Department of Human Genetics and the Meakins-Christie Laboratories, McGill University, Montreal, Quebec, Canada
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
genetic predisposition to disease; quantitative trait loci; microarray analysis; recombinant congenic strains
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Inbred strains of mice differ in their tendency to develop pulmonary fibrosis after bleomycin treatment, and they have been used as the base of genetic investigations to define susceptibility genes. We have previously mapped two quantitative trait loci (QTL), named bleomycin-induced pulmonary fibrosis 1 and 2 (Blmpf1 and Blmpf2), of the propensity to develop fibrosis after bleomycin exposure in F2 mice derived from progenitor strains C57BL/6J (B6) and C3Hf/KAM. Barth et al. (1) used the progenitor strains DBA/2 and BALB/c to map two loci of susceptibility to bleomycin-induced pulmonary fibrosis, which, as they differ from those we have mapped (15), indicates the utility of studies in distinct inbred strains of mice for uncovering susceptibility loci of complex traits.
In this study, we made use of the strain difference in the bleomycin response between B6 (susceptible) and A/J (resistant) mice (31) and of available genomic resources to both map susceptibility to pulmonary fibrosis and to identify a set of potential candidate genes for the trait. Investigations of the A/J strain were undertaken as our previously identified locus Blmpf1, which maps to the major histocompatibility complex (MHC), was defined in B6 (MHC haplotype H2b) and C3Hf/KAM (H2k) mice, and, as the haplotype of the A/J strain (H2a) is different, studies of this strain may reduce the number of MHC-derived fibrosis candidate genes.
To map the susceptibility to pulmonary fibrosis, we used a series of 36 recombinant congenic mouse strains (RCS) derived from B6 and A/J progenitor strains (7, 8). Fourteen of the strains (named AcB) contain a random 13.25% of B6 genes in the A/J strain background, and 22 strains are 13.25% A/J genes in the B6 background (BcA strains). Thus, with this resource, B6 alleles involved in the susceptibility to pulmonary fibrosis are potentially divided among 14 strains of mice, enabling the assessment of the effect of discrete B6 genomic regions on the phenotype. Such RCSs have been used by others to map complex traits of malaria susceptibility and endotoxin-induced lung response among others (5, 6, 8 ).
In addition, among the mapped positional candidates, a set of potential fibrosis susceptibility genes was isolated by identifying the subset of these genes that were differentially expressed between B6 and A/J mice in bleomycin-treated mouse lungs and for which there is a sequence variation between the two strains.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Bleomycin treatment.
Lung damage was elicited by administering bleomycin through osmotic minipumps implanted subcutaneously, as described previously (15). A/J mice were typed for their fibrosis response at 3 wk (5 males and 6 females) or 6 wk (6 males and 6 females) after treatment, and 10 untreated control mice (5 males and 5 females) were killed at the 6-wk time point. Male mice received 100 U bleomycin/kg body wt (2.5 U/mouse), and female mice received 125 U/kg. Male and female mice were treated in separate studies due to the higher drug dose required to produce fibrosis in female mice. An additional 11 A/J and 7 B6 mice were treated with a lower dose of bleomycin (80 U/kg for males and 100 U/kg for females) and killed after 3 wk; 208 RCS mice (a minimum of 3 male and 3 female mice of each of 19 RCS, and a total of 35 male and female mice for 14 additional strains) received 80 U/kg for males and 100 U/kg for females to assay the fibrosis response. A further 815 mice of each of the AcB65 and the BcA 70, 72, 78, 81, 84, and 85 strains were treated with the lower-dose bleomycin protocol to substantiate their phenotypes. AcB/BcA mice were killed at 3 wk after treatment. AcB52, AcB56, and BcA76 strains were not studied due to low availability of these strains.
Histology and fibrosis scoring.
At autopsy, the lungs were removed, and the single left lobe of each mouse was perfused with 10% neutral buffered formalin and submitted for histological processing. Lung sections were stained with Massons trichrome to identify the sites of collagen deposition in the lung. The area of the fibrosing phenotype for each mouse was quantified with image analysis of histological sections as in our previous study (15). Specifically, the area of fibrosis in the left lung lobe was determined from a user-drawn region surrounding the fibrosis (Spot Software) and compared with the area of the entire lobe to yield the percent pulmonary fibrosis for individual mice. Two different users evaluated the percent fibrosis of the mice, and the interuser agreement was r2 = 0.87.
QTL analysis.
Genome scan analyses were performed by using MapManager QTX (version b20) (27). With this software, the set of fibrotic phenotypes (defined as the percentage of the lung with fibrosis by histology) of the 33 RCS mice was compared with their known genotypes to identify the genetic loci influencing this trait in B6 versus A/J mice. Only the RCSs with three or more phenotyped mice were included in the QTL analysis (this yielded 33 strains when both sexes were combined, 27 strains for males only, and 23 strains for females only). In this analysis, marker regression function was used to determine the likelihood ratio statistic for each of 616 markers on 20 chromosomes. For each marker, the resultant likelihood ratio statistic was divided by 4.61 (2 x ln 10) to yield the logarithm of the odds (LOD) score. The thresholds for determining the significance of loci were based on Lander and Kruglyak (22)-proposed linkage standards and on empirically derived limits. From Lander and Kruglyak, we used the mouse backcross value (deemed closest to recombinant congenic mice), which is a suggestive linkage LOD score of 1.9 and a significant LOD score of 3.3. To empirically determine suggestive and significant threshold LOD scores, 10,000 permutations of the phenotype on the genotype were carried out in our data set. With the use of the data of 33 RCSs, the LOD score suggestive of linkage was 1.8, for significant linkage the LOD score was 3.8, and the LOD score indicative of highly significant linkage was 6.9.
Gene expression.
After death, the right lung of each mouse was immediately homogenized in 2 ml TRIzol reagent and placed in dry ice. The homogenates were stored at 85°C until RNA isolation. Total RNA was extracted from A/J lung homogenates according to the manufacturers (Sigma) instructions. The RNA from the right lungs of four or five mice from each group, defined by sex and treatment, was pooled as in Ref. 30 to minimize biological variation in gene expression within a group. One sample of pooled RNA for each group was processed through a RNAEasy column (Qiagen) and submitted for hybridization. The quality of the isolated RNA was assessed and confirmed both before and after pooling by using the Agilent Bioanalyzer (Agilent Technologies; Palo Alto, CA). The experiment was performed with 1 chip/mouse group, represented by its pooled RNA. The gene expression profile of the following groups of A/J mice was measured at the 3-wk time point: males, 100 U bleomycin/kg; females, 125 U/kg; males, 80 U/kg; females, 100 U/kg; and male untreated control mice and female untreated control mice. The gene expression profile of the following groups of mice was measured at the 6-wk time point: males, 100 U/kg; and females, 125 U/kg.
Microarray hybridization was performed by the Affymetrix Gene Chip Core facility at the McGill University and Genome Quebec Innovation Centre. Probe synthesis, hybridization, and washing protocols followed the standardized Affymetrix protocol as reported by Novak et al. (29).
The chips were scanned with a GeneArray Scanner (Agilent Technologies). The resultant gene expression profile was then viewed using Microarray Suite 5.0 (Affymetrix). MOE430A GeneChip arrays containing 22,690 probe sets derived from sequence clusters contained in Build 107, June 2002, of UniGene, which represent 12,422 functionally annotated genes and a set of expressed sequence tags, were used.
Microarray data analysis.
Routines from Bioconductor version 1.4 (http://www.bioconductor.org/) within the R version 1.90 statistical language (17) were used for quality control, normalization, and differential expression. In particular, the quality of the raw microarray data was assessed by inspecting similarities between the intensity distribution and RNA digestion plot for each array. Normalization was performed using the robust probe level model (18). Using mean log intensity versus average log intensity plots, we compared arrays to determine whether different times of postbleomycin exposure, bleomycin dose, and/or gender of the animal influenced gene expression in A/J mice, and we found no significant differences in expression levels, with the exception of the genes on the X and Y chromosomes. This lack of difference justified the pooling of data from these arrays to form two distinct groups: A/J control and A/J bleomycin. A list of significantly differentially expressed genes with P < 0.01 was then generated intrastrain (control vs. bleomycin exposure) with the detection of differential expression performed using the LIMMA package (25, 33). The gene expression data of A/J mice were then compared, using LIMMA analysis, with those reported for B6 mice in response to the same bleomycin treatment (16).
To further assess the lung gene expression profile of A/J mice in response to bleomycin, we used the LIMMA package to analyze the data from NCBI GEO entry GDS350. These data were generated from four Gladstone v2 mouse lung oligo arrays (n = 16,463 probes) hybridized with cDNA from each of four bleomycin-treated A/J mice compared with the pooled RNA of control untreated A/J mice.
The detection of significantly overrepresented Gene Ontology categories was performed using the GOStats package in bioconductor (10). This test of statistical significance considers the number of differentially expressed genes found in each category compared with the total number of genes in the category represented on the chip.
Quantitative real-time PCR.
Four to five micrograms of total RNA from each of four mice of each treatment group were used in a RT reaction to synthesize first-strand cDNA using oligo(dT)1218 Primer and Superscript II RNase H- Reverse Transcriptase (Invitrogen; Carlsbad, CA) in a 20-µl total volume. The lung expression level of each of the genes in Table 1 was determined for bleomycin-treated mice (100 U/kg for female and 80 U/kg for males at the 3-wk time point) and control B6 and A/J mice. These six genes were selected to represent genes of increased and decreased expression in response to bleomycin, as indicated by the arrays. For this analysis, sequence-specific primer sets were designed using Primer 3 (32) or taken from Primerbank (35). Primers were selected to span large introns to amplify only cDNA (see Table 1).
|
Applied Biosystems Real-Time PCR system 7500 was used with the Taqman gene expression assay to test fibroblast growth factor receptor 2 expression. In this assay, each 25-µl reaction contained 1 µl of 1:10 diluted cDNA template, 12.5 µl of TaqMan Universal PCR Master Mix, and 1.25 µl of Assays-on-Demand Gene Expression Assay Mix, which contained forward and reverse primers and labeled probe. The default thermal cycling conditions for PCR were used as instructed by the manufacturer. Relative quantification values were obtained by using Applied Biosystems software and Sca10 reference gene expression.
Sequence comparison.
The markers flanking each of the identified putative loci were located in the Celera (Rockville, MD) mouse genome database (http://www.celera.com/, CDS 13h release), and the number of genes (excluding pseudogenes) mapping to each region was determined. With the use of DNA positions of the flanking markers, the Celera Mouse single nucleotide polymorphism (SNP) reference database (version 3.6) was queried for SNPs within each linkage region. These SNP data were then filtered to uncover the set of SNPs for which B6 and A/J mice have a different allele. These data were further filtered to exclude SNPs appearing in the intronic region or identified as synonymous, as has been used by others (24).
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Because of the risk of lethality from acute toxicity in response to bleomycin, which is not related to the development of fibrosis, we assayed the lung phenotype of B6 and A/J mice 3 wk after delivery of a lower dose of bleomycin (80 U/kg for males and 100 U/kg for females). As the strain difference in fibrosis phenotype of B6 and A/J mice was evident at this dose in both male (B6 = 5.7 ± 1.1% vs. A/J = 0.12 ± 0.18%, P = 1 x 107) and female mice (B6 = 4.7 ± 1.9% vs. A/J = 0.48 ± 0.56%, P = 4.7 x 104), this lower dose was used in the mapping study.
AcB/BcA fibrosis phenotype.
The first step of our strategy for identifying the genes involved in the fibrosis susceptibility of B6 mice relative to the A/J strain was to determine their map position. To accomplish this, we treated a minimum of 3 male and 3 female mice of 19 AcB/BcA strains and 35 mice of additional 14 RCS with bleomycin and killed the mice 3 wk later. The percent fibrosis of the lung, by histology, was used to phenotype for susceptibility/resistance, and the resultant strain distribution pattern of the RCS mice is shown in Fig. 1. As shown in Fig. 1, mice of the BcA strains were generally more sensitive to the development of fibrosis than those of the AcB strains. Two BcA strains (BcA 78 and 81) were highly susceptible to the development of pulmonary fibrosis. These strains had an average of 10% and 15% fibrotic lung tissue, which is 1.8 and 2.7 times the B6-susceptible phenotype at this dose. In addition, BcA strains 68, 69, 73, 74, 79 (females only), 84, and 85 had significantly lower levels of fibrosis compared with B6 mice (<1% fibrosis, all P < 8.6 x 103).
|
Mapping of pulmonary fibrosis susceptibility.
With the use of percent fibrosis as a quantitative phenotypic trait and the genotypic data of all 33 phenotyped AcB and BcA strains combined, 2 linkage regions were detected through linear regression analysis with MapManager QTX software (27) (see Table 2). The identified linkages (on chromosomes 3 and 6) have suggestive LOD scores (22), and they are also suggestively linked to the phenotype according to the permutation test in MapManager QTX. When regression was performed using the dataset from male mice only (n = 27 strains with 3 or more phenotyped male mice), suggestive regions on chromosomes 1, 5, 9, and 12 were detected in addition to the regions on chromosomes 3 and 6. No regions suggested to be linked to the fibrosis phenotype were evident with the data from the female mice (23 strains with 3 or more phenotyped male mice) alone. The difference in linkage results between the sexes is likely attributable to the lower number of phenotyped recombinant congenic female mice compared with males and to the greater range of phenotype in male mice. At each of these putative loci, the presence of B6 alleles increased the fibrotic phenotype of RCS mice, as shown in Table 2.
|
Candidate gene identification.
To propose candidate fibrosis susceptibility genes, we isolated, from among the mapped positional candidates (identified by a composite of Ensembl and Celera), the subset of genes that were differentially expressed between B6 and A/J mice in bleomycin-treated lungs and for which there was a sequence variation between the strains.
Gene expression studies.
It was hypothesized that the genes producing the difference in response to bleomycin between B6 and A/J mice would be differentially expressed in the lungs of these mice after the drug treatment. To identify such genes, expression studies were performed using Affymetrix GeneChip microarrays. The lung response of A/J mice to the drug was ascertained by comparing the gene expression profile of bleomycin-treated mice with that of untreated controls, and, subsequently, this dataset was compared with that of B6 mice (16) to identify strain differences in the response.
The A/J response to bleomycin was measured with six arrays, each of which represents the response of a group of treated mice, as described in MATERIALS AND METHODS, compared with two arrays of gene expression in lung tissue from untreated mice. Nine genes were identified to be differentially expressed (fold 2, P < 0.01, with a maximum fold change = 5.8, P = 0.0003) between control and bleomycin-treated A/J mice (see Table 3). This limited change in gene expression reflects the minimal histological response of A/J mice to bleomycin, which was consistent across the groups of A/J mice evaluated. In support of this finding, we analyzed the data taken from NCBI entry GEO GDS350 of the pulmonary response to bleomycin of A/J mice, and we detected no differentially expressed genes for bleomycin to control comparison (all genes P
0.14). In contrast to the minimal bleomycin response of A/J mice, the B6 gene expression profile, as reported in Haston et al. (16), has 1,768 genes or expressed sequence tags measured to be differentially expressed between controls and bleomycin-treated mice.
|
|
The map positions of the genes measured to be differentially expressed in the lungs of bleomycin-treated B6 mice compared with A/J mice were reviewed to isolate genes located in the putative linkage regions. Two hundred forty-six linkage interval genes were identified to be differentially expressed (P 0.01), and the subset of 18 of these genes, with fold changes in expression
2, is shown in Table 4. These genes are considered to be expression and positional candidates for the genetic basis of bleomycin-induced pulmonary fibrosis in this model.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Using the osmotic minipump delivery method, we showed the mice of the A/J strain to have a minimal fibrotic response to bleomycin, in contrast to the inflammation and fibrosis that develops in B6 mice (13, 31). This mode of bleomycin delivery, developed by Harrison and Lazo (12) and used by us (15), was selected as it has been found to produce a fibrotic phenotype that more closely resembles idiopathic pulmonary fibrosis than the more commonly used experimental method of intratracheal drug delivery (9). The response of the A/J strain to the drug agrees with the findings of Rossi et al. (31), in which bleomycin was delivered intraperitoneally to mice over 4 wk and fibrosis did not result, but differs from the report of Chen et al. (3), in which an intratracheal drug delivery system was used.
With the confirmed strain difference in bleomycin-induced fibrosis susceptibility, recombinant congenic mice were used to map nine loci of the phenotype. Two of the putative linkage regions, on chromosomes 6 and 18, overlap with previously defined QTL of susceptibility to radiation-induced pulmonary fibrosis (14), and the locus on chromosome 6 may coincide with a QTL of bleomycin-induced pulmonary fibrosis that has been reported to be on this chromosome (1). The commonality of the loci supports their existence, but as the present linkage regions were mapped in a limited number of RCSs, confirmatory studies are required. If confirmed, the implication of the same linkage regions in susceptibility to both radiation-induced and bleomycin-induced pulmonary fibrosis may indicate that the phenotype causative genes underlying these loci are not specific to the damaging agent used but are related to the development of fibrosis. In addition, the putative fibrosis linkage region indicated by the marker D3Mit335 overlaps a butylated hydroxytoluene-induced inflammation (lymphocytes) QTL (26), which may indicate this to be a common lung response locus. The linkage regions of bleomycin-induced pulmonary fibrosis susceptibility detected in a B6 x C3Hf/KAM cross did not meet the criterion for suggestive linkage with the present data set; the LOD score of Blmpf1 was 1.7, and the LOD score of Blmpf2 was 1.5.
Our second genomic approach for identifying fibrosis susceptibility genes was to measure the gene expression profile of A/J mouse lungs after drug treatment and compare it with that documented for B6 mice. The B6 gene expression data, reported in Haston et al. (16) and used for comparison in the present investigation, agree with those reported in previous studies of this strain (19, 20) and include the representation of 41 bleomycin-induced differentially expressed B6 genes in a gene cluster (n = 66 genes) defined by Kaminski et al. (19) for fibrosis development. The A/J response to bleomycin, measured by pulmonary gene expression of phenotypically similar groups, also agreed with a separate report for this strain (NCBI GEO). The comparison of the gene expression profile of A/J and B6 mice revealed thousands of genes to be differentially expressed by strain, further indicating the complexity of the fibrosis phenotype.
By combining the genomic approaches of linkage and gene expression (as in Ref. 23), fibrosis-causative candidate genes for the B6:A/J model were proposed. The identified genes are considered candidates as the causal variation leading to the development of bleomycin-induced pulmonary fibrosis with the assumption that the fibrosis-causative gene is differentially expressed in the bleomycin-treated lung. Differential expression is not a necessary condition for implication as causal variation but was used to rank the set of positional candidate genes to facilitate further investigation. From this analysis, the possible pathways to fibrosis in the A/J:B6 model include differences in immune system mediators and in extracellular matrix homeostasis. Specifically, gene candidates from the linkage regions that showed an increase in expression in the lungs of A/J mice compared with B6 mice are involved in immune defense, such as the immunoglobulin heavy chains 1 and 4 and the J558 family, whereas genes such as procollagen 31 and 5
2 and elastin, which are linked to the collagen deposition and turnover, were of relatively increased expression in the lungs of B6 mice. Second, from the analysis of the 21 BcA strains, we were able to detect loci where alleles from the resistant A/J strain increased the fibrotic phenotype of BcA mice, which likely indicates an interaction among loci influences the development of fibrosis. As an example of such an interaction, Fgf1, which maps to a locus where A/J alleles increase the phenotype, showed an increase in expression in the lungs of A/J mice relative to B6, and it could interact with fibroblast growth factor receptor 1 (Fgfr1), which was shown to be increased in B6 mice, to produce increased levels of pulmonary fibrosis in certain BcA strains.
We also assessed the positional candidate genes for DNA sequence variation and, as with gene expression, the functional affect of any sequence variation would have to be confirmed but the existence of a coding or regulatory SNP is potential supporting evidence for causal variation. The Celera database was used as the source of SNPs as it is the most complete documentation of the A/J strain at present, although it may not be a comprehensive review of all B6:A/J sequence variation. With this analysis, a finite set of sequence variation in the candidate genes (shown in Table 4) was uncovered for further testing. Included in this list are the physiological candidate genes lysyl oxidase and interleukin-1 receptor-associated kinase 4. Lysyl oxidase is involved in the cross-linking of collagen, and persistent expression of the gene has been implicated in irreversible fibrosis in bronchiolitis obliterans (34), whereas interleukin-1 receptor expression has been shown to increase with the development of fibrosis in B6 mice (3).
In summary, a combination of genomic approaches was used to identify candidate genes for susceptibility to bleomycin-induced pulmonary fibrosis in a B6:A/J mouse cross. Thirty-three recombinant congenic mouse strains were used to define six intervals, which ranged in size between 9.4 and 29.8 Mb, linked to the trait. Three more putative loci were defined using 21 BcA strains only, and these contain alleles where the A/J genotype increases the fibrosis phenotype. In addition, gene expression studies identified a set of differentially expressed genes mapping to these nine intervals, and a review of SNP data permitted the identification of parental strain gene sequence variations.
The phenotypic trait mapped, susceptibility to bleomycin-induced pulmonary fibrosis, is clinically significant as this lung response limits the dose of bleomycin that can be safely administered and as the induced injury may be a model for the more prevalent condition of idiopathic pulmonary fibrosis. The specific genetic variants reported, if confirmed to influence drug-induced pulmonary fibrosis, could provide insight on the development of this pathology.
![]() |
GRANTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: C. K. Haston, Dept. of Human Genetics and the Meakins-Christie Laboratories, McGill Univ., 3626 St. Urbain, Montreal, Quebec, Canada H2X 2P2 (e-mail: christina.haston{at}mcgill.ca).
1 The Supplemental Material for this article (Supplemental Table S1) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00095.2005/DC1.
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|