CALL FOR PAPERS Stem Cell Biology

Atrophy, inducible satellite cell activation, and possible denervation ofsupraspinatus muscle in injured human rotator-cuff muscle

Deanna Gigliotti,1 Jeff R. S. Leiter,2 Bryce Macek,3 Michael J. Davidson,4 Peter B. MacDonald,5

and Judy E. Anderson1

1Department of Biological Sciences, University of Manitoba, Winnipeg, Manitoba, Canada; 3College of Medicine, Universityof Manitoba, Winnipeg, Manitoba, Canada; 4Department of Radiology, University of Manitoba, Winnipeg, Manitoba,Canada; 5Section of Orthopedics, Department of Surgery, University of Manitoba, Winnipeg, Manitoba, Canada; and 2PanAm Clinic, Winnipeg, Manitoba, Canada

Submitted 20 May 2015; accepted in final form 29 June 2015

Gigliotti D, Leiter JR, Macek B, Davidson MJ, MacDonald PB,Anderson JE. Atrophy, inducible satellite cell activation, and possi-ble denervation of supraspinatus muscle in injured human rotator-cuffmuscle. Am J Physiol Cell Physiol 309: C383–C391, 2015. Firstpublished July 1, 2015; doi:10.1152/ajpcell.00143.2015.—The highfrequency of poor outcome and chronic pain after surgical repair ofshoulder rotator-cuff injury (RCI) prompted this study to explore thepotential to amplify muscle regeneration using nitric oxide (NO)-based treatment. After preoperative magnetic resonance imaging(MRI), biopsies of supraspinatus and ipsilateral deltoid (as a control)were collected during reparative surgery for RCI. Muscle fiber diam-eter, the pattern of neuromuscular junctions observed with alpha-bungarotoxin staining, and the �:ε subunit ratio of acetylcholinereceptors in Western blots were examined in tandem with experimentsto determine the in vitro responsiveness of muscle satellite cells toactivation (indicated by uptake of bromodeoxyuridine, BrdU) by theNO-donor drug, isosorbide dinitrate (ISDN). Consistent with MRIfindings of supraspinatus atrophy (reduced occupation ratio and tan-gent sign), fiber diameter was lower in supraspinatus than in deltoid.ISDN induced a significant increase over baseline (up to 1.8-fold), inthe proportion of BrdU� (activated) Pax7� satellite cells in supraspi-natus, but not in deltoid, after 40 h in culture. The novel applicationof denervation indices revealed a trend for supraspinatus muscle tohave a higher �:ε subunit ratio than deltoid (P � 0.13); this ratioinversely with both occupancy ratio (P � 0.05) and the proportion ofclusters at neuromuscular junctions (P � 0.05). Results implicatepossible supraspinatus denervation in RCI and suggest NO-donortreatment has potential to promote growth in atrophic supraspinatusmuscle after RCI and improve functional outcome.

Pax7; nitric oxide; acetylcholine receptor; supraspinatus; deltoid

“ROTATOR CUFF” MUSCLES (supraspinatus, infraspinatus, teresminor, and subscapularis) function to rotate and abduct the armand stabilize the shoulder joint. Rotator-cuff injury (RCI) fromdamage to cuff muscles or their tendons may be either chronicand age-related (4, 5) or an acute result of sports injury (23,29). RCI most commonly affects the supraspinatus, and sec-ondary pathology, including muscle atrophy and fibro-fattyinfiltration, results in pain, weakness, loss of function, andbiomechanical shoulder-joint instability (3, 12, 15). Most tearsrequire reparative surgery; however, recurrence of tears fol-

lowing surgery is common, with failure rates ranging from 30to 94% (12).

Regulation of muscle satellite cells (SCs), the stem cells formuscle regeneration and growth, is implicated in the responseto and rehabilitation from RCI, since there are reportedly fewerSCs in RCI muscle (12). The SC population and its respon-siveness to an activating stimulus needed for regenerationcould potentially be targeted for new treatments to improverecovery, since surgical repair of a torn tendon may not restorejoint stability and function.

SC are mitotically quiescent and identifiable by their satelliteposition between the plasma membrane and myofiber basallamina (27), and by the proteins they express, including Pax7(4, 6, 27). Transit of a SC from the quiescent G0 state into theG1 (activated) stage of the cell cycle is regulated by hepatocytegrowth factor (5, 35) and nitric oxide (NO) (2, 3). DecliningSC function can lead to myopathy and accompanies conditionssuch as muscular dystrophy and age-related atrophy. However,while strategies such as treatment with the NO-donor drug,isosorbide dinitrate (ISDN), have been explored as possibletreatment for age-related atrophy and muscular dystrophy (19,26, 28), the potential of NO-based treatment for RCI has notbeen explored (3, 26). This study aimed to investigate whetherNO treatment in culture would increase SC activation inpathological supraspinatus compared with uninjured deltoidmuscle (as an internal control).

Acetylcholine receptor (AChR) distribution on the musclefiber membrane and subunit composition reveal the innervationstatus of muscle, histologically and in protein assays, respec-tively. During development, before fibers are innervated,AChRs are arranged in a linear pattern on fibers. With inner-vation, AChRs cluster (45) at neuromuscular junctions (NMJs).Five subunits form each receptor, and the identity of onesubunit changes during development: specifically, the gamma(�) subunit, which shows fetal characteristics, transitions to amore efficient adult epsilon (ε) subunit after innervation. Aftermuscle is denervated, both histological and subunit characterchange: AChR distribution becomes more linear on the sarco-lemma, and the AChR subunit pattern reverts to the fetal, �isoform (45). These functional transitions make AChR distri-bution and composition useful indicators of innervation status.

This study aimed to investigate myogenic and neurogenicaspects of RCI, by examining differences in fiber morphometry

Address for reprint requests and other correspondence: J. E. Anderson,Univ. of Manitoba, 50 Sifton Rd., Winnipeg, MB, Canada R3T 2N2 (e-mail:Judith.Anderson@umanitoba.ca).

Am J Physiol Cell Physiol 309: C383–C391, 2015.First published July 1, 2015; doi:10.1152/ajpcell.00143.2015.

0363-6143/15 Copyright © 2015 the American Physiological Societyhttp://www.ajpcell.org C383

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

and innervation (AChR distribution and subunit composition)and SC responsiveness to an activating stimulus in biopsysamples of supraspinatus muscle in RCI. These features ofatrophy in supraspinatus were compared with those in unin-jured, ipsilateral deltoid muscle of the same participant, toidentify the mechanisms underlying RCI and ultimately findpotential target areas for preventing RCI or improving theoutcome of RCI repair. The hypothesis was that supraspinatusmuscle would demonstrate atrophy, indicators of denervation,a reduced baseline level of SC activation, and increased re-sponsiveness of SCs to activation by ISDN compared with thedeltoid.

MATERIALS AND METHODS

Participants were recruited to this study according to a protocolapproved by the Human Ethics Review Board at the University ofManitoba (protocol no. B2010:074) and the Winnipeg RegionalHealth Authority Research Review Access Committee (WRHA 2010-019) and signed an informed consent form prior to starting the study.Participants had clinical evidence (history, physical exam, and MRIfindings, below) suggesting a rotator-cuff tear and had failed at least6 mo of conservative management of RCI (which usually includesphysiotherapy) at the time they consented for surgery.

MRI. MRI scans took place at the Pan Am Clinic (Winnipeg, MB)and were performed using a 1.5-T Siemens MAGNETOM scanner(Siemens, Erlangen, Germany). The occupation ratio, defined as thecross-sectional area of the supraspinatus muscle belly in relation to thecross-sectional area of the supraspinatus fossa, was used as a quanti-tative measure of atrophy, and calculated from the scapular Y-view(37, 43). An occupation ratio of �50% indicates supraspinatus muscleatrophy (37). The tangent sign (43) is a qualitative indicator ofatrophy. In the Y-view a line is drawn connecting the coracoid processto the scapular spine; a positive tangent sign is observed when thesupraspinatus muscle belly lies completely below that line (37). Toassess fatty infiltration, the classification system of Goutallier (10) forCT scans, recently adapted for MRI (8), was applied to the supraspi-natus and infraspinatus. The scale runs from 0 to 4, describingintramuscular fat, as follows. A grade of 0 indicates no intramuscularfat; 1 indicates small streaks of fat without significant accumulation;2 corresponds to significant fat but less than the amount of muscle; 3

describes equal amounts of fat and muscle; and 4 indicates there ismore fat than muscle. MRI measurements were made first by findingthe Y-view as described above. The image was then examined furtherusing NIH ImageJ software.

Tissue collection. Biopsies of RCI-affected supraspinatus andhealthy, ipsilateral deltoid muscle were obtained during arthroscopicsurgical repair. The deltoid muscle was used as a control to theaffected supraspinatus as it is not affected in patients reporting RCIsymptoms, and the contralateral supraspinatus is often affected eventhough it may be asymptomatic (42). Another study demonstrating areduction in fiber diameter with increasing pathology of the rotatorcuff, from partial- to full-thickness tears (11), also used the deltoidmuscle as a control and reported that in the presence of supraspinatuspathology on one side, changes were more evident in the contralateralsupraspinatus muscle than the deltoid muscle ipsilateral to the RCI. Tominimize bias, the identity of muscle samples was coded until afterdata analysis. Parts of each biopsy were prepared for histology, aculture study of the capacity for NO-induced SC activation, and forprotein study by Western blotting.

Histology. Muscle was fixed in 4% paraformaldehyde in phos-phate-buffered saline and prepared for sectioning (7 �m thick);sections were collected onto slides for morphometry after stainingwith hematoxylin and eosin (H & E) (35). Slides were visualized andphotographed using a digital Sentech camera and imaging software(StCamSWare v1.0.0.9).

A second set of slides was collected for immunostaining to analyzethe pattern of AChR distribution at NMJs identified by direct fluo-rescence (14). In brief, slides were washed, fixed with acetone,incubated with fluorescein-conjugated �-bungarotoxin (Invitrogen,F1176, Carlsbad, CA), and mounted with Vectashield media (VectorLabs H1000, Burlingame, CA). Coded sections were scanned system-atically, and up to 20 nonoverlapping fields (depending on biopsysize) were photographed using a 40X oil immersion lens on a ZeissApoTome (Jena, Germany) for analysis.

SC activation in culture. A portion of each biopsy was divided intwo and cultured in 60-ml petri dishes at 37°C with 5% CO2 for 40 hin basal growth medium (41) with 0.65 �M (200 �g/l) BrdU, with orwithout 1 mM ISDN, based on our previous report (18). Each piecewas initially a maximum of 2–3 mm in any dimension, and due to thenature of fibers collected and cut in the biopsy, small bundles of fibersfanned out, once immersed in the culture medium. The 40-h timepoint was selected to restrict the time for BrdU incorporation intoDNA to only one S-phase of the cell cycle. The need to accommodatethe anticipated time lag in activation (following biopsy manipulation)in participants aged 49–65 yr-of-age was based on previous reports in

Table 1. Summary of demographic data and MRI findings

Value

Demographic dataAge, yr 57.6 � 1.5 (range 45–69)Sex Male, n � 8; Female, n � 5Shoulder affected Right, n � 9; Left, n � 4Height, m 1.74 � 0.03Weight, kg 92.4 � 5.6BMI, kg/m2 30.4 � 1.4Smoking history Yes, n � 4; No, n � 9Injury mechanism Traumatic, n � 9; unknown, n � 4

MRI findingsSS tear 1 PT, 12 FTSS tear thickness, cm 2.24 � 0.30 cmMuscles affected by tear SS, n � 9; SS�IS, n � 1; SS�TM, n � 2;

md � 1Tangent sign Negative, n � 10; positive, n � 1, md � 2Occupancy ratio 0.75 � 0.06, md � 2Goutallier classification 1.18 � 0.23 (range 0 to 3), md � 2

Values are means � SE. Demographic data are from 13 participants. ForMRI findings, 2 participants had tear described and measured in MRIs, butother data were not available; these are recorded as missing data (md). BMI,body mass index; SS, supraspinatus; FT, full tear; PT, partial tear; IS,infraspinatus; TM, teres major.

Fig. 1. Representative preoperative MRI Y-shaped views of participants witha rotator-cuff tear, in which a line is drawn from the coracoid process (C) tothe apex of the scapular spine (S). A: in this MRI, the supraspinatus musclebelly lies above the line, so the tangent sign is considered negative. B: in thisMRI, the supraspinatus muscle belly lies below the line connecting thecoracoid process and the apex of the scapular spine, indicative of a positivetangent sign. Orientation is indicated as superior (sup), inferior (inf), anterior(ant), and posterior (post), and supraspinatus is indicated by an asterisk (*).

C384 SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

old-rodent muscle (1, 17, 18). Following incubation, tissue was fixedand sectioned, and slides were stored at 20°C until use.

Sections were processed for immune detection of BrdU and Pax7using double immunofluorescence to identify mitotically active SCs,as reported (13, 28). Overnight incubation at 4°C with a primaryantibody solution containing mouse anti-BrdU (1:200, 11–170376001, Roche, Basel, Switzerland) and rabbit anti-Pax7 (1:150,

ab34360, AbCam, Cambridge, MA) was followed by quenching ofendogenous peroxidase with 3% hydrogen peroxide (10 min). Sec-tions were washed and incubated (1 h) with secondary antibodies:Dylight 488-conjugated goat anti-rabbit (1:200, 111-487-003) andDylight 649-conjugated goat anti mouse (1:200, 115-497-003, Jack-son ImmunoResearch), washed, and mounted (Vectashield, VectorLabs).

Coded sections were scanned systematically and 20 nonoverlap-ping fields (depending on biopsy size) were photographed using a 40Xoil immersion lens on a Zeiss ApoTome. Fluorescent SCs werecounted and tabulated in Excel as activated (BrdU�/Pax7�) or not(BrdU/Pax7�) in each section, and the proportion of activated SCswas calculated (the number of BrdU�/Pax7� SCs divided by thetotal number of Pax7� SCs). The change in activation induced byISDN was calculated as the proportion of activated SCs after ISDNdivided by the baseline level of activation in the paired untreatedbiopsy from the same participant.

Immunoblotting for AChR subunits. The levels of � and ε AChRsubunit proteins were quantified using Western blotting according toa standard protocol (19) with modifications to detect multiple epitopesfrom the same membrane (39). In brief, samples were placed inRNA-later stabilizing solution (Life Technologies, Burlington, Can-ada) and frozen until protein extraction. Total protein was assayed ona standard curve using a Pierce BCA Protein Assay Kit (Thermo-Fisher Scientific, Rockford, IL). Samples (20 �g protein) were loadedon 9% polyacrylamide gels and run for 1 h before transfer (100 V for1 h) to polyvinylidene fluoride membranes (Fisher Scientific, Ottawa,Canada). Membranes were first probed using goat anti-ε AChR(diluted 1:150, sc-1454, Santa Cruz, Dallas TX) and incubated withbovine anti-goat-HRP secondary antibody (1:300, sc-2350, SantaCruz). Bands were visualized using chemiluminescence and imagedusing a VersaDoc chamber linked to the QuantityOne software pro-gram (BioRad, Hercules, CA). Membranes were quenched in 27%hydrogen peroxide, and serially reprobed, visualized, and quantifiedfor additional epitopes (39), including � AChR subunit using rabbitanti-� AChR (1:250, sc-13998, Santa Cruz) detected using donkeyanti-rabbit-HRP secondary antibody (1:5,000, NA-9340V, GE Health-care Life Sciences, Piscataway, NJ)and -actin using a cocktail ofmouse anti--actin primary antibody (1:1,000, sc-81178, Santa Cruz)and goat anti-mouse-HRP secondary antibody (1:5,000, A2304,Sigma Aldrich, St. Louis, MO). Membranes were stained in PonceauRed (Sigma, St. Louis, MO) to visualize and quantify total proteinloaded in each sample.

Statistics. Data were compiled in Microsoft Excel spreadsheets foranalysis and graphing (mean, SE). Parametric tests of score data,nonparametric Chi-squared tests (to study distributions), and correla-tions and linear regressions were used to analyze the findings col-lected from assays in different parts of the biopsy from each partici-pant, as appropriate, using Excel or JMP-SAS statistical software. Aprobability of P � 0.05 was used to indicate statistical significance.Power analyses were conducted post hoc, using data for the distribu-

Fig. 2. Representative micrographs of hematoxylin and eosin (H&E)-stained muscle. Scale bar, 50 �m. A: representative view of the longitudinal orientationthrough fibers sometimes observed. B: multiple layers of adipocytes indicating fatty infiltration of the muscle tissue. C: large amount of fibrous connective tissuebetween muscle fibers, characteristic of muscle fibrosis.

Fig. 3. Fiber diameter and distribution. A: fiber diameter (mean � SE) insupraspinatus and deltoid muscle biopsies from participants with rotator-cuffinjury (RCI). A total of 875 fibers in the supraspinatus and 772 fibers in thedeltoid were measured (n � 12). *Significant difference (paired t-test, P �0.003). B: frequency distribution of fiber diameter for supraspinatus anddeltoid muscle, plotted in 5-�m bins (x-axis). There was a significant shifttoward a smaller fiber diameter in supraspinatus muscle compared with thedeltoid (Chi-squared statistics, P � 0.005).

C385SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

tion of the AChR staining pattern at NMJs and the �:ε ratio of AChRsubunits.

RESULTS

Participants and surgery. Fifteen participants completed aninformed consent form and were enrolled in the study. Twowere excluded intraoperatively: one had no rotator-cuff tear atarthroscopy, and there were technical difficulties at the time ofbiopsy retrieval in the other. MRI data were missing in twoparticipants; therefore the study included 13 participants over-all, with 11 for MRI analysis. Demographic data are reportedin Table 1. Average time from symptom onset to surgery was142 wk, although one participant had surgery 10 yr after onset;without this participant, the average symptomatic period was103 wk from injury to surgery. Arthroscopy confirmed afull-thickness rotator-cuff tear in 12 participants and a partial-thickness tear in one. Other diagnoses made at surgery in-cluded one labral tear, one partial subscapularis tear, threeType-2 superior labral tear from anterior to posterior (SLAP)lesions, one SLAP I lesion, an AC arthrosis in two participants,and one teres minor with mild to moderate muscle atrophy.Rotator-cuff repair was performed in nine participants (allfull-thickness tears); biceps tenotomy (n � 9), acromioplasty(n � 12), bursectomy (n � 5), and labral debridement (n � 2)were also performed during surgery in these participants.

MRI. Representative MRIs are shown in Fig. 1. MRI find-ings are summarized in Table 1 (n � 11), including the size ofthe rotator cuff tear of supraspinatus, muscles affected by thetear, tangent sign, occupancy ratio, and Goutallier classifica-tion. Two participants had tear extension to the infraspinatustendon. The occupancy ratio was 0.75 � 0.06 (mean � SE)with a median Goutallier classification score of 1 (range: 0–3).

Muscle histology. Fiber diameter and the relative amount ofmuscle, fat, and fibrous connective tissue varied among biop-sies from the different participants (Fig. 2). Typical sectionsfrom the deltoid had many fibers, few to no adipocytes, andonly thin layers of connective tissue within the muscle. A fewbiopsies, typically from supraspinatus muscle, displayed pre-dominantly adipocytes and/or fibrous connective tissue sur-rounding small fibers. The diameter of 875 supraspinatus and772 deltoid fibers was measured before sections were decoded(20–102 fibers per biopsy). Fiber diameter (Fig. 3A) wassignificantly smaller (P � 0.01) in the supraspinatus (17.9 �0.3 �m) than in the deltoid (21.8 � 0.3 �m), and the frequencydistribution of fiber diameter (Fig. 3B) showed significantlymore small-diameter fibers in supraspinatus compared withdeltoid [P � 0.005, Chi-squared (df � 9) � 73.4].

AChR distribution at NMJs. NMJ staining appeared astightly defined spots on fiber membranes in both muscles.NMJs were classified along a range of complexity as one of asinglet, a doublet, a cluster (3 or more dots less than 2.5 �mapart), or a line (three or more dots in a straight, lineararrangement), as illustrated in Fig. 4. There were 42–209 NMJsclassified for each biopsy. In both muscles, the majority ofAChRs appeared as single dots of �-bungarotoxin fluorescencewith a diameter of 1.5–2.5 �m on the sarcolemma.

The pattern of NMJs was analyzed by the proportions ofAChRs distributed as clusters (representing innervated fibers)or lines (representing denervated fibers) in each of the su-praspinatus and deltoid muscles, as visualized with �-bunga-rotoxin fluorescence. Singlet and doublet patterns were notincluded as they could not be distinguished from lines orclusters sectioned in a different plane. The distribution ofstaining pattern (proportion of clusters vs. lines) did not differ

Fig. 4. Representative micrographs of neuromuscular junctions (NMJs) observed by direct fluorescence using �-bungarotoxin staining. Singlet, doublet, andcluster describe one, two, or three or more dots in close proximity, respectively; a line is defined as three or more dots in a row. Pairs of images in fluorescence(top row) and DIC (bottom row) show a characteristic singlet (A), doublet (B), line (C), and cluster (D) from supraspinatus (A and C) and deltoid muscles (Band D). Scale bar, 10 �m.

C386 SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

between supraspinatus and deltoid (P � 0.3). The proportion ofclusters was 0.49 in supraspinatus (923 NMJs photographed)and 0.51 in deltoid (675 NMJs photographed). The proportionof linearly arranged AChRs in the deltoid increased with age(R2 � 0.42, P � 0.02); however, this correlation was not notedfor the RCI supraspinatus.

AChR subunits. Two AChR subunits, � and ε, were quanti-fied by Western blot, as an indicator of innervation status in themuscles, � representing AChRs on denervated fibers and εrepresenting AChRs on normal, innervated fibers (Fig. 5, A andB). The amount of the � subunit protein was highly variable;relative to total protein, neither subunit differed between su-praspinatus and deltoid. Supraspinatus showed a tendencytoward a higher �:ε AChR ratio than the deltoid (P � 0.13, Fig.5C). The ratio of �:ε AChR subunits in supraspinatus increasedwith age (R2 � 0.44, P � 0.02), but this was not the case forthe ratio from deltoid muscle. There was no statistical relation-ship between the �:ε AChR subunit ratio and the proportion ofclustered AChRs at NMJs, analyzed for separate muscles orpooling data for the two muscles.

SC activation. Pax7� cells were most often located in aperipheral position around muscle fibers, as expected for SCs(Fig. 6). The pattern of sarcomeres was retained in fibersegments throughout the biopsy, indicating that conditionsincluding oxygenation were able to maintain tissues in culture.In all sections, the number of nonproliferating Pax7�/BrdU

SCs (Fig. 7) was larger than the number of activated andproliferating Pax7�/BrdU� SCs. In cultured supraspinatusbiopsies, ISDN increased the proportion of BrdU�/Pax7�SCs compared with the baseline level in untreated supraspina-tus (P � 0.03); by contrast, deltoid muscle SCs showed nochange in activation in response to ISDN compared withbaseline.

Scatterplots were used to consider possible statistical rela-tionships between activation (baseline and change with ISDNtreatment), NMJ staining (proportion of clusters), and age insupraspinatus and deltoid together (Fig. 8A) and between NMJstaining, fiber diameter, �:ε AChR subunit ratio, and occu-pancy ratio (MRI) in supraspinatus (Fig. 8B). Correlationswere made on data compiled for each participant and includedvariables from assessment of any of the 3 parts of each biopsy.

There was no statistical relationship between the proportionof clusters at NMJs and the change in SC activation induced byISDN (Fig. 8A1). The proportion of clusters at NMJs decreasedwith increasing age (R2 � 0.24, P � 0.05, Fig. 8A2). BaselineSC activation only tended to be related to the proportion ofclusters at NMJs (R2 � 0.05, P � 0.15, Fig. 8A3). Interest-ingly, there was a negative relationship between baseline SCactivation and the change in SC activation with ISDN, in thatthose biopsies with the lowest baseline had the highest increasein activation after ISDN (R2 � 0.28, P � 0.01, Fig. 8A4).Considering only the RCI supraspinatus biopsies, the propor-

Fig. 5. Acetylcholine receptor (AChR) sub-units. A: Western blots of AChR ε and �subunits (and -actin) prepared from a sin-gle blotted gel of protein extracts of su-praspinatus (SS) and deltoid (D) muscle bi-opsies of participants with RCI. Bands werevisualized from the same gel, blotted,probed, quenched, and reprobed, in series,for each protein (ε-AChR, �-AChR, and-actin) according to our reported protocol(39). B: the original scans of blots (intensityon the y-axis) for each epitope from 2 of thelanes shown in A, from deltoid (D, lane 2)and supraspinatus muscles (SS, lane 5). C:the �:ε AChR subunit ratio in supraspinatusand deltoid muscle (mean � SE) in 11 bi-opsies from participants with RCI. The ratioof �:ε AChR subunits did not differ signifi-cantly between muscles (P � 0.13).

C387SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

tion of clusters at NMJs was not significantly related to meanfiber diameter (R2 � 0.14, Fig. 8B1). The occupancy ratio ofsupraspinatus by MRI was negatively related to the proportionof clusters at NMJs (R2 � 0.39, P � 0.05, Fig. 8B2), negativelyrelated to the �:ε AChR subunit ratio (R2 � 0.18, P � 0.05,

Fig. 8B3), and positively related to the mean fiber diameter insupraspinatus (R2 � 0.15, P � 0.05, Fig. 8B4).

DISCUSSION

The most novel finding of this study was that SCs in thesupraspinatus muscle with RCI responded to activation byISDN whereas those in the control deltoid muscle did not. Inthe same participants, the supraspinatus was atrophic both byMRI assessment and histology, the latter compared with acontrol muscle, the contralateral deltoid. The finding of su-praspinatus atrophy is consistent with previous literature onmouse models of RCI or denervation, and other studies onhuman muscle atrophy and neuromuscular disorders (9, 11, 15,21, 24, 32, 38, 45). To examine the mechanism of atrophy(disuse vs. denervation), the innervation status of RCI su-praspinatus muscle was analyzed for the first time using twoindices in the same biopsy: the pattern of AChR staining inNMJs and the �:ε ratio of AChR subunits. Although neitherindex of innervation status gave significant results, at leastpartly due to a small sample size, the ratio of �:ε AChR proteinsubunits tended to be higher in the RCI supraspinatus than incontrol deltoid. Together with the findings of SC responsive-ness and atrophy, the trend in innervation status is interpretedas leaving open the possibility that denervation plays a role inRCI pathophysiology. For instance, partial denervation of thesupraspinatus muscle (not captured in this small sample aschanges in NMJ clusters or the �:ε ratio of AChR subunits)could theoretically account for the responsiveness of SCs inthat muscle to activation by the ISDN stimulus in culture dueto disuse in RCI. It would be interesting to investigate whetherthere is fiber-type grouping and/or larger motor units in su-praspinatus (due to collateral axon sprouting in compensation

Fig. 6. Micrographs of immunofluorescence staining for BrdU and Pax7. Tissues are from sections of supraspinatus muscle removed from 40-h in culture (withor without ISDN), fixed and stained to examine BrdU incorporation in Pax7� satellite cells. In each panel, the Cy5 channel (top left), GFP channel (top right),merged channels (bottom left) and DIC images (bottom right) illustrate a satellite cell that is positive for Pax7 staining and negative for BrdU staining, foundin a section of an untreated supraspinatus muscle biopsy after culture (A), and a satellite cell that is Pax7� and BrdU�, found in a section of an untreated deltoidmuscle biopsy after culture (B).

Fig. 7. Graph showing the proportion of proliferating, BrdU� satellite cells(Pax7�) as a proportion of total Pax7� satellite cells (mean, SE) in sectionsof supraspinatus (SS) and deltoid (D) muscles prepared after 40 h in culture inthe presence of BrdU, with or without ISDN treatment. ISDN, isosorbidedinitrate. *Significant difference from untreated supraspinatus muscle (P �0.05).

C388 SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

for partial denervation) in those with RCI who progress toreparative surgery. Notably in this study, the ipsilateral deltoidserved as a control, since the contralateral supraspinatus isoften affected asymptomatically in unilateral RCI. As well, thedeltoid is reportedly not affected in patients with symptoms ofRCI (42). Although the potential for some disuse atrophy in theipsilateral deltoid (in accommodating the symptomatic su-praspinatus) cannot be ruled out, the observed statistical dif-ferences between the two muscles would be even more impor-tant if the deltoid were also atrophied compared with deltoid ina healthy age-matched individual. Therefore, this study wasdesigned to biopsy ipsilateral deltoid muscle as a comparisonto the supraspinatus muscle of RCI participants, and control forinfluences of age, life history, and activity, on the two muscles.While all participants had failed at least 6 mo of conservativetreatment when they consented for surgery, the effects ofphysiotherapy treatment on the supraspinatus and deltoid mus-cle of the affected shoulder were beyond the scope of this studyand would be of future interest in relation to present findings onmuscle fiber diameter, SC responsiveness to an activatingstimulus, and innervation status.

SCs in supraspinatus were activated by a NO-donor drug,despite atrophy and possible contribution by denervation,whereas SCs in the control deltoid muscle were not. SCs fromRCI supraspinatus were more activated after culture with ISDNrelative to baseline and did not differ from the baseline ofactivation in deltoid cultures, which did not respond to ISDN.Research on the effect of ISDN on SCs in human muscle isscarce (12, 15), although the therapeutic potential of activatingSCs by drugs that act through the NO-mediated pathway inmouse (3, 18, 19, 26, 28, 31) and zebrafish (44) is currently inclinical trials for muscular dystrophy (NCT01359670 andNCT01168908). One study that reported a decrease in thenumber of proliferating cells in biopsies of patients with full-and partial-thickness tears concluded that RCI muscle showed

a decrease in proliferative capacity (24) as a drawback topositive therapeutic outcome. However, that study did notspecifically examine SC responses to an activating stimulus.Since retention of SC proliferative capacity would enableregeneration despite muscle atrophy, and in mice can persistinto very old age to repair muscle injury (16), these results areparticularly promising.

The lack of SC response to ISDN in cultures of normaldeltoid muscle may be due to muscle-specific differences in thedose response of SCs to NO (3, 33, 40). Since NO levels in amuscle could be anticipated to support optimal SC activity forthe architecture and functional demands of a particular muscle,SCs of healthy adult muscle may not respond to a NO stimulus,or the ISDN concentration used in this study. SCs in muscle ofolder animals would also be theoretically less responsive to anactivating stimulus than in younger animals, as shown in mice(18). Conversely, injured supraspinatus may have a relativelylow level of NO due to disuse (and/or possible partial dener-vation) compared with a healthy muscle, which could increaseSC sensitivity to the NO-donor drug, and raise SC activationup to the baseline level seen in control muscle. This reasoningwould fit with the proposed U-shaped model of SC activationas a function of NO concentration, previously reported (40),since the SC-activation response to ISDN observed in biopsiesfrom individual participants ranged widely. For instance, twosupraspinatus biopsies with the highest level of ISDN-inducedSC activation (Fig. 8A1), a relative 1.8-fold increase frombaseline, were from a 65 year-old male and a 56 year-oldfemale. Both were smokers with relatively high BMI (35 and38.4), and had surgery at 73 and 69 wk post-onset; theirsupraspinatus mean fiber diameter was 16 and 23.7 �m, re-spectively (Fig. 8B1). Two other supraspinatus biopsiesshowed a decrease in SC activation after ISDN to 0.7-fold oftheir respective baseline levels. These participants, both male(53 and 65 yr old), had lower BMI (27 and 25 kg/m2) and

Fig. 8. Scatterplots of regressions for the level of SC activation (baseline and change with ISDN treatment), the pattern of NMJ staining (proportion of clusters),and participant age for findings in supraspinatus and deltoid, combined together (A); and the pattern of NMJ staining, fiber diameter, �:ε AChR subunit ratio,and occupancy ratio (MRI) in supraspinatus muscles (B). Trend lines (with equation and R2 values) are shown for linear regressions. *Significant relationships(slope statistically different from zero). Note that the correlations are on data collected as a set from each participant and may relate to comparison of assessmentsfrom any of the 3 parts of the biopsy, prepared for sectioning (immediate fixation), culturing, or protein extraction.

C389SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

supraspinatus mean fiber diameter (8 and 20 �m, respectively)at surgery (61 and 100 wk post-onset, respectively). Based onrecent reports that semaphorin3A, a putative axon-guidancemolecule, is secreted by SCs after activation and early differ-entiation (7, 7, 30, 34, 36), possible changes in SC-derivedsemaphorin3A with age, injury, or inflammation in humansmay be related to innervation status and the potential forsuccessful surgical repair of RCI.

Muscle atrophy was assessed using both histology and MRIoccupancy ratio. Fiber diameter in supraspinatus was statisti-cally related to occupancy ratio despite the variable intervalbetween MRI and surgery in study participants (36.3 � 4.5wk). RCI and injury to the supraspinatus nerve (25) leads tomuscle atrophy and fibro-fatty infiltration (15) after both par-tial- and full-thickness tears (24), and similar findings werereported in a mouse model of massive RCI (22). Fiber atrophyaccompanied by fibro-fatty infiltration (Goutallier sign), lowoccupancy ratio, and a trend to a higher �:ε ratio of AChRsubunits would be consistent with denervation, as demon-strated in rat experiments (9, 21, 45), clinical reports onneuromuscular and neurogenic disorders (9, 45), and findingsthat supraspinatus nerve damage limits active movement bysupraspinatus muscle (25). It was interesting that �:ε AChR-subunit ratio and the proportion of AChR clusters at NMJsobserved by �-bungarotoxin staining were not statisticallyrelated, possibly as the molecular-level change in subunitcomposition in each receptor complex may occur before orafter redistribution of receptor clusters and NMJ remodeling atthe cellular level (9, 45). As well, the distribution of AChRs inthe staining pattern of NMJs could be influenced by thediameter range in each biopsy (smaller fibers would theoreti-cally increase the number of NMJs sampled in a sectionthrough the motor end-plate region). Nonetheless, the novelapplication of two indices of denervation in arthroscopic biop-sies of muscle in RCI suggests further study of the possibleimplication of denervation in RCI would be useful, nowinformed by power analyses showing for human RCI thatstudying AChR subunits requires n � 61 samples and studyingAChR staining pattern requires n � 124 samples.

Correlation studies, even within this small sample of partic-ipants, found an increase in the �:ε AChR subunit ratio indeltoid muscle with increasing age. Injury and subsequentdisuse in RCI-affected muscle may have obscured such astatistical relationship for the RCI supraspinatus muscle of thesame participants. The possibility of denervation-related mus-cle atrophy would be consistent with the characteristic ofage-related atrophy and the increased susceptibility to andprevalence of injuries such as RCI with age (20). Althoughfindings implicate denervation in RCI, they do not establishdenervation in the etiology of RCI. However, the presentfindings suggest the value of exploring the innervation status ofRCI supraspinatus in the clinic, as well as other factors such asvascularization, to assist decision making about the approachto treatment, since even anatomically effective tendon repairwill not restore function in the absence of innervation. Theseexperiments are the first, to our knowledge, to study thephysiology of human SCs concurrent with cellular and molec-ular indicators of innervation status in RCI supraspinatusmuscles. The clinical potential for a treatment based on NOdelivery to muscle in RCI patients may be useful in promotingsupraspinatus growth as a means of helping restore muscle

strength to achieve maximal functional outcome after tendonrepair.

ACKNOWLEDGMENTS

We are grateful for the expert technical assistance of R. Upadhaya.

GRANTS

Research funding (to J. R. S. Leiter, P. B. MacDonald, and J. E. Anderson)that supported this work was received from a Canadian Orthopedic ResearchLegacy Award from the Canadian Orthopedic Foundation, the AlexanderGibson Fund, and the Department of Surgery at the University of Manitoba.We are grateful for support from a Manitoba Health Research CouncilStudentship (to D. Gigliotti).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: D.G., J.R.L., B.M., P.B.M., and J.E.A. performedexperiments; D.G., J.R.L., B.M., M.J.D., and J.E.A. analyzed data; D.G.,J.R.L., M.J.D., P.B.M., and J.E.A. interpreted results of experiments; D.G. andJ.E.A. prepared figures; D.G. and J.E.A. drafted manuscript; D.G., J.R.L., andJ.E.A. edited and revised manuscript; D.G., J.R.L., B.M., M.J.D., P.B.M.,and J.E.A. approved final version of manuscript; J.R.L., P.B.M., and J.E.A.conception and design of research.

REFERENCES

1. Allen RE, Sheehan SM, Taylor RG, Kendall TL, Rice GM. Hepatocytegrowth factor activates quiescent skeletal muscle satellite cells in vitro. JCell Physiol 165: 307–312, 1995.

2. Anderson J, Pilipowicz O. Activation of muscle satellite cells in single-fiber cultures. Nitric Oxide Biol Chem 7: 36–41, 2002.

3. Anderson JE. A role for nitric oxide in muscle repair: nitric oxide-mediated activation of muscle satellite cells. Mol Biol Cell 11: 1859–1874, 2000.

4. Charge SBP, Rudnicki MA. Cellular and molecular regulation of muscleregeneration. Physiol Rev 84: 209–238, 2004.

5. Chazaud B. Dual effect of HGF on satellite/myogenic cell quiescence.Focus on “High concentrations of HGF inhibit skeletal muscle satellite cellproliferation in vitro by inducing expression of myostatin: a possiblemechanism for reestablishing satellite cell quiescence in vivo”. Am JPhysiol Cell Physiol 298: C448–C449, 2010.

6. Dhawan J, Rando TA. Stem cells in postnatal myogenesis: molecularmechanisms of satellite cell quiescence, activation and replenishment.Trends Cell Biol 15: 666–673, 2005.

7. Do MK, Sato Y, Shimizu N, Suzuki T, Shono J, Mizunoya W,Nakamura M, Ikeuchi Y, Anderson JE, Tatsumi R. Growth factorregulation of neural chemorepellent Sema3A expression in satellite cellcultures. Am J Physiol Cell Physiol 301: C1270–C1279, 2011.

8. Fuchs B, Weishaupt D, Zanetti M, Hodler J, Gerber C. Fatty degen-eration of the muscles of the rotator cuff: Assessment by computedtomography versus magnetic resonance imaging. J Shoulder Elbow Surg8: 599–605, 1999.

9. Gattenlohner S, Schneider C, Thamer C, Klein R, Roggendorf W,Gohlke F, Niethammer C, Czub S, Vincent A, Muller-Hermelink HK,Marx A. Expression of foetal type acetylcholine receptor is restricted totype 1 muscle fibres in human neuromuscular disorders. Brain 125:1309–1319, 2002.

10. Goutallier D, Postel JM, Bernageau J, Lavau L, Voisin MC. Fattymuscle degeneration in cuff ruptures—preoperative and postoperativeevaluation by CT scan. Clin Orthop Rel Res 78–83, 1994.

11. Irlenbusch U, Gansen HK. Muscle biopsy investigations on neuromus-cular insufficiency of the rotator cuff: a contribution to the functionalimpingement of the shoulder joint. J Shoulder Elbow Surg 12: 422–426,2003.

12. Isaac C, Gharaibeh B, Witt M, Wright VJ, Huard J. Biologic ap-proaches to enhance rotator cuff healing after injury. J Shoulder ElbowSurg 21: 181–190, 2012.

13. Janke A, Upadhaya R, Snow WM, Anderson JE. A new look atcytoskeletal NOS-1 and beta-dystroglycan changes in developing muscle

C390 SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from

and brain in control and mdx dystrophic mice. Dev Dyn 242: 1369–1381,2013.

14. Kong JM, Anderson JE. Dystrophin is required for organizing largeacetylcholine receptor aggregates. Brain Res 839: 298–304, 1999.

15. Laron D, Samagh SP, Liu XH, Kim HT, Feeley BT. Muscle degener-ation in rotator cuff tears. J Shoulder Elbow Surg 21: 164–174, 2012.

16. Lee AS, Anderson JE, Joya JE, Head SI, Pather N, Kee AJ, GunningPW, Hardeman EC. Aged skeletal muscle retains the ability to fullyregenerate functional architecture. Bioarchitecture 3: 25–37, 2013.

17. Leiter JR, Peeler J, Anderson JE. Exercise-induced muscle growth ismuscle-specific and age-dependent. Muscle Nerve 43: 828–838, 2011.

18. Leiter JRS, Anderson JE. Satellite cells are increasingly refractory toactivation by nitric oxide and stretch in aged mouse-muscle cultures. Int JBiochem Cell Biol 42: 132–136, 2010.

19. Leiter JRS, Upadhaya R, Anderson JE. Nitric oxide and voluntaryexercise together promote quadriceps hypertrophy and increase vasculardensity in female 18-mo-old mice. Am J Physiol Cell Physiol 302:C1306–C1315, 2012.

20. Lewis JS. Rotator cuff tendinopathy. Br J Sports Med 43: 236–241, 2009.21. Li AM, Ma H, Villarroel A. Acetylcholine receptor gamma-subunits

mRNA isoforms expressed in denervated rat muscle. Mol Neurobiol 37:164–170, 2008.

22. Liu X, Laron D, Natsuhara K, Manzano G, Kim HT, Feeley BT. Amouse model of massive rotator cuff tears. J Bone Joint Surg Am 94: e41,2012 (doi:10.2106/JBJS.K.00620).

23. Longo UG, Berton A, Khan WS, Maffulli N, Denaro V. Histopathologyof rotator cuff tears. Sports Med Arthroscop Rev 19: 227–236, 2011.

24. Lundgreen K, Lian OB, Engebretsen L, Scott A. Lower muscle regen-erative potential in full-thickness supraspinatus tears compared to partial-thickness tears. Acta Orthopaed 84: 565–570, 2013.

25. Mallon WJ, Wilson RJ, Basamania CJ. The association of suprascap-ular neuropathy with massive rotator cuff tears: a preliminary report. JShoulder Elbow Surg 15: 395–398, 2006.

26. Marques MJ, Luz MAM, Minatel E, Santo Neto H. Muscle regenera-tion in dystrophic mdx mice is enhanced by isosorbide dinitrate. NeurosciLett 382: 342–345, 2005.

27. Mauro A. Satellite cell of skeletal muscle fibers. J Biophys Biochem Cytol9: 493, 1961.

28. Mizunoya W, Upadhaya R, Burczynski FJ, Wang GQ, Anderson JE.Nitric oxide donors improve prednisone effects on muscular dystrophy inthe mdx mouse diaphragm. Am J Physiol Cell Physiol 300: C1065–C1077,2011.

29. Nho SJ, Yadav H, Shindle MK, MacGillivray JD. Rotator cuff degen-eration: etiology and pathogenesis. Am J Sports Med 36: 987–993, 2008.

30. Sakaguchi S, Shono JI, Suzuki T, Sawano S, Anderson JE, Do MK,Ohtsubo H, Mizunoya W, Sato Y, Nakamura M, Furuse M, YamadaK, Ikeuchi Y, Tatsumi R. Implication of anti-inflammatory macrophagesin regenerative moto-neuritogenesis: promotion of myoblast migration andneural chemorepellent semaphorin 3A expression in injured muscle. Int JBiochem Cell Biol 54: 272–285, 2014.

31. Sciorati C, Buono R, Azzoni E, Casati S, Ciuffreda P, D’Angelo G,Cattaneo D, Brunelli S, Clementi E. Co-administration of ibuprofen and

nitric oxide is an effective experimental therapy for muscular dystrophy,with immediate applicability to humans. Br J Pharmacol 160: 1550–1560,2010.

32. Simon AM, Hoppe P, Burden SJ. Spatial restriction of AChR gene-expression to subsynaptic nuclei. Development 114: 545-&, 1992.

33. Soltow QA, Lira VA, Betters JL, Long JHD, Sellman JE, Zeanah EH,Criswell DS. Nitric oxide regulates stretch-induced proliferation inC2C12 myoblasts. J Muscle Res Cell Motil 31: 215–225, 2010.

34. Suzuki T, Do MK, Sato Y, Ojima K, Hara M, Mizunoya W, Naka-mura M, Furuse M, Ikeuchi Y, Anderson JE, Tatsumi R. Comparativeanalysis of semaphorin 3A in soleus and EDL muscle satellite cells in vitrotoward understanding its role in modulating myogenin expression. Int JBiochem Cell Biol 45: 476–482, 2013.

35. Tatsumi R, Anderson JE, Nevoret CJ, Halevy O, Allen RE. HGF/SF ispresent in normal adult skeletal muscle and is capable of activatingsatellite cells. Dev Biol 194: 114–128, 1998.

36. Tatsumi R, Sankoda Y, Anderson JE, Sato Y, Mizunoya W, ShimizuN, Suzuki T, Yamada M, Rhoads RP Jr, Ikeuchi Y, Allen RE. Possibleimplication of satellite cells in regenerative motoneuritogenesis: HGFupregulates neural chemorepellent Sema3A during myogenic differentia-tion. Am J Physiol Cell Physiol 297: C238–C252, 2009.

37. Thomazeau H, Rolland Y, Lucas C, Duval JM, Langlais F. Atrophy ofthe supraspinatus belly—assessment by MRI in 55 patients with rotatorcuff pathology. Acta Orthopaed Scand 67: 264–268, 1996.

38. Tomioka T, Minagawa H, Kijima H, Yamamoto N, Abe H, MaesaniM, Kikuchi K, Abe H, Shimada Y, Itoi E. Sarcomere length of tornrotator cuff muscle. J Shoulder Elbow Surg 18: 955–959, 2009.

39. Upadhaya R, Mizunoya W, Anderson JE. Detecting multiple proteinsby Western blotting using same-species primary antibodies, precomplexedserum, and hydrogen peroxide. Anal Biochem 419: 342–344, 2011.

40. Wozniak AC, Anderson JE. Nitric oxide-dependence of satellite stemcell activation and quiescence on normal skeletal muscle fibers. Dev Dyn236: 240–250, 2007.

41. Wozniak AC, Pilipowicz O, Yablonka-Reuveni Z, Greenway S, Cra-ven S, Scott E, Anderson JE. C-met expression and mechanical activa-tion of satellite cells on cultured muscle fibers. J Histochem Cytochem 51:1437–1445, 2003.

42. Yamaguchi K, Ditsios K, Middleton WD, Hildebolt CF, Galatz LM,Teefey SA. The demographic and morphological features of rotator cuffdisease—a comparison of asymptomatic and symptomatic shoulders. JBone Joint Surg Am 88A: 1699–1704, 2006.

43. Zanetti M, Gerber C, Hodler J. Quantitative assessment of the musclesof the rotator cuff with magnetic resonance imaging. Invest Radiol 33:163–170, 1998.

44. Zhang H, Anderson JE. Satellite cell activation and populations on singlemuscle-fiber cultures from adult zebrafish (Danio rerio). J Exp Biol 217:1910–1917, 2014.

45. Zhu ZZ, Qiu Y, Wang B, Yang Y, Qian BP, Zhu F. Abnormal spreadingand subunit expression of junctional acetylcholine receptors of paraspinalmuscles in scoliosis associated with syringomyelia. Spine 32: 2449–2454,2007.

C391SATELLITE CELL ACTIVATION AND AChRs IN ROTATOR-CUFF MUSCLE

AJP-Cell Physiol • doi:10.1152/ajpcell.00143.2015 • www.ajpcell.org

by 10.220.32.246 on October 4, 2017

http://ajpcell.physiology.org/D

ownloaded from