We sought to determine whether the fa (leptin receptor) mutation was a major determinant of the putative obesity effects on respiratory frequency in an intercross between the Brown Norway (low breathing frequency, nonobese strain) and the Zucker (moderately high breathing frequency, with the fa mutation) strains. The hypothesis was that rats bearing one (heterozygote) or two (homozygote) alleles of the Glu296Pro point mutation (fa) would have a uniformly high respiratory frequency in the second filial (F2) generation, compared with wild-type animals. In addition to breathing frequency, tidal volume and minute ventilation were assessed during baseline, acute hypoxic (10% O2-0% CO2-balance nitrogen), hypercapnic (93% O2-7% CO2), hyperoxic (100% O2-0% CO2), and combined (10% O2-3% CO2-balance nitrogen) challenges in fa homozygote (fa/fa; n = 24), fa heterozygote (fa/wt; n = 33), and wild-type (wt/wt; n = 19) animals. Phenotypes were adjusted with stepwise regression analyses for the effects of age, sex, length, and litter size. Broad-sense heritability was estimated by examining the variance of the traits in first filial and F2 generations. ANOVAs were used to determine the mode of inheritance of the fa allele in the F2 generation. As anticipated, weight demonstrated the greatest overall broad-sense heritability (77%) and was the result of the recessive mutation. Breathing parameters during the hypoxic, hypercapnic, and combined challenges demonstrated a wide range of heritability from 5 to 96%, with a very nonuniform proportion of heritability explained by the leptin receptor. At best, for frequency 4.5 min into the hypercapnic hypoxic challenge, ∼20% of the total heritability (∼67%) could be attributed to an effect of the leptin receptor mutation. We conclude that, unlike its major effect on weight, the effect of the fa allele is not a major gene involved in the regulation of breathing frequency.
- minute ventilation
- tidal volume
- respiratory quotient
- Glu296Pro mutation
obesity is a major risk factor for a number of chronic respiratory conditions, including hypoventilation and sleep apnea, and for symptoms of dyspnea and exercise intolerance (13). The pathophysiological mechanisms that link obesity and disorders of respiratory control are often linked to mechanical loading of the respiratory system (13); however, evidence exists for a more direct influence on respiratory control by metabolic factors that also lead to the obese state (27). The discovery of leptin (1, 9, 21), an adipocyte-derived hormone that directly regulates both adiposity and energy homeostasis, intensified the interest in the role of molecular mechanisms in the development of obesity and its comorbidities. One primary mechanism of leptin, produced predominantly by white adipose tissue, is to inhibit neuropeptide Y, a potent stimulator of food intake (14). Thus leptin functions in part by decreasing food intake and partitioning metabolic fuels toward utilization and away from storage (9). Human obesity, which is associated with elevated plasma leptin levels, can be considered a state of resistance to leptin. Inheritance of a complete disruption of leptin receptor or of leptin is rare; thus other modifying peptides may lead to leptin resistance (45).
Leptin is also linked to the control of respiration. Mutant mice that constitutively lack circulating leptin display respiratory depression and elevated arterial Pco2 (44). Treatment with exogenous leptin improves the ventilatory response by increasing minute ventilation (V̇e), in both sleep and waking states (26), and has lead to the hypothesis that the leptin pathway is one common link between obesity and disorders of respiratory control such as sleep apnea (27).
The most common rat model studied for obesity is the Zucker strain. These animals become obese and hyperleptinemic because they have a missense mutation (Glu296Pro point mutation; fa) in the leptin receptor (12). Homozygotes (fa/fa) with this mutation exhibit markedly diminished responsiveness to leptin and are considered an in vivo surrogate model for leptin resistance, a feature of the human metabolic syndrome (15). One consistent finding in studies comparing lean (fa/wt or wt/wt, where wt represents wild type) and obese (fa/fa) Zucker animals is of a higher respiratory frequency in obese animals (see Refs. 16–18, 22, 24, 36, 40), thus leading to the conclusion that obesity per se induces higher breathing frequencies.
In contrast, another rat strain, the Koletsky, which also bears a mutation in the leptin receptor, shows breathing patterns different from the Zucker; obese Koletsky and Zucker rats, with similar fat accumulation and distribution patterns, respond differently to ventilatory challenges (39). In general, the Koletsky rats showed lower breathing frequencies and tidal volumes (Vt) than the Zucker rats, regardless of the respiratory challenge. This physiologically based experiment, which contrasted breathing patterns in rat strains with the obesity phenotype, demonstrated that obesity by itself may not be sufficient to explain the differences in the breathing rate and depth, leading to the conclusion that other sources of variation, such as strain (genetic) divergence, could be responsible for the dissimilarity in breathing patterns between Zucker and the Koletsky animals. If genes distinct from the leptin receptor would control ventilation, then phenotypes would segregate independently of obesity in intercrosses where weight and ventilation parameters differ between parental strains.
To determine which of the two competing mechanisms better explains breathing patterns in context of obesity, we intercrossed the Brown Norway (low breathing frequency, nonobese strain with normal/wild-type alleles at the leptin receptor) and the Zucker (moderately high breathing frequency, homozygous for the fa mutation) strains. In addition to the difference in the level of obesity, these two strains demonstrated substantial differences in frequency of respiration in our laboratory's (40) initial assessment of strain ventilation characteristics; thus this experiment is equivalent to establishment of a dihybrid cross. In physiological terms, we tested whether the second filial (F2) generation rats bearing one (heterozygote: fa/wt) or (homozygote: fa/fa) two alleles of the Glu296Pro point mutation in the leptin receptor (fa) had uniformly high breathing frequency when compared with their wild-type littermates during rest and chemosensory challenges. If an increase in breathing frequency cosegregated with leptin receptor mutation in the F2, then the difference in breathing frequency between the low-frequency strain (Brown Norway) and the moderately high-frequency strain (Zucker) can be attributed to the absence or presence, respectively, of the fa allele. We further decomposed the variability in frequency, Vt, and V̇e in the first filial (F1) and F2 generations into genetic (both additive and dominant) and environmental fractions and estimated the proportion that can be inherited (heritability). We also determined how much of this variability was due to the genotype at the leptin receptor gene, thus revealing the relative influence of the fa allele on respiratory frequency and other features of respiratory control.
MATERIALS AND METHODS
Animal breeding and history of the strains.
Two rat strains, Zucker and Brown Norway, were selected for an F2 cross because of the previously described strain differences in ventilatory and metabolic phenotypes (39, 40). The original source of the Brown Norway colony was Harlan Laboratories (Indianapolis, IN), where brother-sister mating was maintained for 23 generations, and further inbred for another five generations at the Medical College of Wisconsin. We obtained animals for this study directly from the Medical College of Wisconsin and maintained them using brother-sister matings for two additional generations before the intercross.
We derived a Zucker colony from fa/wt heterozygotes, generously provided by Dr. Barbara Horwitz at the University of California, Davis, where seven generations of cousin-cousin matings were maintained. After the founders were obtained, two generations of brother-sister matings were conducted in our laboratory. Heterozygotic fa animals were bred to produce the fatty phenotype, as homozygous fa/fa Zucker rats are obviously obese compared with heterozygous siblings (3). The parental (F0) rats used for testing consisted of Zucker fatty phenotype (Z/ucd/cwru: 7 males, 3 fatty/4 lean; 8 females, 2 fatty/6 lean) and the Brown Norway rat (BN/har/cwru: 8 males, 20 females). Animal breeding and testing were approved by the Louis Stokes DVA Medical Center Animal Care and Use Committee.
Female Brown Norway rats (n = 6) were mated to male Zucker rats (n = 3) to generate Brown Norway-Zucker F1 hybrids (n = 28). A reciprocal cross was not performed as female fatty Zuckers are uniformly poor breeders, although male fatty animals are more likely than not successful breeders (46). The Brown Norway-Zucker F1 hybrids were crossed to generate F2 progeny (n = 73). Ventilatory and metabolic parameters were measured on all F0, F1, and F2 animals.
Animal husbandry and other measures.
All rats were 14–20 wk of age at the time of testing and had been housed at Case Western Reserve University for at least 2 mo before our testing. Most animals were fed LabDiet 5001 (Rodent diet; PMI Nutrition International, Brentwood, MO). The average age of the F0 rats at measurement of the phenotypic parameters was 15 wk for the Brown Norway, 19 wk for the Zucker, and 14 wk for the F1 and F2 intercross animals.
Measurement of ventilatory traits.
Ventilation and metabolism were assessed by whole body plethysmography via the open-circuit method (40). Briefly, the chamber consisted of a 14-cm-diameter Plexiglas cylinder of 8.4 liter volume, with air intake and output ports to allow for different gas mixtures to be flushed through the chamber at a rate of 30 l/min and a low continuous flow of the gas to be drawn through the chamber during the testing period at a rate of 600 ml/min to prevent CO2 buildup and to maintain an ambient chamber temperature. A small opening at the top of the chamber contained a section of PE50 tubing, through which a sample of chamber air was obtained for assessment of oxygen and carbon dioxide concentrations within the chamber for metabolic analysis. Ventilatory parameters evaluated were V̇e or the total volume of air breathed per minute (in units of ml/min), Vt (in units of ml), and breathing frequency (in units of breaths/min); metabolic parameters evaluated were oxygen consumption (V̇o2; in units of ml/min), carbon dioxide production (V̇co2; in units of ml/min), and respiratory quotient (the ratio of V̇co2 to V̇o2). V̇e was computed as breathing frequency multiplied by the Vt. Ventilatory parameters were recorded both by strip chart and with an analog-to-digital converter coupled to a computer. Two setups allowed for two animals to be tested at a time.
The sequence of testing is done only once per animal so that each animal comes to the testing paradigm without prior experience to the sequence of challenges. Hence, random sources of variance in environment present at each point and behavioral responses may influence trait values. Testing was done between 10:00 AM and 1:00 PM. Each animal was allowed 45 min to acclimatize to the chamber. Temperature (via an implanted animal identification-temperature transponder; model IPTT 100, Biomedical Data Systems, Seaford, DE) was measured before and after each testing session. Baseline resting ventilation and carbon dioxide and oxygen concentrations were continuously recorded during this acclimatization period. At the end of the acclimatization period, five measurements were taken over a 15-min period to provide resting values (a1) for ventilation and metabolism (see Table 1). A 5-min presentation of the test gases was conducted in the following order: 10% O2-balance nitrogen (h) and then 100% O2 (o). This was followed by a 20-min rest period in room air and second collection of resting values (a2), after which the animal was then exposed to a 5 min of 7% CO2-93% O2 (c) and then 3% CO2-10% O2-balance nitrogen (b).
Our logic for the sequence of testing was as follows. The first sequence tests a traditional hypoxic exposure, without controlling for hypocapnia, within which there might be a complex of responses deemed early (h1 and h2) and late responses (h4 and h5) (34). The timing of data collection was chosen to represent these transitions from peripheral to central chemosensitivity, which in reality may be quite variable. The reoxygenation with oxygen was chosen to allow us to examine also an early response (o1 and o2), e.g., a “Dejours” test (4), which is known to differ among strains (41). Values at a later time (o4 and o5) would be more likely to represent the effect of hyperoxia at rest and thus an inhibition of carotid body activity at rest at a time more remote from hypoxic exposure. The second series of exposures was chosen to test central drive with hyperoxic hypercapnia. Proceeding onto a hypercapnic hypoxic gas mixture introduces complexity, but the intent was to see whether the transition from hyperoxic hypercapnia to hypoxia in the presence of supplemental carbon dioxide to prevent severe hypocapnia might give insight into interactive effects. The time points of interest were therefore the later values (c4–c5, b4–b5).
During the challenges, ventilatory parameters were continuously recorded, and values representative of each challenge were obtained in the last 15–20 s of each minute during the challenge. We used a respiratory-based software program to score breaths. Sniffs or sighs were not included in the calculations of Vt and breathing frequency. A mean value under each condition was entered for each animal. At the end of testing, the animal was removed from the chamber, and body weight in grams and nose-to-anus body length in centimeters were obtained. We calculated the Lee index using a standard formula [Lee index = (body weight)1/3/body length].
A portion of the liver (all F2 animals) or tail (Brown Norway progenitor animals) was collected, frozen in liquid nitrogen, and stored at −80°C before DNA isolation. We extracted DNA using the DNeasy tissue kit (catalog no. 69506, Qiagen, Valencia, CA). To genotype the fatty locus and distinguish lean (heterozygotes for the fa mutation; fa/wt) from fatty (homozygotes for the fa mutation; fa/fa) Zucker animals, PCR was performed with 100 ng of DNA, 1 μM of each primer (5′-GTT TGC GTA TGG AAG TCA CAG-3′ and 5-ACC AGC AGA GAT GTA TCC GAG-3′), 0.5 U of Taq polymerase (GIBCO BRL), and 200 μM dNTP (Pharmacia) in a final volume of 25 μl (31). The template was initially denatured at 94°C for 5 min. The samples were then run through 30 cycles of 94°C for 30 s, 54°C for 30 s, and 72°C for 30 s. There was a final extension of 72°C for 5 min. Ten microliters of PCR product were digested with MspI at 37°C for 1 h and analyzed on a 1% agarose gel at 135 V for 1.5 h, stained with ethidium bromide. The fa mutation is due to a single A-to-C change at nucleotide 880 in the leptin receptor gene (31). The A-to-C transversion causes a glutamic acid to proline substitution at residue 269. This mutation also introduces an MspI restriction site (CCGG) at the fa allele, which can be visualized by restriction fragment analysis of the leptin receptor PCR product.
Normality of the exposure variables and adjustment for covariates.
We examined metabolic and ventilatory variables for departure from normality using a Kolmogorov-Smirnov test. Outliers that were greater than three standard deviations from the mean value were treated as missing data. If, after removal of outliers, the data were still not normally distributed, the data were transformed with the use of the method described by Healy (10).
Stepwise regression analyses were used to identify significant covariates from among the following variables: age, sex, length, and litter size. Because the fa mutation has an impact on weight and Lee index, these were considered as confounders, and no adjustments were performed for these variables. Exposure variables were only adjusted for covariates that were considered significant at 5% level using the univariate ANOVA routine in SPSS (SPSS, Chicago, IL).
Because the Brown Norway rats are inbred and demonstrate little to no genetic variation between animals, we assume that variation in the trait values between animals (, where σ2 denotes variance and e denotes the environmental component) is largely due to environmental factors. The Zucker strain is considered an outbred strain, and, although the Zucker animals used in this study might have reduced heterozygosity due to seven generations of cousin-cousin followed by two generations of brother-sister mating, a high degree of homozygosity at all alleles in the founders, similar to that for the Brown Norway strain, could not be assumed. Because the Zucker founders have an unknown degree of homozygosity, we had to assume that their trait variation is due to a combination of environmental and genetic factors, and such genetic variation is passed on to the F1 and F2 generations. Given this circumstance, heritability in the broad sense (5) was calculated with the assumption that where g represents the genetic component and c is an unknown value between 0 and 1.
We estimate as a weighted average of variance (Brown Norway) and the absolute value of 2 [variance (F1) − variance (Zucker)] based on sample size. Hence, heritability can be estimated as H2 = [variance (F2) − ]/variance (F2). We calculated standard errors of these estimates of heritability using the delta method (2).
Effect of the fa allele.
For each of the traits, we determined the effect of the fa allele by calculating the phenotype mean for each of the three possible genotypes in the F2 generation. ANOVA was used for comparisons of ventilation and its components, metabolism, and derivative values among F2 animals with homozygous wild-type (wt/wt), heterozygous (fa/wt) and homozygous affected (fa/fa) to obtain specific modes of inheritance. Four different models (recessive, dominant, additive, and overdominant) were fitted to the data with the use of the SPSS statistical software package. For example, a recessive model referred to animals with a fa/fa genotype having a trait value significantly different from the fa/wt and wt/wt animals, with the fa allele being recessive to the wild-type allele. The dominant model was very similar to the recessive, with the wild-type (wt/wt) genotype being recessive to the other two groups. In the additive model, the value of the heterozygote (fa/wt) category was fixed exactly midway between the means of the two homozygotes. Finally, to fit a dominance component where each allele interacted in a multiplicative manner, an overdominance model was also examined. Here, the heterozygote (fa/wt) was allowed to assume a value two times greater than the average of the two homozygote classes, such that the joint effect of the two alleles (in the heterozygote) was greater than expected if the effect of each were considered singly (as in each homozygote). The best-fitting ANOVA model was chosen on the basis of which had the largest F statistic. The threshold for reporting a model was a significance level of P < 0.05 or greater. Because models were run on the 53 traits with heritability estimates, a Bonferroni correction was also performed to estimate a conservative level of statistical significance with an α = 0.05/53 = 0.000943 level.
To determine how the fa allele affected heritability of ventilatory traits, 28 Brown Norway rats, 15 Zucker rats, 28 Brown Norway-Zucker F1 hybrids, and 73 F2 progeny were measured for several ventilatory and metabolic parameters. Among the F2 generation, genotyping of the Glu296Pro mutation in the leptin receptor was necessary to distinguish between homozygous (wt/wt) and heterozygous (fa/wt) lean animals. Data were tested for departure from Hardy-Weinberg equilibrium at the leptin receptor locus with a χ2 test. The data fit Hardy-Weinberg proportions (5).
Before further analysis, phenotypic data were transformed to have an approximately normal distribution and adjusted for the effects of significant covariates. The variations in the mean values and the variabilities for each generation and for the actions of the fa allele after we had adjusted for the effect of covariates are presented in Tables 2–6, at rest and during various chemosensory challenges. Strain and sex effects are also noted when found to be significant. As previously described, frequency, followed by V̇e, was determined to have significant differences between the F0 Brown Norway and the Zucker strains, with the Brown Norway being thinner and having a lower breathing frequency. The differences in breathing frequency between the two strains persisted during the chemosensory challenges (Tables 2–6). Sex effects were present more often at rest and during hyperoxia and to a lesser extent with the hypercapnic hypoxic challenge. Brown Norway and Zucker animals showed no significant differences in age, length, V̇o2, V̇co2, or respiratory quotient (Table 2). In summary, these parental strains have divergence in response to ventilatory challenges, with Zucker rats in general having higher chemoresponsiveness than the Brown Norway animals (Table 2) (39, 40).
F1 animals are 50:50 genomic hybrids between Brown Norway female and Zucker male (fa/fa) animals had trait means that often fell in between the means for the two parental strains. As an example, F0 Brown Norway with zero fa alleles weighed on average 208.63 g vs. 322.29 g for the F0 Zucker with two fa alleles (Table 2). The F1 animals with one copy of the fa allele weighed 260.59 g, a value in between the two progenitor strains. Similarly, the value of the mean breathing frequency at rest in the F1, at 112.13 breaths/min, was in between the values for each progenitor strain (Table 2).
In the F2 generation, we could follow the segregation of the fa mutation and determine whether the ventilatory parameters cosegregated with the dosage of the fa mutation. Therefore, the mean and standard deviations for ventilatory parameters partitioned by leptin receptor genotype in the F2 generation are provided (Tables 2–6). Also provided are the means and standard deviations for all the F2 animals when not differentiated by fa allele status. Thus we tested the hypothesis that rats bearing one (heterozygote) or two (homozygote) alleles of the Glu296Pro point mutation (fa) had a uniformly high respiratory frequency during rest and during chemosensory challenges, compared with wild-type animals. Similar hypotheses were tested for Vt and V̇e. The cosegregation of the leptin receptor genotypes and ventilatory parameters can be modeled with dominant, recessive, additive, and overdominant modes of inheritance, as in a two-way cross of a binary character and a quantitative trait. The approximate model of inheritance can be inferred from the trait means partitioned by the leptin receptor genotype (Tables 2–6). However, as described below, we obtained the best-fitting model using statistical methods, including best fit to a genetic mode of inheritance (Table 7).
Overall broad-sense heritability, which encompasses both additive (allelic) and dominant (genotypic) genetic components, was estimated from covariate-adjusted data. Phenotypes with measurable (greater than zero) heritability values are shown in Table 7. As expected, indexes of weight and mass (Lee index) demonstrated significant heritability (>70%). V̇co2 also exhibited high heritability values (∼55%). Many measures of ventilatory behavior at rest and during chemosensory challenge conditions also demonstrated moderate (>20%) to high (>50%) heritability.
The models for the fa allelic effect with the best fit of the inheritance patterns of dominance, recessiveness, additivity, and overdominance are reported in Table 7. Only variables with significant evidence for heritability were tested, and only those models with P values <0.05 are reported. As expected, a recessive model strongly fit the data for weight and for Lee index. In contrast, the model for length best fit an additive model. In an additive model, the mean for the heterozygote (fa/wt) was at the halfway point between those for the fa/fa and the wt/wt homozygotes. With regard to ventilatory behavior, a variety of models were applicable; however, it should be noted that only one met a conservative level of significance, largely due to the stringent criteria imposed by the number of traits being tested. A variety of modes of inheritance of the leptin receptor allele fit the traits for ventilation.
As the fa allele was tracked in the cross, we could determine the proportion of variance attributable to the leptin receptor locus under a specific mode of inheritance. For example, during the “isocapnic hypoxic” challenge (b4.5) the variation at the fa locus explained 20.1% of the total heritability (∼0.67 or 67%) under an additive model (P < 0.0009; Table 7). In traits like weight and Lee index, the effect of the fa allele at the leptin receptor locus is expectedly large (>35–45%); in contrast, only 10% of the heritability of values of V̇co2 was accounted for by the fa mutation. Generally small contributions of the leptin receptor were observed in ventilatory traits, and this contribution was not consistent within a chemosensory challenge (i.e., at 2, 3, etc. minutes into the challenge). We conclude that inheritance patterns of traits for ventilatory behavior in this intercross are largely unexplained by the fa allele and are likely to represent the contributions of other regions of the rat genome.
The magnitude and direction of the effects of the fa allele are reported by β̂ (Table 7), which is equivalent to a regression coefficient. This parameter, β̂, provides an estimate of the effect of one fa allele in the units of measurement for the variables of interest. An example of a rather strong effect is observed in the trait for frequency during the hypoxic challenge (h3; Table 7). In this instance, the variation at the fa locus explained 17.1% of the total heritability (∼0.76 or 76%) under an additive model (P = 0.002; Table 7), and animals bearing one copy of the Glu296Pro allele would have an increase in breathing frequency of 13.5 breaths/min, whereas those bearing two copies of the Glu296Pro allele would have an increase of 27 breaths/min 3 min into a hypoxic challenge. Other genetic models were also feasible for some of the traits. For instance, at the second baseline (a2), values for frequency fit a model where each allele interacted in a multiplicative manner (overdominance), such that the mean for the heterozygote was greater than the average of the two homozygote categories.
In short, compared with a major effect of the fa allele on weight and body mass (Lee index), its effect on respiratory frequency, Vt, and V̇e was small, and models were unpredictable.
We conducted an intercross of Brown Norway and Zucker strains to explore the inheritance of ventilatory traits (Vt, breathing frequency, and V̇e) during baseline and chemosensory challenges (hypoxia, hypercapnia, and hyperoxia) in the context of a known mutation producing obesity. In contrast to the strong effects of the fa allele on weight and body mass, the effect size on ventilatory behavior at rest and during several different chemosensory stimuli was more consistent with a minor or modifying gene effect. At maximum, the leptin receptor locus accounted for ∼20.1% of the total heritability in breathing frequency (Table 7; heritability = 67% at 4.5 min into the isocapnic hypoxia challenge). During other chemosensory challenges, even smaller amounts of the variation in the breathing frequency were attributed to the leptin receptor allele. Similar observations can be made about Vt and V̇e, which were not as consistently heritable as breathing frequency but nevertheless did show occasional evidence of a genetic basis. Based on these data, we conclude that the fa allele is not a major gene for ventilatory behavior in this intercross.
With regard to the traits of weight and mass, as expected, the fa allele acted in a recessive mode (46). Most importantly, but predictably, the leptin receptor accounted for 46.8% of the total heritability for weight (heritability of 77.5%). Within the Zucker strain, not all of the metabolic parameters relating to fat and glucose metabolism and turnover associate with the fa mutation in a dose-response manner (30). Specifically, in a study of Brown Norway-Zucker F1 generation, an action of one fa allele (fa/wt) was apparent with regard to increased fat content even at 7 days of life but did not necessarily translate to measurable differences in thermoregulatory thermogenesis and plasma concentrations of insulin and triglycerides found in the fa/fa pups (38). In our study, such detailed measurements of fat distribution and biochemical factors were beyond the scope of the study; however, this serves as an example of the complex actions of the leptin receptor gene and how it might contributes to some of the ventilatory values found in the heterozygotes.
Alternately, obesity mechanisms that act independently of the leptin receptor may influence ventilatory traits. In a previous genome scan of obstructive sleep apnea and body mass index in Caucasians, Palmer and Redline (29) did not obtain any evidence for the human ortholog of the leptin receptor on 1q32 being linked to either body mass index or to the apnea-hypopnea index. They did, however, determine some evidence for linkage to the apnea-hypopnea index explained by body mass. Furthermore, they found evidence of linkage to obesity-related traits on chromosomes 2p, 7p, and 12q. In our model, only ∼50% of the total heritability for weight was explained by the leptin receptor; the remainder of the variance is probably due to other genes for obesity that are strain specific. In the literature, the actions of leptin and its receptor have been cited extensively in mediating the effects of obesity on ventilation. Our data demonstrate that the genetic architecture of ventilation in the context of obesity is complex, involving many genes (oligogeny) and/or gene-gene (epistatic) interactions. Evidence to sort which other obesity-related genes may influence ventilation await a genome scan of this intercross.
Because the leptin receptor does control some proportion (∼20%) of the total heritability in isocapnic hypoxia, we need to consider pleiotropic actions of the leptin system and the potential for epistatic effects. To some extent, these issues have been addressed in previous studies. For example, information from knockout mouse models suggests that leptin and leptin receptors affect ventilation. The ob/ob C57BL/6J mouse compared with the null littermate exhibited differences in hypercapnic ventilation before pronounced obesity emerged, but changes in baseline breathing appeared to follow age-dependent increases in body weight (42). In the mouse model of absent leptin, leptin administration improved the hypercapnic response (44). Thus leptin and the leptin system may influence the trajectory of hypercapnic responses independent of fat accumulation and distribution. However, even the presence of obesity is accompanied by differences in ventilatory behavior (32), perhaps because the genetic background of the db/db mouse differs from the ob/ob model, as well as by gender (33). The presence of metabolic derangements such as ketosis would have an effect independent of obesity (32).
Furthermore, the actions of the leptin receptor may modify a number of other systems directly and indirectly involved in respiratory control. There are effects not only on body weight and lipid metabolism but also on other physiological systems, including immune function (7, 25, 35). Relevant to breathing, several systems and pathways are known to be abnormal in Zucker animals. There are differences between lean and fatty Zucker siblings in temperature regulation (19, 20) and in dopaminergic effects on hypoxia (22), as well as in the response to pharmacological interventions, including inhibition of neuronal nitric oxide synthase (23), opioid blockade (16, 36), and a noncompetitive glutamate N-methyl-d-aspartate receptor antagonist (18). Many of these effects are not solely accounted for by differences in respiratory mechanics between lean and obese phenotypes (6). Thus, even the fa allele can have substantial effects on ventilatory behavior and other components of respiratory control that go beyond the known primary targets for leptin in the central nervous system (7).
Finally, the progenitor background appears to have an impact on ventilatory behavior under many conditions of challenge. In the Brown Norway strain, downstream or upstream pathways may differ from those in the Zucker strain in structural or functional mechanisms that are encoded by other parts of the genome. A full genome scan is needed to identify such loci. What is clear is that the obese animals in the F2 generation have ventilatory behaviors that are different from those observed in Zucker obese parental strains. Correspondingly, the wt/wt animals in the F2 generation resemble the Zucker grandparents in ventilatory behavior, which further supports the hypothesis that the obese phenotype is not associated with a unique ventilatory pattern. We suspect ventilatory behavior is a consequence of multiple genes determining some proportion of the variance in ventilation at rest and with chemosensory challenge and that the fa allele and its physiological effects are only one of many components that are manifested by these genetic variants. In summary, our results lead us to conclude that the action of the leptin receptor is insufficient to explain the genetic variation in ventilatory traits.
There are some design limitations. This was a rather small data set, and it was designed specifically to track the fa allele. The power to estimate the effects and/or numbers of other putative genes is limited. Because a cross between any two rat strains represents only a small proportion of the potential genetic variation in this species, our estimates of heritability are only applicable to this cross; other strains may have novel alleles or loci that also modulate ventilation parameters. Second, this was not a reciprocal intercross design because Zucker females homozygotic for the fa allele are very poor breeders (46). Reproductive success with male Zucker homozygotes was more effective when Brown Norway females were exposed to a number of these rats. However, this directional breeding strategy limited us from studying the effects of the Brown Norway Y chromosome and the Zucker XX genotype. Third, because heritability values were estimated from a limited number of F2 animals across a large number of traits, the precision of the estimates and models are modest at best. Fourth, the Zucker strain is not entirely homozygous (inbred) at all loci; therefore, there was additional variance encountered by such heterozygosity. This necessitated a general approach to the estimation of broad-sense heritability (5). Estimates can be refined to address additive genetic effects and inheritance across multiple alleles with a genome-wide scan, accompanied by measurements in a larger number of F2 animals, as statistical power is improved, the degree of heterozygosity is determined empirically, and modes of inheritance are defined for specific regions of the genome other than the fa locus. Finally, researchers could disentangle the confounding effects in obesity and ventilation measures by creating a conditional knockout or knockin for the leptin receptor that can be expressed in selected tissues at particular developmental phases. However, this work is beyond the scope of the present study.
Other, more practical, limitations were related to specific testing procedures, which were implemented based on the need for high throughput at times when litters of particular ages were to be tested, with attention given to uniformity in the testing schedule and high reproducibility of results. Thus animals were brought to the laboratory at a certain time of day and after acclimatization to the testing environment. To encourage animals to be awake during the study, they had ∼5 h of light exposure before coming to the laboratory, and testing was accomplished within the next 3 h. Although the use of plethysmography raises some unavoidable technical issues, this approach is favorable compared with the potential influences of restraint or anesthetic variance. The variability in an animal for frequency and Vt is within the range reported in the literature for rodents (28). In the analysis, problems with the accuracy and reproducibility are accounted for as technical and environmental variance. The identification of strain differences by us and by others indicates that the study of the unanesthetized rodents with body plethysmography is an appropriate first step (8, 11, 43).
Results indicate that attention to the time and manner of collection of phenotype value may be quite important in studies that seek to uncover naturally occurring allelic variations that operate in the regulation of ventilation. For instance, we found that measures of resting breathing taken before and the several minutes after the initial challenge with hypoxia and reoxygenation are modeled with different degrees of heritability. This difference could reflect how genetic background might shape ventilatory behavior during a series of challenges. In a practical sense, one needs to be explicit about the testing paradigm, especially if comparisons are to be made across studies from different laboratories, and the interpretations from linkages studies may need to account for the presence of such differences in proprioception.
In summary, traits for ventilatory behavior at rest and with chemosensory challenge show inheritance, but the relative influence from the fa locus is not consistent with a major gene effect. Other regions of the rat genome and various patterns of inheritance may contribute to adult ventilatory phenotypes.
The Department of Veterans Affairs and National Heart, Lung, and Blood Institute Grants HL-97015, HL-54587, and HL-07567 (training grant) supported this research.
We thank Dr. Robert Elston for advice on the modeling components of this study.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society