Biomaterials 25 (2004) 5347–5352

ARTICLE IN PRESS

*Correspondin

106, Brigham an

Francis Street, B

+1-617-732-6705

E-mail addres1Current addr

University Medic

Pakistan.

0142-9612/$ - see

doi:10.1016/j.bio

Investigation of the activation of a human serum complementprotein, C3, by orthopedic prosthetic particulates

S. Noordina,1, S. Shortkroff a, C.B. Sledgea, M. Spectora,b,*aDepartment of Orthopaedic Surgery MRB 106, Brigham and Women’s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA

bVA Boston Healthcare System, Boston Campus, Boston, MA 02130, USA

Received 12 December 2002; accepted 23 November 2003

Abstract

Myriad molecular, cellular, and physiological processes underlie the inflammatory and osteolytic processes induced by particles of

biomaterials resulting from the wear of implants such as total joint replacement prostheses. The objective this study was to

investigate the role that the complement system may be playing in these phenomena. The aim was to evaluate the degree to which

particles of selected orthopaedic materials—high density and ultrahigh molecular weight polyethylene, polymethylmethacrylate, and

commercially pure titanium—cause the elevation of a key complement molecule, C3a, in an in vitro assay that directly measured the

concentration of C3a. The results demonstrated that HDPE particles, at high concentration, are capable of causing the elevation of

C3a in the in vitro assay. This finding is discussed in the context of other work and the mechanics of the complement system as it

may affect the osteolytic process.

r 2003 Elsevier Ltd. All rights reserved.

Keywords: Complement system; Biomaterials; Particles; Orthopaedics; Prostheses; Osteolysis

1. Introduction

Studies of osteolysis around joint replacement pros-theses have generally focused on the response ofphagocytes to particulate wear debris as the incitingcause [1]. Certain inflammatory mediators such asprostaglandin E2, (PGE2) [2], and interleukin-1 (IL-1)and IL-6 [3], released by macrophages—also referred toas histiocytes when they are tissue resident-phagocytes—as they phagocytose the debris, are known to be potentstimulators of bone resorption [4,5]. Similar processesappear to explain the inflammatory response to othertypes of implants (e.g., temporomandibular disc repla-cement devices [6–8]). There are, however, otherbiological processes that could also be involved in wearparticle-induced inflammation and osteolysis.

g author. Department of Orthopaedic Surgery MRB

d Women’s Hospital, Harvard Medical School, 75

oston, MA 02115, USA. Tel.: +1-617-732-6702; fax:

.

s: mspector@rics.bwh.harvard.edu (M. Spector).

ess: Department of Orthopaedic Surgery, Aga Khan

al College, Stadium Road, P.O. Box 3500, Karachi,

front matter r 2003 Elsevier Ltd. All rights reserved.

materials.2003.11.057

Exposure of blood to foreign surfaces such as thoseon wear particles, particularly the energetic surfaces ofparticles as they are being produced, can cleave certainmolecules engaged in contact-activated host defensesystems, including coagulation, fibrinolysis, and thecomplement system. The complement system [9] iscomposed of a group of more than 30 separate proteinspresent either in plasma, or as receptors on the surfaceof many cells, that have evolved to protect the host frominvasion by foreign materials. Complement activationcan occur by two distinct pathways, the classical andalternative pathways. The classical pathway is usuallytriggered by antigen–antibody complexes, whereasactivation of the alternative pathway occurs sponta-neously at a slow rate and is augmented by substancessuch as the polysaccharides of bacterial cell walls,endotoxins or artificial materials. Both pathways mergeat the formation of an enzymatic complex, a C3convertase, capable of cleaving the third complementcomponent (C3), with generation of C3a and C3b.Surface-bound C3b, in conjunction with other comple-ment factors, is responsible for cleaving C5 into C5a andC5b. The latter combines with C6, C7, C8, and C9 toform the multimolecular C5b-9 complex or terminalcomplement complex, which has cytolytic functions and

ARTICLE IN PRESSS. Noordin et al. / Biomaterials 25 (2004) 5347–53525348

can either react with the foreign surface or beinactivated in the fluid phase [10].

The smaller fragments C3a and C5a are anaphylatox-ins that have multiple biological effects including theinduction of the release of inflammatory mediators(including IL-1) from monocytes, basophils, and mastcells. C3b can facilitate aggregation of macrophages [11]and their adherence to other cells or particles bearingC3b on their surface and thus can play an important rolein phagocytosis. C3b may also stimulate ‘‘frustratedphagocytosis’’ by phagocytes [12], a process whichwould result in degradative enzymes and high-energyoxygen species being secreted around the implantsurface. Macrophage aggregation may lead to macro-phage fusion and the formation of giant cells [13]. Theaccumulation of macrophages and giant cells in theperiprosthetic tissues is one of the hallmarks of theparticle-induced prosthesis failure, and the inflamma-tory products of monocytes and macrophages activatedby complement molecules have been implicated inosteolysis following total joint arthroplasty (TJA).These striking similarities between the mechanismsimplicated in osteolysis and the known biologic activ-ities of C3a and C5a suggest that these inflammatorymediators may play a role in promoting osteolysis. Arecent in vitro study [14] has demonstrated thatpolyethylene particles can in fact result in the elevationof levels of C3b, further prompting the current study, theobjective of which was to evaluate the degree to whichparticles of selected orthopaedic materials—ultrahighmolecular weight polyethylene (UHMWPE), polymethyl-methacrylate (PMMA), and commercially pure titanium(CPT)—cause the elevation of C3a in an in vitro assaythat directly measured the concentration of C3a.

Physiologically significant complement activation thatcan propagate inflammatory reactions may not bedetectable by immunoelectrophoretic techniques. Thehemolytic assay for total complement (CH50) is neithersensitive nor specific for any individual componentproteins since deficiency in any of the components in theclassical or terminal pathway can result in subnormalhemolysis. Measurement of individual components suchas C3a (and to a certain extent other C3 activationproducts) is a more sensitive indicator of C3 activationthan the other techniques. Sensitivity of radioimmu-noassay for C3a is approximately 50 times greater thanthat of other techniques [15].

2. Materials and methods

2.1. Materials

Particulate specimens included: CPT particles with amean particle diameter of 0.8 mm (Johnson Matthey,MA); HDPE with an average particle size of 7 mm and

ultra high molecular weight polyethylene (UHMWPE)in two particle diameters of 18 mm (UHMWPE-18) and130 mm (UHMWPE-130), both having a specific gravityof 0.93 (Shamrock Technologies Inc., Newark, NJ);PMMA particles produced in our laboratory by millingSimplex bone cement (Stryker, Howmedica, Osteonics,Rutherford, NJ). The PMMA particles were passedthrough special filters to ensure uniform particle size andwere then analyzed to obtain their average diameter,which was 7 mm. Zymosan particles, which were ofcomparable diameter, were employed as positive con-trols (Sigma Chemical Co., St. Louis, MO).

2.2. Human serum isolation

The complement source was human serum collectedfrom one healthy volunteer. The fresh blood was left tocoagulate for 20min at room temperature in theCorvacs vacutainer tubes. The last 3ml of blood wasdiscarded to minimize artifacts resulting from thrombo-plastic substances released upon tissue puncture. Theclot was ringed with the wooden end of a sterile swab todetach it from the side of the glass tube. The blood wasthen centrifuged at 2000g for 15min at 4�C. The serumwas aliquoted into plastic centrifuge tubes and immedi-ately frozen at �70�C.

Serum from one individual was used to more clearlyreveal differences in complement activation due to thevarious particles. Since independent experimental runswere performed using different aliquots of the serum,the findings from the three runs using the same particlesrepresent three independent observations.

2.3. Preparation of particulate samples for complement

activation

The CPT particles were suspended in sterile phos-phate buffered saline (PBS) at the required concentra-tions for the complement activation experiments.Because of the hydrophobic nature of the HDPE,UHMWPE and PMMA particles, human serum wasdirectly added to predetermined quantities of theseparticles. Zymosan, a potent activator of the comple-ment system, was prepared as follows: 0.5 g wassuspended in 100ml of saline, boiled for 10min andthen allowed to cool before being centrifuged at 2500g

for 10min. The saline was decanted and the Zymosanpellet was washed by resuspending in 100ml saline andagain centrifuging at 2500g for 10min. The washingprocedure was repeated three times before the finalsuspension was aliquoted into samples of 5mgZymo-san/ml saline and stored at 4�C until used. Prior totheir use in the following experiments, the titanium andZymosan particles were autoclaved, whereas the poly-ethylene and PMMA particles were gas sterilized usingethylene oxide.

ARTICLE IN PRESS

Fig. 1. Bar chart showing the C3a desArg level (mg/ml) for various

particulate materials, as a function of time (n ¼ 3; mean7standard

error of the mean, SEM): commercially pure titanium (CPT), 2.5� 108

particles/ml; high-density polyethylene (HDPE), 4.5� 108 particles/ml;

and Zymosan, 1.25mg/ml. The two HDPE groups with asterisks and

all of the Zymosan groups were significantly different from control

values at respective time periods by ANOVA with po0:0002:Mean7standard error of the mean.

S. Noordin et al. / Biomaterials 25 (2004) 5347–5352 5349

2.4. Incubation of particulate material with human serum

Predetermined volumes of human serum under sterileconditions were incubated with varying concentrationsof the particulates in triplicate in a shaker water bath at37�C for different time periods in 1.5ml polypropylenetubes. During incubation, the samples were removedand vortex mixed each time to ensure that the serumadequately came in contact with the particulate materi-al, particularly in the case of polyethylene and PMMAparticles. In the negative controls, no particulate wasadded. As a positive control for complement activation,Zymosan was added to the serum to give a finalconcentration of 1.25mg/ml.

A separate tube was used for each time point and eachparticulate concentration to ensure sterility as well as tomaintain the same ratio of serum to particulate materialin each tube. At the end of the desired incubationperiod, to all the samples including the negative andpositive controls, Na2EDTA was added to achieve afinal concentration of 10mm to inhibit further comple-ment activation. The samples were then centrifuged at400g for 10min. Serum supernatant, containing noparticulate, was immediately frozen at �70�C for latercomplement assay. In the case of polyethylene andPMMA particulates, the serum was filtered through0.22 mm millipore filters to separate the particles.

2.5. Dose response and kinetics (temporal profile)

experiments

To assess complement activation in the presence ofparticles at constant incubation time, increasingamounts of particles were added to a constant volumeof human serum to achieve constant final volumes. Inaddition, to assess the temporal profile of activation ofcomplement for kinetics experiments, a constantamount of particles was added to a constant volumeof human serum and incubated at 37�C with gentleagitation. After a selected time, the complement activa-tion was determined as described below: 15 and 30min,and 1, 4, and 20 h. Further, because we wanted tocompare the level of complement activation on thesurface of different particulates, we incubated equivalentsurface areas of the different particulates with constantvolumes of human serum. Care was taken to ensure thatthe cumulative surface area of particulates in every tubewas the same.

2.6. Assay for complement activation

Complement activation was assessed by measuringthe concentration of C3a (desArg) in the supernatantusing a radioimmunoassay (RIA) kit (Amersham). Theradioactivity level was determined using a Beckmangamma 5500 counter. The use of RIA for measurement

of activated protein, C3a, made it possible to evaluatemany samples at once, and thus to directly compare atthe same time the potential of several differentparticulates to activate complement. Once C3a isgenerated in the serum, it is rapidly cleaved into C3adesArg by carboxypeptidase N in the serum. C3adesArg differs from C3a by only a single amino acidand both are recognized by the antibody used in theRIA. Since the antibody for the RIA also recognizes theC3a portion of the uncleaved C3, removal of C3 prior tothe assay was performed in the precipitation step.

2.7. Statistical analyses

Statistical significance of results obtained fromsimultaneous testing of three specimens (n ¼ 3) wascalculated by analysis of variance (ANOVA) andFisher’s protected least squares differences (PLSD) test.The results are expressed as mean values 7 standarderror of the mean (SEM) and the criterion forsignificance was pp0:05:

3. Results

3.1. Kinetics experiments

Various prosthetic particulate materials were incu-bated with human serum for selected time periods(Fig. 1) and the serum samples analyzed for the contentof C3a desArg. All particulates gradually increasedcomplement activation with time (Fig. 1; mean7SEM).

ARTICLE IN PRESS

Fig. 3. This chart demonstrates the C3a desArg levels in response to

incubation of UHMWPE particles of different doses with serum

(n ¼ 3). There were no significant differences comparing the values

from the particles and the negative control. The positive control,

Zymosan, level of 11772.5 was significantly higher than the results

from the UHMWPE samples (ANOVA, po0:0001).

Table 1

C3a desArg level in response to incubation of serum with equivalent

surface areas of particles (n ¼ 3)

Particulate material Surface area C3a desArg

(� 103mm2) (mg/ml)

Negative control — 45.279

UHMWPE-18 4.3 33.672

UHMWPE-130 4.5 30.670.6

CPT 4.3 35.274.6

S. Noordin et al. / Biomaterials 25 (2004) 5347–53525350

More C3a desArg generation was observed withexposure to 4.5� 108 HDPE particles/ml than in thenegative controls, and the differences steadily increasedduring the observation period (po0:01 at 4 and 20 h). Incontrast, the changes in the level of C3a desArg in serumincubated with 2.5� 108 CPT particles/ml, did notsignificantly differ from those in the negative controlat any time point (Fig. 1). The positive control,Zymosan, showed exceedingly high levels of C3a desArg(more than 5-fold greater than the negative control)after 15min of incubation with serum. Interestingly, theC3a desArg levels in response to Zymosan activationdecreased over the successive time intervals in relation tothe response at 15min, until 4 h when the maximumcomplement activation was seen. However, it must bepointed out that at all time intervals, exposure toZymosan resulted in significantly higher C3a desArglevels than the negative controls (po0:04), thus validat-ing these experiments.

3.2. Dose response

Increasing amounts of the different prosthetic parti-culate materials shown in Figs. 2 and 3 were incubatedwith human serum for 2 h, after which complementactivation was stopped and the samples were processedfor C3a desArg analyses. The doses of 250, 500, 750, and1000� 106 particles/ml (Fig. 2) correspond to surfaceareas of 16.3, 32.7, 49, and 65.4mm2/ml, respectively.C3a desArg levels activated by the positive control,Zymosan, (11772.5 mg/ml) were statistically signifi-cantly higher (p ¼ 0:002) than the levels found for thenegative control (45.279 mg/ml; Fig. 2), corroboratingthe experimental set-up. HDPE and PMMA particleconcentrations ranging from 250–1000million/ml andCPT particle concentrations up to 5000million/ml did

Fig. 2. C3a desArg levels in response to incubation of particles with

serum (n ¼ 3). There were no significant differences comparing the

values from the particles and the negative control. The positive

control, Zymosan, level was 11772.5, which was significantly different

from the responses to the particles and the negative control value

(ANOVA, po0:0001). PMMA, polymethylmethacrylate.

not activate C3 above the negative control (Fig. 2).Similarly, serum incubated with 10–40millionUHMWPE-18 particles/ml and 0.025–0.2� 106

UHMWPE-130 particles/ml failed to stimulate thecomplement system above the negative control (Fig.3). Inter-group differences also were statistically incon-sequential. Interestingly, for the UHMWPE samples theC3a desArg level at the highest particle concentrationwas lower than that for the second highest concentration(Fig. 3) however, this difference was significant only inthe case of UHMWPE-18 (p ¼ 0:02).

3.3. Incubation of serum with equivalent surface areas of

individual particles

It was only possible to test the UHMWPE-18,UHMWPE-130 and CPT particles at one equivalentsurface area due to limitations involving the density,particle diameter, and availability of these materials atthe time the assays were performed. Complementactivated by these surfaces was less than the negativecontrol (Table 1), with no differences in the C3a desArglevels among the groups.

ARTICLE IN PRESSS. Noordin et al. / Biomaterials 25 (2004) 5347–5352 5351

4. Discussion

The notable finding of the present study was theactivation of the complement protein, C3, by HDPEparticles. This statistically significant finding was madeusing a high concentration of particles (450million/ml)after 4 and 20 h of incubation with serum. Lowerconcentrations of HDPE particles and of particulates ofother orthopaedic biomaterials at shorter incubationtimes did not result in the elevation of C3a desArg levelsabove the negative control values. Difficulties in thehandling characteristics and procurement of the parti-cles precluded direct comparison of comparable doses ofall of the materials investigated. It is not possible toconclude, therefore, how the materials compare withrespect to their activation of complement. A recentinvestigation [14] also found that polyethylene, in theform of UHMWPE particles, activated the productionof C3b using an immunoassay. It is difficult to otherwisedirectly compare those findings with the results of thepresent study because of differences in the methodologyand particle size and dose. Collectively, however, thesestudies make the case for the importance of continuedinvestigation of the role of complement activation in thepolyethylene particle-induced inflammation and osteo-lysis. The large number of such particles, with theattendant large surface areas, that are known to begenerated by selected total joint replacement prosthesesas they undergo wear may exceed in some patients thethreshold value required for meaningful complementactivation.

An interesting observation in the other recent work[14] was that the levels of C3b that were measuredactually decreased with increasing volumes of thepolyethylene particles. The authors suggested that thismight have been due to the adsorption of C3b to theparticle surface. Radioimmunoassays measure the fluidphase fractions of the target proteins. Consequently,differences in the affinity of the test materials for activecomplement fragments may obscure true differences incomplement-activating ability. No data are available onthe relative adsorptive capacities of the materials tested,making corrections for possible error on this scoreimpossible. This observation demonstrates the complex-ity of the system due to the myriad factors that canaffect diffusible levels of activated complement mole-cules. It may be instructive in future work to relate theanaphylatoxin behavior of complement fragments, asthey bind to the surfaces of mast cells and basophils andcause the release of inflammatory mediators, to thebinding of these molecules to biomaterial particles.

In this study it was not possible to determine whetherthe complement activation by the HDPE particles wasdue to surface contact between complement and theparticulate materials only or was partly due to dissol-ving chemicals eluted from the particulates during the

incubation. Hydrophobic polymers such as polyethyleneare characterized by a relative paucity of hydroxylgroups at the surface and the low water wettability ofthese substances makes them strong adsorbers of plasmaproteins [16,17]. Conformational changes brought aboutduring adsorption possibly reveal nucleophilic groups,which provide sites for complement activation. Thesepolymers differ in their surface free energy concentra-tions, and hence their relative affinity to water andplasma proteins. A critical range probably exists withinwhich complement activation is least facilitated [18].

Surface flaws could also contribute to complementactivation by trapping air bubbles [19]. Prior studiesdemonstrated [20] that denucleation of silicone cathetersto eliminate surface microbubbles significantly reducedthe amount of C3a and C5a produced during incubationin plasma. Surface irregularities that could be playingsuch a role have been found on scanning electronmicroscopic examination of polyethylene wear debrisfrom retrieved hip arthroplasty specimens [21].

The question of why increased concentrations of thedifferent particulate materials did not result in anincreased quantity of C3 activation is difficult to answer.It is possible that the incubation period was too short asalso may be suggested by our temporal profile experi-ments. Another explanation is that there is only alimited number of C3 convertases that can form on thesurface of these particles. Addition of more particleswould not result in any further C3 conversion. Ournegative findings of complement activation by PMMAparticles are in agreement with other investigators.PMMA, the most commonly used synthetic polymer inthe manufacture of different intra-ocular lens styles, is arelatively inert material with respect to complementactivation [22,23].

The role of the complement system in particle-inducedinflammatory processes and osteolysis warrants contin-ued study. That this investigation found that polyethy-lene particles, even though at high concentrations, couldactivate a selected complement protein demonstratesthat this system may play a role in the degradativeprocesses associated with the wear of orthopaedicimplants.

References

[1] Maloney WJ, Smith RL. Periprosthetic osteolysis in total hip

arthroplasty: the role of particulate wear debris. J Bone Jt Surg

1995;77-A:1448–61.

[2] Goldring SR, Schiller AL, Roelke M, Rourke CM, O’Neill DA,

Harris WH. The synovial-like membrane at the bone-cement

interface in loose total hip replacements and its proposed role in

bone lysis. J Bone Jt Surg 1983;65A:575–84.

[3] Jiranek WA, Machado M, Jasty M, Jevsevar D, Wolfe HJ,

Goldring S, Goldberg MJ, Harris WH. Production of cytokines

around loosened cemented acetabular components. Analysis with

ARTICLE IN PRESSS. Noordin et al. / Biomaterials 25 (2004) 5347–53525352

immunohistochemical techniques and in situ hybridization.

J Bone Jt Surg 1993;75-A:863–79.

[4] Klein DC, Raisz LG. Prostaglandins: stimulation of bone

resorption in tissue culture. Endocrinology 1970;86:1436–40.

[5] Gowen M, Wood DD, Ihrie EJ, McGuire MKB, Russell RGG.

An interleukin 1 like factor stimulates bone resorption in vitro,

vol. 306. 1983. p. 378–80.

[6] Bosanquet AG, Ishimaru J, Goss AN. The effect of silastic

replacement following discectomy in sheep temporomandibular

joints. J Oral Maxillofac Surg 1991;49:1204–9.

[7] Zardeneta G, Mukai H, Marker V, Milam SB. Protein interac-

tions with particulate teflon: implications for the foreign body

response. J Oral Maxillofac Surg 1996;54:873–8.

[8] Schliephake H, Schmelzeisen R, Maschek H, Haese M. Long-

term results of the use of silicone sheets after diskectomy in the

temporomandibular joint: clinical, radiographic and histopatho-

logic findings. Int J Oral Maxillofac Surg 1999;28:323–9.

[9] Barrington R, Zhang M, Fischer M, Carroll MC. The role of

complement in inflammation and adaptive immunity. Immunol

Rev 2001;180:5–15.

[10] Muller-Eberhard HJ. Molecular organization and function of the

complement system. Ann Rev Biochem 1988;57:321–47.

[11] Ehlenberger AG, Nussenzweig V. The role of membrane receptors

for C3b and C3d in phagocytosis. J Exp Med 1977;145:357–71.

[12] Henson PM. The immunological release of constituents from

neutrophil leucocytes. I. The role of antibody and complement on

nonphagocytosable particles. J Immunol 1971;107:1535–40.

[13] Sutton JS, Weiss L. Transformation of monocytes in tissue

culture into macrophages, epithelioid cells, and multinucleated

giant cells. J Cell Biol 1966;28:803.

[14] DeHeer DH, Engels JA, DeVries AS, Knapp RH, Beebe JD. In

situ complement activation by polyethylene wear debris. J Biomed

Mater Res 2001;54:12–9.

[15] Hugli TE, Chenoweth DE. Biologically active peptides of

complement: techniques and significance of C3a and C5a

measurements. In: Nakamura R, Dito WR, Tucker ES, editors.

Immunoassays: clinical laboratory techniques for the 1980s. New

York: W.H. Freeman & Co; 1980. p. 443–60.

[16] Baszkin A, Lyman DJ. The interaction of plasma proteins with

polymers. I. Relationship between polymer surface energy and

protein adsorption/desorption. J Biomed Mater Res 1980;14:

393–403.

[17] Elwing H, Nilsson H, Svensson KE, Askendal A, Nilsson U,

Lundstrom I. Conformational change of a model protein

(complement factor 3) adsorbed on hydrophobic and hydrophilic

solid surfaces. J Colloid Interface Sci 1988;125:139.

[18] Hoffman AS, Horbett TA, Ratner BD. Interaction of blood and

blood components at hydrogel interfaces. Ann NY Acad Sci

1977;283:372–82.

[19] Ward CA, Koheil A, Johnson WR, Madras PN. Reduction in

complement activation from biomaterials by removal of air nuclei

from the surface roughness. J Biomed Mater Res 1984;18:255–69.

[20] Kalman PG, Ward CA, McKeown NB, McCullough D,

Romaschin AD. Improved biocompatability of silicone rubber

by removal of surface entrapped air nuclei. J Biomed Mater Res

1991;25:199–211.

[21] McKellop HA, Campbell P, Park S, Schmalzried TP, Grigoris P,

Amstutz HC, Sarmiento A. The origin of submicron polyethylene

wear debris in total hip arthroplasty. Clin Orthop 1995;311:3–20.

[22] Tuberville AW, Galin MA, Perez HD, Banda D, Onh R,

Goldstein IM. Complement activation by nylon- and polypropy-

lene-looped prosthetic intraocular lenses. Invest Opthalmol Visual

Sci 1982;22:727–33.

[23] Herzlinger GA, Bing DH, Stein R, Cumming RD. Quantitative

measurement of C3 activation at polymer surfaces. Blood

1981;57:764–70.