Phagocytosis and virulence of different strains of Porphyromonas gingivalis

Abstract

Abstract – In this study 17 strains of Porphyromonas gingivalis, both reference and clinical isolates, were investigated for their in vitro interaction with human polymorphonuclear leukocytes, hydrophobicity, density, and virulence in a mouse model. The results of the phagocytosis, hydrophobicity, and density experiments showed that P. gingivalis strains could be divided into two distinct groups. One group of strains were readily attached and phagocytosed when exposed to the leukocytes. These bacteria were hydrophobic and had a higher buoyant density than the other group, which were poorly phagocytosed, had a low buoyant density, and were hydrophilic. This latter group also exhibited an extracellular meshwork resembling a glycocalyx when examined by electron microscopy. There were also significant differences between strains in the mouse pathogenicity model. Two strains caused an invasive, spreading infection compared with the other 15 strains which produced small, localized abscesses. There was no clear correlation between the results of the phagocytosis assay and the virulence of the bacteria when injected subcutaneously in mice. Resistance to phagocytosis may be important for survival of these bacteria, but it does not in itself imply the ability to cause damage to the host.

Full text

Phagocytosis and virulence of different strains
oi Porphyromonas gingivalis
GORAN SUNDQVIST', DAVID FIGDOR', LENNART HANSTROM=, STIG SORLIN'
AND
GUNNAR SANDSTROM^
Departments
of
^Endodontics
and
"^Periodontology,
University
of
Umea
and
'^Department
of
Cell
and
Microbiology,
Swedish Defense
Research
Establishment, Umea, Sweden
Sundqvist
G,
Figdor
D,
Hanstrom
L,
Sorlin
S,
Sandstrom
G:
Phagocytosis
and
virulence
of
different strains
oi
Porphyromonas
gingivalis.
Scand
J
Dent
Res
1991;
99:
117-29.
Abstract
- In
this study 17 strains
oi
Porphyromonas
gingivalis,
both reference
and
clinical isolates,
were investigated
for
their
in
vitro interaction with human polymorphonuclear leukocytes,
hydrophobicity, density,
and
virulence
in a
mouse model.
The
results
of the
phagocytosis,
hydrophobicity,
and
density experiments showed that
P.
gingivalis
strains could
be
divided into
two distinct groups.
One
group of strains were readily attached
and
phagocytosed when exposed
to
the
leuko cytes. These bacteria were hydrophobic
and h ad a
higher buoyant density than
the
other group, which were poorly phagocytosed,
had
a
low buoyant density,
and
were hydrophilic.
This latter group also exhibited
an
extracellular meshwork resembling
a
glycocalyx when exam-
ined
by
electron microscopy. There were also significant differences between strains
in the
mouse
pathogenicity model.
Two
strains caused
an
invasive, spreading infection compared with
the
other
15
strains which produced small, localized abscesses. There
was no
clear correlation
between
the
results
of
the
phagocytosis assay
and the
virulence
of
the
bacteria when injected
subcutaneously
in
mice. Resistance
to
phagocytosis
may be
important
for
survival
of
these
bacteria,
but it
does
not in
itself imply
the
ab ility
to
cause damage
to the
host.
Key words: glycocalyx; phagocytosis;
P.
gingivalis;
virulence.
G. Sundqvist, Department
of
Endodontics, University
of
Umea, S-901
87
Umea, Sweden.
Accepted
for
publication
5
September
1990.
Porphyromonas
gingivalis
has
been implicated
proteolytic activities which profoundly affect
in
the
pathogenesis
of
advancing lesions
of the
host defense. Thus,
we
have shown that
periodontal disease
in
humans (1-3),
and
complement factors
C3 and C5 and
immu-
slrains
of
this species induce rapidly spread- noglobulins
are
degraded
in
both guinea pigs
ing,
purulent infections which result
in ex- and
humans
(6, 7),
indicating potential
in-
tensive necrotic breakdown
of the
tissues
in
terference with
the
immunological defense
mice
and
guinea pigs (4-6).
The
pathogenic system. Also, plasma proteinase inhibitors
j)otential
of
this species
may be due to a (8),
iron-handling proteins
(9) and key fac-
-
ariety
of
virulence factors. Previous studies tors
of the
plasma proteinase systems
(10)
;
lave
suggested that this bacterium possesses
are
degraded
or
functionally eliminated
by

118
SUNDQVIST ET AL.
P.
gingivalis
enzymes.
In
addition
to
this,
P.
pngivalis possesses collagenolytic aetivity
(11-13),
which
may be of
importance
in
tissue destruction (14,
15).
The extent of tissue damage and invasive-
ness
in
induced infections
in
animals
has
been widely used
as a
measure
of
bacterial
virulence.
A
number
of
studies have shown
that there is
a
considerable variation
in
viru-
lence between strains within the species
and
it has been suggested that this may be due
to
differences
in
collagenolytic
and
proteolytic
activities
(16, 17).
Ho wev er, recent results
indicate that the protease and collagenolytic
activities
do not
account
for the
diflferences
between
the
invasive
and
non-invasive
strains
(18, 19). I t has
also been suggested
that differences
in
virulence might
be
d ue
to
the degree of encapsulation and resistance
to
phagocytosis (20-22).
The aim
of
th is study was
to
examine
the
relationship between
the
virulence
of dif-
ferent P.
gingivalis
isolates
and
their surface
structures
and
the interaction
of
th e various
strains with human polymorphonuclear neu-
trophil leukoeytes.
Material and methods
Microorganisms
and
culture conditions
- The
strains
tested
are
listed
in
Table
1.
The
strain ATCC
33277
was
ob ta in ed from Ameriean Type Collec-
tion Rockville,
MD,
and
strains VV83
and 381
from
T.
J.
M.
VAN STEENBERGEN, Free University.
Amsterdam,
The
Ne therla nds . Other strains were
isolated
by the
authors from dental root canals
or
periodontal pockets.
The
strains were kept
in an
anaerobic
box
with
an
atmosphere
of
5% carbon
dioxide
and 10%
hydrogen
in
nitrogen
on
blood
agar plates
at
37°C.
The
blood agar medium
was
prepared
by the
method
of
HOLDEMAN
et al.
(23).
For
the
experiments, bacteria were grown
for 24
h
in
peptone yeast-glucose broth
(23)
at
37°C,
harvested
by
centrifugation, washed twice
in
phos-
phate-buflered saline (PBS;
pH 7.4) and
adjusted
to
the
concentration used
in
the
various experi-
ments.
Discontinuous density gradient centrifugation
~ The
density
of the
bacterial strains
was
tested
as d e -
scribed
by
PATRICK
&
REID
(24). Briefly,
2 ml of
the bacterial suspension
was
layered
on top of a
discontinuous density gradient
gel of
Percoll
(Pharmacia, Uppsala, Sweden)
and
centrifuged
at
2600
X
g
for 20 min. The
concentration
of
Percoll
in the various gradients was 20%, 40%, 60%,
and
80%.
The
bacteria were recovered after centrifu-
gation
at
the
interface layers
of
these gradients.
Opsonins
and
opsonization procedure
-
Normal
se-
rum
was
obtained from
six
healthy donors
and
pooled. Antiserum
was
prepared
by
immunizing
rabbits with heat-killed cells
of
the
strain
W83.
The sera were stored
at
—
70°C,
and
thawed short-
ly before use. Bacteria were opsonized
by
incuba-
tion
in
20% serum
at
37°C
for 10 m in or 30 min.
Bacteria were centrifuged
(10 min
at
1600x^),
washed
and
suspended
to a
final concentration
of 5
X
10" bacteria
per ml.
In
some experiments
inhibitors
of
proteinase enzymes
of
P.
gingivalis
were present
in the
bacteria-serum mixtures
dur-
ing the opsonization. The following inhibitors were
used: hydrogen peroxide (100 ixM), mercuric chlo-
ride
(10
|iM), p-chloromercuriphenylsulfonic acid
(100
|iM),
N-tosyl-L-lysine chloromethyl ketone
(100
\iM),
tosyl-L phenylalanine chloromethyl ke-
tone (100 |iM). The presence of antibodies against
the strain
in
the
sera
was
demonstrated
by the
indirect fluorescent antibody technique
as
pre-
viously described
(25).
Preparation
of
human polymorphonuclear leukocytes
-
Venous blood from healthy volunteers
was ob -
tained
in
heparinized glass tubes (Becton, Dicki-
son & Co., Rutherford, NJ).
The
bl ood was mixed
with
6%
dextran (Pharmacia, Uppsala, Sweden)
in
0.15
M
sodium chloride solution. After sedi-
mentation
of
erythrocytes during 1
h at
22 °C
the
leukocyte-rich plasma was withdrawn
and
centri-
fuged
at
160x^
for 10
mi n
at
4°C.
When
the
leukocytes were used
in
killing experiments
in t he
anaerobic
box,
they were resuspended
in
0.87%
ice-cold ammonium chloride
to
lyse contaminating
erythrocytes. Leukocytes were thereafter washed
twice in RPMI 1640 medium with 20 mM HEPES
and
2
mM L-glutamine (Gibco, Paisley, Scotland)
and taken into
the
anaerobic
box and
suspended
in RPMI
1640
medium with
2%
human serum
albumin (Sigma Chemical Co.,
St.
Louis,
MO)
at
a density
of
1
x
10' polymorphonuclear leukocytes
(PMNs)
per ml. The
RPMI
1640
medium
had
been stored overnight
in the
box.
The
suspension
of leukocytes
was
kept
on ice for
1
h
before
use.
When used
in the
phagocytic assay with radioac-

PHAGOGYTOSIS AND VIRULENCE OF P GINGIVALIS 119
Table 1
Characteristics
of
P.
gingivalis strains
Strain
33277
381
220-1
220-2
BE-c
329-1
1:15
339-1
332-2
320-3
VV83
B262
295
JBB-c
274
81
102
Phago-
cytosis'
36.3
47.2
27.5
63.3
32.1
35.6
30.0
30.8
11.3
14.4
6.9
8.5
5.7
5.8
4.7
6.2
5.9
Hydrophobicity
Phase
system^
55.4
42.6
71.4
34.5
59.2
73.9
46.6
53.8
90.1
94.2
91.4
92.2
91.2
99.0
96.1
99.9
99.5
Salt
aggregation'
0.03
0.03
0.03
0.03
0.03
0.03
0.03
0.03
4.0
1.0
1.5
0.5
0.4
0.5
0.5
4,0
4.0
Buoyant
density*
40
60
60
40
80
40
40
60
20
20
20
20
20
20
20
20
20
Antibody
titer'
64
32
64
16
128
32
16
32
16
16
16
16
16
16
16
32
32
' Percentage of bacteria associated with leukocytes after 5 min incubation at
37 °G.
Mean of
six
or more
experiments.
'-'Percentage of bacteria remaining in aqueous phase after hexadecane partitioning. Mean of three
experiments.
•'End-point of aggregation was determined using serial dilutions of ammonium sulfate (0.03 M-4.0 M).
•'The density of the bacteria was tested with Percoll discontinuous density gradient; 20%, 40%, 60%,
80%.
Gradient layer in which bacteria were recovered is given.
' Reciprocal antibody titer in the pooled serum.
tive labeled bacteria, the leukocytes were suspend-
ed in 3 ml ice-cold deionized distilled water and
1 ml 0.6 M sodium chloride was added to lyse
erythrocyte contaminants. After 30-40 s 10 ml ice-
cold PBS was added. The leukocytes were washed
and suspended in Gey's solution. Of the leukocytes
in the suspensions the PMNs made up approxi-
mately 90%. Viability was greater than 95% as
assessed by trypan blue exclusion.
Phagocytosis of radioactive labeled bacteria ~ Bacteria
were inoculated from a blood agar plate into 5
ml of peptone-yeast-glucose broth containing 0.05
mCi thymidine methyl ''H-labeled (8 Gi/mmol
specific activity, New England Nuclear Ghemicals,
CmbH, Dreieich, FRG) and incubated for 24 h
ai 37°G. The bacteria were washed twice in PBS
iiid the radioactivity of the bacteria was measured
I, fore use. Phagocytosis was assayed as described
by
ROZENBERG-ARSKA
et al. (26).
Briefly,
0.3 ml of
the suspension of radiolabeled opsonized or unop-
sonized bacteria (5 x 10" GFU/ml) was mixed with
0.3 ml of the leukocyte suspension (1 x 10' cells/
ml) in four polypropylene tubes. The mixtures
were incubated and shaken in a water bath for 5
min and the phagocytosis was stopped by adding
1.0 ml ice-cold Gey's solution. To get a measure
for total amount of added bacteria, tubes were
centrifuged at 1600 x ^ for 15 min and the pellets
were suspended in 1 ml deionized distilled water
and 5 ml Aquasol 2 scintillation liquid was added.
Radioactivity was measured in a liquid scintilla-
tion counter (Beckman Instruments, Inc., Fuller-
ton, GA). Tubes were centrifuged at 160 x^' for 5
min and the pellets were washed three times in
ice-cold Gey's solution to remove bacteria loosely
adhering to the PMNs. Radioactivity of the fmal

120SUNDQVIST ET AL.
pellet representing cell associated bacteria was
measured as above. Phagocytosis was expressed as
the percentage of uptake of total radioactivity.
Phagocytosis and killing under anaerobic conditions -
Anaerobic phagocytosis and killing was examined
in the anaerobic box. All vials and media used
were stored in the box at least overnight. Prepara-
tion of the bacterial suspension and opsonization
was done anaerobically. The bacteria were opson-
ized for
10
min in serum that had been thawed and
kept for
1
h in the box. One ml of
the
suspension of
opsonized bacteria (5 x lO^/ml) was added to
1
ml
of the leukocyte suspension (1 x lO'/ml). A sample
(200 |xl) was added to 1.8 ml ice-cold sterile dis-
tilled water to determine the total number of via-
ble bacteria. Further dilutions were made in dilu-
tion blanks (23) and cultivation was done on blood
agar. The blood agar plates were incubated for 7
days at 37°C. The bacteria-leukocyte mixture was
incubated at 37°C on a tilting table. After 10 min
the mixture was put on ice to stop the phagocyto-
sis.
To estimate phagocytosis the mixture was cen-
trifuged at 160 x ^ for 5 min at +4°C to sediment
leukocytes and leukocyte-associated bacteria. A
sample was taken from the supernatant to calculate
non-phagocytosed bacteria. The pellet was resus-
pended in 1.8 ml RPMI 1640 medium, centrifuged
at 160x
w
for 5 min at 4°C. A sample was taken
from the supernatant to determine non-phagocy-
tosed bacteria which had pelleted at the first cen-
trifugation. The pellet with leukocyte ingested
bacteria was resuspended in 1.8 ml RPMI 1640
medium. A sample (200 |il) was taken and put on
ice for 120 min to calculate the survival of ingested
bacteria at +4°C. The leukocyte suspension was
incubated at 37°C. Samples were taken at 0 and
120 min to calculate the killing of the phagocy-
tosed bacteria.
Determination oj cell-surface hydropliohicity - The
bacterial-cell surface hydrophobicity was mea-
sured in 2 ways: with the salt aggregation test
according to ROZGONYI
et
al.
(27) and with bacte-
rial adherence to the hydrocarbon, hexadecane
(28).
The salt-induced cell aggregation test is based
on the tendency of hydrophobic bacterial cells to
clump at relatively low ionic strength and the fact
that hydrophilic bacteria aggregate at high ionic
strength. The end-point of aggregation for the
strains was determined using serial dilutions of
ammonium sulfate ranging from 0.03 M to 4.0 M.
The highest dilution giving visible aggregation was
scored as a numerical value for bacterial surface
hydrophobicity. Bacteria grown in peptone-yeast-
glucose broth (23) were tested with this assay.
In the hexadecane assay, the bacteria were
.grown for 24 h at 37°C in peptone-yeast-glucose
broth containing 0.05 mCi thymidine methyl ^H-
labeled and washed twice in PBS. Three-milliliter
samples of the washed bacterial suspension (10'/
ml) were transferred to test tubes and 0.2 ml of
hexadecane (Sigma) was added. The layers were
mixed vigorously on a rota-mixer for 15 s and
the hexadecane was allowed to partition while
standing for 15 min. This complete procedure was
repeated twice. After three cycles of rotamixing
and partitioning of cells and hexadecane, the ra-
dioactivity of the aqueous phase was measured.
The values were expressed as the percentage of
bacteria that remained in the aqueous phase com-
pared with that of the original suspension.
Electron microscopy
- The polymorphonuclear leu-
kocyte-bacteria interactions were terminated by
mixing the suspensions with equal volumes of
4%,
glutaraldehyde in PBS at 0°C for 60 min. Further
treatment of
the
specimens was carried out accord-
ing to ROZENBERG-ARSKA et al. (26). The ultra-
structure of the strains was studied with a tech-
nique described by SANDSTROM
et
al.
(29). Colony-
containing agar plates were overlaid with nutrient
broth (Oxoid Ltd., Basingstoke, England), supple-
mented with 2.5% sodium chloride and 0.6% Bac-
to-Agar (Difco laboratories, Detroit, MI). Fixative
(2.5%
glutaraldehyde in 0.1 M cacodylate buffer
(pH 7.2), containing 0.1 M sucrose) was added to
the plate. After being fixed overnight, colonies
were excised, washed in cacodylate buffer, dehy-
drated in a graded series of
ethanol,
treated for 30
min in propylene oxide, and embedded in Epon
812.
Sections were cut with an LKB ultratome
and stained with uranyl acetate and lead citrate.
A
Jeol 100 CX electron microscope was used.
Virulence assay
- The virulence of the strains was
compared in mice using a pathogenicity model
based on that of VAN STEENBERGEN
el
al.
(5). Bacte-
rial cells grown for 4 days on blood agar were
suspended in peptone-yeast-glucose broth (23). To
avoid contact with air, harvesting and preparation
of
the
suspensions were performed in the anaerobic
box. Viable counts were determined by plating
serial dilutions of the suspensions on blood agar
plates.
BALB/c mice weighing about 30 g were
injected subcutaneously in the back with 0.2 ml
of the bacterial suspension. The animals were ex-

PHAGOCYTOSIS AND VIRULENCE OF
P.
GINGIVALIS 121
amined daily over 7 days for the development of
lesions and health status.
Results
The uptake of radiolabeled bacteria by
PMNs is shown in Table 1. There was a low
uptake of nine of the strains. Less than 10%
of the cells of the strains W83, B262, 295,
JBB-c, 274,
81,
102 and only 11.3 and 14.4%
of the cells of the strains 332-2 and 320-3
were associated with PMNs within 5 min. Of
the other strains,
30-63 %
of the cells were
associated with the PMNs after 5 min expo-
sure.
The bacteria had been opsonized in
20%
pooled serum in these experiments. The
strains W83, 381, and ATCC 33277 were
also used unopsonized. The uptake of these
strains was then 6.9%,
41.5%,
and
30.5%,
respectively. Low levels of antibodies to all
the tested strains were present in the pooled
serum. Opsonizing the strain W83 in 50%
pooled human serum or in rabbit antiserum
(titer
1:2048)
did not increase the uptake
by PMNs. The strain W83 was, in some
experiments, opsonized in the presence of:
100 |J.M hydrogen peroxide, 10
]\M.
mercuric
chloride, 100
\iM.
p-chloromercuriphenylsul-
fonic acid, 100 |iM N-tosyl-L-lysine chloro-
methyl ketone (TLCK), and 100 pM tosyl-
L-phenylalanine chloromethyl ketone
(TPCK). The presence of these enzyme in-
hibitors during opsonization did not influ-
ence the subsequent uptake of the bacterium
by PMNs (data not shown).
The uptake and killing of
the
strains W83,
381,
and ATCC 33277 under anaerobic con-
ditions is shown in Table 2. After 10 min
exposure to PMNs, 6.2% of the cells of the
strain W83 were associated with the leuko-
cvtes.
Corresponding figures for the strains
ATCC 33277 and 381 were 37.5% and
61.2%,
respectively. To determine the killing
of the bacteria in the leukocytes, the PMNs
vere further incubated after being washed
fee of n on-associated bacteria. After 120 min
Table 2
Phagocytosis and killing ofV. gingivalis by polymorpha-
nuclear leukocytes under anaerobic conditions
Bacterial
strain
W83
381
33277
Phago-
cytosed''
6.2 ±3.6
61.2±6.2
37.5 ±3.5
Bacteria (%)a
Survival of phagocytosed
bacteria'
+ 37°C
73.7± 11.5
23.7 ±26.8
16.0
±9 .8
+ 4°C
114.0±21.0
118.0 ±20.0
93.0 ±16.0
"
Values are mean
±
SD of three experiments.
"^
Preopsonized bacteria (5 x 10°) were incubated
for 10 min at 37°C with PMNs (1 x 10') in 2 ml
RPMI-HEPES. To estimate phagocytosis, sam-
ples were centrifuged at 160x^ for 5 min and
the percentage of viable bacteria in the superna-
tant was calculated.
' For estimates of survival of phagocytosed bacte-
ria, sediments obtained after centrifugation of
10 min samples were resuspended in RPMI-
HEPES,
incubated at 37°C and +4°G for 120
min and treated with ice-cold distilled water. Via-
ble counts were performed and compared with
viable counts on the 10 min sample.
incubation 16.0% and 23.7% of the cells of
strains ATCC 33277 and 381 were viable,
respectively. The killing rate of strain W83
was lower and 73.7% of the bacterial cells
were still viable after 120 min incubation of
the leukocytes. The bacteria survived when
the leukocytes were kept on ice (Table 2).
To determine if the bacteria associated
with the PMNs were ingested, PMNs which
had been exposed to
P.
gingivalis
were exam-
ined by transmission electron microscopy.
Cells of the strains 381 and 33277 were
phagocytosed by the PMNs within 10 min
(Fig. lA and B). Approximately 10-20 bac-
terial cells in various stages of disintegration
could be seen in sections of most leukocytes.
In some sections the early stages of phagocy-
tosis,
i.e. attachment and engulfment, could
be observed. Of the strain W83, almost all
bacteria were found extracellularly and only
a few bacteria eould be seen within the leu-

122SUNDQVIST ET AL.
Fig. I. Human PMNs af-
ter 5 mm of incubation
with P. gingivalis strains
381 (A), ATCC 33277
(B) and W83 (C).
X
7500.

PHAGOCYTOSIS AND VIRULENCE OF P
GINGIVALIS
123
kocytes after 10 min of phagocytosis (Fig.
1G),
This strain was phagocytosed to a very
limited extent even when the incubation time
was increased to 60 min (data not shown).
All strains exhibited a cell wall structure
with an outer and inner cell membrane typi-
cal of Gram-negative bacteria. An electron
dense external layer in direct contact with
the outer membrane as well as extracellular
vesicles were present on all strains. Between
the cells of strains W83, B262, 295, JBB-c,
274,
81, 102,
320-3,
and 332-2 there was a
spider-web like electron-dense reticula (Fig.
2A),
This feature was either absent or poorly
defined in the remaining strains (Fig, 2B),
There were clear differences between the
strains in the hydrophobic assay (Table 1),
Nine strains were hydrophilic with more
than 90% of the cells being recovered in the
aqueous phase in the two-phase system with
the hydrophobic hydrocarbon, hexadecane.
These strains also aggregated at salt concen-
trations between 0.4 M and 4.0 M, The other
strains aggregated at the lowest used salt con-
centration (0,03 M) and of these strains
34.5-73,9% of the cells remained in the
aqueous phase after hexadecane partition-
ing. The nine strains which were hydrophilic
had a lower density than the other strains
and were recovered in the 20% Percoll layer
or on top of it in the discontinuous density
gradient centrifugation (Table 1), The other
strains were recovered in the 80% gradient
layers.
Marked differences were found when the
strains were compared in the mouse pathoge-
nicity model. The strains W83 and B262
caused abscesses which spread to the ventral
side of the mice with necrosis and rupture of
the skin. Animals injected with these two
strains became cachectic and died at the
third day. For the induction of this type of
lesion the initial number of bacteria in the
inoculum had to be at least 8,0 x 10", The
oilier strains did not cause this type of
infec-
ti)n even when each mouse was injected with
5.,)
X
10'° colony forming units. These strains
produced small abscesses which did not
spread or ulcerate.
Discussion
The results of the phagocytosis, hydrophobi-
city, and density experiments showed that
the
P.
gingivalis
strains could be divided into
two distinct groups. One group of strains
(shaded section. Fig. 3) were readily at-
tached and phagocytosed when exposed to
PMNs, These bacteria were also more hy-
drophobic and had a higher buoyant density
than the other group (unshaded section. Fig.
3),
which were poorly phagocytosed, had a
low buoyant density, and were hydrophilic.
The phagocytosis experiments performed
under anaerobic conditions with a cultiva-
tion technique gave similar results to the
experiments performed aerobically. The
strains 381 and ATCC 33277 were phagocy-
tized and killed by the PMNs to a high extent
while the strain W83 was phagocytized and
killed to a low extent (Table 2), Electron
microscopy showed that PMN-associated
bacteria were phagocytosed by the PMNs
and were in various stages of destruction
(Fig. 1).
Opsonization of the bacteria in normal
human serum had little effect on phagocyto-
sis since there was only a slight increase in
uptake of the strains 381 and ATCC 33277
when they were opsonized, while the strain
W83 was poorly phagocytized whether op-
sonized in the pooled human serum, homolo-
gous rabbit antiserum or exposed to the
PMNs unopsonized. It has recently been
shown that this strain fails to accumulate
C3 during opsonization by serum due to its
proteolytic capacity (30). A large increase in
the number of C3 molecules bound to W83
was,
however, noted in assays carried out in
the presence of the protease inhibitor TLCK
(30).
We found that the presence of this in-
hibitor, and other known inhibitors of the
proteolytic activity oi
P.
gingivalis
(13, 19),

124SUNDQVIST ET AL.
Fig. 2. Elcclron micrographs of
P.
gingivalis strains. A, strain B262, which is representative of a group
having an extensive intercellular matrix. B, strain 1:15 is representative of a group lacking this structure
X
48 000.

PHAGOCYTOSIS AND VIRULENCE
OF
P.
GIJ^GIVALIS
125
PHAGOCYTOSIS
WITH
PMN'S
60 H
40 -
BACTERIA HYDROPHOBICITY
AQUEOUS
PHASE
80-
20
PERCOLL
AT WHICH
BACTERIA
RECOVERED
(%)
80 •
6 0
•
40
20
DENSITY
Fig.
3.
Relationship between results for phagocyto-
sis experiments, determination
of
cell-surface
hy-
drophobicity
and
buoyant cell density
of
various
sirains. Arrowheads show virulent strains.
during opsonization
did not
increase
the
phagocytosis of the strain
W83.
This suggests
that opsonically active
C3 w as not
attached
to
the
true external surface
of
the bacterial
cells as has been shown for encapsulated bac-
teria (31).
Furthermore
it
suggests that the proteases,
while important
as
factors
in
tissue destruc-
tion,
do not
have
an
inhibitory effect
on
opsonophagocytosis.
It has
recently been
shown that some strains
of
P.
gingivalis
may
elaborate factors which impair
the
normal
bactericidal activity
of
the leukocytes since
the culture supernatant of strain 381
has the
ability to modulate the generation of reactive
oxygen species
by
the leukocytes (32).
It
ap-
pears unlikely that this factor
is
indeed
a
protease since
the
factor
was
neither heat
sensitive
nor
was
it
affected
by the
protease
inhibitors tested
(32).
Bacterial hydrophobicity
is an
important
factor
for
resistance
to
phagocytosis.
In
gen-
eral,
a
hydrophilic surface
is
advantageous
for bacteria in avoiding phagocytosis, where-
as hydrophobic bacteria
are
more readily
phagocytosed (33-35).
In
accordance with
this,
we
found that
the
strains which were
recovered
in the
aqueous phase
in the two-
phase system were poorly phagocytized
while
the
strains with affinity
for the
hydro-
phobic hexadecane were readily phagocy-
tized
(Fig. 3 ).
There
was
also
a
correlation
between hydrophobicity
and the
buoyant
density
of
the cells.
The
discontinuous
de n-
sity gradient assay has been used
to
separate
capsulate
and
non-capsulate cells
of
strains
of B.
fragilis
(24). Cells with large capsules
remained,
in
this assay,
in or on top of
the
20%
layer and non-capsulate cells were con-
centrated
in
the higher gradients.
REYNOLDS
et
al.
(36) have recently reported heterogeni-
city
in
encapsulation within
P.
gingivalis
strains
and
have also noted
a
correlation
be-
tween buoyant density
and
capsulation.
Capsule-like structures
are
known
to be
antiphagocytic
in a
wide variety
of
organ-
isms
(37, 38) and our
results indicate that

126SUNDQVIST ET AL.
the differences in phagoeytosis within the
species
P.
gingivalis
may be due to differences
in surface structure. Several authors have
observed capsules or electron-dense material
associated with the outer membrane in
P.
gingivalis
(39—43). Similar structures were
observed in our study, but in addition the
electron microscopic examination revealed
an intercellular matrix forming a meshwork
between the eells of the strains whieh were
poorly phagoeytized. These structures are in-
dicative of the presence of glycocalyces on
these strains (44, 45). The bacterial glycoca-
lyx consists of exopolysaecharides that con-
tain approximately 99% water (46) and is
usually not seen in electron micrographs be-
cause of
its
radical eondensation upon dehy-
dration, which is a normal requirement for
electron microscopy (44). However, our
demonstration of these structures is due to
the use of a preparation technique whieh is
favorable for stabilizing capsule structures
(29).
It has recently been shown that the
Bacteroides
glycocalyx may not only impair
phagocytosis of homologous strains or spe-
cies,
but also inhibit phagoeytosis of associat-
ed organisms in mixed infections when the
glyeocalyx is detaehed into the surrounding
milieu (47). The glyeoealyx has been shown
to be a virulenee factor in pathogens as di-
verse as Actinomyces, Bacteroides, Hemophilus,
Streptococcus, and Staphylococcus (44).
Pili and fimbriae are other surface struc-
tures which may influence the susceptibility
to phagocytosis by human neutrophils. A
direct correlation has been demonstrated be-
tween degree of piliation and susceptibility
to phagoeytosis for bacteria of the species
E. coli, Pseudomonas aeruginosa, and B. fragilis
(48-50).
P.
gingivalis
has been reported to
show the presence of fimbriae on the phago-
cytosis-susceptible strains 381 and ATCC
33277 (51-55). The number of f i m b r i a t e cells
and the density of fimbriae varies within
strains of the species. While strain 381 has
been reported to be near 100% fimbriated
the strain W83 may be fimbriated to a lesser
extent (42). This strain (W83) has also been
reported to lack the 41-43 kD fimbrial
protein present on strain ATCC 33277 (55).
In fimbriate eells, phagocytosis may also be
influenced by the glycocalyx which, if pres-
ent, may hide the fimbriae in its negatively
charged mass of capsule polymer (HANDLEY,
personal communication).
The extent of tissue damage and invasion
in experimental infections in animals has
been widely used as a measure of bacterial
virulence. Several studies have demonstrated
heterogeneity of virulence among different
strains oi
P.
gingivalis
(4-6, 16, 18, 20, 56).
Our results confirmed this, with two strains,
W83 and B262, causing extensive tissue dam-
age and death of the animals compared to the
other 15 strains, which caused local abscess
formation only.
It has been suggested that the degree of
encapsulation and resistance to phagocytosis
are factors which may influence the ability
of strains to be invasive in induced infections
(20-22).
Of the nine poorly phagocytosed
strains which possessed a glyeocalyx-like
structure, only two were capable of produc-
ing an invasive lesion when injected subcuta-
neously. Thus, while resistance to phagocyto-
sis and pronounced encapsulation were fea-
tures of the invasive strains, these factors
alone were not determinant for bacterial vir-
ulence.
In summary, we found that there were
clear differences among P.
gingivalis
strains
in their interaction in vitro with human
leukoeytes. One group of strains whieh
were readily phagoeytized, had a hydro-
phobic surface eharaeter and a high buoy-
ant density. The other strains, which were
more resistant to phagocytosis, were hy-
drophihc and had a lower density. Elec-
tron microscopy revealed a glycocalyx
structure on the poorly phagoeytized
strains. We found no elear correlation be-
tween the results from the phagocytosis as-
say and the virulenee of the baeteria wher
they were injected subeutaneously in mice.

PHAGOCYTOSIS
AND
VIRULENCE
OF
P. GINGIVALIS 127
Acknowledgment
-
This study
was
supported
by t he
Swedish Medical Research Council (project
No.
6270).
References
1.
TANNER
ACR,
SOCRANSKY
SS,
GOODSON
JM.
Microbiota of periodontal pockets losing crest-
al alveolar bone.
J
Periodont
Res 1984;
19:
279-91.
2.
LoESCHE WJ,
SYED
SA,
SCHMIDT
E,
MORRISON
EC.
Bacterial profiles
of
subgingival plaques
in periodonlkis.
J
Penodo?itol
1985, 56: 447-56.
3.
SLOTS
J.
Bacterial specificity
in
adult period-
ontitis.
A
summary
of
recent work.
J
Clin
Penodontol
19 86 ; 13 :
570-7.
4.
KASTELEIN
P,
VAN
STEENBERGEN
TJM,
BR A S
JM,
DE
GRAAFF
J.
An
experimentally induced
phlegmonous abscess
by a
strain
of
Bacteroi-
des gingivalis
in
guinea pigs
and
mice. Antonie
Van
Leeuwenhoek
1981;
47:
1-9.
5. VAN
STEENBERGEN
TJM,
KASTELEIN
P,
TOUW
JJA,
DE
GRAAFF
J.
Virulence
of
black-pig-
mented Bacteroides strains from periodontal
pockets
and
other sites
in
experimentally
in-
duced skin lesions
in
mice.
J
Periodont
Res
1982;
17:
41-9.
6. SuNDqvisT GK,
CARLSSON
J,
HERRMANN
BF ,
HoFLiNG
JF,
VAATAINEN
A.
Degradation
in
vivo
of the C3
protein
of
guinea-pig comple-
ment
by a
pathogenic strain
of
Bacteroides
gingivalis.
Scand
J
Dent Res 1984;
92:
14-24.
7.
SuNDQ^viST G,
CARLSSON
J,
HERRMANN
B,
TARNVIK
A. Degradation
of
human immuno-
globulins
G
and
M
and
complement factors
C3
and C5
by
black-pigmented Bacteroides.
J
Med
Microbiol 1985;
19:
85-94.
8.
CARLSSON
J,
HERRMANN
BF,
HOFLING
JF,
SuND5)viST
GK.
Degradation
of
the
human
proteinase inhibitors alpha-1-antitrypsin
and
alpha-2-macroglobulin
by
Bacleroides gingi-
valis.
Infect Immun 1984;
43:
644-8.
9.
CARLSSON
J,
HOFLING
JF,
SUND^VIST
GK. De-
gradation
of
albumin, haemopexin, haptoglo-
bin
and
transferrin,
by
black-pigmented
Bac-
teroides species.
J
Med
Microbiol
1984;
18:
39-46.
10.
NiLssoN
T,
CARLSSON
J,
SUNDQVIST
G. Inacti-
vation
of
key factors
of the
plasma proteinase
cascade systems
by
Bacteroides gingivalis.
In-
fect Immun 1985;
50:
467-71.
11.
MAYRAND
D,
GRENIER
D. Detection
of
coUa-
genase activity
in
oral bacteria.
Can
J
Micro-
biol 1985; 31: 134-8.
12. VAN
STEENBERGEN
TJM, DE
GRAAFF
J. Pro-
teolytic activity
of
black-pigmented Bacteroi-
des strains. FEMS Microbiol Lett
1986;
33:
219-22.
13.
SuND^visT G,
CARLSSON
J,
HANSTROM
L. Col-
lagenolytic activity of black-pigmented Bacte-
roides species.
J
Periodont
Res 1987;
22:
300-6.
14.
ROETERINK
CH, VAN
STEENBERGEN
TJM, DE
JONG
WFB,
DE
GRAAFF
J. Histopathological
effects
in the
palate
of
th e
rat
induced
by
injec-
tion with different black-pigmented Bacteroi-
des strains.
J
Periodont
Res
1984; 1 9:
292-302.
15.
ROETERINK
CH, VAN
STEENBERGEN
TJM, DE
GRAAFF
J.
Histopathological changes
in the
hind foot
of
the mouse induced
by
black-pig-
mented Bacteroides strains.
J
Med
Microbiol
1985;
20:
355-61.
16.
GRENIER
D,
MAYRAND
D. Selected character-
istics
of
pathogenic
and
nonpathogenic strains
of Bacteroides gingivalis.
J
Clin Microbiol
1987;
25:
738-40.
17.
MCKEE
AS,
MCDERMID
AS,
WAIT
R,
BASKER-
viLLE
A,
MARSH
PD.
Isolation of colonial vari-
ants
of
Bacteroides gingivalis
W50
with
a
r e-
duced virulence.
J
Med
Microbiol
1988;
27:
59-64.
18.
NEIDERS
ME,
CHEN
PB,
SUIDO
H, et al.
Heterogeneity
of
virulence among strains
of
Bacteroides gingivalis.
J
Periodont
Res 1989;
24:
192-8.
19.
GRENIER
D,
CHAO
G,
MCBRIDE
BC.
Charac-
terization
of
sodium dodecyl sulfate-stable
Bacteroides gingivalis proteases
by
polyacryl-
amide
gel
electrophoresis.
Irifect
Immun
19 89 ;
57:
95-9.
20.
CHEN
PB,
NEIDERS
ME,
MILLAR
SJ,
REYN-
OLDS
HS,
ZAMBON
J J .
Eflect
of
immunization
on experimental Bacteroides gingivalis infec-
tion
in
a
murine model.
Infect
Immun 1987;
55:
2534-7.
21. VAN
STEENBERGEN
TJM,
DELAMARRE
F G A ,
NAMAVAR
F,
DE
GRAAFF
J .
Diflerences in viru-
lence within
the
sp ec ie s Bacteroides gingivalis.
Antonie van
Leeuwenhoek
1987; 53: 233-44.
22.
HAAPASALO
M,
SHAH
H,
GHARBIA
S,
SEDDON
S, LouNATMAA
K.
Surface properties
and
ultrastructure
of
Porphyromonas gingivalis
W50
and
pleiotropic mutants. Scand
J
Dent
Res 1989;
97:
355-60.

128
SUNDQVIST
ET
AL,
23,
HOLDBMAN
LV,
CATO
EP,
MOORE
WEC.
An-
aerobe laboratory
manual.
4th ed,
Blacksburg,
VA:
VPI
Anaerobe Laboratory, Virginia
Polytechnic Institute
and
State University,
1977,
24,
PATRICK
S,
REID
JH,
Separation ofcapsulate
and non-capsulate Bacteroides fragilis
on a
discontinuous density gradient,
J
Med
Micro-
bid 1983;
16:
239-41,
25,
SUNDQVIST G, JOHANSSON
E,
Bactericidal eflfect
of pooled human serum
on
Bacteroides mela-
ninogcnicus, Bacteroides asaccharolyticus
and
Actinobacillus actinomycetemcomitans, Scand
J DenlRes 1982;
90:
29-36,
26,
ROZENBERG-ARSKA
M,
SALTERS
MEC, VAN
STRIJPJAG, GEUZE
JJ ,
VERHOEF
J, Electron
microscopic study
of
phagocytosis
of
Escheri-
chia coli
by
human polymorphonuclear leuko-
cytes.
Infect Immun
1985;
50:
852-9,
27,
RozGONYi
F,
SziTHA
KR,
LjuNGH
A,
BALODA
SB,
HjERTEN
S,
WADSTROM
T,
Improvement
of
the
s al t aggregation tesl
to
study bacterial
cell-surface hydrophobicity, FEMS Microbiol
Leil 1985; 30: 131-8,
28,
ROSENBERG
M,
GUTNICK
D,
ROSENBERG
E,
Adherence
of
bacteria
to
hydrocarbons:
A simple method
for
measuring cell-surface
hydrophobicity, FEMS
Microbiol
Lett 1980;
9:
29-33.
29,
SANDSTROM
G,
LOFGREN
S,
TARNVIK
A, A cap-
sule-deficient mutant
of
Francisella tularensis
LVS exhibits enhanced sensitivity
to
killing
by
serum
but
diminished sensitivity
to
killing
by
polymorphonuclear leukocytes. Infect Immun
1988;
56: 1194-1202,
30,
ScHENKEiN
HA,
Failure
of
Bacteroides gingi-
valis AV83
to
accumulate bound
C3
following
opsonization with serum,
J
Periodont
Res
1988;
24:
20-7,
31,
WILKINSON
BJ,
PETERSON
PK,
QUIE
PG, Cryp-
tic peptidoglycan
and th e
antiphagocytic
ef-
fect
of the
Staphylococcus aureus capsule:
model
for
th e antiphagocytic effect of bacterial
cell surface polymers. Infect Immun
1979;
23:
502-8,
32,
YoNEDA
M,
MAEDA
L,
AONO
M, Suppression
of bactericidal activity of human polymorpho-
nuclear leukocytes
by
Bacteroides gingivalis.
Infect Immun
1990;
58:
406-11,
33, VAN OSS
CJ, Phagocytosis
as a
surface
phe-
nomenon, Ann Rev
Microbiol
1978; 32: 19-39,
34,
STENDAHL
O, T h e
physicochemical basis
of
surface interaction between bacteria
and
phagocytic cells. Role
of
th e
envelope
in the
survival
of
bacterial infection.
In:
EASMON
CSF,
ELJASZEWIZ
J,
BROWN
MR,
LAMBERT
PA,
eds.
Medical
microbiology.
Vol.
3,
London:
Academic Press,
1983;
137-52,
35,
OHMAN
L,
MALUSZYNSKA
G,
MAGNUSSON
K-
E,
STENDAHL
O,
Surface interaction between
bacteria
and
phagocytic cells. Prog Drug
Res
1988;
32:
131-47,
36,
REYNOLDS
HS,
VAN
WINKELHOEE
AJ,
SCHIF-
FERLE
RE,
CHEN PB, ZAMBON
J J , Relationship
of encapsulation
of
Bacteroides gingivalis
to
invasiveness,
J
Dent Res 1989;
68: 328,
37,
ONDERDONK
AB,
KASPER
DL,
CISNEROS
RL,
BARTLETT
JG,
The
capsular polysaccharide
of Bacteroides fragilis
as a
virulence factor:
Comparison
of
the pathogenic potential
of
en -
capsulated
and
unencapsulated strains,
J
Infect
Dis 1977; 136:
82-9,
38,
BROOK
I,
Ro le
of
encapsulated anaerobic
b ac -
teria
in
synergistic infections,
CRC
Crit
Rev
Microbiol 19 87 ;
14:
171-93,
39,
MANSHEIM
BJ,
KASPER
DL, Purification and
immunochemical characterization
of
th e outer
membrane complex
of
Bacteroides melanino-
genicus subspecies asaccharolyticus,
J
Infect
Dis 1977;
135:
787-99,
40,
LiSTGARTEN
MA,
LAI C-H,
Comparative
ultrastructure
of
Bacteroides melaninogenicus
subspecies,
J
Periodont
Res 1979;
14:
332-40,
41,
Woo DDL,
HOLT
SC,
LEADBETTER
ER,
Ultrastructure
of
Bacteroides species: Bacte-
roides asaccharolyticus, Bacteroides fragilis,
Bacteroides melaninogenicus subspecies mela-
ninogenicus,
and B,
melaninogenicus subspe-
cies intermedius,
J
Infect
Dis 1979; 139:
534-46,
42,
HANDLEY
PS,
TiPLER
LS, An
electron micro-
scope survey
of
the surface structures
and hy -
drophobicity
of
oral
and
non-oral species
of
the bacterial genus Bacteroides, Arch
Oral Biol
1986;
31: 325-35,
43,
OKUDA
K,
FUKUMOTO
Y,
TAKAZOE
I,
SLOTS
J,
GENGO
RJ, Capsular structures
of
black-
pigmented Bacteroides isolated from human.
Bull
Tokyo
Dent Coll 1987;
28: l-11,
44,
CosTERTON JW,
IRVIN
RT,
CHENG
K-J, The
role
of
bacterial surface structures
in
patho-
genesis,
CRC
Crit
Rev
Microbiol
1981;
8:
303-38.
45,
LAMBE
J R
DW,
MAYBERRY-CARSON
KJ,
FER-

PHAGOCYTOSIS
AND
VIRULENCE
OF
P.
GINGIVALIS
129
GusoN KP, CosTERTON J W . Morphological sta-
bilization
of
the
glycocalyces
of
23
strains
of
five Bacteroides species using specific antisera.
Can
J
Microbiol
1984;
30:
809-19.
46.
SUTHERLAND
IW.
Bacterial exopolysacchar-
ides:
their nature
and
production.
In:
SUTHER-
LAND
IW, ed.
Surface carbohydrates
of the
prokary-
otic cell.
New
Yo rk : Academic Press,
19 7 7;
27-96.
47.
VERINGA
EM,
FERGUSON
JR DA,
LAMBE
JR
DW,
VERHOEF
J.
Trospectomycin enhances
surface phagocytosis
of
Bacteroides
and
Staphylococcus
by
altering
the
bacterial
gly-
cocaly.x. Zfntralbl
Bakteriol
1989; 271: 311-20.
48.
SiLVERBLATT
FJ,
DREYER
JS, ScHAUER
S.
Ef-
fect
of
pil i
on
susceptibility
of
Escherichia coli
to phagocytosis.
Infect Immun
1979; 24: 218-23.
49.
SPEERT
DP,
LOH
BA,
CABAL
DA,
SALIT
IE.
Nonopsonic phagocytosis
of
nonmucoid Pseu-
domonas aeruginosa
by
human neutrophils
and monocyte-derived macrophages
is
corre-
lated with bacterial piliation
and
hydrophobi-
city.
Infect Immun
1986; 53: 207-12.
50.
PRUZZO
C,
DAINELLI
B,
RICCHETTI
M. Pili ated
Bacteroides fragilis strains adhere
to
epithelial
cells
and are
mo re sensitive
to
phagocytosis
by
human neutrophils than nonpilated strains.
Infect Immun
1984; 43: 189-94.
51.
OKUDA
K,
SLOTS
J,
GENCO
RJ. Bacteroides
gingivalis, Bacteroides asaccharolyticus,
and
Bacteroides melaninogenicus subspecies: cell
surface morphology
and
adherence
to
erythro-
cytes
and
human buccal epithelial cells. Curr
Microbiol 19 81 ;
6:
7-12.
52.
YosHiMURA F,
TAKAHASHI
K,
NODASAKA
Y, S U-
ZUKI
T.
Purification
and
characterization
of a
novel type
of
fimbriae from
the
oral anaerobe
Bacteroides gingivahs.
J
Bacteriol
1984;
160:
949-57,
53.
PEROS
WJ,
ETHERDEN
I,
GIBBONS
RJ,
SKOBE
Z.
Alteration
of
fimbriation
and
cell hydro-
phobicity
by
sublethal concentrations
of
tetra-
cycline.
J
Periodont
Res 1985;
20:
24-30.
54.
IsoGAi
H,
IsoGAi
E,
YosHiMURA
F,
SUZUKI
T,
KAGOTA
W
TAKANO
K. Specific inhibition
of
adherence
of
an oral strain
of
Bacteroides
gin-
givalis
381
to
epithelial cells
by
monoclonal
antibodies against
the
bacterial fimbriae. Arch
Oral Biol 1988; 33: 479-85.
55.
LAMY
L,
MOUTON
C. Analysis
of
fimbriae
on
11 Bacteroides gingivalis strains
by
immu-
noblotting.
J
Dent Res 1988;
67: 239.
56.
DAHLEN
G,
SLOTS
J.
Experimental infections
by Bacteroides gingivalis
in
non-immunized
and immunized rabbits.
Oral Microbiol Immunol
1989;
4:
6-11.