Quantitative trait loci mapping of the mouse plasma proteome (pQTL)

Abstract

A current challenge in the era of genome-wide studies is to determine the responsible genes and mechanisms underlying newly identified loci. Screening of the plasma proteome by high-throughput mass spectrometry (MALDI-TOF MS) is considered a promising approach for identification of metabolic and disease processes. Therefore, plasma proteome screening might be particularly useful for identifying responsible genes when combined with analysis of variation in the genome. Here, we describe a proteomic quantitative trait locus (pQTL) study of plasma proteome screens in an F(2) intercross of 455 mice mapped with 177 genetic markers across the genome. A total of 69 of 176 peptides revealed significant LOD scores (≥5.35) demonstrating strong genetic regulation of distinct components of the plasma proteome. Analyses were confirmed by mechanistic studies and MALDI-TOF/TOF, liquid chromatography-tandem mass spectrometry (LC-MS/MS) analyses of the two strongest pQTLs: A pQTL for mass-to-charge ratio (m/z) 3494 (LOD 24.9, D11Mit151) was identified as the N-terminal 35 amino acids of hemoglobin subunit A (Hba) and caused by genetic variation in Hba. Another pQTL for m/z 8713 (LOD 36.4; D1Mit111) was caused by variation in apolipoprotein A2 (Apoa2) and cosegregated with HDL cholesterol. Taken together, we show that genome-wide plasma proteome profiling in combination with genome-wide genetic screening aids in the identification of causal genetic variants affecting abundance of plasma proteins.

Figure 1

Experimental overview of pQTL analysis. (A) Schematic of the intercross between B6.Ldlr^−/− and FVB.Ldlr^−/− mice. Genome-wide scans and proteomic screens were obtained from 455 F₂ mice. Examples of proteomic screens acquired by MALDI-TOF mass spectrometry of (B) the parental strains, (C) intensity differences between the parental strains (FVB minus B6) with maximal absolute differences at m/z 8713 and m/z 3494, (D) the F₁ generation, and (E) two representative animals of the F₂ generation.

Figure 2

Heatmaps visualizing pQTLs. Results are shown for (A) AUC-based (176 phenotypes) and (B) m/z-based (1841 phenotypes) analysis. Heatmaps are visualizing significant pQTLs with their chromosomal position and m/z. Color code indicates the magnitude of calculated LOD scores. M/z 8713 and m/z 3494 were followed up in functional studies.

Figure 3

Genetic effects of chromosome 11 on m/z 3494, hemoglobin and red blood cell parameters. (A) LOD-score plot of m/z 3494 before (closed line) and after (dotted line) fine mapping with a marker at Hba (red line). (B) Genotypic effects of Hba SNP on MALDI-TOF spectra in the F₂ (means of B6.Ldlr^−/−, red; FVB.Ldlr^−/−, blue; and heterozygote, black). (C) MALDI-TOF spectra in the F₀. (D) Free hemoglobin in plasma. Blood count of B6.Ldlr^−/− (n = 17) and FVB.Ldlr^−/− (n = 16): (E) hemoglobin, (F) red blood cells (RBC), (G) hematocrit (HCT), and (H) mean corpuscular volume (MCV). (I) Haptoglobin in plasma of B6.Ldlr^−/− (n = 6) and FVB.Ldlr^−/− (n = 6) mice. (J) Osmotic resistance of B6.Ldlr^−/− and FVB.Ldlr^−/− RBC to low-salt concentrations. (K) Sequence comparison of duplicated Hba genes in B6.Ldlr^−/− and FVB.Ldlr^−/−. (L) Ribbon diagram of an Hba chain (Protein DataBank identification: 3HRW). The three polymorphic sites mapped close to the heme-binding site. This figure was created using the ICM-browser (Internal Coordinate Mechanisms, www.molsoft.com). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4

Genetic effects of chromosome 1 on m/z 8713, HDL and Apoa2. (A) LOD-score plot of m/z 8713 and physical position of Apoa2 (red line). (B) Genotypic effects of D1Mit111 on MALDI-TOF spectra in the F₂ (means of B6.Ldlr^−/−, red; FVB.Ldlr^−/−, blue; and heterozygote, black). (C) MALDI-TOF spectra (m/z region 8800–8900) in the F₀. Note the 10-Da mass shift between B6.Ldlr^−/− and FVB.Ldlr^−/−, which is also present in B. (D) LOD-score plot for HDL cholesterol. (E) Genotypic effects of D1Mit111 on HDL cholesterol in the F₂ and (F) in the parentals. (G) LOD-score plot for Apoa2 mRNA. (H) Genotypic effects of D1Mit111 on Apoa2 mRNA in the F₂ and (I) in the parentals. (J) MALDI-TOF spectra of isolated HDL. (K) Peak intensity at m/z 8713 in HDL-depleted plasma (broken line) and isolated HDL (dotted line). Fractions were obtained from pooled plasma of FVB mice that also served as reference (solid line). (L–N) Analysis of isolated Apoa2 from HDL of B6.Ldlr^−/− and FVB.Ldlr^−/− mice by LC-MS/MS coupled with Orbitrap Velos confirming expected sequence variants. *P < 0.05, ***P < 0.001.