Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Acute lung injury can encompass a continuum of pulmonary severity, often progressing through worsening stages of endothelial and epithelial permeability and arterial hypoxemia. The most severe cases of acute lung injury culminate in respiratory distress, a condition that remains a serious clinical challenge in critical care medicine because of its high incidence and substantial mortality (1, 2).

To initiate studies to identify the major gene(s) influencing differences in acute lung injury survival, we established a mouse model using ozone as the pulmonary irritant. Within hours, ozone can induce alveolar damage that resembles the exudative phase of acute lung injury and that, with continued exposure, is fatal. The more than 3-fold difference in mean survival times (MSTs) between sensitive A/J (A) and resistant C57BL/6J (B6) inbred strains permitted an examination of the genome to identify genetic loci controlling this phenotype. Quantitative trait locus (QTL) analysis studies with backcross (3) and F₂ (4) mice generated from these strains identified three genetic loci (on chromosomes 11, 13, and 17) linked to ozone-induced acute lung injury (currently designated Aliq1, Aliq2, and Aliq3, respectively, for acute lung injury QTL), and several potential modifiers on chromosomes 3, 5, 6, 7, and 12.

To examine whether similar genes might be involved in the response to different irritants, we expanded studies to include exposure of mice to two other aerosolsTeflon fumes and nickel sulfate. Teflon powder, when heated to 420°C, generates an ultrafine particle capable of inducing oxidant lung injury (5, 6) and death (7). High-temperature combustion processes (i.e., welding, smelting, and ore roasting) commonly used in industrial settings can generate substantial nickel levels (10) and, like ultrafine Teflon, have been directly related to serious occupational lung injury and death (13, 14). Interestingly, although ozone, ultrafine Teflon particles, and fine nickel sulfate aerosol have no obvious similarity in physicochemical properties, strains of inbred mice independently exposed to each of these three irritants displayed a similar survival time pattern; the A strain was sensitive and the B6 strain was resistant (15). In addition, exposure of first generation progeny from the A and B6 strains (B6AF₁) demonstrated that the survival time associated with ozone-, Teflon-, or nickel-induced acute lung injury was inherited as a dominant trait (15).

To further examine the genetic components influencing survival time differences to nickel-induced acute lung injury between A and B6 mice, we now report on the QTL analysis of a large backcross population generated from these strains. This analysis identified significant linkage to chromosome 6 and two QTLs on chromosomes 1 and 12 suggestive of linkage. Two additional regions on chromosomes 9 and 16 contributed to the overall response. By comparing the disparity in MSTs of mice with the sensitive or resistant haplotypes for the QTLs on chromosomes 6, 9, 12, and 16, we determined that these four loci could account for the difference in survival between the parental A and B6 strains. Including the QTL on chromosome 1 further increased the MST difference in mice with sensitive and resistant haplotypes. Similar regions on chromosomes 6 and 12 were also identified for ozone-induced acute lung injury survival, suggesting that some genes may play a role in response to both pulmonary irritants.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Experimental Design

To identify chromosomal regions linked to nickel-induced acute lung injury, backcross progeny derived from mating sensitive A and resistant B6 strains of mice were phenotyped, genotyped, and then analyzed for evidence of linkage after continuous exposure to 150 µg/m³ nickel inhalation. Initially, a cohort consisting of the most sensitive (survival time  66 h, n = 55) and most resistant (survival time  112 h, n = 54) backcross mice were examined. This group was genotyped for microsatellite markers polymorphic for the A and B6 strains and distributed throughout the genome. Results of phenotyping and genotyping were analyzed for linkage using MAPMAKER/QTL (16). Those loci reaching a lod score (log of the odds ratio, which equals the log of the "likelihood of linkage divided by the likelihood of no linkage") for suggestive linkage (i.e.,  1.9) were typed for additional markers (20). To increase the likelihood of detecting major and minor QTLs, all 307 backcross mice were included in the genomewide scan. The phenotype and genotype data were analyzed by QTL Cartographer (21, 22) to identify QTLs and to determine the empirical threshold levels for significant and suggestive linkages to the nickel-induced lung injury phenotype. An equation by Wright (23) was used to determine the number of independently segregating loci. Haplotype analysis was performed to quantify the genetic contribution of the QTLs to the overall survival time; mice with the same allelic combinations for the different QTLs were grouped and MSTs calculated and compared. Chromosomal regions of interest were examined for known genes of likely relevance to the phenotype to identify positional candidate genes.

Mice

Male and female A, C57BL/6 (B6), and (B6 × A)F₁ (F₁) control mice, and backcross mice (n = 307), generated from F₁ × A (n = 162) and A × F₁ (n = 145) matings, were obtained from the Jackson Laboratory (Bar Harbor, ME). Mice were maintained in a viral and pathogen-free environment with 12-h alternating cycles of darkness and artificial light.

Nickel Atmosphere Generation, Exposure, and Phenotyping

Mice were placed in a 0.32-m³ stainless-steel inhalation chamber (capable of complete air exchange every 2 min) and exposed continuously to filtered room air containing 152 ± 11 µg/m³ (0.2 µm mass median aerodynamic diameter) nickel. Nickel aerosol was generated from 50 mM nickel sulfate hexahydrate (Ni₂SO₄ · 6H₂O) solution as described previously (15). The nickel concentration in the chamber was determined using the methylglyoxime method (24). Exposures of mice were initiated approximately 1 wk after receipt from the Jackson Laboratory. Backcross mice were tested in a total of six separate exposures performed at various times throughout the year. Group sizes ranged from 30 to 67 mice per exposure. All backcross exposures also contained at least two control A and B6 mice (either sex), approximately equal numbers of each backcross mating scheme (A × F₁ and F₁ × A), and males and females of each group. Survival times were recorded within 5% for each animal.

DNA Preparation/Genotype Analysis

After each animal's death, the liver was removed, immediately frozen in liquid nitrogen, and stored at 40°C for subsequent genotype studies. Genomic DNA was isolated (Wizard DNA; Promega, Madison, WI) and samples were analyzed for purity and DNA concentration using a Beckman DU-64 spectrophotometer. A fraction of each DNA sample was diluted to 10 ng/µL for use in microsatellite analysis. Polymerase chain reactions (PCRs) were performed to genotype backcross progeny for microsatellite markers located throughout the mouse genome. Primer pairs (Research Genetics, Huntsville, AL) were chosen on the basis of known polymorphisms between the A and B6 strains. PCR was performed in 15-µL reactions in 96-well plates (MJ Research, Watertown, MA) using a four-block thermocycler (Model PTC-225; MJ Research). The final concentration for each reaction was: 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, 0.2 mM of each deoxynucleotide triphosphate (Promega), 1 × RediLoad (Research Genetics), and 0.132 µM of each microsatellite primer. This reaction mixture was added to 100 ng of genomic DNA (10 µL) and 0.6125-U of Taq DNA polymerase (GIBCO BRL, Grand Island, NY). Final mixtures were initially denatured at 94°C for 3 min, followed by 35 to 37 cycles of 94°C for 30 s, 55°C for 45 s, and 72°C for 30 s. A final elongation step at 72°C for 7 min was followed by refrigeration (4°C) until genotyped by gel analysis. The PCR products were differentiated on agarose (GIBCO BRL) gels and visualized by ethidium bromide staining. The agarose concentration (2.5 to 6%) depended on the size of the allelic variants, and a difference of 4 to 5% in size between A and B6 alleles could be resolved.

Estimated Number of Genes

To estimate the number of independent genes segregating with the response to nickel-induced acute lung injury, the following formula of Wright was used: n = (P₂  F₁)² /4(|V_N2  V_F1|), where n is the estimated number of segregating loci; P₂ and F₁ are the MSTs of A and B6AF₁ mice, respectively; and V_N2 and V_F1 are computed variances of the (F₁ × A) backcross and F₁ cohorts, respectively (23). This estimate assumes that the genes are unlinked, and that each gene is semidominant and contributes equally to the phenotype.

QTL Mapping, Linkage Analysis, and Gene Interactions

To initially identify possible QTLs that influence survival of nickel-induced lung injury, 77 microsatellite markersdistributed at 20- to 30-cM intervals across the genomewere typed for the 55 most sensitive (survival times  66 h) and 54 most resistant (survival times  112 h) backcross mice (representing the 109 phenotypic extremes). After generation of a linkage map (MAPMAKER/EXP, V3.0b), all phenotype and genotype data were analyzed for linkage using MAPMAKER/QTL, V1.1b (16- 19). For this analysis, the theoretic levels for significant and suggestive linkage proposed by Lander and Kruglyak (20) were used to identify potential QTLs in the polar responders.

To determine the significance of these QTLs in the total backcross population and to better ensure detection of minor contributing QTLs, all 307 mice were then typed for the original 77 microsatellite markers, plus an additional 32 microsatellite markers in the putative QTL intervals. After generation of a linkage map (MAPMAKER/EXP), all phenotype and genotype data were analyzed for QTLs using MAPMAKER/QTL and QTL Cartographer, Model 3 (21, 22), computer programs. Initial analysis of the phenotype distribution indicated skewness so the data set was log-transformed to approximate a normal distribution. QTL Cartographer analysis of the total backcross cohort established significant ( = 0.05) and suggestive ( = 0.1) linkages at empirical levels determined by 10,000 permutations of the original data set (21, 22).

Using the cumulative search function of MAPMAKER/QTL, the major locus was fixed to remove its variance and the genome was rescanned to identify additional loci explaining lesser portions of the genetic variance. After identification of each additional locus, the combined loci were sequentially fixed to further determine potential linkages. Similar methods with cumulative search were also used to investigate evidence of gene interactions between the putative QTLs by assessing differences between additive (sum of individual QTL lod scores) and cumulative (derived from the cumulative search) lod scores. In addition, total backcross data were examined using Epistat (25), a computer program designed to directly detect gene interactions (i.e., epistasis).

Haplotype Analysis

To gain insight into the contribution of each QTL and combination of QTLs to the overall survival phenotype, haplotype analysis was performed. This procedure directly quantifies a difference in survival time (in hours) associated with a particular haplotype. For this analysis, each QTL was assumed to be fully penetrant and located at the microsatellite marker nearest the peak lod score for the identified QTL regions (i.e., D1Mit213, D6Mit183, D8Mit65, D9Mit299, D12Mit112, and D16Mit152). MSTs for groups of mice with the same haplotype (i.e., sensitive alleles) at each QTL or combination of QTLs were calculated and then compared with the MSTs of mice with the opposing haplotype (i.e., resistance alleles) to determine the contributions of these QTLs to the overall phenotype.

Data Analysis

Survival times are presented in hours as the means ± standard error (SE). To evaluate susceptibility to nickel sulfate-induced acute lung injury, mean responses between groups were assessed using analysis of variance followed by Student-Newman-Keuls a posteriori multiple comparison test of significance. Comparisons of MSTs for mice with specific allelic combinations (haplotypes) were performed using the t test. Statistical significance for all comparisons of means was accepted at P < 0.05. Significant and suggestive experiment-wise (empirical) lod scores were determined by 10,000 permutations of the backcross data set using QTL Cartographer, based on the methods of Churchill and Doerge (26, 27).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

QTL Analysis and Gene Interactions

QTL analysis of a data set assumes a normal distribution of the phenotype. Because the phenotype distribution exhibited skewness, survival times were log-transformed to produce a normal distribution before performing QTL analysis (Figure 1).

View larger version (32K): [in this window] [in a new window]   Figure 1.   Survival time distributions of backcross mice before and after log-transformation. The survival times of mice were grouped into 10-h intervals. Log-transformation eliminated the left-skewness of the original data set to meet the assumption of a normal distribution for MAPMAKER/QTL analysis.

Phenotypic extreme responders. To initially identify possible QTLs influencing survival to nickel-induced lung injury, 77 microsatellite markers were typed for the 55 most sensitive (survival times  66 h) and 54 most resistant (survival times  112 h) backcross mice (representing the 109 phenotypic extremes) and results were analyzed by MAPMAKER/QTL. First, the theoretical levels for significant and suggestive linkage proposed by Lander and Kruglyak (20) were used to identify potential QTLs in the polar responders. Regions reaching suggestive linkage (i.e., lod scores  1.9) were identified on chromosomes 1, 6, 8, 9, 12, and 16, with lod scores for these six putative QTLs ranging from 2.1 to 2.8 (Table 1). Assuming loci are acting independently, these QTLs explain 62% of the genetic variance in the phenotypic extreme cohort.

Total backcrosses. To determine the significance of these QTLs in the total backcross population and to better ensure detection of minor contributing QTLs, all 307 mice were typed for the original 77 microsatellite markers plus an additional 32 microsatellite markers in the six putative QTL intervals. Using QTL Cartographer analysis of the total backcross cohort, significant ( = 0.05) and suggestive ( = 0.1) linkages were established at empirical levels determined by 10,000 permutations of the original data set (21, 22). Accordingly, experiment-wise levels for this data set were set at a lod score  2.6 for significant linkage and  2.3 for suggestive linkage. The major findings of this analysis are displayed in Figure 2. The QTL on chromosome 6 was significantly linked, reaching a peak lod score of 3.0 at D6Mit183. In keeping with the previous nomenclature, we have designated this QTL on chromosome 6 as Aliq4. A QTL on chromosome 1 reached the level for suggestive linkage, with a peak corresponding to 4 cM distal to D1Mit213 and a lod score of 2.5. In addition, proximal chromosome 12 had two intervals reaching the level suggestive of linkage (both with lod scores of 2.3), exhibiting peaks at D12Mit185 and D12Mit112. Reanalysis and verification of the genotypes for markers in this region did not eliminate the dual peaks. Assuming independence of loci, the three QTLs had an additive lod score of 7.8 and explained 11.9% of the genetic variance (Table 1). The QTLs on chromosomes 8 (peak at 6 cM distal to D8Mit65, lod score 2.2), 9 (D9Mit227, lod score 1.6), and 16 (D16Mit152, lod score 1.6), initially identified as suggestive loci in the phenotypic extreme backcrosses, did not reach experiment-wise suggestive linkage in the total backcross population.

View larger version (17K): [in this window] [in a new window]   Figure 2.   QTL analysis results for chromosomes 1, 6, and 12 from the total backcross population (n = 307). Open circles represent positions of the designated MIT microsatellite markers. Solid lines (at lod score 2.6) and dashed lines (at lod score 2.3) denote significant or suggestive linkage, respectively, as determined empirically by 10,000 permutations of the original data set using QTL Cartographer (21, 22). The Aliq4 interval (peak at microsatellite marker D6Mit183) is shown on chromosome 6.

Cumulative search and gene interactions. The cumulative search function of MAPMAKER/QTL was performed on the phenotypic extreme and total populations, using the QTLs with at least suggestive lod scores in the original analysis (Table 1). The QTLs on chromosomes 1 and 6 had the largest lod scores. After fixing D6Mit183, D1Mit213 + 2 cM gave the highest cumulative lod score. Sequentially fixing loci to remove their variance, the QTLs on chromosomes 9, 8, 16, and then 12 were identified as contributing loci. In total, these six QTLs had a cumulative lod score of 12.5 and explained 41.6% of the genetic variance. Using the same strategy in the total backcross population, the QTL on chromosome 6 had the highest single lod score. After fixing this QTL (D6Mit183), D1Mit213 + 4 cM and then D12Mit112 were sequentially identified as contributing QTLs. Together, these three QTLs had a cumulative lod score of 8.3 and accounted for 12.2% of the genetic variance in the total backcross population (Table 1). MAPMAKER/ QTL and the computer program Epistat (25) were used to examine the total backcross data set for gene interactions among the putative QTLs. However, no evidence for significant gene interactions was detected.

Haplotype Analysis

To determine the contribution of each QTL and QTL combination to the overall phenotype, MSTs of mice with the same haplotype (i.e., sensitive versus resistance alleles) were calculated and then compared with the MSTs of mice with the opposing haplotype (Figure 3). For each backcross, only a homozygous A (AA) or heterozygous (H) genotype could be obtained in the typing of microsatellite markers. When the MSTs were determined for the groups of mice containing either the AA or H genotype at each of the markers representing the QTLs, the largest difference in MST was found for D6Mit183. Mice heterozygous at that locus survived an average of 12 h longer than did those mice homozygous for the A allele. Different haplotype combinations at two QTLs showed the greatest MST difference for mice heterozygous for QTLs on chromosomes 1 and 6, with H-H mice surviving an average of 25 h longer than did mice homozygous for A at both loci (Figure 3).

View larger version (27K): [in this window] [in a new window]   Figure 3.   Survival time differences of backcross mice with resistance versus sensitivity haplotypes at microsatellite markers representing the putative QTLs. Group on far right represents survival times of A/J (A) and C57BL/6J (B) control strains. Open bars, backcross mice with AA genotype (represented as A); filled bars, backcross mice with AB or BA genotype (represented as H), or BB genotype for C57BL/6J control mice. Number within each bar represents the number of backcross mice with the designated genotype or haplotype. Values are means ± SE. All comparisons of sensitivity (open bars) versus resistance (  filled bars) haplotypes were significant (P < 0.05, t test).

Continuing this analysis, examination of different combinations of three QTLs revealed that mice heterozygous for markers on chromosomes 6, 12, and 16 had a MST 40 h longer than did AA mice at these markers. The best agreement between phenotype and genotype for four QTLs was noted with haplotypes H-H-H-AA for markers on chromosomes 6, 12, 16, and 9, results that directly correlated with QTL results. Mice with this haplotype survived an average of 52 h longer than did mice that were AA-AA-AA-H for the four QTLs, respectively. Analysis of the different haplotypes for five QTLs showed a MST difference of 75 h for mice with H-H-H-H-AA at chromosomes 1, 6, 12, 16, and 9, respectively, compared with mice that were AA-AA-AA-AA-H for these markers. Adding in the genotype at the QTL on chromosome 8 decreased the MST difference between the haplotypes (AA-AA-AA-AA-AA-H versus H-H-H-H-H-AA) from 75 h to 70 h for QTLs on chromosomes 1, 6, 8, 12, 16, and 9, although the number of mice in each group was low (i.e., five mice had a sensitive haplotype for these QTLs and four had a resistant haplotype).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In previous studies, a significant difference in the ability of common laboratory inbred mice to survive a nickel-induced acute lung injury was identified; the A strain of mice was sensitive and the B6 strain was considerably more resistant (15). To further explore a genetic basis for the susceptibility differences, we now present results on QTL analysis of a large backcross population generated from the A and B6 strains. Initial analysis in the most sensitive and resistant backcross mice suggested six chromosomal regions requiring further analysis. Examination of the total backcross population identified a significant QTL on chromosome 6, with a peak lod score occurring at the microsatellite marker D6Mit183 (located 26.5 cM distal to the centromere) and two suggestive QTLs located on chromosomes 1 and 12. Although not reaching significance for linkage, haplotype analysis strongly supported two additional loci on chromosomes 9 and 16 contributing to the overall response. Using a modification of the Wright formula (23), five genes were estimated to be independently segregating with nickel-induced acute lung injury survival, thus in agreement with the QTL results.

To quantify (in hours) the contribution of each QTL individually and in combination to the overall survival time, the MSTs for mice with each specific haplotype were determined and compared. Those mice with the necessary opposing allelic combinations for QTLs on chromosomes 6, 9, 12, and 16 had a MST difference (52 h) that could account for the total difference in MSTs between the parental A and B6 strains (Figure 3). An even greater difference in MSTs was determined for mice with opposing resistance or sensitivity haplotypes at QTLs on chromosomes 1, 6, 9, 12, and 16. These five QTLs in haplotype analysis are in accord with the calculated estimate for the number of genes segregating with the phenotype.

The increased survival of mice with the resistance haplotype for the five QTLs (i.e., those on chromosomes 1, 6, 9, 12, and 16) was above that observed between the two parental strains and may relate to the QTL on chromosome 9. Although this QTL reached a lod score of only 1.6, it had a greater effect on the MSTs than either the QTL on chromosome 1 (lod score 2.5) or chromosome 8 (lod score 2.2) when its effects were added to those for QTLs on chromosomes 6 and 12. This QTL is of interest because its actions are in opposition to the other resistance QTLs in the B6 strain (i.e., the QTL appears to contain a susceptibility gene in the B6 strain). Thus, the A strain likely has resistance alleles for the chromosome 9 QTL. This scenario would mean that an A allele for this gene, when placed onto the B6 strain, would impart further resistance. In fact, many of the backcross mice in this study showed survival times significantly longer than that of the B6 parental mice. In addition, although a QTL on chromosome 9 was not identified in the analyses with ozone, many backcross and F₂ offspring showed more resistance than did the B6 parental strain (3, 4). Conversely, this would also imply that the transfer of a sensitivity allele for the chromosome 9 QTL (from B6) onto the A strain background could impart further sensitivity. This was not seen in the nickel- or ozone-exposed backcross mice, suggesting that the QTL on chromosome 9 may require a B6 allele at another QTL for its effects.

Several approaches can be pursued to narrow the identified QTL intervals. For example, increasing the number of mice to map the QTLs more precisely is an option; however, segregating QTLs contributes phenotypic noise, making it difficult to ascertain whether a given mouse has inherited a specific QTL. Thus, because many recombinants are likely to be uninformative, this strategy will probably not lead to significant refinement of the QTLs. A method to separate the effects of the different loci, and thereby allow individual QTL intervals to be resolved, is to generate congenic strains for each QTL. Once constructed, these strains can be bred to produce multicongenic lines to evaluate additive and epistatic effects of the QTLs to the overall phenotype.

Another method that can be undertaken concurrent to these strategies is a positional candidate-gene approach (28). Possible candidate genes for the putative QTLs are listed in Table 2. In the interval spanning a 1-lod-unit decrease on either side of the major QTL peak (D6Mit183, Aliq4), four genes located within this interval are of immediate interest as possible genes involved in nickel-induced acute lung injury, including aquaporin-1 (Aqp1), surfactant protein (SP)-B (Sftpb), thromboxane A synthase 1 (Tbxas1), and transforming growth factor (TGF)- (Tgfa). We have recently initiated studies to examine Tgfa as a gene candidate. Strong physiologic evidence of a major role for TGF- in acute lung injury resistance was provided in tests performed with mice overexpressing human TGF- in the lung. These mice were exposed to ultrafine Teflon particles (9) or nickel sulfate aerosol (Hardie and colleagues, in preparation) and demonstrated a significant increase in MST compared with their littermate controls. Thus, Tgfa is a positional candidate for Aliq4. Using microarray analysis of lungs retrieved from B6 mice after nickel inhalation, SP-B gene expression was found to decrease significantly with continuous exposure (29). Mice homozygous for gene-targeted SP-B succumb rapidly after birth to respiratory failure, indicating a pivotal role for this protein in alveolar function (30). Thus, SP-B appears to be an excellent positional candidate for further investigation as a susceptibility gene for this acute lung injury model. Other positional candidate genes for the QTLs are listed in Table 2.

A closer look at all the QTL regions for both nickel- and ozone-induced acute lung injuries identified two QTLs with overlap. A region near Aliq4 was identified in the analysis of the F₂ phenotypic extreme-responders for ozone-induced acute lung injury (4). This region also maps near QTLs implicated in airway hyperresponsiveness to acetylcholine (31) or methacholine (32), a locus the latter study designated Bhr5 for bronchial hyperresponsiveness QTL 5. The suggestive QTL on chromosome 12 for nickel-induced lung injury also coincides with a similar region identified with ozone (3). Thus, results suggest that genes on mouse chromosomes 6 and/or 12 may be involved in shared pathways important in nickel- and ozone-induced lung injuries.

In summary, QTL analysis of a large backcross population of mice identified significant linkage of a QTL on chromosome 6 (Aliq4) to nickel-induced lung injury, with modifier genes suggested on chromosomes 1, 12, and possibly 9 and 16. Results from analysis of specific haplotypes strongly support the combined effects of five QTLs in the overall phenotype, a result in agreement with the calculated number of genes independently segregating with the phenotype. Promising candidate genes reside within the QTL intervals and represent targets for additional studies. Two chromosomal intervals identified for nickel-induced acute lung injury were near regions also identified for ozone, suggesting that some of the same genes are involved in both responses. Resolution of the responsible genes within the QTL regions identified in this study could provide valuable insight into the pathways controlling acute lung injury.