| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Centenary Institute of Cancer Medicine and Cell Biology, Newtown, Australia
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The cause(s) of SLE are unknown, but environmental and genetic factors
are involved. The environmental factors that may trigger the disease
include infections, antibiotics (especially those in the sulfonamide
and penicillin groups), and UV light. SLE is known to cluster within
families, and the risk of a first-degree relative of a patient
developing disease is about 4% (i.e., the 
s is about
20) (4). As the disease exhibits complex genetics and
significant heterogeneity in terms of the clinical features at
presentation, the introduction of animal models, principally mouse
models, has generated considerable interest.
We have previously reported a novel model of SLE in which a single dose
of pasteurized Mycobacterium bovis administered i.v. to
prediabetic nonobese diabetic (NOD) mice prevented the spontaneous
onset of type 1 (autoimmune) diabetes but precipitated a systemic
"autoimmune rheumatic disease" (ARD) similar to SLE (5, 6). This syndrome was characterized by hemolytic anemia (HA),
antinuclear autoantibodies (ANA), increased severity of sialadenitis,
and glomerular immune complex deposition. The specificity of the ANA
responses in these mice was directed against dsDNA and the Smith
(Sm)/ribonucleoprotein complex, of which the 28-kDa polypeptide
appeared to be immunodominant (7). The IgG subclass
involved in the anti-Sm response was primarily IgG2a, while the
subclass of the response against dsDNA was mixed, with IgG2a and IgG2b
being present in equal amounts. The anti-dsDNA and anti-Sm
reactivities were not mediated by polyreactive Abs because neither Ag
could cross-compete plasma Ab binding to the other in competitive
ELISA. The role of polyclonal B cell activation was examined by
measuring total 
-globulin as well as IgG reactive with other nuclear
Ags including Ro60, Ro52, and La, which although not a major component
of the autoantibody responses in these mice did show small but
significant increases following immunization with M. bovis.
Thus, polyclonal stimulation, while likely to be occurring, was not
directly responsible for production of anti-Sm Abs
(7). The pattern of renal disease seen in NOD mice treated
with bacillus Calmette-Guérin (BCG) is similar to mild focal
lupus nephritis and is characterized by segmental proliferation of a
minority of glomeruli and widespread C3c deposition.
This model of SLE has two major advantages. First, it is the only mouse model for which the environmental trigger has been identified. Therefore, it introduces the opportunity for a detailed analysis of environment/genome interactions in determining disease liability. The second advantage is that the mouse strain involved, the NOD mouse strain, has been extensively genetically characterized, which permits the direct comparison of genetic loci identified in linkage studies of SLE in this model with loci previously identified in linkage analysis of type 1 diabetes.
The coexistence of these two distinct autoimmune diseases within the same mouse strain, and the reciprocal switching between these two phenotypes by a single environmental trigger (mycobacterial exposure), raises the possibility that genetic susceptibility for diabetes and SLE may be conferred by a single collection of genes. That is, a single inherited disease (autoimmunity) can be expressed as either phenotype, depending on, for example, environmental modifiers. Recent genomic studies of autoimmune diseases, both in humans and in mice, have provided significant support for this "common gene" hypothesis (8, 9, 10, 11). The linkage analysis reported here allows comparison with the data obtained from mapping diabetes in the same strain and SLE in other mouse models.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
NOD/Lt, BALB/c, and C57BL/6J mice were obtained from the Animal Resource Center (Canning Vale, Australia). Breeding of specific crosses was performed within the animal facility at the Centenary Institute (Sydney, Australia). Mice were housed in clean conditions, and sentinel mice were tested by serology at 4-mo intervals for the following pathogens: mouse hepatitis virus, rotavirus, ectomelia, mouse CMV, polyoma virus, murine adenovirus, lymphocytic choriomeningitis virus, mouse pneumonia virus, reovirus, Sendai virus, Theiler’s murine encephalitis virus, Bacillus piliformis, Mycoplasma pulmonis, Bordetella bronchiseptica, Corynebacterium kutscheri, Klebsiella species, Pasteurella multocide, Pasteurella pneumotropica, Pseudomonas aeruginosa, Staphylococcus aureus, Streptococcus pneumoniae, Citrobacter freundii, and Salmonella species. No sentinel mice tested positive for any of these pathogens.
M. bovis injection
Mice were injected at 8 wk of age with 0.4 mg of CSL (Melbourne,
Australia) dried BCG (M. bovis) vaccine (living) for
percutaneous use after pasteurization at 65°C for 45 min. In vitro
culture of this quantity of BCG indicated that it was equivalent to
4 x 106 CFU before inactivation.
Phenotyping
Assessment of anemia. Mice were bled 6 mo after BCG injection by retroorbital venepuncture. Hematocrits were determined by centrifuging the blood in heparinized capillary tubes (Becton Dickinson, Parsippany, NJ) at 1000 x g for 30 min. Direct Coombs’ tests were performed by washing the packed red cells three times in PBS plus 0.3% BSA and incubating them in round-bottom microtiter plates with anti-IgG serum (Southern Biotechnology Associates, Birmingham, AL) for 2 h at 37°C. Plates were then assessed visually for evidence of agglutination.
Assessment of antinuclear autoantibodies. Sera were diluted to 1:10 in PBS and incubated on HEp-2 slide monolayers (Quantafluor, Chaska, MN). Slides were washed in PBS and incubated with 20 µg/ml human-adsorbed anti-IgG FITC-conjugated antiserum (Southern Biotechnology Associates). After washing, the presence of ANA was detected by fluorescence microscopy. Samples that contained ANA were then retested at 1:100. Sera from affected NOD mice and unaffected BALB/c mice were used as positive and negative controls, respectively. Samples were declared positive for ANA if they showed moderate to strong immunofluorescent staining of HEp-2 cell nuclei at a dilution of 1:100. Samples were declared negative for ANA if they showed no evidence of immunofluorescent staining of HEp-2 cell nuclei at a dilution of 1:10.
Detection of glomerular immune complex deposition. Unlike that in NZB/W mice, the pattern of renal disease in NOD mice treated with BCG is similar to mild focal lupus nephritis and is characterized by widespread C3c deposition and segmental proliferation of a minority of glomeruli. As this form of nephritis is associated with only mild proteinuria, measurement of urinary protein with clinical dip-sticks was found not to be a robust diagnostic test. Although the diagnosis can be readily made by electron microscopy, the very large numbers of samples in this study precluded this option. In view of the relative subtlety of the histological changes as well as the expected increase in variation in the backcross compared with examining genetically identical parental or F1 mice, glomerular C3c deposition was used to detect the presence of renal disease because it was the most robust assay available.
Mice were killed at 32 wk, and their tissues were harvested for histology and DNA extraction. Kidneys were snap frozen, and C3c glomerular deposition was detected on cryostat sections following incubation with 1:100 dilution of FITC-conjugated goat anti-mouse C3c serum (Nordic Immunology, Tilburg, The Netherlands) in PBS for 30 min at room temperature. Slides were then washed three times for 5 min with PBS, mounted, and examined by fluorescence microscopy.
Genotyping
DNA of BC1 progeny was extracted from livers and subjected to an autosomal genome-wide scan at a 15-centiMorgan (cM) resolution using 150 microsatellite markers polymorphic between BALB/c and NOD/Lt strains chosen from the Whitehead Institute simple sequence length polymorphism library (12). A list of all microsatellite markers typed is available on request. Analysis of simple sequence length polymorphism was performed as previously described (13).
Linkage analysis
Genotyping errors were identified manually as double
recombinants or by the error checking function of Mapmaker/EXP
(14) and were reamplified. Recombination distances between
markers were calculated from recombination frequencies using the
Mapmaker/EXP program (14). Lengths of chromosomes and
order of markers were checked against published maps (15)
(http://www-genome.wi.mit.edu/; http://www.informatics.jax.org/). Two
approaches to linkage analysis were used. To calculate the linkage of
individual markers to the autoimmune phenotypes, a 2 x 2
contingency table (
2 test of independence) was
applied. The significance of linkage achieved by this method was tested
by creating experimentwise thresholds using permutation analysis
(16). The thresholds were simultaneously valid for all
markers of the genome and were calculated by permuting the phenotype
data 10,000 times. The highest 2 value
throughout the whole genome for each permutation was collected and
subsequently sorted to produce a null distribution of the test
statistic. The distribution allowed the calculation of experimentwise
95% (p < 0.05 of type 1 error in the entire
genome) and 37% (p < 0.63 of type 1 error
throughout the genome) critical thresholds. Markers with experimental
2 surpassing the experimentwise 95% critical
threshold were considered significantly linked to the trait, while
those surpassing the 37% threshold were considered to show a trend
toward linkage (i.e., suggestive linkage) to the trait.
In an alternative approach, interval analysis of linkage to the
autoimmune phenotypes was conducted on two linkage analysis programs.
The MapManager/QT program (17) was initially used to
perform interval linkage analysis on the autoimmune phenotypes by
entering the trait values as 0 for unaffected and 1 for affected mice
(Ref 18 ; Kenneth Manly, unpublished observations). The
degree of linkage to the autoimmune phenotypes was reported using a
likelihood ratio statistic (LRS) (19). The program was
used to calculate permutation-derived threshold values for significant
and suggestive linkage (according to Lander and Kruglyak’s
specifications; Ref. 20) for the interval analysis,
utilizing the methods established by Churchill and Doerge (16, 17, 20). One thousand permutations of phenotypes in the dataset
were performed. Qualitative trait analysis was also performed on
Mapmaker/QTL 2.0b using the "penetrance scan" function (Ref.
21 ; new version supplied by Mark Daly of the Whitehead
Institute of Biomedical Research), which optimizes a set of penetrances
for each genotypic class. Trait values were entered as 0 or 1 for
unaffected and affected mice, respectively. Results of the penetrance
scan are given as a logarithm of odds (LOD) score, representing the
likelihood that data have arisen due to the effect of a qualitative
trait locus with optimized sets of penetrances rather than the effect
of chance (under the null hypothesis that the penetrances for all
genotypic classes are equal). Significance thresholds used were those
suggested by Lander and Kruglyak (20) for analysis of a
mouse backcross; viz LOD 
3.3 for the threshold for significant
linkage and LOD 1.9 for the threshold suggestive of linkage.
These values were validated by substitution of the following specific
values from our genetic study into equation 1 (µ(T) =
[C +
2
GT2)
(T)) in Lander
and Kruglyak’s paper (20): C (number of
chromosomes) = 19, (measure of S(x)
fluctuation, which represents total crossover rate for a backcross with
1 df) = 1, G (genome size in Morgans) = 15.3 (see
Results), T (pointwise significance level given
in 2 (1 df)) = 3.841, and
(T) = 0.05. From this equation, we expected the
pointwise significance level of p = 0.05 to be
surpassed approximately seven times by chance (µ(T)) in
each of the linkage analyses. Thus, it was determined that a pointwise
significance level of p = 0.0001 and p
= 0.0036 would only be surpassed 0.05 and 1 time throughout the genome
by chance, respectively. These figures correspond to LOD values of 3.3
(p = 0.0001) and 1.8 (p
= 0.0036), respectively, which are similar to those suggested by Lander
and Kruglyak (20). These values did not change when
µ(T) was calculated separately for each trait.
Mapmaker/QTL and MapManager/QT have individual advantages, and the two programs apply slightly different methods to calculate linkage to binary traits. Mapmaker/QTL is usually regarded as the tool of choice for the analysis of murine data sets, and its "penetrance scan" function is ideally suited for performing interval linkage analysis of binary traits. One disadvantage of the program is its tendency to produce artificially elevated LOD scores in large intervals (>15 cM) (21), which greatly limits its utility for calculating significance thresholds by permutation analysis. MapManager/QT has the advantage of being able to accurately calculate permutation-derived thresholds specific for individual datasets using methods derived from Churchill and Doerge (16).
Disease genes were proposed within a region when the significance thresholds were surpassed in at least two of the three analyses performed.
Other statistical analyses
Qualitative differences between samples were examined using the
4-fold table (2) test unless the expected
value in any cell was 5 or less, in which case the Fisher’s exact test
was used. Quantitative differences between samples were compared using
the Mann-Whitney U (rank sum) test. A goodness of fit
(2) test was used to compare proportions of
affected animals with those predicted by modeling.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
NOD/Lt, C57BL/6, BALB/c, and hybrid mice (produced by mating NOD/Lt mice with either C57BL/6 or BALB/c mice) were injected i.v. at 8 wk with 0.4 mg heat-killed M. bovis. After 6 mo, the mice were bled to detect HA (by measurement of hematocrit) and ANA (by HEp-2 immunofluorescence) and subsequently killed for removal of kidneys for immunofluorescent staining of glomerular C3c deposits.
The hematocrits of all animals tested were normally distributed about a
mean of 53% with a SD of 3.8%. Therefore, anemia was defined as a
drop in the hematocrit to 1.645 SD below the mean (i.e., hematocrit
<47%; p < 0.05). While the great majority of female
NOD/Lt mice became anemic (9 of 11; 82%), only one of 14 (7%) female
C57BL/6 mice and none of 16 female BALB/c mice did so (Table I
; p < 0.0001; 4-fold
table 2 test). Similarly, although the sera of
10 of 11 (91%) female NOD/Lt mice had moderate to strong
immunofluorescent staining of HEp-2 cell nuclei at a dilution of 1:100,
none of the sera from 14 C57BL/6 and 16 BALB/c female mice did so
(Table I; p < 0.0001; 4-fold table
2 test). The differences in C3c staining
deposition were less clear cut, as nine of 10 (90%) female NOD/Lt mice
showed strong glomerular staining compared with four of nine (44%)
C57BL/6 and three of 10 (30%) BALB/c female mice (Table I;
p < 0.05; 4-fold table 2
test).
|
;
p < 0.0001; 4-fold table 2
test). While NOD/BALB hybrid mice were free of glomerular C3c
deposition, it was present in about half of NOD/B6 hybrid mice. The
presence of glomerular immune complex deposition in female NOD/B6
F1 hybrid mice showed a significant deviation
from expected values when the two directions of mating were compared
(Table I; 11 of 14 vs 3 of 11; 79% vs 27%; p < 0.05;
4-fold table 2 test). These two crosses were
repeated, and this difference was not confirmed (3 of 10 vs 7 of 14;
30% vs 50%; NS; 4-fold table 2 test) and is
therefore unlikely to be of importance. No evidence was found for any
other sex-linked affects in these crosses.
The penetrances of HA, ANA, and glomerular immune complex deposition in
NOD mice were higher in female mice than male mice, although this trend
only reached significance for glomerular immune complex deposition
(p < 0.0005, 4-fold table
2 test) and was not apparent in B6 or BALB/c
mice or the hybrid progeny. As the phenotypic differences were more
strongly marked in the BALB/c crosses than those using C57BL/6 mice,
the NOD/BALB strain combination was selected for the linkage study. As
considerable difficulties were experienced with conflicts arising
between male hybrid mice, only female backcross mice were selected for
analysis.
Production and phenotypic analysis of backcross mice
NOD/Lt mice were crossed with NOD/BALB hybrid mice in all four possible directions. At 8 wk of age, 960 segregating BC1 female mice were injected i.v. with 0.4 mg heat-killed M. bovis and 6 mo later were bled to detect HA (by measurement of hematocrit and direct Coombs’ test) and ANA (by HEp-2 immunofluorescence) and then killed for removal of kidneys and immunofluorescent staining of glomerular C3c deposits.
The hematocrits of the BC1 progeny were not normally distributed, but
formed a bimodal distribution with left-skewing of the main population
(Fig. 1A). The threshold
between the two populations occurred at 44%, which correlated well
with the presence or absence of Coombs’ Abs. While 36 of 45 (80%)
mice with hematocrits below 44% tested positive in the direct Coombs’
test, only 47 of 820 (6%) mice with hematocrits of 44% or above did
so. For the purposes of linkage analysis, HA was declared to be present
in the mice with Coombs’ Abs in their sera and hematocrits below 44%.
The control sample was selected from the mice without Coombs’ Abs in
their sera and hematocrits of 44% or above, with a strong bias for
hematocrits around the mode of the main peak in the population
distribution (Fig. 1A). Mice with very high hematocrits were
avoided because diabetics were over-represented in this sample due to
dehydration and hemoconcentration secondary to hyperglycemic diuresis.
As the hematocrits of BC1 progeny were not normally distributed, they
did not represent a quantitative trait suitable for quantitative trait
locus (QTL) mapping. Furthermore, as the threshold between normal
values and anemic values correlated well with the presence of Coombs’
Abs, a binomial linkage analysis of hematocrits as a separate trait
would be unlikely to contribute information independent of that
obtained from mapping HA.
|
ANA were detected in the sera of BC1 progeny by immunofluorescence
detection of IgG bound to HEp-2 nuclei at a dilution of 1:10, and
positive samples were retested at a dilution of 1:100. Mice were
declared positive if their sera were positive or strongly positive at
1:100 (134 of 903; 15%) and negative if their sera tested negative at
1:10 (154 of 903; 17%; Fig. 1B).
Glomerular C3c deposition was detected by immunofluorescence microscopy of renal cryostat sections in 68 of 818 (8.3%) BC1 progeny.
Complete phenotypic data for HA, ANA, and glomerular immune complex
deposition were available for 804 BC1 female progeny. The goodness of
fit (2) test was used to compare the observed
frequencies of coinheritance of more than one phenotype with those
expected on the basis of the observed frequencies of each individual
phenotype, assuming the null hypothesis that the traits segregated
independently. The null hypothesis was rejected
(p < 0.0001; 2 =
19.2, 2 df), almost entirely due to greater than expected coinheritance
of HA and glomerular immune complex deposition. There was no evidence
of coinheritance of ANA and glomerular immune complex deposition (Fig. 2), consistent with other reports
(22) but not with a proposed critical role of ANA in the
pathogenesis of glomerular immune complex deposition (reviewed in Ref.
23 ; paradox discussed in Ref. 24).
|
Construction of linkage map
A genome map was created by typing 221 female BC1 progeny at each of 150 polymorphic microsatellite markers distributed throughout the autosomal genome. Recombination distances between each marker were determined using the Mapmaker/EXP program (14), and the lengths of chromosomes and the order of markers were checked against published maps (15). The total autosomal genome length (excluding centromeric and telomeric portions) obtained here was 1529 cM, compared with 1187 cM reported for the MIT map, consistent with suppression of recombination in the intraspecific cross used by Dietrich et al. (15). The gene orders obtained from this data set conformed well to those previously published, with the following exceptions: D1Mit438 mapped 0.9 cM distal to D1Mit9 (order reversed); D3Mit10 mapped 2.4 cM distal to D3Mit11 (colocalized on the MIT map); D9Mit206 colocalized with D9Mit67 (2.2 cM more distal on MIT map); D14Mit225 mapped 2.6 cM distal to D14Mit160 (order reversed); D14Mit77 mapped 4.4 cM distal to D14Mit136 (colocalized on MIT map); and D17Mit83 colocalized with D17Mit16 (1 cM more distal on MIT map).
Genotypic and linkage analysis of HA
An autosomal genome scan of 34 anemic (Coombs’ positive,
hematocrit <44%) and 53 unaffected female BC1 progeny was conducted
using 133 polymorphic microsatellite markers with an average marker
separation of 15 cM. Linkage analysis was initially performed at each
marker using a 2 x 2 contingency table, and the results were
compared with the 37% (suggestive) and 95% (significant)
experimentwise statistical thresholds derived from a permutation
analysis of 10,000 iterations of the data as specified by Churchill and
Doerge (16). Genotype frequencies differed significantly
(2 > 12.3) between affected and unaffected
mice at loci on chromosomes 16 and 17 and exhibited results suggestive
of linkage (2 > 6.5) on distal chromosome 1
and at the most proximal marker tested on chromosome 14 (Table II). While NOD alleles contributed to
susceptibility in a recessive fashion on chromosomes 1 and 17, BALB
alleles contributed to susceptibility on chromosomes 14 and
16.
|
3.3)
and suggestive linkage (LOD 1.9) were applied. The Mapmaker/QTL
analysis confirmed the results obtained using contingency tables, in
that it identified regions of significant linkage to HA on chromosomes
16 and 17 (Fig. 3). The peak linkage on
chromosome 16 occurred between D16Mit58 and
D16Mit5 (LOD = 3.9). The region of significant linkage
on chromosome 17 extended over about 20 cM, with a peak occurring at
D17Mit83 (Hsp70–1) and D17Mit16
(LOD = 4.8), which lie within the H2. A second peak was
identified between D17Mit20 and D17Mit153
(LOD = 3.8), with a minimum LOD of 3.0 between the two peaks. The
linkage region on distal chromosome 1 was also identified as suggestive
of linkage with Mapmaker/QTL, appearing as two small peaks, one between
D1Mit108 and D1Mit36, and the other between
D1Mit36 and D1Mit406. Analysis of the data set
using MapManager/QTL was applied using permutation derived statistical
thresholds based on 1000 iterations of the data. This analysis again
confirmed linkage on distal chromosome 16 and proximal chromosome 17
(data not shown)
|
Genotypic and linkage analysis of ANA
An autosomal genome scan of 52 female BC1 progeny with no evidence
of plasma ANA (by immunofluorescence at a dilution of 1:10) and 52
female BC1 progeny with strong plasma ANA (positive at 1:100) was
conducted using 127 polymorphic microsatellite markers with an average
marker separation of 15 cM. Linkage analysis was initially performed at
each marker using a 2 x 2 contingency table and results compared
with the 37% (suggestive) and 95% (significant) experimentwise
statistical thresholds derived from a permutation analysis of 10,000
iterations of the data. Genotype frequencies differed significantly
(2 12.0) between affected and unaffected
mice at loci on chromosomes 1, 10, and 17 (Table III). Suggestive linkage
(2 6.5) was also identified on chromosome
5. The NOD alleles contributed susceptibility in a recessive fashion on
chromosomes 1, 5, and 17, while the BALB/c allele contributed
susceptibility in a dominant fashion on chromosome 10 (Table III).
|
3.3; Fig. 3). Regions with suggestive linkage were also identified on chromosomes
1 and 5. Linkage peaked on distal chromosome 1 at D1Mit399
(LOD = 2.8), on chromosome 5 between D5Mit239 and
D5Mit65 (LOD = 3.0), and on proximal chromosome 10
between D10Mit213 and D10Mit257 (LOD = 4.9).
LOD scores exceeded the Lander and Kruglyak threshold for significant
linkage over most of chromosome 17, with the proximal half of the
chromosome surpassing a LOD of 4.0. The maximum LOD score on this
chromosome (7.0) was obtained at D17Mit83, D17Mit16, and
D17Mit24, which are associated with the H2.
Analysis of the data set with MapManager/QT confirmed the regions of
significant (LRS 12.8) linkage on chromosomes 10 and 17. This
analysis also confirmed the chromosome 1 linkage identified by the
2 analysis. Linkage reached significance with
a sharp peak at D1Mit399 (LRS = 13.5), although two
other peaks exhibiting a trend toward linkage (LRS 6.7) were
identified, one centered on D1Mit445 (LRS = 11.8) and
the other on D1Mit135 (LRS = 7.9), consistent with
similar results obtained with Mapmaker/QTL.
In summary, all three analyses identified genomic regions controlling susceptibility to ANA occurring in NOD mice treated with M. bovis. The first, here named Bana1 (BCG-induced ANA, locus 1), mapped to proximal chromosome 17 and may be the H2, although the breadth of the peak is suggestive of the existence of multiple genes in this region, which can influence the development of ANA in this model. The second locus, here named Bana2 (BCG-induced ANA, locus 2), mapped to proximal chromosome 10, between D10Mit213 and D10Mit257. A third locus, here named Bana3 (BCG-induced ANA, locus 3) and mapped to D1Mit399 on distal chromosome 1, achieved significance when analyzed with the contingency table and MapManager/QT, but not when analyzed with Mapmaker/QTL.
Genotypic and linkage analysis of autoantibodies
Because of the likelihood factors shared in the etiologies of
Coombs’ Abs and ANA, and the apparent colocalization of genes for both
traits on chromosomes 1 and 17, an autosomal genome scan was performed
comparing 94 female BC1 progeny with autoantibodies (either positive
for ANA at a dilution of 1:100 or direct Coombs’ test positive) and 54
unaffected female BC1 progeny (negative for ANA at 1:10 and Coombs’
negative). Linkage analysis was initially performed at each marker
using a 2 x 2 contingency table and results compared with the
37% (suggestive) and 95% (significant) experimentwise thresholds
derived from a permutation analysis of 10,000 iterations of the data.
Genotype frequencies differed significantly (2
12.2) at loci on chromosomes 1 and 17. Loci on chromosomes 5, 10,
and 16 demonstrated a trend toward linkage (2
6.5). Linkage peaked on chromosome 1 at D1Mit396 with a
2 of 17.4, while the H2-associated
markers on chromosome 17 (D17Mit83, D17Mit16, and
D17Mit24) reached a 2 of 42.2. The
NOD alleles contributed susceptibility in a recessive fashion on
chromosomes 1, 5, and 17, while the BALB/c alleles contributed to
susceptibility in a dominant fashion on chromosomes 10 and 16 (Table IV).
|
3.3; Fig. 3) and also
identified regions suggestive of linkage on chromosomes 10 and 16. The
pattern of LOD scores produced by the genotype frequencies of loci on
chromosome 1 was very complex, with at least three peaks; one between
D1Mit12 and D1Mit445 (LOD = 4.3), one
proximal to D1Mit396 (LOD = 4.5), and the third between
D1Mit396 and D1Mit398 (LOD = 4.5). Almost
the whole of chromosome 17 exceeded the threshold for significance,
with a maximum LOD occurring at D17Mit83, D17Mit16, and
D17Mit24 (LOD = 9.6). Additional regions surpassing the
threshold for suggestive linkage (LOD 1.9) were identified on
chromosomes 10 and 16. Maximum linkage on chromosome 10 was found
between the two most proximal markers (D10Mit213 and
D10Mit257; LOD = 2.3) and on chromosome 16 between the
two most distal markers (D16Mit5 and D16Mit94
LOD = 2.7). The MapManager/QT analysis of the data set confirmed
significant linkage (LRS 13.2) on chromosomes 1 and 17 and
indicated a trend toward linkage (LRS 6.8) on chromosome 16
(data not shown).
In summary, all three analyses identified linkage regions controlling
susceptibility to autoantibodies in M. bovis-treated NOD
mice on chromosomes 1 and 17. The first, here named Babs1
(BCG-induced autoantibodies, locus 1) mapped to proximal chromosome 17
and may be the H2. Linkage on chromosome 1 was complicated
as MapManager/QT placed the maximum peak between D1Mit445
and D1Mit396, while the Mapmaker/QTL analysis flanked this
region with two other peaks, although the trough between them was not
sufficient (i.e., 1 LOD) to define them as separate loci. Thus, only
a single locus could be attributed to chromosome 1, here named
Babs2 (BCG-induced autoantibodies, locus 2).
Genotypic and linkage analysis of glomerular immune complex deposition
An autosomal genome scan of 44 female BC1 progeny with glomerular
immune complex deposits (by immunofluorescence of C3c) and 52
unaffected female BC1 progeny was conducted using 127 polymorphic
microsatellite markers with an average marker separation of 15 cM.
Linkage analysis was initially performed at each marker using a 2
x 2 contingency table. No locus exceeded the 95% (significant)
experimentwise threshold derived from a permutation analysis of 10,000
iterations of the data (2 12.0). Genotype
frequencies showed a trend suggestive of linkage (37% threshold;
2 6.6) at loci on chromosomes 1, 4, 12, 16,
and 17. The NOD alleles contributed susceptibility in a recessive
fashion on chromosomes 1, 4, 12, and 17, while the BALB/c allele
contributed susceptibility in a dominant fashion on chromosome 16
(Table V).
|
In summary, no linkage region controlling susceptibility to glomerular C3c deposition in NOD mice treated with M. bovis surpassed a significant level of linkage with any of the three analyses applied. It is speculated that the existence of background glomerular immune complex deposition in the BALB/c outcross partner reduced the power of this section of the study.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The class III genes have been implicated on a number of grounds including: 1) the role of the early complement components in the removal of immune complexes from the circulation, 2) the presence of deletions or silent alleles of C4A and C4B in the susceptible haplotypes such as B8-SC01-DR3-DQ2.1, and 3) the strong association of the rare cases of absolute component deficiencies of C1q, C1r-C1s, or C4 with susceptibility to SLE even though the risks associated with partial C4 deficiency or C2 deficiency are much lower (28). However, linkage disequilibrium with class III null alleles does not fully explain the association of SLE with class II genes within the HLA gene complex (reviewed in Ref. 25), as exemplified by the very strong association between DR4 and hydralazine-induced SLE in women.
In addition to the HLA-linked SLE susceptibility genes, other genetic
factors are involved. With the exception of the C1 complement component
genes (on chromosome 12p13), which are involved in immune complex
handling and are associated with anti-dsDNA autoantibody production
and SLE nephritis (29), little is known about them except
that they demonstrate a dominant pattern with incomplete penetrance in
families (reviewed in Ref. 28) and may include the genes
encoding the Ag receptors on T (- and 
-chains, 14q11.2;
ß-chain, 9p22-9cen; -chain 7p15-7p14; Ref. 30) and B
cells (heavy chain, 14q32; 
light chain, 2p12; light chain,
22q11.2; Ref. 31). Three recent genome-wide screens of SLE
susceptibility genes in affected pedigrees and affected sib pair
families (32, 33, 34) identified two regions significantly
(LOD 3.6) (20) linked to SLE: 6p11-p21, mapping
adjacent to the MHC; and 16q13. These studies also identified six
additional regions with LOD scores suggestive of linkage (LOD
2.2) (20): 1q41, 1q23, 13q32, 14q21-q23, 20q13, and 20p12;
and nine other regions with LOD scores exceeding 1.0.
As with SLE, susceptibility to type 1 diabetes in humans appears to depend on multiple genes encoded within the MHC and elsewhere in the genome (reviewed in Ref. 35). IDDM1 maps to the MHC on chromosome 6p21 and susceptibility is contributed by DRB1, DPB1, DQB1, and DQA1 as well as several other minor genes within the MHC. The contributions of these genes is best seen in the light of ancestral haplotypes; while the incidence of type 1 diabetes in Caucasian communities is around 0.03%, about 6% of individuals with the high-risk MHC haplotype (DR3/DR4) develop diabetes. Linkage to the insulin gene (IDDM2) on 11p15 was also originally identified on the basis of candidature and has been associated with (36), and attributed to (37, 38), allelism in a VNTR locus 5' to the insulin gene, which affects the level of insulin gene expression in the thymus. In addition to IDDM1 and IDDM2, a score of other linkage regions have been identified in genome-wide screens of affected sib pair families (39, 40), although only IDDM4 (11q13), IDDM5 (6q25), and IDDM8 (6q26) have been confirmed in other studies (reviewed in Ref. 35 ; reasons for failure to replicate linkages discussed in Refs. 35 and 41).
The coexistence of SLE and type 1 (autoimmune) diabetes in patients is
very rare, and its frequency has never been demonstrated to deviate
significantly from that expected to occur by chance alone if the two
diseases were completely independent. Similarly, the combination of HA
with type 1 diabetes appears to be exceptional. ANA, in contrast, are
found in about 15% of children at onset of type 1 diabetes,
while are present in only 1% of normal controls (42, 43). Despite the rarity of the coexistence of SLE and type 1
(autoimmune) diabetes, there is good evidence that, at least for loci
within the MHC, some alleles confer susceptibility to both diseases.
For example, two DR3-associated ancestral haplotypes that confer
susceptibility to SLE (B8,SC01,DR3,DQ2,Dw24 and B18,F1C30,DR3,DQ2,Dw25)
also confer susceptibility to type 1 diabetes (44). In a
recently published controlled correlation of 23 published genome-wide
scans of autoimmune or immune-mediated diseases, Becker et al.
(8, 9) found that 65% of positive human linkages
mapped nonrandomly into 18 distinct clusters. In addition to the MHC,
type 1 diabetes and SLE susceptibility genes colocalized at 7 other
IDDM loci, including IDDM2 and
IDDM8.
At least three major possibilities could account for these results.
First, the 18 clusters identified by Becker could represent areas of
the genome in which genes that affect immune function have aggregated
through gene duplication and/or selective advantage in a manner similar
to the evolution of the MHC. Second, they could represent genes
critical to immune function at which different alleles contribute to
susceptibility to different immune-related diseases. The third
possibility is that certain alleles at each of these loci contribute to
susceptibility to autoimmunity per se and that environmental,
stochastic, and other genetic factors affect the final phenotypic
expression of that tendency to autoimmunity. There are at least two
possible ways to directly test the last possibility (the "common
gene" hypothesis). The first method involves identifying the alleles
responsible for susceptibility to each disease for at least one locus.
To date, this has not been achieved, although the possibility that
IDDM11 on 14q (45) and the SLE linkage in the
same region (33) represents association with the TCR
-chain gene provides a potential opportunity. The second method is
to identify the environmental or genetic modifiers acting on a discrete
locus. The model presented here offers an opportunity to identify in
NOD mice, candidate regions conferring susceptibility to both diabetes
and SLE, depending on exposure to the mycobacterial trigger, with a
view to their isolation by congenesis.
Evidence of colocalization of autoimmune disease susceptibility genes in animal models was first reported by Vyse and Todd (10). In a review, they correlated the linkage data from a score of linkage studies of autoimmune diseases in mice and reported that three of nine murine SLE genes and two of three orchitis genes colocalized with diabetes genes, events unlikely to have occurred by chance. Further correlative support has come from independent linkage analyses published since 1996 that have continued to identify further genomic regions associated with susceptibility to multiple autoimmune diseases. For example, Silveira et al. (46) mapped gastritis in mice following neonatal thymectomy to two loci on distal mouse chromosome 4: Gasa1 and Gasa2. Gasa1 mapped to the same genomic region as the SLE susceptibility gene Nba1 (47) and the diabetes susceptibility gene Idd11 (13), and Gasa2 mapped to the same region as the diabetes susceptibility gene Idd9 (48). Support for the common gene hypothesis was also provided by two hypothetico-inductive studies. In the first, Teuscher et al. (49) sought a relationship between the susceptibility of BALB/cBy to two immunization-induced autoimmune diseases: experimental allergic encephalomyelitis (EAE) and experimental allergic orchitis. They studied cosegregation of susceptibility alleles within a BC1 generation by assaying the male progeny of a subsequent backcross for experimental allergic orchitis and the female progeny for EAE. These results demonstrated that the loci conferring susceptibility were probably linked and were consistent with the same locus conferring susceptibility to both diseases. Perhaps the most compelling study was that by Encinas et al. (50), in which they examined diabetes-resistant congenic lines of NOD for susceptibility to EAE. One NOD mouse line, NOD.B6-Idd3, which carried a 0.15-cM region of B6 genome encoding the Il2 gene, was resistant to EAE induction while wild-type NOD mice were susceptible to disease. This result suggested that NOD sequences in this region, perhaps those of the NOD Il2 allele itself, confer susceptibility to the spontaneous autoimmune disease type 1 diabetes as well as to the experimentally induced autoimmune disease EAE.
Here, we have mapped genetic susceptibility to SLE induced in NOD mice by injection with M. bovis, and it is probably of value to consider these data in the light of the common gene hypothesis. Diabetes in NOD mice and SLE in BXSB, NZB/W, and related strains have been extensively mapped, and the regions of apparent colocalization of susceptibility genes reported (10, 11). Restriction of these candidate regions to those which either surpassed Lander and Kryglyak’s thresholds for significance, had been replicated in other studies, or confirmed by congenesis, identified three major genomic regions contributing susceptibility to both diseases: Idd1/Lbw1/Sle4 on chromosome 17, Idd11/Sle2/Nba1/Lbw2 on chromosome 4, and Idd7/Lbw5/Sle3 on chromosome 7. Both lupus and diabetes susceptibility have also been mapped to mouse chromosome 1, although the 95% confidence interval of the chromosome 1 diabetes susceptibility gene Idd5 (51) mapped at least 20 cM proximal to the SLE susceptibility genes identified with the NZB/NZW strains (Sle1/Nba2/Lbw7) (52). In contrast, of the three SLE susceptibility genes mapped to chromosome 1 with the BXSB model, one of these (Bxs1) maps to a similar region as Idd5 while another (Bxs3) maps close to the Sle1/Nba2/Lbw7 cluster of SLE susceptibility genes (53). Therefore, it is possible that the BXSB strain contains both sets of autoimmunity genes on chromosome 1.
We considered the hypothesis that the NOD alleles of genes in the
implicated regions of chromosomes 17, 4, and 7 conferred susceptibility
to both diabetes and SLE and would therefore be identified in a linkage
study of SLE in NOD mice. The first of these regions, that on
chromosome 17, was found to be strongly linked to HA (Bah1),
ANA (Bana1), and autoantibodies (Babs1) in this
study. The strongest association was with autoantibodies for which a
2 of 42.2 (p <
8.08E-11) was obtained at the MHC. This result is consistent with the
critical role of the MHC in SLE in humans and the NZB/W (22, 54, 55) and BXSB (56) models and in diabetes in humans,
NOD mice (57) and BB rats (58) (Fig. 4). Here, homozygosity for the NOD allele
of the MHC conferred maximum susceptibility to lupus, whereas in the
NZB/W-related strains heterozygosity conferred greatest risk (22, 55, 59, 74, 80) and in the BXSB strain either homozygosity or
heterozygosity conferred equal risk (56). Analogous to the
situation with DR3 in human autoimmunity, the same H2
haplotype that is associated with susceptibility to diabetes conferred
susceptibility to SLE in NOD mice. Therefore, this finding supports the
common gene hypothesis, at least as applied to the MHC. In contrast,
significant linkage to SLE in this model was not found on proximal
chromosome 1, chromosome 4, or chromosome 7 (or indeed any other
non-MHC diabetes susceptibility locus). One possible explanation for
this finding is that BALB/c mice may carry susceptibility alleles at
these loci, and as a result may not be segregating in this cross. There
is some experimental support for the idea that BALB/c mice carry
autoimmunity susceptibility alleles on distal chromosome 4 as we have
mapped to BALB/c gastritis susceptibility alleles to this region
(46). Although this problem could have been avoided by
using C57BL/6 mice as breeding partners in these crosses, this choice
was precluded by the finding reported here of extensive glomerular
immune complex deposition in F1 crosses between
NOD and B6 mice.
|
).
Thus, linkage to ANA on distal chromosome 1 was found in an NZM/Aeg
backcross (Sle1; Ref. 59), a segregating
intercross of NZB F1 mice with NZW mice
(Nba2; Ref. 47), and an
F2 intercross of NZB and NZW mice
(Lbw7; Ref. 54). It is of particular interest
that this region contained the only NZB/W lupus susceptibility locus to
be identified in crosses with both the C57BL/6J and BALBcJ resistant
strains (60). Therefore, the study reported here further
confirms the major importance of this region in conferring
susceptibility to ANA, despite significant differences between the
nature of the models and mouse strains involved. The relevance to human
disease of this robust linkage is strongly reinforced by genome-wide
screens that have identified linkage of lupus to the syntenic region of
human chromosome 1, 1q23, (32, 33). The second region identified in this study was that on chromosome 16 (Bah2) at which BALB/c sequences conferred susceptibility to HA. The genetic marker with strongest linkage to HA on chromosome 16 used in this study (D16Mit58) has previously been identified as showing the strongest linkage to neonatal thymectomy-induced autoimmune ovarian dysgenesis (Aod1; Ref. 61).
The only other genomic region controlling susceptibility to autoimmune disease identified in this study was that on proximal chromosome 10 (Bana2) at which the BALB/c allele also conferred susceptibility to disease. Significant linkage of this locus to any autoimmune disease has not been previously identified, in contrast to the three other genomic regions associated with SLE in this study.
In summary, with the exception of the H2, this study failed to provide direct support for the common gene hypothesis, as the non-MHC-linked loci identified as conferring susceptibility to SLE did not colocalize with those previously implicated in diabetes. The NOD haplotype at the H2 conferred susceptibility to both type 1 (autoimmune) diabetes and SLE, and therefore this region remains a locus for which a congenic approach may allow further study of the interaction between mycobacterial exposure and the expression of autoimmune disease. A striking finding was the nonrandom localization of SLE susceptibility genes, which in three of four cases mapped to regions previously implicated in autoimmunity, thus confirming previous observations of a similar nature. The significance of this clustering remains to be fully explained, but may reflect the genomic structure of loci impacting on immunological function.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 M.A.J. and P.A.S. contributed equally to this study.
3 Address correspondence and reprint requests to Dr. Alan Baxter, Centenary Institute of Cancer Medicine and Cell Biology, Locked bag 6, Newtown NSW 2042, Australia.
4 Abbreviations used in this paper: SLE, systemic lupus erythematosus; NOD, nonobese diabetic; HA, hemolytic anemia; ANA, antinuclear autoantibodies; ARD, autoimmune rheumatic disease; BCG, bacillus Calmette-Guérin; LRS, likelihood ratio statistic; LOD, logarithm of odds; QTL, quantitative trait locus; EAE, experimental allergic encephalomyelitis; Sm, Smith; cM, centiMorgan.
Received for publication December 27, 1999. Accepted for publication May 19, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
) and the anemic (
) and control
(
) samples selected for genomic analysis of HA. B,
Histogram of ANA as detected by immunofluorescent staining of HEp-2
cell nuclei at a dilution of 1:10 and 1:100 in sera of BC1 population
(
