HomeLarge Type Edition
HOME ARCHIVE SEARCH TABLE OF CONTENTS
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
PubMed
Right arrow PubMed Citation
The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 59:B118-B125 (2004)
© 2004 The Gerontological Society of America

Quantitative Trait Loci Specifying the Response of Body Temperature to Dietary Restriction

Brad A. Rikke1, John E. Yerg, III1, Matthew E. Battaglia1, Tim R. Nagy2, David B. Allison3 and Thomas E. Johnson1

1 Institute for Behavioral Genetics, University of Colorado at Boulder.
2 Department of Nutrition Sciences, University of Alabama at Birmingham.
3 Department of Biostatistics, Section on Statistical Genetics, University of Alabama at Birmingham.


   Abstract
Top
 Abstract
Methods
 Results
 Discussion
 References
 
Dietary restriction (DR) retards aging and mortality across a variety of taxa. In homeotherms, one of the hallmarks of DR is lower mean body temperature (Tb), which might be directly responsible for some aspects of DR-mediated life extension. We conducted a quantitative trait locus (QTL) analysis of the response of Tb to DR in mice using a panel of 22 LSXSS recombinant inbred strains, tested in two cohorts. Tb in response to DR had a significant genetic component, explaining ~35% of the phenotypic variation. We mapped a statistically significant QTL to chromosome 9 and a provisional QTL to chromosome 17, which together accounted for about two thirds of the genetic variation. Such QTLs could be used to critically test whether the response of Tb to DR also affects the response of life extension. In addition, this study demonstrates the feasibility of trying to map QTLs that affect other physiological responses to DR, including the life extension response. Importantly, the genes underlying such QTLs would be causal factors affecting these responses and could be identified by positional cloning.

THE ability of dietary restriction (DR) to extend mammalian life span was first demonstrated in the 1930s (1). Later studies, beginning in the 1970s, demonstrated that DR also has a profound ability to delay aging-related physiological declines and diseases (2). However, the physiological effects of DR have turned out to be nearly as broad and complex as aging itself, making it difficult to elucidate the molecular mechanism(s) underlying DR's benefits. One, as yet, untried means of uncovering the connections between DR, altered physiology, and reduced rates of aging is the forward-genetic approach.

Forward genetics typically begins with detecting genetic variation followed by genetic mapping in which single genes or quantitative trait loci (QTLs) are detected and localized. Subsequently, positional cloning can be used to identify the genes underlying these QTLs (3,4). Positional cloning involves identifying the genes based on their physical location and is "hypothesis free", i.e., it does not require predictions of molecular mechanism or mode of action, although informed guesses can expedite the process. This approach is well suited to identifying genes that specify complex traits and has been used to identify genes involved in Alzheimer's disease, asthma, and diabetes, as well as many others (4).

Previous studies have shown that one of the hallmarks of DR in mammals is lower body temperature (Tb) (2,5), and recently we demonstrated that laboratory mice exhibit a large amount of strain variation in their Tb response to DR (6). We also demonstrated that this is a direct effect of DR and ruled out several alternative interpretations (see Discussion). In this article, we combine these previous results with studies from a second cohort to demonstrate that this strain variation has a statistically significant genetic component. Moreover, we map a statistically significant QTL specifying the response of Tb to DR to chromosome 9 (Tbdr1) and a provisional QTL to chromosome 17.


   METHODS
Top
 Abstract
 Methods
Results
 Discussion
 References
 
All mice were housed under specific pathogen-free conditions. The strains, husbandry, and diet were the same as previously described (6). The mice in the present study (Cohort 2) were all female and singly housed in 26 x 12 x 14 cm polypropylene cages for the entire study. AL food intake was measured once each week at the same time. The DR mice were fed 60% of the intake of ad libitum-fed (AL) controls of the same strain and age. The DR rations were precisely weighed out for each individual mouse. The DR mice were fed double rations on Mondays and Wednesdays and a triple ration on Fridays throughout the entire study; whereas the DR mice in Cohort 1 were initially fed single rations every day through the first temperature trial and were then switched to the Monday/Wednesday/Friday feeding schedule prior to their second temperature trial.

To minimize strain variation due to a large change in AL food intake due to refilling each food hopper only when it was nearly empty (we found that AL mice eat considerably more when the food is fresh from the container), we replenished the food for each AL mouse each week (typically adding ~30 g to bring the total back up to ~200 g). The new food was added to the top of the old food, so that the mice only had access to the older food at the bottom of the hopper. Because the older food is mixed each week as it is being weighed, the food being eaten from the bottom of the hopper is a mix of food that is 1–6 weeks old (averaging 31/2 weeks) for each mouse. Therefore, the food being eaten by the AL mice was not particularly old, and the weekly replenishment helped maintain the freshness at a more constant level from mouse to mouse and from week to week.

The number of mice tested and the mean food intake per strain are shown in Table 1. The mice tested in Trial 1 are the same mice tested in subsequent trials unless a mouse died and/or a previously untested mouse was added. Also, as shown in Table 1, we observed an excess in the mortality of the DR mice that was largely due to strain R20, with three R20 mice dying shortly after measuring their voluntary runwheel activity. Excluding the R20s, the DR and AL mortalities were similar (11 DR and 9 AL deaths). Mice that died did not have mean Tbs that were unusually different from the other mice for that strain and diet.


View this table:
[in this window]
[in a new window]
  		Table 1. Mice Used in Cohort 2.

 
Temperature Measurement
For Cohorts 1 and 2a, Tbs were measured by passive radio telemetry using Bio Medic Data System's IPTT-100 transponders (Seaford, DE) as previously described (6). For nearly all of the mice in Cohort 2b (started 4 months after Cohort 2a), we used Bio Medic's next-generation IPTT-200 transponders. The linearity of the IPTT-200 transponders was measured between 20°C and 38°C at 1° intervals using a water bath. No deviation from linearity was found (R2 =.99) based on readings collected from 17–20 transponders at each temperature.

Because some transponders were clearly out of calibration, over the course of several days immediately following each temperature trial, we directly compared 3–4 transponder readings from each mouse with its rectal temperature (Thermalert TH-5, 1.7 cm probe, Physitemp, Clifton, NJ). This also minimized the potential for mouse-to-mouse variation due to the exact placement of the transponders on the mice. The mean difference between the transponder readings and rectal temperatures [after adjusting for the nonlinearity of the IPTT-100 transponders (6)] was then used to correct the transponder readings obtained during the temperature trial. The mean Tb of each mouse was then calculated as previously described (6).

Temperature Trial 1, consisting only of Cohort 2a, began May 23, 2001 at 3:30 PM and ended after the 3:30 PM time point on May 30 (7 days). Temperatures were collected at 6-hour intervals. Twenty-nine temperatures (1.5%) were missed either inadvertently or due to transponders that stopped working. For reasons previously described (6), missing values were imputed for the AL mice using multiple linear regression and for the DR mice by averaging the temperatures flanking the missing temperature. Room temperature during Trial 1 averaged 24.1°C (SD [standard deviation] = 0.2).

In Trial 2, Tb was measured every 4 hours on Cohorts 2a and 2b beginning at 4:30 PM on July 11, 2001 and ending after the 4:30 PM time point on July 18 (7 days). There were 112 temperatures (2.0%) imputed largely due to IPTT-200 transponders that stopped working. For data missing over several consecutive time points, we imputed the average Tbs from the other DR mice for that strain. Room temperature during Trial 2 averaged 23.3°C (SD = 0.3°C).

In Trial 3, Tb was measured every 6 hours on Cohorts 2a and 2b beginning at 3:30 PM on December 7, 2001. The 9:30 PM time point on December 7 was missed for all mice. Measurements were suspended after the 9:30 AM time point on December 10 and restarted at the 3:30 PM time point on December 14. Testing continued for the next 7 days, ending after the 3:30 PM time point on December 21 (10.25 days total). There were 93 temperatures (1.8%) imputed due to transponder failures or mice that were inadvertently missed. Room temperature during Trial 3 averaged 22.7°C (SD = 0.1).

We initially used a 6-hour time interval in Trial 1 because it is much less labor intensive than the 4-hour interval that we used in our study of Cohort 1, but would still be nearly as accurate for measuring mean Tb. Our results from this temperature trial, however, did not correlate with our results from Cohort 1 as closely as we were expecting. Therefore, we used a 4-hour interval in Trial 2 to match what we did with Cohort 1. The Trial 2 results, however, were not much different from the Trial 1 results. For this reason, and to do a more direct test for differences due to aging or time on DR, we used the same time points in Trial 3 as we used in Trial 1. For the purpose of genetic mapping, averaging across trials with different time points helps to ensure the robustness of the results.

For genetic mapping, we used Map Manager QTXb17, released October 2002 (7). This program includes built-in functions that we used to control for background effects of other loci, control for differences in the strain variance, and search for epistatic interactions. Because the LSXSS recombinant inbreds (RIs) were derived from noninbred parents (8), a substantial number of markers were segregating for alternate alleles [i.e., alleles different from those in the inbred LS (ILS) and inbred SS (ISS) strains and typically much less frequent (6)]. These alternate alleles often caused the permutation test to fail because QTXb17 inappropriately tried to recode the alternate alleles as ISS or ILS. Therefore, to carry out the permutation testing we used an older version of Map Manager, QTb27, which excluded the strains at each marker having an alternate allele (appropriate since marker regression was only done on the ISS and ILS alleles at each locus). The permutation tests were conducted using 10,000 permutations, 1 cM intervals, an additive regression model, and no control for other QTLs (except where noted). At the QTL locations, D9Mit4 and D9Mit256 were segregating for alternate alleles, whereas D9Mit315, D17Mit49, and D17Mit10 were not.

Our initial mapping was conducted using 72 markers with an average spacing of ~20 cM across the genome. We then genotyped the strains at additional microsatellite markers on chromosomes 9 and 17 to further define the QTL regions. These markers were assayed using the polymerase chain reaction as previously described (9) and were included in the permutation testing.

Other Measurements
Home-cage activity was monitored using infrared motion detectors mounted above the cage (10). To maximize viewing area, the food was moved to the cage floor and the water bottle was replaced with a 15 ml tube with a sipper. The data were collected in 15-minute bins over 18–24-hour intervals (usually 24 hours) for 2 consecutive days using two detectors per cage. Eight cages were monitored at a time, typically 1 DR and 1 AL mouse each from four different strains. To adjust for differences in detector sensitivity, we normalized the readings of each detector to an arbitrarily designated reference detector. Immediately after monitoring home-cage activity, we monitored the 8 mice for voluntary runwheel activity for 2 days (typically) using Nalgene runwheels as previously described (6) (results to be published separately). Home-cage and runwheel activity were monitored from October 1, 2001 to January 2, 2002, when the Cohort 2a mice were 11–13 months of age and the Cohort 2b mice were 7–9 months of age. The activity measurements were suspended during Temperature Trial 3.

Feces:food ratios were measured on all but 8 mice in Cohort 2 during Temperature Trial 2. The fecal droppings were collected once at the end of a 2-week period from 9, 10 to 23, 24 July 2001 and isolated as previously described (6). Nine feces:food ratios were excluded as outliers (<4%), largely due to feces weights that were more than 50% different from the next closest weight for that strain and diet.

Calorimetry was conducted on droppings collected at the end of a 2–4-week period encompassing Temperature Trial 3. Two mice per strain and diet were sampled, with the exception of R4 (3 AL mice), R6 (1 DR mouse), R7 (1 AL mouse), and R40 (3 AL mice). Calorie content was determined twice on each sample using a Parr 1261 semimicrobomb per manufacturer's instructions (Parr Instrument Company, Moline, IL).

The mice in this study were also used to measure tail growth and hair regrowth after plucking a dorsal area of ~1 cm2 (Cohort 2 only) (results to be published). No other measures were conducted on these mice and no blood was drawn.

Calculations
Correlations, regressions, and multiple linear regressions were conducted as previously described (6). The RI heritability (h2RI, narrow sense) of the response of Tb to DR was estimated from a one-way analysis of variance as described by Belknap and colleagues (11) using the residuals from DR mean Tb after regressing on AL mean Tb (trials combined and cohorts equally weighted). To calculate the residuals for the individual mice (to estimate the variance within strain), the predicted value of DR mean Tb was calculated using the AL strain mean for Tb as the independent variable in the regression equation. The lower limit of the 95% confidence interval (CI) was estimated using the F critical value at the lower limit of statistical significance; the upper limit was estimated using an F critical value having an infinite number of degrees of freedom within strain (12) (J. Belknap, personal communication).

The percent of genetic variance explained by a QTL tends to be overestimated when statistical power (the probability of detecting a QTL) is less than 1 (13). In this study, the estimated power to detect a QTL the size of Tbdr1 at a statistically significant level, corresponding to a logarithm of the odds of linkage (LOD) score of 2.7 and a single-marker regression p value of.00038, was estimated from Figure 1 of Belknap and colleagues (11) to be 0.85. In the same manner, the power to detect a QTL the size of the chromosome 17 locus at a provisional level (LOD 1.2, corresponding to a single-marker regression p value of.022) was estimated to be 0.60. These estimates take into account the effective heritability obtained when testing 8 mice per strain (Figure 1 of ref. 14). The degree of overestimation was then calculated from the formula (1/Power)0.6 (15). This formula suggested that Tbdr1 was overestimated by 10% and the chromosome 17 locus by 36%. Therefore, the percent of genetic variance explained by Tbdr1 was reduced from 54% to 49% and the chromosome 17 locus from 23% to 17%.



View larger version (23K):
[in this window]
[in a new window]
  		Figure 1. Scatter plots of the LSXSS strain means for the response of Tb to dietary restriction (DR). Panel A shows the strain variation for Cohorts 1 and 2 combined. The DR grand mean for Tb was 33.5°C and the standard deviation (SD) after regressing on AL mean Tb was 0.71; the correlation between the DR and ad libitum-fed (AL) strain means was 0.22 (p =.16, one tailed). Panels B, C, and D show the strain variation in Cohort 2 at different times during DR. The filled circles are the strain means for Tb in Cohort 2a tested at 4, 6, and 11 months after starting DR. Open circles are the strain means in Cohort 2b tested at 2 and 7 months after starting DR. In panel B, the DR grand mean for Tb was 33.4°C, and the correlation between the DR and AL strain means was.05 (p =.42, one tailed). In panel C, the DR grand mean for Tb was 33.2°C, and the correlation between the DR and AL strain means was 0.49 (p =.010, one tailed). In panel D, the DR grand mean for Tb was 32.7°C, and the correlation between the DR and AL strain means was 0.40 (p =.11, one tailed)

 
The 95% CI for the location of Tbdr1 was estimated by adapting Equation 4 from Darvasi and Soller (16). Briefly, the 95% CI of a QTL mapped in an F2 panel can be approximated using the formula: 530/(N x VQTL), where N is the number of F2 mice and VQTL is the phenotypic variance explained by the QTL. When h2RI equals 0.35, testing an average of 8 mice per strain and 22 strains is equivalent to testing 83 F2 mice (14). VQTL expected in an F2 panel is one half of VQTL measured in an RI panel (14); therefore, the expected VQTL for Tbdr1 would be 0.245 (i.e., 0.49/2). The CI is then divided by 4 due to the fourfold expansion of an RI recombination map (17), yielding the final equation 530/(83 x 0.245 x 4) = 7 cM.


   RESULTS
Top
 Abstract
 Methods
 Results
Discussion
 References
 
We asked whether a significant genetic component underlies strain variation in the response of Tb to DR in the LSXSS RIs by testing 20 strains in Cohort 1 (6) and then retesting those 20 strains plus 2 additional strains in Cohort 2. All mice were females and singly caged. DR was started at 2 months of age. The DR mice were fed 60% of what the AL controls consumed, as determined separately for each strain and cohort and based on weekly measurements throughout the study.

For both cohorts combined, the strain variation of DR mean Tb was highly significant relative to the variation within strains (p < 10-10) and ranged from 32.0°C to 34.5°C (SD = 0.7; Figure 1A). The AL-fed mice exhibited only a modest amount of strain variation in mean Tb that ranged from 36.1°C to 37.1°C (SD = 0.27; Figure 1A); this variation did not quite reach statistical significance (p =.07). The DR grand mean for Tb of 33.5°C (strains equally weighted) was a phenomenal 3.3°C lower than the AL grand mean for Tb.

To assess the long-term stability of the strain variation, we measured the mean Tb of the Cohort 2 mice at three different times during DR (Figure 1B–D). Cohorts 2a and 2b are shown grouped by age rather than trial because age had a greater effect on mean Tb during DR than the trial differences. The range and standard deviation (SD = 0.95) of the DR strain variation after 2–4 months on DR (Figure 1B) was nearly identical to what we observed in Cohort 1 after 2–4 months on DR [SD = 1.0 (6)]. After 6–7 months on DR (Figure 1C) there was no decrease at all in the amount of strain variation (SD = 0.96), nor was there an appreciable decrease after 11 months on DR (SD = 0. 78, Figure 1D; only Cohort 2a was tested at 11 months). Thus, the marked strain variation in the response of Tb to DR persisted for at least 11 months.

Heritability
Heritability is the relative contribution of genetic versus nongenetic factors underlying variation in a trait. For heritability and QTL analyses, the response of Tb to DR was defined as the DR mean Tb of each strain after regression on the AL strain means for Tb; this adjustment was used because there was a consistent, though small, positive correlation between the DR and the AL strain means in all five temperature trials conducted on Cohorts 1 and 2 [(6); Figure 1B–D]. The calculated heritability (h2RI) was 35% and statistically significant, with a 95% CI of 21%–51%. Similarly, the heritability of the DR mean Tbs without regressing on AL mean Tb was 37% (95% CI of 23%–53%). In contrast, the AL mean Tbs did not exhibit significant heritability (h2RI = 7%).

We found that the heritability estimate of the response of Tb to DR would have been considerably higher than 35% if each cohort had been considered separately (temperature trials combined). Heritability was 47% in Cohort 1 and 65% in Cohort 2. The heritability estimate was reduced when the cohorts were combined because some strains (such as R23 and R17) had relatively high Tbs during DR in Cohort 1 but relatively low Tbs during DR in Cohort 2, whereas other strains exhibited the opposite pattern [(6) and Figure 1B]. The correlation between the DR strain means for Tb measured in Cohort 2 versus Cohort 1 was only 0.36 (p =.06, one tailed). The heritability in Cohort 2 also dropped from 67% after 2–4 months on DR down to 49% after 7–11 months on DR, suggesting an effect of aging or time on DR. However, this effect was negligible in our overall estimate of heritability considering that the heritability in Cohort 2 for the different ages combined was still 65%.

QTL Analysis
With both cohorts combined, genetic mapping on the response of Tb to DR resulted in the identification of a sharply defined QTL on chromosome 9 that exceeded the permutation test threshold of 2.7 LOD for statistical significance [i.e., 0.05 genome-wide protection against false positives as determined empirically based on 10,000 permutations of the data (18)]. The peak LOD score of 3.1 occurred at marker D9Mit4 (Figure 2), which had a single-marker regression p value of.00014. The tightly linked markers D9Mit256 and D9Mit315 gave supportively high LOD scores of 2.1 and 2.0, respectively. This QTL, which will be referred to as Tbdr1 for "body temperature response to dietary restriction, QTL 1," explained 49% of the genetic variation (after adjusting for imperfect statistical power, see Methods). The phenotypic effect size predicted for substituting an S allele (from the SS parental line) for an L allele (from the LS parental line) was -0.7°C (additive model). This QTL also appeared to be robust in that the single-marker regression p value was less than.05 in all temperature trials conducted on both cohorts. For Cohort 1, these p values were.01 and.005 for Trials 1 and 2, respectively. For Cohort 2, the p values were.02,.045, and.01 when Cohorts 2a and 2b were grouped according to time on DR: 2–4 months, 6–7 months, and 7–11 months, respectively (the Cohort 2b data at 7 months is used twice so that the mapping could be done using 22 strains at each time point).



View larger version (31K):
[in this window]
[in a new window]
  		Figure 2. Logarithm of odds (LOD) plot of chromosome 9 showing the location of Tbdr1. The thick, solid line is the LOD plot of the response of Tb to dietary restricted (DR) for both cohorts combined. The dots correspond to the markers shown below the X-axis. The thick, dashed line is the LOD plot of the mean Tbs of the ad libitum-fed (AL) mice obtained for both cohorts combined. The thin, solid line is the LOD plot of the response of Tb to DR based on Cohort 1 alone. The thin, dashed line is the plot based on Cohort 2 alone. The dotted line across the top is the threshold for statistical significance (2.7 LODs) based on 10,000 permutations. The X-axis shows the megabase positions and just below are the approximate cM positions from the Mouse Genome Informatics database (http://www.informatics.jax.org/). The thick bar on the X-axis is the 95% confidence interval for the location of Tbdr1.

 
We also repeated the mapping while controlling for the marker on each chromosome having the highest LOD score for the response of Tb to DR. This had virtually no effect on the mapping of Tbdr1, as even the highest single-marker regression p value at the Tbdr1 locus was only.0007. Therefore, it did not appear likely that Tbdr1 was an artifact due to a correlation by chance with genotypes on a different chromosome (19).

We tested all pairs of marker loci for epistatic interactions and found no pairs that reached a provisional level of statistical significance. We also asked whether an epistatic interaction might explain the mapping of Tbdr1 by repeating the mapping while controlling for the background effects of 26 different marker pairs having the highest LOD scores for an interaction effect, with all chromosomes represented at least once. In each case, the single-marker regression p value at the Tbdr1 locus was still less than.003. Therefore, the mapping of Tbdr1 did not appear to be an artifact due to a chance correlation with an epistatic interaction.

We also identified a locus on chromosome 17 (linked to markers D17Mit49 and D17Mit10 at 23 and 24.5 cMs, respectively) that exceeded the permutation test threshold of 1.2 LOD for being a provisional QTL [i.e., p <.05 but no correction for testing multiple markers (20)]. The peak LOD score was 1.5, and the single-marker regression p value was.0096. The peak LOD was still 1.5 when the mapping was repeated while controlling for the background effect of Tbdr1, indicating that the mapping of Tbdr1 and the chromosome 17 locus arose independently. The chromosome 17 locus explained 17% of the genetic variance, with a phenotypic effect size predicted for substituting an S allele for an L allele of -0.3°C (additive model). Mapping of the chromosome 17 locus appeared to be robust in that the single-marker regression p value was.03 in Cohort 1 and.05 in Cohort 2.

In contrast to the robustness of Tbdr1 and the chromosome 17 locus across cohorts, there were two provisional QTLs mapped using Cohort 1 alone, linked to D6Mit48 and D12Mit44, having single-marker regression p values of.003 and.01, respectively, that were not replicated in Cohort 2 (p =.9 and.15, respectively). There was also a provisional QTL suggested using Cohort 2 alone (linkage to D14Mit37, p =.01) that was not seen in Cohort 1 (p =.8).

Effects of Other Variables
We repeated the QTL mapping while controlling for the background effects of D9Mit105, the marker at the peak LOD of a provisional QTL on chromosome 9 affecting AL mean Tb (Figure 2). The peak LOD of Tbdr1 and the chromosome 17 locus were unaffected (Table 2), indicating that the mapping of these loci was not due to effects associated with D9Mit105.


View this table:
[in this window]
[in a new window]
  		Table 2. Peak LOD Scores Associated With the Tb Response to DR After Controlling for Other Variables.

 
We also asked if the mapping of Tbdr1 and the chromosome 17 locus might be explained by the strain variation in variables such as AL fat, absolute AL food intake, DR body weight during the temperature trials, the efficacy of nutrient extraction during DR (feces:food ratio), and DR home cage activity. We found that none of these traits had provisional or significant QTLs that mapped to chromosome 9 or 17. To explore further, we regressed the response of Tb to DR on each variable and repeated the mapping. With the exception of home-cage activity, controlling for the traits in this manner had no effect on the statistical significance of the Tbdr1 locus (Table 2). The p value after controlling for home-cage activity was still only.001, and thus home-cage activity did not appear to be sufficient to explain Tbdr1 either.

Controlling for these traits also did not cause the p value for the provisional QTL on chromosome 17 to rise above.04 (Table 2). When we controlled for the strain variance in the Tb response, the chromosome 17 locus achieved genome-wide statistical significance. These results indicate that the mapping of Tbdr1 and the chromosome 17 locus was not due to strain variation in any of the tested variables.

We also found that DR had no appreciable effect on calorie extraction. As we found previously from testing a different set of strains (6), there was no significant difference between the calorie content (total energy) of feces from DR mice in Cohort 2 (3710 cal/g, SE [standard error] = 11, N = 22) and the calorie content of feces from the AL controls (3719 calories, SE = 12, N = 22) (p =.28, paired-sample t test, one tailed). Furthermore, there was no correlation between the strain means for the calorie content of the DR feces (no significant correlation with the AL calorie content, R = -.015, p =.94) and the Tb response to DR, as measured during Temperature Trial 3 of Cohort 2 (R =.25, p =.26). Likewise, there was no correlation between the feces calorie content and Tb during DR when comparing the individual DR mice (R = -.05, p =.75, N = 41).

Tbdr1 Candidate Genes
The 95% CI defining the genomic location of Tbdr1 was just 7 cM (see Methods), which corresponded to the 2-LOD support interval between markers D9Mit228 and D9Mit6 (Figure 2). Based on the mouse genome sequence, this region contains ~250 known and predicted genes (http://www.ncbi.nlm.nih.gov/mapview); therefore, this region still needs to be narrowed down considerably before candidate genes can be tested. Nevertheless, given the large number of gene expression microarray studies that have been conducted on DR, we asked whether these studies might suggest intriguing candidate genes. We found that altered expression due to DR has been reported for 9 of the genes within the 95% CI of Tbdr1. These genes encoded 2 heat shock proteins, 2 apolipoprotein A proteins, 1 histone protein, 1 zinc finger protein, 1 dopamine receptor, 1 cytochrome P450, and 1 src tyrosine kinase (Table 3).


View this table:
[in this window]
[in a new window]
  		Table 3. Genes With Altered Expression During DR That Are Within the 95% Confidence Interval of Tbdr1.

 

   DISCUSSION
Top
 Abstract
 Methods
 Results
 Discussion
References
 
This study is the first to demonstrate that variation in a physiological response to DR is significantly heritable and amenable to QTL mapping. For both cohorts combined, the heritability of mean Tb during DR was 35% (95% CI of 21%–51%). This result also implies that 65% of the strain variation was nongenetic. Part of this nongenetic component appears to be due to significant nonreplicable, gene-by-environment effects of unknown origin such that the heritability estimates of each cohort alone (47% and 65%) were considerably higher than the heritability for the cohorts combined. The higher heritability in Cohort 2 compared with Cohort 1 might be due to improvements in methodology, such as the in vivo calibration of the temperature transponders and reduced week-to-week variation in AL food intake.

We identified two likely QTLs: a statistically significant QTL, Tbdr1, in a 7 cM region flanking D9Mit4 (95% CI) on chromosome 9 and a provisional QTL on chromosome 17, near D17Mit49 and D17Mit10. Although we obtained a statistically significant result at the chromosome 17 locus when controlling for the strain variance in the Tb response, multiple-hypothesis and post hoc considerations suggest that this QTL needs further confirmation.

Consistent with our previous findings regarding the strain variation in the Tb response (6), neither Tbdr1 nor the chromosome 17 locus appeared to be the result of the strain variation in AL fat, absolute AL food intake, DR body weight during the temperature trials, the efficacy of nutrient extraction during DR, or DR motor activity. Because the DR rations were by design directly proportional to the AL food intake of each strain, finding no effect of absolute AL food intake also indicates that there was no effect of the absolute differences in DR food, calorie, protein, vitamin, and mineral intakes (6).

Identifying the genes underlying Tbdr1 and the chromosome 17 locus will require further refinement of the candidate region, which can be accomplished by developing congenic mice (17,30). In addition, gene expression microarray studies could provide an important means of quickly identifying candidate genes. In this regard, the Cab140 and Dnaja4 genes that we identified within the 95% CI for the location of Tbdr1 are especially intriguing. Cab140 is a temperature-sensitive gene encoding the GRP170 heat shock protein, whose levels are lowered by approximately 40% in the hepatocytes of mice fed 55% AL and appears to be posttranslationally modified (22). Dnaja4 is a temperature-sensitive gene that encodes for an Hsp40 heat shock protein. In response to DR (76% AL), the mRNA expression of Dnaja4 was lowered almost threefold in skeletal muscle, which was the largest decrease among all of the genes surveyed by Lee and colleagues (25). The lowering of Cab140 and Dnaja4 expression occurs in concert with a marked lowering of other molecular chaperones/heat shock proteins (24), and it has been proposed that this lowering could produce longevity effects by increasing protein turnover and apoptosis (31). Although it would be premature to focus exclusively on these genes, the analysis illustrates how genetic mapping and microarray studies can complement each other to help identify interesting candidate genes very quickly.

The Tbdr1 locus and the chromosome 17 locus together accounted for about two thirds of the genetic variance in the response of Tb to DR. We are currently testing whether these loci influence other responses to DR, such as body weight, growth, and fertility responses. It will also be feasible in the future, through the development of congenic mice, to critically test whether these loci affect life extension in response to DR. Such QTLs also represent a promising step towards using positional cloning to bridge the long-standing gap between the physiology of DR responses and their underlying molecular mechanism.


   Acknowledgments
 
We are grateful to John Belknap for clarifying the methods to estimate the statistical power and heritability of the RIs, and to Beth Bennett, Chris Downing, Sam Henderson, Chris Link, and Nengjun Yi for comments on the manuscript. Excellent technical assistance was provided by Lena Gordon (genotyping), Jaclyn Francese (animal care), Ted Pokrywka (animal care), Christine Martin (animal care and data collection), and Julia Rifkin (animal care and data entry). Primary funding was provided by the Ellison Medical Foundation through a Senior Scholar Award to T. E. Johnson. Additional support was obtained from the National Institutes of Health via RO1 AA11984, KO1 AA00195, P30DK05633, and the UAB Clinical Nutrition Research Center (P30DK56336).

Address correspondence to Brad A. Rikke, PhD, Institute for Behavioral Genetics, Campus Box 447, University of Colorado, Boulder, CO 80309-0447. E-mail: rikke{at}colorado.edu


   Footnotes
 
Decision Editor: James R. Smith, PhD

Received September 2, 2003

Accepted November 13, 2003


   References
Top
 Abstract
 Methods
 Results
 Discussion
 References
 

  1. McCay CM, Crowell MF, Maynard LA. The effect of retarded growth upon the length of the life span and upon the ultimate body size. J Nutr.. 1935;10:63-79.
  2. Weindruch R, Walford RL. The Retardation of Aging and Disease by Dietary Restriction. Springfield, Illinois: Charles C. Thomas; 1988:8–29.
  3. Rikke BA, Johnson TE. Towards the cloning of genes underlying murine QTLs. Mamm Genome.. 1998;9:963-968.[Medline]
  4. Glazier AM, Nadeau JH, Aitman TJ. Finding genes that underlie complex traits. Science.. 2002;298:2345-2349.[Abstract/Free Full Text]
  5. Walford RL, Spindler SR. The response to calorie restriction in mammals shows features also common to hibernation: a cross-adaptation hypothesis. J Gerontol Biol Sci.. 1997;52A:B179-B183.
  6. Rikke BA, Yerg III JE, Battaglia ME, Nagy TR, Allison DB, Johnson TE. Strain variation in the response of body temperature to dietary restriction. Mech Ageing Dev.. 2003;124:663-678.[Medline]
  7. Manly KF, Cudmore Jr RH, Meer JM. Map Manager QTX, cross-platform software for genetic mapping. Mamm Genome.. 2001;12:930-932.[Medline]
  8. DeFries JC, Wilson JR, Erwin VG, Petersen DR. LSXSS recombinant inbred strains of mice: initial characterization. Alcohol Clin Exp Res.. 1989;13:196-200.[Medline]
  9. Rikke BA, Johnson DK, Johnson TE. Murine albino-deletion complex: high-resolution microsatellite map and genetically anchored YAC framework map. Genetics.. 1997;147:787-799.[Abstract]
  10. Henderson ND, Tritto T, Fulker DW, DeFries JC. Home cage activity in mice selected for high- and low open-field ambulation. Behav Genet.. 1998;28:471.
  11. Belknap JK, Mitchell SR, O'Toole LA, Helms ML, Crabbe JC. Type I and type II error rates for quantitative trait loci (QTL) mapping studies using recombinant inbred mouse strains. Behav Genet.. 1996;26:149-160.[Medline]
  12. Sokal RR, Rohlf FJ. Biometry: The Principles and Practice of Statistics in Biological Research. New York: WH Freeman and Company; 1995:215–216.
  13. Beavis WD. QTL analyses: power, precision, and accuracy. In: Paterson AH, ed. Molecular Dissection of Complex Traits. Boca Raton, Florida: CRC Press; 1998:145–162.
  14. Belknap JK. Effect of within-strain sample size on QTL detection and mapping using recombinant inbred mouse strains. Behav Genet.. 1998;28:29-38.[Medline]
  15. Bergeson SK, Warren RK, Crabbe JC, Metten P. Erwin VG, Belknap JK. Chromosomal loci influencing chronic alcohol withdrawal severity. Mamm Genome.. 2003;14:454-463.[Medline]
  16. Darvasi A, Soller M. A simple method to calculate resolving power and confidence interval of QTL map location. Behav Genet.. 1997;27:125-132.[Medline]
  17. Silver LM. Mouse Genetics: Concepts and Applications. New York: Oxford University Press; 1995:44–49,218–220.
  18. Churchill GA, Doerge RW. Empirical threshold values for quantitative trait mapping. Genetics.. 1994;138:963-971.[Abstract]
  19. Williams RW, Gu J, Qi S, Lu L. The genetic structure of recombinant inbred mice: high-resolution consensus maps for complex trait analysis. Genome Biol.. 2001;2:RESEARCH0046.[Medline]
  20. Lander E, Kruglyak L. Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet.. 1995;11:241-247.[Medline]
  21. Lee CK, Weindruch R, Prolla TA. Gene-expression profile of the ageing brain in mice. Nat Genet.. 2000;25:294-297.[Medline]
  22. Dhahbi JM, Cao SX, Mote PL, Rowley BC, Wingo JE, Spindler SR. Postprandial induction of chaperone gene expression is rapid in mice. J Nutr.. 2002;132:31-37.[Abstract/Free Full Text]
  23. Miller RA, Chang Y, Galecki AT, Al-Regaiey K, Kopchick JJ, Bartke A. Gene expression patterns in calorically restricted mice: partial overlap with long-lived mutant mice. Mol Endocrinol.. 2002;16:2657-2666.[Abstract/Free Full Text]
  24. Cao SX, Dhahbi JM, Mote PL, Spindler SR. Genomic profiling of short- and long-term caloric restriction effects in the liver of aging mice. Proc Natl Acad Sci U S A.. 2001;98:10630-10635.[Abstract/Free Full Text]
  25. Lee CK, Klopp RG, Weindruch R, Prolla TA. Gene expression profile of aging and its retardation by caloric restriction. Science.. 1999;285:1390-1393.[Abstract/Free Full Text]
  26. Han E, Hilsenbeck SG, Richardson A. Nelson JF. cDNA expression arrays reveal incomplete reversal of age-related changes in gene expression by calorie restriction. Mech Ageing Dev.. 2000;115:157-174.[Medline]
  27. Lee CK, Allison DB, Brand J, Weindruch R, Prolla TA. Transcriptional profiles associated with aging and middle age-onset caloric restriction in mouse hearts. Proc Natl Acad Sci U S A.. 2002;99:14988-14993.[Abstract/Free Full Text]
  28. Wada Y, Kayo T, Koizumi A. Characterization of gene expression profile associated with energy restriction-induced cold tolerance of heart. Microsc Res Tech.. 2002;59:313-316.[Medline]
  29. Jervis KM, Robaire B. Effects of caloric restriction on gene expression along the epididymis of the Brown Norway rat during aging. Exp Gerontol.. 2003;38:549-560.[Medline]
  30. Bennett B, Beeson M, Gordon L, Carosone-Link P, Johnson TE. Genetic dissection of quantitative trait loci specifying sedative/hypnotic sensitivity to ethanol: mapping with interval-specific congenic recombinant lines. Alcohol Clin Exp Res.. 2002;26:1615-1624.[Medline]
  31. Dhahbi JM, Cao SX, Tillman JB, et al. Chaperone-mediated regulation of hepatic protein secretion by caloric restriction. Biochem Biophys Res Commun.. 2001;284:335-359.[Medline]




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Services
Right arrow Download to citation manager
PubMed
Right arrow PubMed Citation

HOME ARCHIVE SEARCH TABLE OF CONTENTS