Mammalian Genome

, Volume 15, Issue 6, pp 460–471

Identification of a congenic mouse line with obesity and body length phenotypes

Authors

    • Rowe Program in Human Genetics, Section of Neurobiology/Physiology/BehaviorUniversity of California
    • Department of PediatricsUniversity of California
  • Steven Stone
    • Myriad Genetics
  • Sally Chiu
    • Department of PediatricsUniversity of California
  • Adam L. Diament
    • Department of PediatricsUniversity of California
  • Pablo Corva
    • Department of Animal ScienceUniversity of Mar del Plata Balcarce
  • Donna Shattuck
    • Myriad Genetics
  • Robyn Riley
    • Myriad Genetics
  • Steven C. Hunt
    • Cardiovascular Genetics Research ProgramUniversity of Utah School of Medicine
  • Juliet Easlick
    • Department of PediatricsUniversity of California
  • Janis S. Fisler
    • Department of NutritionUniversity of California
  • Juan F. Medrano
    • Department of Animal ScienceUniversity of California
Article

DOI: 10.1007/s00335-004-2353-9

Cite this article as:
Warden, C.H., Stone, S., Chiu, S. et al. Mamm Genome (2004) 15: 460. doi:10.1007/s00335-004-2353-9

Abstract

Our primary objective was to discover simplified mouse models corresponding to human obesity linkages. We used the B10.UW– H3bwe Pax1unat/Sn (B10.UW) congenic strain, a subcongenic strain with a reduced UW strain donor region, and their C57BL/10SnJ background strain. The congenic and subcongenic UW strain donor regions are on mouse Chr 2. We measured body length [anal-nasal (AN) length], summed fat depot weights normalized for body weight (Adiposity Index, AI), and percentage of body weight that is lipid. The B10.UW congenic and subcongenic strains have significantly smaller AN lengths (p < 0.0001) and have a significantly lower AI and percentage of body weight as fat than the background strain (p < 0.0001). In an F2 intercross of the congenic and background strains, AN and AI were both linked to the distal half of the donor region with LOD scores greater than 19 and 5, respectively. F2 haplotypes identified a minimal region for AN linkage of 0.8 megabases (Mb) that is estimated to express four genes in the current Celera mouse genome assembly. We narrowed the most likely location of the obesity gene to 15 Mb whose homologous genes are all located on human Chr 20 in the region surrounding the centromere. Since a previous study identified human obesity linkage peaking near the centromere, then the B10.UW mice may exhibit obesity due to the homologous gene.

Mouse obesity models have been successfully used to identify genes that cause rare forms of human obesity (Chagnon et al. 2003a). Isolation of genes for complex quantitative traits is more difficult, but congenic mouse strains have proven to be a valuable tool (Aitman et al. 1999; Bodnar et al. 2002; Floyd et al. 2003; Ma et al. 2002). These strains are produced by breeding donor and background strains, then selecting animals that retain a chromosomal region from the donor during repeated backcross breeding to the background strain. The objective is to produce mice that are virtually identical except for a small selected target donor chromosomal region harboring genes that produce a distinguishable phenotype. A few congenic strains have been identified for obesity traits (Diament and Warden 2004; Lembertas et al. 1997; Stoehr et al. 2004; York et al. 1999), but since linkage studies in experimental crosses identified at least 168 chromosomal regions containing obesity genes (Chagnon et al. 2003b), then there are likely to be many congenics with obesity phenotypes. We reasoned that many congenics made for other experimental purposes, such as isolation of histocompatibility antigens, will also include obesity genes and are excellent “ready-made” experimental models to identify obesity genes.

Several studies using different inbred strains have identified quantitative trait loci (QTLs) for obesity on mouse Chr 2 (Corva et al. 2001; Lembertas et al. 1997; Mehrabian et al. 1998; Moody et al. 1999). The congenic strain B10.UW– H3bwe Pax1und/Sn (which will be abbreviated as B10.UW congenic) contains natural alleles of the UW/Le strain on a mouse Chr 2 donor segment. The UW mouse strain is named for the underwhite mutation that was discovered in the strain, but the underwhite gene itself maps to Chr 15 and is not present in this congenic. The B10.UW mice include four known alleles in the donor region, three of which have visible phenotypes. They include alleles for wellhaarig (we) at 73.5 cM, histocompatibility 3b (H3b) at 81.4 cM, the undulated allele of paired box gene 1 (Pax1un) at 82 cM, and the black and tan allele of agouti, (avt) at 89 cM. Wellhaarig is a spontaneous mutation with recessive inheritance that exhibits curly whiskers at 2–3 days and a wavy first coat most evident from 10 to 21 days (Graff et al. 1986). The undulated allele of Pax1 is a recessive allele where homozygotes have a shortened and usually kinked tail (Balling et al. 1993). This is due to a point mutation in the paired-box sequence of Pax1 (Balling et al. 1988). The black and tan allele of agouti produces a dominant black dorsum and a cream or yellow belly (Poole and Silvers 1976). The dorsal pigment pattern is distinct from that of other agouti alleles that result in banded hairs.

This report provides data demonstrating that the B10.UW congenic exhibits an obesity phenotype and locating an obesity gene to mouse Chr 2 in a region homologous to the region surrounding the centromere of human Chr 20. In addition, a body length gene is placed within approximately 0.8 Mb of the Pax1 gene on Chr 2.

Materials and methods

Mice

B10.UW– H3bwe Pax1unat/Sn mice (B10.UW congenic) and C57BL/10SnJ (B10) mice were purchased from The Jackson Laboratory. Three independent experiments compared obesity phenotypes of congenic and background strains. Our initial survey used both males and females of the B10.UW congenic and other Chr 2 congenic strains and the B10 background to identify candidate obesity model congenic strains as previously described (Diament and Warden 2004). We next mapped phenotypes to the donor region, using male and female F2 mice derived from a cross of B10.UW congenic × B10 mice. For the third independent experiment, B10.UW subcongenic mice were produced by breeding the B10.UW congenic mice and B10 mice to produce a new subcongenic strain with a reduced congenic interval that excluded the black and tan allele of agouti (at) from the UW strain, but that includes the black hair of the a allele of agouti derived from B10.

All mice used for experiments were bred at UC-Davis. In all three experiments, mice were weaned at 3 weeks of age, fed Purina Stock Chow 5001 (Dean’s Animal Feed, San Carlos, Calif.), separated by gender, and housed, three to five mice in each polycarbonate cage. We chose to use group housing for these mice even though it increases animal-to-animal variation, and may even interact with genotype to influence phenotypes, because it was cost effective and would allow us to identify statistically significant genotype effects in our animals. All animals were kept in the same room with a 14 hour light/10 hour dark cycle, 21 ± 2°C temperature, and greater than 25% humidity. Mice were given access to food and deionized water ad libitum.

At 12 weeks of age, mice were fed one of three diets as described below for each experiment: (a) the Stock Chow 5001, or (b) a defined low-fat diet (#D10001, Research Diets, Inc., New Brunswick, NJ), or (c) a defined moderate-fat diet (#D12266B, Research Diets, Inc., New Brunswick, NJ). The Stock Chow 5001 contains approximately 12% energy from fat and 60% energy from carbohydrate. The defined low-fat diet contains 11.5% fat and 68% carbohydrate. The defined moderate-fat diet contains 31.8% fat as butter and corn oil and 51.4% carbohydrate. The mice were on the respective diet for approximately 8 weeks and killed at approximately 20 weeks of age. Mice in the first and second experiment as described above were placed on either a defined low-fat diet or moderate fat diet, while mice in the third experiment either remained on chow or were switched to a defined low-fat or moderate-fat diet.

Phenotyping

We determined fat depot weights, Adiposity Index, and percentage fat as previously described (Warden et al. 1995). Briefly, mice were anesthetized by isoflurane and bled retro-orbitally. Body weight and anal-nasal (AN) length were measured, and mice were killed by cervical dislocation. Three white adipose tissue depots (femoral, gonadal, and retroperitoneal) were removed and weighed in experiments 1 and 2, and mesenteric white adipose tissue was included as a fourth dissected depot in experiment 3. Gonadal adipose tissue was stored at −80°C, while the rest of the adipose depots were returned to the carcass for determination of body composition by chemical extraction. The animal carcass was used to measure precise body composition by the method of Bell and Stern (1977). Percentage body fat was corrected for the removal of the gonadal depot by using the equation, percentage body fat = (carcass % fat + 0.8 (gonadal weight)) × 100/body weight Phenotype measures included anal-nasal length (AN), live weight, BMI [live weight (g)/length2 (cm2)], weights of individual fat pads, total fat pad weight, AI, (total fat pad weight/live weight) and percentage body fat. For all three experiments, femoral, gonadal, and retroperitoneal adipose depots were used to calculate AI to maintain consistency. Where values for percentage body fat are not available, AI is used as a surrogate. Data from the first experiment as described above show strong positive correlation between percentage body fat and AI, with r = 0.93 and p < 0.001.

Genotyping

Genomic DNA was extracted from kidney by using the Dneasy Tissue Kit (QIAGEN Inc., Valencia, Calif.). To determine the extent of the donor region in the B10.UW congenic, we genotyped congenic and B10 mice for 63 fluorescent dye-labeled di-, tri-, and tetranucleotide microsatellite repeat markers. The markers covered Chr 2 from D2Mit354 (0 cM) to D2Mit200 (110 cM) with an average genetic spacing of 1.8 cM. Genetic distance between markers and their relative orders were derived originally from the European Collaborative Interspecific Backcross (EUCIB) genetic map (Rhodes et al. 1998). Subsequently, marker order was confirmed by physical assignment in the Celera (www.celera.com) and Ensembl (www.ensembl.org) mouse genome assemblies. All dinucleotide repeat markers contained GTTT extensions at the 5′ ends of the non-dye-coupled PCR primers, to reduce the variability of the addition of non-templated nucleotides at the 3′ ends of the labeled products (Brownstein et al. 1996). Fluorescent D2Mit markers were sized on an ABI 377 or with agarose gel electrophoresis. Since the UW/Le donor strain DNA was not available, we assumed that microsatellite alleles were derived from the donor strain if the allele in the congenic was a size different from the allele in the B10 background strain. If the allele size in the congenic was the same as in B10, then the marker was either uninformative (i.e., the same allele was present in UW/Le and B10) or the marker resided on genomic DNA derived from B10. Therefore, it was not possible to precisely determine the size of UW donor region, but the large number of non-polymorphic markers both proximal and distal to UW-derived alleles is consistent with the hypothesis that we have crossed the breakpoint between donors and background-derived genomic DNA.

Statistical analysis

All comparisons of background, congenic and subcongenic strains were analyzed with Statview 5.0.1(SAS Institute, Gary, N.C.) and expressed as mean ± SD. Two- and three-way ANOVA was used to examine main effects of genotype, sex, and diet as well as interaction effects where applicable. Group differences within a given sex and diet were compared by one-way ANOVA followed by the Tukey-Kramer post-hoc test for multiple comparisons or Student’s t-test for pairwise comparisons. A p < 0.05 was considered significant.

Results

The B10.UW congenic is a model for body length and obesity

The B10.UW congenic mouse strain contains Chr 2 UW donor strain alleles on the B10 background. Since obesity in mice frequently exhibits sexual dimorphism and may also exhibit gene × diet interactions (Brockmann and Bevova 2002), we examined phenotypes of weight, length, and adiposity in male and female B10.UW congenic and the B10 background strain after being fed defined low- or moderate-fat diets. We observed that the B10.UW congenic mice are shorter (Table 1) and weigh less (data not shown) than B10 mice in both sexes and diets. There is no overall significant difference between congenic and background strains for BMI, indicating that weight of the congenics is generally proportional to size (length).
Table 1

Phenotypes of B10 and B10.UW congenic mice

Diet

Sex

Gen

N

AN (cm)

BMI (g/cm2)

% fat

AI

FWAT (g)

GWAT (g)

RWAT (g)

L fat

M

B

10

  9.9 ± 0.2a

0.28 ± 0.03a

16.3 ± 5.5a

0.033 ± 0.012a

0.28 ± 0.13a

0.53 ± 0.26a

0.13 ± 0.08a

  

C

  5

  8.9 ± 0.2b

0.27 ± 0.01a

12.5 ± 1.5a

0.020 ± 0.010a

0.14 ± 0.08b

0.25 ± 0.12b

0.05 ± 0.04a

M fat

M

B

15

10.2 ± 0.3a

0.30 ± 0.04a

25.2 ± 10.1a

0.056 ± 0.022a

0.64 ± 0.34a

0.92 ± 0.45a

0.30 ± 0.20a

  

C

  5

  9.0 ± 0.1b

0.28 ± 0.01a

19.4 ± 4.0a

0.041 ± 0.010a

0.32 ± 0.08a

0.50 ± 0.14a

0.13 ± 0.06a

L fat

F

B

15

  9.8 ± 0.2a

0.22 ± 0.02a

17.4 ± 9.7a

0.029 ± 0.015a

0.23 ± 0.12a

0.35 ± 0.21a

0.06 ± 0.05a

  

C

  5

  8.5 ± 0.2b

0.26 ± 0.01b

  8.6 ± 1.9a

0.014 ± 0.004b

0.11 ± 0.03b

0.13 ± 0.04b

0.02 ± 0.01a

M fat

F

B

15

  9.9 ± 0.2a

0.24 ± 0.03a

21.8 ± 9.9a

0.042 ± 0.024a

0.40 ± 0.27a

0.56 ± 0.45a

0.12 ± 0.12a

  

C

  5

  8.8 ± 0.2b

0.24 ± 0.01a

11.8 ± 2.3b

0.020 ± 0.005a

0.16 ± 0.03a

0.20 ± 0.07a

0.03 ± 0.01a

  

p values

        
  

Strain

 

<0.0001

NS

0.0015

0.0006

0.004

0.0003

0.0016

  

Sex

 

<0.0001

<0.0001

NS

0.0163

0.029

0.0072

0.0015

  

Diet

 

0.0002

NS

0.009

0.0006

0.001

0.0093

0.0063

  

Strain × sex

 

NS

0.007

NS

NS

NS

NS

NS

All values are mean ± SD. There are no significant strain × diet, sex × diet, or 3-way interactions. NS = p-value >0.05. AN length, BMI and AI were determined as described in materials and methods. Gen, genotype; B, background; C, congenic; L fat, defined low-fat diet; M fat, defined moderate-fat diet; AN, body length in cm; BMI, body mass index; % fat, percentage of body weight that is triglyceride; AI, Adiposity Index; FWAT, femoral white adipose tissue; GWAT, gonadal white adipose tissue; RWAT, retroperitoneal white adipose tissue.a,b Genotypes (background, congenic) not sharing a similar superscript within a given diet and sex are significantly different (p < 0.05) by one-way ANOVA followed by Student’s t-test.

We then tested the hypothesis that B10.UW congenic mice are an obesity model by examining phenotypes of fat mass that are corrected for body weight such as AI and percentage of body weight that is fat. There was a significant strain effect on body fat measures, with B10.UW congenic mice having a lower AI and less percentage fat than B10 mice (Table 1). Analysis of individual fat pad weights showed results similar to AI, indicating that there are no differential effects on the various adipose depots.

Determining the size of the B10.UW congenic donor region

Congenic donor regions have highly variable sizes that can be much larger than predicted on the basis of the number of generations of backcross to the background strain. Mouse Chr 2 contains many positional candidate obesity genes. We sought to determine which of these genes may have alleles underlying phenotypes of the B10.UW congenic mice by first determining the borders of the congenic donor region. We used microsatellite markers to infer the position of the donor region of the B10.UW congenic mice. As described in the materials and methods, we typed D2Mit markers in B10 and congenic mice. Table 2 lists the markers used, gives their positions as determined from the Celera and Ensembl mouse genome assemblies, and indicates which markers were polymorphic between B10 background and the B10.UW congenic, indicating the donor region. We ordered the markers by position in the genome assemblies because a high fraction of these markers are incorrectly ordered in the Mouse Genome Database, which is based on consensus linkage maps. Our results are consistent with the presence of a UW donor region of at least 43.3 Mb, extending from an as yet undefined start proximal of marker D2Mit103 at 115.1 Mb to an end distal to marker D2Mit27 at 158.4 Mb (Table 2). This corresponds to approximately 55.7–90 cM on the genetic map. Thus, the B10.UW congenic includes both the agouti and mahogany/attractin alleles of UW as well as hundreds of additional UW alleles on the B10 background.
Table 2

Donor strain regions of B10.UW mice and the subcongenic strain

Marker

Jax position (cM)

Ensembl position (MB)

Celera position (MB)

Allele size difference

Congenic donor region

Subcongenic donor region

D2Mit354

0.5

4.9

1.9

No

  

D2Mit2

4.0

7.5

4.5

No

  

D2Mit178

9.0

17.8

14.5

No

  

D2Mit521

15.3

26.3

22.8

No

  

D2Mit65

24.0

NMa

29.7

No

  

D2Mit240

30.0

49.6

47.5

No

  

D2Mit58

51.4

109.8

106.4

No

  

D2Mit420

54.6

NM

NM

Yes

x

x

D2Mit103

55.7

118.6

115.1

Yes

x

x

D2Mit305

60.1

NM

NM

Yes

x

x

D2Mit62

65.0

119.8

NM

Yes

x

x

D2Mit104

66.0

120.9

117.5

Yes

x

x

D2Mit421

56.8

NM

121.8

No

x

x

D2Mit16

69.5

126.4

123.0

No

x

x

D2Mit224

74.0

130.9

127.5

Yes

x

x

we

73.5

NM

NM

Yes

x

x

D2Mit78

73.5

132.1

128.7

Yes

x

x

D2Mit258

78.0

132.1

NM

Yes

x

x

D2Mit77

74.9

133.3

129.8

Yes

x

x

D2Mit46

76.3

135.7

132.2

No

x

x

D2Mit28

78.2

NM

135.0

Yes

x

x

D2Mit353

63.4

140.0

136.5

Yes

x

x

D2Mit138

79.4

141.7

138.2

Yes

x

x

D2Mit401

79.7

144.0

140.4

Yes

x

x

D2Mit21

80.0

144.4

140.9

Yes

x

x

D2Mit194

81.4

145.6

142.0

Yes

x

x

D2Mit491

81.7

146.0

NM

Yes

x

x

D2Mit403

81.7

146.4

142.9

Yes

x

x

D2Mit168

81.7

146.5

143.0

Yes

x

x

D2Mit109

81.7

147.2

143.7

Yes

x

x

D2Mit280

81.7

147.8

151.8

No

x

x

Pax1

82.0

149.1

145.6

Yes

x

x

D2Mit492

67.8

150.0

146.4

Yes

x

x

D2Mit57

82.0

150.0

146.7

No

x

x

D2Mit281

82.0

150.0

146.8

Yes

x

x

D2Mit407

68.9

150.3

146.9

Yes

x

x

D2Mit47

83.0

150.4

147

No

x

x

D2Mit408

68.9

151.2

147.7

No

x

x

D2Mit110

83.0

151.3

NM

Yes

x

x

D2Mit283

83.5

152.0

148.5

No

x

x

D2Mit261

83.5

152.7

148.9

No

x

x

D2Mit449

69.9

153.3

154.4

No

x

x

D2Mit424

71.0

153.3

154.5

No

x

x

Tcf15

83.9

NM

155

Yes

x

x

D2Mit309

71.0

154.1

155.3

No

x

x

D2Mit284

86.0

154.7

155.9

No

x

x

D2Mit450

72.1

154.9/155

156

No

x

x

D2Mit139

86.0

155.4

156.4

Yes

x

 

D2Mit493

72.1

155.8

156.8

Yes

x

 

D2Mit286

87.0

156.4

157.4

Yes

x

 

a

89.0

156.9

157.9

Yes

x

 

D2Mit27

87.0

NM

158.4

Yes

x

 

D2Mit494

73.2

157.5

158.5

No

  

D2Mit451

73.2

157.5

158.6

No

  

D2Mit262

87.0

157.7

158.8

No

  

D2Mit55

91.0

159.8

160.9

No

  

D2Mit411

77.6

161.5

162.5

No

  

D2Mit263

92.0

164.2

165.3

No

  

D2Mit29

94.0

166.1

167.2

No

  

a NM=Not mapped

A gene influencing obesity maps to the Chr 2 donor region of the B10.UW congenic

All congenic strains may contain unlinked donor strain alleles that can potentially influence the observed trait. Therefore, we performed an F2 intercross of the congenic and background strains to confirm that the observed obesity traits map to the donor region and to fine map potential QTL. 228 male and female F2 mice from a cross of B10 and B10.UW congenic mice were sacrificed after 5 weeks on the defined moderate-fat diet. All mice were scored for the recessive undulated tail phenotype caused by Pax1un and for the dominant black and tan at allele of agouti. There was highly significant linkage of AI, but not BMI to the Pax1un undulated mutation (Table 3). There was no linkage of AI to the at allele of agouti (data not shown). Comparisons of Pax1un/Pax1un with Pax1/Pax1un heterozygotes and Pax1/Pax1 (homozygous for the Pax1 alleles from B10) were both significant for AI and AN length, confirming that both traits map to the UW donor region. Each of the three fat depots measured exhibited very strong linkage to the Pax1un mutation, reiterating that there is no evidence the obesity genes present in this model have differential effects on one depot versus another.
Table 3

Linkage of phenotypes to Pax1un in congenic donor region in F2 mice

Sex

Genotype

N

AN (cm)

AI

BMI (g/cm2)

FWAT (g)

GWAT (g)

RWAT (g)

M

Wildtype tail

  90

9.9 ± 0.2a

0.057 ± 0.017a

0.31 ± 0.03a

0.45 ± 0.17a

0.94 ± 0.36a

0.38 ± 0.15a

 

(Pax1/Pax1 or Pax1/Pax1un)

       

M

Undulated tail

  19

9.1 ± 0.45b

0.031 ± 0.014b

0.29 ± 0.03b

0.19 ± 0.09b

0.42 ± 0.21b

0.13 ± 0.09b

 

(Pax1un/Pax1un)

       

F

Wildtype tail

100

9.7 ± 0.3a

0.028 ± 0.012a

0.24 ± 0.02a

0.21 ± 0.11a

0.38 ± 0.25a

0.82 ± 0.07a

 

(Pax1/Pax1 or Pax1/Pax1un)

       

F

Undulated tail

  19

8.8 ± 0.3b

0.018 ± 0.012b

0.25 ± 0.03b

0.12 ± 0.08b

0.22 ± 0.17b

0.04 ± 0.04b

 

(Pax1un/Pax1un)

       
 

p-values:

       
 

Genotype

 

<0.0001

<0.0001

NS

<0.0001

<0.0001

<0.0001

 

Sex

 

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

 

Sex × genotype

 

NS

0.0038

<0.0001

0.0005

0.0008

<0.0001

NS = p > 0.05. A11 values are means ± SD. AN length, BMI, and AI were determined as described in materials and methods. AN, body length in cm; AI, Adiposity Index; BMI, body mass index; FWAT, femoral white adipose tissue; GWAT, gonadal white adipose tissue; RWAT, retroperitoneal white adipose tissue. Genotypes (wildtype tail, undulated tail) not sharing a similar superscript within a given diet and sex are significantly different (p < 0.05) by one-way ANOVA followed by Student’s t-test.

High-resolution mapping was performed by additional genotyping on DNA extracted from 69 (the 20% heaviest and leanest) mice. Genotypes were determined for 10 microsatellite markers. Body length (AN) was linked to the distal half of the donor region with the highly significant LOD score of 19.2. There was no linkage of AI to the first two markers in the proximal half of the congenic donor region, while linkage to all other markers in the distal half of the congenic donor region was highly significant, with LOD scores exceeding 5 (Figure 1). A LOD plot for BMI in the same mice revealed a QTL that did not reach statistical significance (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00335-004-2353-9/MediaObjects/fig1.jpg
Fig. 1

Localization of the B10.UW AN length and obesity gene and homology to human Chromosome 20. The top portion of this figure shows results for QTL mapping. Selective genotyping was performed in the F2 cross (N = 228). The mice were ranked based on AI, and 69 mice (32 males and 37 females) from the extreme ends of the F2 distribution were typed. In order to determine the location of the QTL within the donor segment on Chr 2, interval mapping was performed by regression analysis (Haley and Knott 1992). Although selective genotyping introduces a bias in the estimated QTL effects, it does not affect the F distribution under the null hypothesis (Stone et al. 1999). The regression analyses were run with SAS software (SAS Institute, Inc., Cary, N.C.). The regression models included the effect of sex and the additive (a) and dominance (d) effects of the putative QTL. The regression analysis between markers was performed at 1-cM intervals. The results were expressed as LOD scores, LOD = 4.605 × Likelihood Ratio test (LR), where LR = n×loge (RSSreduced/RSSfull) (n is the sample size, and RSSfull, and RSSreduced are the Residual Sum of Squares of the complete regression model and the model with the additive (a) and dominance (d) terms omitted, respectively (Haley and Knott 1992). The x-axis shows the megabase positions of the markers along mouse Chr 2. The known donor positions of the B10.UW congenic and subcongenic are depicted by black bars below the QTL map. The homologous human chromosomes derived from the Celera mouse-human synteny maps (www.celera.com) are shown below. An expanded map shows the mouse megabase positions in the region of the UW donor chromosome most likely to include at least one obesity gene. The proximal border is at the LOD-1 location from the AI QTL peak above, while the distal border is at the position of the end of the subcongenic. The corresponding homologous locations from human Chr 20, derived from the Celera mouse–human synteny maps, are shown below, along with the corresponding human cytological map positions. Finally, positions of potential candidate genes are shown at the bottom of the figure.

Obesity of male and female congenic and subcongenic strains

We sought to further localize the genes underlying obesity by producing a subcongenic strain where the donor region is reduced by selection to include UW alleles of Pax1un and B10 alleles of agouti. Although there is no published evidence that common alleles of agouti influence obesity, it remains possible that some will. The subcongenic and congenic strains had similar AN lengths and were both significantly shorter than the background (Table 4). Although there are some apparent small differences for BMI, AI, and the separate fat pads between congenic and subcongenic strains, perhaps owing to experiment-to-experiment variation, the overall consistent result is that both congenic strains are leaner than the background strain.
Table 4

Phenotypes of background, congenic, and subcongenic mice

Diet

Sex

Gen

N

AN (cm)

BMI (g/cm2)

AI

FWAT (g)

GWAT (g)

RWAT (g)

MWAT (g)

Chow

M

B

7

9.9 ± 0.23a

0.28 ± 0.01a

0.032 ± 0.003a

0.24 ± 0.03a

0.50 ± 0.07a

0.14 ± 0.02a

0.17 ± 0.04a

  

C

14

8.9 ± 0.30b

0.28 ± 0.02a

0.015 ± 0.005b

0.10 ± 0.03b

0.21 ± 0.08b

0.02 ± 0.02b

0.07 ± 0.04b

  

SC

20

8.9 ± 0.31b

0.30 ± 0.03b

0.018 ± 0.005b

0.13 ± 0.03c

0.25 ± 0.08b

0.04 ± 0.03b

0.11 ± 0.07ab

L fat

M

B

10

9.7 ± 0.16a

0.30 ± 0.01a

0.047+0.007a

0.42 ± 0.09a

0.70 ± 0.15a

0.23 ± 0.06a

0.31 ± 0.11a

  

C

10

9.0 ± 0.14b

0.30 ± 0.02a

0.034 ± 0.013b

0.26 ± 0.10b

0.44 ± 0.21b

0.13 ± 0.08b

0.19 ± 0.06b

  

SC

9

8.8 ± 0.40b

0.32 ± 0.02a

0.031 ± 0.005b

0.25 ± 0.05b

0.43 ± 0.08b

0.10 ± 0.04b

0.14 ± 0.05b

M fat

M

B

9

9.8 ± 0.13a

0.30 ± 0.02a

0.056 ± 0.007a

0.45 ± 0.10a

0.94 ± 0.21a

0.27 ± 0.05a

0.27 ± 0.07a

  

C

8

8.8 ± 0.21b

0.31 ± 0.02ab

0.039 ± 0.009b

0.30 ± 0.07b

0.54 ± 0.15b

0.10 ± 0.06b

0.18 ± 0.06a

  

SC

10

9.0 ± 0.35b

0.33 ± 0.02b

0.049 ± 0.012ab

0.38 ± 0.12ab

0.73 ± 0.27ab

0.21 ± 0.06a

0.23 ± 0.10a

Chow

F

B

9

9.7 ± 0.20a

0.23 ± 0.01a

0.023 ± 0.010a

0.17 ± 0.06a

0.29 ± 0.17a

0.04 ± 0.03a

0.18 ± 0.07a

  

C

7

8.7 ± 0.42b

0.26 ± 0.02b

0.010 ± 0.004b

0.10 ± 0.03b

0.10 ± 0.05b

0.00 ± 0.01b

0.06 ± 0.03b

  

SC

15

8.6 ± 0.57b

0.26 ± 0.02b

0.015 ± 0.006b

0.12 ± 0.03b

0.16 ± 0.09b

0.01 ± 0.02b

0.09 ± 0.04b

L fat

F

B

10

9.5 ± 0.11a

0.24 ± 0.01a

0.027 ± 0.010a

0.20 ± 0.06a

0.34 ± 0.15a

0.05 ± 0.03a

0.15 ± 0.09a

  

C

10

8.8 ± 0.25b

0.26 ± 0.01a

0.025 ± 0.007a

0.19 ± 0.06a

0.27 ± 0.10ab

0.04 ± 0.03a

0.14 ± 0.06a

  

SC

10

8.6 ± 0.33b

0.27 ± 0.20a

0.019 ± 0.005a

0.15 ± 0.04a

0.22 ± 0.06b

0.01 ± 0.02b

0.10 ± 0.04a

M fat

F

B

11

9.7 ± 0.20a

0.22 ± 0.02a

0.028 ± 0.011a

0.21 ± 0.09a

0.34 ± 0.17a

0.05 ± 0.03a

0.15 ± 0.04a

  

C

4

8.6 ± 0.52b

0.26 ± 0.01b

0.028 ± 0.007a

0.22 ± 0.07a

0.31 ± 0.12a

0.04 ± 0.02a

0.16 ± 0.10a

  

SC

10

8.4 ± 0.37b

0.27 ± 0.02b

0.016 ± 0.012a

0.12 ± 0.08a

0.18 ± 0.17a

0.02 ± 0.03a

0.07 ± 0.05b

p-value:

          

Strain

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

   

Sex

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

   

Diet

NS

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

   

Strain × sex

NS

<0.0001

.004

<0.0001

.001

<0.0002

NS

   

Strain × diet

NS

NS

NS

NS

NS

.003

0.03

   

Sex × diet

NS

<0.0001

<0.0001

<0.0001

<0.0001

<0.0001

0.0004

   

All values are mean ± SD. There are no significant three-way interactions. NS = p-value >0.05. AN length, BMI, and AI were determined as described in materials and methods. Gen, genotype; B, background; C, congenic; SC, subcongenic; L fat, defined low-fat diet; M fat, defined moderate-fat diet; AN, body length in cm; AI, Adiposity Index; % fat, percentage of body weight that is triglyceride; BMI, body mass index; FWAT, femoral white adipose tissue; GWAT, gonadal white adipose tissue; RWAT, retroperitoneal white adipose tissue; MWAT, mesenteric white adipose tissue.a,b,c Genotypes (background, congenic, subcongenic) not sharing a similar superscript within a given diet and sex are significantly different (p < 0.05) by one-way ANOVA followed by the Tukey-Kramer post-hoc test.

Determining the size of the UW donor region of the B10.UW subcongenic

We further defined the UW donor region of the B10.UW subcongenic because it retains obesity phenotypes while having a smaller donor region than the B10.UW congenic. Thus, it can be used to refine the positional location of the underlying obesity gene. Since the B10.UW subcongenic was developed by selection for two visible phenotypes, the black hairs of the B10 allele of agouti and the undulated tail UW allele of Par1 (Pax1un), we typed two additional polymorphisms in the established B10.UW subcongenic line. We typed one polymorphism proximal to Par1 and one polymorphism proximal to agouti. First, we confirmed that the subcongenic retains the we allele of wellhaarig that is proximal to Pax1 and is also present in the B10.UW congenic. Next, we demonstrated that the microsatellite marker D2Mit139, which is proximal to agouti, has a UW genotype in the congenic and a B10 background genotype in the subcongenic. The subcongenic was not genotyped for the other markers polymorphic between UW and B10, and thus the UW donor region may be smaller than shown in Table 2, since Table 2 assumes that all UW DNA between we and the proximal end is retained and that all DNA up to D2Mitl39 is also UW derived donor DNA. Nevertheless, the subcongenic has a relatively large donor region of approximately 36.3 Mb that includes several candidate genes (Fig. 1).

Localization of a body length gene

AN length had a LOD score exceeding 19.2 for linkage to the B10.UW donor region in the F2 animals, with the peak over the Pax1 gene (Fig. 1). Therefore, we examined haplotypes of the 69 F2 mice genotyped at high resolution to determine positional location of the body length AN gene. We determined whether the body length gene exhibits additive or dominant/recessive inheritance from the 49 F2 mice that are non-recombinant within the B10.UW donor region. Mice that were “BB” homozygotes (B10/B10) had AN lengths of 9.9 ± 0.2 cm (SD), with a range from 9.5 to 10.1 (n = 10). Mice with the “BD” heterozygous haplotype had AN lengths of 9.9 ± 0.2 cm (n = 28), with a range from 9.0 to 10.4. Mice with the “DD” haplotype (B10.UW/B10.UW homozygotes) had AN lengths of 8.7 ± 0.4 cm with a range from 8.1 to 9.4 cm (n = 11). There were no significant differences between males and females and no sex × genotype interactions. Mice with the “BB” and “BD” haplotypes were both significantly different from “DD” haplotypes with p < 0.0001, while there was no significant difference between “BB” and “BD” haplotypes. Thus, the body length phenotype exhibits a dominant/recessive inheritance, with “B” genotype dominant for longer AN, and “D” genotype recessive for shorter AN length. We identified three male mice that were recombinant immediately proximal or distal to the Pax1 gene (Table 5). All three mice had undulated tails, indicating that they are Pax1un homozygotes. Mice 4603 and 4714 have identical genotypes of B10.UW “DD” homozygotes from the proximal D2Mit104 marker until Pax1 and were recombinant to B10.UW/B10 “BD” heterozygotes at the following D2Mit492 marker. However, they differ for body length, since mouse 4603 had an AN length of 9.0 cm and mouse 4714 had an AN length of 10.0 cm. This suggests that the AN length gene is located in the 0.8 megabases between D2Mit492 and Pax1, where both mice exhibit their only recombination. Genotypes from mouse 4908 are consistent with this placement, since it is both short (AN = 9.2 cm) and has “DD” genotypes from Pax1 thru D2Mit286 with a recombination immediately proximal to Pax1.
Table 5

Haplotypes indicate a gene 0.8 megabase distal to Pax1 influences body length

 

Mouse number

Marker

4603

4714

4908

D2Mit104

DD

DD

BD

D2Mit258

DD

DD

BD

D2Mit21

DD

DD

BD

D2Mit194

DD

DD

BD

D2Mit491

DD

DD

BD

D2Mit403

DD

DD

BD

D2Mit109

U (inferred DD)

U (inferred DD)

BD

Pax1

DD = undulated tail

DD = undulated tail

DD = undulated tail

D2Mit492

BD

BD

DD

D2Mit407

BD

BD

DD

D2Mit139

BD

BD

DD

D2Mit286

BD

BD

DD

Body length (AN, cm)

9.0

10.0

9.2

Inferred body length genotype

DD

DD

BB or BD

Three individual mice with informative recombinations are shown from the F2 cross summarized in Table 3. DD = B10.UW congenic donor region homozygotes, BD = B10.UW donor region/B10 heterozygotes, U = no data. The undulated Pax1un/Pax1un genotype is a recessive B10.UW congenic donor genotype (DD). Although D2Mit109 was not genotyped in mice 4603 and 4704, we infer that it is “DD” genotype since the flanking markers D2Mit403 and Pax1 are “DD” genotype.

Discussion

Results of three separate experiments demonstrate that the B10.UW congenic strain is a model for variations in body length and obesity. We separated obesity effects from confounding effects of body length and weight. We first demonstrated that the congenic has significantly lower AI and percentage body fat than the B10 background; that is, UW alleles decrease obesity (Table 1). Second, we tested the hypothesis that phenotypes of the B10.UW congenic mice are due to alleles in the UW donor region of Chr 2 in an F2 intercross of background B10 and congenic B10.UW. From this experiment, we observed highly significant UW donor strain genotype effects on AI, AN length (peak LOD score 19.2, Fig. 1) and weight (data not shown), but not BMI. We then used a subcongenic strain to confirm the existence of both a body length and an obesity gene in the donor region near Pax1. It is interesting to speculate that one gene may influence both body length and weight, since at least one prior report (Reed et al. 2003) has identified a QTL for body length and adiposity in an F2 intercross of 129/P3 J and C57BL/6ByJ strains. The peak for length appears to be immediately proximal to the agouti locus. The adiposity QTL peak is more spread out, but the maximal LOD score is approximately co-incident with the body length QTL. Discovery of the underlying gene would be made easier if a single gene or allele influences both length and adiposity, since the length trait is mapped to a region of less than 1 Mb.

Large regions of mouse and human chromosomes include orthologous genes in the same orders. This property of synteny has long been used to predict where QTL identified in one species will be found in the other. However, completion of the mouse and human genome sequences has provided a much more detailed knowledge of the precise positions and organization of these synteny regions. For instance, it is now clear that even regions found on the same chromosomes in mice and humans frequently include several distinct syntenic blocks. Donor region mapping revealed that the B10.UW congenic has a minimal donor region of 433 Mb (Table 2). Examination of the Celera mouse–human synteny maps reveals that at least 30 of the 433 Mb of the minimal B10.UW donor region are homologous to human Chr 20.

The combined data from subcongenic and F2 QTL mapping provide a much more precise localization of the underlying obesity gene than would be available from the congenic only (Fig. 1). The LOD score plot can determine an approximate 90% confidence interval as the peak LOD-1 (shown in Fig. 1). Thus, the LOD plot suggests that the obesity gene is in the distal half of the donor region, while the subcongenic eliminates all genes distal to D2Mit139, which is approximately 3 cM or 1.5 Mb proximal to agouti (Table 2). Thus, the most likely position of the obesity gene is from 138.5 Mb to 156.8 Mb on the Celera assembly (Ensembl 141 Mb to 155.8 Mb). This region contains approximately 160 genes and excludes mahogany/attractin. Included in these genes are several whose predicted functions make them candidate genes: proprotein convertase subtilisin/kexin type 2 (Pcsk2), which is a serine-type protease at 81.4 cM (141.7 Mb, Celera) that influences growth and neuroendocrine peptide processing (Zhu et al. 2002); acetyl-Coenzyme A synthetase 2-like (ADP forming, Acas2l) which is a lipogenic enzyme that may influence energy balance and that is located at 148.8 Mb (84 cM, Sone et al. 2002); and somatostatin receptor 4 (Smstr4), which is a G-protein, coupled receptor involved in somatostatin signaling that is located at 146.7 Mb (84.0 cM). Smstr5 has been reported to influence insulin secretion and glucose homeostasis (Strowski et al. 2003). Foxa2, which is a member of a transcription factor family including genes involved in causing diabetes, is located at 146.3 Mb. While Foxa2 is an obesity or diabetes candidate gene, it is unlikely that alleles of Foxa2 could influence body length. Thus, the combined effects of all our experiments narrow the region most likely to have the UW obesity gene to approximately 15 Mb. Candidate genes that might underlie an obesity QTL on mouse Chr 2 have also been identified by oligonucleotide microarray expression profiles of mRNA from 111 F2 mice of a DBA/2J × C57BL/6J cross (Schadt et al. 2003). These mice have a fat pad mass QTL whose peak is immediately distal to the agouti gene. Several differentially expressed genes were identified that map distal to agouti. While the proximal portion of the QTL overlaps with the B10.UW congenic donor region, the differentially expressed candidate genes identified all reside outside the subcongenic donor region and are not candidate genes for the B10.UW obesity effects.

At least ten published papers have reported obesity or obesity-related QTLs on mouse Chr 2 and/or human Chr 20. These reports may be repeated identification of the same or linked genes. In Table 6, we examine the hypothesis that some of the studies have overlapping QTLs, and others are non-overlapping. Table 6 provides estimates for the positions of peak linkage for each study as physical position on mouse Chr 2. Many QTLs are only approximately defined, and evidence for overlap is necessarily imprecise. Nevertheless, the results are consistent with the hypothesis that the various QTLs are due to more than one gene and that several QTLs overlap with that observed in the B10.UW mice. The overlap with B10.UW mice appears to be particularly marked for one study of human obesity where the peak linkage occurred near the centromere (Hunt et al. 2001). The B10.UW mice also clearly have an overlapping obesity linkage to that observed in F2 mice of the NZB × SM cross (Lembertas et al. 1997).
Table 6

Mouse Chr 2 physical positions of all published human Chr 20 and mouse Chr 2 QTLs

Reference

Species

Published peak location

Estimated mouse Chr 2 location

Overlaps with B10.UW subcongenic

Figure or table from reference including peak

Phenotype for QTL

LOD or p-value

Lembertas   et al. 1997

Mouse

D2Mit28 to D2Mit22

135–154 Mb

Yes

Figure 1

% fat

3.6

Mehrabian   et al. 1998

Mouse

D2Mit57 to D2Mit25

146.6–174.1 Mb

Yes

Figure 3

% fat

5.8

Hunt   et al. 2001

Human

46–58 cM (human)

∼21–-∼37.5 Mb (human)

Yes

Figure 5

BMI

4.2

   

145.1–164.3 Mb (mouse)

    

Moody   et al. 1999

Mouse

1 LOD confidence   interval is   50–85 cM

112.5–172.2 Mb

Yes

Figure 1

Heat loss, Hlq2

3.5

  

73–88 cM (+/+ mice)

127.3–156.8 Mb

Yes

Figure 3B

Weight gain, 2–9   weeks.

8.0

Corva   et al. 2001

Mouse

53–68 cM   (hg/hg mice)

115.0–145.6 Mb

Yes

Figure 3A

Weight gain, 2–9   weeks, Q2Ucd2

7.4

Hirayama   et al. 1999

Mouse

D2Mit156 ± 5 cM

34.8–62.6 Mb

No

Figure 3, cited   peak marker

Body weight

5.9

     

+/−

  

Taylor and   Phillips1997

Mouse

GcgChgb

61.2–131 Mb

No

Figure 2

Adiposity

5.1

Dong   et al. 2003

Human

D20S178D20S149

43.3–52.3 Mb (human)

No

Table 1

BMI, % Fat

2.6

   

168.9–177.1 Mb (mouse)

    

Lembertas   et al. 1997

Human

D20S176MC3R

44.0–51.6 Mb (human)

No

Figure 3

% Fat

p < 0.004

   

169.4–176.0 Mb (mouse)

    
  

PLC1D20S17

36.5–41.6 Mb (human)

No

Figure 3

% Fat

p < 0.01

   

163.5–167.5 (mouse)

    

Cheverud   et al. 1996

Mouse

No peak. Significant   markers include

120.9 Mb and 154.6 Mb

Yes

Table 4

Weight gain   early and late

3.9

  

D2Mit17 and D2Mit22

     

Mouse chromosomal location corresponds to the Celera mouse assembly in Mb. We estimated the 1 LOD (LOD-1) confidence intervals as this provides the most stringent test for overlap between the various studies. Although the quantitative trait genes underlying each QTL may be located outside these intervals, those QTLs that overlap at the higher stringency of 1 LOD are more likely to be due to the same gene(s). Some of the included papers analyze their data using several different methods, which can result in overlapping but different QTL peaks. Hence, for several papers we report more than one peak. The interval used from Moody et al. (1999) is taken from Fig. 1. Moody et al. also identified QTLs for fat depots and fat mass, but none were present on Chr 2. Intervals used for Corva et al. (2001) are taken from Fig. 3 for 2 to 9-week weight gain, with separate peaks for hg/hg and +/+ mice as they are overlapping but not identical. Another less significant QTL in Corva et al. is more proximal and clearly does not overlap with the B10.UW congenic region. Cheverud et al. (1996) genotyped only four markers on Chr 2 and did not provide a LOD plot, so we cannot determine a 1 LOD support interval. Rather, we provide the locations of two significant microsatellite markers identified with the likelihood that the 1 LOD confidence interval is larger. Hunt et al. (2001) examined linkage assuming dominant or recessive inheritance. These two models produced peaks at somewhat different, but overlapping, positions. A model summarized in Fig. 5 of Hunt et al. (2001) includes both dominant and recessive pedigrees and includes heterogeneity. These data are consistent with complex obesity linkage in humans. Indeed, Hunt et al concluded that their data were consistent with the presence of two obesity genes on Chr 20. Results of Dong et al. (2003) are from Table 2. They determined QTLs using conditional analysis from families with BMI threshold of 27 or greater. Stated peak linkage was 65–83 cM human Chr 20. Lembertas et al. (1997) report (Fig. 3) two putative QTLs for percentage fat, both with p-values that are suggestive but not significant and a separate, more statistically significant QTL for fasting insulin levels between D20S120 and D20S171 (78–88 cM human).

Our data are the first to show that a body length gene must be very close to Pax1. Indeed, examination of haplotypes in the F2 mice reveals three mice homozygous for Pax1un, but with recombinations at the markers immediately proximal or distal to Pax1. Thus, the most likely position of the body length “AN” gene is immediately distal of Pax1 to immediately proximal of D2Mit492 (Table 5). Pax1 itself is not a candidate “AN” length gene because mouse 4714 has an undulated tail but has a length consistent with the “H” haplotype. Only four genes are in this region according to the most recent Celera mouse assembly. These include a beat shock protein (HSP90 beta); Foxa2, also called hepatocyte nuclear factor 3b (Hnf3b); and two unnamed, uncharacterized genes. None has been previously associated with body length, although knockout studies have demonstrated that Foxa2 is a transcription factor that influences adipocyte differentiation (Wolfrum et al. 2003) and glucose-induced insulin release (Wang et al. 2002) and may be a potential candidate underlying the obesity phenotypes.

Finally, if the results we have obtained with mouse strains congenic for Chr 2 are relevant to the corresponding region of human Chr 20, these data suggest that it will be difficult to isolate the underlying human genes because there may be more than one obesity gene and that there may be complex environmental and genetic interactions between these linked genes. Therefore, studies in mice may provide a productive approach to the identification of the genes on human Chr 20 whose alleles influence human obesity.

Acknowledgments

This work was supported by National Institutes of Health grant R01 DK52581, NIH training grants PHS DK07355 and T32 GM007377, and the UC Davis Clinical Nutrition Research Unit, NIH DK35747. We thank Susan Bennett and Noreen Shibata for their excellent work managing the mouse colonies.

Copyright information

© Springer-Verlag 2004