INFECTION AND IMMUNITY, Sept. 1987, p. 2148-2154 Vol. 55, No. 90019-9567/87/092148-07$02.00/0Copyright © 1987, American Society for Microbiology

Bacterial Antigens Induce Collagenase and Prostaglandin E2Synthesis in Human Gingival Fibroblasts through a Primary Effect

on Circulating Mononuclear CellsJOAN K. HEATH,lt* SUSAN J. ATKINSON,' ROSALIND M. HEMBRY,1 JOHN J. REYNOLDS,'

AND MURRAY C. MEIKLE" 2

Cell Physiology Department, Strangeways Research Laboratory, Cambridge CB] 4RN,l and Department of Orthodontics,Eastman Dental Hospital and Institute of Dental Surgery,

London WCJX 8LD,2 United Kingdom

Received 3 February 1987/Accepted 11 June 1987

Our previous work suggests that one mechanism through which connective tissue breakdown might occur inperiodontal diseases is the production of metalioproteinases, including collagenase, by gingival fibroblasts. Inthis study we investigated whether highly purified preparations of lipopolysaccharide (LPS) and lipoteichoicacid (LTA) from a number of putative periodontal pathogens could induce monolayer cultures of humangingival fibroblasts to synthesize collagenase and prostaglandin E2. Using both biochemical assays andimmunocytochemical techniques, we found that cells synthesized only very small amounts of collagenase indirect response to LPS or LTA (0.1 to 20.0 ,ug/ml). At the highest dose of both antigens, prostaglandin E2production was increased. We then studied whether LPS and LTA could signal collagenase production byinteracting primarily with a different cell type. Our results show that LPS and LTA were each able to stimulatecultures of human blood mononuclear cells (greater than 95% monocytes) to release collagenase-inducingcytokines. By indirect immunocytochemistry, we found that a large proportion of human gingival fibroblastswas activated to produce collagenase by supernatants from LPS- and LTA-stimulated mononuclear cells,whereas gingival fibroblasts cultured with supernatants from unstimulated mononuclear cells were not.Furthermore, in a population of activated fibroblasts we demonstrated, using a double-labeling technique, thatsome cells made collagenase and the specific tissue inhibitor of metalloproteinases (TIMP) simultaneously. Asyet, the collagenase-inducing signals remain poorly characterized but the interleukins-1 and tumor necrosisfactors seem likely candidates.

We previously showed that explants of gingiva frompatients undergoing periodontal surgery synthesized threemetalloproteinases (MPs), collagenase, gelatinase, andstromelysin (formerly called proteoglycanase), when main-tained in culture for several days (15). Though all of theseenzymes were synthesized as inactive precursors (latentforms), some of the collagenase in the culture medium hadbecome activated and therefore had the potential to mediateconnective tissue destruction. Collagenase is a specific pro-teinase, able to cleave only interstitial collagens (types I, II,and III), but the other MPs demonstrate wider substratespecificity (11, 23), such that the concerted action of thethree enzymes is capable of degrading all of the componentsof connective tissue matrices. Interstitial collagens may alsobe degraded in vitro by proteinases of the other classes(serine, aspartic acid, and cysteine), and it remains to beestablished whether collagen degradation in vivo takes placeas a result of the cooperative action of all of theseproteinases or whether the activity of a single proteinase isdominant in specific conditions. MPs have neutral pH op-tima and have been shown to mediate extracellular matrixdegradation in a number of in vitro model systems (6, 12),and it has been suggested that MPs have an important role in

* Corresponding author.t Present address: Department of Bone Biology and Osteoporosis

Research, Merck Sharp & Dohme Research Laboratories, WestPoint, PA 19486.

mediating connective tissue breakdown in both physiologicaland pathological degradative processes (24). Indeed, wehave proposed that MPs are largely responsible for thedestruction of periodontal tissue that takes place during theprogress of periodontal diseases (22).When we grew gingival fibroblasts from explants of tissue,

we found that the cells no longer secreted collagenase intothe medium. Instead, we detected significant levels of tissueinhibitor of metalloproteinases (TIMP; 15). This was consis-tent with our previous studies carried out with other con-nective tissue cells, such as chondrocytes (10, 30) andosteoblasts (14), and supports the idea that MPs are onlysynthesized by cells in response to specific inducing signals.Such signals appear to persist when cells are grown inexplant culture but are lost when cells are grown as purifiedpopulations. In this study, we tried to mimic the inducingstimulus present in explant cultures of gingiva from peri-odontitis patients by incubating the fibroblasts with antigeniccell wall components from a number of putative periodontalpathogens. We investigated whether highly purified lipopoly-saccharide (LPS) or lipoteichoic acid (LTA) could directlyinduce MP and prostaglandin E2 (PGE2) synthesis in gingivalfibroblasts or whether their action was mediated by anothercell type. To do this, we incubated human peripheral bloodmononuclear cells with purified LPS and LTA and tested thesupernatants on fibroblasts for collagenase-inducing activity.We chose mononuclear cells because they are characteristi-cally found at sites of inflammation and are thought to

2148

on August 31, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

COLLAGENASE SYNTHESIS INDUCED IN GINGIVAL FIBROBLASTS

mediate degradative effects in other connective tissue dis-eases, notably rheumatoid arthritis (25).

MATERIALS AND METHODS

Preparation of antigens from various bacterial strains.Actinobacillus actinomycetemcomitans (NCTC 9710) andBacteroides gingivalis (W50) were obtained from P. Marsh,Public Health Laboratory, Porton Down, England. Strepto-coccus mutans (NCTC 10449) and Veillonella sp. wereobtained from the Central Public Health Laboratory, Col-indale, London, England, and Streptococcus sanguis(ATCC 10558) was kindly supplied by B. Rosan, Universityof Pennsylvania, Philadelphia. LPS from A. actinomyce-temcomitans, B. gingivalis, and Veillonella sp. and LTAfrom S. mutans and S. sanguis were highly purified byMichael Wilson, Department of Clinical Pathology and Im-munology, Eastman Dental Hospital, London, England, bypreviously published methods (18, 19).

Gingival fibroblast cultures. Fibroblasts were grown fromhuman gingival explants as described previously (15). Cellsfor subculturing were incubated with trypsin (500 ,ug/ml)-EDTA (200 ,ug/ml) in phosphate-buffered saline (PBS) andseeded into 75-cm2 flasks (106 per flask) with 10 ml ofDulbecco modified Eagle medium (DMEM) containing 100 Uof penicillin per ml, 100 pig of streptomycin per ml, and 10%fetal calf serum (GIBCO Laboratories, Grand Island, N.Y.)and grown to confluence. For experiments, the cells werepassaged into either 3-cm (diameter) dishes (5 x 105 cells perdish) or multiwell trays (area, 2 cm2; 105 cells per well).Having reached confluence, the cells were washed exhaus-tively with fresh DMEM to remove fetal calf serum andcultured in fresh serum-free DMEM (controls) or the sampleto be tested, and the medium was changed every 48 h.Supernatants were then assayed for the presence of colla-genase or PGE2 as described below. Differences betweengroups were analyzed by the Student t test.

Blood mononuclear cell cultures. Human peripheral blood(50 ml), obtained by venipuncture from healthy volunteers,was dispensed into heparinized vessels and diluted 1:1 withCa2+- and Mg2+-free PBS containing 1 mM EDTA. Themononuclear cells were separated by centrifugation on aFicoll-Paque (Pharmacia, Uppsala, Sweden) gradient as de-scribed by Boyum (4). The cells were washed in sterile PBS,suspended in DMEM containing antibiotics and 10% heat-inactivated rabbit serum (Wellcome Research Laboratories,Beckenham, England) and cultured at 37°C in a humidifiedatmosphere of 5% C02-95% air at a concentration of 4 x 106cells per 3-cm dish. After 2 h, the nonadherent cells werediscarded, and in some cultures adherent cells were strainedfor naphthal acetate-nonspecific esterase by the method ofKoski et al. (20) to estimate the proportion of monocytespresent and checked for viability by exclusion of TrypanBlue. The adherent mononuclear cells (more than 95%monocytes) were then cultured for 24 or 48 h with thevarious antigen preparations (20 ,ug/ml). The mononuclearcell-conditioned medium (MCM) was then clarified by cen-trifugation and used in experiments at a concentration of20%.

Collagenase assay. Collagenase activity in supernatantsfrom gingival fibroblasts was measured by the release of'4C-labeled peptides from thermally reconstituted trypsin-resistant fibrils of 14C-acetylated rat skin type I collagen (28).One unit of collagenase hydrolyzes 1 pig of reconstitutedfibrils per min at 35°C. Procollagenase was activated by

including 4-aminophenylmercuric acetate in the fibril assay(final concentration, 0.67 mM).

Assay for PGE2. Culture supernatants were assayed forPGE2 by specific radioimmunoassay. Portions of media wereadded to 60 nCi of [n_3H]PGE2 (Amersham International,Amersham, United Kingdom) and rabbit anti-PGE2 bovineserum albumin (Steranti Research, St. Albans, United King-dom) in sodium phosphate buffer (100 mM, pH 7.0) contain-ing 1 mM EDTA, 0.1% gelatin, and 0.1% sodium azide. Afterincubation at 4°C for 18 h, bound and free PGE2 wereseparated on dextran-coated charcoal (Steranti Research),incubated at 4°C for 10 min, and spun at 1,000 x g for 10 min,and the supernatant was counted.

ConA-stimulated MCM. Supernatants from human periph-eral blood monocytes cultured for 24 h with 100 ,ug ofconcanavalin A (ConA) per ml provided positive controls forthe experiments. Such preparations contained interleukin-1(IL-1; 10, 27) and probably other cytokines capable ofinducing collagenase and PGE2 synthesis in connectivetissue cells, such as tumor necrosis factor a (7). To stimulatethe human gingival fibroblasts, the preparations were rou-tinely used at a concentration of 20%.Immunocytochemical localization of collagenase and TIMP.

The specificity of the antisera used and the, methodology forthe intracellular immunocytochemical localization of collavgenase and TIMP have been published in full (16, 17).Briefly, cultures of human gingival fibroblasts on glass coverslips in 3-cm dishes were washed twice in DMEM andincubated with fresh DMEM containing 5 F.M monensin forthe last 3 h of the experiment. The cells were then washedtwice in PBS, fixed for 5 min in 4% paraformaldehyde, andpermeabilized with 0.1% Triton X-100 in PBS for 5 min.After repeated washing, the cells were incubated at roomtemperature with either sheep anti-rabbit collagenase immu-noglobulin G or sheep anti-human TIMP immunoglobulin Gfor 30 min. The cells were then washed and treated with afluorescein isothiocyanate (FITC)-labeled pig Fab to sheepFab monomer (pig FITC) for 30 min. The cover slips werewashed with PBS and mounted in Citifluor (Citifluor Ltd.,London, England). Cells were observed in a Zeiss Photo-microscope III with standard FITC and TRITC filters andcounted over at least eight random fields so that a total ofapproximately 250 cells was counted on each cover slip. Ifthe Golgi apparatus was visible within a cell, it was scoredpositive irrespective of brightness. Photographs were takenon either Kodak Ektachrome film uprated to 1600 ASAduring processing or TriX film developed for 12 min.To determine whether gingival fibroblasts made both

collagenase and TIMP simultaneously, 10 mg of the sheepantibody immunoglobulin G to rabbit collagenase was la-beled with tetramethylrhodamine isothiocyanate (TRITC),and 10 mg of the sheep antibody to human TIMP was labeledwith FITC by the method of The and Feltkamp (29). Cellsgrown on cover slips were fixed and permeabilized asdescribed above and incubated for 30 min with a mixture ofthe antibodies (both at a 1:100 dilution in PBS). The cellswere then washed repeatedly in PBS, mounted, and assessedas described above.

RESULTS

Direct effects of antigens. Supernatants from unstimulatedmonolayer cultures of human gingival fibroblasts containedvery low levels of collagenase when assayed by the 14C-acetylated collagen fibril method (Table 1). At the end of the

2149VOL. 55, 1987

on August 31, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

2150 HEATH ET AL.

TABLE 1. Collagenase production by human gingival fibroblasts in response to direct stimulation withvarious bacterial antigens and ConA-stimulated MCM

Mean (±SEM) collagenase production (U/ml) on daysa:Antigen (concn, ,ug/ml) 2-4 4-6

Controls in DMEM alone 0.063 ± 0.01 0.046 ± 0.014 0.139 ± 0.008

A. actinomycetemcomitans LPS0.1 0.075 ± 0.014 0.029 ± 0.003 0.055 ± 0.0101.0 0.091 ± 0.023 0.018 ± 0.003 0.057 ± 0.00110.0 0.074 ± 0.010 0.024 ± 0.002 0.150 ± 0.00420.0 0.062 ± 0.009 0.030 ± 0.005 0.162 ± 0.012

B. gingivalis LPS0.1 0.102 ± 0.026 0.027 ± 0.001 0.108 ± 0.0301.0 0.122 ± 0.OO9' 0.035 ± 0.030 0.104 ± 0.03010.0 0.062 ± 0.007 0.080 ± 0.023 0.116 ± 0.01920.0 0.069 ± 0.001 0.019 ± 0.001 0.067 ± 0.030

Veillonella sp. LPS0.1 0.071 ± 0.017 0.032 ± 0.009 0.026 ± 0.0101.0 0.088 ± 0.020 0.023 ± 0.004 0.022 ± 0.00310.0 0.075 ± 0.015 0.029 ± 0.005 0.049 ± 0.00220.0 0.091 ± 0.006 0.028 ± 0.003 0.046 ± 0.003

S. mutans LTA0.1 0.079 ± 0.007 0.050 ± 0.10 0.052 ± 0.0011.0 0.083 ± 0.023 0,019 ± 0.002 0.031 ± 0.00110.0 0.114 ± 0.005b 0.034 ± 0.010 0.211 ± 0.020b20.0 0.116 ± 0.019b 0.018 ± 0.110 0.195 ± 0.020b

S. sanguis LTA0.1 0.101 ± 0.003b 0.018 ± 0.001 0.035 ± 0.0101.0 0.117 ± 0.014b 0.032 ± 0.001 0.022 + 0.00810.0 0.146 ± 0.016" 0.033 ± 0.004 0.023 ± 0.00520.0 0.123 ± 0.003b 0.023 ± 0.001 0.035 ± 0.002

ConA-stimulated MCM (20%) 2.10 ± 0.13b 2.62 ± 0.26b 2.63 ±0.13b

aEach value represents the mean of four replicate cultures.b Significantly higher than control value (P < 0.05; Student t test for unpaired groups).

6-day experiment, the cumulative collagenase production collagenase per ml per 48 h in response to ConA-stimulatedwas only 0.25 U/ml. When the cells were exposed to highly MCM. However, in response to the highest dose of thepurified LPS from either A. actinomycetemcomitans, B. antigens (20 ,ug/ml), human gingival fibroblasts did increasegingivalis, or Veillonella sp. and LTA from either S. mutans their output of PGE2 into the medium (Table 2).or S. sanguis at a dose range of 0.1 to 20 pLg/ml, it was found Immunocytochemical procedures were also used to inves-that some values were statistically greater than control levels tigate whether cultures ofhuman gingival fibroblasts respond(P < 0.05), though the absolute amounts of enzyme pro- to purified bacterial antigens by synthesizing collagenase. Induced were still so small as to be unlikely to have biological a typical experinment with fourth-passage fibroblasts, it wassignificance (Table 1). That the cells were capable of induc- found that 8.5% of unstimulated cells contained brightlytion was confirmed by the inclusion of a pQsitive control in staining Golgi when stained for collagenase by the two-stepthe experiment; the cells made approximately 2.5 U of (indirect) method after monensin treatment and 36% of cells

TABLE 2. Production of PGE2 by human gingival fibroblasts in response to purified preparations of LPS or LTA

Mean (±SEM) PGE2 production (ng/ml) on daysa:Antigen (concn, ,ug/ml) 24 4-6

DMEM controls 4.3 ± 0.6 10.5 ± 0.9 5.3 ± 1.2A. actinomycetemcomitans LPS (20) 25.2 ± 1.9 15.1 ± 2.2 20.3 ± 3.8Veillonella sp. LPS (20) 116.0 ± 30.0 91.5 ± 10.0 34.2 ± 8.2B. gingivalis LPS (20) 154.0 ± 26.0 45.0 ± 5.1 161.0 ± 24.0S. sanguis LTA (20) 27.2 ± 9.4 7.5 ± 1.3L 6.1 ± 1.2bConA-stimulated MCM (20%) 325.0 ± 50.0 217.0 ± 26.0 87.5 ± 10.6

aEach value represents the mean of four replicate cultures.b Except for these, all experimental values were significantly higher than those of corresponding controls (P < 0.05).

INFECT. IMMUN.

on August 31, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

COLLAGENASE SYNTHESIS INDUCED IN GINGIVAL FIBROBLASTS

0-2 0-4Days in culture

0-2 0-4

Days In culture

TABLE 3. Secretion of PGE2 by human gingival fibroblasts inresponse to mononuclear cell supernatants

Mean (±SEM) PGE2 production (ng/ml) on daysa:Antigen (concn) 0-2 2-4 16

Controls in 3.2 + 1.1 4.1 ± 0.7 5.0 ± 0.6DMEM alone

Unstimulated 506.0 ± 105.0 97.3 ± 18.0 70.0 ± 4.0MCM (20%)

A. actinomyce- 1,066.0 ± 86.0 775.0 ± 175.0 624.0 ± 42.0temcomitansLPS-stimulatedMCM (20%o)

ConA-stimulated 812.0 ± 98.0 907.0 ± 48.0 451.0 ± 35.0MCM (20%)a Each value represents the mean of four replicate cultures. All experimen-

tal values were significantly higher than those of corresonding controls (P <0.01).

0-6

0-6

FIG. 1. Collagenase production by human gingival fibroblasts inresponse to supernatants from human peripheral blood mononuclearcells cultured with antigenic components from bacterial cell wallscompared with that of unstimulated cells and cells cultured withConA-stimulated MCM. For panel a, cells were fourth-passagefibroblasts plated into multiwell dishes at a density of 105 cells perwell and grown to confluence before testing. For panel b, cells werethird-passage fibroblasts plated into 3-cm (diameter) dishes (5 x 105cells per dish) and grown to confluence before testing, as describedin the text. Each point represents the mean value of four replicatecultures the standard error of the mean (where standard-errorbars are not shown, they are smaller than the symbols). Symbols: A,

ConA-stimulated MCM; A, unstimulated MCM; El, B. gingivalisLPS-stimulated MCM; *, S. mutans LTA-stimulated MCM; 0, A.actinomycetemcomitans LPS-stimulated MCM.

stained brightly with anti-human TIMP. In response to 4 ,ugof LPS per ml, 13.4% of the cells were positive for collagen-ase synthesis and 48% were positve for TIMP, but neither ofthese values was significantly higher than those of thecontrols. In an analogous experiment, only 3.1% of fourth-passage unstimulated cells were positive for collagenase and20.6% were positive for TIMP, and in the presence of 4 p.g ofLPS per ml, 1.8% of cells stained for collagenase and 19.7%stained for TIMP. These values were again not significantlydifferent from those of the unstimulated cells in the experi-ment.

Indirect effects of antigens. Supernatants from humangingival fibroblasts that had been cultured with medium fromhuman mononuclear cells stimulated with either LPS orLTA contained relatively high levels of collagenase com-pared with unstimulated controls or cells incubated withunstimulated MCM (Fig. la and b). In both of these exper-iments, the response of cells to unstimulated MCM wasindistinguishable from that of cells incubated in mediumalone; in contrast, ConA-stimulated MCM induced substan-tial collagenase secretion. The graphs show that LPS from B.gingivalis and A. actinomycetemcomitans (20 ,ug/ml) andLTA from S. mutans (20 ,ug/ml) all induced human mono-nuclear cells to synthesize a cytokine(s) which, when testedat 20% strength on human gingival fibroblasts (final concen-tration of LPS or LTA, 4 p,g/ml), stimulated collagenaseproduction for 6 days in culture to levels approaching thoseinduced by ConA-stimulated MCM. Compared with those ofcontrol cells, the levels of collagenase produced were signif-icantly higher for days 2 to 4 and 4 to 6 in all cases (P < 0.01),and for S. mutans LTA-stimulated MCM, significantly morecollagenase was also produced during the first 2 days ofculturing (P < 0.02). Similar data (not shown) were obtainedwhen LPS from Veillonella sp. or LTA from S. sanguis wasused in the experiments. All of the preparations of stimu-lated MCM were tested before use and none were found tocontain collagenase. The cells also made PGE2 in responseto A. actinomycetemcomitans LPS-stimulated MCM (Table3). Endogenous production of PGE2 by human gingivalfibroblasts was very low (< 5 ng/ml per 48 h), and this couldbe stimulated substantially by supernatants from untreatedmononuclear cells. However, much higher levels of PGE2were secreted in response to A. actinomycetemcomitansLPS- or ConA-stimulated MCM. For these experiments, itwas noted that the carryover of PGE2 in unstimulated MCMwas approximately 4 nglml, and in LPS-stimulated MCM itwas 20 ng/ml.When collagenase production was investigated by im-

munocytochemical localization, the response to LPS-stimu-lated MCM was very marked (Fig. 2). In one experiment(Fig. 2), 76.4% of the cells were induced to make collagenasecompared with 8.2% spontaneously synthesizing the enzyme(unstimulated cultures). Many of these cells were also acti-vated by LPS-stimulated MCM to make TIMP; 80.9% werescored positive for TIMP, compared with 36.1% for controlcultures, suggesting that some cells made both collagenaseand TIMP at the same time. To demonstrate whether anycells did indeed make both collagenase and TIMP, a double-

a

E0

Coc0

0V0

c00

CO

C5

00

0

Cu20

0.4 -

0.2 -

5.0

4.0

3.0

2.0

b

1.0o

VOL. 55, 1987 2151

1 .

on August 31, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

2152 HEATH ET AL.

FIG. 2. Immunolocalization of intracellular collagenase in hu-man gingival fibroblasts stimulated with LPS-stimulated MCM.Human gingival fibroblasts stimulated for 48 h with 20% LPS-stimulated MCM with monensin added for the final 3 h of theincubation period were fixed, permeabilized, and stained with sheepanticollagenase immunoglobulin G followed by pig FITC. Each cellhad a brightly fluorescent vesiculated Golgi apparatus and numeroussecretory vesicles (arrow). In this experiment, 76.4% of the cellssynthesized collagenase in response to LPS-stimulated MCM and80.9% of the cells made TIMP, compared with LPS-treated cells,13.4% of which stained for collagenase and 48.0% of which stainedfor TIMP. In unstimulated cultures, 8.2% of the cells stained forcollagenase and 36.1% stained for TIMP. Bar, 20 ,um. In ananalogous experiment, 53.6% of the cells synthesized collagenase inresponse to LPS-stimulated MCM and 25.3% of the cells madeTIMP, compared with control cells, 3.1% of which stained forcollagenase and 20.6% of which stained for TIMP.

labeling procedure was used. In one such experiment, 16%of the cells cultured with LPS-stimulated MCM synthesizedcollagenase and TIMP simultaneously (Fig. 3).

DISCUSSION

Our results suggest that antigens from the walls of bacteriacommonly present in dental plaque may induce connectivetissue breakdown via a mechanism which involves a primaryinteraction with circulating mononuclear cells. Though hu-man gingival fibroblasts did respond directly to the LPSpreparations in terms of PGE2 production, they were onlyinduced to synthesize collagenase in response to superna-tants from LPS- or LTA-stimulated mononuclear cells. Thiswas demonstrated in two ways: by assaying the fibroblastsupernatants for collagenase activity by using the reconsti-tuted collagen fibril assay and by localizing the enzymeimmunocytochemically. There are disadvantages to relyingon the collagen fibril assay alone for accurate estimation ofcollagenase synthesis by cell cultures. The main problem isthe likelihood of low levels of collagenase remaining unde-tected in supernatants containing relatively high levels ofTIMP. This occurs because for collagenase to be detected inthe fibril assay it must first be converted from its proenzymeform to its active form. This is usually achieved by including4-aminophenylmercuric acetate in the assay or by trypsinactivation. Once the enzyme is activated, it is avidly boundby free TIMP, yielding irreversible TIMP-collagenase com-plexes which cannot be measured in activity assays (5).However, such problems are overcome in assays based onthe use of the anti-rabbit collagenase antiserum since the

antibodies recognize all forms of the enzyme (latent, acti-vated, or bound [16]). The immunocytochemical approachhas the added advantage of providing information concern-ing the proportion of cells which become activated by agiven stimulus.

Using a double-labeling technique, we were able to dem-onstrate for the first time the simultaneous synthesis ofcollagenase and TIMP by a marked proportion of stimulatedcells. We envisage that for resorption to occur the level ofcollagenase synthesized and activated must be sufficient toexhaust the local concentration of TIMP so that binding tothe collagen substrate can take place. Little information isavailable at present to explain how collagenase becomesactivated in vivo, but one possible mechanism is through theenzyme plasmin. Werb et al. (31) demonstrated that colla-genase-producing rheumatoid synovial cells required incu-bation with plasmin before they degraded collagen films, andGavrilovic et al. (12) found that stimulated rabbit articularchondrocytes did not degrade collagen films unless plas-minogen was added. A mechanism involving a plasminogen-plasmin conversion gains further support from the observa-tions that osteoblasts in culture respond to parathyroidhormone by making both collagenase (15) and plasminogenactivator (13).We found that both LPS and LTA from a number of

different bacterial strains were equally effective in inducingmononuclear cells to produce cytokines able to stimulategingival fibroblasts to synthesize collagenase. Our conclu-sion is that if a cytokine-mediated mechanism plays animportant role in connective tissue destruction during peri-odontitis, then the disease can probably be triggered by alarge variety of bacteria or by combinations of differentbacteria. This is in contrast with those theories whichassume that only one or a small number of bacterial strainsis able to mediate the onset and progress of the disease. Wethink that the disease is more likely to be initiated in asusceptible individual by a critical number of organisms andthat the population of bacteria always comprises the mixture

FIG. 3. Dual localization of collagenase and TIMP in stimulatedhuman gingival fibroblasts. Cells were treated as described in thelegend to Fig. 2 but stained with collagenase-TRITC and TIMP-FITC together. Bars, 10 ,um. (a) Cells viewed through FITC filters.All three cells had a fluorescent juxtanuclear Golgi apparatus,indicating the presence of TIMP. (b) Cells viewed through TRITCfilters. The two lower cells had juxtanuclear rhodamine fluores-cence, indicating the presence of collagenase; these cells secretedcollagenase and TIMP simultaneously. The third cell (arrow) had nocollagenase fluorescence; it secreted only TIMP. Of 194 cellscounted on this cover slip, 16% stained for both collagenase andTIMP, 37.6% stained for collagenase only, and 9.3% stained forTIMP only. On another cover slip (data not shown), cells werestimulated with LPS directly (4 ,ug/ml); of 162 cells counted, only1.2% made collagenase and TIMP. Unstimulated cultures containedno cells making both collagenase and TIMP.

INFECT. IMMUN.

on August 31, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

COLLAGENASE SYNTHESIS INDUCED IN GINGIVAL FIBROBLASTS

of strains best able to exploit the environmental conditionspertaining at any one time.

In all of our experiments we included ConA-stimulatedMCM (a source of human IL-1) as a positive control.Recently, human IL-1 has been cloned (2, 21) and its diverseproinflammatory effects have been found to reside in twomolecules of distinct amino acid sequence: IL-la (pI, 5) andIL-ip (pl, 7). It is known that mononuclear cells synthesizeIL-1 in response to bacterial antigens (1, 9). Indeed, Marchet al. (21) used Escherichia coli LPS (10lOg/ml) to stimulatethe production of mRNA for IL-la and IL-ip in humanperipheral blood monocytes to enable them to construct acDNA library for their cloning experiments. They found thatmRNA for IL-ip was about 10 times more abundant thanmRNA for IL-la in this system.

In summary, we have presented evidence that one way inwhich periodontal pathogens might mediate connective tis-sue destruction during periodontal disease is through theability of antigens from their cell walls to stimulate cytokineproduction in circulating mononuclear cells. The cytokinesthen induce MP synthesis in resident gingival fibroblasts,and matrix degradation is initiated extracellularly. Thecytokines responsible remain poorly characterized, butmuch circumstantial evidence suggests that IL-la and IL-ipare likely to be involved. It is also possible that tumornecrosis factors play a significant role. It has recently beenestablished that both tumor necrosis factor ax, produced byantigen-activated macrophages, and tumor necrosis factor ,(formerly known as lymphotoxin) also exhibit many of theactivities previously described for IL-1 (8, 9, 27), includingthose on connective tissues (3, 7, 26).

ACKNOWLEDGMENTS

We thank our colleagues in the Cell Physiology Department forhelpful discussions, Jane Mead and Lisa Andrews for typing themanuscript, and Michael Wilson for generously providing the bac-terial antigens.These studies were funded by the Medical Research Council

(United Kingdom).

LITERATURE CITED

1. Arend, W. P., and R. J. Massoni. 1986. Characteristics ofbacterial lipopolysaccharide induction of interleukin 1 synthesisand secretion by human monocytes. Clin. Exp. Immunol. 64:656-664.

2. Auron, P. E., A. C. Webb, L. J. Rosenwasser, S. F. Mucci, A.Rich, S. M. Wolff, and C. A. Dinarello. 1984. Nucleotidesequence of human monocyte interleukin 1 precursor cDNA.Proc. Natl. Acad. Sci. USA 81:7907-7911.

3. Bertolini, D. R., G. E. Nedwin, T. S. Bringan, D. D. Smith, andG. R. Mundy. 1986. Stimulation of bone resorption and inhibi-tion of bone formation in vitro by human tumour necrosisfactors. Nature (London) 319:516-518.

4. Boyum, A. 1968. Separation of leukocytes from blood and bonemarrow. J. Clin. Lab. Invest. 21(Suppl.):77-89.

5. Cawston, T. E., G. Murphy, E. Mercer, W. A. Galloway, B. L.Hazleman, and J. J. Reynolds. 1983. The interaction of purifiedrabbit bone collagenase with purified rabbit metalloproteinaseinhibitor. Biochem. J. 211:313-318.

6. Chen, W.-T., K. Olden, B. A. Barnard, and F.-F. Chu. 1984.Expression of transformation-associated protease(s) that de-grade fibronectin at cell contact sites. J. Cell Biol. 98:1546-1555.

7. Dayer, J. M., B. Beutler, and A. Cerami. 1985. Cachectin/tumour necrosis factor stimulates collagenase and prostaglandinE2 production by human synovial cells and dermal fibroblasts. J.Exp. Med. 162:2163-2168.

8. Dayer, J. M., B. de Rochemonteix, B. Burrus, S. Demczuk, andC. A. Dinarello. 1986. Human recombinant interleukin-1 stimu-lates collagenase and prostaglandin E2 production by humansynovial cells. J. Clin. Invest. 77:645-648.

9. Dinarello, C. A. 1984. Interleukin-1. Rev. Infect. Dis. 6:51-95.10. Duncan, S. J., and J. J. Reynolds. 1983. The effects of razoxane

(ICRF 159) on the production of collagenase and inhibitor(TIMP) by stimulated rabbit articular chondrocytes. Biochem.Pharmacol. 32:3853-3858.

11. Galloway, W. A., G. Murphy, J. D. Sandy, J. Gavrilovic, T. E.Cawston, and J. J. Reynolds. 1983. Purification and character-ization of a rabbit bone metalloproteinase that degradesproteoglycan and other connective tissue components. Bio-chem. J. 209:741-752.

12. Gavrilovic, J., J. J. Reynolds, and G. Murphy. 1985. Inhibitionof type I collagen film degradation by tumour cells using aspecific antibody to collagenase and the specific tissue inhibitorof metalloproteinases (TIMP). Cell Biol. Int. Rep. 9:1097-1107.

13. Hamilton, J. A., S. R. Lingelbach, N. C. Partridge, and T. J.Martin. 1984. Stimulation of plasminogen activator in osteo-blast-like cells by bone resorbing hormones. Biochem. Biophys.Res. Commun. 122:230-236.

14. Heath, J. K., S. J. Atkinson, M. C. Meikle, and J. J. Reynolds.1984. Mouse osteoblasts synthesize collagenase in response tobone resorbing agents. Biochim. Biophys. Acta 802:151-154.

15. Heath, J. K., M. Gowen, M. C. Meikle, and J. J. Reynolds. 1982.Human gingival tissues in culture synthesize three metal-loproteinases and a metalloproteinase inhibitor. J. PeriodontalRes. 17:183-190.

16. Hembry, R. M., G. Murphy, T. E. Cawston, J. T. Dingle, andJ. J. Reynolds. 1986. Characterization of a specific antiserum formammalian collagenase from several species: immunolocaliza-tion of collagenase in rabbit chondrocytes and uterus. J. CellSci. 81:105-123.

17. Hembry, R. M., G. Murphy, and J. J. Reynolds. 1985. Im-munolocalization of tissue inhibitor of metalloproteinases(TIMP) in human cells. Characterization and use of a specificantiserum. J. Cell Sci. 73:105-119.

18. Huff, E. 1982. Lipoteichoic acid, a major amphiphile of gram-positive bacteria that is not readily extractable. J. Bacteriol.149:399-402.

19. Kiley, P., and S. C. Holt. 1980. Characterization of the lipopoly-saccharide from Actinobacillus actinomycetemcomitans Y4 andN27. Infect. Immun. 30:862-873.

20. Koski, I. R., D. G. Poplack, and R. M. Blaese. 1976. Anon-specific esterase stain for the identification of monocytesand macrophages, p. 359-362. In B. R. Bloom and J. R. David(ed.), In vitro methods in cell-mediated and tumor immunity.Academic Press, Inc., New York.

21. March, C. J., B. Mosley, A. Larsen, D. P. Cerretti, G. Braedt, V.Price, S. Gillis, C. S. Henney, S. R. Kronheim, K. Grabstein,P. J. Conlon, T. P. Hopp, and D. Cosman. 1985. Cloning,sequence and expression of two distinct human interleukin-1complementary DNAs. Nature (London) 315:641-647.

22. Meikle, M. C., J. K. Heath, and J. J. Reynolds. 1986. Advancesin understanding cell interactions in tissue resorption. Rele-vance to the pathogenesis of periodontal diseases and a newhypothesis. J. Oral Pathol. 15: 239-250.

23. Murphy, G., C. G. McAlpine, C. T. Poll, and J. J. Reynolds.1985. Purification and characterization of a bone metallopro-teinase that degrades gelatin and types IV and V collagen.Biochim. Biophys. Acta 831:49-58.

24. Murphy, G., and J. J. Reynolds. 1985. Current views of collagendegradation. Bioessays 2:55-60.

25. Nouri, A. M. E., G. S. Panayi, and S. M. Goodman. 1984.Cytokines and the chronic inflammation of rheumatic disease. I.The presence of interleukin-1 in synovial fluids. Clin. Exp.Immunol. 55:295-302.

26. Saklatvala, J. 1986. Tumour necrosis factor stimulates resorp-tion and inhibits synthesis of proteoglycan in cartilage. Nature(London) 322:547-549.

27. Saklatvala, J., L. M. C. Pilsworth, S. J. Sarsfield, J. Gavrilovi,

VOL. 55, 1987 2153

on August 31, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

2154 HEATH ET AL. INFECT. IMMUN.

and J. K. Heath. 1984. Pig catabolin is a form of interleukin 1.Biochem. J. 224:461-466.

28. Sellers, A., and J. J. Reynolds. 1977. Identification and partialcharacterization of an inhibitor of collagenase from rabbit bone.Biochem. J. 167:353-360.

29. The, T. A., and T. E. W. Feltkamp. 1970. Conjugation offluorescein isothiocyanate to antibodies. II. A reproduciblemethod. Immunology 18:875-881.

30. Trechsel, U., G. Dew, G. Murphy, and J. J. Reynolds. 1982.Effects of products from macrophages, blood mononuclear cellsand of retinol on collagenase secretion and collagen synthesis inchondrocyte culture. Biochim. Biophys. Acta 720:364-370.

31. Werb, Z., C. L. Mainardi, C. A. Vater, and E. D. Harris. 1977.Endogenous activation of latent collagenase by rheumatoidsynovial cells. Evidence for a role of plasminogen activator. N.Engl. J. Med. 296:1017-1023.

on August 31, 2020 by guest

http://iai.asm.org/

Dow

nloaded from