layer. A prior immunocytochemistry report in human EOM has shown the expression of SERCA1 in all orbital layer fibers and most SIFs and some MIFs of the global layer, in contrast to limb muscle where SERCA1 is limited to fibers expressing fast myosin. ²² Likewise, the sulfonamide-resistant isoform of carbonic anhydrase (CA3) is normally expressed at higher levels in slow oxidative fibers, ²³ but we detected higher expression in the global layer. Second, the orbital layer showed preferential expression of the cardiac muscle contractile elements, α-cardiac myosin (MYH6) and cardiac troponin (TNNT2). Previous immunohistochemistry studies have shown similar preferential presence of α-cardiac myosin in orbital multiply innervated fibers only or in both orbital and global multiply innervated fibers, depending on the species. ^{5 24 25 26} Our data also suggest muscle-layer–specific expression patterns for some cytoskeletal components. Most of these relate to the organization of the Z-discs and thin filaments. Calponin 3 (CNN3), nebulette (NEBL), cysteine and glycine-rich protein 3 (cardiac LIM protein, CSRP3), and desmin (DES) were expressed at higher levels in the orbital layer, whereas α-actinin 3 (ACTN3) expression was higher in the global layer. 

Some genes involved in myogenesis, or known to upregulate in denervated muscle or regenerating muscle, also exhibited muscle-layer–specific expression patterns. Those higher in global layer included aquaporin 1 (AQP1), known to be downregulated during myogenic differentiation; SPARC/osteonectin, a transcript essential for myoblast fusion and differentiation ²⁷ ; osteopontin (SPP1), which is upregulated in regenerating muscle ²⁸ ; and myostatin (GDF8), which negatively regulates skeletal muscle growth. Transcripts with higher expression in orbital layer included thrombospondin 4 (THBS4), which exhibits expression patterns that correlate with myogenesis and innervation status ²⁹ ; follistatin, which promotes Pax-3 expression and thereby transiently delays muscle differentiation ³⁰ ; and insulin-like growth factor binding protein (IGFBP3), which is a muscle growth stimulatory factor. ³¹  

Microarray Data Validation

Using qPCR, orbital- and global-layer–enriched genes were arbitrarily selected to validate the microarray data. PCR data were normalized using GAPDH. The results, presented in Table 2 , show that the expression patterns obtained with qPCR were consistent with the microarray results for all the 10 arbitrarily selected transcripts. Expression of one of the genes, adenosine monophosphate deaminase 1 (AMPD1), which catalyzes the deamination of adenosine monophosphate AMP to inosine monophosphate (IMP) in skeletal muscle and plays a critical role in energy metabolism was higher in the global layer. We also examined the differential expression of AMPD1 using an enzyme activity analysis. ³² The orbital layer fibers were almost devoid of adenylate deaminase reaction product (data not shown), which is consistent with a nearly threefold lower level of gene expression of AMPD1 in the orbital layer, as detected by both microarray and qPCR. Also, the expression differences for actinin 3 and CSRP3 were confirmed using both qPCR and in situ hybridization, which detected the actinin 3 isoform transcript only in the global layer fibers, whereas it detected the CSRP3 signal was only in the orbital layer (data not shown), consistent with the microarray data. 

Discussion

Prior gene expression profiling studies have established that EOM exhibits substantial modifications to the skeletal muscle prototype in adapting to its novel role in eye movement control. ^{16 17 33 34} In the present study, we used LCM, linear RNA amplification, and genome-wide expression profiling technology to uncover a broad pattern of differential gene expression between the orbital and global layers of monkey MR muscles and between the two horizontal recti muscles. Data suggest that the orbital and global layers are nearly as different from one another as EOM is from other skeletal muscles. Of the transcripts screened, 181 (representing 167 distinct genes and ESTs) were identified as differentially expressed between orbital and global layers of the monkey MR muscle at the twofold level, and 14 of these differed by more than fivefold. This most likely is a conservative estimate of orbital-global differences, since the probes on the human microarrays used here do not always have exact sequence homology (yielding false negatives) with the monkey and we used a very stringent data-acceptance threshold. By contrast, the extremes of typical skeletal muscle, slow (red) versus fast (white), appear to exhibit only modest differences. ³⁵ Although our findings concur with gene/protein expression data from the few prior reports of muscle layer specificity, thereby validating our approach, most of the transcripts identified as orbital or global layer enriched have not been previously reported. Taken together, these data support the emerging concept that the two EOM layers represent distinct functional compartments with fundamentally unique roles in eye movement control. Using the same analytical criteria, no significant differences were detected between MR and LR, consistent with their roles as antagonists in both conjugate and disjunctive horizontal gaze. 

Patterned Sarcomeric Transcript Differences and Division of Labor

Our DNA microarray data show that the morphologic differences between orbital and global layers ¹ are accompanied by distinctive gene expression profiles. We suggest that the patterned expression differences detected in our study can be attributed to the divergent functional roles of the two muscle layers. The active-pulley hypothesis states that each EOM has dual points of insertion: The global layer inserts on the globe to execute eye movements, whereas the orbital layer inserts on and positions the muscle pulley to determine each EOM’s rotational axis. ³⁶ Although this division of labor may not be black and white, because electrophysiological evidence indicates transmission of orbital layer muscle force to the muscle insertion, ³⁷ some electromyographic (EMG) studies provide evidence that there may be muscle-layer–specific oculomotor motor unit pools. ³⁸ Global layer EMGs show pulse–step activity during saccades, the pulse opposing the formidable viscous load from the relaxing antagonist muscle and the step opposing an elastic load that is a consequence of eccentric fixation positions. By contrast, orbital layer EMGs show only the step change of activity during saccades. In addition, fibers in the orbital layer are nearly continuously active throughout the oculomotor range, whereas most global layer fibers become silent only slightly out of their field of action (i.e., on-direction). ³⁹ A correlate of the active pulley hypothesis is that the orbital and global layer fibers must be adapted to very different mechanical loads. The muscle-layer–specific transcriptional data obtained in our study add to the understanding of the division of labor in EOM. 

It is well established that the contractile protein profile of EOM is complex, including heterogeneous expression of multiple myosin heavy chain isoforms within individual muscle fibers and utilization of nearly all known striated muscle myosin isoforms within each EOM. ^{2 3 4 5 6 25 26 40 41 42 43} In agreement with the orbital–global differences in fiber type composition, the contractile protein isoforms were prominent among transcripts with muscle layer specificity. For example, our microarray data are in agreement with prior reports of enrichment of embryonic (MYH3) myosin in the orbital layer fibers ^{3 25 26 42} and with the expression of EOM-specific myosin (MYH13) in both muscle layers. ^{4 44} The finding of higher expression of slow myosin (MYH7) and troponin (TNNC1 and TNNT2) and α-cardiac (MYH6) myosin transcripts in the orbital layer correlates with a higher proportion of multiply innervated fibers (MIFs) in the orbital layer (20% versus 10% in global layer), ¹ since slow contractile protein isoforms and α-cardiac myosin are expressed by MIF types. 

There has not yet been a direct physiologic assessment of the relative contractile properties of the EOM orbital and global layers. However, in addition to the distribution of contractile protein transcripts noted earlier, expression levels of other sarcomeric proteins support the concept that orbital and global layers differ in contraction speed. The fast-twitch fiber myosin binding protein, tropomodulin (TMOD4), and α-actinin-3 (ACTN3) were expressed at considerably higher levels (>7-fold) in the global layer. α-Actinin-3 associates with Z-discs of fast-twitch fibers and may aid in generation of forceful contractions at high velocity. ⁴⁵ Taken together, the distribution of ACTN3 and other sarcolemmal proteins represents an adaptation to the functional requirements of the two muscle layers: fast, but transient, muscle contractions in the global layer for eye movement and slower, but sustained, contractions in the orbital layer for pulley positioning and movement. The sustained activity of the orbital layer is reflected in high myofiber oxidative capacity and vascularity. We did not, however, detect significant transcriptional evidence of layer differences in mitochondrial content or fatigability either here or in our prior studies comparing whole EOM with hindlimb muscle. ^{16 17 34} This finding may reflect the relative importance of transcriptional versus translational control mechanisms in regulation of myofiber energetics. ^{46 47}  

Usually, muscle fiber types are defined on the basis of contraction velocity and fatigue resistance. We have shown that the EOM expression signature is fundamentally distinct from that of other skeletal muscle, but most differentially expressed transcripts are not directly related to the contraction velocity–fatigability properties that are basic determinants of fiber type. ^{16 17 34} A similar theme was apparent in the data in our study, since nearly all of the orbital- or global-layer–enriched transcripts could not be linked to myofiber contractile or energy metabolism traits. Instead, orbital–global gene expression differences included a range of functions, such as cell signaling and cytoskeleton and extracellular matrix components. Layer specificity in these particular protein classes suggests that the sarcolemmal organization and mechanisms for transfer of force from sarcomere to tendon may differ in the two muscle layers. The novel sparing of EOMs in the dystrophin-glycoprotein complex-based muscular dystrophies, including the differential response of orbital and global layers in dystrophin- and utrophin–deficient mice, ⁴⁸ supports the concept of muscle group and layer specialization in cytoskeletal and extracellular matrix proteins. 

Further dissection of the significant laminar differences in signal transduction transcripts may yield insights into the regulation of orbital–global layer phenotypes. At the extreme, we detected orbital layer enrichment of several signal transduction transcripts that are traditionally considered to be retina-specific (PDC, NDP, and RGS1.6) but apparently have been adopted to serve a unique, but as yet not understood, role in EOM. Also, orbital layer showed higher expression of neuroserpin and tissue plasminogen activator (PLAT). Neuroserpin is a protease inhibitor secreted by axons in the synaptic region, where it controls plasminogen activator activity, and it is widely expressed in brain, particularly during the period of synaptic specification and refinement. ⁴⁹ As knowledge of the molecular and cell biology, biochemistry, and physiology of EOM layers and fiber types accumulates, integrated models of EOM function that account for the diversity of traits identified in this study will become possible. 

Retention of Embryonic Traits in the Orbital Layer and Identification of a Putative Regulatory Mechanism

The developmental myosin heavy chain isoforms (MYH3 and MYH8) are present during transitional states during skeletal myogenesis and are subsequently replaced by adult fast-twitch myosin isoforms. Expression of developmental myosin heavy chain isoforms by adult limb muscles is pathologic and is interpreted as evidence of either on-going regeneration or arrested muscle development. By contrast, MYH3 is normally expressed in the orbital layer of adult EOMs. ^{3 25 42} Several other embryonic traits are preferentially retained by orbital layer fiber types, including the embryonic γ-acetylcholine receptor subunit at singly innervated fibers (SIFs) and multiply innervated fibers (MIFs) neuromuscular junctions ⁷ and neural cell-adhesion molecule and β1-syntrophin on the extrasynaptic sarcolemma. ^{50 51} We extend the same theme by showing orbital layer enrichment of transcripts that are typically upregulated in regenerating or denervated muscle, such as thrombospondin 4 ²⁹ and follistatin. ⁵² Follistatin has also been shown to be a negative regulator of myostatin. ⁵³ The lower expression of myostatin (GDF8), which negatively regulates skeletal muscle growth, in the orbital layer extends the breadth of orbital layer transcripts that may be part of the same theme. It has been suggested that the EOMs are constantly remodeling as part of requisite repair mechanisms for this highly active muscle, ^{54 55} but there is no evidence that this process is responsible for the large-scale, continuous expression of embryonic traits in the orbital layer. Although the retention of embryonic-like properties likely has adaptive value for orbital layer fiber types, the identity of the responsible regulatory mechanisms represents a central unresolved question in the EOM biology. 

A plausible mechanism for regulation of embryonic traits in orbital layer fiber types emerged from the patterned variations in gene expression seen in this study. The orbital layer shows higher expression of some critical Z-line assembly and structure-related transcripts. One transcript expressed at higher levels (approximately fivefold) in the orbital layer is cardiac muscle LIM protein (CSRP3), which is a component of the Z-disc. CSRP3 is an essential part of a complex that triggers downstream effector pathways after mechanical stretch of striated muscle. ⁵⁶ Regulation of gene transcription by stretching is a highly conserved function in both cardiac and skeletal muscle. This stretch-regulated signaling cascade triggers a hypertrophic program that leads to an increase in sarcomeres and upregulates genes normally expressed only at embryonic or fetal stages. ⁵⁷ The differential expression of CSRP3 (higher in the orbital layer fibers) may mechanistically couple the continuous load of the pulley elastic tissues to the maintained expression of multiple embryonic muscle traits in the orbital layer of adult EOM. The orbital layer enrichment of four and a half LIM domains 1 (FHL1; also known as SLIM1) s corroborates the stretch-related upregulation of genes, since induction of FHL1 mRNA expression has been associated with postnatal skeletal muscle growth and stretch-induced muscle hypertrophy. ⁵⁸  

Summary

Skeletal muscle fiber traits are not static, but rather adapt to the loads and activation patterns that they experience. Taken together, the transcriptional differences identified herein between the orbital and global layers reflect the different loads and usage patterns for the two layers, as predicted by the active pulley hypothesis. We further propose that the retention of embryonic traits in the orbital layer is a consequence of its load acting through a CSRP3 signal transduction cascade. The magnitude and breadth of the orbital–global layer expression differences strongly suggests that oculomotor control systems may drive two distinct motor output pathways, each comprising separate motoneurons and muscle fibers, one output path adapted to determining pulley position and the other to movement of the eye. Muscle group differences in gene expression also have been associated with differing response to disease, ^{59 60} suggesting that transcriptional differences between the orbital and global layers may confer differential sensitivity to metabolic and neuromuscular disease. 

 Supported by National Eye Institute Grants R01 EY09834 (JDP), R01 EY12779 (JDP), R01 EY015306 (JDP), and P30 EY11371 and a Walt and Lilly Disney Award for Amblyopia Research (JDP) from Research to Prevent Blindness. The DNA microarray core facility is supported by National Institutes of Health Grant P30 CA43703 and support for primate use is by National Institutes of Health Grant P51 RR000165.

 Submitted for publication December 19, 2003; revised March 1 and May 3, 2004; accepted May 12, 2004.

 Disclosure: S. Khanna, None; G. Cheng, None; B. Gong, None; M.J. Mustari, None; J.D. Porter, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

(A) Photomicrograph of a 10-μm-thick cross-section of monkey MR stained with hematoxylineosin, illustrating LCM. Top: Section including both orbital (O) and global (G) layers before capture; bottom: same section after capture of the orbital layer fibers. (B) Photograph of gel showing amplified RNA from the global (glob) and orbital (orb) layers, illustrating the uniformity of the cRNA size among the four samples.

Table 1.

 

Transcripts Differentially Expressed between Orbital and Global Layers

Accession No.	Transcript (symbol)	Difference* (n-fold)
				MAS	RMA
Transcription factor
BC004958	Leucine-rich repeat (in FLII) interacting protein 1 (LRRFIP1)	−2.50	−2.11
BE784583	Activated RNA polymerase II transcription cofactor 4 (PC4)	−2.00	−2.14
AF055376	v-Maf musculoaponeurotic fibrosarcoma oncogene homologue (avian) (MAF)	−2.55	−2.34
NM_005360	v-Maf musculoaponeurotic fibrosarcoma oncogene homologue (avian) (MAF)	−3.14	−2.51
N50083	CCAAT/enhancer binding protein (C/EBP), delta (CEBPD)	−3.46	−2.94
Cell signaling/growth factor
M33478	Phosducin (PDC)	8.66	7.71
NM_000266	Norrie disease (pseudoglioma) (NDP)	6.28	4.26
BF339445	Chimerin 1 (CHN1)	3.71	3.06
NM_004414	Down syndrome critical region gene 1 (DSCR1)	3.16	2.78
R40892	Protein phosphatase 1E (PP2C domain containing) (PPM1E)	2.65	2.70
NM_013233	Serine threonine kinase 39 (STE20/SPS1 homologue, yeast) (STK39)	2.73	2.69
NM_005627	Serum/glucocorticoid regulated kinase (SGK)	3.29	2.61
M92934	Connective tissue growth factor (CTGF)	2.75	2.50
NM_005863	Neuroepithelial cell transforming gene 1 (NET1)	2.22	2.45
NM_007325	Glutamate receptor, ionotrophic, AMPA 3 (GRIA3)	4.44	2.27
NM_005822	Down syndrome critical region gene 1–like 1 (DSCR1L1)	2.46	2.27
NM_005536	Inositol(myo)-1 (or 4)-monophosphatase 1 (IMPA1)	2.14	2.26
NM_021785	Retinoic acid–induced 2 (RAI2)	2.68	2.25
U50748	Leptin receptor (LEPR)	2.23	2.21
BC003623	Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide (YWHAZ)	2.16	2.19
BE043700	LIM domain protein (RIL)	2.67	2.17
M31159	Insulin-like growth factor–binding protein 3 (IGFBP3)	2.47	2.16
AA707199	Neurotrophic tyrosine kinase, receptor, type 2 (NTRK2)	2.15	2.16
NM_014624	S100 calcium binding protein A6 (calcyclin) (S100A6)	2.00	2.13
BC005916	Pleiotrophin (heparin binding growth factor 8, neurite growth-promoting factor 1) (PTN)	2.25	2.13
BF589322	Thrombospondin (FLJ14440)	2.81	2.10
BF438173	Follistatin (FST)	2.81	2.07
AB044807	PDZ domain protein (Drosophila inaD-like) (INADL)	2.25	2.06
AL565812	Pleiotrophin (heparin binding growth factor 8, neurite growth-promoting factor 1) (PTN)	2.61	2.04
NM_001769	CD9 antigen (p24) (CD9)	2.23	2.02
NM_013409	Follistatin (FST)	3.15	2.00
AU154408	p21/Cdc42/Rac1-activated kinase 1 (STE20 homologue, yeast) (PAK1)	−2.39	−2.01
AA809487	Calcium/calmodulin-dependent protein kinase (CaM kinase) II delta (CAMK2D)	−2.65	−2.16
AI700962	Protein phosphatase 1, catalytic subunit, beta isoform (PPP1CB)	−2.44	−2.34
N32557	Protein phosphatase 1, regulatory (inhibitor) subunit 14C (PPP1R14C)	−4.15	−3.11
AA502643	Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, epsilon polypeptide (YWHAE)	−3.79	−3.40
NM_013314	B-cell linker (BLNK)	−6.19	−3.46
NM_018557	Low-density lipoprotein-related protein 1B (deleted in tumors) (LRP1B)	−36.16	−4.70
Cell cycle/apoptosis
BF686442	Prothymosin alpha (gene sequence 28) (PTMA)	2.74	2.44
N92917	Cell division cycle 42 (GTP binding protein, 25kDa) (CDC42)	−2.89	−2.64
Transporter/channel
NM_002899	Retinol-binding protein 1, cellular (RBP1)	2.38	2.25
AF016535	ATP-binding cassette, sub-family B (MDR/TAP), member 1 (ABCB1)	3.07	2.18
NM_021614	Potassium intermediate/small conductance calcium-activated channel, subfamily N, member 2 (KCNN2)	2.08	2.17
AF016535	ATP-binding cassette, sub-family B (MDR/TAP), member 1 (ABCB1)	2.86	2.13
NM_020689	Solute carrier family 24 (sodium/potassium/calcium exchanger), member 3 (SLC24A3)	−2.73	−2.21
BF111326	Potassium inwardly rectifying channel, subfamily J, member 2 (KCNJ2)	−2.74	−2.24
NM_004171	Solute carrier family 1 (glial high affinity glutamate transporter), member 2 (SLC1A2)	−3.06	−2.33
AL518391	Aquaporin 1 (channel-forming integral protein, 28 kDa) (AQP1)	−2.70	−2.42
AF131796	LAT1-3TM protein (LAT1-3TM)	−4.26	−2.64
R62432	Solute carrier family 24 (sodium/potassium/calcium exchanger), member 3 (SLC24A3)	−3.81	−2.98
AL157452	Solute carrier family 1 (glial high-affinity glutamate transporter), member 2 (SLC1A2)	−30.25	−11.73
ECM
NM_003248	Thrombospondin 4 (THBS4)	4.77	3.85
NM_001996	Fibulin 1 (FBLN1)	11.82	2.26
AI813758	Collagen, type III, alpha 1 (Ehlers-Danlos syndrome type IV, autosomal dominant) (COL3A1)	2.10	2.16
AY029208	Collagen, type VI, alpha 2 (COL6A2)	3.65	2.16
AF059274	Chondroitin sulfate proteoglycan 5 (neuroglycan C) (CSPG5)	−2.94	−2.40
M83248	Secreted phosphoprotein 1 (osteopontin, bone sialoprotein I, early T-lymphocyte activation 1) (SPP1)	−15.60	−5.11
AI808090	Sparc/osteonectin, cwcv and kazal-like domains proteoglycan (testican) 3 (SPOCK3)	−6.14	−5.52
Cytoskeletal
NM_001839	Calponin 3, acidic (CNN3)	2.25	2.23
AL157398	Nebulette (NEBL)	1.99	2.11
NM_001927	Desmin (DES)	2.50	2.05
NM_007246	Kelch-like 2, Mayven (Drosophila) (KLHL2)	−2.46	−2.00
AL537395	Junctophilin 3 (JPH3)	−2.94	−2.18
Muscle-related
NM_002470	Myosin, heavy polypeptide 3, skeletal muscle, embryonic (MYH3)	11.82	9.72
X79857	Troponin T2, cardiac (TNNT2)	8.29	7.84
NM_003476	Cysteine and glycine-rich protein 3 (cardiac LIM protein) (CSRP3)	7.58	5.23
D00943	Myosin, heavy polypeptide 6, cardiac muscle, alpha (cardiomyopathy, hypertrophic 1) (MYH6)	4.44	4.40
AI160292	Myosin, heavy polypeptide 7B, cardiac muscle, beta (MYH7B)	4.97	3.13
U29538	Four-and-a-half LIM domains 1 (FHL1)	2.86	2.86
NM_003063	Sarcolipin (SLN)	3.06	2.80
AF098518	Four-and-a-half LIM domains 1 (FHL1)	3.10	2.76
NM_000257	Myosin, heavy polypeptide 7, cardiac muscle, beta (MYH7)	2.85	2.58
AF220153	Four-and-a-half LIM domains 1 (FHL1)	2.38	2.51
AF020769	Troponin C, slow (TNNC1)	2.66	2.31
AF063002	Four-and-a-half LIM domains 1 (FHL1)	2.17	2.24
NM_004533	Myosin-binding protein C, fast type (MYBPC2)	−12.39	−2.04
NM_020164	Aspartate beta-hydroxylase (ASPH)/junctin	−3.52	−2.99
NM_000036	Adenosine monophosphate deaminase 1 (isoform M) (AMPD1)	−4.26	−3.67
NM_005259	Growth differentiation factor 8 (GDF8)	−5.15	−3.96
AF177173	Tropomodulin 4 (muscle) (TMOD4)	−7.06	−4.40
NM_001104	Actinin, alpha 3 (ACTN3)	−34.57	−16.46
NM_005181	Carbonic anhydrase III, muscle specific (CA3)	−32.45	−18.47
Cell defense/immune response
NM_014333	Immunoglobulin superfamily, member 4 (IGSF4)	4.09	4.05
NM_000700	Annexin A1 (ANXA1)	2.30	2.13
NM_014333	Immunoglobulin superfamily, member 4 (IGSF4)	2.93	2.05
AA209239	Lipase protein (LOC57406)	−2.41	−2.13
NM_002852	Pentaxin-related gene, rapidly induced by IL-1 beta (PTX3)	−3.62	−2.68
Metabolism
NM_000689	Aldehyde dehydrogenase 1 family, member A1 (ALDH1A1)	5.48	5.27
NM_001444	Fatty acid–binding protein 5 (psoriasis-associated) (FABP5)	2.80	2.40
NM_003645	Solute carrier family 27 (fatty acid transporter), member 2 (SLC27A2)	2.30	2.06
NM_003837	Fructose-1,6-bisphosphatase 2 (FBP2)	−2.04	−2.01
BE328402	Hypothetical protein KIAA1434 (KIAA1434) AKA hypothetical protein FLJ11085	−2.94	−2.18
NM_020299	Aldo-keto reductase family 1, member B10 (aldose reductase) (AKR1B10)	−2.83	−2.23
NM_001442	Fatty acid–binding protein 4, adipocyte (FABP4)	−2.84	−2.60
Nucleic acid protein modification
NM_016937	Polymerase (DNA directed), alpha (POLA)	5.47	3.71
AB033064	KIAA1238 protein (KIAA1238)	4.00	3.30
NM_000930	Plasminogen activator, tissue (PLAT)	3.28	3.29
AB033064	KIAA1238 protein (KIAA1238)	3.88	2.98
AI760919	KIAA1238 protein (KIAA1238)	2.77	2.37
NM_000919	Peptidylglycine alpha–amidating monooxygenase (PAM)	2.13	2.28
NM_003769	Splicing factor, arginine/serine-rich 9 (SFRS9)	2.33	2.06
BF038548	Peptidylglycine alpha–amidating monooxygenase (PAM)	2.28	2.05
BG028213	Mitochondrial ribosomal protein L50 (MRPL50)	−2.44	−2.05
AF312735	5′-nucleotidase, cytosolic III (NT5C3)	−2.55	−2.31
AF019638	Guanine deaminase (GDA)	−2.75	−2.37
Other/unknown
U87460	Putative endothelin receptor type B–like protein [Homosapiens], mRNA sequence	8.91	8.47
Other/unknown (continued)
BC002778	Hypothetical protein BC002778 (LOC93408)	6.83	6.82
W22117	HT017 protein (HT017)	4.97	5.05
NM_030804	Hypothetical protein DKFZp434E2135 (DKFZP434E2135)	5.29	4.35
NM_001584	Chromosome 11 open reading frame 8 (C11orf8)	4.99	4.29
AI816291	Hypothetical protein FLJ20152 (FLJ20152)	4.30	3.85
W73230	Hypothetical protein Ells1 (Ells1)	5.33	3.60
NM_019000	Hypothetical protein FLJ20152 (FLJ20152)	3.61	3.41
NM_002112	Histidine decarboxylase (HDC)	3.01	3.36
BF062942	KIAA1000 protein (KIAA1000)	4.78	2.95
NM_004796	Neurexin 3 (NRXN3)	23.55	2.92
AF225421	HT017 protein (HT017)	3.16	2.87
AI129949	Neurexin 3 (NRXN3)	4.44	2.85
NM_004595	Spermine synthase (SMS)	3.08	2.64
NM_005025	Serine (or cysteine) proteinase inhibitor, clade I (neuroserpin), member 1 (SERPINI1)	3.62	2.62
NM_013989	Deiodinase, iodothyronine, type II (DIO2)	2.56	2.61
AJ251830	p53-induced protein PIGPC1 (PIGPC1)	8.30	2.51
AI694320	Hypothetical protein FLJ25270 (FLJ25270)	4.25	2.32
NM_005114	Heparan sulfate (glucosamine) 3-O-sulfotransferase 1 (HS3ST1)	3.07	2.26
NM_000130	Coagulation factor V (proaccelerin, labile factor) (F5)	4.67	2.25
NM_016140	Brain specific protein (CGI-38)	2.74	2.19
NM_030965	Alpha-N-acetylneuraminyl 2,3-betagalactosyl-1,3)-N-acetyl galactosaminide alpha-2,6-sialyltransferase E (SIAT7E)	3.81	2.13
R55784	Homo sapiens cDNA FLJ38396 fis, clone FEBRA2007957, mRNA sequence	3.57	2.05
BG291649	Hypothetical protein MGC45416 (MGC45416)	2.30	2.02
AL574912	Hypothetical protein MGC46520 (MGC46520)	2.08	2.01
BE965029	Homo sapiens cDNA: FLJ21533 fis, clone COL06072, mRNA sequence	−2.46	−2.04
NM_018212	Enabled homologue (Drosophila) (ENAH)	−2.47	−2.06
BC000892	Myelin-associated glycoprotein (HSPC154)	−2.22	−2.06
NM_018447	30-kDa protein (LOC55831)	−2.47	−2.06
AB007880	KIAA0420 gene product (KIAA0420) aka N-ethylmaleimide-sensitive factor attachment protein, beta	−2.33	−2.07
M36532	Carbonic anhydrase II (CA2)	−4.09	−2.08
AB051458	Golgi autoantigen, golgin subfamily a, 2-like, Y-linked (KIAA1671)	−2.86	−2.12
AA424065	Homo sapiens cDNA FLJ32429 fis, clone SKMUS2001014, mRNA sequence	−2.44	−2.14
NM_022469	Hypothetical protein FLJ21095 similar to protein related to DAC and cerberus (FLJ21095)	−3.16	−2.14
AA040057	Protocadherin 20 (PCDH20)	−2.16	−2.19
BF675754	Hypothetical protein MGC22014 (MGC22014)	−2.30	−2.23
AL589605	Hypothetical protein DKFZp451A175 (DKFZp451A175)	−2.32	−2.23
W72338	UDP-glucose ceramide glucosyltransferase (UGCG)	−2.74	−2.36
N64750	Homo sapiens cDNA FLJ37441 fis, clone BRAWH2006543, mRNA sequence	−2.30	−2.48
AI147310	Homo sapiens, clone IMAGE:4814672, mRNA, mRNA sequence	−3.41	−2.79
AI452457	Homo sapiens cDNA FLJ36149 fis, clone TESTI2025174, weakly similar to FYN-BINDING PROTEIN, mRNA sequence	−3.38	−2.80
NM_022443	Myeloid leukemia factor 1 (MLF1)	−3.16	−2.82
H40786	Homo sapiens cDNA FLJ39578 fis, clone SKMUS2003034, mRNA sequence	−3.10	−2.85
AI767756	Heparan sulfate 6-O-sulfotransferase 2 (HS6ST2)	−2.94	−2.99
AF191654	Nudix (nucleoside diphosphate linked moiety X)-type motif 4 (NUDT4)	−4.14	−3.14
T15657	Homo sapiens cDNA: FLJ23075 fis, clone LNG05768, mRNA sequence	−4.67	−3.34
AW024437	Tetratricopeptide repeat-containing protein (LOC118491)	−4.46	−3.38
AW043602	KIAA1946 protein (KIAA1946)	−3.77	−3.55
BF689253	Hypothetical protein FLJ25348 (FLJ25348)	−5.12	−3.63
AI124053	Hypothetical protein FLJ35961 (FLJ35961)	−3.77	−3.69
AI376433	Peptidoglycan recognition protein (KIAA1912)	−5.22	−4.63
EST
BF110588	Band 4.1-like protein 4	3.28	3.13
AW294729	ESTs	3.73	3.05
AI971535	Unnamed protein product [Homo sapiens], mRNA sequence	7.03	3.00
AA772920	ESTs	3.61	2.38
AI979261	Unknown (protein for IMAGE:2899977) (Homo sapiens), mRNA sequence	2.62	2.27
AA746863	Band 4.1–like protein 4	2.39	2.26
BC003096	EST, moderately similar to RIL_HUMAN LIM protein RIL (Reversion-induced LIM protein) (H. sapiens)	9.15	2.25
BF589413	ESTs	2.30	2.21
D60329	ESTs	2.74	2.19
EST (continued)
BF110551	ESTs, highly similar to S50128 glutamate receptor 3 isoform flip- human (H. sapiens)	2.22	2.11
BG503479	ESTs, weakly similar to CBF1 interacting corepressor (Homo sapiens) AKA Human recepin	2.49	2.08
AL134451	ESTs	3.04	2.06
AA628511	ESTs, weakly similar to A55479 CDK6 inhibitor p18-human (H. sapiens)	2.15	2.05
AI590238	ESTs	2.39	2.01
BE671491	ESTs	−2.14	−2.01
AI147384	Unnamed protein product (Homo sapiens), mRNA sequence	−2.83	−2.04
AI798207	ESTs	−2.56	−2.06
AA977975	Unknown (protein for MGC:46487) (Homo sapiens), mRNA sequence	−16.73	−2.12
BF513800	ESTs, weakly similar to hypothetical protein FLJ20234 (Homo sapiens)	−2.74	−2.25
T65537	ESTs	−9.52	−2.44
AW880875	ESTs	−2.76	−2.47
AA002211	ESTs	−14.77	−2.97
R41565	ESTs	−4.38	−3.86
AI741629	ESTs, weakly similar to T17307 hypothetical protein DKFZp566O084.1- human (H. sapiens)	−7.25	−4.09
AI333373	ESTs, weakly similar to RIKEN cDNA 2310038H17 [Mus musculus]	−7.22	−5.65

*  Transcripts listed met acceptance criteria (consistent call with all replicates and change of ≥2-fold, using both RMA and MAS 5.0). A positive difference denotes higher expression in the orbital layer.

Figure 2.

 

(A) Hierarchical clustering of the gene probes identified as differentially expressed between the orbital (O1, O2) and global (G1, G2) layers of monkey MR muscles. Transcripts identified are those at the intersection of data obtained with the Affymetrix MAS and RMA algorithms. The two independent replicates of each layer are represented. The scale at the top denotes normalized expression levels (red, high expression; blue, low expression). (B) Functional grouping of the differentially expressed genes for EOM layers. One hundred three transcripts were identified with twofold or higher expression in the orbital layer and 78 had higher expression in the global layer. Genes with unknown functions and those that could not be assigned to one of the functional categories were classified as other.

Figure 3.

 

Hierarchical clustering of expression levels for the subset of muscle-related transcripts across the replicates of orbital (O) and global (G) layers. Red: higher expression; blue: lower expression (refer to the scale in Fig. 2 ). Note the concordance in the expression levels between the two replicates of each layer for most probes.

Table 2.

 

qPCR Analysis of Select Transcripts Differentially Expressed between the Orbital and Global Layers

Accession No.	Gene Name (symbol)	Primers*	Difference (n-fold), †
						Array (RMA)	qPCR
X79857.1	Troponin T2, cardiac (TNNT2)	5′-CAGCGCATCCGGAATGAG-3′	8.29	24.93
		5′-TGTAACCCCCAAAATGCATCA-3′
NM_003476	Cysteine and glycine-rich protein 3 (CSRP3)	5′-CCAGCAACCCTTCCAAATTC-3′	5.23	20.11
						5′-GCCACCTCCCATAACCTTCTC-3′
NM_004796.1	Neurexin 3 (NRXN3)	5′-GAGCGGCCACAAGAAACAGA-3′	2.92	12.97
		5′-GGCGTGGATAGAAATAACTGTGAA-3′
NM_001996.1	Fibulin 1 (FBLN1)	5′-GCTTGCCTTGCCATGAGAA-3′	2.26	12.01
		5′-GCCCATGCGGAAAACCA-3′
AF063002	Four-and-a-half LIM domains 1 (FHL1) adenosine	5′-ACTGCGTGGATTGCTACAAGAA-3′	2.24	6.54
						5′-GACTGGGTGGCTCACTCTTGAC-3′
NM_000036.1	Monophosphate deaminase 1 (isoform M) (AMPD1)	5′-CCAATGCAATTCCACTTTACCA-3′	−3.66	−2.85
						5′-CTTGCCACTTCGCACATATCA-3′
NM_005259.1	Growth differentiation factor 8 (GDF8)	5′-GAATCACGATGCTGTCGTTACC-3′	−3.96	−2.98
						5′-TCCAGAGCAGTAATTGGCCTTATA-3′
M36532.1	Carbonic anhydrase II (CA2)	5′-GCTAAACCGGGCCTTCAGA-3′	−2.08	−4.06
		5′-GGAAGGAGGCCACGAGGAT-3′
NM_005181	Carbonic anhydrase III (CA3)	5′-CATGAGAATGGCGAGTTCCA-3′	−18.47	−7.91
		5′-AACAGGCAGGATGGGTCAAA-3′
NM_001104.1	Actinin, alpha 3 (ACTN3)	5′-ACAGCGCACGATCAGTTCAA-3′	−16.46	−12.61
		5′-TACGTCTGGCAGATCTTCTGGAT-3′

*  Forward and reverse primer sequences used for qPCR.

†  Microarray expression changes (n-fold) represent the average of two independent replicates of orbital layer versus global layer, calculated using RMA. qPCR data represent the mean of triplicate analyses, using the forward and reverse primers indicated. For both microarray and qPCR data, positive changes denote higher expression in the orbital layer.

The authors thank Marty Veigl and Patrick Leahy and the Gene Expression Array Core Facility for assistance with DNA microarray; Scott Shnider and Jason Feuerman for assistance with data analysis; Claire Doerschuk, Sarah E. Richer, and the Pediatric Imaging Center and Department of Pediatrics at Case Western Reserve University for access to and training on the Leica LCM; and Denise Hatala for invaluable technical assistance in specimen preparation for LCM.