1 Department of Pathology, University of Michigan School of Medicine, Ann Arbor
2 Geriatrics Center, University of Michigan School of Medicine, Ann Arbor
3 University of Michigan Institute of Gerontology
4 Ann Arbor Department of Veterans Affairs Medical Center
5 Department of Human Genetics, University of Michigan School of Medicine, Ann Arbor, Michigan 48109-0940
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
epistasis; UM-HET3 mice; late-life hormone levels; age-specific
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Most of the insulin-like growth factor I (IGF-I) in the circulation is produced by the liver in response to the pituitary peptide growth hormone (GH) and plays a prominent role in the determination of body size (9, 44), as well as being an important factor in the regulation of reproductive pathways (26, 39, 48, 54), energy balance (11, 22, 51, 52), and cell proliferation and death (6, 24, 25). In addition, IGF-I appears to play a role in the pathogenesis of common proliferative diseases (30).
Leptin is an adipocyte-derived protein hormone originally thought to function solely as a satiety signal (53), although it is now apparent that leptin can act both centrally and in the periphery to regulate a host of physiological processes. This includes growth, fertility, and glycemic control, as evidenced by leptin deficient ob/ob mice (36), as well as regulation of the hypothalamic-pituitary-adrenal (HPA) axis (19), bone growth (17), and T-cell immunity and the inflammatory response (16).
Thyroxine (T4) and triiodothyronine (T3) are synthesized in the thyroid gland using the amino acid tyrosine and the trace mineral iodine as building blocks and are key regulators of the basal metabolic rate (BMR) and heat production in mammals (13). Increased levels of either hormone will result in a corresponding increase in BMR and heat production via their effect on mitochondrial respiratory pathways (10, 13, 33). Moreover, the thyroid hormones are also key mediators of vertebrate growth and development (41, 44) and have been linked to oxidative defense mechanisms (12, 20, 21).
The glucocorticoids (GC) are a class of steroid hormones released during times of physiological and psychological stress, with corticosterone (CORT) being the primary bioactive GC in most species of rodent. GCs mobilize energy stores for immediate use (40) and by providing copious short-term energy may be invaluable for surviving sudden emergencies (46). Indeed, mild elevations in GC levels are beneficial for health (5). Nevertheless, chronic activation of this pathway does occur at the expense of other physiological processes and has been linked to significant impairments in growth (44), reproduction (35, 38, 50), and immune function (15).
In this study, 1,108 mice derived from a cross between CB6F1 females and C3D2F1 males were genotyped and sampled at the ages of 4, 15, and 17 mo to obtain blood or fecal material for the determination of IGF-I, leptin, T4, and CORT levels. This population is the genetic equivalent of a large sibship and thus allows us to identify QTL that have detectably large effects on hormone levels. By measuring T4, IGF-I, and leptin at both 4 and 15 mo, we were also able to determine whether the genetic effects noted might vary with age.
![]() |
MATERIAL AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
All mice were weaned at 3 to 4 wk of age, housed in same-sex cages, and given free access to food and water for the duration of the study. To ensure the specific pathogen-free status of the study population, groups of sentinel mice were exposed to spent bedding from the study population on a quarterly basis and were later evaluated serologically for the presence of specific viral and bacterial pathogens. The animals were also examined for pinworm. All test results were negative over the course of the study.
Mice were entered into the study in cohorts of 30 per month with all mice housed in the Cancer Center and Geriatrics Center Building at the University of Michigan Medical Center. For the determination of serum hormone levels, all mice were immunized with erythrocytes at the ages of 4 and 15 mo as part of another protocol and bled by tail venipuncture 2 wk after each immunization to assess antibody production as an index of their immunoresponsiveness. The remaining serum samples were used for the quantification of circulating T4, leptin, and IGF-I levels in this study. In addition, at 12 mo of age skin biopsy samples were obtained under brief Metofane anesthesia as part of another protocol, and 13 fecal pellets were collected from approximately one-third (n = 321 females) of the mice at 17 mo of age for the quantification of fecal corticosteroid levels for use in this study. Fecal samples were collected from each mouse for five consecutive days to control for the possibility of significant day-to-day variation in circulating CORT levels due to the influence of the estrus cycle (4). The mean 5-day fecal corticosteroid level for each mouse was used for the analysis.
Hormone measurements.
Serum samples for hormone measurements were taken by tail venipuncture between the hours of 7 and 11 AM and stored at -70°C for up to 3 yr prior to assay. Serum T4 levels were determined using a monoclonal solid-phase radioimmunoassay (RIA) kit (ICN Pharmaceuticals, Costa Mesa, CA) run at one-quarter volume according to the manufacturers instructions. Each sample was assayed in duplicate and diluted up to 1:7 with phosphate-buffered saline (PBS) if necessary to achieve adequate sample volume. Serum IGF-I levels were quantified via a double-antibody RIA kit (Diagnostic Systems Laboratories, Webster, TX) run at one-quarter volume according to the manufacturers instructions. Prior to assay, 10 µl of serum from each individual was subjected to an acid-ethanol extraction procedure using the materials provided in the kit. All samples were assayed in duplicate. Serum leptin levels were quantified with a double-antibody RIA kit (Linco Research, St. Charles, MO) according to the manufacturers instructions, except that all volumes were reduced by a factor of 4. Each sample was assayed in duplicate using dilutions up to 1:10.4 with PBS if necessary to achieve adequate sample volume. The inclusion of two pooled serum controls run in each assay (n = 13) indicated that the mean (±SD) intra-assay coefficients of variation (CV) were 6.95 ± 5.55%, 5.89 ± 4.75%, and 6.12 ± 4.53% for T4, IGF-I, and leptin, respectively, and that the inter-assay coefficient CV was less than 25% for all. The lower limits of detection were 150 ng/ml, 0.2 ng/ml, and 0.625 µg/dl for IGF-I, leptin and T4, respectively.
Fecal corticosteroid levels were quantified with a double-antibody 125I-corticosterone RIA kit (ICN Biomedicals) after hormone extraction using an ethanol-based procedure as described in (23). This procedure has been validated previously for use in multiple species of rodent as a specific, precise, and accurate integrated measure of circulating corticosteroid concentrations (23). All samples were assayed in duplicate, and the assay was performed according to the manufacturers instructions at one-quarter volume. The 25 ng/ml standard provided with the kit was diluted 1:2 with steroid diluent (provided with the kit) to allow the inclusion of a 12.5 ng/ml standard (the lower limit of detection for this assay). Corticosteroid concentrations are presented as nanograms corticosteroids per gram dry weight of feces. As an indicator of the day-to-day variation in fecal CORT levels over the 5-day sampling period, the intra-individual CV of the 5-day mean fecal CORT measure was calculated for each mouse. The result of this analysis indicated that fecal CORT levels typically varied by 36% within a five-day sampling period (mean ± SD; 35.95 ± 15.62%). In addition, the inclusion of two pooled fecal extracts in each assay (n = 20) indicated that the mean (±SD) intra-assay CVs were 3.73 ± 3.62% and 8.45 ± 7.31% for the low and high control, respectively. The interassay CV was 8.7% for the low control and 18.8% for the high control.
Genotyping.
Genomic DNA was isolated from a 1-cm section of tail collected from 4-wk-old individuals using a phenol-extraction method. The DNA concentration, ability to sustain PCR amplification, and electrophoretic size distribution for each of the preparations were tested prior to simple-sequence length polymorphism (SSLP) genotyping using an ALFexpress automated sequence analyzer (Pharmacia, Piscataway, NJ) using previously described methods (28). Primer pairs were either purchased from Research Genetics or were synthesized locally (University of Michigan Molecular Biology Core Facility). Polymorphic loci were selected using data provided by the Mouse SSLP Database (Whitehead/MIT Center for Genome Research, Cambridge, MA) or the Mouse Genome Database (Jackson Laboratory, Bar Harbor, ME). Chromosomal localization and marker order were calculated using the MapMaker QTX software package (Whitehead Institute, Cambridge, MA).
Statistical analyses.
A single point genome-wide search was performed for each of the seven hormone measures to detect QTL that may be associated with each measure. To make the analysis consistent for all partially and fully informative markers, four-way informative markers were split into two sets of bi-allelic markers that were informative for either the maternally or paternally transmitted alleles. One-way ANOVA models, with each hormone measure as the dependent variable and a bi-allelic marker as the independent variable, were used to perform a genome-wide search for all of the 164 bi-allelic markers. The statistical significance for each marker-trait combination was calculated empirically using a permutation-based technique. This allows the generation of an "experiment-wise" acceptance criterion that takes multiple comparisons into account to avoid type I error inflation. A null distribution for permutation analysis was generated based on 1,000 random shuffles of the original phenotypic data. In addition, genetic marker- and age-related interactions, as well as the amount of variability of each hormone explained by genetic effects, were studied using ANOVA.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
Using a permutation method to derive experiment-wise significance criteria, we found 12 significant gene/trait associations, as shown in Table 2. IGF-I levels were affected by loci on chromosomes 1, 3, 8, 10, and 17; leptin levels were affected by a locus on chromosome 3; T4 levels were modulated by loci on chromosomes 4, 15, and 17; and fecal CORT levels were influenced by a locus on chromosome 1. Each marker was only associated with a single trait measured at a specific age with one exception: D4Mit155 was significantly associated with serum T4 levels in both 4- and 15-mo-old individuals. The marker D17Mit46 is also listed twice in Table 2, but it is the maternal allele that is associated with alterations in IGF-I levels (at 15 mo), and the paternal allele associated with differences in T4 levels in 4-mo-old mice. Four significant markers, i.e., D3Mit227, D3Mit86, D15Mit63, and D1Mit206, are not shown in Table 2, because they are each within 16 cM of another marker with a stronger level of association with the trait concerned and thus are unlikely to represent an independently segregating allele with functional effect.
|
In addition to the statistically significant associations shown in Table 2, we have compiled a Supplementary Table listing each of the markers for which the genome scan provided suggestive evidence at a less conservative criterion of P < 0.20. This Supplementary Table can be viewed at http://www-personal.umich.edu/millerr/Harper%20suppl%20table%201.htm, and is also available at the Physiological Genomics web site.1
It is possible polymorphic alleles might modify more than one of the tested hormones but that only one of the associations is sufficiently strong to reach the experiment-wise significance threshold. To examine this question, further analysis was performed to determine whether any of the QTL listed in Table 2 had an influence on the level of the other hormones measured in this study. We used a Bonferroni-corrected P value, P = 0.004, as our criterion for rejecting the null hypothesis that the alleles listed in Table 2 had no effect on any of the other hormones. The results of this post hoc analysis are presented in Table 3, which also includes the original associations from Table 2 for completeness. Overall, these results suggest that some of the chromosomal segments containing significant QTL show associations with the production or metabolism of multiple hormones or of the same hormone at more than one age. It is of particular interest to note that the B6 allele on chromosome 8 is associated with high IGF-I and low T4 levels, whereas the B6 allele on chromosome 15 is associated with low levels of both of these hormones, suggesting that the two alleles may affect different aspects of the hormonal response. Further work would be needed to determine whether the associations noted in Table 4 reflect the actions of single QTL or of multiple QTL linked to one another on the indicated chromosomes, and to investigate the extent to which changes in one hormone might bring out secondary, compensatory alterations in the levels of other hormones in our test battery.
|
|
In the second approach, a full-factorial ANOVA was performed that used all of the significant markers as factors in the model, and the P value of the interaction term, p(I), that involved all of the significant markers was determined. A nonsignificant interaction term is interpreted as no genetic interaction, or epistatic effect, for the markers tested. The results of these analyses are also presented in Table 4 and indicate that there was no significant evidence for epistasis for any of the hormones tested, with the possible exception of serum IGF-I levels in 15-mo-old mice. In this case, the second-order interaction between D8Mit51 and D3Mit25 was significant at P = 0.05. However, it is possible this finding is due simply to chance. Neither the third-order interaction, nor the two remaining second-order interactions, are significant at P = 0.10, and the sum of the estimated and pooled genetic effects are virtually identical (173 vs. 187 ng/ml). Figure 1 illustrates this additivity for serum IGF-I levels in 15-mo-old mice. For each of the three significant markers, the median concentration of serum IGF-I is shown with the effect of the B6 or C allele at each of the three loci. At D3Mit25 and D8Mit51 inheritance of the B6 allele is associated with an increase in serum IGF-I, whereas at D17Mit46 inheritance of the C allele has the same effect. For mice that inherit the B6 allele at D3Mit25 and D8Mit51 and the C allele at D17Mit46, their serum IGF-I concentration is 187 ng/ml greater than those who inherited the opposite alleles at these loci (Fig. 1).
|
Age-related effects on hormone measures.
Circulating levels of three of the hormones were measured at two ages (4 and 15 mo) in most of the mice. This allowed us to test whether the effect of a given allele on circulating IGF-I, leptin, or T4 levels varies with age. For a more systematic exploration of this issue, we performed a repeated measures ANOVA for each of the 10 markers with an experiment-wise P < 0.05, as well as for the two leptin QTL for which experiment-wise P < 0.10. For most of the markers the [age x genotype] interaction term failed to reach statistical significance, providing no evidence for an age-specific effect. There were five instances, however, in which the strength of the genetic effect was indeed age sensitive. In four of five of these cases, the effect of the genetic polymorphism is more dramatic at the older age than it is in young mice. In particular, for IGF-I the markers D3Mit25 and D17Mit46, and for leptin the markers D11Mit289 and D13Mit64, all had a significant effect on circulating hormone levels in 15- vs. 4-mo-old mice. On the other hand, the marker D10Mit230 had a greater effect on circulating IGF-I levels in 4- vs. 15-mo-old mice. Figure 2 shows two examples of situations in which the effect of an allele is equally apparent at both ages tested (IGF-I, D8Mit51; leptin, D3Mit127) and two others for which the allele effect is apparent only at the older age (IGF-I, D3Mit25; leptin, D11Mit289). In this example, for IGF-I p(I) = 0.40 for the marker D8Mit51, whereas p(I) = 0.01 for the marker D3Mit25. This suggests that the effect of D3Mit25 on circulating IGF-I levels changes with age but that D8Mit51 does not; a suggestion that is clearly borne out by the illustration of these effects in Fig. 2.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Previous studies have also identified QTL for serum IGF-I (8, 45) and leptin (8) in mice. There was, however, little overlap between the QTL identified in the previous studies and the QTL identified here. For example, the chromosome 14 QTL for serum leptin identified by Brockman and colleagues (8) was documented using a cross between DBA.2 mice and mice of a stock, D6Ui, that had previously been selected for extreme growth rate. The significant and suggestive QTL for serum leptin identified in this study reside on chromosomes 2, 3, 11, and 13, but it is difficult to compare the these results to Brockmans findings because of the differences in the genetic backgrounds of the stocks examined.
Brockman and colleagues (8) also reported that there were two QTL for serum IGF-I levels in their stock of mice: one on chromosome 10 and the other on chromosome 18. In contrast, Rosen and colleagues (45) reported a significant association between serum IGF-I levels and QTL on chromosomes 6, 10, and 15 in a population of 4-mo-old female (C57BL/6 x C3H/He)F2 mice. Interestingly, the QTL identified on chromosome 10 in each of these studies map to the same region as the one identified by us and in each case are in close proximity to the mouse Igf1 gene. A recent study (47) has shown that QTL regulating levels of gene expression are often linked to the structural genes in question and are often clustered in proximity to other QTL regulating similar physiological systems. It thus seems plausible that this area of mouse chromosome 10 marked by D10Mit230 in our study could contain multiple genes with effects on IGF-I levels, perhaps including polymorphic regions with effects on transcription of Igf1 per se. Similarly, the IGF-I QTL identified on chromosome 15 by Rosen et al. (45) maps to approximately the same region as a suggestive QTL on chromosome 15 identified in this study.
To the best of our knowledge this study is the first to identify any QTL for either T4 or CORT in mice, although two QTL for serum cortisol levels have been reported for humans (42). Interestingly, both of these human QTL are on human chromosome 1q and map to a region that is orthologous to that containing the locus D1Mit105 in mice, linked in our study to mouse CORT levels (http://www.informatics.jax.org/reports/homologymap/mouse_human.shtml).
For several of these loci, likely candidate genes can readily be identified. In addition to the linkage of D10Mit230 to the mouse Igf1 locus mentioned earlier, we note that the IGF-I QTL linked to D17Mit46 is in close proximity to the IGF-II receptor (Igf2r) gene. Furthermore, the markers D4Mit155 and D15Mit100 are included in regions on each of the two chromosomes that contain genes integral to the metabolic control of thyroid hormones: type I iodothyronine deiodinase (Dio1) on chromosome 4; and the thyroglobulin releasing hormone receptor (Trhr) and thyroglobulin (Tgn) on chromosome 15. Each of the chromosomal regions identified by this low-resolution study, however, may well contain additional genes with direct or indirect effects on hormone levels, and additional work will be needed to determine the specific locus (or loci) responsible for the effects documented in Table 2.
Each of the 12 loci with an experiment-wise P 0.10 was tested formally to see whether we could exclude the hypothesis of age independence. In five cases we were able to reject the null hypothesis, and found that in four of these five cases the genetic effect was stronger in older than in young animals. For example, the loci at D3Mit25 and D17Mit46 had only negligible effects on the serum level of IGF-I in 4-mo-old mice but a very large effect in 15-mo individuals. However, the opposite relationship was observed for the locus D10Mit230, which had an appreciable effect on serum IGF-I levels in 4-mo-old mice but little effect in the 15-mo-old individuals. Thus five of the QTL seem to have an age-sensitive effect, and four of these only manifest their effect later in life. It will be interesting to learn whether these affect different metabolic pathways than do loci whose effect on hormone levels is established in young adults and unvarying thereafter.
In a previous study, mice generated using an identical breeding scheme indicated that there were multiple QTL for the regulation of T-cell subset distributions in this stock and that the effects of these loci could either be stable or age specific (29), similar to what we observed in this study. Interestingly, the QTL at D13Mit57 identified by Jackson and colleagues (29) for the regulation of CD4 levels in 18-mo-old mice and the QTL at D13Mit64 identified in this study for the regulation of serum leptin in 15-mo-old mice are only 20 cM apart on chromosome 13 and only exhibit their effects in middle-aged mice (15 and 18-mo-old). This suggests that the regions of chromosome 13 containing and/or flanked by these two loci may harbor genes that are capable of modulating the pace of age-related changes across multiple physiological systems. This segment of chromosome 13 bears a host of genes known to affect proliferative, regenerative, and cell death pathways. It will be of considerable interest to follow-up on these findings to identify other potential loci with significant age-related effects on multiple systems.
![]() |
DISCLOSURES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
![]() |
ACKNOWLEDGMENTS |
---|
![]() |
FOOTNOTES |
---|
Address for reprint requests and other correspondence: R. A. Miller, 1500 East Medical Center Drive, 5316 CCGCB Box 0940, Ann Arbor, MI 48109-0940 (E-mail: millerr{at}umich.edu).
10.1152/physiolgenomics.00063.2003.
1 The Supplementary Material for this article (a table listing each of the markers for which the genome scan provided suggestive evidence at a less conservative criterion of P < 0.20) is available online at http://physiolgenomics.physiology.org/cgi/content/full/00063.2003/DC1.
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|