ABSTRACT
Muscle-restricted transcription of sarcomeric actin genes is negatively
controlled by the zinc finger protein YY1, which is down-regulated at the protein level during myogenic differentiation. To identify cellular proteins that might mediate the function/stability of YY1 in muscle cells,
we screened an adult human muscle cDNA library using the yeast two-hybrid cloning system. We report the isolation and characterization of a novel protein
termed YAF2 (YY1- associated factor 2) that interacts with YY1. The YAF2 cDNA encodes a 180
amino acid basic protein (pI 10.5) containing a single N-terminal C2-X10-C2 zinc finger. Lysine clusters are present that may function as a nuclear
localization signal. Domain mapping analysis shows that the first and second
zinc fingers of YY1 are targeted for YAF2 protein interaction. In contrast to
the down-regulation of YY1, YAF2 message levels increase during
in vitro
differentiation of both rat skeletal and cardiac muscle cells. YAF2 appears to
have a promyogenic regulatory role, since overexpression of YAF2 in C2
myoblasts stimulates myogenic promoter activity normally restricted by YY1. Co-transfection of YY1 reverses the stimulatory effect of YAF2. YAF2 also
greatly potentiates proteolytic cleavage of YY1 by the calcium- activated protease m-calpain. The isolation of YAF2 may help in understanding the
mechanisms through which inhibitors of myogenic transcription may be
antagonized or eliminated by proteolysis during muscle development.
Several research groups have independently cloned the cDNA encoding the multi-functional transcription factor YY1 (UCRBP, [delta], NF-E1) (
1
-
4
). Nuclear factors identified in other studies as FACT-1 (
5
), CF-1 (
6
), CBF (
7
), LBF (
8
), MAPF-1 (
9
) and NMP-1 (
10
) have since been found to be identical or similar to YY1. This broad body of
literature indicates that YY1 regulates, by transactivation and
transrepression, the expression of many cellular as well as viral genes. The
functional versatility of YY1 can be attributed in part to the wide occurrence
of its binding site in the promoter regions, both proximal and distal, of
numerous genes. In addition, selection of YY1 binding sites from a random pool
of oligonucleotides revealed a high degree of flexibility in the DNA
recognition mode of YY1 (
11
). The
cis
-acting elements recognized by YY1 appear to bear no dyad symmetry and can
act in an orientation-dependent manner in certain promoter contexts (
12
). Studies also showed that a properly positioned YY1 binding site is capable of
functioning as an initiator element, promoting transcriptional initiation
in vitro
in concert with RNA polymerase II and TFIIB (
13
).
These diverse and unique transcriptional properties of YY1 suggest that there
must exist multiple cellular factors that interact with and mediate the
functions of YY1. Indeed, YY1 has been found to interact, either physically or
functionally, with the c-Myc oncoprotein (
14
), the adenovirus E1A product (
4
), the GC box binding factor Sp1 (
15
), the E1A-associated factor p300 (
16
), TFIIB (
13
) and immunophilins (
17
). The interplay between YY1 and E1A proteins is particularly intriguing, since
E1A appears to act as a molecular switch in the transactivation and transrepression functions of YY1 (
4
). This finding also indicates that, like p53 (
18
), E2F (
19
) and Rb (
20
), YY1 may be preferentially targeted by certain viral oncoproteins. However,
the relationship, if any, between cell cycle control and the action of YY1
remains to be elucidated.
Studies of YY1 have frequently been based on the use of immortalized cell lines,
which have been found to express YY1 constitutively at high level. This
presumably implies a functional importance of YY1 in proliferating cell types.
On the other hand, our studies have provided the first evidence that YY1
protein is developmentally down-regulated during muscle development (
5
,
21
,
22
). Down-regulation of YY1 during muscle development is consistent with the finding
that YY1 antagonizes the myogenic function of SRF by binding competition (
5
,
21
,
23
). Thus, a high level of YY1 expression in proliferating myoblasts suppresses
promoter activity of the sarcomeric actin genes, which can be reversed by down-regulation of YY1 in post-mitotic muscle cells (
22
). Although the mechanism underlying down-regulation of YY1 during muscle development remains unclear, our previous
study suggested the potential involvement of a protease-mediated mechanism (
21
). As a further step toward understanding the myogenic regulation of YY1, we
have, using the yeast two-hybrid cloning system, undertaken the identification of muscle regulatory proteins that may modulate the function/stability of YY1. We here report the isolation from two human muscle cDNA libraries of a novel zinc finger factor termed YAF2 (YY1-associated factor 2), which physically interacts with YY1 and selectively potentiates
proteolytic cleavage of YY1 by the myogenic calcium-activated protease m-calpain (
24
-
26
). Overexpression of YAF2 in myoblasts further stimulates promoter activity of
muscle-restricted genes. Thus, although many YY1 binding proteins have been
characterized, YAF2 is unique in its promyogenic effect and may play a
regulatory role in muscle development.
Skeletal and cardiac muscle tissues (from 10 rats for each preparation) were
isolated from 2-day-old Sprague-Dawley rats. Minced skeletal muscle (from hind limbs) and
cardiac ventricles were each gently agitated in 15 ml 0.05% trypsin + 1 mM EDTA
at 4oC overnight. Excess trypsin solution was removed following overnight agitation and tissue fragments incubated at 37oC for 10 min, after which 10 ml minimal essential medium (MEM) containing 10% horse
serum was added to inactivate trypsin. Collagenase (Worthington) and DNase I
(Sigma) were then added to a final concentration of 1 mg/ml and 0.1 mg/ml
respectively for another 10-20 min incubation. Tissue fragments were triturated for three to four
cycles with the addition of 10 ml medium for each cycle. Skeletal myoblasts
were preplated twice (1 h each) and cardiac myocytes preplated once (3 h) to
deplete fibroblasts. Skeletal myoblasts were plated in MEM + 10% fetal bovine serum on Primaria
culture dishes (Falcon) at a density of 3120 cells/mm2. Cardiac myocytes were plated in MEM + 10% horse serum at a density of 1040
cells/mm2. Unattached cells were removed and fresh medium added 16 h after plating. To
facilitate cardiac differentiation, medium was switched to MEM supplemented with 5 [mu]g/ml insulin, 1 [mu]g/ml transferrin and 5 ng/ml selenite 48 h after plating (designated as 0 h). Cultured mouse C2 myoblasts were
maintained in MEM + 10% fetal bovine serum and 50 [mu]g/ml gentamicin at 37oC in a humidified 5% CO2 atmosphere. Myoblasts were transiently transfected using the calcium phosphate method and luciferase assays were as described previously (
22
).
The
Nco
I-
Bam
HI YY1 fragment derived from pGEM-human YY1 (
4
) was inserted into the
Nco
I and
Bam
HI sites of the GAL4 DNA binding domain vector pDG1, which was generated from
pAS1 (
27
) by deleting the
Eco
RI-
Nde
I HA epitope contained in pAS1. pGAD-YAF2(1.5 kb) was isolated by yeast two-hybrid screening (see below). pGAD-YAF2(0.8 kb) was constructed by inserting the 0.8 kb
Eco
RI fragment of YAF2 into the
Eco
RI site of pGAD424 (Clontech) in the sense orientation. pGAD-YAF2 (frameshift) was made by end-filling the
Bam
HI fusion site of pGAD-YAF2(1.5 kb), which introduces 4 bp at the fusion junction and thus changes the YAF2 reading frame. pBluescript-YAF2-8 was rescued and excised from the [lambda]ZAPII vector according to the manufacturer's manual
(Stratagene). The
Nco
I-
Bam
HI YAF2 fragment derived from pBluescript-YAF2-8 was inserted into the
Nco
I and
Bam
HI sites of pGEX (
28
), generating pGEX-YAF2, which overexpresses GST-YAF2 fusion protein in bacteria in the presence of IPTG. The pGEX-TFIIB construct was kindly provided by M.-J.Tsai (Baylor College of Medicine). pET-YY1 (full-length and DM3), pTC21, pMSV-YY1, p(G)5E1B-Luc and pSK-Luc were as described previously (
22
). pTK-Luc was constructed by inserting the
Xba
I-
Bgl
II fragment of the thymidine kinase promoter from pBLCAT2 into pGL2-Basic (Promega) digested with
Nhe
I and
Bgl
II. pET-YY1 C-terminal truncation mutants were constructed by BAL31 deletion. pCMV-YAF2 was constructed by inserting a full-length
Eco
RI YAF2 fragment into pcDNA3 (InVitrogen) in the sense orientation.
The screening protocol was detailed previously (
29
). The yeast host SFY526, which contains an integrated copy of the
GAL1
-
lacZ
reporter gene, was first transformed with pDG-YY1 using the lithium acetate protocol (
30
). Cells were grown in medium lacking tryptophan and transformed with an adult
human skeletal muscle two-hybrid cDNA library (Clontech) constructed in a GAL4 activation domain
vector. Yeast colonies were selected on 150 mm agar plates lacking tryptophan
and leucine over a period of 5 days. Colonies were transferred to filter paper
and processed by the [beta]-galactosidase filter assay (
29
). Candidate colonies were picked, grown up and further examined by secondary
and tertiary filter assays. Plasmid DNA was isolated from positive yeast colonies, amplified in bacteria and reintroduced into yeast with either pDG1 or pDG-YY1 DNA. The [beta]-galactosidase assay was again performed to ensure YY1-dependent activity. Three separate large scale transformations were
performed, examining over 1 000 000 yeast colonies.
An adult human skeletal muscle-derived [lambda]ZAPII cDNA library (Stratagene) was screened with the 5' 0.8 kb YAF2 DNA fragment labeled by random priming. Briefly, 6 * 105 plaques were lifted with nitrocellulose filters. Probed filters were first rinsed in 2* SSC + 1% SDS at room temperature for two 5 min cycles, following which
filters were washed in 2* SSC + 1% SDS at 65oC for two 30 min cycles. The filters were finally washed in 0.2* SSC + 1% SDS at 68oC for 30 min, air dried and autoradiographed. Subsequent
secondary and tertiary screening were performed under the same condition.
Ten micrograms of pBluescript-YAF2-8 were linearized with
Bam
HI and the DNA digest was treated with 10 [mu]g proteinase K at 50oC for 30 min. This was followed by a double phenol/chloroform extraction and ethanol precipitation. The pelleted DNA was dissolved in
DEPC-treated H2O and transcribed with T7 RNA polymerase at 37oC for 1 h. RNA products were cleaned up by phenol/chloroform extraction and
ethanol precipitation.
In vitro
translation was carried out as directed by the manufacturer using nuclease-treated rabbit reticulocyte lysates (Promega). 35S-Labeled proteins were resolved by 12% SDS-PAGE. Dried gels were processed for autoradiography.
Bacterially expressed GST-YAF2 and GST-TFIIB fusion proteins were produced in
Escherichia coli
strain BL21 by 1 mM IPTG induction for 2 h. Cells were lysed in freshly
prepared 50% urea supplemented with 5 mM DTT and heated in a 60oC water bath for 30 min. Insoluble material was removed by centrifugation
at 10 000 r.p.m. for 10 min. The supernatant was dialyzed in a cold room in TBS
(100 mM Tris, pH 8, 100 mM NaCl) for 4 h. Clarified dialysate was applied to a
5 ml glutathione-agarose column. The column was washed extensively with TBS and bound
proteins were eluted with 20 mM reduced glutathione. Eluted proteins were dialyzed again in the cold to remove glutathione. Bacterially expressed YY1 was purified by nickel chelate affinity
chromatography as described previously (
22
).
Twenty microliters of glutathione agarose and 50 [mu]l bacterial protein lysate containing GST fusion proteins were mixed and
agitated in a cold room for 30 min. Agarose beads were washed three times with
TBS containing 0.05% Tween-20 (TTBS) and then incubated with 50 [mu]l of the protein lysate to be examined. The mixture was agitated for
another 30 min in a cold room and the agarose beads washed three times with
TTBS. Washed beads were boiled in 50 [mu]l SDS sample buffer and the supernatant analyzed by SDS-PAGE and Western blot.
The
in vitro
cleavage assay was done in a total volume of 20 [mu]l. In each reaction, 2.5 [mu]g purified recombinant YY1 were used as substrate. GST-YAF2 or GST-TFIIB was preincubated with YY1 on ice for 10 min,
following which 2 [mu]g purified m-calpain (Calbiochem) and CaCl2 (1 mM final) were added to initiate the cleavage reaction. Cleavage reactions
were continued at 37oC for 20 min. Reactions were terminated with SDS-PAGE sample buffer supplemented with [beta]-mercaptoethanol. Samples were boiled for 5 min,
electrophoresed by SDS-PAGE and then analyzed by Western blot using YY1 antibody.
Total RNA was isolated using the guanidinium thiocyanate method. Isolated RNA
was dissolved in formamide, mixed with ethidium bromide (final concentration
0.1 [mu]g/ml) and separated on a 1.2% agarose gel containing 37% formaldehyde. RNA
was transferred to nylon membrane (BioRad) and immobilized by UV crosslinking. Membrane was blocked in prehybridization solution (1% SDS, 10% dextran sulfate, 1 M NaCl) for 5 h and then incubated with the
random-primed 0.9 kb YAF2 DNA probe at 55oC overnight. Probed membrane was washed twice with 2* SSC + 1% SDS at 65oC for 30 min, followed by one wash with 0.2* SSC + 1% SDS at 68oC for 30 min. The membrane was air dried and
autoradiographed.
Western blotting was performed as described previously (
31
). In brief, proteins were electrophoresed by 10% SDS-PAGE and electrotransferred to Immobilon-P membrane (Millipore), which was then probed with a primary
antibody. A chemiluminescence kit (Amersham) was used to detect transferred
proteins.
The yeast host SFY526 was co-transformed with plasmid pDG-YY1 and a human muscle cDNA library (Clontech) constructed in the
GAL4 activation domain vector pGAD10. Colonies were transferred to filter paper
and processed for [beta]-galactosidase assay. More than 1 000 000 colonies were screened in three separate
transformations. Blue colonies were picked and further purified by secondary
and tertiary assays. This stringent screening resulted in the isolation of a
cDNA clone designated YAF2, for YY1-associated factor 2, containing a 1.5 kb cDNA insert. To identify the minimal functional domain encoded by the 1.5
kb YAF2 clone, the 5' 0.8 kb region was subcloned into pGAD424, generating pGAD-YAF2(0.8 kb) (Fig.
1
). The GAD-YAF2 reading frame was also altered by insertion of 4 bp at the fusion site, creating pGAD-YAF2(frameshift). [beta]-Galactosidase enzyme assays showed that while pGAD-YAF2(0.8 kb) retained the ability to interact
with YY1, the YAF2 frameshift construct failed to stimulate [beta]-galactosidase activity (data not shown), suggesting that the observed augmentation of [beta]-galactosidase activity is most likely mediated by a
specific protein-protein interaction between YAF2 and YY1.
DNA sequence analysis of the 0.8 kb region of the YAF2 clone revealed a novel
sequence not deposited in the GenBank and EMBL compilations. This region
appears to contain a potential open reading frame, although the start codon was
not found within this region (see below). The 0.8 kb YAF2 DNA fragment was then
used as a hybridization probe to screen a [lambda]ZAPII cDNA library (Stratagene) made from human skeletal muscle cDNAs.
Approximately 6 * 105 plaques were screened, resulting in the isolation of several cDNA clones, among
which a 1.6 kb clone designated YAF2-2 and a 1.0 kb clone designated YAF2-8 were further analyzed (Fig.
1
). Sequence analysis of the YAF2-8 clone revealed that it contains an additional 5' 250 bp region not present in the original YAF2 clone. The complete
sequence of this [lambda]ZAPII-derived YAF2-8 clone is shown in Figure
2
, with the deduced amino acid sequence. An open reading frame begins with the
putative ATG codon at nucleotide position 254. This ATG codon is the likely
translation start site, as the sequence flanking the initiation codon conforms
to the Kozaki (
32
) consensus sequence (overlined in Fig.
2
). The open reading frame ends with the stop codon at nt 794 (denoted by an
asterisk), indicating that YAF2 is composed of 180 amino acids with a predicted
molecular mass of 19.88 kDa. The partial YAF2 clone derived from the yeast two-hybrid library thus encodes a truncated protein lacking 18 N-terminal amino acids, which nonetheless retains the putative zinc
finger domain (see below). This finding also indicates that the 18 amino acid
peptide of YAF2 is not required for interaction with YY1.
Examination of the YAF2 protein sequence revealed a potential zinc finger of the
Cys2-X10-Cys2 variety in the N-terminal region (underlined in Fig.
2
). A limited sequence homology (WDCSVC) is noted between the sequence
encompassing the first cysteine pair of YAF2 and the zinc finger motifs of
several nucleoporins and Ran binding protein 2 (
33
,
34
), although the significance of this homology remains unclear. We have found that the recombinant YAF2 protein exhibits a moderate affinity for DNA-cellulose (data not shown). This DNA binding property of YAF2 may be due
to the preponderance of lysine (23/180) and arginine (12/180) residues,
particularly near the N-terminal half of the protein. The abundance of these two amino acids also
contributes to a calculated isoelectric point of 10.5. Several clusters of
basic amino acid residues may constitute a nuclear localization signal, in
particular the lysine clusters located at amino acids 75-79 and 94-100. Also notable in the sequence of YAF2 protein is its high
serine content (29/180), especially in the C-terminal portion.
Both the YAF2 and YAF2-8 cDNA clones were isolated from human skeletal muscle cDNA libraries. We
therefore examined whether expression of YAF2 message might be developmentally
regulated in primary cultures of neonatal rat skeletal myotubes and ventricular
cardiac myocytes. As shown in Figure
3
, total RNAs from the cultured muscle cells exhibited a YAF2 message of ~2 kb in size. That this YAF2 message was detected under stringent wash
conditions (0.2* SSC, 1% SDS at 68oC) suggests that the YAF2 sequence is highly conserved between human
and rat. Interestingly, the levels of YAF2 mRNA were found to increase during
both skeletal and cardiac differentiation. The down-regulation of YY1 and up-regulation of SRF during skeletal myoblast differentiation (
21
,
22
) are shown as proper controls of myogenic culture condition (Fig.
3
, bottom). Thus, in contrast to the down-regulation of YY1 during myogenesis, YAF2 exhibits a myogenic expression
pattern similar to that of the promyogenic regulatory protein SRF. It should be
mentioned that although several cellular factors have been shown to interact
with YY1, YAF2 is unique in that its expression is up-regulated during sarcomeric muscle developement.
To verify that the YAF2-8 cDNA clone indeed encodes a protein of the expected size, we transcribed
the linearized pBluescript-YAF2-8 construct with T7 RNA polymerase and translated the resultant
transcripts in a reticulocyte lysate in the presence of 35S-labeled methionine. Figure
4
shows that translation of the YAF2-8 transcript produced a protein with an apparent molecular mass of ~25 kDa relative to a set of protein markers (lanes 2 and 3). This
protein was not observed in the lysate lacking YAF2 transcript (lane 1). This
in vitro
translation study indicates that the start and termination codons noted above
are most likely functional. To facilitate purification and characterization of
YAF2 protein, the coding region of YAF2 was fused in-frame with the glutathione S-transferase (GST) gene carried in the bacterial expression vector
pGEX (
28
). Overexpression of the GST-YAF2 fusion protein was induced with 1 mM IPTG and the protein examined
by SDS-PAGE. Figure
5
A shows that bacterially expressed GST-YAF2 fusion protein (lane 2) exhibits an apparent molecular mass of 50 kDa, whereas the GST carrier protein (lane 3) migrates at a position expected for its size (26 kDa). The GST-YAF2 fusion protein was further used for the protein binding assay
described below.
The ability of the yeast two-hybrid system to demonstrate protein-protein interactions has been illustrated by many groups (
29
,
36
,
37
). Isolation of YAF2 using YY1 as the target gene in a yeast genetic approach indicates that YAF2 and YY1 interact
in vivo
. To provide biochemical evidence for this interaction, we utilized YAF2 protein
affinity beads to capture YY1 from crude nuclear extracts as well as from bacterial lysates containing overexpressed YY1. The GST-YAF2 fusion protein was first bound to glutathione-agarose to generate GST-YAF2 protein affinity beads (
28
). These affinity beads were then incubated with either crude nuclear extracts
or bacterial lysates, followed by extensive washing. Retained proteins were
probed with YY1 antibody by Western blot. Figure
5
B showed that the GST-YAF2 affinity beads (lanes 1, 3 and 5), but not the GST beads (lanes 2, 4
and 6), captured YY1 in the binding assay. That YAF2 captured YY1 from crude
nuclear extracts strengthens the notion that the interaction is quite specific.
The affinity of the protein interaction appears to be quite high, since
inclusion of 0.5 M NaCl and 0.2% Tween-20 in the washing buffer (lanes 5 and 6) failed to disrupt the YAF2-YY1 protein complex.
To further illustrate the specificity of YY1-YAF2 interaction, a series of YY1 deletion mutants was generated (see
22
for YY1 domains), purified and examined by the GST binding assay. Among the
seven YY1 mutants examined, four (N271, N200, N170 and N152) failed to interact
with YAF2 (Fig.
6
). The human YY1 protein contains four cysteine zinc finger motifs located at
positions 298, 327, 355 and 385 (
4
). Our assay thus indicates that the region encompassing the first and second
zinc fingers of YY1 is essential for YAF2 interaction, whereas the third and
fourth fingers are dispensable, even though they are required for the DNA
binding activity of YY1 (
22
). Use of the N-terminal deletion mutant DM3 further reveals that the N-terminal half of YY1 is not required for YAF2 interaction. It should
be mentioned that, similar to our conclusions here, the first and second zinc
fingers of YY1 have also been found to be critical and sufficient for Sp1
interaction (
15
). These findings suggest that the first two zinc fingers of YY1 may adopt a
unique conformation suitable for protein-protein interactions.
Our previous study suggested that YY1 may be proteolytically degraded in
differentiated muscle cells (
21
). We have recently found that YY1 but not SRF is selectively cleaved by the
calcium-activated neutral protease m-calpain (data not shown). m-Calpain (calpain II) requires millimolar levels of calcium for
catalytic activity, as opposed to [mu]-calpain (calpain I), which requires micromolar levels of calcium (
24
). In contrast to the inhibitory role of YY1 in myogenesis, the activity of m-calpain was found to be up-regulated and essential during myogenic differentiation (
25
,
38
,
39
). Since protein complex formation is frequently associated with functional or
structural alterations, we examined the effect of the YAF2-YY1 protein interaction on m-calpain-mediated cleavage of YY1. Western blotting was used to
monitor the protein cleavage process. Figure
7
shows that incubation of YY1 with m-calpain in the presence of 1 mM CaCl2 generated a 40 kDa cleavage product. Preincubation of GST-YAF2 with YY1 dramatically enhanced the cleavage of YY1, as evidenced by a decrease in full-length YY1 and accumulation of the 40 kDa cleavage product in a
dose-dependent fashion. On the other hand, preincubation of YY1 with the basal
transcription factor TFIIB (fused to GST), which has been shown to interact
with YY1 (
13
), had no effect on the cleavage of YY1 by m-calpain. That YAF2 itself may possess a protease activity was ruled out by
the demonstration that incubation of YY1 and YAF2 in the presence of Ca2+ (without m-calpain) caused no degradation of YY1 (see left three lanes in Fig.
7
). Furthermore, YAF2 has no appreciable effect on trypsin- or proteinase K-mediated limited cleavage of YY1 (data not shown), indicating a
degree of specificity in the concerted action of YAF2 and m-calpain.
YY1 suppresses myogenic transcription in part through its inhibitory effect on
the expression of the sarcomeric [alpha]-actin genes (
21
,
22
). Direct repression of the skeletal [alpha]-actin promoter, in particular, can be achieved by a high level of
YY1, which precludes SRF from occupying the promoter SRE1 site through binding
competition. Since YAF2 appears to be up-regulated during myogenic differentiation (Fig.
3
) and promotes the degradation of YY1 by m-calpain (Fig.
7
), it is expected to exert a positive effect on myogenic transcription. This
possibility was examined by transiently transfecting pCMV-YAF2 into C2 myoblasts. Figure
8
shows that co-transfected YAF2 activated luciferase activity driven by the skeletal [alpha]-actin promoter (pSK-Luc) in a dose-dependent fashion. We also examined the effect of
YAF2 on promoters lacking YY1 binding sites. Neither the construct driven by
the thymidine kinase promoter (pTK-Luc) nor by the adenovirus E1B minimal promoter [p(G)5E1B-Luc] was significantly affected by transfected YAF2 (Fig.
8
). We also found that co-transfection of YY1 reversed the positive effect of YAF2, supporting the
hypothesis that the function of YAF2 may be to inactivate YY1 in part through
proteolysis. This finding is thus consistent with the developmental up-regulation of YAF2 and down-regulation of YY1 (Fig.
3
).
This work was supported by grants from the American Heart Association (94-004) and the Muscular Dystrophy Association (PNU0694).
*To whom correspondence should be addressed. Tel: +1 716 829 3106; Fax: +1 716
829 2725; Email: chunglee@acsu.buffalo.edu
REFERENCES
Return