Automated Methods for the Analysis of Skeletal Muscle Fiber Size and Metabolic Type - International Review of Cell and Molecular Biology - page 310

for both TRMU and PDXK), whereas the antibody for GPATCH3 labeled
fibers that are negative for TRMU and PDXK. As our data from human
skeletal muscle biopsies suggest that TRMU is expressed in Type I fibers,
it therefore seems likely that PDXK may also be prevalent in Type I
fibers, whereas GPATCH3 may be preferentially expressed in Type II pro-
teins. Thus, identification of serial sections would be a useful feature for
image-analysis software to be developed for histospatial phenotyping.
This novel approach, applied to the images from the HPA, has several
advantages. First, the “wet work” associated with the HPA project has
already been completed, to a large degree, and will likely be finished around
2015; thus, an enormous treasure trove of protein expression information is
being assembled, which will likely yield enhanced understanding of tissue-
specific and, in the case of skeletal muscle, fiber-specific protein expression.
Second, the results are at the level of protein, rather than messenger RNA;
since proteins are directly responsible for cellular and physiological func-
tions, and since there can often be discrepancies in mRNA versus protein
expression for a given gene, the HPA data will likely be very relevant to fur-
thering our understanding of tissue function in health and disease. The use of
antibody labeling, however, is not without caveats; a major concern, for
example, is the degree to which the antibodies specifically label the intended
proteins ( Berglund et al., 2008 ); indeed, a common feature of the HPA
entries are that positive staining may only be obtained with a subset of
the antibodies that are tested for a particular protein. Thus, additional lab-
oratory testing will be required to confirm the fiber-specific protein expres-
sion inferred from the approach.
A logical extension of the approach is to use bioinformatic strategies to
explore potential protein expression/interaction networks that are likely to
control or influence skeletal muscle differentiation, function, growth, or
pathologies. Our initial exploration involved using the STRING v9.0
( http://string.embl.de ) database, which incorporates protein-protein inter-
action (PPI) data from over 1000 different organisms ( Franceschini et al.,
2013 ). As expected, our initial query using the fiber-specific gene list iden-
tified relationships between proteins known to be components of muscle
sarcomeric structures,

, isoforms of SERCA2 (e.g.,
ATP2A1), and the troponin subunits. Furthermore, using STRING's
gene-enrichment option, this cellular component cluster was predicted to
be sarcomeric (GO:0030017) and its molecular function to be a structural
constituent

such as MyHC- b

(GL:008307) of the muscle systems biological process

(GL:0003012; Fig. 7.16 A-C). The histospatial phenotyping and