Treponema denticola Outer Membrane Inhibits Calcium Flux in Gingival Fibroblasts

Abstract

Treponema denticola is a cultivable oral spirochete which perturbs the cytoskeleton in cultured cells of oral origin, but intracellular signalling pathways by which it affects actin assembly are largely unknown. As the outer membrane (OM) of Treponema denticola disrupts actin-dependent processes that normally require precise control of intracellular calcium, we studied the effects of an OM extract on internal calcium release, ligand-gated and calcium release-activated calcium channels, and related mechanosensitive cation fluxes in human gingival fibroblasts (HGF). Single-cell ratio fluorimetry demonstrated that in resting cells loaded with Fura-2, baseline intracellular Ca²⁺ concentration ([Ca²⁺]_i) was not affected by treatment with OM extract, but normal spontaneous [Ca²⁺]_i oscillations were dramatically increased in frequency for 20 to 30 min followed by complete blockade. OM extract inhibited ATP-induced and thapsigargin-induced release of calcium from intracellular stores by 40 and 30%, respectively. Addition of Ca²⁺ to the extracellular pool following depletion of intracellular Ca²⁺ by thapsigargin and extracellular Ca²⁺ by EGTA yielded 59% less replenishment of [Ca²⁺]_i in OM extract-treated than in control HGF. In cells loaded with collagen-coated ferric oxide beads to stimulate integrin-dependent calcium release, baseline [Ca²⁺]_i was nearly doubled but was not significantly different in control and OM extract-treated cells. Magnetically generated tensile forces on the beads induced >300% increases of [Ca²⁺]_i above baseline. Cells preincubated with OM extract exhibited dose-dependent and time-dependent reductions in stretch-induced [Ca²⁺]_i transients, which were due to neither loss of beads from the cells nor cell death. The T. denticola OM inhibitory activity was eliminated by heating the OM extract to 60°C and by boiling but not by phenylmethylsulfonyl fluoride treatment. Thus nonlipopolysaccharide, nonchymotrypsin, heat-sensitive protein(s) in T. denticola OM can evidently inhibit both release of calcium from internal stores and uptake of calcium through the plasma membrane, possibly by interference with calcium release-activated channels.

Some pathogenic bacteria which parasitize humans have evolved ways to exploit host cell cytoskeleton-regulating signaling pathways for their own survival. Much of the ground-breaking research in this area has centered on enteroinvasive and other enteropathogenic bacteria (16, 27, 33) and more recently on periodontal pathogens like Actinobacillus actinomycetemcomitans and Porphyromonas gingivalis, which are able to invade human cells in vitro (21, 24, 32, 38). The periodontal pathogen Treponema denticola attaches to human cells of oral origin (12, 14, 37, 39). It perturbs actin assembly in human gingival fibroblasts (HGF), degrades endogenous HGF plasma membrane fibronectin, and causes HGF to detach from the extracellular matrix (3, 14, 39, 40). It induces reduction in F-actin and desmoplakin expression, disrupts barrier function, and blocks volume regulation in oral epithelial cells (12, 37). As actin assembly, volume regulation, and the integrity of cytoskeletal components of cell junctions depend on intact calcium signalling pathways (8, 20), we set out to determine how calcium-regulating and integrated intracellular signalling pathways are related to cytoskeletal perturbation by T. denticola. We selected HGF as target host cells in this study because (i) they are key cells which function in the homeostasis and wound-healing capacity of the gingiva exposed to surface components of T. denticola during periodontal infections; (ii) their F-actin rearranges upon exposure both to T. denticola and to outer membrane (OM) extracts of T. denticola, and a time course for F-actin depolymerization has been established (3, 40); and (iii) methods for studying intracellular calcium regulating pathways of both resting and mechanically stimulated HGF, and specific features of the pathways themselves, are rather well established (1, 18).

In nonexcitable stromal cells, such as fibroblasts and osteoblasts which maintain the integrity of the periodontium, a number of different mechanotransduction systems convert externally applied forces to signals that regulate cellular metabolism. Applied mechanical force stimulates actin assembly (28) among other essential physiological responses such as rapid bone remodelling and cell division (9, 30). These metabolic responses to mechanical force are mediated in part by the generation of second messengers including intracellular calcium ions (1, 26). Indeed, transient changes in intracellular Ca²⁺ concentration [Ca²⁺]_i and elevation of inositol-1,4,5-trisphosphate (IP₃) are early responses of cells to shear forces and strain (7, 26). To study second messengers under precisely controlled, predetermined conditions, we considered recent reports that HGF exhibit an increase in cytosolic Ca²⁺ by flux of Ca²⁺ through mechanosensitive calcium-permeable channels (18). This is part of a coordinated response along with actin assembly and cross-linking when tractional forces are applied magnetically via collagen-coated ferric oxide beads attached to the plasma membrane (17). Our specific aim was to determine potential mechanisms for the effect of a T. denticola OM extract on spontaneous oscillations and mechanosensitive fluxes of intracellular calcium in HGF.

MATERIALS AND METHODS

T. denticola culture conditions and preparation of T. denticola OM extracts.

T. denticola type strain ATCC 35405 stock cultures were maintained in a complex spirochete broth medium containing brain heart infusion, tryptic peptone, yeast extract, and volatile fatty acids and supplemented with rabbit serum as previously described (11, 40). This strain has been shown to stimulate F-actin rearrangement, plasma membrane blebbing, and degradation of endogenous fibronectin in HGF (3, 14, 40). OM extracts were prepared by our modification (40) of a previous method (10, 29). The bacteria were harvested at late stationary phase, washed, resuspended in phosphate-buffered saline (pH 7.2) containing 10 mM MgCl₂, and extracted in Triton X-100. After repeated centrifugation, the supernatant was dialyzed for several days until the OM precipitated, and it was centrifuged at 25,000 × g. The pellet was resuspended in the original volume of distilled water and stored frozen until used. The dry weight of the extract was determined after freeze-drying, and the protein content was determined (Bio-Rad assay), using bovine albumin (Sigma Chemical Co., St. Louis, Mo.) as a standard.

The OM extract was also tested for peptidase activity by using the chromogenic peptides N-succinyl-l-alanyl-l-alanyl-l-prolyl-l-phenylalanine-p-nitroanilide (SAAPNA) and N-benzoyl-dl-arginine-p-nitroanilide (BAPNA; Sigma) (14). The undiluted extract contained SAAPNA-degrading activity equivalent to 3 μg of chymotrypsin per ml and BAPNA-degrading activity equivalent to 1 μg of trypsin per ml.

For all experiments (except the dose-response experiment), a 1/10 dilution (in calcium buffer) of this OM preparation was used. It contained 0.27 mg (dry weight) per ml and a final assay concentration of 60 μg of protein per ml. When pretreatment of HGF with OM extract was specified, usually for 40 or 45 min as described below for individual experiments, the OM extract remained in the cell culture wells for the duration of the experiment. In some experiments, the OM extract was pretreated by boiling for 10 min to denature proteins; by preincubating with 170 μg of phenylmethylsulfonyl fluoride (PMSF; Sigma) per ml for 1 h (3, 14), which is known to inhibit the chymotrypsin-like activity of T. denticola; or by heating to 60°C for 30 min, to which the activity is relatively resistant. We have found that T. denticola cells and OM extract retain 77.5% ± 13.4% and 50.1% ± 18.8% (mean ± standard deviation), respectively, of their SAAPNA-degrading activity at this temperature.

Cell culture.

HGF were derived from primary explant cultures as described previously (1). Briefly, cells from passages 6 to 19 were grown as monolayers in T-75 flasks (Costar, Mississauga, Ontario, Canada) containing alpha minimal essential medium, 15% heat-inactivated fetal bovine serum (Flow Laboratories, Maclean, Va.), and a 1:10 dilution of an antibiotic solution (0.17% [wt/vol] penicillin V, 0.1% [wt/vol] gentamicin sulfate, 0.01 μg of amphotericin per ml [Sigma]). The cells were maintained at 37°C in a humidified incubator containing 5% CO₂ and were passaged with 0.01% trypsin (Gibco BRL, Burlington, Ontario, Canada). Twenty-four hours before each experiment, cells were harvested with 0.01% trypsin and ∼50,000 cells were plated onto 0.1-mm-thick, 31-mm-diameter, round glass coverslips (no. 0; Biophysica Technologies, Sparks, Md.) in 35-mm-diameter petri dishes (model 1008; Falcon, Becton Dickinson, Mississauga, Ontario, Canada).

Intracellular calcium.

We measured [Ca²⁺]_i as described previously (8). Briefly, cells on coverslips were incubated at 37°C with 3 μM Fura-2/AM (Molecular Probes, Eugene, Oreg.) for 20 min and then at 23°C for 10 min. The calcium-free buffer was bicarbonate free and contained 150 mM NaCl, 5 mM KCl, 10 mM d-glucose, 1 mM MgSO₄, 1 mM NaHPO₄, and 20 mM HEPES (pH 7.4) with an osmolarity of 291 mosM. For experiments requiring external calcium, 1 mM CaCl₂ was added to this buffer (calcium buffer). Whole-cell [Ca²⁺]_i measurements were obtained with an inverted microscope optically interfaced to an epifluorescence spectrofluorimeter and analysis system (Photon Technology International, London, Ontario, Canada). Fura-2 was excited at alternating (approximately 100-Hz) wavelengths of 346 and 380 nm from dual monochromators with slit widths set at 2 nm. Emitted fluorescence was passed through a 530/20-nm barrier filter. A variable-aperture, intrabeam mask was used to restrict measurements to single cells. Estimates of [Ca²⁺]_i independent of the precise intracellular concentration of Fura-2 were calculated from ratios of dual-excitation emitted fluorescence. Spontaneous oscillations in [Ca²⁺]_i in response to exposure to T. denticola OM extract were measured in resting cells and compared with oscillation patterns of HGF in OM extract-free medium. As HGF are known to respond with a well-characterized, immediate, severalfold rise in [Ca²⁺]_i upon mechanical stretching (1), separate experiments were conducted to determine the effect of OM extract on mechanosensitive calcium transients and to explore mechanisms accounting for these effects.

Force generation at the HGF plasma membrane.

The electromagnetic-bead model generates tensile forces on the HGF plasma membrane and was used as described by Glogauer et al. (18). Briefly, a suspension of ferric oxide microparticles (hereafter referred to as beads; Fe₃O₄; Aldrich Chemical Co., Milwaukee, Wis.) coated with collagen was added at 10 μl/ml for a 10-min incubation period followed by three washes to remove unbound beads, achieving a high degree of membrane surface area coverage (18). The collagen-coated beads attach through α₂β₁ integrin receptors. A single, 1-s magnetic field was applied by an electromagnet which induces a strong force in the x-y plane of the cell. The force was applied parallel to the dorsal surface of the cell and exerted only a minimal vectorial component orthogonal to the dorsal surface. The voltage and current applied to the electromagnet were calibrated so that the applied force was 2 N/cm².

An image analyzer (Bioquant, Nashville, Tenn.) was used as previously described (18) to determine if incubation with the OM extract caused bead detachment from the cell membrane. Briefly, cells were incubated with beads for 10 min. Unbound beads were removed, and cells were incubated with the OM extract for various time periods. Cells were exposed to a 0.1 pN/μm² force to apply tension to the bead-membrane complex. The total cell area covered with beads was divided by total cell area and expressed as a percentage. Differences of total cell area covered by beads in experimental and control wells were computed.

Mechanism of T. denticola OM perturbation of calcium flux.

To examine the effect of T. denticola OM on internal calcium stores, two agonists of internal calcium release in resting cells were used. ATP (100 μM; Sigma) and thapsigargin (1 μM; Sigma) were added to resting HGF in calcium-free buffer with EGTA (5 mM; Sigma). Cells that were preincubated with OM for 40 min at 20°C were compared with untreated cells. Influx of extracellular calcium following depletion of intracellular stores was also studied. Control and OM extract-treated (45 min) HGF were exposed to 1 μM thapsigargin in 1 mM EGTA buffer for 30 min to deplete internal Ca²⁺ followed by 1 mM EGTA for 10 min to chelate residual extracellular Ca²⁺. Then 1 mM CaCl₂ was added, and [Ca²⁺]_i transients due to calcium influx were measured.

Specific inhibitors were used to identify the pathways responsible for the changes observed in [Ca²⁺]_i transients in response to membrane stretching in the presence of OM (incubated for 40 min). Gadolinium chloride, a putative stretch-activated channel blocker (41), (1 mM; 60-s incubation prior to force application; Aldrich) was used to assess the possible potentiation of blockade by OM on mechanosensitive channels. Similarly, to study the possible potentiation of blockade of internal calcium release after membrane stretching, cells were incubated with thapsigargin (Sigma) 30 min prior to force application. This protocol ensures complete block of the thapsigargin-sensitive intracellular stores (8). Cells treated with either (i) gadolinium chloride and OM extract or (ii) thapsigargin and OM extract (for 40 min) were compared with cells that were treated with either gadolinium chloride or thapsigargin alone. Cellular membrane integrity of OM extract-treated and control HGF was compared by measuring intracellular Fura-2 fluorescence at the isosbestic point (356 nm), as Fura-2’s fluorescence at this wavelength is independent of [Ca²⁺]_i. In addition, changes in [Ca²⁺]_i in response to ionomycin (3 μM) in control and OM-pretreated HGF were compared to determine if they would clear excess intracellular Ca²⁺ equivalently.

A Fura-2 quenching experiment was conducted to determine whether the observed OM extract’s suppression of calcium transients in response to mechanical stretching was due to inhibition of mechanosensitive cation-permeable channels. Control and OM extract-treated (45 min) cells were incubated in 1 mM MnCl₂ in the usual Ca²⁺-containing buffer and then subjected to force-induced plasma membrane stretching. The effect of Mn²⁺ influx on the quenching of Fura-2 fluorescence emission was determined.

Data analysis.

Although the amplitude of the induced calcium pulses varied among cells (around 50 to 500 nM of [Ca²⁺]_i above baseline), the amplitude was reasonably constant for individual cells responding to repeated force application when a recovery period of more than 15 min was allowed. Means and standard errors of the means (SE) were calculated for the variables associated with [Ca²⁺]_i measurements. Calcium measurements were restricted to (i) baseline [Ca²⁺]_i, (ii) percentage change of the transient [Ca²⁺]_i above baseline, (iii) net change in [Ca²⁺]_i, (iv) time to peak [Ca²⁺]_i, and (v) area under the peak of [Ca²⁺]_i transient. For experiments with resting cells, calcium spikes were defined as calcium transients that were >10 nM above baseline values and that returned to baseline values. The frequency of calcium spiking was expressed as spikes/minute. Paired Student’s t tests were performed to compare cells before and after treatment with OM extract.

RESULTS

Calcium signalling in resting cells.

In HGF, baseline [Ca²⁺]_i was constant over the time course of the experiment (1 to 70 min) and OM treatment did not affect baseline [Ca²⁺]_i over this duration in resting cells not stimulated by mechanical stretching (Table 1). Unstimulated cells exhibited spontaneous, periodic calcium transients in calcium buffer. These calcium oscillations exhibited spiking characteristics as they returned to baseline levels between events. Over a 90-min period of recording, all untreated control cells showed some degree of calcium spiking, but the oscillations varied considerably in magnitude (10 to 200 nM above baseline). Exposure of cells to OM extract caused a significant increase in the frequency of spiking only for the first 20 to 30 min, followed by a significant and sustained reduction in frequency and magnitude thereafter (Fig. 1). Measurement of Fura-2 fluorescence at the isosbestic point (356 nm) showed no reduction of photon counts (i.e., dye loss) after 30 min of incubation with the extract, indicating that membrane integrity was not altered by the OM extract. Cells preincubated for 30 min with OM extract responded equally to ionomycin (3 μM) as untreated controls (Fig. 2). In both controls and OM extract-treated cells, [Ca²⁺]_i returned close to basal levels within 4 min, indicating that altered patterns of [Ca²⁺]_i caused by the OM extract were probably not due to an effect on the OM-pretreated cells’ clearance of excess intracellular Ca²⁺.

TABLE 1.

Changes in [Ca²⁺]_i due to the binding of collagen-coated beads to HGF

Condition	Mean baseline [Ca2+]i (nM)a ± SE
	1 minb	35 min	70 min
Without collagen-coated beads
Control	84.3 ± 2.2	84.5 ± 2.6	84.2 ± 2.9
OM treated	81.2 ± 1.9	80.5 ± 3.0	80.7 ± 2.9
With collagen-coated beads (no stretching)
Control	149.6 ± 5.5	158.9 ± 2.9	158.0 ± 2.2
OM treated	155.1 ± 6.6	159.0 ± 7.6	152.2 ± 10.6

^aMeasured by Fura-2 ratio fluorimetry. 

^bCells were treated with OM extract (0.1 concentration) for 30 min and incubated with or without collagen-coated ferric oxide beads for indicated times. Collagen-coated beads induced a relatively rapid increase of basal calcium, but OM extract had no effect on integrin-dependent increase of [Ca²⁺]_i in this case, when the plasma membranes were not stretched. 

FIG. 1.

T. denticola OM extract induces an early increase of spiking frequency that is followed by significant (P < 0.05) diminution at 70 min. Cells were preincubated with OM extract from T. denticola (▴) or untreated (control; ▪). Spikes were defined as calcium transients that were >10 nM above baseline values and that returned to baseline levels. Frequency was measured as the number of spikes for 5 cells per 20 min. Data are means ± SE, with each data point representing an average of 40 cells in a 20-min time interval in eight independent experiments.

FIG. 2.

Increase of [Ca²⁺]_i in single cells that were untreated (control) or treated with OM extract for 30 min and then incubated with ionomycin (3 μM; arrow). The baselines of the two traces were at the same level (100 nM), but the OM-treated trace has been offset vertically by 50 nM merely to facilitate visualization of the two responses. Note that both control and OM-treated cells respond robustly to ionomycin and show a sharp reduction of intracellular calcium within 50 s.

As the OM evidently affected alteration of spontaneous calcium oscillations and therefore the regulation of intracellular calcium pools, two agonists capable of generating internal calcium release were studied. ATP (100 μM) and thapsigargin (1 μM) were incubated with cells in calcium-free buffer containing 5 mM EGTA to ensure that the induced calcium transients were attributable solely to release from internal calcium stores. The regular buffer was exchanged for EGTA-containing buffer immediately before addition of the agonist. Cells treated with ATP demonstrated a 408% ± 79% increase of [Ca²⁺]_i above basal levels (84 ± 2 nM [n = 6]), but pretreatment of cells with OM extract (40 min before EGTA buffer exchange; incubation at 20°C) followed by stimulation with ATP showed a 40% reduction of the amplitude of calcium transients (260% ± 88% above baseline [n = 6]). In an identical experimental design, the mean calcium transient induced by thapsigargin was ∼30% less in the OM-treated cells ([Ca²⁺]_i; thapsigargin, 390% ± 48% above baseline [n = 4]; OM followed by thapsigargin, 268% ± 36% above baseline [n = 7]; P < 0.05).

When CaCl₂ was added to HGF that had been previously depleted of internal Ca²⁺ following incubation with thapsigargin and the extracellular chelator EGTA, [Ca²⁺]_i transients rose to a peak and then fell to a higher baseline (from approximately 35 to 100 nM Ca²⁺). The maximum influx of Ca²⁺, presumably through calcium release-activated channels, was significantly diminished in the OM extract-treated cells (238.1% ± 52.2% above baseline [n = 4]) compared with control cells (585.8% ± 96.5% above baseline [n = 5]; P = 0.02).

Ligation of β-integrins by extracellular matrix ligands can regulate [Ca²⁺]_i by integrin-gated calcium-permeable channels (34) and could thus be affected by components in the OM extract. Cells incubated with collagen-coated beads as ligand, but not stretched, exhibited a 78% increase in basal calcium (Table 1), and this increase was remarkably stable over 70 min of monitoring. Treatment of cells with OM extract exerted no significant effect on increased [Ca²⁺]_i induced by integrin ligation at all time periods.

Calcium response to membrane stretching.

After incubation of cells with collagen-coated ferric oxide beads, the membranes of HGF were stretched by application of a magnetic field, which is a reproducible method to cause an immediate increase in [Ca²⁺]_i. Cells exhibited an increase of 330% ± 63% of [Ca²⁺]_i above baseline, with a time to peak [Ca²⁺]_i in 23 ± 3 s. The Fura-2 fluorescence was susceptible to quenching by Mn²⁺ in both control and OM extract-treated cells, indicating unidirectional ion flow from the bathing medium to the inside of the cell. We estimated the total Ca²⁺ flux over the duration of the increased calcium permeability by measuring the area under the calcium pulse (10,000 ± 2,280 nM/s). These control values were used to determine if the OM extract modulated [Ca²⁺]_i responses. We found a rapid inhibition of stretch-induced [Ca²⁺]_i within 1 min after incubation with OM extract (Fig. 3). Over the time course, the calcium response was progressively reduced and reached a minimum for cells which had been incubated with OM extract for 35 min. In these cells, the [Ca²⁺]_i increase was markedly reduced to 142% ± 22% above baseline (P = 0.006; n = 10 independent experiments). The time to reach the peak was significantly faster than controls (14 ± 2 s), and the area under the pulse was reduced >65% (3,480 ± 2,950 nM/s).

FIG. 3.

T. denticola OM extract reduces the amplitude of stretch-induced calcium transients. Cells were incubated with collagen-coated ferric oxide beads and were either untreated (control) or treated with OM extract for 1 min. [Ca²⁺]_i was measured in cells that were mechanically stretched by application of magnetic fields.

As repeated stretching of the control cells showed that the responses were relatively constant over time when a recovery period of >15 min was allowed between stretches (Fig. 4), it was evident that the OM extract-induced inhibition was not simply an artifact due to desensitization of calcium-permeable channels because of repeated stretching. Treatment of cells with increasing concentrations of the OM extract preparation showed a dose-response relationship (Fig. 5).

FIG. 4.

T. denticola OM extract inhibits stretch-induced calcium responses within 20 min after incubation with OM extract, and the inhibition persists for up to 70 min (P < 0.05). Cells were treated as described for Fig. 3. Measurements of peak calcium increases induced by mechanical stretching were made at the indicated times, and the percent change compared to the increase obtained at time zero was computed. The responses of control cells did not change significantly over 70 min. Data are shown as mean ± SE percent of control values (n = 10 cells per data point).

FIG. 5.

T. denticola OM extract causes a dose-dependent reduction of the amplitude of stretch-induced calcium transients in fibroblasts. Cells were loaded with ferric oxide beads and incubated with OM extract at the indicated dilutions for 30 min. Data are means ± SE percent change of peak calcium transient compared with untreated controls (n = 10 cells per data point).

It is conceivable that the magnetic force may have removed beads from cells that were treated with OM extract and that the reduced calcium transients may be an experimental artifact caused by lower numbers of attached beads. After application of a 0.1 pN/μm² force, ∼70% of cell surface area was coated with collagen-coated beads. Cells were incubated with OM extract to determine whether the proteolytic activity of the extract was capable of removing the collagen-coated beads. There was no significant loss of beads from the membranes of cells preincubated with OM extract for up to 1 h and exposed to a constant magnetic force for 10 min (range in means from 60% ± 12% to 74% ± 11% cell surface area covered with beads at four time points over 1 h). Thus, the OM extract did not inhibit force-induced calcium transients simply because of bead removal.

Mechanism of OM suppression of force-induced [Ca²⁺]_i response.

For cells incubated in nominally Ca²⁺-free buffer containing 5 mM EGTA, there was no increase of [Ca²⁺]_i after force application in either controls or in cells preincubated with OM extract for 30 min (Fig. 6). We next compared the inhibition of stretch-induced calcium responses in Ca²⁺-containing buffer after treatment with OM extract (30 min) or gadolinium chloride (GdCl₃; 30 min), a putative stretch-activated channel blocker (41). Cells treated with GdCl₃ alone showed a 40% reduction of [Ca²⁺]_i in response to force compared with untreated controls, while cells treated with OM extract showed a 60% reduction of [Ca²⁺]_i (Fig. 6). Treatment with both GdCl₃ and OM extract produced only a small additional inhibition compared with GdCl₃ alone. We examined whether the OM extract inhibited calcium flux after stretch because of depletion of intracellular calcium stores. Cells treated with thapsigargin alone exhibited a 60% reduction of the stretch-induced [Ca²⁺]_i increase (Fig. 6), but there was no significant difference in cells treated with both thapsigargin and OM extract.

FIG. 6.

Histogram of stretch-induced calcium responses in cells that were treated with OM extract for 30 min (OM), 5 mM EGTA before stretching (EGTA), EGTA plus OM extract for 30 min, 1 mM gadolinium chloride (GdCl₃), gadolinium chloride plus OM extract for 30 min, 1 μM thapsigargin (Tg), or thapsigargin plus OM extract for 30 min. OM extract reduces the amplitude of stretch-induced calcium transients >3-fold. Data are means ± SE percent of control values (n = 20 per experimental condition).

Preliminary characterization of Ca²⁺ inhibitory activities of OM extract.

At the concentration used here, the OM extract was able to stimulate rearrangement of stress fibers in >60% HGF in 90 min, as determined by fluorescence microscopy using rhodamine-phalloidin (40). Heating the extract at 60°C for 30 min diminished the trypsin-like BAPNA-degrading activity to <10%, but the SAAPNA-degrading activity remained at 50% of values for unheated extract, indicating that heating did not fully destroy the chymotrypsin-like activity of OM extract whereas other enzymes were denatured. Heating the OM extract to 60°C reversed the OM extract’s usual suppression of force-induced [Ca²⁺]_i increases (Fig. 7). The chymotrypsin inhibitor PMSF inhibited SAAPNA-degrading activity, but the OM extract that was pretreated with PMSF still inhibited force-induced [Ca²⁺]_i transients to the same level as untreated OM extracts, indicating that chymotrypsin-like enzymes probably did not mediate the force-induced calcium inhibition. Boiling the extract for 10 min reduced both its peptidase and stress fiber rearrangement activities to negligible levels, and it eliminated the OM extract’s suppression of the force-induced calcium response (Fig. 7).

FIG. 7.

Histogram of repeated stretch-induced calcium responses of fibroblasts to OM extract treated by heating, boiling, or chymotrypsin inhibitor. Heat (60°C for 30 min) or boiling (10 min) abrogated the inhibitory effect of OM extract on stretch-induced calcium transients. The chymotrypsin inhibitor PMSF did not affect inhibition. OM extract was treated by heating, boiling, or PMSF before incubation with cells and stretching as described. Cells were measured before (control) and after incubation with OM extract for 1, 18, 35, 52, 69 min. Data are means ± SE percent of control values (N = 10 cells per bar).

DISCUSSION

In this investigation, which determined that the periodontal pathogen T. denticola can perturb calcium signalling in HGF, the OM extract was not globally cytotoxic but instead produced specific lesions of discrete calcium signalling pathways. Although the T. denticola OM extract contains a mixture of potentially cytotoxic molecules, our studies suggest that protein(s) from T. denticola interacted with calcium-permeable channels that are required for mechanotransduction and for replenishing intracellular calcium stores. These conclusions, however, depend on the demonstration that physiological cell functions were largely preserved and were not simply secondary to cell death. We found robust calcium responses to ionomycin, normal responses to α₂β₁ integrin-induced calcium flux (34), the ability to clear calcium after ionomycin treatment or membrane stretch, and the maintenance of the Fura-2 fluorescence at the isosbestic point. On the basis of these findings, we are confident that the cells were not undergoing cell death, plasma membrane leakage, or a general cytotoxic reaction in response to the OM extract. This interpretation is consistent with our previous findings that both fibroblasts and epithelial cells remaining attached to the substratum after challenge with whole T. denticola cells are viable as measured by lactate dehydrogenase release, propidium iodide staining, colony formation, and limiting dilution assays (3, 12).

Intracellular calcium.

Spontaneous calcium transients and oscillations have been observed in a variety of cell types (5, 6) including fibroblasts (19). We observed spontaneous, low-frequency spiking in the HGF studied here. The frequency of spontaneous cytoplasmic calcium ion transients was altered after incubation with OM extract. There was a dramatic but short-lived (30-min) increase of spiking in the early phase followed by a sustained decrease compared with control cells. A number of theories have been developed to account for the control of spontaneous and hormonally induced calcium spiking and oscillations (5, 36) in which IP₃ synthesis and IP₃-sensitive and IP₃-insensitive calcium stores have been invoked. In this case, progressive cytoplasmic depletion of calcium could have been caused in part by diminished IP₃, as we have found in concurrent work that T. denticola and the OM extract suppress inositol phosphate (IP) responses in HGF (40).

In HGF, the most likely mechanism for control of spontaneous oscillations involves a calcium feedback system that relies on leakage of calcium from the extracellular pool to replenish intracellular stores (1). We currently do not have definitive data to identify the regulatory factors that determine spiking. However, several findings suggest that the observed perturbations in spiking may be explained by dysfunction of calcium release-activated channels which in turn lead to depletion of intracellular stores, perhaps in conjunction with IP suppression. First, there was an initial increase in spiking frequency of calcium transients. This result would be anticipated if inadequate calcium was leaking from the extracellular bathing medium, and consequently increased frequency of quantal release from intracellular stores would be expected to occur (25), consistent with our observations. Second, the gradual elimination of spiking over time would be expected if the intracellular stores were gradually depleted or their release blocked. Third, OM extract substantially reduced thapsigargin and ATP-releasable calcium when cells were switched to EGTA buffer, indicating that the intracellular stores were indeed depleted. Fourth, replenishment of cytoplasmic Ca²⁺ following addition of extracellular CaCl₂ to Ca²⁺-depleted cells was diminished following OM extract treatment. Fifth, the time of elimination of calcium spiking coincided with the time of maximum suppression of the stretch-induced [Ca²⁺]_i, a phenomenon that is dependent in part on release from intracellular stores (18). Sixth, this period in the time course also coincides with suppression of IP responses (40), which is significant in physiological mobilization of intracellular Ca²⁺ and possibly regulation of Ca²⁺-permeable membrane channels. Collectively these data are consistent with a model in which proteins from T. denticola interact with calcium release-activated channels and block their activity. However, the nature of these channels in HGF has not been clearly established.

Mechanosensitive calcium flux.

The previous development and characterization of the collagen-coated magnetic bead model for application of controlled forces to collagen receptors and the cytoskeleton (18) provided us with a novel strategy to examine the effects of the T. denticola OM extract on highly predictable Ca²⁺ responses to mechanotransduction in single cells. Application of force to HGF induces a reproducible increase in [Ca²⁺]_i which is a result of an influx of Ca²⁺ through putative stretch-activated channels (35) and subsequent calcium release from internal stores (18). The unidirectional flow of cations through putative cation-permeable mechanosensitive channels was demonstrated here by manganese quenching of Fura-2 fluorescence. Although the fibroblasts studied here are heterogeneous, single-cell analyses showed that there was a remarkable consistency of the calcium responses to both stretch and attenuation of these responses by the OM extract. Indeed, a 1-min preincubation of cells with the OM extract prior to force application induced a dramatic reduction in the amplitude of the [Ca²⁺]_i response to force. This observation points to the acute action of the OM extract on mechanosensitive calcium signalling. In 30-min experiments, the inhibition by the OM extract was at least 20% greater than the inhibition produced by gadolinium chloride, a low-potency mechanosensitive ion channel blocker that is not wholly specific (31, 35). However, this finding cannot be interpreted as indicating that the OM extract directly affects mechanosensitive ion-permeable channels. The minor increase in inhibition when GdCl₃ and OM extract were combined probably derived from the OM extract’s effect on Ca²⁺ release from internal stores. As the OM extract is known to contain proteolytic molecules with peptidase activities similar to chymotrypsin and trypsin, we verified that the OM-treated collagen-coated beads were not removed by the force application, an important point confirming that the suppressed calcium response was not merely due to bead loss from proteolysis. These findings are consistent with T. denticola’s relatively weak collagenolytic activity and the resistance of native integrins to trypsin.

Although the chymotrypsin-like enzyme of T. denticola is considered a major potential virulence factor (37) and is implicated in HGF detachment from the extracellular matrix (3, 37), it is not apparently the OM component that caused the suppression of mechanosensitive calcium signalling. The susceptibility to 60°C heat and the insensitivity to PMSF pretreatment both argue against a role for chymotrypsin-like enzymes, as these characteristics are inconsistent with SAAPNA- and fibronectin-degrading activities of T. denticola (14). We have also generated preliminary data indicating that purified native 95-kDa chymotrypsin-like protease from T. denticola fails to suppress the calcium responses to membrane stretching (data not shown). Further, lipopolysaccharide-like molecules were evidently not important in that the suppression of calcium responses by the OM extract was eliminated by boiling. We conclude that other proteins, perhaps a class of OM proteins with toxigenic effects, may interact with a family of cation channels that are responsible for some of the Ca²⁺ permeability following applied physical force. One candidate could be the major surface protein antigen (Msp) of T. denticola, which is known to bind some extracellular matrix proteins (15). Partially purified Msp lacking chymotrypsin activity depolarizes and increases conductance of cultured HeLa cells, possibly by the integration of this protein into the membrane and the creation of short-lived, high-conductance, nonspecific ion channels (22).

In experiments involving treatments with both thapsigargin and OM extract, we noted no additional inhibition of stretch-induced calcium response with thapsigargin. A significant part of the whole cell calcium response to membrane stretch is attributable to calcium-induced calcium release from intracellular stores that can be inhibited by thapsigargin (18). The failure to find additional inhibition with thapsigargin is consistent with the notion that the OM extract affects not only IP-dependent Ca²⁺ mobilization but also the blockade of calcium release-activated channels that “leak” to refill intracellular calcium stores (6). This conjecture is also consistent with the calcium oscillation data discussed above.

Bacterial exploitation of host cell signalling pathways has led to several reports of increased [Ca²⁺]_i in resting cells responding to bacterial contact. These studies have concentrated mostly on invasive or diarrheagenic enteric pathogens which induce the accumulation of actin filaments adjacent to adherent or invading bacteria (4, 13, 16, 27). In contrast, we have found profound diminution of both spontaneous and induced calcium responses of fibroblasts to extracts of T. denticola, an indigenous pathogen which does not generally invade cells and which causes a reduction in actin filaments in host cells (3, 12, 37, 40). The OM extract from T. denticola evidently contains non-chymotrypsin-like proteins that can directly interfere with calcium signalling in mechanotransduction and in the regulation of intracellular calcium (Fig. 8).

FIG. 8.

Putative mechanisms for T. denticola OM perturbation of intracellular calcium ion fluxes in HGF. Channels and sources marked by X’s and thin arrows, representing Ca²⁺ release from internal stores and Ca²⁺ flow through calcium release-activated calcium (CRAC) channels, are blocked by nonproteolytic OM protein(s). Inhibition of mobilization of intracellular Ca²⁺ may be, in part, secondary to diminished IP responses (40) and calcium-activated Ca²⁺ release. In their simplest interpretation, the findings suggest a hypothesis that OM proteins interfere rapidly and directly with CRAC channels and perhaps with receptors for some agonists of the IP pathway at the external interface of the plasma membrane prior to and concomitant with disruption of F-actin at the ventral interface with the substratum (40) and cytoskeletal proteins which complex with integrins upon mechanical stretching by magnetic forces acting on the bound, collagen-coated ferric oxide beads (Bead) (17). T. denticola OM has no effect on integrin-gated (Ligand-gated) Ca²⁺-permeable channels, and the data suggest little direct effect on stretch-activated ion-permeable channels (thick arrows). Proteolytic enzymes in the OM have little if any immediate effect on intracellular Ca²⁺ and IPs but may indirectly affect actin subsequently by degrading extracellular matrix proteins like endogenous fibronectin (Fn) and components of the substratum (ECM [extracellular matrix]).

Therefore, future studies should aim to isolate and determine the impact of specific T. denticola surface proteins on virulence by virtue of their specific inhibition of calcium transients and related signalling pathways. Conceivably, such proteins may bind to or interfere with a family of calcium-permeable ion channels, including calcium release-activated calcium channels, and thereby perturb calcium homeostasis in T. denticola-affected cells. These perturbations would likely impact on actin-dependent functions such as cellular locomotion (2) and phagocytosis (23) which are crucial for physiological wound remodelling in response to chronic infections like periodontitis.