Speed congenic strains provide a powerful approach to confirm and physically confine QTL within intervals defined by molecular markers. In the current study, approximately 20% of the CAST genome harboring all previously detected growth and carcass composition QTL was isolated on an HG or B6 background through the development of 14 speed congenic strains. Two distinct speed congenic panels were developed, the first provided a comprehensive isolation of all MMU2 QTL between the B6, HG and CAST strains and the second targeted all QTL outside of MMU2. Each successfully characterized strain exhibited phenotypic differences relative to control mice. These strains represent important resources and provide the genetic resource to positionally clone numerous quantitative trait genes.
One criticism of the speed congenic approach is the potential for QTL to reside among unwanted donor alleles not eliminated during backcrossing. In this case differences between the congenic and control strains would be due in part or whole to these contaminating alleles. We took several precautions to reduce the probability of this occurrence. First, our control strains were developed from mice undergoing the same selection as all of the other congenics. Therefore, it is possible that any unwanted QTL or mutations arising during congenic construction are shared between all strains. More importantly, we have knowledge of all large previously detected growth and obesity QTL in the current cross [24
]. Using this information we increased the density of markers in each QTL region (MMU1, 2, 5, 8, 9, 11 and 17; Table and Additional File 1
), ensuring the absence of CAST alleles at each of these QTL. This approach, termed "QTL-Marker-Assisted Counter Selection" or QMACS, has been previously used to characterize QTL for hypnotic sensitivity to ethanol [31
]. In that study, only markers flanking QTL were typed, not genome-wide markers. In contrast, we selected not only against known QTL, we also screened for genome wide heterozygosity increasing the probability that effects observed are due to genetic variation within each donor region.
Although great effort was put forth to eliminate non-donor region direct genetic effects, other factors such as maternal genotype (maternal genotype for each congenic versus control dams differed) and environmental effects could confound our results. Maternal genotype effects on growth and obesity have been observed in a number of mouse crosses [8
] and their existence in the current cross cannot be discounted. However, our congenics provide the ideal foundation genomic resource to test for the influence of any of these possible effects. Future fine mapping experiments can be designed to randomize the influences of any contaminating donor alleles and environmental differences, as well as test for maternal genotype effects.
MMU2 is a hotspot for growth and obesity QTL. Over 30 QTL have been identified in various experiments [1
]. Several previously reported or novel MMU2 QTL have been isolated and characterized using congenic strains [7
]. Our findings are no different and indicate that MMU2 is highly complex with regards to genes affecting growth and obesity. The overlapping nature of our MMU2 strains allowed us to parse the chromosome into five regions (Regions I–V) (Figure ). The data support the presence of at least one QTL in each of the five regions (Figure ). Each of the five pleiotropically impact both growth and obesity, although to varying degrees.
In addition to the large number of MMU2 QTL, the presence of hg adds complexity by either eliciting interactions with the same QTL or by inducing the expression of novel QTL. The 2P unique region (Region I) contains an hg modifier with large effects on growth and smaller effects on obesity (Table ). In contrast, interactions between 2PM/2M QTL and hg primarily affect fat deposition (Table ). As illustrated in Figure the 2M donor region exhibits strong sex effects on the rate of lipid storage, dependent on background. In control mice, HGC females have a higher AI than males, while the opposite is seen in B6C mice. Interestingly, the 2M hg modifier QTL abolished the hg induced sexual dimorphism in adiposity (Figure ). Although these results provide insight into the nature of hg modifier QTL, it should be noted, that the actual number and precise location of loci is still unclear.
In addition to MMU2, three other congenics (HG9, HG11 and HG17) captured hg
modifier QTL. In classical terms, they are QTL which modify the expressivity of growth and obesity in HG mice [18
]. These QTL are novel since they represent epistasis between hg
and the modifier gene. QTL-hg
epistasis implies that Socs2
and the hg
modifiers are in the same biological pathway. Therefore, these QTL are likely due to polymorphism in genes interacting with Gh
, responsive to Gh
or which in some way modulate Gh
function. This information will significantly aid QTL cloning by providing another filter to screen candidates. Cloning these QTL has major implications to improve our understanding of Gh
and its regulation of growth and adiposity and in the administration of human Gh
HG1 mice displayed differences only in growth traits. Originally, Q1Ucd1 had small effects and these results illustrate the power of congenic strain analysis to isolate small effect QTL, although it is likely this QTL represents the lower boundary of detection using only 20–30 congenic mice. The most notable candidate genes located within the HG1 interval are the signal transducer and activator of transcription 1 (Stat1) and Stat4.
Two of the congenics had major alterations in the deposition of adipose tissue, HG8 and HG9. The HG8 donor region promotes leanness. Congenic females displayed a reduction of over 25% in AI. In contrast, the HG9 strain is an obese mouse model. The strain is quite novel for two reasons. First, the effect size is large; AI was 57% higher in HG9 females and 30% higher in males. Secondly, it is dependent on hg for its expression. The HG9 strain represents a major epistasis-based obese mouse model and promises to aid in the understanding of obesity and specifically the modulation of adipose tissue deposition by Gh. Studies are currently underway to identify the causative mutation and to characterize the effects of age and diet on obesity in this strain as well as testing for other physiological consequences such as alterations in food intake and insulin sensitivity.
The HG11 strain is of particular interest for a number of reasons. First, HG11 congenic mice demonstrated significant strain by sex interactions for a number of traits. Males were generally larger, faster growing and longer and the converse was seen in females. The confounding effects of sex are likely the reason for the discrepancies between the congenic and genome scan results, where both sexes were analyzed together. Secondly, Carp2
was found to interact with hg
and MMU11 is saturated with genes involved in the central Gh
intracellular signaling pathway, such as Gh
. Thirdly, MMU11 growth QTL overlapping HG11 have been identified in a number of crosses using different mouse strains [37
]. Given the well documented sexually dimorphic nature of Gh
secretion and Gh
induced gene expression [42
], it is probable that the underlying mutation in the HG11 congenic may reside in a gene enhancing Gh
induced sex-specific effects. The structural Gh
gene itself and Stat5b
are excellent candidates. The potential role of Gh
would include polymorphism that alters protein function in the absence of Socs2
or that causes transcriptional deregulation of Gh
. Additionally, functional variation in Stat5b
may explain the sex-specific phenotypes in HG11 mice since it is the primary transcription factor responsible for Gh
induced sex-specific liver gene expression [44
]. Future studies aimed at identifying the HG11 QTG, will certainly include a thorough characterization of Gh
sequence and expression patterns in congenic mice.
modifier QTL located within the HG17 strain had large effects on growth, body length and carcass components. The most intriguing candidate gene located in the congenic is Sstr5
. Somatostatin is a potent inhibitor of Gh
is one of five receptors which mediate these effects [45
]. Ubiquitous and pancreatic beta cell specific knockout of Sstr5
leads to alterations in insulin secretion and glucose homeostasis [47
We sequenced candidate hg modifier genes to complement the characterization of the speed congenics on MMU2, 9, 11 and 17. A limited number of studies have identified variation within CAST coding sequence; so sequencing candidates gave us the opportunity to estimate the SNP frequency in coding sequence relative to B6. This information will be vital to future fine mapping studies, which will include gene expression analysis. The high SNP frequency (0.312%; 295 SNP in 94.492 kbp) in CAST genes may lead to a higher rate of false positives using DNA microarrays or quantitative real-time PCR assays, most of which are based on B6 sequence. Therefore, these sequence data can be used to guide the development of gene expression assays to confirm differential expression for candidates and suggests that genes identified by downstream experiments should also be sequenced.
More importantly, sequencing hg
modifier candidates allowed us to identify nonsynonomous polymorphism, which may underlie QTL. SIFT and/or PolyPhen predicted alterations in protein function for 15 of the 56 total nsSNP (27%) in nine of the 44 total genes (20%) (Table ). It has been shown that predicting function based on evolutionary conservation using programs such as SIFT and PolyPhen is very accurate [49
]. Four of the nine genes (Ubr1
, Ptpns1, Ubce7ip5
; all of which are on MMU2) are of particular interest. It has been suggested the Socs2
may rely on ubiquitination dependent proteasomal degradation to inhibit Gh
are both involved in protein ubiquitination and Ubr1
knockout mice show an 20% decrease in body weight partly due to a reduction in adipose tissue [50
knockout mice displayed a 10% reduction in body weight [52
] and Mmp9
knockout mice show a reduction in size and drastically reduced bone length [53
]. All of these genes are excellent functional and positional hg
modifier candidate genes and this information can be incorporated into future fine mapping studies.