Vaccine-Induced Autoimmunity in the
Advances in Veterinary Medicine Vol 41, pp 733-747 - HARM HOGENESCH,
JUAN AZCONA-OLIVERA, CATHARINE SCOTT-MONCRIEFF, PAUL W. SNYDER, AND
LARRY T. GLICKMAN
Departments of Veterinary Pathobiology
and Veterinary Clinical Sciences, Purdue University, West Lafayette,
II. Materials and Methods
B. Vaccination Schedule
C. Viral Serology
G. Lymphocyte Blastogenosis Assay
H. Enzyme-Linked Immunosorbent Assay (ELISA)
J. Statistical Analysis
A. Viral Serology
B. Clinical Observations, Hematology, and Endocrinology
Vaccines are widely used in human and veterinary medicine as an effective
and economic method to control viral and bacterial diseases. Although
generally considered safe, vaccination is occasionally accompanied
by adverse affects. Many adverse affects related to vaccination are
acute and transient, for example, fever, swelling at the site of the
inoculation, and allergic reactions. In contrast, reports of autoimmune
disease following vaccination are relatively rare. In most instances,
it is difficult, if not impossible, to ascertain that vaccination
caused or precipitated the autoimmune disease. In a recent report,
the Advisory Committee on Immunization Practices in people concluded
that there is a causal relation between diptheria-tetanus-pertussis
(DTP) and measles-mumps-rubella (MMR) vaccination and arthritis, but
no evidence of a causal relationship between these vaccinations and
other autoimmune diseases such as autoimmune hemolytic anemia and
Guillain-Barre syndrome (Centers for Disease Control and Prevention,
1996). Cohen and Shoenfeld (1996) also stated that the relation between
vaccination and autoimmunity is obscure. They added that there is
a need for experimental studies to address this subject (Cohen and
There has been a growing concern among dog owners and veterinarians
that the high frequency with which dogs are being vaccinated may lead
to autoimmune and other immune-mediated disorders (Dodds, 1988; Smith,
1995). The evidence for this is largely anecdotal and based on case
reports. A recent study observed a statistically significant temporal
relationship between vaccination and subsequent development of immuno-mediated
hemolytic anemia (IMHA) in dogs (Doval and Ciger, 1996). Although
this does not necessarily indicate a causal relationship, it is the
strongest evidence to date for vaccine-induced autoimmune disease
in the dog.
We are investigating the effect of vaccination on dogs in a series
of experimental studies. The goals of these experiments are (1) to
determine if vaccination of dogs affects the function of the immune
system and, in particular, if vaccination results in autoimmunity;
(2) to delineate the mechanisms by which vaccination results in autoimmunity
if this occurs; and (3) to develop alternative vaccination strategies
that will not be accompanied by adverse effects. The issue that is
the focus of this and ongoing studies in our laboratory is somewhat
different from that examined by Duval and Ciger (1996). In their study,
a statistically significant temporal relationship between the onset
of IMHA and prior vaccination suggested that vaccination caused IMHA
or accelerated preexisting IMHA in adult dogs. Although not documented,
it is likely that these middle-aged dogs had received multiple vaccines
prior to the last vaccination. Why this last vaccination suddenly
triggered the onset of IMHA is unknown. In contrast, our studies examine
if vaccination of dogs at a young age causes alterations in the immune
system, including the production of autoantibodies, that could eventually
lead to autoimmune disease in susceptible individuals. In this paper,
we report on the findings of the first study in which a group of vaccinated
dogs and a group of unvaccinated dogs were followed for 14 weeks after
the first vaccination.
II. Materials and Methods
Two pregnant Beagle dogs were purchased from a commercial breeder.
The animals whelped in the Animal Facility of the Purdue University
School of Veterinary Medicine and the pups were weaned at 6 weeks
of age. Five pups were assigned to one of two groups, a vaccinated
and an unvaccinated group, based on body weight, gender, and litter
of origin. The vaccinated and unvaccinated group of dogs were housed
in separate rooms.
The dogs were examined daily. Rectal temperature and body weight were
recorded twice a week. Blood samples were collected from the jugular
vein prior to each vaccination and 2, 5, 7, and 14 days following
vaccination for hematology, endocrinology, and viral serology. Blood
samples collected on days 5 and 14 following vaccination were also
used for lymphocyte phenotyping and lymphocyte proliferation assays,
and blood samples collect at 7 days following vaccination were used
for the detection of autoantibodies.
B. Vaccination Schedule
The dogs in the vaccinated group were injected subcutaneously with
a commercially available multivalent vaccine, Vanguard-5 CV/L (Pfizer,
Croton, CT) at 8, 10, 12, 16, and 20 weeks of age according to the
instructions of the manufacturer. They were injected subcutaneously
with an inactivated rabies vaccine, Imrab-2 (Rhone-Mericux, GA) at
16 weeks of age. The unvaccinated group of dogs received subcutaneous
injections of sterile saline at the same time points.
Both groups of dogs were injected subcutaneously with 1 mg of keyhole
limpet hemocyanin (KLH, Calbiochem) in RIBI-adjuvant at week 20.
C. Viral Serology
Serum samples collected at 6 weeks of age and 0, 2, 5, 7, and 14 days
after each vaccination were assayed for the presence of antibodies
to canine distemper virus by serum neutralization test, and for antibodies
against canine parvovirus by hemagglutination inhibition test. Serum
samples were analyzed for antibodies against rabies virus at 16 and
20 weeks of age by a rapid fluorescent focus inhibition test.
Blood samples were collected at 0, 2, 5, 7, and 14 days after each
vaccination for hematocrit, corrected white blood cell count and differential,
and platelet counts.
Plasma and serum samples collected at 0, 2, 5, 7, and 14 days after
each vaccination were assayed for curtisol, triiodothymonine (T3),
and thyroxine (T4) by radioimmunoassay.
Lymphocyte phenotyping was used. Whole blood was stained with a panel
of mouse monoclonal antibodies, followed by F(ab')2 goat anti-mouse
IgG (Jackson Research Laboratories). The monoclonal antibodies used
were CA2.1D6 (anti-CD21), CA15.8G7 (anti-TCRoB), CA20.8H1 (anti-TCRv81,
12.125 (anti-CD4), and 1.140 (anti-CD8). The characteristics of these
monoclonal antibodies have been described (Gebhard and Carter, 1992;
Moore et al., 1995). Following red blood cell lysis and fixation in
2% paraformaldehyde, the cells were analyzed by flow cytometry.
G. Lymphocyte Blastogenosis Assay
Heparinized blood samples were diluted 1:10 in RPM1-1640 and distributed
in the wells of a 96-well plate. Triplicate samples were incubated
for 96 hours in the presence of medium only, 2.5 and 5 pg/ml PHA,
5 and 10 pg/ml Concenavalin A (Con A) and 1 and 10 pg/ml PWM. During
the last 24 hours of incubation the wells were pubed with 0.5 uCi
of H-thymidine. The cells were harvested with a 96-well cell harvester,
and the incorporation of radioactivity was measured in a TopCount
scintillation counter (Packard Instrument Co., Meriden, CT).
H. Enzyme-Linked Immunosorbent Assay (ELISA)
The presence of antibodies reactive with homologous and heterologous
antigens in serum samples collected at 22 weeks of age was analyzed
by an indirect ELISA. High-binding ELISA plates (Costar, Cambridge,
MA) were coated with 10 pg/ml of antigen in 0.1 M bicarbonate buffer.
The wells were rinsed and incubated for 1 hour with phosphate-buffered
saline (PBS)/0.1% Tween. Serum samples were diluted 1:10 in PBS and
added to the wells in triplicate. Following incubation, the wells
were rinsed and incubated with alkaline phosphatase labeled goat anti-dog
IgG (Kirkegnard and Perry, Gaithersburg, MD). Alkaline phosphatase
activity was measured after addition of p-NPP substrate at 405 nm
in a microplate reader (Molecular Devices, Menlo Park, CA).
Essentially the same procedure was used to measure the presence of
antibodies against KLH. Alkaline phosphatase labeled anti-dog IgM
and IgG were used as secondary reagents.
At 22 weeks of age, the dogs were killed by intravenous injection
of barbiturates, and a complete necropsy performed. Tissue samples
were collected in 10% buffered formalin and processed for light microscopic
examination. The tissues that were examined included the spleen, lymph
nodes, tonsils, thymus, Psyer's patches, adrenal glands, thyroid glands,
pituitary gland, pancreas, heart, lung, kidney, liver, and brain.
J. Statistical Analysis
Data were analyzed for significant differences between groups by Student's
t test or repeated measures ANOVA and a significant change over time
using a repeated measures ANOVA.
A. Viral Serology
None of the pups had detectable antibodies against canine distemper
virus and canine parvovirus at 6 weeks of age and against rabies virus
at 16 weeks of age. The unvaccinated dogs remained seronegative for
these three viruses during the course of the study. The dogs that
were immunized developed titers against CDV (maximum titers ranged
from 1:48 to 1:1024), CPV-2 (1:320 to 1:1280), and rabies (1:25 to
B. Clinical Observations, Hematology, and Endocrinology
No differences between the unvaccinated and vaccinated groups were
found for rectal temperature, body weight, and hematologic values.
There were no significant differences between unvaccinated and vaccinated
dogs for concentrations of cortisol, T3, and T4. However, a significant
(p<0.02) change was observed over time for each of these three
hormones. The plasma concentration of cortisol decreased from a mean
of 41.1 ng/ml at 8 weeks of age to 17.6 ng/ml at 22 weeks of age.
The concentration of T4 also decreased, from 31.1 ng/ml at 8 weeks
of age to 22.8 ng/ml at 22 weeks of age. The concentration of T3 increased
from 0.63 ng/ml at 8 weeks of age to 1.1 ng/ml at 22 weeks of age.
No differences were observed unvaccinated and vaccinated dogs for
lymphocyte subpopulations or for the proliferative response to any
of the mitogens tested.
The response of both groups of dogs to KLH was similar. There was
no statistically significant difference in the KLH-specific IgM and
IgG concentrations in the serum (not shown).
At 8 weeks of age, antibodies against homologous and conserved heterologous
antigens were negligible in the serum of the dogs. At 22 weeks of
age there was a significant increase of IgG antibodies reactive with
10 of 17 antigens in the vaccinated dogs versus no increase in the
unvaccinated dogs (Table I). The increase of optical density was modest
for 8 of these 10 antigens, but a large increase was observed for
fibronectin and laminin. All vaccinated dogs developed high levels
of fibronectin-specific IgG antibodies. Similar levels of IgG anti-fibronectin
antibodies were observed when bovine fibronectin was substituted by
human or mouse fibronectin (not shown). The concentration of anti-fibronectin
antibodies began to increase after the second vaccination in three
dogs and after the third vaccination in the other two vaccinated dogs,
and reached a maximum level after the fourth vaccination (Fig. 1).
To determine if the antibodies had a preferential reactivity with
a particular part of the bironectin molecule, we tested the reactivity
of serum samples with two fragments of the fibronectin. The 30-kDa
fragments contains the heparin-binding domain of fibronectin, whereas
the 45-kDa fragment contains the collagen-binding domain. As shown
in Fig. 2, little reactivity was observed with the 45-kDa fragment,
but significant reactivity was observed with the 30-kDa fragment.
High levels of anti-laminin antibodies were observed in the serum
of three of the five vaccinated dogs at 22 weeks of age. One dog had
high levels at 17 weeks of age, whereas the other two dogs did not
devlop high levels until the end of the study.
High levels of antibodies reactive with skeletal muscle myosin and
myoglobin were observed in both groups of dogs at 22 weeks of age.
The antibody levels increased at 11 weeks of age in three dogs, at
13 weeks of age in another three dogs, and at 17 weeks of age in the
remaining four dogs.
Gross and light microscopic examination of the tissues of the dogs
revealed no significant lesions. The thyroid gland of one of the vaccinated
dogs had a small lymphoid nodule with obliteration of adjacent thyroid
In this study, we exhaustively evaluated the effects of vaccination
with a multivalent vaccine and a rabies vaccine on the immune system
of young dogs. Vaccination did not cause immunosuppression or alter
the response to an unrelated antigen (KLH). In contrast to an earlier
study (Mastro et al., 1986), but in agreement with other work (Phillips
and Schultz, 1987), we did not observe a transient lymphopenia in
the dogs at any time. However, vaccination did induce autoantibodies
and antibodies to conserved heterologous antigens. The pathogenic
significance of these autoantibodies is presently uncertain. We did
not find any evidence of autoimmune disease in the vaccinated dogs,
but the study was terminated when the dogs were 22 weeks of age, well
before autoimmune diseases usually become clinically apparent. It
is likely that genetic and environmental factors will trigger the
onset of clinical autoimmune disease in a small percentage of the
animals that develop autoantibodies. For practical and economic reasons,
only a small number of dogs can be followed in an experimental study,
and clinical autoimmune disease may, therefore, never be observed.
The principal value of an experimental study is that it enables us
to determine the frequency of autoantibody responses and the mechanism(s)
that cause vaccines to induce autoantibodies.
We used two vaccines, a multivalent vaccine and an inactivated rabies
vaccine of a particular commonly used brand. We consider it unlikely
that the observed autoantibodies were specifically induced in response
to those brands of vaccine and this phenomenon will likely occur with
other commercial vaccines. In a follow-up study, we have observed
similar autoimmune phenomena in dogs immunized with the multivalent
vaccine only and in dogs immunized with the rabies vaccine only (unpublished
There was a marked increase of autoantibodies to the skeletal muscle
proteins, myoglobin and myosin, in both groups of dogs. The reason
for the appearance of these antibodies is uncertain, but it may be
the result of the frequent blood sampling of the dogs. The dogs were
bled five times following each vaccination, and some tissue trauma
We examined the thyroid and adrenal cortical function in the dogs,
and did not find evidence of any abnormality. Autoimmune thyroiditis
is one of the most common autoimmune diseases of dogs, and it has
been suggested that the apparent increase of this condition in dogs
is related to the increased frequency of vaccination with modified
live vaccines. There was no increase of anti-thyroglobulin antibodies
in the vaccinated animals, or other evidence of thyroid dysfunction.
However, the lymphoid nodule found in the thyroid gland of one of
the vaccinated dogs may be an early manifestation of thyroiditis,
a common lesion in purpose bred Beagles (Fritz et al., 1970).
The most strikingly increased concentrations of autoantibodies were
directed against fibronectin and laminin. Fibronectin is widely distributed
in the body as a component of the extracellular matrix and plasma.
The anti-fibronectin antibodies were reactive with fibronectin of
bovine, murine, and human origin. Although we have not yet demonstrated
that they also react with canine fibronectin, this is very likely,
since fibronectin is highly conserved between species. Anti-fibronectin
antibodies have been found in human patients with systemic lupus erythematosus
(SLE) and rheumatoid arthritis, and a patient with a poorly defined
connective tissue disease (Henane et al., 1986; Atta et al., 1994,
1995; Girard et al., 1995). The anti-fibronectin antibodies in four
human SLE patients were directed against the collagen-binding domain
(Atta et al., 1994), in contrast to the anti-fibronectin antibodies
in the vaccinated dogs, which showed no affinity for this domain.
The anti-fibronectin antibodies in the human patient with connective
tissue disease showed reactivity with the cell-binding domain of fibronectin
(Girard et al., 1995).
Anti-fibronectin antibodies have been experimentally induced in rabbits
by immunization with human fibronectin in complete Freund's adjuvant
(Murphy-Ullrich et al., 1984). The antibodies were reactive with both
human and rabbit fibronectin. The rabbits subsequently developed a
glomernlopathy with granular deposits suggestive of immune complexes
in the glomcrular basement membrane. Anti-fibronectin antibodies have
been induced in mice by multiple injections of homologous fibronectin
without adjuvant (Murphy-Ulrich et al., 1986). The titer of anti-fibronectin
antibodies was much lower in mice immunized with native fibronectin
than in mice immunized with de-natured fibronectin. However, in both
groups, immune complexes were present in the serum and in the glomerali
(Murphy-Ullrich et al., 1986). Light microscopic examination of the
glomerali of the kidneys of vaccinated dogs did not reveal evidence
of glomerunfpathy, but we cannot exclude the possibility of sub-light
Anti-laminin antibodies were prevalent in the serum of three of the
five vaccinated dogs. Anti-laminin antibodies are increased in human
patients with SLE, rheumatoid arthritis, and vasculitis. Injection
of polyclonal anti-laminin antibodies into rats resulted in glocnerulopathay
and proteinuria (Abrahamson and Caulfield, 1982)> Anti-laminin
antibodies have also been implicated in glomaerular disease in rats
induced by mercuric chloride (Aten et al., 1995).
The mechanisms that may underlie the production of autoantibodies
following vaccination are unknown, but at least four mechanisms can
be proposed: cross-reactivity with vaccine-components, somatic mutation
of immunoglobulin variable genes, "bystander activation"
of self-reactive lymphocytes, and polyclonal activation of lymphocytes.
Perhaps the simplest and most likely mechanism is that of cross-reactivity
of vaccine and self-antigens. (Schattner and Rager-Ziaman, 1990),
the most likely sources of cross-reactive epitopes are bovine serum
and cell culture components. These are present in almost all vaccines
as residual components of the cell culture necessary to generate vaccine
viruses and may purposely be added to the vaccine as a stabilizer.
In the presence of an adjuvant, these bovine products stimulate a
strong immune response and induce antibodies that cross-react with
conserved canine antigens. Thus, the strong response to fibronectin
in the vaccinated dogs is most likely the result of the injection
of bovine fibronectin contaminants in the vaccine. Indeed, this is
essentially identical to the protocol used to produce anti-fibronectin
antibodies in rabbits with human fibronectin in complete Freund's
adjuvant (Murphy-Ullrich et al., 1984), as mentioned above. The lower
response to other antigens (e.g., cardiolipin and laminin) may be
due to a lower concentration of these antigens in the vaccine or lower
During every immune response, self-reactive B and T lymphocytes are
generated and activated. This is the result of somatic mutation and
bystander activation. Under normal conditions, this will not lead
to significant production of autoantibodies, because of the selection
process in the germinal centers of lymph nodes. In the germinal centers
only B cells that successfully compete for interaction with antigen
presented on the surface of follicular dendritic celia will be allowed
to survive (MacLennan, 1994). These B cells generally have high-affinity
receptors for the antigen to which the immune response was induced.
B cells with low affinity for the antigen or affinity for other antigens,
including self-antigens, will undergo programmed cell death. The B
cells with high-affinity receptors express bc1-2, which may rescue
them from programmed cell death (MacLennan, 1994). This mechanism
was elegantly demonstrated in mice immunized with a nominal antigen
phosphorylcholine (Ray et al., 1996). A single point mutation in the
hypervariable region of the expressed immunoglobulin genes was sufficient
for the phosphorylcholine-specific B cells to acquire specificity
for DNA. However, it was only possible to demonstrate DNA-specific
B cells by fusing germinal center B cells with celia that expressed
high levels of bc1-2, thereby rescuing them from programmed cell death
(Ray et al., 1996). An increased expression of bc1-2 was observed
in thymic lymphoid follicles of patients with myasthenia gravis, suggesting
that failure to delete self-reactive B cells in these patients may
lead to autoimmune disease (Shiono et al., 1997). While this may seem
an attractive hypothesis to explain autoimmune phenomena in human
beings and dogs, there is currently no evidence that this is a common
Finally, polyclonal activation of lymphocytes, including activation
of self-reactive lymphocytes, is a possible mechanism of vaccine-induced
autoimmunity. Certain viruses and bacteria have superantigen or mitegen
activity (Schwarts, 1993). This could also be the case for the microbial
products included in the vaccines. The present study does not support
this mechanism. Firstly, antibodies were observed against 10 of 17
antigens tested. Secondly, the anti-fibronectin antibodies did not
react with any portion of the fibronectin molecule, but instead, reacted
most strongly with the heparin binding domain. These observations
indicate that the appearance of autoantibodies in the serum of vaccinated
dogs is an antigen-driven process and not caused by polyclonal activation.
As argued earlier, the main antigens implicated are cell culture contaminants
and bovine serum components.
In the dog, certain autoimmune diseases occur more frequently in particular
breeds of dogs, indicating genetically determined susceptibility (Dodds,
1983; Happ, 1995). There is abundant evidence from studies in rodents
and human beings that the magnitude of the antibody response and the
susceptibility to autoimmune disease are in part genetically determined
(Schwartz, 1993). It is likely that genetic factors also determine
the susceptibility to vaccine-induced autoimmunity. That this is indeed
the case is suggested by the finding that only three of the five vaccinated
dogs developed a strong anti-laminin antibody response and that the
kinetics of the anti-fibronectin response differed between individual
animals. Identification of susceptibility genes will be important,
because it may shed light on the pathogenesis of the autoimmunity.
In addition, it will provide genetic tests that will enable dog breeders
to monitor the susceptibility of their breeding stock to vaccine-induced
Although the pathogenic significance of the vaccine-induced autoantibodies
is still unclear, there are a number of ways to prevent their induction.
Not vaccinating dogs is not a viable option, because the benefits
of vaccination clearly outweigh the still uncertain risks of immune-mediated
disease. However, since bovine serum components in the vaccine may
be responsible for the majority of autoantibodies, elimination of
these bovine components may avoid this problem. This could be accomplished
by substituting homologous serum for bovine serum. However, as mentioned
earlier, anti-fibronectin antibodies may still be induced by immunization
with homologous fibronectin. New generations of vaccines, especially
naked DNA vaccines, are free of serum components, and these should
not induce autoantibodies. A recent study in mice indicates that DNA
vaccination does not induce or accelerate autoimmune disease (Mer
et al., 1997). Finally, mucosal vaccines are less likely to induce
autoantibodies than parenterally administered vaccines. Depending
on the formulation of the vaccine, soluble serum components are less
likely to be absorbed via the mocosal surface, and, in fact, may induce
tolerance instead of autoantibodies (Weiner et al., 1994).
In conclusion, we have demonstrated that vaccination of dogs using
a routine protocol and commonly used vaccines, induces autoantibodies.
The autoantibody response appears to be antigen driven, probably directed
against bovine antigens that contaminate vaccines as a result of the
cell culture process and/or as stabilizers. The pathogenic significance
of these autoantibodies has not yet been determined.
The authors thank Cheryl Anderson and Julie Tobelski-Crippen for animal
care and technical support, and Nita Glickman for data management.
This work is supported by the John and Winifred Hayward Foundation.