MICROBIAL ECOLOGY
IN HEALTH AND DISEASE VOL. 4 293-301 (199 1)
Salivary IgA Antibodies Against Bacteria Incriminated as
Periodontal Pathogens in Kenyan Adolescents:
Correlation with Disease Status and Demonstration of
Antibody Specificity
J. M. A. WILTON*?, J. M. SLANEYt, J. A. C. STERNEt, D. BEIGHTONf and N. W. JOHNSON?
?Medical Research Council Dental Research Unit, Periodontal Diseases Programme, London Hospital Medical
College, 30132 Newark Street, London El 2AA and $Dental Research Unit, Royal College of Surgeons, London
Hospital Medical College, Turner Street. London El 2AA, UK
Received 22 February 1991; revised 10 April 1991
Kenyan adolescents, without destructive periodontal disease, possess salivary IgA antibodies to Actinobacillus actinomycetemcomitans strain Y4, Porphyromonas gingivalis strain W83, Prevotella intermedia, Capnocytophaga ochracea
and Streptococcus mutans. There were strong correlations between the levels of antibodies to each species tested, to the
total antibacterial IgA response and to the levels of the total IgA isotype. The mean antibody levels to all the organisms
tested was significantly positively correlated with the Bleeding Index and negatively with the Plaque Index. Reduction
of the saliva with periodate decreased the IgA binding to each species tested, suggesting that a part of the specific IgA
antibodies may be associated with bacteria binding salivary glycoproteins. The association between the IgA antibodies
to each species was unaffected by periodate reduction.
KEY Woms-IgA;
Salivary; Bacteria; Periodontal; Antibodies.
INTRODUCTION
Specific salivary IgA antibodies against bacteria
incriminated in human periodontal diseases have
not been widely studied.26 Plasma-derived antibody levels, particularly IgG, are found in different
types of periodontal disease such as juvenile periodontitis (JP) or adult periodontitis (AP).9 Since the
bacteria incriminated in periodontal diseases are
found in subgingival plaque and IgG antibodies
interact directly with them in the crevice or pocket,
most attention has been paid to systemic antibody
responses.
It is generally accepted that salivary IgA antibodies exert their presumed protective effect on
supragingival plaque bacteria-the concept of the
‘Salivary Domain’ first postulated by Lehner.14 It
has been assumed that they will not necessarily
correlate with the presence of organisms in subgingival plaque, whether or not these organisms are
associated with periodontal diseases. The presence
of salivary IgA antibodies should, however, reflect
*Author to whom correspondence should be addressed.
0891460X/9 1/05029349 $05.00
0 1991 by John Wiley Kt Sons, Ltd.
the general ability of the host to respond to a commensal flora. Pathogenic members of the flora that
elicit an enhanced, selective secretory IgA response
would provide evidence for microbial specificity.
Elevated levels of salivary IgA antibodies against
Actinobacillus actinomycetemcomitans have been
found in saliva from patients with JP20323and
against Bacteroides gingivalis and Treponema denticola in patients with APs compared to controls. In
contrast, Mansheim et a1.16 showed no differences
in salivary IgA antibodies to B. asaccharolyticus
between patients with JP and rapidly progressive
periodontitis (RPP) and controls.
In the investigations of disease cited above all
control subjects had salivary IgA antibodies to
presumptive pathogens and, in the study of Eggert
et aL5 antibodies to Streptococcus salivarius, an
organism not incriminated in periodontal diseases,
were also elevated in patients with AP. Suspected
periodontal pathogens such as black pigmented
bacteroides, Fusobacterium nucleaturn, and Capnocytophaga spp. can be isolated from the tongue,
tonsils and saliva of children’ and black pigmented
294
bacteroides and fusobacteria from these sites in
healthy adults.25 These bacteria, as well as those
incriminated in the aetiology of dental caries,2,'
will elicit a secretory IgA immune response.
We have studied the secretory IgA responses to a
number of plaque bacteria incriminated as periodontal pathogens in saliva from a population of
rural Kenyan adolescents with gingivitis but no
evidence of destructive periodontal disease. The aim
of the study was to assess the secretory IgA response
to suspected periodontal pathogens and relate the
immune responses to clinical status.
J. M. A. WILTON Er AL.
Collection and processing of serum samples
MATERIALS AND METHODS
Venous blood was taken from a single, healthy
adult at the MRC DRU by venepuncture of the
antecubital fossa and allowed to clot at room
temperature. After incubating the clot at 4°C for
4 h, the serum was separated by centrifugation and
stored in 100-pl aliquots at -70°C until used in the
assays. This serum was then used as a reference
serum for between-plate comparisons of IgA antibody activity. Each aliquot was only thawed once
and any residue was discarded, a fresh sample being
used for each assay. For cultural and ethical reasons
it was not possible to obtain serum from the Kenyan
study population.
Study population
Preparation of bacteria
The following bacteria were used as a source
The studies reported here form part of the ongoing longitudinal studies on periodontal diseases of antigens: Actinobacillus actinomycetemcornitans
which are part of the Primary Health Care Project, (AA) strain Y4 and Capnocytophage ochracea (CO)
Kenya. The study was conducted in a rural area (Dr N. S. Taichman, University of Pennsylvania,
of Kenya in the Northern Division of Machakos USA), Porphyromonas gingivalis (PG) (formerly
District. The sample comprised 49 persons aged Bacteroides gingivalis2') strain W83 and Prevo15-1 9 yr (23 male and 27 female) which had been tella intermedia (PI) (Dr G. Bowden, University of
selected as an addition to a larger cohort of 100 Manitoba, Canada), Streptococcus mutans (SM),
persons in this age group. The larger group had been strain Guy's serotype c (Dr J. M. A. Wilton). The
selected from the population by a random cluster bacteria were maintained on Blood agar base No
sampling methodi5 and the clinical characteristics 2 (Oxoid, UK) with 5 per cent horse blood in
of the study population have been described by an anaerobic incubator (Don Whitley, UK) and
Baelum et al.' In the present study the clinical subcultured into BM medium22 for antigen prepindices used for comparison with the IgA antibody aration. After 2 4 d of incubation, the cultures were
levels were supragingival dental plaque, gingival checked for purity by Gram's staining and the
bleeding, loss of periodontal attachment and pocket cultures centrifuged at 8000g for 30 min. The bacterial pellets were washed three times in 0.05M
depth as measured by Baelum et al.'
carbonate/bicarbonate buffer, pH 9.2, resuspended
in the same buffer and heated for 45 min at 60°C.
Collection andprocessing of saliva samples
The suspensions were then diluted to an absorbance
of 1.0 at 640nM and further diluted 1 : 10. This
Unstimulated whole saliva (3-7 ml) was collected suspension was then used to coat 96-well vinyl
from each subject by expectoration into a sterile cluster microtitre plates (Costar, USA-obtained
plastic bottle and the container stored on ice until from Northumbria Biologicals Ltd, Cramington,
transport to the laboratory, usually within 6 h of Northumberland, UK). Each well received 100 pl of
collection. Samples were centrifuged at 1000g for bacterial suspension and the plates were incubated
15min, the salivary sediment discarded and the at 37°C for 4 h and then at 4°C for 24 h. The plates
clarified saliva was divided into 1.0-ml aliquots. It were then washed five times with phosphate bufwas then frozen at -20°C until being placed into fered saline (PBS) containing Tween 20 (50 pl/litre)
liquid nitrogen for air transport to the UK. The using a Skatron Microwash plate washer (Skatron,
samples were then stored at - 20°C until thawed for No w a y ) .
use in the antibody assays. As a control for the
reproducibility of the assay, saliva was collected
in the same way from healthy adult laboratory per- Estimation of total salivary IgA
Microtitre plates were coated with rabbit
sonnel at the MRC Dental Research Unit (DRU),
pooled and stored at - 20°C until used in the assays. anti-human IgA (A092 Dako, UK) in carbonate/
295
IgA SALIVARY ANTIBODIES TO PLAQUE BACTERIA
bicarbonate buffer at 4°C for 18 h. The plates were
washed five times and non-specific binding sites
were blocked by the addition of 1 per cent normal
rabbit serum (X902 Dako, UK) and incubated for
18 h at 4°C. The plates were again washed five times
and stored at - 20°C until used in the assay. Saliva
samples were diluted from 1 : 500 to 1 : 8000 by
doubling dilution. As a standard for IgA concentration, human serum protein calibrator 1x908
Dako, UK) was diluted to give 10 dilutions ranging
from 10 to 200 ng/ml. Saliva samples and calibrator
were dispensed in duplicate onto the anti-IgAcoated plates and incubated for 2 h at 4°C. After
washing three times, swine anti-rabbit immunoglobulins, conjugated with horseradish peroxidase
(P217 Dako, UK) was added at a dilution of 1 : 500.
The plates were incubated for 2 h at 4"C, washed
three times and freshly prepared substrate solution
(0.4 mg/ml 0-phenylenediamine [Sigma, UK] in
citrate-phosphate buffer, pH 5.0 with 0.4 p1 of
added hydrogen peroxide) was added. After the
colour had developed, the reaction was stopped
with 6 N sulphuric acid to give a final concentration
of 1.5 N. The plates were read in a Titertek Multiskan MCC ELISA reader (Flow, UK) at 492 nM.
Solid phase enzyme-linked assay (ELISA) for
salivary IgA antibodies
Non-specific binding sites on the. plates coated
with bacteria were blocked by incubating the plates
with a solution of 1 per cent bovine serum albumin
(Sigma, UK) in PBS-Tween 20. The saliva samples
were added in duplicate, diluted from 1 : 10 to
1 : 160 by doubling dilution in this solution. The
undiluted reference serum sample was given a value
of 1500 arbitrary units (AU) and diluted to give ten
dilutions between 1 : 50 and 1 : 1200. Sample controls at 1 : 10 dilution were always included as were
controls of diluent alone. The plates were incubated
for 2 h at 37°C and washed as for the salivary IgA
estimations. Rabbit anti-human IgA was diluted in
1 per cent normal swine serum (901 Dako, UK) and
added at a dilution of 1 : 4000 to the test, reference
and diluent control wells and normal rabbit serum
diluted in 1 per cent normal swine serum (1 :4000)
was added to the sample control wells. The plates
were further incubated for 90min at 37°C and
washed. Peroxidase-conjugated swine anti-rabbit
immunoglobulins at a dilution of 1 : 500 were added
to all test, reference and control wells for 90 min at
37°C. After washing, freshly prepared substrate was
added to each well, the reaction was stopped and
the plates were read as already outlined in the description of the method for the estimation of total
salivary IgA.
Periodate treatment of saliva samples
Where the effect of reduction by periodate12
was investigated, salivary samples were diluted in
the standard diluent containing 0.01 M sodium
periodate (BDH, UK). Samples diluted with the
standard diluent alone were always run on the same
plate in parallel and both sets of samples were then
assayed for IgA antibodies as detailed above.
Data analysis
As already described, each plate contained ten
dilutions in duplicate of wells treated with a reference serum. Titersoft Diagnostic Software (Flow,
UK) was used to estimate the parameters of a
logistic regression using the optical densities of these
wells. These parameters were then used to give an
estimate of the relative antibody levels (RALs) in
the test sample wells with respect to the reference
serum.
For each sample, RALs were estimated from five
dilutions in duplicate. These were then converted
into a single estimate of RAL for the sample as
follows:
Write rij to be the estimated RAL for the ith
dilution, jth duplicate,
( i = 1,. . .,5),j= 1 or 2
Write di to be the amount by which r!jwas diluted.
For each dilution i, the data were discarded where
the optical densities of one or both of the wells were
outside the range covered by the reference serum.
Where both wells provided an estimate of the RAL,
the following calculations were performed. For each
dilution i, the data were discarded where the optical
densities of one or both of the wells were outside
the range covered by the reference serum. Where
both wells provided an estimate of the RAL, the
following calculations were performed:
rijx di+xij, the estimated RAL in the original
sample.
mi = (xil + x i , ) / 2 , the mean for dilution i.
vi = (xil- x i 2 ) ,the variance for dilution i. Where
the difference xil -xi2 was less than 0.01 it was
set to 0.01.
296
J. M. A. WILTON ETAL.
Table 1 . Summary statisticsfor the mean levels of the clinical indices for each subject
Attachment level
Pocket depth
Plaque index
Bleeding index
n
Mean
Median
SD
Minimum
Maximum
49
49
49
49
0.0292
1.8827
1.0516
0.3933
0.0000
1.8400
1.0313
0.3710
0.0867
0.2592
0.3831
0.1841
0.0000
1.5200
0.4167
0.0991
0.5200
2.6900
1.9352
0.8839
Table 2. Correlations between LRALs for each organism, MLSAL and
total salivary IGA levels
PG
PI
SM
Y4
co
MLSAL
IgA
PG*
PI
SM
Y4
CO
MLSAL
0.883
0.897
0.858
0.804
0.943
0.500
0.834
0.866
0.864
0.950
0.499
0.841
0.835
0.939
0.528
0.813
0.932
0.559
0.916
0.417
0.535
*See text for abbreviations.
+
plaque and bleeding observed at all sites in the
mouth were calculated. Summary statistics for these
means across all subjects are shown in Table 1 . It
can be seen from these data that the subjects had
little or no destructive periodontal disease, although
the data for the plaque and bleeding indices are
indicative of gingivitis of varying severity.
Thus the final estimate R of the RAL for the sample
Pearson correlations between the specific LRALs
is given by a weighted average of the estimates for for each subject were calculated. Because of the
each dilution, where the smaller the difference in the high values of these correlations a mean log specific
duplicates, the greater the weight given to the mean antibody level (MLSAL) was calculated. This was
for that dilution.
done by dividing the LRAL for each organism by its
On initial examination of the data it became population mean, and then calculating the mean of
apparent that the distributions of RAL for P. these values for each subject. Table 2 shows the
gingivalis were skewed, while the distributions of Pearson correlations between the LRAL for each
log RAL were approximately normal. Log RALs organism, the MLSAL and the total IgA level in
(LRALs) were therefore used in all statistical saliva. The correlations between the specific LRALs
analyses involving specific antibody levels.
and the MLSAL are all highly statistically signifiAll statistical calculations on these data were cant, as is the relationship between total salivary
performed using the software package MINITAB IgA and MLSAL (P<O.OOI).
(3081 Enterprise Drive, State College, PA 16801,
Pearson correlations between the specific
USA).
LRALs, MLSAL and total salivary IgA, and the
subject means of the clinical indices are shown in
Table
3. It can be seen that the antibody levels were
RESULTS
negatively correlated with the average plaque index
For each subject, the mean of the values of the whilst the average attachment level, pocket depth
clinical indices for attachment level, pocket depth, and bleeding index were all positively correlated.
(4) m = ml/vl +m2/v2 +m3/v3 m4/v4+m5/v5
(omit dilutions where the data were not available).
(5) d= l / vl + l/v2+ l/v3+ l/v4+ l/v5 (omit dilutions where the data were not available.
(6) R=m/d
297
IgA SALIVARY ANTIBODIES TO PLAQUE BACTERIA
Table 3. Correlations between antibody measurements and mean clinical
indices
PG*
PI
SM
Y4
co
MLSAL
IgA
Attachment
level
Pocket
depth
Plaque
index
Bleeding
index
0.310
0.168
0.379
0.289
0.262
0.302
0.175
0.196
0.102
0.304
0,165
0.115
0.194
-0.000
-0'247
-0.213
-0.196
-0-184
-0.210
-0.23 1
- 0.08 1
0.360
0.233
0.482
0.354
0.240
0.354
0.209
*See text for abbreviations.
Table 4. Analysis of variance on LRALs of pooled control saliva
Source
DF
SS
MS
F
Organism
Error
Total
4
55
59
17-3486
2.6324
19.9810
4.3371
0.0479
90-62
n
PG*
PI
SM
Y4
CO
12
12
12
12
12
Mean
6.3933
5.4090
5.3730
5.9527
6.7414
SD
0.2431
0.1473
0.1591
0.3351
0.1445
Variance
0.05910
0.02170
0-02531
0.11229
0.02088
Individual 95% confidence intervals
for mean based on pooled SD
--_--+ _--___-_+
+--___-___
+
(--*-)
* --)
* --)
((-
(-
-___-_
Pooled SD = 0.2 188
* --)
(-- * -)
+_________ +--_______ +_________ +
5.50
6.00
6.50
7.00
*See text for abbreviations.
One subject had no detectable salivary IgA and no
antibody levels were found in this isotype against
any of the bacteria tested and he was excluded from
the subsequent analyses.
An estimate of the experimental error was derived
by performing a one-way analysis of variance,
allowing for different means for each organism,
for the estimates of LRALs from the pooled saliva
controls. The results are shown in Table 4. The
estimated error mean square, assuming the variation to be the same for each organism, was 0-0479.
To investigate further the relationship between
the specific LRALs, a two-way analysis of variance
allowing for subject means and organism means was
performed. This used data from 44 of the 49 subjects
Table 5. Two-way analysis of variance on LRALs in 44
subjects
~~
Source
DF
Organism
Subject
Error
Total
43
172
219
4
SS
MS
F
61.689
119-434
17.536
198.659
15.422
2.778
0.102
151.20
27.24
for whom there were no missing values. The results
are shown in Table 5 . The estimated mean squares
were 15.422 for the variation between organisms,
298
J. M. A. WILTON E T A L .
Log Ellea Unltr
A
0 6
d p
1 0
d p
I
1 6
1 8
2 0
2 9
3 2
4 6
d p
d p
d p
d p
d p
d p
Subject Number: d-diiuent, pqmlodate.
Figure 1. Antibacterial IgA antibody levels before and after periodate treatment ineight subjects. Diluent = 1 per
cent BSA in PBS-Tween 20; periodate=0.01 M sodium periodate in diluent. Pint. = Prevotella intermedia;
P.ging = P . gingivalis; S.sang = Streptococcus sanguis; S m u t = Streptococcus mutans; Y4 (AA strain) and 652
(AA strain) = Actinobacilfus aciinomyceiemcomiians strains; Capno = Capnocytophaga ochracea
2.778 for the variation between subjects, and 0.102
for the residual variation. This residual variation
is composed of the experimental error (estimated
as 0.0479) and the subject-organism interaction
(estimated as 0.1024.0541). This latter can be
interpreted as the specific antibody effect, since it
is variation not accounted for by the difference
between subject means, organism means or experimental error. Thus the ratio between variation due
to subject means and variation due to specific
antibody was around 50 to 1.
It was evident that it was not meaningful to
investigate further the relationship between specific
LRALs and the values of the clinical indices. The
relationship between MLSAL and the subject
mean clinical indices was therefore examined using
multiple linear regression. This showed that of
the four clinical indices, the bleeding index was
significantly positively correlated ( P < 0.025) with
MLSAL, the plaque index was significantly negatively correlated ( P < 0.001) with MLSAL, while
the subject means for attachment level and pocket
depth were not correlated with MLSAL. The multiple correlation (R”) for the regression was 0.3.
These results suggest that the assay might reflect
the overall ability of a subject to mount an antibody
response to all the organisms tested. Since the antibodies were assayed in whole saliva it is possible that
some of the IgA-specific binding to the bacteria in
the assay might be associated with non-antibody
specific binding of salivary glycoproteins to the
bacteria. The serum used in the ELISA was specific
for a chains and, assuming that there is no crossreactivity of this antiserum with other moieties
present in saliva, then it is evident that the IgA
must be physically associated in some way with the
glycopro teins.
In order to investigate this, a subset of eight saliva
samples, selected to include subjects with high, low
and intermediate MLSALs, was assayed with and
without treatment with periodate to destroy any
carbohydrate-binding sites of the salivary glycoproteins. These data are shown in Figure l . It can be
seen that there was a reduction in virtually every
case of the amount of IgA binding to the bacteria.
Two-way analyses of variance were performed separately for the LRALs with and without treatment
with periodate. These are shown in Table 6. It can
IgA SALIVARY ANTIBODIES TO PLAQUE BACTERIA
Table 6. Two-way analyses of variance on LRALs with
and without treatment with periodate
~
Source
~~
DF
SS
MS
Samples without periodate treatment
4
8.7475
2.1869
7
12.2676
1.7525
Error
28
1.8036
0.0644
Total
39
22.8187
Organism
Subject
Samples with periodate treatment
4
9.6619
2.4155
7
7.7357
1,1051
Error
28
1.4945
0.0534
Total
39
18.8921
Organism
Subject
F
33.96
27.21
45.33
20.69
be seen that after treatment with periodate the
between-organism variation was increased and the
between-subject variation was decreased. However,
the residual variation was marginally decreased for
the periodate-treated data, and was for each data set
only marginally greater than the variation due to
experimental error estimated earlier. There was thus
no evidence that treatment with periodate enhanced
the ability of the assay to measure specific IgA
antibody.
DISCUSSION
We have shown that saliva from Kenyan adolescents with gingivitis contained IgA antibodies
against a number of oral bacteria incriminated in
the aetiology of periodontal diseases and dental
caries. The average specific antibody responses to
these bacteria were also significantly correlatednegatively with the supragingival plaque index and
positively with the bleeding index. The biological
relevance of these associations is unclear at present
but the negative correlation might indicate that the
greater the amount of plaque the more salivary antibody would be bound to the bacteria, leading to
lower levels in saliva. The positive correlation might
be associated with the inflammation of gingivitis
where greater amounts of plasma-derived IgA
would be released into the saliva from gingival
crevicular fluid. The antiserum used in this study
was specific for Fca and would not distinguish
between secretory and serum IgA. The strong associations between the levels of specific antibodies
to the individual bacterial species have not been
previously reported. Most studies of salivary IgA
299
antibodies only investigated single organisms such as
A . actinomycetemcomitans Y420323
or B. asaccharolyticus.'6 Where more than one organism or antigens from different organisms such as s. mutans
glucosyltransferase and p o l i o ~ i r uor
s ~P.~gingivalis,
B. fragilis and Escherichia coli lipopolysaccharides
and S. mutans glucosyltransferase, antigen 1/11and
cell wall carbohydrate were ~ t u d i e d no
, ~ relationships such as those demonstrated here were reported.
It is possible that there is a common antigen,
possessed by the bacteria used in this study, and that
such an antigen(s) would elicit the IgA antibodies
detected, but this is unlikely. Another explanation
for theassociations in this group ofsubjects would be
that there is no periodontal destruction and thus no
infection with any suspected pathogens; the levels of
antibodies are thus what might be expected from
exposure to a commensal flora where no organism is
dominant in otherwise healthy subjects.
The whole saliva used for the assays will contain
both mucin and non-mucin glycoproteins. Some of
these specifically agglutinate plaque bacteria such as
S. mutans and S. mitior by lectin-like interactions'*
and S. sanguis by binding to lipoteichoic acid."
Salivary mucins with blood group antigen A, B and
0 specificity bind selectively to S. mutans.'O The
aggregation of Lepiotrichia buccalis by salivary
mucins is totally inhibited by amino sugars such as
N-acetyl D-galactosamine, suggesting blood group
A re~ognition.'~IgA can be associated with the
high molecular weight, non-mucin glycoproteins as
shown by reactions of anti-IgA antibodies with
these glycoprotein~~
and inhibition of bacterial agglutination, caused by glycoproteins, with anti-IgA
antibodies.
Periodate oxidation reduced IgA antibody binding to bacteria in virtually every case indicating
that at least part of the binding of salivary IgA is
mediated by carbohydrate moieties. Since the decrease was detected by the diminution in the binding
of an Fca-specific antibody, the original antibody
binding must have been either due to uncomplexed
IgA or IgA complexed to carbohydrates with their
own recognition sites for the bacterial surface. The
one subject with no detectable salivary IgA or IgA
antibodies would rule out any binding of salivary
proteins cross-reactive with IgA. Although the
assumption must be that IgA binding to surface
antigens is mediated by the F(ab'),, whether or
not IgA is complexed, it is possible that the carbohydrate residues on the constant region of the
F(ab'), as well as residues in the hinge region are
essential, at least in part, for specific binding of the
300
J. M. A. WILTON E T A L .
bacterlai antigen with
specific IgA
iectin bound to bacteria
non-specific binding of
IgA to bacteria bound iectin
- periodate effect
lectin bound to bacteria
with no non-specific binding
of IgA- no periodate effect
binding of igA by
carbohydrate residues in
the Fc region
-periodate effect
Figure 2. Possible binding between IgA, lectin and bacteria
whole IgA to antigen. The different interactions
between the bacterial-binding moieties (including
IgA antibodies) is shown in Figure 2. Oxidation of
carbohydrate residues of the Fca, secretory piece
and the J chain of dimeric secretory IgA could also
inhibit antigen binding by F(ab'),. IgA2 has a
greater carbohydrate content than IgAl and lacks
N-acetylgala~tosamine'~so that there may be differential susceptibility to periodate oxidation
between the two subclasses. It has been claimed that
secretory IgA in saliva is susceptible to periodate
oxidation6 but that study did not use purified,
uncomplexed IgA and the susceptibility could well
have been that of the carbohydrate residues of a
complexed glycoprotein.
Since the levels of both specific IgA and of the
glycoprotein agglutinins and bacterial aggregating
factors will vary from subject to subject, some
caution must be used in unequivocally ascribing
IgA binding from whole saliva as being due to
uncomplexed, antigen-specific, secretory IgA. For
mucosal protection it might be advantageous for
IgA to combine its antigen specificity with other
proteins with combining sites for bacterial surfaces.
In future studies of IgA antibody activity in mixed
secretions such as saliva, the possibilities of other,
non-IgA binding moieties in the interpretation of
results will have to be considered. Until the periodate susceptibility of the IgA molecule has been
established, assays of whole saliva for antibody
activity should include periodate treatment of the
saliva.
In summary, we have demonstrated that Kenyan
adolescents with gingivitis and no periodontal
destruction all possess antibodies of the IgA isotype
in saliva. These antibodies are directed against
suspected periodontal pathogens which have been
incriminated in destructive periodontitis as well as
organisms which may cause dental caries. Whether
the clinical associations or the associations of the
IgA antibody levels to the various bacteria that we
have demonstrated will be maintained if destructive
disease develops will have to await further studies.
ACKNOWLEDGEMENTS
We thank Drs Vibeke Baelum, Firoze Manji and Ole
Fejerskov for their painstaking clinical measurements
and many stimulating discussions.We thank the Director
of KEMRI, Dr Steven Kinote, the field workers Benson
and Dominic and James Simwa for laboratory assistance.
This work was supported by DANIDA, WHO and the
Medical Research Council.
REFERENCES
1. Baelum V, Fejerskov 0, Manji F. (1988). Periodontal diseases in adult Kenyans. Journal of Clinical
Periodontology 1 5 , 4 4 5 4 5 2 .
2. Beighton D, Manji F, Baelum V, Fejerskov 0,
Johnson NW, Wilton JMA. (1989). Associations
between salivary levels of Streptococcus mutans,
Streptococcus sobrinus, lactobacilli and caries experience in Kenyan adolescents. Journal of Dental
Research 68, 1242-1246.
3. Brown TA, Mestecky J. (1985). Immunoglobulin A
subclass distribution of naturally occurring salivary
antibodies to microbial antigens. Infection and
Immunity 49,459462.
4. Eggert FM. (1979). Saliva and amniotic fluid agglutinin and aggregating factors. Evidence for complexing between high molecular weight non-mucin
glycoproteins and immunoglobulins. Biochemical
Society Transactions 7 , 194-196.
5. Eggert FM, Maenz L, Tam YC. (1987). Measuring
the interaction of human secretory glycoproteins
to oral bacteria. Journal of Dental Research 66,
42W24.
6. Eggert FM. (1980). The nature of secretory agglutinins and aggregating factors. 11. Biochemical and
immunochemical properties of factors in human
saliva and amniotic fluid. International Archives of
AIlergy and Applied Immunology 61,203-2 12.
7. Ericson T, Bratthall D, Rundegren J. (1979). Bacterial agglutination induced by saliva. Inhibition
studieswith anti-Ig antisera and serum components.
IgA SALIVARY ANTIBODIES TO PLAQUE BACTERIA
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
In: Kleinberg I, Ellison SA, Mandel ID (eds) Saliva
and Dental Caries. IRL Press, London, p. 243.
Frisken KW, Tagg JR, Laws AJ. (1987). Suspected
periodontopathic microorganisms and their oral
habitats in young children. Oral Microbiology and
Immunology 2,60-64.
Genco RJ, Slots J. (1984). Host responses in periodontal diseases. Journal of Dental Research 63,
4446.
Gibbons RJ, Quereshi JV. (1978). Selective binding
of blood group-reactive salivary mucins by Streptococcus mutans and other oral organisms. Infection
and Immunity 22,665471.
Hogg SD, Embery G. (1981). Aggregation of oral
streptococci by a mucous protein from saliva. In:
Frank RM, Leach SA (eds) Surface and Colloid
Phenomena in the Oral Cavity; Methodological
Aspects. IRL Press, London.
Kabat EA, Mayer MM. (1961). Experimental
Immunochernistry, 2nd edn. Thomas, Illinois, pp.
542-550.
Kondo W, Sato M, Sat0 N. (1978). Properties of the
human salivary aggregating factor for Leptotrichia
buccalis cells. Archives of Oral Biology 23,453478.
Lehner T. (1969). In: Fourth Proceedings of the International Academy of Oral Pathology. Gordon and
Breach, New York, p 109.
Manji F, Fejerskov Baelum V. (1989). Patterns of
dental caries in an adult rural population. Caries
Research 23,5542.
Mansheim BJ, Stenstrom ML, Low SB, Clark WB.
(1980). Measurement of serum and salivary antibodies to the oral pathogen Bacteroides asaccharolyticus in human subjects. Archives of Oral Biology 25,
553-557.
Mestecky J, Russell MW. (1986). IgA subclasses. In:
Monographs in Allery, 19. Karger, Basel, p 277.
301
18. Rundegren J, Ericson T. (1981). An evaluation of
salivary agglutinins. Journal of Oral Pathology 10,
261-268.
19. Russell RRB, Johnson NW. (1987). The prospects
for vaccination against dental canes. British Dental
Journal 162,29-34.
20. Sandholm L, Tolo K, Olsen I. (1987). Salivary IgG,
a parameter of periodontal disease activity. Journal
of Clinical Periodontology 14,289-294.
21. Shah HN, Collins MD. (1988). Proposal for reclassification of Bactetoides asaccharolyticus, Bacteroides
gingivalis and Bacteroides endodontalis in a new
genus. International Journal of Systemic BacteriOlOgy24,128-131.
22. Shah HN, Williams RAD, Bowden GH, Hardie JM.
(1976). Comparison of the biochemical properties
of Bacteroides melaninogenicus from human dental
plaque and other sites. Journal of Applied Bacteriology 41,473492.
23. Smith DJ, Ebersole JL, Taubman MA, Gadulla L.
(1985) Salivary IgA antibody to Actinobacillus actinomycetemcomitans in a young adult population.
Journal of Periodontal Research 20,8-11.
24. Smith DJ, Taubman MA, Ebersole JL. (1987).
Ontogeny and senescence of salivary immunity.
Journalof Dental Research 66,451456.
25. Van der Velden U, van Winkelhoff AJ, Abbas F de
Graaff J. (1986). The habitat of periodontopathic
organisms. Journal of Clinical Periodontology 13,
243-248.
26. Wilton JMA, Curtis MC, Gillett IR, Griffiths GS,
Maiden MFJ, Sterne JAC, Wilson DT, Johnson
NW. (1989). Detection of high-risk individuals for
periodontal diseases: laboratory markers from
analysis of saliva. Journal of Clinical Periodontology
16,475483.