altered integrin mechanotransduction in human nucleus pulposus

7/27/2019 Altered Integrin Mechanotransduction in Human Nucleus Pulposus

1/10

ARTHRITIS & RHEUMATISM

Vol. 60, No. 2, February 2009, pp 460469

DOI 10.1002/art.24248

2009, American College of Rheumatology

Altered Integrin Mechanotransduction in Human Nucleus

Pulposus Cells Derived From Degenerated Discs

Christine Lyn Le Maitre,1 Jennie Frain,2 Jane Millward-Sadler,2 Andrew P. Fotheringham,2

Anthony John Freemont,2 and Judith Alison Hoyland2

Objective. Several studies have demonstrated bio-

logic responses of intervertebral disc (IVD) cells to

loading, although the mechanotransduction pathways

have not been elucidated. In articular chondrocytes,

which have a phenotype similar to that of IVD cells, a

number of mechanoreceptors have been identified, with51 integrin acting as a predominant mechanorecep-

tor. The purpose of this study was to investigate the role

of integrin signaling in IVD cells during mechanical

stimulation and to determine whether RGD integrins

are involved.

Methods. Human nucleus pulposus (NP) cells

derived from nondegenerated and degenerated discs

were subjected to dynamic compressive loading in the

presence of an RGD inhibitory peptide. Expression of

the 51 heterodimer in IVD tissue was examined by

immunohistochemistry and possible alternative mech-

anoreceptors by real-time quantitative polymerase

chain reaction.Results. Aggrecan gene expression was decreased

following loading of NP cells from nondegenerated and

degenerated discs. This response was inhibited by treat-

ment with an RGD peptide in cells from nondegener-

ated, but not degenerated, IVDs. Immunohistochemis-

try demonstrated that expression of the 51

heterodimer was unaltered in degenerated IVD tissue as

compared with normal IVD tissue.

Conclusion. Our results indicate that the mech-

anotransduction pathways are altered in cells from

degenerated IVDs. Mechanosensing in NP cells from

nondegenerated discs occurs via RGD integrins, possi-

bly via the 51 integrin, while cells from degenerated

discs show a different signaling pathway that does not

appear to involve RGD integrins.

The intervertebral disc (IVD) is an importantload-bearing structure, with daily activities exposing theIVD to dynamic oscillatory loads (1). These forces arisefrom a combination of gravity and muscle tension duringmovement and result in various types of pressure beingexperienced within the IVD. Abdominal and back mus-cles stabilize the spine and resist gravitational forces toallow spinal movement, but by doing so, they exertadditional compressive forces on the lumbar discs (2).These combined forces result in a wide variety ofmechanical stimuli, including compression, hydrostaticpressure, shear stress, torsion and flexion of the spine,electrokinetic effects, and volumetric changes (38).

The biologic response to these stimuli variesaccording to the type of cells in the IVD, the nature ofloading experienced, and the duration of the load.Several studies have demonstrated biologic responses toa number of different types of load in IVD cells, withmany focusing on the effects of compressive and hydro-static loading (3,4,914). However, the mechanotrans-duction pathways have not been elucidated. Many celltypes detect movement or forces via a transmembranereceptor, which triggers subsequent remodeling of the

cytoskeleton, leading to changes in cell metabolism vianumerous downstream events, such as posttranslationalmodifications (15). In articular chondrocytes, whichshare phenotype similar to that of IVD cells, a numberof transmembrane receptors have been identified thatcan fulfill this role, including CD44, anchorin II, andintegrin receptors (16,17).

Integrins are a family of transmembrane glycosy-

1Christine Lyn Le Maitre, PhD: Sheffield Hallam University,Sheffield, UK; 2Jennie Frain, PhD, Jane Millward-Sadler, PhD, An-drew P. Fotheringham, MSc, Anthony John Freemont, MD, FRCP,FRC Path, Judith Alison Hoyland, PhD: University of Manchester,Manchester, UK.

Address correspondence and reprint requests to Judith Ali-son Hoyland, PhD, Tissue Injury and Repair Group, School of Clinicaland Laboratory Sciences, Faculty of Medical and Human Sciences,The University of Manchester, Stopford Building, Oxford Road, Man-chester M13 9PT, UK. E-mail: [email protected].

Submitted for publication April 10, 2008; accepted in revisedform October 13, 2008.

460


2/10

lated proteins that mediate dynamic interactions be-tween the cell surface and the immediate surroundingenvironment (18). Structurally, they are a large family ofobligate heterodimers comprising 2 noncovalently asso-ciated subunits, referred to as the and subunits,

which combine to form 24 distinct integrin receptors.The integrins can be grouped into subfamilies based onligand interactions. One such family is the RGD inte-grins, where RGD (arginine-glycine-aspartic acid) is theligand motif that these integrins recognize and interact

with in extracellular matrix (ECM) molecules such asfibronectin. The RGD integrins are those containing the5, 8, v, and IIb subunits (18) and include the classicfibronectin receptor 51.

Integrins (particularly, the 51 heterodimer)have been identified in articular chondrocytes (19) andare responsible for the initial mechanotransduction re-sponse within these cells (17,2022). Stimulation ofhuman articular chondrocytes in the presence of aninhibitory RGD peptide has been shown to block theusual molecular response observed following loading, inthat there was an increase in cell proliferation andproteoglycan synthesis (23), membrane hyperpolariza-tion (24), and increased aggrecan and decreased matrixmetalloproteinase 3 (MMP-3) gene expression (25).Interestingly, the 51 integrinassociated pathway hasbeen shown to activate an altered intracellular signalingpathway in osteoarthritic (OA) chondrocytes followingmechanical stimulation (24), which raises the possibilityof altered mechanotransduction mechanisms between

healthy and diseased cells. Consistent with this is ourrecent study showing that the application of hydrostaticload to IVD cells resulted in an anabolic response incells derived from nondegenerated discs, with no re-sponse seen in cells derived from degenerated discs (14),

which suggests that mechanotransduction pathways mayalso be altered in degenerated IVDs.

Few studies have investigated integrin expressionin the IVD. Yu et al (26) demonstrated that expressionof the 5 integrin subunit decreased following applica-tion of cyclic hydrostatic pressure to porcine IVD cells,suggesting this may be part of a mechanotransductionpathway. However, no studies have investigated the

expression of the 51 heterodimer or its functionalityin human IVDs, and few studies have investigated theexpression of the individual integrin subunits in humanIVDs. Nettles et al (27) demonstrated the production of5 and 1 individual subunits in degenerated humananulus fibrosus (AF) tissue (albeit only 3 samples) and inporcine nucleus pulposus (NP) and AF. More recently,Gilchrist et al (28) investigated porcine IVD cell attach-

ment on various matrices in the presence of integrinsubunitblocking antibodies. They showed inhibition offibronectin binding by blockade of the 1 subunit andsuggested that this was due to blockade of the 51heterodimer (the classic fibronectin receptor). However,

no studies have investigated the role of integrin signalingin IVD cells during mechanical stimulation, and to date,it is unclear whether RGD integrins are involved in the

IVD response to mechanical stimulation.In order to assess this, we subjected human NP

cells derived from nondegenerated and degenerateddiscs to dynamic compressive loading in the presence ofan RGD inhibitory peptide and then investigated theexpression of51 heterodimer and integrin subunits byimmunohistochemistry, together with the possible ex-pression of alternative mechanoreceptors by polymerasechain reaction (PCR).

PATIENTS AND METHODS

Tissue selection and grading of IVDs. Human IVDtissue was obtained either at the time of surgery or atpostmortem examination. Informed consent of the patient orthe relatives of the deceased was obtained. The study wasapproved by the local research ethics committee.

Postmortem IVD tissue. Nine IVDs were recoveredfrom 6 subjects within 18 hours of death (Table 1). Theseconsisted of full-thickness wedges of IVD of 120 of arc, whichwere removed anteriorly, allowing well-orientated blocks oftissue for histologic assessment.

Degenerated IVD tissue. Nine patients (n 15 samples)

were selected based on the presence of magnetic resonanceimagingdiagnosed degeneration and progression to anteriorresection, either for spinal fusion or for IVD replacementsurgery because of chronic low back pain. Some patientsunderwent fusion at more than one level because of instability.Patients with a history of sciatica sufficient to warrant theirseeking medical opinion or patients experiencing classic sciat-ica were excluded.

Tissue preparation. A block of tissue, incorporatingthe AF and the NP in continuity, was fixed in 10% neutralbuffered formalin, decalcified in EDTA, and processed toparaffin wax. Hematoxylin and eosinstained sections wereanalyzed for the degree of morphologic degeneration (29). Ascore of 03 indicates a histologically normal (nondegener-ated) IVD, a score of 47 indicates evidence of moderate

degeneration, and a score of 812 indicates severe degenera-tion. For purposes of the loading and gene expression studies,samples were classified as nondegenerated or degenerated,but these 3 categories were used for immunohistochemicalanalysis.

Creation of IVD constructs. Normal human cadavericIVD tissue (n 3; predetermined using power calculations)was obtained from the spines of cadavers: one from an18-year-old man with L3/4 disc grade 1, another from a

INTEGRIN MECHANOTRANSDUCTION IN IVD CELLS 461


3/10

47-year-old man with L4/5 disc grade 2, and the third froma 79-year-old man with L5/S1 disc grade 2. Degenerated

human IVD tissue (n

3) was obtained from patients under-going spinal surgery: one from a 75-year-old man with L5/S1disc grade 7, another from a 35-year-old woman with L5/S1disc grade 10, and the third from a 43-year-old man with L4/5disc grade 10.

NP tissue was finely minced and digested with 300350PUK/ml of Pronase (Calbiochem, Nottingham, UK) in Dul-beccos modified Eagles medium (DMEM)Hams F-12 for 1hour at 37C and then washed in DMEMHams F-12. NP cellswere isolated in 0.25% type II collagenase (Invitrogen, Paisley,UK) and 0.01% hyaluronidase (Sigma, Poole, UK) for 4 hoursat 37C. Cells were then expanded in monolayer and used atearly passage (2). Cells were trypsinized and resuspended in1.2% medium-viscosity sodium alginate (Sigma) in 0.15MNaCl at a density of 1 106 cells/ml. Sixty microliters ofcell/alginate suspension was added to the inner (5 mm diam-eter) foam ring of BioFlex 6-well compression plates (FlexcellInternational, from Dunn Labortechnik, Asbach, Germany).Constructs were then formed by polymerization with 200 mMCaCl2 to yield 3.1-mm high samples. CaCl2 was removed,constructs were washed twice with 0.15MNaCl and medium, 2ml of complete medium was added, and the constructs wereincubated for 2 weeks at 37C in a humidified atmospherecontaining 5% CO2 to allow cell redifferentiation (30) prior toloading.

Inhibitory peptide treatments. One hour prior toloading, constructs were left untreated or were treated witheither 50 g/ml of GRGDSP (Calbiochem), a peptide thatcompetes for integrin ligand RGD binding sites or 50 g/ml ofGRADSP (Calbiochem), a control peptide for GRGDSP. Alltreatments were performed in triplicate.

Application of compressive load. BioFlex compressionplates containing the alginate/NP cell constructs were loadedonto a Flexercell FX-4000C compression loading system (Flex-cell International), 300 l of medium was applied per sample,which ensured no leakage or drying out of samples duringloading, and compressive load was applied for 2 hours at0.350.95 MPa, where the applied stress was oscillated sinu-soidally between these 2 limits at a frequency of 1 Hz in thepresence or absence of peptide inhibitors. The FlexercellFX-4000C functions via the application of compressed gas tothe base of the BioFlex compression plates (which have aflexible membrane), resulting in compression of the sample(contained within the foam circle) against the fixed platen atcontrolled rates of pressure (31). Control plates were placed inthe incubator at 37C without application of load. Samples

were harvested 1 hour after loading to assess cell viability andaggrecan gene expression.

Cell viability. Cell viability was assessed using carboxy-fluorescein diacetate succinimidyl ester (CFSE-DA) and pro-pidium iodide (PI) staining, where CFSE-DA stains viable cellsgreen, and PI stains nonviable cells red. Alginate constructswere incubated at 37C for 10 minutes in 1 ml of phosphatebuffered saline containing 10 l of 6 mMPI and 2 l of 5 mMCFSE-DA (both from Sigma). Cells were visualized with aninverted microscope equipped with a dual-wavelength bypassfilter (450 nm for CFSE-DA/520 nm for PI). Emission of greenfluorescence and red fluorescence indicated viable cells andnonviable cells, respectively. Total cell numbers were countedmanually and tallied as viable and nonviable cells to allow forcalculation of the percentage of viability.

RNA extraction, complementary DNA (cDNA) synthe-sis, and real-time quantitative PCR for aggrecan gene expres-sion. Prior to TRIzol extraction, alginate constructs werewashed in 0.15M NaCl and dissolved in dissolving buffer (55mMsodium citrate, 30 mMEDTA, 0.15MNaCl, pH 6) at 37Cfor 30 minutes. RNA was then extracted from triplicateconstructs using TRIzol reagent (Gibco, Paisley, UK) accord-ing to the manufacturers instructions. Reverse transcriptionwas performed as described previously (30), and real-timequantitative PCR was used to investigate the effects of dy-namic compressive loading on aggrecan gene expression. Real-time quantitative PCR was performed as described previously(30), using 5-TCGAGGACAGCGAGGCC-3 as the forwardprimer, 5-ATGGAACACGATGCCTTTCACCACGA-3 (with FAM at the 5 end and minor groove binder at the 3 end)as the probe,and 5-TCGAGGGTGTAGCGTGTAGAGA-3 asthe reverse primer (Applied Biosystems, Warrington, UK).Data were analyzed using the 2Ct method to calculate thedifference in cycle thresholds and using the housekeeping generibosomal subunit 18S (Pre-Developed Assay Reagents; Ap-plied Biosystems) and unloaded controls for normalization (30).

Localization of5 and 1 integrin subunits and 51heterodimer in human IVDs. Tissue samples from 24 IVDswere selected for immunohistochemical analysis. These con-sisted of 6 nondegenerated (grades 03), 8 moderately degen-

Table 1. Characteristics of the paraffin-embedded intervertebral

disc samples used for integrin immunohistochemistry

Sample

Sample

source

Age/sex

of subject

Histology

grade*

1 Postmortem 37/M 1

2 Postmortem 47/M 13 Postmortem 47/M 14 Postmortem 61/M 25 Postmortem 47/M 26 Postmortem 37/M 37 Postmortem 37/M 48 Postmortem 37/M 49 Surgical / 510 Postmortem 61/M 511 Surgical 75/M 612 Surgical / 613 Surgical 75/F 714 Surgical 46/M 715 Surgical 58/M 816 Surgical /F 817 Surgical 58/M 918 Surgical / 9

19 Surgical 58/M 1020 Surgical 22/M 1021 Surgical 73/F 1022 Surgical 46/M 1123 Surgical /F 1224 Surgical /M 12Rib control Postmortem 47/M NA

* Histologic scoring was performed as previously described (29), basedon the degree of degeneration: grades 03 normal (nondegener-ated), grades 47 moderate degeneration, and grades 812 severedegeneration. NA not applicable.

462 LE MAITRE ET AL


4/10

erated (grades 47), and 10 severely degenerated (grades

812) discs (Table 1). Immunohistochemistry was also used tolocalize the 51 heterodimer in the NP cells cultured in thealginate constructs. The immunohistochemistry protocol wefollowed has previously been described (30). Antibodies usedwere mouse monoclonal primary antibodies against humanintegrin 5 subunit (MAB1956; 1:100 dilution), human inte-grin 1 subunit (MAB1965; 1:400 dilution), and human 51heterodimer (MAB1969; 1:600 dilution) (all from Chemicon,Southampton, UK). Negative controls in which mouse IgG(Dako, Cambridge, UK) replaced the primary antibody (at anequal protein concentration) were included.

Image analysis. All slides were examined under a LeicaRMDB microscope (Leica, Cambridge, UK), and images werecaptured using a digital camera and a Bioquant Nova imageanalysis system (Bioquant Image Analysis, Nashville, TN).

Each section was divided into the NP, the inner AF, and theouter AF, and each was analyzed separately. Within each area,200 cells were counted, and the number of immunopositivecells was expressed as a proportion of this. Alginate/NPconstructs were analyzed for cells displaying 51 immunopos-itivity.

Gene expression of mechanoreceptors. Tissues from 19lumbar IVD samples were used for gene expression analysis.These consisted of 10 nondegenerated (grades 03; age range3079 years) and 9 degenerated (grades 4 12; age range 2974years) discs. RNA was extracted using TRIzol reagent, andcDNA was synthesized using BioScript RNase H reversetranscriptase (Bioline, London, UK) and random hexamers(Roche, Lewes, UK). Real-time PCR was performed to ana-lyze expression of the integrin subunits 1, 2, 5, v, and 1together with CD44 and the housekeeping gene GAPDH.Primers were designed using Amplify 1.2 software (available athttp://engels.genetics.wisc.edu/amplify), and gene specificitywas confirmed by BLAST searches of GenBank databasesequences (Table 2). Twenty-five microliters of the PCRreaction product consisted of 5 l of cDNA (50 ng/l), 0.125 lof HotStar Taq DNA polymerase (Qiagen, West Sussex, UK),0.25 l of forward primer (100 M), 0.25 l of reverse primer(100 M), 0.5 l of 40 mM dNTPs (Roche), and 2.5 l ofbuffer, which was brought to 25 l with sterile distilled H2O.

PCR was then performed on a PTC-200 Peltier ther-

mal cycler (Labsystems, Epsom, UK) with an initial Taqactivation step at 95C for 10 minutes, followed by 3542 cyclesconsisting of denaturation at 94C for 30 seconds, annealing at55.157.4C for 30 seconds, and extension at 72C for 1 minute(Table 2). A final incubation at 72C for 10 minutes was thenperformed. PCR products were visualized on a 1% TrisborateEDTA weight/volume agarose gel using an ultraviolettransilluminator and the UV Grab program GeneSnap fromSyngene (SLS, Manchester, UK).

Statistical analysis. Loading study data. Data weredetermined to be nonparametric using the Shapiro-Wilke test.The Kruskal-Wallis test with Mann-Whitney U post hoc testswere used to assess the effect of dynamic compressive load oncell viability and aggrecan gene expression on nondegeneratedand degenerated samples in the 3 treatment groups.

Immunohistochemistry data. Data were nonparametric,and hence, Kruskal-Wallis tests with Mann-Whitney U posthoc tests were performed to compare the numbers of immu-nopositive cells in degenerated groups versus nondegeneratedIVDs (scores 03) for each area of the IVD. In addition,Wilcoxon paired sample tests were used to compare theproportions of immunopositive cells in the different regions ofthe IVDs.

RESULTS

Application of mechanical load to human disc

cells and inhibition of the RGD integrins. All alginateconstructs maintained cell viability at 80%, with no

difference observed between samples subjected to dy-namic compressive loading and unloaded controls andno difference with any treatment in cells derived fromnondegenerated discs. A small decrease in cell viability

was observed in loaded disc cells derived from degener-ated IVDs compared with unloaded disc cells (P 0.05),

with no difference seen between treatments (Figure 1).Application of compressive loading at 0.350.95

Table 2. Primer sequences, conditions, and product sizes for standard real-time polymerase chain reaction (PCR)

Gene

Primer sequences

PCR conditionsProduct

sizeAccession

numberForward (533) Reverse (335)

GAPDH CCCATCACCATCTTCCAGG GGCCATCCACAGTCTTCTG Annealing at 57.4C;

35 cycles

354 bp M33197

1 integrin GGCTGTTGGTGAATTCAGTG GCTGCAGGTAGACGTAGAC Annealing at 57.4C;36 cycles

300 bp NM_181501

2 integrin GTGCCTGCAGAAGAATATGG GGTAGCCTACATCGCAGGC Annealing at 57.4C;42 cycles

475 bp NM_002203

5 integrin GGCTGTTGGTGAATTCAGTG GTCTACGTCTACCTGCAGC Annealing at 57.4C;36 cycles

286 bp NM_002205

v integrin GAGCAGCAAGGACTTTGGG GGGTACACTTCAAGACCAGC Annealing at 57.4C;36 cycles

619 bp NM_002210

1 integrin CCTTATGGACCTGTCTTACTC GCTGAAATTCTTCAGTAACTGC Annealing at 57.4C;38 cycles

589 bp NM_002211

CD44 TCCCAGTATGACACACATATTGC CACCTTCTTCGACTGTTGAC Annealing at 55.1C;38 cycles

549 bp NM_000610



5/10

MPa and a frequency of 1 Hz to NP cells derived from

nondegenerated or degenerated discs significantly de-creased aggrecan gene expression (P 0.05) (Figure 2).No difference in the response to loading was observed inNP cells that had been pretreated with the controlpeptide. However, in NP cells from nondegenerateddiscs pretreated with the RGD peptide, no change inaggrecan gene expression was observed following appli-cation of dynamic compressive load. In contrast, theresponse to dynamic compressive loading in NP cellsfrom degenerated discs pretreated with the RGD pep-tide was not altered from that seen in untreated cells or

cells treated with control peptide; that is, a significantdecrease in aggrecan gene expression in NP cells fromdegenerated discs was observed in all 3 treatment groups(P 0.05) (Figure 2).

Localization of5 and 1 integrin subunits and

the 51 heterodimer in human IVDs. Immunoreactiv-ity for 5 and 1 subunits and 51 heterodimer wasobserved in nondegenerated and degenerated IVDs.

Positive control tissue (human rib cartilage) stained foreach integrin and the heterodimer. IgG controls werenegative (Figure 3). Staining was particularly prominentin the chondrocyte-like cells of the NP and the inner AF(Figure 3), with significantly lower numbers of immu-nopositive cells in the outer AF (P 0.05). No signifi-cant difference was observed in the number of immu-nopositive cells for the 5 subunit, the 1 subunit, or the51 heterodimer identified in the NP or inner AF of

nondegenerated discs as compared with IVDs withmoderate or severe histologic degeneration (P 0.05for all comparisons). Immunopositive staining for 51heterodimer was observed in cells cultured in alginate(Figure 3).

Gene expression of mechanoreceptors. Althoughgene expression was detected for CD44 and the integrinsubunits in both nondegenerated and degenerated discs,a differential expression pattern was observed. Integrinsubunits 1 and 5 and CD44 were strongly expressed inall samples, while v was moderately expressed in the

Figure 1. Percentage of viable cells following 2 hours of compressive

loading (0.350.95 MPa at a frequency of 1 Hz) in untreated, controlpeptidetreated, and inhibitory peptidetreated cells from interverte-

bral discs (IVDs) obtained at surgery from patients undergoing spinal

fusion or IVD replacement (degenerated) and at postmortem exami-nation from normal subjects (nondegenerated). Values are the mean

SEM. P 0.05 versus unloaded cells from degenerated discs.

Figure 2. Aggrecan gene expression in human nucleus pulposus cells derived from interverte-

bral discs (IVDs) that were left untreated or were treated with control peptide or integrininhibitory peptide and then subjected to compressive loading for 2 hours at 0.350.95 MPa at a

frequency of 1 Hz. RNA was extracted 1 hour after loading. IVDs were obtained at surgery from

patients undergoing spinal fusion or IVD replacement (degenerated) and at postmortemexamination from normal subjects (nondegenerated). Values are the mean SEM. P 0.05

versus unloaded cells.

464 LE MAITRE ET AL


6/10

majority of samples. Integrin subunits 2 and 1 wereonly weakly expressed in nondegenerated and degener-ated NP samples, with little expression detected in AFsamples (Figure 4).

DISCUSSION

The intervertebral disc is an important load-

bearing structure in the human body that experiences a

Figure 4. Expression of GAPDH, 1 integrin, 2 integrin, 5 integrin, v integrin, 1 integrin, and CD44 genes in nucleus pulposus (NP)

and anulus fibrosus (AF) cells derived from intervertebral discs (IVDs) obtained at surgery from patients undergoing spinal fusion or IVDreplacement (degenerated; n 9) and at postmortem examination from normal subjects (nondegenerated; n 10), as determined by

real-time polymerase chain reaction analysis. Red X indicates unavailable sample.

Figure 3. Photomicrographs demonstrating immunohistochemical localization of integrin subunits 5

and 1 and the heterodimer 51 in the nucleus pulposus from intervertebral discs (IVDs) obtained at

surgery from patients undergoing spinal fusion or IVD replacement (degenerated) and at postmortem

examination from normal subjects (nondegenerated). Cells extracted from nondegenerated and degener-ated IVDs cultured in alginate for 2 weeks showed immunopositivity for 51. IgG negative controls

showed no staining. Bars 380 m.



7/10

number of mechanical forces during daily activities.Cells within the IVD respond to these stresses with amultitude of anabolic and catabolic responses. As such,understanding the mechanisms of mechanosensing

within the human IVD is of paramount importance to

understanding the regulation of IVD matrix homeosta-sis. In this study, we investigated the role of integrinsignaling in mechanotransduction pathways in the IVD.

Application of dynamic compressive loading resulted ina significant decrease in aggrecan gene expression inboth nondegenerated and degenerated discs. Interest-ingly, this response was inhibited by the application of anRGD peptide in nondegenerated, but not degenerated,discs.

Having determined that RGD-binding integrinsplay a role in the mechanotransduction response innondegenerated IVDs, we then investigated the expres-sion and localization of the potential mechanoreceptor51 integrin and its individual subunits (5 and 1)

within both nondegenerated and degenerated humanIVD samples. Our results showed that human NP and

AF cells express both the 5 and 1 integrin subunitsand the 51 heterodimer and that this expression wasnot altered with degree of degeneration. Additionally,

we showed that both degenerated and nondegeneratedIVDs express the 1 integrin subunit and CD44, asdetermined by PCR. Neither 1 integrins nor CD44 actsthrough an RGD motif: 1 integrins interact with colla-gen, whereas CD44 is a hyaluronan receptor, but eithercould act as potential alternative mechanoreceptors in

degenerated IVDs.In this study, cells were subjected to dynamiccompressive loading for 2 hours at 0.350.95 MPa at afrequency of 1 Hz, which Wilke et al (7) demonstrated tobe the physiologic loading pattern experienced during

jogging. In vivo, the IVD is routinely subjected tocompressive loads that can exceed 2 MPa and areaffected by posture and muscle activity (32). The direc-tional compressive forces that are the result of spinalmovement will also lead to fluid flow and changes inhydrostatic pressures within the IVD, all of which willhave an influence on IVD cellular metabolism. Withincreasing degeneration, there is a decrease in hydration

of the IVD, which is likely to result in a change in thetype and magnitude of forces that the individual cellsexperience.

We previously investigated the effect of hydro-static loading on NP cells isolated from human IVDs(14). Hydrostatic loading is a major force experiencedin the IVD, particularly in the nondegenerated disc,

which behaves mechanically like a fluid. In the present

study, however, we focused on the application of com-pressive load to cells derived from nondegenerated anddegenerated NP IVDs. Compressive loading is a forcethat is regularly experienced by the IVD. In addition,there is a growing number of studies investigating the

effect of compression on the different regions of theIVD and showing that the duration and amount ofloading have an effect on proteoglycan/aggrecan expres-sion (10,11,3335).

The present study showed that the application ofdynamic compressive loading to NP cells cultured inalginate resulted in decreased aggrecan gene expressionin cells from both nondegenerated and degenerateddiscs. Previous studies investigating the application ofcompressive loading to animal IVD cells and tissueshave found both increases and decreases in aggrecangene expression, depending on the method of applica-tion, the magnitude, frequency, duration, and the periodof rest following loading p rior to assessment(4,10,11,13,34,3638). The present study is the first toapply compressive loading to human NP cells in a3-dimensional matrix.

In contrast to our findings, compressive loadingof animal IVD tissue (both in vivo and in vitro) usingpressures similar to those used in our study has beenshown to result in increased aggrecan gene expression(10,11,34,3638). However, a number of studies to datehave shown differential responses, depending on thefrequency (1,10,39,40) and duration (41) of the appliedload as well as on the time of sample harvest following

loading (37), and this may explain the different re-sponses to a similar magnitude of loading that have beenobserved between these studies. Indeed, Maclean et al(10) showed with rat tail compression studies that com-pressive loading at 1 MPa at a frequency of 1 Hz resultedin a greater up-regulation of catabolic gene expressionthan matrix gene expression, while a lower frequencyresulted in a greater anabolic gene response.

Additionally, the majority of studies investigatingcompressive loading have been in vivo studies, where theforces actually perceived by the cells, particularly in thehealthy NP, may be different, rather than in vitro studiesusing isolated cells cultured in alginate. It is therefore

possible that the deformation experienced by the cells inour study was higher than that observed in cells pro-tected by the extensive ECM in vivo, thus giving rise tothe catabolic response. Furthermore, since these cells

were embedded in a 3-dimensional matrix, they wouldnot have experienced any hydrostatic pressure, whereasthey would have experienced hydrostatic pressure in

vivo. It is also of note that our study is the first to

466 LE MAITRE ET AL


8/10

investigate the effects of compressive loading on humancells, which could also explain the observed differences.

Interestingly, the response of aggrecan gene ex-pression following compressive loading of NP cells inalginate was inhibited in NP cells from nondegenerated

discs by pretreatment with the inhibitory peptide(GRGDSP) but not the control peptide (GRADSP).Integrins containing 5, v, IIb, and 8 subunits areinhibited by RGD inhibitory peptides, which suggeststhat in NP cells derived from nondegenerated discs, theRGD integrins act as mechanoreceptors when cellsembedded in alginate are subjected to compressiveloading. RGD peptide inhibitors have previously beenused to investigate the role of integrins in mechanotrans-duction pathways in articular chondrocytes (24,25),

where the stimulation of aggrecan gene expression andthe inhibition of MMP-3 gene expression observedfollowing application of cyclic stretch at 0.1 MPa at afrequency of 0.33 Hz for 20 minutes was inhibited bypretreatment with RGD peptides (25). In articular chon-drocytes, the RGD integrin receptor proposed to be themain mechanoreceptor is the 51 heterodimer (17,2024). The expression of51 by NP cells in vivo and in

vitro following culture in alginate for 2 weeks wouldsuggest that this could possibly be the integrin involvedin the mechanotransduction response in cells from non-degenerated IVDs. A number of studies have investi-gated the effect of alginate culture of chondrocytes onmatrix deposition and have shown that within 2 weeks,cells have synthesized components of both pericellular

and territorial matrix (42,43). Thus, it is likely that the51 integrin is operating via interaction with the newlysynthesized pericellular and territorial matrix producedin this culture system.

It is noteworthy that in NP cells derived fromdegenerated discs, pretreatment with the RGD peptidefailed to inhibit their response to compressive loading,and a significant decrease in aggrecan gene expression

was still observed following application of load. Thissuggests that in degenerated discs, an alternative mech-anotransduction pathway may be involved that replacesthe RGD mechanoreceptor signaling seen in nondegen-erated discs. In cells derived from OA cartilage, the

intracellular signaling that is activated following me-chanical stimulation is altered as compared with normalpathways, but the cells still sense the mechanical signalthrough the 51 integrin (24). Our findings suggestthat in the degenerated IVD, this is not the case. Wehave shown that although degenerated disc cells expressthe 51 integrin, it does not appear to be involved inthe mechanotransduction pathway in these cells. The

51 integrin is also involved in the extracellular matrixinteraction with fibronectin; thus, the expression of the51 integrin in degenerated discs may play a greaterrole in ECM interactions than mechanotransduction.

Gilchrist et al (28) demonstrated that binding to

fibronectin in porcine IVDs was blocked by inhibition ofthe 1 integrin subunit, suggesting expression of thefibronectin receptor (51 integrin) in these disc cells.Fibronectin expression has been shown to be increasedduring disc degeneration (44) and, thus, may affect theavailability of the 51 integrin for mechanosensingand/or may affect integrin-mediated signaling, as hasbeen shown in OA chondrocytes with other integrinECM relationships (45). Our findings indicate that IVDcells in degenerated discs may be using a differentreceptor to sense and respond to mechanical signaling,since these disc cells retained their response to mechan-ical stimuli in the presence of the RGD peptide. Thismechanosensing could occur through a non-RGD inte-grin, such as 1, or a nonintegrin transmembrane mol-ecule, such as CD44, both of which we have shown to bestrongly expressed by human IVD cells.

In OA, the increased production of mediators ofinflammation, including interleukin-1, tumor necrosisfactor , and nitric oxide, have been postulated to affectintegrin-mediated signaling, either by modulating chon-drocyte integrin expression (46) or by regulating intra-cellular signaling (47). In the degenerated IVD, anincrease in these mediators of inflammation, particularlyinterleukin-1, is also seen (30,48), and mechanical load-

ing has been shown to initiate signaling pathways thatinvolve calcium (49) and nitric oxide (50). It is possiblethat these mediators could be involved in the alteredmechanotransduction seen in our study, as has previ-ously been reported for OA chondrocytes (24). Interest-ingly, we have recently shown that the application of0.81.7 MPa of hydrostatic pressure at a frequency of 0.5Hz for 2 hours resulted in a differential response in NPcells derived from nondegenerated versus degenerateddiscs, where degenerated discs failed to respond to thisloading regimen (14). Together with the data presentedherein, this suggests that different mechanotransductionpathways may be involved in sensing different types of

mechanical load and that these pathways are alteredduring disc degeneration.

In conclusion, we have demonstrated that whensubjected to compressive loading, cells isolated fromnondegenerated and degenerated human discs recog-nize mechanical stimulation through different mechano-receptors. The inhibition of the aggrecan gene responseto mechanical loading by RGD peptides in cells from



9/10

nondegenerated discs suggests that these cells recognizethe mechanical stimulus through RGD integrins in amanner similar to that of articular chondrocytes. Expres-sion of the 51 heterodimer would suggest that this isa likely candidate as a mechanoreceptor in these NP

cells. The 51 integrin is also expressed in degeneratedIVDs, with no change in expression observed during discdegeneration. However, the presence of RGD peptidesdid not inhibit the response of these cells to compressiveloading, suggesting that cells in the degenerated discs donot use the 51 integrin or other RGD integrins asmechanoreceptors during this loading stimulus. Thus,

we have demonstrated differences in the mechanotrans-duction pathways of nondegenerated and degeneratedIVDs. Further elucidation of the signaling events acti-

vated by mechanical stimuli in IVD cells derived fromnondegenerated and degenerated discs will lead to abetter understanding of how IVD matrix homeostasis ismaintained by mechanical stimuli in health and disease.

AUTHOR CONTRIBUTIONS

Professor Hoyland had full access to all of the data in thestudy and takes responsibility for the integrity of the data and theaccuracy of the data analysis.Study design. Le Maitre, Frain, Hoyland.

Acquisition of data. Le Maitre, Frain, Fotheringham.Analysis and interpretation of data. Le Maitre, Frain, Millward-Sadler, Fotheringham, Freemont, Hoyland.Manuscript preparation. Le Maitre, Millward-Sadler, Freemont, Hoy-land.Statistical analysis. Le Maitre, Hoyland.

REFERENCES

1. Kasra M, Goel V, Martin J, Wang ST, Choi W, Buckwalter J.Effect of dynamic hydrostatic pressure on rabbit intervertebral disccells. J Orthop Res 2003;21:597603.

2. Adams M, Bogduk N, Burton K, Dolan P. The biomechanics ofback pain. London: Churchill Livingstone; 2002.

3. Iatridis JC, Mente PL, Stokes IA, Aronsson DD, Alini M. Com-pression-induced changes in intervertebral disc properties in a rattail model. Spine 1999;24:9961002.

4. Lotz JC, Colliou OK, Chin JR, Duncan NA, Liebenberg E.Compression-induced degeneration of the intervertebral disc: anin vivo mouse model and finite-element study. Spine 1998;23:2493506.

5. Lotz JC, Chin JR. Intervertebral disc cell death is dependent onthe magnitude and duration of spinal loading. Spine 2000;25:147783.

6. Li H, Wang Z. Intervertebral disc biomechanical analysis using thefinite element modeling based on medical images. Comput MedImaging Graph 2006;30:36370.

7. Wilke HJ, Neef P, Caimi M, Hoogland T, Claes LE. New in vivomeasurements of pressures in the intervertebral disc in daily life.Spine 1999;24:75562.

8. Sato K, Kikuchi S, Yonezawa T. In vivo intradiscal pressuremeasurement in healthy individuals and in patients with ongoingback problems. Spine 1999;24:246874.

9. Hsieh AH, Lotz JC. Prolonged spinal loading induces matrix

metalloproteinase-2 activation in intervertebral discs. Spine 2003;28:17818.

10. Maclean JJ, Lee CR, Alini M, Iatridis JC. Anabolic and catabolicmRNA levels of the intervertebral disc vary with the magnitudeand frequency of in vivo dynamic compression. J Orthop Res2004;22:1193200.

11. Maclean JJ, Lee CR, Alini M, Iatridis JC. The effects of short-term load duration on anabolic and catabolic gene expression inthe rat tail intervertebral disc. J Orthop Res 2005;23:11207.

12. Iatridis JC, Maclean JJ, Roughley PJ, Alini M. Effects of mechan-ical loading on intervertebral disc metabolism in vivo. J Bone JointSurg Am 2006;88 Suppl 2:416.

13. Hutton WC, Toribatake Y, Elmer WA, Ganey TM, Tomita K,Whitesides TE. The effect of compressive force applied to theintervertebral disc in vivo: a study of proteoglycans and collagen.Spine 1998;23:252437.

14. Le Maitre CL, Frain J, Fotheringham AP, Freemont AJ, HoylandJA. Human cells derived from degenerate intervertebral discsrespond differently to those derived from non-degenerate inter-

vertebral discs following application of dynamic hydrostatic pres-sure. Biorheology 2008;45:56375.

15. Setton LA, Chen J. Mechanobiology of the intervertebral disc andrelevance to disc degeneration. J Bone Joint Surg Am 2006;88

Suppl 2:527.16. Loeser RF. Chondrocyte integrin expression and function. Bio-

rheology 2000;37:10916.17. Millward-Sadler SJ, Salter DM. Integrin-dependent signal cas-

cades in chondrocyte mechanotransduction. Ann Biomed Eng2004;32:43546.

18. Hynes RO. Integrins: bidirectional, allosteric signaling machines.Cell 2002;110:67387.

19. Loeser RF, Sadiev S, Tan L, Goldring MB. Integrin expression byprimary and immortalized human chondrocytes: evidence of adifferential role for 11 and 21 integrins in mediating chon-drocyte adhesion to types II and VI collagen. OsteoarthritisCartilage 2000;8:96105.

20. Millward-Sadler SJ, Wright MO, Lee H, Nishida K, Caldwell H,Nuki G, et al. Integrin-regulated secretion of interleukin 4: a novelpathway of mechanotransduction in human articular chondrocytes.J Cell Biol 1999;145:1839.

21. Millward-Sadler SJ, Khan NS, Bracher MG, Wright MO, SalterDM. Roles for the interleukin-4 receptor and associated JAK/STAT proteins in human articular chondrocyte mechanotransduc-tion. Osteoarthritis Cartilage 2006;14:9911001.

22. Chowdhury TT, Salter DM, Bader DL, Lee DA. Integrin-medi-ated mechanotransduction processes in TGF-stimulated mono-layer-expanded chondrocytes. Biochem Biophys Res Commun2004;318:87381.

23. Chowdhury TT, Appleby RN, Salter DM, Bader DA, Lee DA.Integrin-mediated mechanotransduction in IL-1 stimulatedchondrocytes. Biomech Model Mechanobiol 2006;5:192201.

24. Millward-Sadler SJ, Wright MO, Lee H, Caldwell H, Nuki G,Salter DM. Altered electrophysiological responses to mechanicalstimulation and abnormal signalling through 51 integrin inchondrocytes from osteoarthritic cartilage. Osteoarthritis Carti-lage 2000;8:2728.

25. Millward-Sadler SJ, Wright MO, Davies LW, Nuki G, Salter DM.

Mechanotransduction via integrins and interleukin-4 results inaltered aggrecan and matrix metalloproteinase 3 gene expressionin normal, but not osteoarthritic, human articular chondrocytes.

Arthritis Rheum 2000;43:20919.26. Yu SJ, Qiu GX, Burton Y, Sandra R, Cari W, Albert Y. Expres-

sion of integrin 5 and actin in the cells of intervertebral discunder cyclic hydrostatic pressure in vitro. Zhonghua Wai Ke ZaZhi 2005;43:16058. In Chinese.

27. Nettles DL, Richardson WJ, Setton LA. Integrin expression incells of the intervertebral disc. J Anat 2004;204:51520.

468 LE MAITRE ET AL


10/10

28. Gilchrist CL, Chen J, Richardson WJ, Loeser RF, Setton LA.Functional integrin subunits regulating cell-matrix interactions inthe intervertebral disc. J Orthop Res 2007;25:82940.

29. Sive JI, Baird P, Jeziorsk M, Watkins A, Hoyland JA, FreemontAJ. Expression of chondrocyte markers by cells of normal anddegenerate intervertebral discs. Mol Pathol 2002;55:917.

30. Le Maitre CL, Freemont AJ, Hoyland JA. The role of interleu-kin-1 in the pathogenesis of human intervertebral disc degenera-tion. Arthritis Res Ther 2005;7:R73245.

31. Flexcell International. Flexcell culture systems. URL: http://www.flexcellint.com/gallery.htm.

32. Schultz A, Andersson G, Ortengren R, Haderspeck K, NachemsonA. Loads on the lumbar spine: validation of a biomechanicalanalysis by measurements of intradiscal pressures and myoelectricsignals. J Bone Joint Surg Am 1982;64:71320.

33. Korecki CL, Maclean JJ, Iatridis JC. Dynamic compression effectson intervertebral disc mechanics and biology. Spine 2008;33:14039.

34. Wang DL, Jiang SD, Dai LY. Biologic response of the interverte-bral disc to static and dynamic compression in vitro. Spine2007;32:25218.

35. Walsh AJ, Lotz JC. Biological response of the intervertebral discto dynamic loading. J Biomech 2004;37:32937.

36. Wenger KH, Woods JA, Holecek A, Eckstein EC, Robertson JT,Hasty KA. Matrix remodeling expression in anulus cells subjectedto increased compressive load. Spine 2005;30:11226.

37. Maclean JJ, Roughley PJ, Monsey RD, Alini M, Iatridis JC. Invivo intervertebral disc remodeling: kinetics of mRNA expressionin response to a single loading event. J Orthop Res 2008;26:57988.

38. Maclean JJ, Lee CR, Grad S, Ito K, Alini M, Iatridis JC. Effects ofimmobilization and dynamic compression on intervertebral disccell gene expression in vivo. Spine 2003;28:97381.

39. Gokorsch S, Nehring D, Grottke C, Czermak P. Hydrodynamicstimulation and long term cultivation of nucleus pulposus cells: anew bioreactor system to induce extracellular matrix synthesis bynucleus pulposus cells dependent on intermittent hydrostaticpressure. Int J Artif Organs 2004;27:96270.

40. Kasra M, Merryman WD, Loveless KN, Goel VK, Martin JD,Buckwalter JA. Frequency response of pig intervertebral disc cells

subjected to dynamic hydrostatic pressure. J Orthop Res 2006;24:196773.

41. Huang J, Ballou LR, Hasty KA. Cyclic equibiaxial tensile straininduces both anabolic and catabolic responses in articular chon-drocytes. Gene 2007;404:1019.

42. Hauselmann HJ, Masuda K, Hunziker EB, Neidhart M, Mok SS,

Michel BA, et al. Adult human chondrocytes cultured in alginateform a matrix similar to native human articular cartilage. Am JPhysiol 1996;271(3 Pt 1):C74252.

43. Beekman B, Verzijl N, Bank RA, von der Mark K, TeKoppele JM.Synthesis of collagen by bovine chondrocytes cultured in alginate:posttranslational modifications and cellmatrix interaction. ExpCell Res 1997;237:13541.

44. Oegema TR, Johnson SL, Aguiar DJ, Ogilvie JW. Fibronectin andits fragments increase with degeneration in the human interverte-bral disc. Spine 2000;25:27427.

45. Probstmeier R, Pesheva P. Tenascin-C inhibits 1 integrin-depen-dent cell adhesion and neurite outgrowth on fibronectin by adisialoganglioside-mediated signaling mechanism. Glycobiology1999;9:10114.

46. Jobanputra P, Lin H, Jenkins K, Bavington C, Brennan FR, NukiG, et al. Modulation of human chondrocyte integrins by inflam-matory synovial fluid. Arthritis Rheum 1996;39:14302.

47. Clancy RM, Rediske J, Tang X, Nijher N, Frenkel S, Philips M, etal. Outside-in signaling in the chondrocyte: nitric oxide disruptsfibronectin-induced assembly of a subplasmalemmal actin/rho

A/focal adhesion kinase signaling complex. J Clin Invest 1997;100:178996.

48. Le Maitre CL, Hoyland JA, Freemont AJ. Catabolic cytokineexpression in degenerate and herniated human intervertebraldiscs: IL-1 and TNF expression profile. Arthritis Res Ther2007;9:R77.

49. Pritchard S, Erickson GR, Guilak F. Hyperosmotically inducedvolume change and calcium signaling in intervertebral disk cells:the role of the actin cytoskeleton. Biophys J 2002;83:250210.

50. Benallaoua M, Richette P, Francois M, Tsagris L, Revel M, CorvolM, et al. Modulation of proteoglycan production by cyclic tensilestretch in intervertebral disc cells through a post-translationalmechanism. Biorheology 2006;43:30310.


altered integrin mechanotransduction in human nucleus pulposus

Documents