Altered metabolism of growth hormone receptor mutant mice: a combined NMR metabonomics and microarray study

Abstract

Background: Growth hormone is an important regulator of post-natal growth and metabolism. We have investigated the metabolic consequences of altered growth hormone signalling in mutant mice that have truncations at position 569 and 391 of the intracellular domain of the growth hormone receptor, and thus exhibit either low (around 30% maximum) or no growth hormone-dependent STAT5 signalling respectively. These mutations result in altered liver metabolism, obesity and insulin resistance.

Methodology/principal findings: The analysis of metabolic changes was performed using microarray analysis of liver tissue and NMR metabonomics of urine and liver tissue. Data were analyzed using multivariate statistics and Gene Ontology tools. The metabolic profiles characteristic for each of the two mutant groups and wild-type mice were identified with NMR metabonomics. We found decreased urinary levels of taurine, citrate and 2-oxoglutarate, and increased levels of trimethylamine, creatine and creatinine when compared to wild-type mice. These results indicate significant changes in lipid and choline metabolism, and were coupled with increased fat deposition, leading to obesity. The microarray analysis identified changes in expression of metabolic enzymes correlating with alterations in metabolite concentration both in urine and liver. Similarity of mutant 569 to the wild-type was seen in young mice, but the pattern of metabolites shifted to that of the 391 mutant as the 569 mice became obese after six months age.

Conclusions/significance: The metabonomic observations were consistent with the parallel analysis of gene expression and pathway mapping using microarray data, identifying metabolites and gene transcripts involved in hepatic metabolism, especially for taurine, choline and creatinine metabolism. The systems biology approach applied in this study provides a coherent picture of metabolic changes resulting from impaired STAT5 signalling by the growth hormone receptor, and supports a potentially important role for taurine in enhancing beta-oxidation.

Figure 1. Structure of the intracellular domain of the Growth Hormone Receptor (GHR).

The mutations in the GHR have been made in the intracellular domain (ICD) of the receptor. (A) The wild-type has intact signaling through JAK2, MAPK and STAT5. (B) In mutant 569 the ICD has been truncated at residue 569 and two distal tyrosines were mutated to phenylalanines resulting in only 30% of wild-type STAT5 signaling in response to GH. (C) Mutant 391 has been truncated at residue 391 and has no STAT5 signaling ability, while normal JAK2 and MAPK signaling is maintained.

Figure 2. Identification of marker genes differentiating between the groups and physiological characterisation of GHR mutant mice.

(A) A heatmap of classifier gene expression in liver tissue of wild-type and GHR mutant mice at 42 days age from the GeneRaVE analysis, clustered according to similarity of expression. Gene expression has been represented as a scale between red and blue, with red indicating over expression and blue representing under expression. Gene abbreviations are used according to current nomenclature. (B) A picture of 10 month old male mice used in this study. (C) The weight curves of male mice from 2 months to 10 months. (D) Subcutaneous fat accumulation of male mice from 2 months to 13 months. (E) Perirenal fat accumulation of male mice from 2 months to 13 months. (C-E) green = wild-type, red = mutant 569, blue = mutant 391, black = GHR−/−.

Figure 3. 1D proton NMR spectra of mouse urine at 298 K.

Spectra of one individual of four months age from each mouse strain are shown. Top: wild-type, middle: 569 mutant, bottom: 391 mutant. The identity of relevant metabolites is indicated above each spectrum.

Figure 4. Statistical analysis of metabolites in mouse urine.

(A) Principal components analysis - scores plot of PC1 versus PC2. Each data point represents one mouse urine sample, and the distance between points in the score plot is an indication of the similarity between samples. Green circles: wild-type, red squares: 569 mutants, blue triangles: 391 mutants. The borders of the groups are also highlighted by lines in the corresponding colors. (B) Principal components analysis - loadings plot of PC1 versus PC2. Each data point represents one bucket (with the chemical shift indicated explicitly). The plot indentifies which spectral regions (and thus which chemical compounds) are responsible for the differences between the spectra observed in the scores plot. The loadings coefficients in each dimension are correlation coefficients that indicate how strongly each metabolite is correlated with the observed variance in the respective dimension. t = 0 means no correlation, and t = 1 means total correlation. Several significant metabolites are indicated explicitly. 2-OG: 2-oxoglutarate, DMA: dimethylamine, TMA: trimethylamine, TMAO: trimethylamine-N-oxide. The model consists of 7 PCs and represents data from 48 samples. Regions containing water, urea, and ethanol signals were excluded from the PCA. (C) Partial Least Squares-Discriminant Analysis - scores plots of PLS1 versus PLS2. The coding of groups is same as in panel (A). (D) Partial Least Squares-Discriminant Analysis - loadings plot of PLS1 versus PLS2. The chemical shift of each bucket as well as selected metabolites are indicated explicitly. Abbreviations as in panel (B). (E) Metabolic trajectories of the three mouse strains. Depicted is the change in the first PLS component PLS1 with the age of the mice. The mean metabolic trajectories (obtained by averaging the PLS1 scores of mice with similar age) are indicated by thick lines. The wild-type trajectory is colored green, the 569 mutant red, and the 391 mutant blue. The 2σ standard deviation from each mean trajectory is indicated by areas shaded in the respective color. As can be seen, the metabolic trajectory of the 569 mutant mice moves with age from the position of the wild-type mice to the position of the 391 mutant mice.

Figure 5. Statistical analysis of metabolites in murine liver tissue.

(A) 1D proton HR-MAS NMR spectrum of murine liver tissue. The identity of relevant metabolites is indicated. Glu: glutamate, Gln: glutamine (B) Principal components analysis - scores plot of PC1 versus PC2. Each data point represents one liver tissue sample. Green circles: wild-type, red squares: 569 mutants, blue triangles: 391 mutants, orange diamonds: wild-type on high-fat diet. (C) Principal components analysis - loadings plot of PC1 versus PC2. Each data point represents one bucket (with the chemical shift indicated explicitly). Several significant metabolites are indicated explicitly. The model consists of 3 PCs and represents data from 18 samples. Regions containing water, lactate and glutamate/glutamine signals were excluded from the PCA.