Improving the Identification and Coverage of Plant Transmembrane Proteins in Medicago Using Bottom-Up Proteomics

Abstract

Plant transmembrane proteins (TMPs) are essential for normal cellular homeostasis, nutrient exchange, and responses to environmental cues. Commonly used bottom-up proteomic approaches fail to identify a broad coverage of peptide fragments derived from TMPs. Here, we used mass spectrometry (MS) to compare the effectiveness of two solubilization and protein cleavage methods to identify shoot-derived TMPs from the legume Medicago. We compared a urea solubilization, trypsin Lys-C (UR-TLC) cleavage method to a formic acid solubilization, cyanogen bromide and trypsin Lys-C (FA-CTLC) cleavage method. We assessed the effectiveness of these methods by (i) comparing total protein identifications, (ii) determining how many TMPs were identified, and (iii) defining how many peptides incorporate all, or part, of transmembrane domains (TMD) sequences. The results show that the FA-CTLC method identified nine-fold more TMDs, and enriched more hydrophobic TMPs than the UR-TLC method. FA-CTLC identified more TMPs, particularly transporters, whereas UR-TLC preferentially identified TMPs with one TMD, particularly signaling proteins. The results suggest that combining plant membrane purification techniques with both the FA-CTLC and UR-TLC methods will achieve a more complete identification and coverage of TMPs.

Keywords: Medicago truncatula; TRAMDOMI algorithm; detergent-free purification; liquid chromatography; mass spectrometry; transmembrane domain; transmembrane protein.

FIGURE 1

Flow chart summarizing the two solubilization and protein cleavage methodologies used in this study. Red box: Negligible insoluble material remained after using the formic acid solubilization, cyanogen bromide, and trypsin Lys-C (FA-CTLC) method (2a, blue arrow), whereas significant insoluble material remained after using the urea solidilization, trypsin Lys-C (UR-TLC) method (2b, red arrow). The remaining insoluble material after UR-TLC was solubilized and digested using the FA-CTLC method. Mass spectrometry (MS) analysis showed the presence of a wide range of proteins in the pellet (Supplementary Table 2). Subsequent centrifugation of this re-solubilized and FA-CTLC-treated material showed that negligible insoluble material remained (2c, green arrow).

FIGURE 2

Analysis of the proteins identified by MS-MS after UR-TLC or FA-CTLC. (A) The reproducibility of protein identifications between the three biological repeats following UR-TLC treatment. (B) Reproducibility of protein identifications between the three biological repeats following FA-CTLC treatment. (C) A comparison of proteins identified by each treatment. The schematic diagrams were made by the Venny online tool (http://bioinfogp.cnb.csic.es/tools/venny/).

FIGURE 3

Distribution of proteins identified after applying the FA-CTLC or UR-TLC methods to Medicago microsomal membrane (MM) preparations from three biological repeats based on transmembrane domain (TMD) number and grand average of hydropathy (GRAVY) score. (A) The proteins identified after analyzing the UR-TLC- or FA-CTLC-treated samples from three biological repeats were submitted to the TMHMM server. The total predicted number of transmembrane protein (TMP) groups in the UR-TLC and FA-CTLC samples was 2,817 (35.45%) and 2,784 (39.11%), respectively. The predicted TMD distribution of the Medicago proteins in the UniProt database is shown in the inset panel as a comparison. There were 23.26% proteins predicted to be TMPs. *p ≤ 0.05, **p ≤ 0.01 (two-tail Student’s t–test). Error bars = standard error, n = 3. (B) The GRAVY scores were calculated (Kyte and Doolittle, 1982) from the proteins identified after analyzing the UR-TLC or FA-CTLC base on the previous published literature. Approximately 20% of proteins identified by FA-CTLC displayed a GRAVY score greater than zero and 17% of proteins identified by UR-TLC. Proteins with a hydrophobicity scores above 0 are more likely to be TMPs.

FIGURE 4

The FA-CTLC method preferentially identifies peptides with TMD motifs. (A) A comparison of the total TMDs identified numbers from both purification methods. The FA-CTLC methods identified 9.3-fold more TMDs than the UR-TLC method. (B,C) The peptides identified in the MTR_7g005910 transporter of Medicago after using the FA-CTLC (B) or UR-TLC method (C). The peptides that were identified by MS were colored in black. The schematic diagrams were made using the Protter online tool (Omasits et al., 2014) (http://wlab.ethz.ch/protter/start/).

FIGURE 5

Functional analysis of the proteins preferentially identified using either the FA-CTLC or UR-TLC methods. (A) The proteins preferentially identified after using FA-CTLC (i.e., with >5 TMDs) were predominantly transporters. (B) The proteins preferentially identified after using UR-TLC (i.e., with one TMD) were predominantly signaling proteins. The category in which the FA-CTLC method had significant difference with p-value < 0.05 was labeled with *.

FIGURE 6

The predicted subcellular location of the proteins identified after using FA-CTLC or UR-TLC. (A,B) Between 73 and 75% of the proteins identified in the MM preparations were cytoplasmic proteins. Of the proteins identified to be TMPs, there was no significant difference in the identity of the proteins predicted to reside in the nuclear, chloroplast, or mitochondrial membranes. A t-test confirmed a significance difference (p < 0.05, n = 3) between the two methods in identifying proteins where the subcellular location could not be assigned (the “unassigned” category).