Previous Article | Next Article ![]()
Infection and Immunity, January 2003, p. 126-131, Vol. 71, No. 1
0019-9567/03/$08.00+0 DOI: 10.1128/IAI.71.1.126-131.2003
Copyright © 2003, American Society for Microbiology. All Rights Reserved.
Center for Molecular Medicine, Karolinska Institutet, 171 76 Stockholm, Sweden,1 Laboratory for Immunogenetics, Central Institute for Tuberculosis, Moscow 107564, Russia,2 McGill Center for the Study of Host Resistance, Montreal General Hospital, Montreal, H3G 1A4 Quebec, Canada3
Received 2 January 2002/ Returned for modification 12 February 2002/ Accepted 24 September 2002
|
|
|---|
|
|
|---|
Mouse models of TB have proved to be valuable for studies of antimycobacterial immunity and of the genetic control of susceptibility and resistance (16). For example, the Nramp1 gene, which has provided great insight in our understanding of macrophage-mycobacterium relationships, was first discovered in a mouse model of susceptibility to Mycobacterium bovis BCG (9, 28). Numerous inbred mouse strains have been tested to determine their survival times after challenge with virulent Mycobacterium tuberculosis (1, 17, 20). Among these strains, I/StSnEgYCit (I/St) mice display the shortest survival time, while A/SnYCit (A/Sn) mice survive significantly longer. In addition, I/St mice display more severe and rapid disease progression than A/Sn mice in terms of body weight loss, mycobacterial loads in lungs and spleens, and lung histopathology (21). We previously reported the results of backcross-1 (BC1) ([A/Sn x I/St] F1 x I/St) analysis of the body weight loss in this strain combination after intravenous challenge with virulent M. tuberculosis H37Rv (15). In BC1 female mice, variations in this clinically relevant and easily measurable phenotype were linked to distal chromosome 3 (D3Mit215 logarithm of the likelihood ratio [LOD] = 3.9) and proximal chromosome 9 (D9Mit89 LOD = 6.8) and were suggestively linked to chromosome 8 (D8Mit289 LOD = 3.0) and chromosome 17 (D17Mit175 LOD = 3.0). In males, there was suggestive linkage to chromosome 5 (D5Mit233 LOD = 3.0) and chromosome 10 (D10Mit133 LOD = 2.3) (15). In the present study we more precisely defined the position of the quantitative trait locus (QTL) on chromosome 9 by using BC1 and performed an analysis of the linkage of the chromosomal regions listed above to body weight loss and to survival time postchallenge by using F2 segregation.
|
|
|---|
Mice and infection. Inbred mice belonging to strains I/St and A/Sn, as well as (A/Sn x I/St) F1 and (A/Sn x I/St) and (I/St x A/Sn) F2 hybrid mice and BC1 mice ([A/Sn x I/St] F1 x I/St), were kept under conventional conditions in the animal facilities of the Central Institute for Tuberculosis, Moscow, Russia, in accordance with guidelines of the Russian Ministry of Health no. 755. F2 and BC1 mice that were 2 to 3 months old were infected intravenously via the lateral tail vein with 5 x 105 CFU of M. tuberculosis H37Rv. Survival time was monitored daily, and body weight was monitored weekly. The BC1 mice were the same animals used for a previous study in which the general location of several QTLs was detected (15).
Genotyping. Genotypes of the simple sequence length polymorphisms D3Mit29, D3Mit215, D3Mit199, D5Mit182, D5Mit232, D5Mit233, D5Mit312, D5Mit366, D6Mit287, D6Mit129, D8Mit289, D8Mit291, D10Mit130, D10Mit133, D9Mit204, D9Mit89, D9Mit23, D9Mit94, D9Mit142, D9Mit207, D9Mit47, D9Mit269, D9Mit113, D17Mit101, D17Mit175, D17Mit28, D17Mit93, and DXMit170 (MapPairs; Research Genetics, Huntsville, Ala.) were determined by using PCR performed with isolated spleen DNA, followed by product separation on 6% polyacrylamide gels and detection by autoradiography. Female and male F2 mice were genotyped for 19 and 17 of the markers, respectively.
Statistical analysis.
The markers were assigned to and mapped within the chromosomes by multipoint linkage analysis by using Mapmaker/Exp, version 3.0 (14). Analysis of multipoint linkage between genotype markers and the quantitative phenotypes square-root-transformed survival time and relative body weight loss ([body weight at zero time - body weight on day 20 postinfection]/body weight at zero time) for identification of QTLs was performed by using interval mapping and subsequent MQM mapping within the MapQTL software package (26). To compensate for gender effects when the sample containing both genders was analyzed, the male-female difference in the phenotype mean was subtracted from the male phenotype data. Interval mapping based on single QTL models was used for initial detection of putative QTLs. MQM mapping analyzes approximate multiple QTL models by assigning a marker near the QTL to a cofactor. This cofactor takes over the effect of the QTL on other loci and enhances the sensitivity for detection and exclusion of additional QTLs and artifactual QTLs, respectively. Cofactors were assigned to putative QTLs with LOD scores of >1.0 as detected by interval mapping. MQM mapping with forward selection of cofactors was performed; cofactors were added or dropped sequentially by using the threshold LOD score of 1.0 in each test. A free model of inheritance of the I/St allele was studied. The final cofactors were tested for representing real QTLs (rather than artifactual QTLs) by moving each cofactor along its linkage group to see whether the QTL remained at the original locus. Genome-wide significance for LOD scores was determined by permutation tests (10,000 runs of simulated data). The critical LOD values suggested by the tests were for suggestive linkage 1.9 (one QTL expected at random in a genome scan) and for significant linkage 2.8 (one QTL expected at random once every 20th genome scan; i.e., genome-wide P value of
0.05) (13). Threshold correction for testing the hypothesis of linkage (i) to two modestly correlated phenotypes and (ii) within marker genotype subgroups gave LOD values of
3.0 (genome-wide P value of
0.02) that were regarded as significant.
Distribution normality and variance equality were tested by using the normality test (test of skewness and kurtosis) and Bartlett's test, respectively. Relative body weight loss was normally distributed. However, survival time was not normally distributed, and therefore it was square root transformed in order to normalize its distribution. Statistical significance for inbred strains was determined by using the two-tailed t test. Correlations were calculated by using Pearson's correlation coefficient. The statistical significance of a difference in phenotype effect of loci genotypes was determined by using multivariate three-way analysis of variance (MANOVA) including the correlated outcome variables relative body weight loss and survival time, the predictor variable gender, and two of the factors genotype at D9Mit89, D3Mit215, D17Mit175, and DXMit170 and followed by a two-way analysis of variance. For MANOVA, a P value of
0.01 was considered significant. These statistical tests were performed with STATA, version 6 (Stata Corporation, College Station, Tex.), and StatView, version 5 (SAS, Cary, N.C.).
|
|
|---|
![]() View larger version (26K): [in a new window] |
FIG. 1. QTL linkage analysis of TB severity following infection of mice with M. tuberculosis H37Rv: LOD score plots for gender-adjusted postinfection relative body weight loss and survival time from analyses of BC1 (A and D) and F2 (B, C, and E) animals. The LOD score peaks between D9Mit23 and D9M142 in panel A were not designated since they were not supported by a marker.
|
The linkage analysis was expanded by genotyping 406 (A/Sn x I/St) F2 infected mice (176 females and 230 males) for microsatellites that were previously significantly or suggestively linked to body weight loss in the BC1 mice (15). The survival times of the F2 mice ranged from 20 to 85 days for the males and from 18 to 71 days for the females (Fig. 2), whereas the relative body weight losses at day 20 postinfection ranged from -22.1 to 32.0% for the males and from -13.9 to 30.6% for the females. The MANOVA supported gender differences for both traits (P = 0.013). The values for both traits in individual F2 mice were distributed between the values for I/St and F1 hybrid mice. For relative body weight loss a few F2 mice had values slightly greater than those of the F1 mice. This supports the hypothesis that there is multigenic control of these traits, as suggested previously in the analysis of BC1 mice. The survival time correlated well with the relative body weight loss (for males, r = 0.57 [95% confidence interval, 0.48 to 0.65; P < 0.0001]; for females, r = 0.62 [95% confidence interval, 0.51 to 0.70; P < 0.0001]). The results of the linkage analysis for F2 mice are shown in Fig. 1B, C, and E. The intermarker distances observed were similar to those reported for the BC1 mice (15) and in public maps (Release 16 May 1999 of Whitehead Institute/MIT Center for Genome Research [http://www-genome.wi.mit.edu/cgi-bin/mouse/index, 2001 Chromosome Committee Reports, Mouse Genome Database, The Jackson Laboratory [http://www.informatics.jax.org/ccr/]). Because of the gender-specific linkages in the BC1 mice, we analyzed males and females separately.
![]() View larger version (12K): [in a new window] |
FIG. 2. Segregation of mortality in (A/Sn x I/St) F2 mice following M. tuberculosis H37Rv challenge. Animals that died early after challenge formed a rather homogeneous group (indicated by bars). Defining these animals as susceptible and the rest of the animals as resistant gave the following segregation results: 95 susceptible and 81 resistant female mice and 104 susceptible and 126 resistant male mice.
|
Other peak LOD scores related to relative body weight loss and survival time were observed in males and in females at the chromosome 17 markers D17Mit28 and D17Mit175, respectively (Fig. 1C). D17Mit28 and D17Mit175 are neighboring markers in the H-2 complex region (intermarker distance, 6.6 cM), and their corresponding peaks, although shifted, might reflect linkage with the same putative QTL. Numerous recombination hot spots in the vicinity of and within the H-2 complex on chromosome 17 (2, 29) are known to provide different crossover frequencies in males and females, and this might explain the gender-related linkage differences. Importantly however, LOD score peaks for both traits and for both genders are present at this highly polymorphic chromosomal region. Gender-adjusted data for relative body weight loss indicated that there was suggestive linkage to D17Mit175 (LOD = 2.3; genome-wide P = 0.062). This result was supported by the MANOVA (P = 0.00093). Suggestive linkage was also indicated with regard to the tbs1 locus (LOD = 1.8; genome-wide P = 0.15) (Fig. 1E), which was supported by the MANOVA (P = 0.016). No linkage for relative body weight loss or survival time was found to the markers genotyped on chromosomes 5, 6, 8, and 10 (LOD scores, <1.0).
The means ± standard errors for the phenotypes corresponding to each genotype category at the tbs2, chromosome 17, and tbs1 loci for the F2 mice of both genders are presented in Table 1. When the tbs2 genotype was considered, heterozygous female mice (genotype i/a) had the lowest body weight loss and survived longer than females bearing the homozygous allelic combination a/a or i/i, thus following the pattern of F1 hybrids (15). In turn, i/a heterozygous males displayed an intermediate level of TB severity. Differences in control of disease progression between females and males could be due to sex hormones. This hypothesis is supported by reports of adrenal hyperplasia followed by adrenal atrophy and cytokine-mediated changes in the hypothalamic-pituitary-adrenal axis after M. tuberculosis H37Rv challenge (10) and of estrogen-induced differences in the development of pulmonary Mycobacterium avium infections in mice (25). F2 mice having the tbs2a/a homozygous genotype were characterized by the greatest relative body weight loss and the shortest survival time irrespective of gender (Table 1). The course of TB is less severe in A/Sn (tbs2a/a) mice than in I/St mice (tbs2i/i). However, tbs2a/a F2 mice control TB worse than their tbs2i/i counterparts (Table 1), suggesting that the effect of tbs2 in parental strain A/Sn might be masked by epistasis. One likely participant in such nonallelic interactions is the chromosome 17 locus that confers better protection to the carriers of the a/a genotype (Table 1). In females, the tbs1 locus behaves like tbs2, providing a heterozygous advantage. The tbs1 locus also potentially participates in epistatic interactions.
|
View this table: [in a new window] |
TABLE 1. Phenotypes of F2 mice according to genotype
|
![]() View larger version (15K): [in a new window] |
FIG. 3. Effect of the chromosome X locus on the distribution of TB severity according to genotype at the chromosome 17 locus (A) and the tbs2 locus (B). The means and standard errors for gender-adjusted postinfection relative body weight loss in (A/Sn x I/St) F2 mice are shown. The numbers above the symbols are the numbers of mice.
|
Even though modeling of interactions between loci is difficult, it was possible to gauge the relationship between the QTLs reported here by examining the distribution of the phenotypes according to combinations of genotypes at different loci. It is clear from the data (Table 1) that the chromosome 17 locus is genotypically related to resistance and susceptibility in the same direction, as the parental strains I/St and A/Sn, while the tbs1 and tbs2 loci display heterozygous advantage, as the F1 hybrids, with the fairly common situation that when the F2 mice carry the genotype of one parent (here a/a), they behave like the other parent (here i/i) after infection. MANOVA of the (A/Sn x I/St) F2 mice showed support for the gender-tbs2-tbs1 interaction (P = 0.0080) and a trend for the gender-chromosome 17 locus-tbs1 interaction (P = 0.022). Figure 4 shows the distribution of gender-adjusted relative body weight loss according to genotype combinations between the tbs1-chromosome 17 (D17Mit175) and tbs2-tbs1 loci. The most interesting finding is that genotype a/a at tbs1 enhances the effect of the D17Mit175 genotype. Linkage analysis with data only for the mice having the tbs1a/a genotype increased the peak LOD score at D17Mit175 from 2.0 to 4.3 (genome-wide P < 0.00005) for relative body weight loss. Similarly, including only animals heterozygous at tbs2 resulted in an LOD score of 2.2 (genome-wide P = 0.047) at tbs1.
![]() View larger version (25K): [in a new window] |
FIG. 4. Interaction among the tbs1, chromosome 17, and tbs2 loci in control of TB severity in (A/Sn x I/St) F2 mice. The TB severity measurements used to determine the LOD scores and graphs were gender adjusted postinfection relative to body weight loss, and the TB severity measurements used for the MANOVA were adjusted postinfection relative to body weight loss and survival time. The means and standard errors for mice grouped on the basis of their genotypes at the three loci are shown. The numbers above the symbols are the numbers of mice. (A) LOD score for D17Mit175, including mice that were tbs1a/a. MANOVA was performed for the gender-chromosome 17 locus-tbs1 interaction. (B) LOD score for D3Mit215, including mice that were tbs2i/a. MANOVA was performed for the gender-tbs2-tbs1 interaction.
|
The genetic control of TB is generally assumed to be a complex phenomenon (reviewed in references 11 and 19), for which there is strong evidence of involvement of several QTLs. Kramnik et al. (12) mapped the QTL sst1 (susceptibility to tuberculosis 1) gene to a 9-cM interval on mouse chromosome 1 clearly distal to the Nramp1 gene by using the C3H-C57BL/6 strain combination. sst1 controls several parameters of M. tuberculosis Erdman-triggered disease. The sst1 locus can be considered a particularly important TB susceptibility locus, since it simultaneously influences survival time, bacterial loads in lungs and spleens, and the degree of lung pathology following TB challenge. Nevertheless, the authors clearly showed that TB control in their model is also multigenic and depends upon the presence of other, unknown loci outside the sst1 region. Using a similar strain combination, DBA/2 and C57BL/6, Mitsos et al. identified three QTLs that control survival time after challenge with M. tuberculosis H37Rv, Trl-1, Trl-2, and Trl-3 (tuberculosis resistance locus), at distal chromosome 1, proximal chromosome 3, and proximal chromosome 7, respectively (18). Trl-1 and sst1 are close to each other but not close enough to suggest that they reflect the same gene.
So far, different QTL analyses of TB have not revealed overlap in the genomic locations of the QTLs found. However, the tbs2 and chromosome 17 loci overlap the two murine Leishmania major resistance loci, lmr2 and lmr1, respectively, identified in the BALB/c-C57BL/6 strain combination (22). In addition, lmr2 and lmr1 were found to interact in the L. major model, like the tbs2 and chromosome 17 loci in our model (23). This suggests the possibility that these two models might be influenced by similar genes and thus strengthens our suggestion that the tbs2 and chromosome 17 loci reflect genes involved in TB control. Even though suggesting candidate genes for such QTLs might be premature, it is appealing to explore the human syntenic regions for the tbs2, chromosome 17, and tbs1 loci for association with human TB. Additionally, the results support the derivation of congenic lines of mice that carry particular genotypic combinations that might improve the chances of isolating the candidate genes.
We thank J. W. van Ooijen for his assistance during the genetic analysis.
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»