Current supportive therapy has positively impacted survival outcome; however, mortality from ALI and ARDS is still unacceptably high (4
). Using different agents to induce ALI mortality, we have carried out genetic studies with a long-term objective of identifying major gene(s) affecting survival. Results from these studies, obtained by recombinant inbred (RI) analysis and separate QTL analyses of backcross and F2
mice derived from A- and B-strains of inbred mice, identified multiple QTLs (i.e., Aliq1-4
) significantly linked to ALI survival, with several additional QTLs suggestive of linkage (20
). The present studies were initiated to further assess the genetic factors involved in ALI survival, to delineate the contributions of each of these significant and suggestive QTLs linked to ALI survival time, and to establish refined mouse models to identify the gene(s) comprising the major effect QTL Aliq1
As an initial strategy, additional QTL analyses were performed on total backcross and F2
datasets. The backcross population had not been analyzed previously as a single population and prior threshold values were established based on theoretical values (25
). Permutations of the total backcross population set an empirical threshold LOD score at 2.6, which was less than the proposed theoretical value for a backcross (3.3). This analysis supported previous QTLs and added Aliq5
, previously identified as suggestive QTLs on chromosomes 7 and 12, respectively. These QTLs were further supported by results from QTL genotype analysis and by in vivo
results with corresponding CSSs.
Using QTL genotype analysis, MSTs of mice carrying the same sensitive or resistant genotypes at each single and combination of QTLs predicted that a three-QTL model could explain nearly all of the phenotypic variance between the A and B progenitor strains. In addition, a four-QTL model yielded a group of mice for which up to 89% (16/18) died within the sensitive A-strain range. This analysis identified Aliq1 as the best single QTL model; mice A-A for D11Mit263 (n = 133) had a MST of 13.7 hours. Including the markers for chromosomes 7 and 12 (and either chromosome 6 or 13) into the QTL genotype analysis identified mice with a MST of about 10 hours. Groups of mice for the two-, three-, and four-QTL models averaged a survival time that fell within the normal range of the sensitive A-strain (i.e., 4–12 h). QTL genotype analysis results generally supported QTL analysis results, providing additional strength to the earlier QTLs.
QTL genotype analysis is a method to quantify the genotype to phenotype relationship. However, the method has an inherent level of uncertainty. For example, when assessing all mice that are A-A at one specific marker, this total group will include mice with differing genotypes at other important QTL markers, and at markers throughout the rest of the genome. Adding additional markers to the analysis decreases the number of mice with the desired genotype (e.g., A-A or B-B for all markers), but also decreases the genetic variability of the mice within the group. Therefore, the method can detect general trends of phenotype–genotype associations and can estimate the number of interacting QTLs contributing to the overall phenotypic difference between the parental strains. The three-QTL model parsimoniously agrees with separate estimates for the number of genes segregating with the survival time phenotype and also agrees with the three highly significant or significant linkages identified in the QTL analysis of the combined backcross plus F2 dataset.
Because a complete line of CSSs derived from the A and B strains (all 19 autosomes and the X- and Y-chromosomes) are available, we examined whether resistant B-strain mice that contained the corresponding individual chromosomes from the sensitive A-strain (each linked to a decrease in ALI survival time) would have increased sensitivity in ozone. Results of QTL analysis and QTL genotype analysis predicted that B.A-11 mice (carrying Aliq1
, the QTL with the largest LOD score and percent variance explained) would have the greatest increase in susceptibility (i.e., decreased MST). The decreased MST of about 4 hours for B.A-11 mice suggests that Aliq1
controls around a third of the phenotypic difference between B and A parental strains. This in vivo
result compares favorably to QTL analysis, which determined that Aliq1
explained 42% of the variance in backcrosses derived from these strains (20
). B.A-13 mice (carrying the sensitive Aliq2
alleles) demonstrated the most sensitivity in ozone, with a MST of 13.1 hours. Thus, chromosome 13 controls at least half (53%) of the phenotypic variance between progenitor A and B strains. Neither QTL nor QTL genotype analysis forecasted this, and results suggest that one or more additional QTLs may occur on chromosome 13. Other putative susceptibility QTLs (chromosomes 1, 6, 7, and 12) were also confirmed using the appropriate CSSs. However, Aliq3
on chromosome 17—identified only by RI analysis—may not, on its own, contribute to overall phenotype.
Lung W:D weight ratios were also determined in B.A-6 and B.A-13 CSSs to correlate survival time to lung injury. Results were consistent with MSTs for these strains. The initial 3-hour ratio showed the slowest gain in lung edema for all strains, suggesting a certain innate ability to initially withstand the insult. After 3 hours, the slopes of all curves increased due to an increase in lung water; however, as predicted by its MST, the A-strain slope was by far the steepest. Thus, these data support the premise that the A, B.A-6 and B.A-13 strains lack one or more resistance genes that initially slow the progression of lung edema in the B strain. Alternatively, compared to the A strain, the B, B.A-6 and B.A-13 strains lack one or more sensitivity genes that cause or allow lung edema to progress. Extending these W:D weight ratio curves beyond the last exposure times recorded, identified that a W:D weight ratio of about 7.0 () correlated well with the MST of each line. From the projected intersections of W:D weight ratios and MSTs, it is evident that the different lines of mice succumb with similar lung W:D weight ratios, but the time to reach this threshold differs among the lines. This finding gives hope to the possibility that identifying a gene or set of genes that allow one to withstand such alveolar flooding may yield insight into pathologic mechanisms that could ultimately direct pharmacologic intervention.
Given that Aliq1 and Aliq4 were both susceptibility QTLs for ALI survival, we expected that B.A-6,11 double CSS mice would be more sensitive than B.A-11 CSS mice. In addition, QTL genotype analysis also predicted an increased sensitivity by demonstrating that backcross and F2 mice that were A-A for both QTL peak markers on chromosomes 6 and 11 had a MST of 12.8 hours, compared to 20.9 hours for B-strain control mice. Contrary to our hypothesis that the QTL effects would be additive, the B.A-6,11 double CSS mice totally lost the susceptibility phenotype seen in the separate CSSs, surviving ozone-induced ALI as long as the progenitor B-strain. This result was further supported by a reversal of the B.A-11 increase in W:D weight ratio by the double CCS mice, when compared to that of the resistant B-strain. Given that QTL genotype analysis only takes into account the genotypes at specific QTL markers, it is understandable how this method would not detect an interaction that eliminates the separate QTL effects, especially an effect that is due to other markers on chromosomes 6 and 11.
As with all Aliq loci, it is too soon to know whether the QTLs on chromosomes 6 and 11 represent individual genes or a combination of closely linked genes. Nor is it known whether the chromosome 6 and 11 QTLs act directly (i.e., causal) or indirectly (i.e., unmask suppressive loci), either alone or with other QTLs located on these or other chromosomes, to generate the CSS phenotypes. Therefore, numerous scenarios can be conceived to explain the loss of phenotype in the B.A-6,11 double consomic mice. QTL genotype analysis predicted an increased sensitivity in the combined B.A-6,11 line; thus, the loss of phenotype suggests that an epistatic effector(s) lies distant to the identified QTL peak marker(s). However, this effector could reside outside the QTLs on chromosomes 6 or 11 or elsewhere in the genome, depending on nature of the interaction. For example, the reversal of phenotype could be due to a combination of genes on chromosomes 6 and 11 interacting to suppress the susceptibility effects produced by genes on the individual chromosomes. Just as likely, the gene(s) on one chromosome may unmask the inhibitory effect of another gene, either within or outside of a QTL region, thereby blocking the sensitivity seen in the single consomic lines. Regardless, results clearly demonstrate that significantly different phenotypes can be generated from gene–gene interactions, which are directly or indirectly relevant to loci on these chromosomes. The exact mechanisms will ultimately depend on identification of the major genes involved. Generating and testing a panel of double congenic lines for Aliq1 and Aliq4 will help determine whether the loss of sensitivity is due to loci within or outside the respective QTL intervals, and whether these loci act directly or indirectly.
To validate the Aliq1 effect and to move towards identification of the quantitative trait gene(s) for Aliq1, we generated reciprocal congenic lines of mice (B.A11 and A.B11) with overlapping regions around the putative QTL interval. A significant change in survival time was successfully captured in both directions. B.A11-5 had an MST of 15.9 hours, accounting for 33% of the difference between A and B controls in these exposures (), and similar to the difference seen in the B.A-11 CSS. A.B11-1 and A.B11-2 congenic lines demonstrated a significant increase in survival time (26% and 29%, respectively) over A-strain controls. Results from the B.A-derived congenic lines suggest that the QTL effect is distal to D11Mit67 at 96.82 Mb (Build 36) and results from the A.B-derived congenic lines suggest that the QTL effect is proximal to D11Mit336 at 110.44 Mb. This effectively reduces the Aliq1 interval by 56%, from 30.6 Mb to 13.6 Mb. Interestingly, this region maps to, and extends beyond, the distal most region of the 1.5-LOD confidence interval identified by QTL analysis. Results from the reciprocal congenic lines give additional support for Aliq1 containing a gene (or a set of closely linked genes) contributing to overall survival and provide further evidence that a positional cloning analysis for Aliq1 should be possible.
Mining through the known genes mapping to the refined Aliq1
interval identified several genes associated with inflammation, epithelial damage or repair, ALI, or water imbalance. Among these, many contain exonic SNPs between the A and B strains that result in amino acid changes (), as well as SNPs in 3′, 5′, and/or splice sites. One excellent candidate gene is Angiotensin converting enzyme (Ace
), which maps near the center of the interval (105.81 Mb) and has been implicated in ALI/ARDS in humans and animals (11
). Recent cohort studies demonstrated a significant association between an ACE
insertion (I allele) or deletion (D allele) polymorphism and the susceptibility and outcome of patients with ARDS (11
). Specifically, the D/D genotype for the ACE
gene was significantly associated with mortality in the ARDS group compared with a control cohort (41
), and patients carrying the ACE
I/I genotype had a significantly increased survival rate (11
). It is important to note that the insertion/deletion polymorphism in ACE
is a 287-bp noncoding intronic sequence; therefore, it likely cosegregates with the causative locus. One additional point of interest with Ace
as a candidate gene for Aliq1
is that the Aliq2
two-LOD support interval on chromosome 13 contains the angiotensin II receptor, type 1a, further supporting a role for the renin-angiotensin system in ALI survival. Several additional positional candidate genes for Aliq1
are listed in , along with their potential roles in differential ALI survival.
In summary, several methods have been utilized to further assess previously identified QTLs for their individual and collective contributions to ozone-induced ALI survival. QTL genotype analysis on the combined A- and B-strain–derived backcross and F2 data demonstrated that the collective effects of three QTLs could explain much of the phenotypic difference seen between the A and B parental strains. These results were consistent with minimal gene estimates and results from QTL analysis of the combined dataset. B.A-derived CSSs for seven purported suggestive and significant ALI QTLs supported all but Aliq3 on chromosome 17 as susceptibility QTLs. Chromosome 13 and, by extension Aliq2, had the most effect on ALI survival time. To examine whether the predicted QTL genotype interaction between the Aliq1 and Aliq4 leads to increased sensitivity, a double CSS line (B.A-6,11) was generated and tested in ozone. The double CSS lost its sensitivity phenotype, demonstrating a similar MST to the resistant B-strain. The implications of this loss of phenotype await resolution of the individual QTLs. Reciprocal congenic lines for Aliq1 captured the QTL effect and refined the QTL interval. Results from these studies confirm the importance of several QTLs in ALI survival and set the stage for further studies to resolve its complexity.