INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

HYPOXIC VENTILATORY SENSITIVITY varies among human subjects and represents a continuum from low to high responsive subgroups (2, 7, 10, 11, 34). Hirshman et al. (7) demonstrated a five- to sixfold difference between extreme ventilatory responses in a group of 44 subjects. Although phenotypes in these subjects varied positively with height, weight, and the ventilatory response to hypercapnia, the bimodal attribute of the distribution suggested that a major genetic determinant underlies individual variation in hypoxic ventilatory sensitivity.

Our laboratory has demonstrated a similar continuum among many inbred mouse strains (32). This suggested that the genetic diversity, which regulates variation in hypoxic ventilation, is conserved across species, representing the basis for comparative mapping studies between mouse and human genomes (27). The C3H/HeJ (C3) and C57BL/6J (B6) inbred strains have been the focus of our research because these strains exhibit breathing differences during a broad array of inspirate challenges, including acute hypoxia. Although C3 and B6 mice were comparable with respect to their hypoxic minute ventilation (E), the strains differed in breathing pattern; i.e., C3 mice were characterized by a slow, deep hypoxic ventilatory response (HVR) relative to the rapid, shallow phenotype of B6 mice. Moreover, the HVR profile of first-generation offspring (i.e., B6C3F₁/J) of C3 and B6 progenitors represented a third phenotype distinguished by hypoventilation, which was attributable to a marked attenuation in hypoxic tidal volume (VT). By using quantitative genetic approaches, our studies further showed that variation in HVR characteristics, including E and VT, was inherited by second-generation offspring in segregation patterns consistent with two-gene models (31).

The purpose of the present study is to link differential HVR characteristics between C3 and B6 mice to candidate regions of the mouse genome. To achieve this objective, a genome-wide screen was performed using 176 microsatellite markers to analyze DNA samples from 52 B6C3F₂ (F₂) offspring. The F₂ offspring were previously characterized with respect to eupneic, hypoxic, and hypercapnic ventilatory responses; a quantitative genetic analysis of these results are reported elsewhere (31). The microsatellite markers represented simple sequence repeat fragments polymorphic between C3 and B6 mice and were evenly distributed across the mouse genome. The hypothesis of the present study suggests that phenotypic differences in hypoxic E and VT among the parental strains and the B6C3F₁/J (F₁) offspring are regulated by major genetic determinants. The results demonstrate that a candidate genomic region on mouse chromosome 9 explains a substantial proportion of the phenotypic variation in hypoxic E and VT and therefore represents a significant quantitative trait locus (QTL) for differential HVR in this genetic model.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Male and female F₁ mice were purchased from Jackson Laboratory (Bar Harbor, ME) to establish breeding colonies at the Johns Hopkins School of Public Health. Intercross progeny were generated from F₁ progenitors and weaned at 4-5 wk of age. From the breeding colonies, randomly selected male F₂ offspring (n = 69) were housed in cages of four to six animals for an additional 6-12 wk. Water and mouse chow (Agway Pro-Lab RMH 1000) were provided ad libitum. All animal protocols were reviewed and approved by the Animal Care and Use Committee of the Johns Hopkins School of Public Health.

HVR. The magnitude and pattern of ventilation were determined by whole body plethysmography that used unrestrained and unanesthetized conditions. The methods and results have been described in detail and are reported elsewhere (31, 32). Briefly, ventilatory function was evaluated during a sequence of acute (3-5 min) inspiratory challenges including hypoxic (inspired O₂ fraction = 0.10) admixtures in mild hypercapnia (inspired CO₂ fraction = 0.03). During hypoxia, the E responses were comparable between C3 and B6 parental strains but were significantly (P < 0.01) lower in F₁ mice. There was a consistent strain difference in breathing frequency (f) between C3 and B6 mice during eupneic, hypoxic, and hypercapnic conditions, and F₁ mice resembled the B6 progenitor. In contrast, VT responses were greatest in C3 mice, and the F₁ offspring demonstrated markedly lower VT responses during hypoxic conditions compared with B6 mice. These observations, among others, compelled us to characterize the HVR profile of F₁ mice as a third phenotype representing hypoxic hypoventilation.

DNA isolation. After measurements of HVR were made, DNA samples were isolated from 52 randomly selected F₂ mice. Each animal was euthanized with pentobarbital sodium to obtain kidney tissue. To isolate high-molecular-weight DNA (22), sample tissue was gently homogenized using a Dounce homogenizer in an isotonic, high-pH solution that included Nonidet P-40. To release genomic DNA, nuclear membranes were lysed in a sodium dodecyl sulfate-proteinase K solution. After incubation at 42°C for 3 h, DNA was separated from other cellular components by extraction using a phenol-chloroform-isoamyl alcohol solution. DNA was precipitated with ice-cold 95% ethanol and 0.15 M potassium chloride. Sample DNA was spooled on a glass rod, rinsed with 70% ethanol, and stored in Tris-EDTA buffer (pH = 8.0). The purity of the sample DNA was determined using spectrophotometry and by calculating the ratio of optical densities (OD) at wavelengths of 260 and 280 nm (i.e., a target OD₂₆₀-to-OD₂₈₀ ratio of 1.8).

PCR amplification. Primers surrounding DNA microsatellite markers (n = 176) polymorphic between C3 and B6 mice were purchased from Research Genetics (Huntsville, AL). The primers were used to amplify sample DNA. One primer from each pair was end-labeled with -³²P using the following reaction conditions: 5.4 µl of sterile H₂O, 3.6 µl of 5× kinase buffer [final concentration equal to 0.25 M Tris (pH = 9.0), 0.05 M MgCl₂, 0.05 M dithiothreitol, and 0.25 mg/ml bovine serum albumin], 4.3 µl of primer (10 µM), 0.7 µl of T4 polynucleotide kinase (7 units), and 4 µl of [-³²P]ATP. The solution was incubated at 37°C for 1 h and diluted with 26 µl of sterile H₂O followed by column purification to remove unincorporated -³²P.

Marker selection. At the time of this study, ~500 PCR-based markers were available throughout the mouse genome that differed between C3 and B6 strains. Markers were selected to establish 10- to 15-centimorgan (cM) intervals between two loci and to provide coverage of the entire mouse genome with 95% confidence. For F₂ offspring, a sample size of 50 mice theoretically excludes 10-15 cM of the genome for each marker using a set of ~150 well-spaced markers (i.e., given the mouse genome is ~1,500 cM long). Although this estimate was used in the present study, three to six additional markers were used to saturate genomic regions showing suggestive linkage [log-likelihood values (LOD) > 2.8]. Therefore, the linkage analysis in the present study incorporated ~95% of the mouse genome with marker intervals of <15 cM. In genomic regions showing suggestive linkage, the size of the gene-containing region was reduced to 1- to 3-cM intervals.

Linkage analysis. Linkage analysis was performed with the use of the computer software programs Mapmaker-EXP and Mapmaker-QTL (13, 14). The distances between loci map assignments were established using Mapmaker-EXP. Because the order of DNA markers was generally predetermined, the results obtained from Mapmaker-EXP concerning distances between loci were compared with known distances defined for each marker. Interval mapping of phenotypic data was accomplished using Mapmaker-QTL. This analysis incorporated high-density genetic maps to distinguish between plausible QTL and weak effects from distant loci. Subroutines and algorithms for Mapmaker-QTL were based on assumptions that quantitative traits are normally distributed. Therefore, power-transformed data sets were shared between Mapmaker-EXP and Mapmaker-QTL to ascertain linkage and compute LOD scores. Linkage was inferred when the log of the odds ratio (i.e., called a LOD score) in favor of linkage over nonlinkage was 1,000 to 1 (i.e., LOD > 3.3). Alternatively, nonlinkage was deduced when the odds against linkage were 100 to 1 (i.e., LOD < 1.8). The candidate region surrounding putative QTL was examined for genes with biological relevance to hypoxic ventilatory control. Finally, comparative mapping was performed to determine homologies between the mouse and human genomes.

Supplementary data analysis. Correlation and linear regression analyses were performed to estimate the strength of association between hypoxic E and its components. The paired components of E are either VT and f or mean inspiratory flow (VT/TI, where TI is inspiratory time) and inspiratory fraction (TI/Ttot, where Ttot is the total respiratory time). In the latter case, VT/TI and TI/Ttot components are used to identify the "driver" and "timer" elements (18), respectively, as a general model of respiratory control mechanisms. Although the paired components are obvious mathematical correlates of E, dissociation between components acts to accentuate one trait to explain the greatest proportion of genetic variance among individuals.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In Fig. 1, cosegregation analysis was used to examine the relationship between hypoxic E responses and either VT or f components of an individual response. As suggested by correlation coefficients, both VT and f demonstrated significant (P < 0.0001) positive associations with hypoxic E. The VT component accounted for ~72% of the variance in hypoxic E, whereas f explained ~49% of the variance.

View larger version (18K): [in this window] [in a new window]   Fig. 1.   Cosegregation plots depict the relationship between either tidal volume (VT, left) or breathing frequency (f, right) and minute ventilation (E, the product of VT and f) for B6C3F2 mice (n = 69 F2 animals) during acute hypoxic inspirate challenge [i.e., inspired O2 fraction (FIO2) = 0.10 and inspired CO2 fraction (FICO2) = 0.03 in N2]. Correlation coefficients and linear regression analysis suggest that individual variation in hypoxic E cosegregates with both VT and f components (P < 0.001).

In Fig. 2, cosegregation was used to examine the relationship between hypoxic E responses as a function of both VT/TI and TI/Ttot components. The correlation coefficients suggested that only VT/TI responses were significantly (P < 0.0001) associated with hypoxic E; that is, there was relatively little association (P > 0.01) between hypoxic TI/Ttot and E responses. In this case, the VT/TI component accounted for ~92% of the variance in hypoxic E.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   Cosegregation plots depict the relationship between either mean inspiratory flow (VT/TI, left) or inspiratory fraction (TI/Ttot, right) and E (the product of VT/TI and TI/Ttot) for 69 F2 mice during acute hypoxic inspirate challenge (i.e., FIO2 = 0.10 and FICO2 = 0.03 in N2). Correlation coefficients and linear regression analysis suggest that individual variation in hypoxic E tightly cosegregates with hypoxic VT/TI (P < 0.0001) but not with TI/Ttot responses (P > 0.01).

As shown in Fig. 3, LOD scores were plotted as a function of the relative marker distance for hypoxic E, VT, and VT/TI responses. On the basis of peak LOD scores of 4.5, 4.0, and 5.1 for hypoxic E, VT, and VT/TI responses, respectively, a putative QTL was established for a candidate region in close proximity to D9Mit207 on mouse chromosome 9. For each of the three HVR characteristics, the putative QTL explained ~30% or more of the phenotypic variation among F₂ offspring.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   LOD (i.e., log likelihood) score graphs illustrate linkage results of 52 F2 mice between DNA markers on mouse chromosome 9 and hypoxic E (top), VT (middle), and VT/TI (bottom) responses. The peak LOD score for each hypoxic response is located in proximity to D9Mit207, which is a marker ~33 centimorgans (cM) from the centromere. Two potential candidate genes are also depicted: Drd2, dopamine D2 receptor; Chrna5, nicotinic acetylcholine receptor 5-subunit.

In Fig. 4, individual responses for F₂ mice were plotted for homozygous and heterozygous genotypes at the D9Mit207 locus of chromosome 9. The allelic frequencies were not significantly different from Mendelian proportions ( = 1.04; P > 0.05). For each HVR characteristic, the response distributions of the heterozygotes were significantly (P < 0.01) lower compared with the two other homozygous groups. In addition, response distributions of the homozygotes with alleles derived from the C3 progenitor spanned the distributions of the other two groups.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Hypoxic E (left), VT (middle), and VT/TI (right) responses for 52 F2 mice are depicted as a function of genotype at the D9Mit207 locus. For each hypoxic response, the heterozygotes (shaded symbols) are significantly (P < 0.01) lower compared with the other two groups. In addition, the variation for C3-like homozygotes (open symbols) spans the range of responses extended by the other two groups. Means ± SE for each genotypic group (i.e., B, B6-like homozygotes; H, heterozygotes; A, C3-like homozygotes) are shown to the right. *P < 0.01.

In Table 1, peak LOD scores were reported for the corresponding hypoxic ventilatory traits. The genome-wide screen produced one other putative QTL (LOD score = 3.3) on mouse chromosome 3 in the vicinity of D3Mit17 for hypoxic VT/TI responses. This QTL was shown to account for ~26% of the variance in the hypoxic VT/TI responses among the F₂ mice. Two other suggestive QTL (LOD scores = 2.8) were shown to link HVR response variation to mouse chromosomes 2 and 6. 

In Fig. 5, the allelic interaction between D9Mit207 and D3Mit17 loci was illustrated for the hypoxic VT/TI responses in F₂ mice. The allelic frequencies were not significantly different from Mendelian proportions ( = 7.24; P > 0.05). In addition, a two-way ANOVA suggested that there was a significant effect of both D9Mit207 (F statistic_df=2 = 10.36; P < 0.001) and D3Mit17 (F statistic_df=2 = 6.92; P < 0.01) loci in determining variation in hypoxic VT/TI responses among F₂ mice.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   Hypoxic VT/TI responses are depicted as a function of genotypes at two loci represented by D9Mit207 and D3Mit17 markers. The D3Mit17 genotype interacts with the D9Mit207 locus and influences hypoxic VT/TI responses in C3-like homozygotes at the D9Mit207 locus.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study demonstrates a putative QTL on mouse chromosome 9 that confers variation in HVR responses among F₂ mice derived from C3 and B6 progenitors. Collectively, the evidence suggests that the QTL influences hypoxic E by regulating the VT component of the response (Fig. 1, left). In addition, individual variation in hypoxic VT/TI responses tightly cosegregates with hypoxic E responses (Fig. 2, left). Furthermore, VT/TI emerges to be the HVR characteristic that demonstrates the strongest linkage relationship with a genomic region proximal to D9Mit207, which is a marker ~33 cM from the centromere on mouse chromosome 9 (Fig. 3). This genotype-phenotype association suggests that the QTL regulates individual variation in hypoxic E by altering a mechanism identified with the neural drive component of breathing (18). In contrast, the respiratory timing components of hypoxic E, including f, TI, expiratory time, and TI/Ttot, are not linked to the QTL on mouse chromosome 9 in this segregant offspring class (Table 1).

Although the C3 and B6 parental strains are comparable with respect to hypoxic E, the responses of F₁ mice are significantly attenuated relative to both progenitors due to reduced hypoxic VT and VT/TI responses (31). One hypothesis, which may explain the variation in HVR, is supported by two major strain differences between C3 and B6 parental strains: 1) a reduced lung volume and compliance in the B6 progenitor (33) and 2) a muted ventilatory response to chemical inspirate challenge in the C3 progenitor (32). The F₁ mice inherit a phenotypic profile that includes a smaller lung volume from the B6 progenitor and a muted chemical response from the C3 progenitor. Therefore, one testable hypothesis suggests that the attenuated HVR of F₁ mice is attributable to two major genetic determinants, which are dominant for reduced lung size and diminished chemical responsiveness.

Inheritance studies suggest that the attenuated hypoxic VT and VT/TI response phenotypes of F₁ mice are conferred to second-generation offspring (31). In addition, similar phenotypic profiles are detectable in selected recombinant inbred strains derived from C3 and B6 progenitors (29). Collectively, these results support the hypothesis that a small number of heritable traits culminate to produce a phenotypic profile representing hypoxic hypoventilation. Recent evidence combining backcross and intercross offspring of C3 and B6 strains suggest that the inheritance patterns for HVR variation are consistent with models that incorporate two unlinked genes (31). The results from the present study demonstrate that, in addition to a locus on mouse chromosome 9, there is another putative QTL on mouse chromosome 3 linked to variation in hypoxic VT/TI in this model. This QTL is positioned in a genomic region adjacent to D3Mit17, which is a locus ~72 cM from the centromere. The D3Mit17 locus differs from the putative QTL assigned to mouse chromosome 3, Itbq1 (located ~25 cM from the centromere), which determines differences in inspiratory timing at baseline (30). Indeed, our laboratory has shown very little association between hypoxic and baseline ventilatory responses in this genetic model (29, 31).

The segregation of hypoxic E, VT, and VT/TI responses of F₂ mice based on their genotype at the D9Mit207 locus does not appear to be consistent with dominant or codominant genetic models (15). Although the allelic frequencies are consistent with Mendelian proportions, the average HVR in the heterozygotes are significantly attenuated compared with the average HVR responses of the homozygotes (Fig. 4). Likewise, the variation of the C3-like homozygotes spans the range of the other two groups. It is difficult to explain this segregation pattern based on Mendelian principles. A study by Kingsley et al. (12) constructed a similar linkage map of mouse chromosome 9 using the B6 progenitor. These investigators concluded that there was no evidence of segregation distortion or chromosome rearrangement in the candidate genomic region described in the present study. Collectively, these observations support the hypothesis that a second influential genetic determinant may be interacting with the gene at the D9Mit207 locus. Alternatively, the results are consistent with an underdominance genetic model (36, 37).

To test whether a second gene is interacting with the D9Mit207 locus, hypoxic VT/TI responses are reclassified on the basis of genotypes at both the D9Mit207 and D3Mit17 loci. In Fig. 5, the variation in hypoxic VT/TI responses is shown to be independent of the genotype at the D3Mit17 locus for B6-like homozygous and heterozygous genotypes at the D9Mit17 locus; that is, the B6-like homozygotes are consistently high, whereas the heterozygotes are consistently low. However, the D3Mit17 genotype interacts with the D9Mit207 locus and influences hypoxic VT/TI responses in C3-like homozygotes at the D9Mit207 locus. The gene-gene interaction is manifested in B6-like homozygotes at the D3Mit17 locus to produce an attenuated hypoxic VT/TI response. Although larger numbers of F₂ mice are essential to discern the significance of the potential interaction between the two QTL, the present data suggest that a gene on chromosome 3 modifies the genetic effects on HVR assigned to the QTL on chromosome 9.

The LOD plots in Fig. 3 appear to incorporate a secondary peak associated with D9Mit2. Although a two-QTL model is rejected for the genomic region between D9Mit2 and D9Mit208 using Mapmaker-QTL, this broader genomic region cannot be excluded for future consideration. In addition to ~100 known genes in the 19-cM interval between D9Mit2 and D9Mit208 (17 and 36 cM from the centromere, respectively), there are likely many more novel genes in this candidate genomic region. The current working hypothesis considers genes with biological relevance to differential hypoxic VT/TI as likely candidates. More specifically, we postulate that the candidate genes in this region influence HVR variation by modifying the VT response achieved during acute hypoxic exposure.

The most obvious candidate gene in the genomic region of mouse chromosome 9 is the Drd2 gene, which is located ~28 cM from the centromere and encodes the dopamine D₂ receptor. The effect of D₂ receptors on hypoxic ventilatory control has been investigated in many different species, including mice. For example, Olson and Saunders (19) demonstrated that intraperitoneal administration of dopamine and levodopa at relatively high doses caused a depression in acute HVR. In a subsequent study (20), these investigators demonstrated an augmented acute HVR in mice after administration of a dopamine antagonist, droperidol. In general, these results are consistent with the acute effects specific to D₂-receptor antagonists such as domperidone. That is, D₂-receptor blockade releases the inhibitory effects of dopamine acting via D₂ receptors, resulting in an increase in acute hypoxic ventilation. The results of Iturriaga et al. (9) suggested that the release of this modulatory role of dopamine, acting through D₂ receptors, resulted from increased excitation of carotid chemosensory activity.

More recently, Huey et al. (8) investigated acute HVR in D₂-receptor-deficient homozygous mice on a B6 and 129-hybrid genetic background. Although there were substantial gender effects, these investigators concluded that hypoxic and/or hypercapnic ventilation was significantly greater in D₂-deficient mice compared with wild-type littermates. In many cases, genotypic differences in ventilation were derived from a significantly greater VT response in D₂-deficient mice. The consistency between the study of Huey et al. and the results of this study suggest that a DNA polymorphism in the Drd2 gene exists between C3 and B6 mice. This possibility is further supported by results of Cotzias and colleagues (3, 28), showing strain-dependent differences in neurological responses to levodopa between C3 and B6 strains.

Another candidate gene more proximal to the D9Mit207 locus on chromosome 9 is the Chrna5 gene, which is located ~32 cM from the centromere (4, 9a). Other members of the nicotinic acetylcholine receptor gene family are also located on chromosome 9, including the Chrna3 and Chrnb4 genes. Collectively, the genes encode neuronal nicotinic acetylcholine receptor subunits, which are widely distributed throughout the central and peripheral nervous system. Nicotinic acetylcholine receptor subunits form pentameric channels resulting in heterogeneous expression and functional diversity (5, 23, 35). Pharmacological studies suggested that C3 and B6 mice responded differently to acute nicotine administration in a dose-dependent fashion (1, 17). Although the strains did not differ in nicotinic or -bungartoxin binding in many regions of the central nervous system (16), a dose-dependent increase in ventilatory response to nicotine was greater in B6 compared with C3 mice (17). Whether DNA variation in nicotinic acetylcholine receptor subunits, such as Chrna5 or others on chromosome 9, plays a role in differential HVR between C3 and B6 strains requires future studies.

Considering the observation that the heterozygotes demonstrate an attenuated hypoxic VT/TI response relative to both parental strains is not easily understood using simple Mendelian models. One speculative explanation with respect to the present linkage results suggests that underdominance of the heterozygotes is attributable to parental DNA variation in subunit composition of heteromeric channels. Functional deficiencies may emerge, for example, if DNA variation in the Chrna5 gene regulates differential heterologous expression of the 5-subunit. Recent studies suggest that 5 substitution for 7 or in combination with -subunits produces dysfunctional nicotinic acetylcholine receptor complexes (5, 35). Therefore, a cogitative explanation for underdominance linked to the D9Mit207 locus suggests that variation in the Chrna5 gene modifies heteromeric composition resulting in an amplification of dysfunctional nicotinic acetylcholine receptors. The observation that 7-subunits have been shown in nerve fibers of the carotid body (26) provides neuroanatomic support for this provisional hypothesis.

Perspective. In a 1992 review article honoring Pierre Dejours, Shea and Guz (25) proposed that individual variation in the magnitude and pattern of breathing is likely attributable to the interaction between lung mechanics and chemosensitivity. An individual outcome results in a breathing strategy that optimizes the work of breathing. With acute hypoxic challenge, individual variation in chemosensitivity may emerge to overcome the limits imposed by variation in lung mechanics, and optimal work of breathing may be restored. Powell and colleagues (21) outline specific mechanisms associated with the time course of hypoxic ventilation. These investigators proposed at least three distinct mechanisms operating within the first several minutes to influence acute HVR. The most immediate response is an augmentation of ventilation mediated by peripheral chemoreception and glutamanergic mechanisms in the nucleus tractus solitarius. Short-term potentiation and depression are mechanisms that follow the acute response. The latter two stages involve other respiratory control centers, including respiratory rhythm generation, which are not specific to peripheral chemoreception. Although it is impossible to delineate between the mechanisms described above using our experimental protocol, it is clear that C3 and B6 parental strains vary with respect to respiratory timing, lung mechanics, and ventilatory responsiveness to chemical challenge.