0099-2399/91/1700-0021/$02.00/0 Printed in U.S.A.
JOURNAL OF ENDODONTICS Copyright 9 1990 by The American Association of Endodontists
VOL. 17, NO. 1, JANUARY 1991
Modeling Bacterial Damage to Pulpal Cells In Vitro C. T. Hanks, DDS, PhD, S. A. Syed, PhD, R. G. Craig, PhD, J. M. Hartrick, BS, and T. E. Van Dyke, DDS, PhD
One objective of the present study was to determine whether fragmented bacterial components (centrifugal fractions) would diffuse across dentin disks in the "in vitro pulp chamber (IVPC). Treponema denlicola and Fusobacterium nucleatum were chosen for this study because as plaque-associated organisms they have been found in the floor of cavity preparations beneath composites (9), are reasonably easy to homogenize, and contain substances which are toxic to eukaryotic cells. F. nucleatum contains lipopolysaccharides (LPS) (10) and releases at least two metabolites, butyric acid and ammonia, which may reach toxic concentrations (11). Strains of T. denticola also contains LPS (12) as well as proteases which hydrolyze trypsin, chymotrypsin, and collagenase substrates (13, 14). A second objective of the present study was to compare the cytotoxic effects of bacterial fractions on L929 cells in the IVPC with the effects on the same cells in monolayer cultures. Chemotaxis and endotoxin assays were run to confirm these biological activities in the bacterial fractions.
There is increasing evidence that access to patent dentinal tubules by bacteria and their products rather than trauma from restorative materials is responsible for subsequent pulpitides. The purpose of this study was to compare the relative cytotoxicity of centrifugal fractions of two bacteria, Fusobacterium nucleatum and Treponema denticola, on L929 cells in monolayer cultures and in the "in vitro pulp chamber." Neutrophilic chemotaxis assays and Limmulus assays were performed to verify biological activity of the various fractions of these bacteria. It was found that T. denticola inhibits new protein synthesis in cultured cells to a much greater extent than F. nucleatum, but that only F. nucleatum fractions are chemoattractive for human neutrophils in the absence of serum. While the chemical nature and molecular weights of the "toxic" materials were not determined, it appeared that eukaryotic protein synthesis inhibition caused by the 7". denticola pellet fraction in the in vitro pulp chamber was at least 1000 times less than that caused by the same concentrations in monolayer cultures.
MATERIALS AND METHODS
Preparation of Bacterial Samples The Gram-negative anerobic microorganism, F. nucleatum (strain FN1), was a clinical isolate from human subgingival plaque of a periodontitis patient and was grown in Schaedler broth (Oxoid Ltd., Basingstoke, UK) under anerobic conditions. T. denticola, a small Gram-negative spirochete (strain no. 35405; American Type Culture Collection, Rockville, MD), was grown anerobically in a modified spirochete broth containing 5% heat-inactivated rabbit serum and yeast extract. Each bacterial species was submitted to the same centrifugal fractionation procedure after the initial growth period (Fig. 1). Sonication of the bacterial cell suspension was performed with a Sonifier Cell Disruptor (model W185D; Heat Systems-Ultrasonics, Inc., Plainview, NY) in an ice bath to control heating. Intermittent checking with phase contrast microscopy confirmed homogenization. After centrifugation, there were three fractions of each organism: homogenate (HOM), 16,000 x g cell wall pellet (PEL) and supernatant (SUP). All subsequent procedures for F. nucleatum utilized 0.05 mol per 1 Tris of HC1 buffer (pH 7.0). For T. denticola, 0.1 mol per I of Sorensen's phosphate buffer (pH 7.0) was used. Dialysis at 4~ was performed against the respective buffers in tubing (Spectra/POR; Spectrum Medical Industries, Inc., Los Angeles, CA) which had a molecular mass cutoff porosity of 3500 daltons. When ready for testing, the lyophilized bacteria were reconstituted to the required protein con-
It was suggested by Zander (1) among others that access to patent dentinal tubules by bacteria and their products, rather than irritation by restorative materials, was responsible for subsequent pulpal reactions. Several lines of evidence which implicate bacteria include a lack of significant inflammation under numerous restorative materials when the cavities or restorations are surface sealed with a agent such as zinc oxideeugenol (2) or when restorative materials are placed in cavity preparations in germ-free animals (3). When placed on exposed dentin of monkey teeth, oral bacteria and plaque extracts including lipopolysaccharides can cause necrosis and an acute inflammatory infiltrate (4). Mj6r et al. (5) pointed out that tissue culture assays are more "sensitive" than usage tests in animal teeth so that the same materials will appear more toxic to cultured cells than to pulpal cells when the material is placed in a cavity preparation. Previous studies in this laboratory (6, 7) and others (8) have demonstrated reduced cytotoxicity due to composites and other dental materials in culture systems where dentin is interspersed between the material and the cell test system as compared with systems in which monolayer cultures are used.
21
22
Hanks et al.
Journal of Endodontics
centrations. For the H O M and PEL fractions, much of the material remained suspended.
Cytotoxicity Tests in Monolayer Cultures and IVPC The cell test system was a culture of L929 cells grown either in monolayer in 96-well dishes (Costar 3596; wells 6.5 mm in diameter) or on polycarbonate (PC) filters directly subjacent to dentin disks (Fig., 2) in the IVPC (6, 7). In the 96-well dishes, cells were plated at 80,000 cells per well (0.33-cm 2 area) and maintained in Ham's F-12 medium with glutamine plus 10% donor calf serum (Flow Laboratories, McLean, VA) for 24 h. After washing the monolayer with a balanced salt solution consisting of 145 mmol per l of NaCI, 5 mmol/l of KC1, 1 mmol/1 of MgSO4, 10 m m o l / l of Hepes, and 10 m m o l / 1 of glucose (pH 7.4) at 37~ serial dilutions of the bacterial fractions in complete medium plus 5 uCi/ml of [3H]leucine (60 Ci/mmol: ICN Radiochemicals, Irvine, CA) were added to the monolayer. Four wells were used for each dilution of each bacterial fraction. After 24 h, the medium was again suctioned off, the monolayer was washed with balanced salt solution, and then lysed with a Triton X-100/NaOH solution. FUSOBACTERIUM NUCLEATUM (FN) TREPONEMA DENTICOLA (TO)
g, 4 ~
4000 30mn.
CellPellet Sonicate to disrupt ce
s
i
Homogenate
16,000g, 4~ 1 hour dialysis
I
lyophilizalion
I
' pellet
i supernate
I
dia ysis
dialysis
I
lyophilization
I
,I I
Aliquots were counted by liquid scintillation and were analyzed for protein content (BCA Protein Assay Reagent: Pierce Chemical Co., Rockford, IL). The PC filters (13 m m diameter, 3-um pore size Poretics Corp., givermore, CA) were placed subjacent to the dentin disks in the IVPC apparatus and the PC filters and dentin disks were treated with 70% ethanol, several changes of double distilled water, and then medium, all under vacuum, to hydrate the system and to remove air bubbles. Twelve dentin disks were used for each bacterial species in the IVPC. The dentin disks had been cut and treated with EDTA as described previously (10). The average (and SD) dentin thickness for tests with T. denticola was 0.47 m m (0.07) with an average (and SD) hydraulic conductance (Lp) of 0.005 (0.001) #1 cm -2 H_,O-~ min -1 (10). The average (and SD)dentin thickness for tests for F. nucleatum was 0.42 m m (0.08) with an average (and SD) Lp of 0.005 (0.002). The IVPC was assembled so that the dentin disk and PC filter were between a top reservoir and a lower culture chamber (Fig. 2). Cells were then plated on the underside of the PC filters at a concentration of 40,000 cells per well (0.1559 cm 2 in 40 ul of medium) for each chamber and allowed to adhere to the filter surface for 1 h. After plating, more medium was added to each chamber and the cells were allowed to grow for 2 days. At the time of the test for toxicity of the bacterial fractions, the medium in the IVPC lower chambers which fed the cells also contained 5 ~Ci per ml of [-~H]leucine. The cells on the PC filters were continuously labeled for another 24 h in the same manner as the monolayer cultures. The IVPC then was disassembled and cells on each PC filter were lysed and analyzed. The dentin disks were washed overnight under hydraulic pressure and retreated with ethanol under vacuum and then rehydrated in medium under vacuum for the next bacterial fraction. The statistical methods used in these studies were Student's t test for comparison of samples from two bacterial fractions and one-way analysis of variance for comparison of samples of more than two fractions.
lyophilization
I Chemotaxis Studies
FIG 1. Scheme for preparation of bacterial fractions by centrifugation and dialysis.
FIG 2. One chamber of 12 chamber "in vitro pulp chamber" utilized for testing bacterial fractions. L929 cells are plated on PC filters on the undersurface of dentin disks (D). Culture medium fills the chamber (M) beneath the dentin disk.
Each bacterial fraction sample was diluted with Gey's gelatin veronal buffer (GGVB) to give a series of dilutions from 0.01 to 100 #g per ml of the original suspension or solution (15). The positive control for these tests was the chemotactic substance, formyl - L- methionyl - L- leucyl - L- phenylalanine (Sigma Chemical Co., St. Louis, MO), at a concentration of 2 • 10 -8 tool per 1 in GGVB. The negative control substance was 2 x 10 -8 mol per I o f phosphate-buffered saline in GGVB. The human peripheral neutrophils were taken from healthy patients by venipuncture and separated with a discontinuous gradient of two Ficoll-Hypaque mixtures (Mono-Poly resolving medium and Histopaque 1077) as described previously (16). Then, the various dilutions of the bacterial fractions as well as the neutrophil suspensions, in the absence of serum, were placed in modified Boyden chambers to test for chemotaxis (15). The chambers were incubated at 37~ in a humid 5% CO_, incubator for 2 h, and then the 5-um pore size cellulose nitrate filters were removed, fixed with methanol and 4% formalin, stained with Mayer's hematoxylin, and mounted with glass coverslips for analysis.
Vol. 17, No. 1, January 1991
Modeling Bacterial Damage
Endotoxin Activity
The HOM, PEL, and SUP fractions of both bacteria were also assayed for endotoxin activity using the Limutus Amebocyte Lysate assay (E-Toxate; Sigma) in the absence of serum. RESULTS High concentrations of all three fractions of F. nucleatum showed significant (p < 0.01) toxicity for mammalian protein synthesis in monolayer cultures. In general, this toxicity was evident at concentrations in the range of 100 ul per ml of bacterial protein. There appeared to be minor peaks of increased protein synthesis at 1 #g per ml of H O M and at 10 #g per ml of PEL and SUP fractions. The yield of bacterial protein for T. denticola cultures was about one-tenth that of F. nucleatum, but the toxic effects of the centrifugal fractions were considerably greater. Although the HOM fraction had little effect on L929 cell protein synthesis in monolayer cultures at bacterial concentrations of less than about 1 ug per ml, higher bacterial concentrations caused increasing inhibition of L929 protein synthesis (Fig. 3). The effect plateaued at about 13% of the untreated control values. The T. denticola PEL and SUP fractions appeared to be even more toxic, so that 0.25 to 2.5 ~g per ml of bacterial protein reduced m a m m a l i a n protein synthesis to about 50 and 70%, respectively, of the control values, and 25 to 250 ug per ml reduced mammalian protein synthesis to 2% of control values for both fractions. The reductions of L929 protein synthesis by bacterial fractions above 1 ug per ml were all statistically significant (p < 0.01). For Figs. 3 and 4, the lines for each fraction simply connect mean values for each concentration and are not meant to indicate exact effects of bacterial fraction concentrations. The PEL and SUP fractions of each bacterium were placed in the reservoirs o f the IVPC, with confluent cultures of L929 cells growing on PC filters directly beneath the dentin disks. To simplify comparisons, only 1 and 10 ug per ml concentrations of the SUP and PEL fractions of each bacterium were
23
placed as 100-~1 aliquots in each of four reservoirs of the IVPC above the dentin disks (Fig. 5). For both SUP and PEL fractions of T. denticola and E. nucleatum, the 10 ug per ml depressed L929 protein synthesis more than did 1 #g per mL However, because of the large variability of readings from well to well, these differences were not statistically significant by the t test. Comparison of the monolayer results and the IVPC results for the T. denticola PEL fraction exemplifies the effect of the dentin barrier on "toxic molecules" derived from bacteria (Fig. 6). The curve demonstrates the rapid reduction of L929 protein synthesis in monolayer cuiture as the protein concentration of the PEL fraction increased. The bars represent 1 and 10 ug per ml concentrations of the PEL fraction placed in the IVPC reservoirs and the resultant effects upon new protein synthesis in L929 cells. At concentrations of 1 and 10 ug per ml, the PEI. fraction in the IVPC reservoir resulted in depression of 85% to 73%, respectively, of the negative control protein synthesis values, rather than 43 % to 15 %, respectively, as in the monolayer. Chemotaxis studies with human neutrophils gave a different picture. All three fractions of F. nucleatum showed much greater chemoattraction for neutrophils than the negative PBS
.,m~ ~~176 "%
o
SO
8
~t 9 FNPs 9 FNSOP
0
OOOl
~91
O~
1
l
10
100
I00o
LOG BACT. PROT. CONC. (ug[r~)
F~G 4 Effects of three centrifugal fractions of F. nucleatum (F.N,) on new protein synthesis by monolayers of L939 cells over a 24-h period.
120
9 i U(;M(
T
~00
...........
80
140
9- ~ ~
9
10 U(~ ML
t~'O
%'%%%%%%
IUU
'II~
80 6O
60 40
40 20
;20
Q o
D
DDDI OC*~
01
1
l
lO
1oo lOO0
ID sop
Iiii T~ , ~
~N SUP
BACTERIAL FRACTION
LOG BACT. PROT. CONC. (ug/ml)
FIG 3. Effects of three centrifugal fractions of T. denticola (TD) on new protein synthesis by monolayers of L929 cells over a 24-h period.
FIG 5. Effects of 1 ~g per ml and 10/~g per ml of two fractions (SUP and PEL) of two bacteria (TD and FN) on new protein synthesis by L929 cells in the IVPC over a 24-h period.
24
Hanks et al.
Journal of Endodontics
controls, in the absence of serum in the Boyden chamber (Fig. 7). The greatest response was with the post-16,000 x g PEL fraction at 100 #g per ml, followed by the HOM fraction at the same concentration. The highest bacterial protein concentration tested for any fraction was 100 ag per ml, and this concentration of F. nucleatum PEL fraction gave a neutrophilic chemotactic response which was statistically no different from that of the positive F M L P control. On the other hand, none of the T. denticola fractions under the same conditions, even at 100 ug per ml concentration, showed any tendency for neutrophilic chemoattraction (Fig. 8). The Limulus Amebocyte Lysate assay on selected concentrations of the bacterial fractions showed that both bacteria contained endotoxin. T. denticola H O M (10 ug per ml), SUP (10 ug per ml), and PEL (15 ~g per ml) were the lowest concentrations of each fraction to show gelation. For F. nucleatum, PEL (1 ug per ml) also caused gelation, while SUP and HOM, even at 10 #g per ml did not.
DISCUSSION In this article, we have demonstrated that the centrifugal fractions derived from 7". denticola were more cytotoxic than similar concentrations of fractions of F. mtcleatum to monolayers of L929 cells in terms of protein synthesized in a 24-h period. When placed in the IVPC, diffusion through 0.4- to 0.5-mm dentin substantially reduced the toxic effects of most of these fractions.
60 Z ~= O. Z I-'~ n-
O
.~
5e 40
[] 9
30
[]
[]
FNPEL FN-HOM FN-SUP PBS FMLP
1
20
Ii m ,B 0.01
0.1
I
10
O0
2x10^-8
CONC. (ug/ml) F~G 6. Effects of 1 #g per ml (&) and 10 #g per ml (O) of T. dentico/a PEL fraction in IVPC (bars) superimposed on graph of various concentrations of the same fraction in L929 cell monolayer cultures, All results are in terms of percentage of negative control values,
80
20
o
ooo1 OOl Ol
1
lo
loo
looo
LOG BACT, PROT. CONG.(ug/mJ)
FIG 7. Comparison of effects of various concentrations of three fractions of F. nuc/eatum to positive (FMLP) and negative (PBS) controls in tests of human neutrophilic chemotaxis in Boyden chambers,
6O Z
50 D. (.'3 40 Z
9 TD-HOM [ ] TD SUP m PBS [ ] FMLP
,~ 3o r r
20
001
0 1
1
10
100
2 x 10^'8
CONC. (ug/ml) FiG 8, Comparison of effects of various concentrations of two fractions of T. dentico/a to positive (FMLP) and negative (PBS) controls in tests of human neutrophilic chemotaxis in Boyden chambers.
In order to investigate possible diffusion of microbial products through dentinal tubules, bacteria were homogenized and fractionated. A mechanical method (sonication) of homogenization as well as lyophilization were the only methods used to avoid degradation of the biological activity of the bacterial fractions. The IVPC is intended for testing dental restorative materials and bacterial fractions or both in a setting which simulates clinical conditions, e.g. presence of remaining dentin in the cavity preparation. Unless the nature and concentration of the "toxic" molecules are known, however, it is impossible to accurately determine the extent of the toxic effects. However, even with these limitations, it is possible to gain an impression of the net effects from studies with biological mixtures. For example, in Fig. 6 both 1 and 10 ~g per ml of the T. denlico]a PEL fraction are superimposed on the curve representing the effects of various concentrations of the same fraction in the monolayer cultures. If the bars representing these two concentrations are moved to the left until they intersect the curve derived from monolayer results, the net effect of the dentin barrier appears to be a dilution of the toxic effect of this fraction by more than 1000 times. There was more variability of values for the IVPC than for the monolayer cultures, even when the dentin disks were selected for minimal variation in thickness and hydraulic conductance. Prior to diffusion experiments, filtration rate and hydraulic conductance were stabilized empirically by treatment of dentin disks with 0.01 mol/1 of Tris-HCl buffered 70% ethanol under house vacuum (560 mm Hg) for 30 min. This treatment probably had several effects on the dentin disks such as replacement of air bubbles in the tubules, fixation of residual proteins in the dentinal tubules, and fixation of bacterial proteins, thus preventing infection. During the experiments, the variation in diffusion can be attributed to both variability in the size and distribution of dentinal tubules in the disks and the variation of the diffusion and adsorption of diverse molecules from each centrifugal fraction. Both F. nuclealum and T. denticola have extractable lipopolysaccharides (10, 12). In the present study, the Limulus assay confirmed the presence of these materials in all centrifugal fractions, Recently, Segal et al. (17) demonstrated that Actinobacillus actinomycetemcomitans LPS diffuses through 0.2-mm dentin disks, but also reduces the hydraulic conductance of these disks over a 24-h period. LPS from Escherichia coli requires serum for stimulation of D N A synthesis in cultured primary mouse embryonic fibroblasts (18). However, even in the presence of serum, Aleo et al. (19) reported that
Vol. 17, No. 1, January 1991
E. coli LPS does not stimulate DNA synthesis in L929 cells, but reduces cell viability (trypan blue incorporation) in culture over a 3-day period. Loss of membrane integrity might be explained by LPS activation of complement which generates both a complex of factors which lyse cell membranes and chemotatic factors for neutrophils. There were dramatic differences in the ability of centrifugal fractions of the two organisms to depress protein synthesis in vitro. Both the PEL and SUP fractions of T. denticola were very effective. While proteolytic activity has been described for 7". denticola (13, 14), it is not known if these enzymes are the molecules responsible for in vitro toxicity. Boehnringer et al. (20) have shown that a chromatographic fraction from T. denticola sonicates has the ability to suppress DNA, RNA, and protein synthesis in cultured human fibroblasts. This activity was associated with a single chromographic peak of about 50,000 daltons and was heat labile, suggesting it was protein and not lipopolysaccharide in nature. In the present study, only fractions o f F . nucleatztm caused a chemotactic response to peripheral human neutrophils in the absence of serum. The PEL fraction had been washed five times with Tris-HCl buffer so that serum complement appeared not to have played a role in the assay results. The PEL fraction of F. nucleatum, at a concentration of 100 ug per ml of bacterial protein, had chemoattractive activity similar to 2 x 10-8 M FMLP. For clinical relevance, further studies are required to compare this with the level of chemoattraction in the presence of serum and the number of organisms represented by these levels of endotoxin activity. This study has focused on mechanisms by which F. nzzcleaturn and T. denticola might cause pulpal damage if they or their products were introduced into the dentinal tubules either beneath restorations or through root dentin. These organisms appear to cause their principle tissue damage by at least two different methods. T. denticola releases toxic molecules, perhaps hydrolytic enzymes, which are capable of depressing macromolecular synthesis in fibroblastic cells. This biological activity is essentially neutralized by the diffusion gradient of 0.5 m m of remaining dentin in the bottom of the cavity preparation. Much less macromolecular synthetic depression is caused by direct contact with components (fractions) of F. nucleatum. However, more tissue injury may be caused by neutrophilic chemotaxis to F. nucleatum, or by complement activation due to LPS released by either organism. In order to assess more properly the contributions of these organisms to pulpal injury, however, it will be important in future studies to determine (a) the concentrations at which individual molecules have their biological effects, and (b) the concentrations at which these molecules reach the pulp across the gradient formed by the solution in the dentinal tubules.
Modeling Bacterial Damage
25
This work was supported by the National Institute of Dental Research Grant DE07987. The authors appreciate the assistance of M. L. Diehl and M K. Gardner (experiments).
Drs. Hanks, Syed, Craig, and Hartrick are affiliated with the School of Dentistry, University of Michigan, Ann Arbor, MI. Dr Van Dyke is affiliated with Department of Periodontology, Emory University, School of Dentistry, Atlanta, GA.
References 1. Zander HA. Pulp response to restorative materials. J Am Dent Assoc 1959;59:911-5. 2. Cox CF, Keall CL, Keall HJ, Ostro E, Bergenholtz G. Biocompatibility of surface-sealed dental materials against exposed pulps. J Prosthet Dent 1987;57:1-8. 3. Watts A. Bacterial contamination and the toxicity of silicate and zinc phosphate cements. Br Dent J 1979;146:7-13. 4. Bergenholtz G, Staffan A, Lindhe J. Experimental pulpitis in immunized monkeys. Scand J Dent Res 1977;85:396-406. 5. Mjor IA, Hensten-Pettersen A, Skogedal O. Biological evaluation of filling materials. A comparison of results using cell culture techniques, implantation tests and pulp studies. Int Dent J 1977;27:124-9. 6. Hanks CT, Craig RG, Diehl ML, Pashley DH. Cytotoxicity of dental composites and other materials in a new in vitro device. J Oral Pathol 1988;17:396-403. 7. Hanks CT, Diehl ML, Craig RG, Makinen P-L, Pashley DH. Characterization of the "in vitro pulp chamber" using the cytotoxicity of phenol. J Oral Pathol Med 1989;18:97-107. 8. Meryon SD. The influence of dentin on the in vitro cytotoxicity testing of dental resotrative materials. J Biomed Mat Res 1984;18:771-9. 9. Mejare B, Mejare I, Edwardsson S. Bacteria beneath composite restorations--a culturing and histobacteriological study. Acta Odontol Scand 1979;37:267-75. 10. Sveen K. Rabbit polymorphonuclear leukocyte migration in vivo in response to lipopolysaccharides from Bacteroides, Fusobacterium, and Veilonella. Acta Pathol Microbiol Immunol Scand [B] 1977;85:381-387. 11. Van Steenbergen TJM, Van Der Mispel LMS, Degraff J. Effects of ammonia and volatile fatty acids produced by oral bacteria on tissue culture cells. J Dent Res 1986;65:909-12. 12. Sela MN, Weinberg A, Borinksky R, Holt SC, Dishon T. Inhibition of superoxide production in human polymorphonuciear leukocytes by oral treponemal factors. Infect Immun 1988;56:589-94. 13. Makinen KK, Syed SA, Makinen P-L, Loesche WJ. Benzoylarginine peptidase and iminopeptidase profiles of Treponema denticola strains isolated from the human periodontal pocket. Curr Microbio11986;14:85-9. 14. Grenier D, Vitto V-J, McBride BC. Cellular location of a Treponema denticola chymotrypsinlike protease and importance of the protease in migration through the basement membrane. Infect Immun 1990;58:347-51. 15. Van Dyke TE, Reilly AA, Horoszewicz H, Gagliardi N, Genco RJ. A rapid, semi-automated procedure for the evaluation of leukocyte locomotion in the micropore filter assay. J Immunol Meth 1979;31:271-82. 16. Kalmar JR, Arnold RR, Warbington ML, Gardner MK. Superior leukocyte separation with a discontinuous one-step FicolI-Hypaque gradient for the isolation of human neutrophils. J Immunol Methods 1988;100:275-281. 17. Segal H, Stevens RH, Trowbddge H, Pash[ey DH. Permeability of human dentin to bacterial endotoxin. J Dent Res 1990;69:355. 18. Smith GL. Synergistic action of bacterial lipopolysaccharides in serumstimulated DNA synthesis in mouse embryo fibroblasts. Proc Soc Exp Biol Med 1976;153:187-192. 19. Aleo JJ, De Renzis FA, Farber PA, Varboncoeur AP. The presence and biologic activity of cementum-bound endotoxin. J Periodonto11974;45:672-5. 20. Boehringer H, Taichman NS, Shenker BJ. Suppression of fibroblast proliferation by oral spirochetes. Infect Immun 1984;45:155-9.
JOURNAL OF ENDODONTICS Copyright 9 1990 by The American Association of Endodontists
VOL. 17, NO. 1, JANUARY 1991
Modeling Bacterial Damage to Pulpal Cells In Vitro C. T. Hanks, DDS, PhD, S. A. Syed, PhD, R. G. Craig, PhD, J. M. Hartrick, BS, and T. E. Van Dyke, DDS, PhD
One objective of the present study was to determine whether fragmented bacterial components (centrifugal fractions) would diffuse across dentin disks in the "in vitro pulp chamber (IVPC). Treponema denlicola and Fusobacterium nucleatum were chosen for this study because as plaque-associated organisms they have been found in the floor of cavity preparations beneath composites (9), are reasonably easy to homogenize, and contain substances which are toxic to eukaryotic cells. F. nucleatum contains lipopolysaccharides (LPS) (10) and releases at least two metabolites, butyric acid and ammonia, which may reach toxic concentrations (11). Strains of T. denticola also contains LPS (12) as well as proteases which hydrolyze trypsin, chymotrypsin, and collagenase substrates (13, 14). A second objective of the present study was to compare the cytotoxic effects of bacterial fractions on L929 cells in the IVPC with the effects on the same cells in monolayer cultures. Chemotaxis and endotoxin assays were run to confirm these biological activities in the bacterial fractions.
There is increasing evidence that access to patent dentinal tubules by bacteria and their products rather than trauma from restorative materials is responsible for subsequent pulpitides. The purpose of this study was to compare the relative cytotoxicity of centrifugal fractions of two bacteria, Fusobacterium nucleatum and Treponema denticola, on L929 cells in monolayer cultures and in the "in vitro pulp chamber." Neutrophilic chemotaxis assays and Limmulus assays were performed to verify biological activity of the various fractions of these bacteria. It was found that T. denticola inhibits new protein synthesis in cultured cells to a much greater extent than F. nucleatum, but that only F. nucleatum fractions are chemoattractive for human neutrophils in the absence of serum. While the chemical nature and molecular weights of the "toxic" materials were not determined, it appeared that eukaryotic protein synthesis inhibition caused by the 7". denticola pellet fraction in the in vitro pulp chamber was at least 1000 times less than that caused by the same concentrations in monolayer cultures.
MATERIALS AND METHODS
Preparation of Bacterial Samples The Gram-negative anerobic microorganism, F. nucleatum (strain FN1), was a clinical isolate from human subgingival plaque of a periodontitis patient and was grown in Schaedler broth (Oxoid Ltd., Basingstoke, UK) under anerobic conditions. T. denticola, a small Gram-negative spirochete (strain no. 35405; American Type Culture Collection, Rockville, MD), was grown anerobically in a modified spirochete broth containing 5% heat-inactivated rabbit serum and yeast extract. Each bacterial species was submitted to the same centrifugal fractionation procedure after the initial growth period (Fig. 1). Sonication of the bacterial cell suspension was performed with a Sonifier Cell Disruptor (model W185D; Heat Systems-Ultrasonics, Inc., Plainview, NY) in an ice bath to control heating. Intermittent checking with phase contrast microscopy confirmed homogenization. After centrifugation, there were three fractions of each organism: homogenate (HOM), 16,000 x g cell wall pellet (PEL) and supernatant (SUP). All subsequent procedures for F. nucleatum utilized 0.05 mol per 1 Tris of HC1 buffer (pH 7.0). For T. denticola, 0.1 mol per I of Sorensen's phosphate buffer (pH 7.0) was used. Dialysis at 4~ was performed against the respective buffers in tubing (Spectra/POR; Spectrum Medical Industries, Inc., Los Angeles, CA) which had a molecular mass cutoff porosity of 3500 daltons. When ready for testing, the lyophilized bacteria were reconstituted to the required protein con-
It was suggested by Zander (1) among others that access to patent dentinal tubules by bacteria and their products, rather than irritation by restorative materials, was responsible for subsequent pulpal reactions. Several lines of evidence which implicate bacteria include a lack of significant inflammation under numerous restorative materials when the cavities or restorations are surface sealed with a agent such as zinc oxideeugenol (2) or when restorative materials are placed in cavity preparations in germ-free animals (3). When placed on exposed dentin of monkey teeth, oral bacteria and plaque extracts including lipopolysaccharides can cause necrosis and an acute inflammatory infiltrate (4). Mj6r et al. (5) pointed out that tissue culture assays are more "sensitive" than usage tests in animal teeth so that the same materials will appear more toxic to cultured cells than to pulpal cells when the material is placed in a cavity preparation. Previous studies in this laboratory (6, 7) and others (8) have demonstrated reduced cytotoxicity due to composites and other dental materials in culture systems where dentin is interspersed between the material and the cell test system as compared with systems in which monolayer cultures are used.
21
22
Hanks et al.
Journal of Endodontics
centrations. For the H O M and PEL fractions, much of the material remained suspended.
Cytotoxicity Tests in Monolayer Cultures and IVPC The cell test system was a culture of L929 cells grown either in monolayer in 96-well dishes (Costar 3596; wells 6.5 mm in diameter) or on polycarbonate (PC) filters directly subjacent to dentin disks (Fig., 2) in the IVPC (6, 7). In the 96-well dishes, cells were plated at 80,000 cells per well (0.33-cm 2 area) and maintained in Ham's F-12 medium with glutamine plus 10% donor calf serum (Flow Laboratories, McLean, VA) for 24 h. After washing the monolayer with a balanced salt solution consisting of 145 mmol per l of NaCI, 5 mmol/l of KC1, 1 mmol/1 of MgSO4, 10 m m o l / l of Hepes, and 10 m m o l / 1 of glucose (pH 7.4) at 37~ serial dilutions of the bacterial fractions in complete medium plus 5 uCi/ml of [3H]leucine (60 Ci/mmol: ICN Radiochemicals, Irvine, CA) were added to the monolayer. Four wells were used for each dilution of each bacterial fraction. After 24 h, the medium was again suctioned off, the monolayer was washed with balanced salt solution, and then lysed with a Triton X-100/NaOH solution. FUSOBACTERIUM NUCLEATUM (FN) TREPONEMA DENTICOLA (TO)
g, 4 ~
4000 30mn.
CellPellet Sonicate to disrupt ce
s
i
Homogenate
16,000g, 4~ 1 hour dialysis
I
lyophilizalion
I
' pellet
i supernate
I
dia ysis
dialysis
I
lyophilization
I
,I I
Aliquots were counted by liquid scintillation and were analyzed for protein content (BCA Protein Assay Reagent: Pierce Chemical Co., Rockford, IL). The PC filters (13 m m diameter, 3-um pore size Poretics Corp., givermore, CA) were placed subjacent to the dentin disks in the IVPC apparatus and the PC filters and dentin disks were treated with 70% ethanol, several changes of double distilled water, and then medium, all under vacuum, to hydrate the system and to remove air bubbles. Twelve dentin disks were used for each bacterial species in the IVPC. The dentin disks had been cut and treated with EDTA as described previously (10). The average (and SD) dentin thickness for tests with T. denticola was 0.47 m m (0.07) with an average (and SD) hydraulic conductance (Lp) of 0.005 (0.001) #1 cm -2 H_,O-~ min -1 (10). The average (and SD)dentin thickness for tests for F. nucleatum was 0.42 m m (0.08) with an average (and SD) Lp of 0.005 (0.002). The IVPC was assembled so that the dentin disk and PC filter were between a top reservoir and a lower culture chamber (Fig. 2). Cells were then plated on the underside of the PC filters at a concentration of 40,000 cells per well (0.1559 cm 2 in 40 ul of medium) for each chamber and allowed to adhere to the filter surface for 1 h. After plating, more medium was added to each chamber and the cells were allowed to grow for 2 days. At the time of the test for toxicity of the bacterial fractions, the medium in the IVPC lower chambers which fed the cells also contained 5 ~Ci per ml of [-~H]leucine. The cells on the PC filters were continuously labeled for another 24 h in the same manner as the monolayer cultures. The IVPC then was disassembled and cells on each PC filter were lysed and analyzed. The dentin disks were washed overnight under hydraulic pressure and retreated with ethanol under vacuum and then rehydrated in medium under vacuum for the next bacterial fraction. The statistical methods used in these studies were Student's t test for comparison of samples from two bacterial fractions and one-way analysis of variance for comparison of samples of more than two fractions.
lyophilization
I Chemotaxis Studies
FIG 1. Scheme for preparation of bacterial fractions by centrifugation and dialysis.
FIG 2. One chamber of 12 chamber "in vitro pulp chamber" utilized for testing bacterial fractions. L929 cells are plated on PC filters on the undersurface of dentin disks (D). Culture medium fills the chamber (M) beneath the dentin disk.
Each bacterial fraction sample was diluted with Gey's gelatin veronal buffer (GGVB) to give a series of dilutions from 0.01 to 100 #g per ml of the original suspension or solution (15). The positive control for these tests was the chemotactic substance, formyl - L- methionyl - L- leucyl - L- phenylalanine (Sigma Chemical Co., St. Louis, MO), at a concentration of 2 • 10 -8 tool per 1 in GGVB. The negative control substance was 2 x 10 -8 mol per I o f phosphate-buffered saline in GGVB. The human peripheral neutrophils were taken from healthy patients by venipuncture and separated with a discontinuous gradient of two Ficoll-Hypaque mixtures (Mono-Poly resolving medium and Histopaque 1077) as described previously (16). Then, the various dilutions of the bacterial fractions as well as the neutrophil suspensions, in the absence of serum, were placed in modified Boyden chambers to test for chemotaxis (15). The chambers were incubated at 37~ in a humid 5% CO_, incubator for 2 h, and then the 5-um pore size cellulose nitrate filters were removed, fixed with methanol and 4% formalin, stained with Mayer's hematoxylin, and mounted with glass coverslips for analysis.
Vol. 17, No. 1, January 1991
Modeling Bacterial Damage
Endotoxin Activity
The HOM, PEL, and SUP fractions of both bacteria were also assayed for endotoxin activity using the Limutus Amebocyte Lysate assay (E-Toxate; Sigma) in the absence of serum. RESULTS High concentrations of all three fractions of F. nucleatum showed significant (p < 0.01) toxicity for mammalian protein synthesis in monolayer cultures. In general, this toxicity was evident at concentrations in the range of 100 ul per ml of bacterial protein. There appeared to be minor peaks of increased protein synthesis at 1 #g per ml of H O M and at 10 #g per ml of PEL and SUP fractions. The yield of bacterial protein for T. denticola cultures was about one-tenth that of F. nucleatum, but the toxic effects of the centrifugal fractions were considerably greater. Although the HOM fraction had little effect on L929 cell protein synthesis in monolayer cultures at bacterial concentrations of less than about 1 ug per ml, higher bacterial concentrations caused increasing inhibition of L929 protein synthesis (Fig. 3). The effect plateaued at about 13% of the untreated control values. The T. denticola PEL and SUP fractions appeared to be even more toxic, so that 0.25 to 2.5 ~g per ml of bacterial protein reduced m a m m a l i a n protein synthesis to about 50 and 70%, respectively, of the control values, and 25 to 250 ug per ml reduced mammalian protein synthesis to 2% of control values for both fractions. The reductions of L929 protein synthesis by bacterial fractions above 1 ug per ml were all statistically significant (p < 0.01). For Figs. 3 and 4, the lines for each fraction simply connect mean values for each concentration and are not meant to indicate exact effects of bacterial fraction concentrations. The PEL and SUP fractions of each bacterium were placed in the reservoirs o f the IVPC, with confluent cultures of L929 cells growing on PC filters directly beneath the dentin disks. To simplify comparisons, only 1 and 10 ug per ml concentrations of the SUP and PEL fractions of each bacterium were
23
placed as 100-~1 aliquots in each of four reservoirs of the IVPC above the dentin disks (Fig. 5). For both SUP and PEL fractions of T. denticola and E. nucleatum, the 10 ug per ml depressed L929 protein synthesis more than did 1 #g per mL However, because of the large variability of readings from well to well, these differences were not statistically significant by the t test. Comparison of the monolayer results and the IVPC results for the T. denticola PEL fraction exemplifies the effect of the dentin barrier on "toxic molecules" derived from bacteria (Fig. 6). The curve demonstrates the rapid reduction of L929 protein synthesis in monolayer cuiture as the protein concentration of the PEL fraction increased. The bars represent 1 and 10 ug per ml concentrations of the PEL fraction placed in the IVPC reservoirs and the resultant effects upon new protein synthesis in L929 cells. At concentrations of 1 and 10 ug per ml, the PEI. fraction in the IVPC reservoir resulted in depression of 85% to 73%, respectively, of the negative control protein synthesis values, rather than 43 % to 15 %, respectively, as in the monolayer. Chemotaxis studies with human neutrophils gave a different picture. All three fractions of F. nucleatum showed much greater chemoattraction for neutrophils than the negative PBS
.,m~ ~~176 "%
o
SO
8
~t 9 FNPs 9 FNSOP
0
OOOl
~91
O~
1
l
10
100
I00o
LOG BACT. PROT. CONC. (ug[r~)
F~G 4 Effects of three centrifugal fractions of F. nucleatum (F.N,) on new protein synthesis by monolayers of L939 cells over a 24-h period.
120
9 i U(;M(
T
~00
...........
80
140
9- ~ ~
9
10 U(~ ML
t~'O
%'%%%%%%
IUU
'II~
80 6O
60 40
40 20
;20
Q o
D
DDDI OC*~
01
1
l
lO
1oo lOO0
ID sop
Iiii T~ , ~
~N SUP
BACTERIAL FRACTION
LOG BACT. PROT. CONC. (ug/ml)
FIG 3. Effects of three centrifugal fractions of T. denticola (TD) on new protein synthesis by monolayers of L929 cells over a 24-h period.
FIG 5. Effects of 1 ~g per ml and 10/~g per ml of two fractions (SUP and PEL) of two bacteria (TD and FN) on new protein synthesis by L929 cells in the IVPC over a 24-h period.
24
Hanks et al.
Journal of Endodontics
controls, in the absence of serum in the Boyden chamber (Fig. 7). The greatest response was with the post-16,000 x g PEL fraction at 100 #g per ml, followed by the HOM fraction at the same concentration. The highest bacterial protein concentration tested for any fraction was 100 ag per ml, and this concentration of F. nucleatum PEL fraction gave a neutrophilic chemotactic response which was statistically no different from that of the positive F M L P control. On the other hand, none of the T. denticola fractions under the same conditions, even at 100 ug per ml concentration, showed any tendency for neutrophilic chemoattraction (Fig. 8). The Limulus Amebocyte Lysate assay on selected concentrations of the bacterial fractions showed that both bacteria contained endotoxin. T. denticola H O M (10 ug per ml), SUP (10 ug per ml), and PEL (15 ~g per ml) were the lowest concentrations of each fraction to show gelation. For F. nucleatum, PEL (1 ug per ml) also caused gelation, while SUP and HOM, even at 10 #g per ml did not.
DISCUSSION In this article, we have demonstrated that the centrifugal fractions derived from 7". denticola were more cytotoxic than similar concentrations of fractions of F. mtcleatum to monolayers of L929 cells in terms of protein synthesized in a 24-h period. When placed in the IVPC, diffusion through 0.4- to 0.5-mm dentin substantially reduced the toxic effects of most of these fractions.
60 Z ~= O. Z I-'~ n-
O
.~
5e 40
[] 9
30
[]
[]
FNPEL FN-HOM FN-SUP PBS FMLP
1
20
Ii m ,B 0.01
0.1
I
10
O0
2x10^-8
CONC. (ug/ml) F~G 6. Effects of 1 #g per ml (&) and 10 #g per ml (O) of T. dentico/a PEL fraction in IVPC (bars) superimposed on graph of various concentrations of the same fraction in L929 cell monolayer cultures, All results are in terms of percentage of negative control values,
80
20
o
ooo1 OOl Ol
1
lo
loo
looo
LOG BACT, PROT. CONG.(ug/mJ)
FIG 7. Comparison of effects of various concentrations of three fractions of F. nuc/eatum to positive (FMLP) and negative (PBS) controls in tests of human neutrophilic chemotaxis in Boyden chambers,
6O Z
50 D. (.'3 40 Z
9 TD-HOM [ ] TD SUP m PBS [ ] FMLP
,~ 3o r r
20
001
0 1
1
10
100
2 x 10^'8
CONC. (ug/ml) FiG 8, Comparison of effects of various concentrations of two fractions of T. dentico/a to positive (FMLP) and negative (PBS) controls in tests of human neutrophilic chemotaxis in Boyden chambers.
In order to investigate possible diffusion of microbial products through dentinal tubules, bacteria were homogenized and fractionated. A mechanical method (sonication) of homogenization as well as lyophilization were the only methods used to avoid degradation of the biological activity of the bacterial fractions. The IVPC is intended for testing dental restorative materials and bacterial fractions or both in a setting which simulates clinical conditions, e.g. presence of remaining dentin in the cavity preparation. Unless the nature and concentration of the "toxic" molecules are known, however, it is impossible to accurately determine the extent of the toxic effects. However, even with these limitations, it is possible to gain an impression of the net effects from studies with biological mixtures. For example, in Fig. 6 both 1 and 10 ~g per ml of the T. denlico]a PEL fraction are superimposed on the curve representing the effects of various concentrations of the same fraction in the monolayer cultures. If the bars representing these two concentrations are moved to the left until they intersect the curve derived from monolayer results, the net effect of the dentin barrier appears to be a dilution of the toxic effect of this fraction by more than 1000 times. There was more variability of values for the IVPC than for the monolayer cultures, even when the dentin disks were selected for minimal variation in thickness and hydraulic conductance. Prior to diffusion experiments, filtration rate and hydraulic conductance were stabilized empirically by treatment of dentin disks with 0.01 mol/1 of Tris-HCl buffered 70% ethanol under house vacuum (560 mm Hg) for 30 min. This treatment probably had several effects on the dentin disks such as replacement of air bubbles in the tubules, fixation of residual proteins in the dentinal tubules, and fixation of bacterial proteins, thus preventing infection. During the experiments, the variation in diffusion can be attributed to both variability in the size and distribution of dentinal tubules in the disks and the variation of the diffusion and adsorption of diverse molecules from each centrifugal fraction. Both F. nuclealum and T. denticola have extractable lipopolysaccharides (10, 12). In the present study, the Limulus assay confirmed the presence of these materials in all centrifugal fractions, Recently, Segal et al. (17) demonstrated that Actinobacillus actinomycetemcomitans LPS diffuses through 0.2-mm dentin disks, but also reduces the hydraulic conductance of these disks over a 24-h period. LPS from Escherichia coli requires serum for stimulation of D N A synthesis in cultured primary mouse embryonic fibroblasts (18). However, even in the presence of serum, Aleo et al. (19) reported that
Vol. 17, No. 1, January 1991
E. coli LPS does not stimulate DNA synthesis in L929 cells, but reduces cell viability (trypan blue incorporation) in culture over a 3-day period. Loss of membrane integrity might be explained by LPS activation of complement which generates both a complex of factors which lyse cell membranes and chemotatic factors for neutrophils. There were dramatic differences in the ability of centrifugal fractions of the two organisms to depress protein synthesis in vitro. Both the PEL and SUP fractions of T. denticola were very effective. While proteolytic activity has been described for 7". denticola (13, 14), it is not known if these enzymes are the molecules responsible for in vitro toxicity. Boehnringer et al. (20) have shown that a chromatographic fraction from T. denticola sonicates has the ability to suppress DNA, RNA, and protein synthesis in cultured human fibroblasts. This activity was associated with a single chromographic peak of about 50,000 daltons and was heat labile, suggesting it was protein and not lipopolysaccharide in nature. In the present study, only fractions o f F . nucleatztm caused a chemotactic response to peripheral human neutrophils in the absence of serum. The PEL fraction had been washed five times with Tris-HCl buffer so that serum complement appeared not to have played a role in the assay results. The PEL fraction of F. nucleatum, at a concentration of 100 ug per ml of bacterial protein, had chemoattractive activity similar to 2 x 10-8 M FMLP. For clinical relevance, further studies are required to compare this with the level of chemoattraction in the presence of serum and the number of organisms represented by these levels of endotoxin activity. This study has focused on mechanisms by which F. nzzcleaturn and T. denticola might cause pulpal damage if they or their products were introduced into the dentinal tubules either beneath restorations or through root dentin. These organisms appear to cause their principle tissue damage by at least two different methods. T. denticola releases toxic molecules, perhaps hydrolytic enzymes, which are capable of depressing macromolecular synthesis in fibroblastic cells. This biological activity is essentially neutralized by the diffusion gradient of 0.5 m m of remaining dentin in the bottom of the cavity preparation. Much less macromolecular synthetic depression is caused by direct contact with components (fractions) of F. nucleatum. However, more tissue injury may be caused by neutrophilic chemotaxis to F. nucleatum, or by complement activation due to LPS released by either organism. In order to assess more properly the contributions of these organisms to pulpal injury, however, it will be important in future studies to determine (a) the concentrations at which individual molecules have their biological effects, and (b) the concentrations at which these molecules reach the pulp across the gradient formed by the solution in the dentinal tubules.
Modeling Bacterial Damage
25
This work was supported by the National Institute of Dental Research Grant DE07987. The authors appreciate the assistance of M. L. Diehl and M K. Gardner (experiments).
Drs. Hanks, Syed, Craig, and Hartrick are affiliated with the School of Dentistry, University of Michigan, Ann Arbor, MI. Dr Van Dyke is affiliated with Department of Periodontology, Emory University, School of Dentistry, Atlanta, GA.
References 1. Zander HA. Pulp response to restorative materials. J Am Dent Assoc 1959;59:911-5. 2. Cox CF, Keall CL, Keall HJ, Ostro E, Bergenholtz G. Biocompatibility of surface-sealed dental materials against exposed pulps. J Prosthet Dent 1987;57:1-8. 3. Watts A. Bacterial contamination and the toxicity of silicate and zinc phosphate cements. Br Dent J 1979;146:7-13. 4. Bergenholtz G, Staffan A, Lindhe J. Experimental pulpitis in immunized monkeys. Scand J Dent Res 1977;85:396-406. 5. Mjor IA, Hensten-Pettersen A, Skogedal O. Biological evaluation of filling materials. A comparison of results using cell culture techniques, implantation tests and pulp studies. Int Dent J 1977;27:124-9. 6. Hanks CT, Craig RG, Diehl ML, Pashley DH. Cytotoxicity of dental composites and other materials in a new in vitro device. J Oral Pathol 1988;17:396-403. 7. Hanks CT, Diehl ML, Craig RG, Makinen P-L, Pashley DH. Characterization of the "in vitro pulp chamber" using the cytotoxicity of phenol. J Oral Pathol Med 1989;18:97-107. 8. Meryon SD. The influence of dentin on the in vitro cytotoxicity testing of dental resotrative materials. J Biomed Mat Res 1984;18:771-9. 9. Mejare B, Mejare I, Edwardsson S. Bacteria beneath composite restorations--a culturing and histobacteriological study. Acta Odontol Scand 1979;37:267-75. 10. Sveen K. Rabbit polymorphonuclear leukocyte migration in vivo in response to lipopolysaccharides from Bacteroides, Fusobacterium, and Veilonella. Acta Pathol Microbiol Immunol Scand [B] 1977;85:381-387. 11. Van Steenbergen TJM, Van Der Mispel LMS, Degraff J. Effects of ammonia and volatile fatty acids produced by oral bacteria on tissue culture cells. J Dent Res 1986;65:909-12. 12. Sela MN, Weinberg A, Borinksky R, Holt SC, Dishon T. Inhibition of superoxide production in human polymorphonuciear leukocytes by oral treponemal factors. Infect Immun 1988;56:589-94. 13. Makinen KK, Syed SA, Makinen P-L, Loesche WJ. Benzoylarginine peptidase and iminopeptidase profiles of Treponema denticola strains isolated from the human periodontal pocket. Curr Microbio11986;14:85-9. 14. Grenier D, Vitto V-J, McBride BC. Cellular location of a Treponema denticola chymotrypsinlike protease and importance of the protease in migration through the basement membrane. Infect Immun 1990;58:347-51. 15. Van Dyke TE, Reilly AA, Horoszewicz H, Gagliardi N, Genco RJ. A rapid, semi-automated procedure for the evaluation of leukocyte locomotion in the micropore filter assay. J Immunol Meth 1979;31:271-82. 16. Kalmar JR, Arnold RR, Warbington ML, Gardner MK. Superior leukocyte separation with a discontinuous one-step FicolI-Hypaque gradient for the isolation of human neutrophils. J Immunol Methods 1988;100:275-281. 17. Segal H, Stevens RH, Trowbddge H, Pash[ey DH. Permeability of human dentin to bacterial endotoxin. J Dent Res 1990;69:355. 18. Smith GL. Synergistic action of bacterial lipopolysaccharides in serumstimulated DNA synthesis in mouse embryo fibroblasts. Proc Soc Exp Biol Med 1976;153:187-192. 19. Aleo JJ, De Renzis FA, Farber PA, Varboncoeur AP. The presence and biologic activity of cementum-bound endotoxin. J Periodonto11974;45:672-5. 20. Boehringer H, Taichman NS, Shenker BJ. Suppression of fibroblast proliferation by oral spirochetes. Infect Immun 1984;45:155-9.
