Discordance between protein and transcript levels detected by selected reaction monitoring

Abstract

Expression levels between transcript and protein are not always correlated. In the present study, the abundance of protein PDR9/ABCG37 in 3 Arabidopsis pdr9/abcg37 mutant alleles was evaluated using selected reaction monitoring analysis. The results showed that protein and mRNA expression levels were similar in 2 mutant alleles. The mRNA expression levels in another mutant, determined by both semi-quantitative and quantitative RT-PCR, were similar to the wild-type, although the abundance of protein was about half the abundance of the wild-type. These results suggested that using only mRNA expression levels to infer protein abundance, compare mutants or responses to various stimuli may lead to incorrect interpretation and conclusions.

Abbreviations

Fe

iron

iTRAQ

isobaric tags for relative and absolute quantification

LC/MS

high performance liquid chromatography coupled with mass spectrometry

SRM

selected reaction monitoring

Introduction

During the last decade, several quantitative proteomic approaches using high performance liquid chromatography coupled with mass spectrometry (LC/MS) have been developed. These methods can be classified into two groups: label-free and labeling approaches.¹ Label-free approaches such as spectral counting² and exponentially modified protein abundance index (emPAI),³ quantify proteins by counting the numbers of fragmented spectra in MS/MS analysis. These methods have some advantages for estimating the relative abundance of proteins in samples since there is no limit on the number of samples used and they have lower costs than labeling approaches. However, it is sometimes difficult to obtain high-accuracy data, because samples to be compared are separately analyzed one by one by LC/MS. Reproducibility is highly dependent on the conditions of the analysis, such as the LC separation column used for peptides and the temperature. Furthermore, labeling approaches such as isobaric tags for relative and absolute quantification (iTRAQ), tandem mass tags (TMT) and stable isotope labeling by/with amino acids in cell culture (SILAC) enable comparison of two or more samples by quantifying proteins by one-shot analysis.⁴ The iTRAQ approach, specifically, has been adapted to compare and quantify proteins in various plant species.^5-10 Digested peptides in each protein sample are labeled with different iTRAQ reagents and samples are then pooled. Therefore, samples can be analyzed by LC/MS at the same time. However, labeling approaches for relative quantification also have disadvantages. For example, proteins containing peptides with 100% matched amino acid sequences are not distinguishable from each other, and sometimes certain peptides not be efficiently labeled. Therefore, label-free and labeling approaches have been mostly used to find marker proteins but not for absolute protein quantification.

Selected reaction monitoring (SRM), also called multiple reaction monitoring, is a mass spectrometry technique that has been recently developed to quantify the absolute abundance of proteins by quantitative proteomics.¹ SRM analysis can provide absolute abundances of target proteins in cells or samples using synthesized peptides with stable isotope labeling as an internal control. In addition, SRM analysis provides high accurate protein abundances among homologous genes with high sequence similarity.^11,12

Using iTRAQ approach we have recently found that the abundance of PDR9/ABCG37 in Arabidopsis roots increased due to iron (Fe) deficiency.¹³ Here, I confirmed that this increase in PDR9/ABCG37 occurred only with Fe deficiency and not with other metal deficiencies by SRM analysis. In addition, mRNA expression levels and the absolute abundance of PDR9/ABCG37 were analyzed in 3 allelic mutants. The mRNA expression levels of PDR9/ABCG37 were not detected in two of the mutant alleles, which was consistent with SRM results. However, the abundance of protein in one of the mutant alleles was approximately half of the wild-type allele, although mRNA expression levels were similar to wild-type. These results indicate that monitoring of both protein and transcript levels in mutant helps correctly understanding their phenotypes and cellular reactions.

Results and Discussion

Comparison of quantitative values between protein and transcript expression levels

Previous iTRAQ analysis showed that the abundance of PDR9/ABCG37 (At3g53480) in microsomal fractions of Arabidopsis roots had a 2.4-fold increase with Fe deficiency but did not respond to zinc (Zn) and manganese (Mn) deficiencies (Table 1).¹³ However, if the digested peptide sequence from the target protein has exactly the same amino acid sequence as homologous proteins, it is impossible to determine their origin. In case of the amino acid sequence of PDR9/ABCG37, it has 79% identity to the sequence of PDR5/ABCG33 (At2g37280). Although iTRAQ is a powerful method for discovering interesting proteins, protein quantification may not be accurate. In the present study, the absolute abundances of PDR9/ABCG37 were determined by the SRM analysis using LC/MS. The flowcharts of iTRAQ and SRM analyses are shown in Fig. 1. Among the trypsin-digested peptides assigned to PDR9/ABCG37 by iTRAQ and shot-gun analysis, two unique peptides (PDR9pep1: “LAPEIDATTK” and PDR9pep 2: “HIIEYFESVPEIPK”) were selected and synthesized with stable isotope labeling. The peptides selected for SRM analysis have unique sequences to the target protein and showed high intensities with LC/MS analysis.^14,15 Microsomal fractions prepared from Arabidopsis roots grown on basal, 0-Zn, 0-Mg, and 0-Fe deficient media for 10 days and digested by trypsin with the synthesized peptides (Fig. 1). Then, trypsin-digested peptides were analyzed using LC/MS after peptide purification by C18 resin. In LC/MS analysis, PDR9pep1 was identified at a higher intensity than PDR9pep2 and the intensity peak of PDR9pep2 was not enough for absolute quantification (data not shown). Therefore, PDR9pep1 was used as internal control in SRM analysis. The synthesized peptides were eluted from the separation column with endogenous targeted peptides in LC using the same retention time, prior to the introduction into the MS. In MS, they were separately detected with endogenous targeted peptides because the isotope labeled peptide is 8 Da heavier than the endogenous peptide. The absolute abundance of endogenous peptides was determined using the peak area of known concentration of isotope labeled peptides (Fig. 1). The abundance of PDR9/ABCG37 had a 1.9-fold increase as a response to Fe deficiency (Fig. 2). To compare these results with previous iTRAQ data, the relative values of the SRM analysis were calculated and compared to basal values (Table 1). Results showed that the SRM data followed the same trend as the iTRAQ data, although the relative abundance of PDR9/ABCG37 response to Fe deficiency was overestimated in the iTRAQ analysis. Furthermore, quantitative RT-PCR analysis showed that the PDR9/ABCG37 transcript level also increased with Fe deficiency (Table 1). These results indicate that SRM analysis is a useful method for estimating endogenous protein abundances.

Table 1.

Relative values of SRM and iTRAQ analysis as a response to Fe deficiency Absolute protein abundances and mRNA expression levels provided by SRM analyses or qRT-PCR were respectively calculated in relation to the basal medium. iTRAQ values refer to a previous report.¹³

Method	0-Fe/basal
iTRAQ (Zargar et al. 2014)	2.387 ± 0.153**
SRM (This study)	1.892 ± 0.085**
qRT-PCR (This study)	2.837 ± 0.527**

Data are means ± SD of 8 and 3 biological replicates for iTRAQ and SRM analysis, respectively. Data are means ± SD of two independent biological replicates with two technical replicates each for qRT-PCR analysis. ** indicates P < 0.01.

Figure 1.

Workflows of iTRAQ analysis and SRM analysis. In iTRAQ analysis, each digested protein sample was labeled with four different iTRAQ reagents {i.e. 114 (red), 115 (blue), 116 (green) or 117 (orange)}. iTRAQ labeled peptides were combined and fractionated by strong cation exchange (SCX) into 7 fractions. Then, desalted iTRAQ labeled peptides using C18 resin were analyzed by mass spectrometry (LTQ-Orbitrap XL).¹³ Protein quantification and identification were performed using reporter ions from each iTRAQ reagents and fragmented peptides, respectively. The identified proteins with relative expression values were provided using MASCOT software. In SRM analysis, proteins including the target protein (containing blue peptide), were digested by trypsin with isotope labeled peptides (red) as internal controls. Digested peptides were analyzed by mass spectrometry (TSQ-Vantage) after peptide desalting. The absolute abundance of endogenous protein was calculated from the peak area of internal control using PinPoint software.

Figure 2.

Absolute quantifications of PDR9/ABCG37 protein in wild-type roots grown on basal, Zn, Mn or Fe deficient medium. Absolute abundances of PDR9/ABCG37 protein in the roots of Col-0 grown on basal, 0-Zn, 0-Mn or 0-Fe medium were quantified using synthesized PDR9/ABCG37 unique peptides with a stable isotope label as an internal control (means ± SD).

Determination of expression levels in mutant lines by SRM analysis

The absolute abundance of PDR9/ABCG37 in 3 Arabidopsis PDR9/ABCG37 T-DNA insertion mutant lines, pdr9/abcg37-1 (SALK_050885), pdr9/abcg37-2 (SALK_035704), and pdr9/abcg37-3 (SALK_063955), was determined (Fig. 3A). Semi-quantitative RT-PCR analysis of the PDR9/ABCG37 gene revealed that PDR9/ABCG37 transcripts in pdr9/abcg37-1 and pdr9/abcg37-3 mutants were not detectable, although shorter transcripts were slightly detected in the pdr9/abcg37-3 mutant (Fig. 3B). Furthermore, PDR9/ABCG37 transcripts did not significantly decrease in the pdr9/abcg37-2 mutant (Fig. 3B). mRNA expression levels were also examined in 3 mutant lines by quantitative RT-PCR (Fig. 3C). Results were consisted with the semi-quantitative RT-PCR, suggesting that pdr9/abcg37-1 and pdr9/abcg37-3 mutants are likely null alleles of PDR9/ABCG37. In addition, the pdr9/abcg37-2 mutant might have no effect on the PDR9/ABCG37 protein expression.

Figure 3.

PDR9/ABCG37 expressions at mRNA and protein levels in pdr9/abcg37 mutants. (A) Schematic of the T-DNA insertion positions and primers for semi-quantitative or quantitative RT-PCR on PDR9/ABCG37 genome sequence. (B) For semi-quantitative RT-PCR, PDR9/ABCG37 and Tubulin4 (TUB4) transcripts were amplified by PCR in Col-0, pdr9/abcg37-1, pdr9/abcg37-2, and pdr9/abcg37-3 roots grown on basal medium. PDR9/ABCG37 transcript was amplified using primer F1 and R1 for 28 PCR cycles and with primer F1 and R2 for 35 PCR cycles. TUB4 transcript was amplified by 28 PCR cycles. (C) For quantitative RT-PCR, PDR9/ABCG37 transcripts in Col-0, pdr9/abcg37-1, pdr9/abcg37-2, and pdr9/abcg37-3 were amplified using primer F and R (means ± SD). Expression in Col-0 roots was adjusted to 1 relative unit. (D) Absolute abundances of PDR9/ABCG37 protein in the roots of Col-0, pdr9/abcg37-1, pdr9/abcg37-2, and pdr9/abcg37-3 mutants were quantified using synthesized unique peptides with a stable isotope label in the amino acid sequence of PDR9/ABCG37 (means ± SD).

To analyze these mutants at the protein expression level, the absolute abundance of PDR9/ABCG37 protein was measured in each mutant line by SRM analysis (Fig. 3D). The abundance of protein significantly decreased in pdr9/abcg37-1 and pdr9/abcg37-3 mutant lines, and decreased to approximately half of the abundance of the PDR9/ABCG37 protein in the pdr9/abcg37-2 mutant line. These results clearly showed that expression levels are not always consistent between mRNA and SRM analysis. Overall, it can be concluded that the SRM analysis is a valuable method to obtain highly accurate values of protein expression levels. The accurate quantification of a target protein in a mutant is important to explain and understand phenotypes and cellular reactions.¹²

Low correlation between transcript and protein expression levels

Sometimes, transcript and protein expression levels are not positively correlated.^16,17 Böhmer and Schroeder compared transcriptome and proteome changes in Arabidopsis suspension cells in response to the phytohormone abscisic acid using DNA microarray and iTRAQ analysis.¹⁸ As the result, transcriptome and proteome changes showed poor correlation. High increases or decreases in transcript levels were not coincident with changes at the protein expression level. Correlation analysis between proteome and transcriptome changes in Arabidopsis roots as a response to phosphate deficiency, showed that some down-regulated proteins did not significantly change at the transcript level, although most proteins showed good correlations.⁸ In this study, changes in PDR9/ABCG37 expression levels between transcript and protein were not correlated in a mutant line (Fig. 3). These differences could be caused by protein accumulation due to posttranslational regulation or due to the time lag between the mRNA and protein expression profiles, affecting circadian rhythmicity, in addition to technical errors. Therefore, it is important to evaluate the expression levels of genes of interest at the protein level too.

Immunoblotting analysis has been used, as alternative approach, to evaluate protein expression levels. However, some protein isoforms are not individually detectable because they have high sequence homology and it is, therefore, difficult to make specific antibodies. Recently, to measure high accurate protein abundances, SRM analysis has been applied to the field of plant science.^19-22 Lehmann and colleagues have successfully quantified the abundance of four sucrose-phosphate synthase isoforms by SRM analysis, although no antibodies could separately detect the different isoforms.²³ These authors showed that this method was highly sensitive since each isoform was detectable at about 10 fmol (approximately 1.2 ng). Data in this study also showed that SRM analysis could measure less than 1 fmol PDR9/ABCG37 per 1 μg microsomal fraction (Figs. 2 and 3D). Therefore, SRM analysis enables individual quantification of not only homologous proteins but also proteins in very low abundances.⁶

Conclusions

The success and failure of protein identification methods is depend on the accuracy of the protein sequence database. Recently, it has been possible to perform proteomic studies in various non-model organisms because of developments in genome sequence technology such as next-generation sequencing. Nevertheless, protein identification in model organisms still has advantages because of the existence of more accurate amino acid sequences. On the other hand, SRM analysis can be applied to non-model plants,^6,11,24 if the targeted peptides are selected. This indicates that complete genome sequences of the organisms of interest are not required. SRM analysis as the advantage of being a high-sensitivity method in quantitative proteomics, because only targeted peptides are intensively analyzed during MS analysis. Furthermore, SRM analysis allows the quantification of absolute abundances of posttranslational modifications such as phosphorylation states.^25-27 Although less common in the field of plant science, this method also has the potential to be used for other purposes, such as the quantification of stoichiometric changes in protein complexes dependent on environmental stimuli, or the quantification of developmental stages.

Materials and Methods

Growth condition of Arabidopsis seedlings

A. thaliana ecotype Columbia (Col-0), pdr9-1, pdr9-2, pdr9-3 mutants were sterilized and germinated on MGRL (basal) medium with/ without Zn, Mn or Fe (0-Zn, 0-Mn or 0-Fe) containing 1.0% (w/v) sucrose and 1.2% purified agar (Nacalai Tesque, Kyoto, Japan) (Zargar, et al. 2014). Seedlings were grown vertically for 10 days or 4 days and then transplanted at 22°C under 16-h light/ 8-h dark conditions. T-DNA insertion mutants SALK_050885 (pdr9-1), SALK_035704 (pdr9-2) and SALK_063955 (pdr9-3) were obtained from the Arabidopsis Biological Resource Center at Ohio State University.

RNA extraction and quantitative real time (RT)-PCR

Total RNA was extracted from 10-day-old roots using RNAeasy Plant Mini kit (QIAGEN, CA, USA). RNA was reverse-transcribed into cDNA using ReverTra Ace (Toyobo) according to the manufacturer's recommendations. PDR9/ABCG37 (At3g53480) and TUB4 (At5g44340) were amplified by PCR using the following primers for semi-quantitative RT-PCR: 5′-CACCATGGCTCATATGGTTGGAGC-3′ (sense; F1) and 5′-TCCTGGAGAAAGTCTGCAAC-3′ (antisense; R1) or 5′-TCTTCGTTGGAAGTTGAGTTTG-3′ (antisense; R2) for PDR9/ABCG37 and 5′-GAGGGAGCCATTGACAACATCTT-3′ (sense) and 5′-GCGAACAGTTCACAGCTATGTTCA-3′ (antisense) for TUB4. PDR9/ABCG37 was amplified in 25 cycles using F1 and R1 and in 30 cycles using F1 and R2. TUB4 was amplified in 25 cycles. Quantitative RT-PCR was performed using LightCycler®480 SYBR Green I Master (Roche) in a total volume of 20 μL, including 12.5 ng of cDNA. The following primers were used: 5′- TCAGATATCACAGGAGCTTTCCGTC-3′ (sense; F1) and 5′- TTCACAGTAGCCTGAGACTCTAGC-3′ (antisense; R1) for PDR9/ABCG37 and 5′- CCCAAAAGCCAACAGAGAGAA-3′ (sense) and 5′- ACACACCATCACCAGAGTCCA-3′ (antisense) for ACT8. The samples were normalized first to ACT8, and then the relative target gene expression was determined by performing comparative ΔΔC_T. Two independent biological replicates with 2 technical replicates each. Statistical analysis was conducted using Student's t-test.

Absolute quantification by the SRM analysis

Col-0 and pdr9/abcg37 mutants were vertically grown on basal medium or Zn0 medium for 10 days at 22°C under 16-h light/8-h dark conditions. In total, 1,200 seeds were grown on basal, 0-Zn, 0-Mn or 0-Fe medium. Roots (∼0.4-g fresh weight) were homogenized with buffer A (50 mM HEPES-KOH, pH 7.5, 5 mM EDTA, 400 mM sucrose, protease inhibitor cocktail). The homogenates were centrifuged at 1,000 × g at 4°C for 20 min, and the supernatants were centrifuged at 8,000 × g at 4°C for 20 min. The supernatants were centrifuged at 100,000 × g at 4°C for 60 min to prepare the microsomal fractions. The pellets were washed by buffer A twice under the same condition and then dissolved in 30 μl of iTRAQ dissolution buffer (Applied Biosystems, CA, USA). The protein concentration was determined using NanoDrop (Thermo Scientific, Germany).

Aliquots of 200 μg of microsomal protein fraction were digested with 10 μl of trypsin (1 mg ml⁻¹) with 20 μl of stable isotope-labeled peptide (5 pmol μl⁻¹) following reduction by dithiothreitol and alkalization by iodoacetamide. The unique peptides of PDR9/ABCG37 (PDR9pep1: “LAPEIDATTK” and PDR9pep2: “HIIEYFESVPEIPK”), which were confirmed by BLAST search, were synthesized with a stable isotope at C-terminal K (¹³C6 and¹⁵N2). The digested peptides were analyzed using TSQ Vantage-HTS-PAL-Advance LC system following purification by Sep-Pak, a C-18 resin (Waters, MA, USA). The peptides were loaded onto the column (100-μm internal diameter, 15 cm, L-Column; CERI, Saitama, Japan) using a Paradigm MS4 HPLC pump (Michrom BioResources, Auburn, CA, USA) and an HTC-PAL autosampler (CTC Analytics, Switzerland). Buffers were 0.1% (v/v) acetic acid in water (A) and 100% (v/v) acetonitrile in water (B). A linear gradient from 5% to 45% B for 20 min was applied, and peptides eluted from the column were introduced directly into a TSQ Vantage mass spectrometer (Thermo Scientific, Germany) at a flow rate of 500 nl min⁻¹ and a spray voltage of 2.0 kV. For the absolute concentrations of PDR9/ABCG37, the peptide transitions PDR9pep1 {precursor ion (m/z): 529.8 and product ions (m/z): 349.2, 420.2, 535.3, 648.4, 777.4, 874.5, 945.5, 1058.6} were used and 3 biological replicates were calculated from the peak areas of obtained spectra using PinPoint 1.0 (Thermo Scientific, Germany). Statistical analysis was conducted using Student's t-test.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

I thank to Rie Kurata and Mami Kobayashi (NAIST) for help with the SRM and qRT-PCR analysis, respectively.

Funding

This work was supported by a Grant-in-Aid for Scientific Research on Innovative Areas (No. 23119512 to Y.F.) from the Ministry of Education, Culture, Sports, Science and Technology of Japan; a Grant-in-Aid for Scientific Research from Nara Institute of Science and Technology supported by The Ministry of Education, Culture, Sports, Science and Technology, Japan.