Download Selected Reaction Monitoring (SRM) to determine protein

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

SR protein wikipedia , lookup

Gene expression wikipedia , lookup

G protein–coupled receptor wikipedia , lookup

Ancestral sequence reconstruction wikipedia , lookup

Proteasome wikipedia , lookup

List of types of proteins wikipedia , lookup

Protein (nutrient) wikipedia , lookup

Expression vector wikipedia , lookup

Magnesium transporter wikipedia , lookup

Protein wikipedia , lookup

Bottromycin wikipedia , lookup

Protein structure prediction wikipedia , lookup

Peptide synthesis wikipedia , lookup

Metalloprotein wikipedia , lookup

Intrinsically disordered proteins wikipedia , lookup

Interactome wikipedia , lookup

Protein moonlighting wikipedia , lookup

Nuclear magnetic resonance spectroscopy of proteins wikipedia , lookup

Protein adsorption wikipedia , lookup

Western blot wikipedia , lookup

QPNC-PAGE wikipedia , lookup

Two-hybrid screening wikipedia , lookup

Protein–protein interaction wikipedia , lookup

Cell-penetrating peptide wikipedia , lookup

Proteomics wikipedia , lookup

Ribosomally synthesized and post-translationally modified peptides wikipedia , lookup

Self-assembling peptide wikipedia , lookup

Protein mass spectrometry wikipedia , lookup

Transcript
Plant Physiology Preview. Published on December 2, 2013, as DOI:10.1104/pp.113.225524
Running Head: SRM to determine protein abundance in Arabidopsis
CORRESPONDING AUTHOR: A. Harvey Millar
ARC Centre of Excellence in Plant Energy Biology and Centre for Comparative
Analysis of Biomolecular Networks, The University of Western Australia (M316), 35
Stirling Highway , Crawley, WA 6009, Australia
Tel: +61 8 64887245
Fax: +61 8 64884401
E-mail: [email protected]
Research Area : Breakthrough Technologies
Downloaded from on June 17, 20171- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Copyright 2013 by the American Society of Plant Biologists
Selected Reaction Monitoring (SRM) to determine protein abundance in
Arabidopsis using the Arabidopsis Proteotypic Predictor (APP).
Nicolas L. Taylor1,2, Ricarda Fenske1,2, Ian Castleden3, Tiago Tomaz1,2, Clark J.
Nelson1,2, and A. Harvey Millar1,2
1
ARC Centre of Excellence in Plant Energy Biology and 2Centre for Comparative
Analysis of Biomolecular Networks (CABiN), Centre of Excellence in Computational
Systems Biology, Bayliss Building M316, The University of Western Australia, 35
Stirling Highway, Crawley WA 6009, Western Australia, Australia.
One-Sentence Summary: The Arabidopsis Proteotypic Predictor (APP) and SRM
mass spectrometry enable the quantitation of protein abundance in knockout and
complemented lines of Arabidopsis when antibodies are unavailable.
Downloaded from on June 17, 20172- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
This work was supported by the Australian Research Council (ARC) ARC Centre of
Excellence for Plant Energy Biology (CE0561495). AHM is supported by the
Australian Research Council (ARC) as an ARC Future Fellow.
Downloaded from on June 17, 20173- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
ABSTRACT
In reverse genetic knockout (KO) studies that aim to assign function to specific genes,
confirming the reduction in abundance of the encoded protein will often aid the link
between genotype and phenotype. However, measuring specific protein abundance is
particularly difficult in plant research where only a limited number of antibodies are
available. This problem is enhanced when studying gene families or different proteins
derived from the same gene (isoforms), as many antibodies cross-react with more than
one protein. We show that utilizing Selected Reaction Monitoring (SRM) mass
spectrometry allows researchers to confirm protein abundance in mutant lines, even
when discrimination between very similar proteins is needed. Selecting the best
peptides for SRM analysis to ensure protein or gene-specific information can be
obtained requires a series of steps, aids and interpretation. To enable this process in
Arabidopsis we have built a web-based tool, the Arabidopsis Proteotypic Predictor to
select candidate SRM transitions when no previous mass spectrometry evidence
exists. We also provide an in depth analysis of the theoretical Arabidopsis proteome
and its use in selecting candidate SRM peptides to establish assays for use in
determining protein abundance. To test the effectiveness of SRM mass spectrometry
in determining protein abundance in mutant lines we selected two enzymes with
multiple isoforms, aconitase and malate dehydrogenase. Selected peptides were
quantified to estimate the abundance of each of the two mitochondrial isoforms in
wild type, KO, double KO and complemented plant lines. We show that SRM protein
analysis is a sensitive and rapid approach to quantify protein abundance differences in
Arabidopsis for specific and highly related enzyme isoforms.
Downloaded from on June 17, 20174- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
INTRODUCTION
Upon the completion of the Arabidopsis thaliana genome in 2000 (AGI, 2000) only
~10% of the 25500 initially predicted genes had an experimentally assigned function
(Alonso and Ecker, 2006). In the past decade, forward and reverse genetics have
provided the basis to experimentally verify the functions of 9663 (35%) of the 27416
genes (The Multinational Coordinated Arabidopsis thaliana Functional Genomic
Project Annual Report 2011). Key to this advance has been the large collections of TDNA insertional mutants (O'Malley and Ecker, 2010). Researchers generally aim to
obtain a knockout of gene expression or a severe reduction in the gene’s function by
T-DNA insertion (O'Malley and Ecker, 2010). However, an examination of published
literature on over 1084 T-DNA insertion mutants, only 44% of insertions resulted in
no transcript, while 42% resulted in reduced transcript abundance and 14% showed no
change or an increase in transcript abundance (Wang, 2008). Of these studies, only
136 reported information on abundance of the protein encoded by the gene of interest
and of these 80% showed no protein expression while one in five showed either no
effect on or reduced protein abundance or the presence of a truncated protein product
(Wang, 2008). As 20% of gene functions assigned to phenotypes are assumed to
result from protein knockouts that do not occur but instead result from other changes
in protein abundance, it has become an increasing requirement to determine the
abundance of the gene’s product in these genetically altered lines.
The classical means of assessing protein abundance has been quantitative western
blotting using antibodies raised to the specific polypeptide of interest (Aebersold et
al., 2013). Polyclonal antibodies typically recognize a series of primary or secondary
structures of a polypeptide and while they can be highly sensitive in immunodetection assays, they can also cross-react with multiple proteins due to common
epitopes. In most assays it is not possible to know which epitopes are responsible for
the immuno-reactivity observed. Monoclonal antibodies recognize a single primary
sequence or secondary structure of a polypeptide, providing greater clarity in the
immuno-reaction, but at a significantly greater cost and require immune-reactivity of
each peptide in an animal model.
Selected reaction monitoring (SRM) mass spectrometry provides an alternative
approach that allows researchers to target their protein of interest in a complex
Downloaded from on June 17, 20175- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
mixture and estimate its abundance by quantitation of peptides derived from its
enzymatic digestion. In this approach a triple quadropole (Q or q) mass spectrometer
(QqQ) transmits a peptide precursor ion in Q1 that is then fragmented in q2 and a
single peptide fragment ion is selected in Q3 for quantitation. The precursor and
fragment ion pair is referred to as a transition and its abundance is used to quantify
the abundance of a specific peptide derived from a protein. During an SRM
experiment, sequential gating of precursor and product in a triple quadrupole mass
spectrometer (QqQ-MS) allows millions of precursor/fragment ion combinations
(transitions) to be assessed in complex peptide mixtures generated by proteolysis of
protein extracts from cells. To further confirm the origin of the quantifier product ion,
two qualifier ions are also examined from the same precursor ion. This combination
of filters gives SRM approaches their power in complex samples and allows
quantification of many different proteins over 4 orders of magnitude in crude whole
protein extracts from plant tissue samples (Picotti and Aebersold, 2012). SRM mass
spectrometry, also referred to as Mass Westerns, has previously been used in plants to
quantify a number of proteins including: sucrose-phosphate synthase isoforms in
Arabidopsis (Lehmann et al., 2008), sucrose synthase isoforms and N-metabolism
enzymes in Medicago (Wienkoop et al., 2008), a basic amino acid carrier involved in
arginine metabolism in rice (Taylor et al., 2010), cytosolic and organelle markers in
Arabidopsis (Ito et al., 2011) and the plasma membrane transportome in Arabidopsis
(Monneuse et al., 2011). Label-free quantitation in this manner requires
reproducibility in sample extraction, digestion, liquid chromatography and ionisation
and has been widely reviewed (Lange et al., 2008, Picotti and Aebersold 2012).
To assist Arabidopsis researchers to design SRM assays, Fan et al. (2012) calculated
transitions for Arabidopsis experimental protein spectra submitted to EBI’s PRIDE
database. These candidate transitions are made available through the web-based tool
MRMaid (www.mrmaid.info, (Fan et al., 2012)) that can be searched by entering an
accession number. Whilst MRMaid provides a valuable resource of transitions for a
subset of Arabidopsis proteins with matching data in PRIDE, the next challenge is to
establish SRM analysis for the other ~20000 Arabidopsis proteins. A key
consideration in the selection of candidate SRM transitions for these ~20000 proteins
is whether they are likely to be detectable by LC-MS, so-called ‘proteotypic’
peptides. MRMaid currently only allows unique peptides that have already been
Downloaded from on June 17, 20176- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
experimentally observed to be selected as candidate transitions, in some cases, such as
limited numbers of available detectable peptides or when studying a group of closely
related proteins, redundant peptides as candidate transitions may in fact be favorable,
desirable or simply unavoidable. This likely to be the case, for example, when
attempting to measure proteins translated from splice variants of a gene, highly
related enzyme isoforms or multiple mature proteins derived by post-translational
processing which may accumulate in different parts of a cell. In these cases, it is
likely that only very small regions of the proteins differ from one another and thus
robust quantitation is likely to require information from a set of unique and nonunique peptides analyses.
In this study we have attempted to systematically address a series of the issues
associated with designing a SRM experiment in Arabidopsis by first carrying out an
in depth analysis of the theoretical in silico trypsin digested Arabidopsis proteome.
We have examined; the number of candidate SRM peptides in each Arabidopsis
protein, the number of times these candidate transitions occur in the Arabidopsis
proteome as both tryptic and non-tryptic fragments, the presence of potential trypsin
missed cleavage sites, the presence of potential oxidised Met and a prediction of
whether the candidate transition is likely to be proteotypic. We have collated these
data
into
a
searchable
web
http://www.plantenergy.uwa.edu.au/APP/)
based
called
database
the
(available
Arabidopsis
at
Proteotypic
Predictor (APP) to assist researchers in the design of SRM experiments in
Arabidopsis. To show the utility of this approach we studied two abundant isoforms
of the TCA cycle, mitochondrial aconitase (mACO1) and malate dehydrogenase
(mMDH1), and two lower abundance isoforms for the same enzymatic steps, mACO2
and mMDH2 (Taylor et al., 2011). We quantified peptides of each of the isoforms, of
each protein, in WT and KO plants, and in the case of mMDH, in dKO and
complemented plants. We use the SRM analysis to demonstrate the relative sensitivity
and reliability of this method to distinguish between similar isoforms of proteins. We
also use SRM analysis to measure changes in protein abundance in genetic knockouts
of each or all isoforms of a protein.
Downloaded from on June 17, 20177- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
RESULTS
Theoretical trypsin digestion of all Arabidopsis proteins to define targets for
SRM mass spectrometry.
In choosing peptides for use in SRM mass spectrometry it is vital to determine
whether they have been observed previously by LC-MS, are theoretically detectable
by LC-MS and their uniqueness within the genome. To assess this the 1618830
redundant set of theoretical trypsin peptides representing 27383 proteins and having a
range of peptides from 1 to 755 per protein (Figure 1A.) generated from an in silico
trypsin digest of the Arabidopsis proteome (Tair10pep) were assessed for how many
times they occurred at a single gene locus (Figure 1B.). 939417 (58%) peptides were
unique as a translated product from a single genome locus, whereas 679413 (42%)
peptides occurred more than once in Arabidopsis.
Of the 939417 unique peptides, 560651 contained a potential trypsin missed cleavage
site, that is they contain an arginine (R) or lysine (K) residue at a position other than
the carboxy-terminus of the peptide. While such peptides are regularly observed in
peptide mass spectrometry, they are not consistently obtained during trypsin digestion
and are thought to be relatively poor candidates for selection as SRM transitions. This
leaves 378766 or 40% of the total theoretical trypsin peptides that contain no missed
cleavage site and are unique, and reduces the total number of unique proteins that
could assessed by unique peptides SRM mass spectrometry approaches to 23777
(87%) (Figure 2A.). If a requirement is made for three peptides per protein to be
amenable for analysis, this would further reduce the number of proteins that could be
assessed by SRM approaches to 21115 (77%).
To determine how these numbers of peptides/proteins may correspond to peptides that
can be detected by LC-MS and thus are proteotypic, we analyzed their occurrence in
two large Arabidopsis data sets, Baerenfaller et al (2008) (106155 peptides) and
Castellana et al (2008) (72195 peptides), and our own data set enriched for peptides
from organellar proteins (9205 peptides). When comparing these sets to the number of
unique theoretical peptides without miss cleavages (378766) we found that 69128
peptides were present in the three datasets, 28776 peptides were only found in the
Baerenfaller et al (2008) data set, 7869 were unique to the Castellana et al (2008) data
set and only 512 were unique to our own data. Overall, 309638 unique peptides
Downloaded from on June 17, 20178- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
without miss cleavages (81%) have not been observed to date in Arabidopsis (Figure
3). The observed, gene locus specific peptides correspond to 12104 non-redundant
proteins (Baerenfaller et al (2008) 8871 non-redundant proteins, Castellana et al
(2008) 5080 non-redundant proteins, and from our own data 269 non-redundant
proteins) with a range of peptides per protein of 1-132 (Figure 2B.). This represents
less than 45% of the Arabidopsis proteome. If we further exclude peptides that have
had only one or two peptides observed, this reduces the number of non-redundant
proteins to 7067 (~26% of the At proteome). While this shows the potential
limitations imposed by this approach, it also identifies a large set of proteins for
which SRM analysis is theoretically and experimentally feasible today. The MRMaid
database currently provides candidate transitions for 7165 Arabidopsis proteins, while
candidate transitions would have to be manually collated for the remaining ~5000
observed peptides from the three dataset above.
A key consideration in attempts to designing an SRM experiment beyond this
observed peptide set and those supplied by MRMaid is to predict the likelihood that
other peptides could be observed if they were analyzed in a targeted assay. To address
this the proteotypic predictor software packages PeptideSieve (Mallick et al., 2007),
CONSeQuence (Eyers et al., 2011) and STEPP (Webb-Robertson et al., 2010) were
used to assess their ability to more broadly predict proteotypic peptides in
Arabidopsis. Using 1000 randomly selected observed proteotypic peptides and 1000
peptides not observed to date but from the same proteins as the known proteotypic
peptides, we found that STEPP (FDR = 36.7%) was the best predictor of proteotypic
peptides of Arabidopsis. (Supplemental Material, Figure S1, Supplemental Material,
Tables S1 and S2). Interestingly whilst the machine learning methods used in
PeptideSieve (FDR = 37.5%) and CONSeQuence (FDR = 69.4%) were trained on
yeast peptides (Mallick et al., 2007; Eyers et al., 2011) the STEPP predictor was
trained using a combination of three bacterial proteomes (Salmonella typhimurium,
Shewanella oneidensis and Yersinia pestis) (Webb-Robertson et al., 2010).
Comparing the amino acids composition of the proteotypic and non-proteotypic data
sets using the Pepstats package from EMBOSS showed that the presence and absence
of a number of amino acids was influencing detection (Supplemental Material, Table
S3.). The presence of the positively charged amino acids arginine, histidine and
Downloaded from on June 17, 20179- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
lysine, the non-polar methionine, tryptophan, leucine and phenylalanine and cysteine
appear to have a negative influence on detection whereas the presence of the
negatively charged amino acids aspartic and glutamic acids, the non-polar alanine,
tyrosine and valine and glycine and proline appear to have a positive influence on
detection (Supplemental Material, Table S3.).
Using a Bayesian classifier approach that incorporates known proteotypic peptides
and known non-proteotypic peptides and the three proteotypic predictions programs
we were able to create an Arabidopsis proteotypic predictor (APP). Using this
predictor against the same 1000 randomly selected known proteotypic and 1000
peptides not previously observed but from the same proteins, we found APP had a
false discovery rate (FDR) of 30.8%. This represent a ~6% improvement over the
STEPP predictor (Webb-Robertson et al., 2010) for the prediction of Arabidopsis
proteotypic peptides. Using APP we can classify 168494 Arabidopsis peptides to be
proteotypic. These predicted peptides correspond to 21830 proteins (with a range of
peptides per protein = 1-148) (Figure 2C.) or ~80% of the Arabidopsis proteome. If
we further include peptides that have been observed in previous studies this increases
the number of non-redundant proteins to 22132 (~81%, with a range of peptides per
protein = 1-177) (Figure 2D.)), and if we require 3 or more peptide per protein this
reduces the number of non-redundant proteins to 17159 (~63% of the Arabidopsis
proteome).
The APP can be queried by entering an AGI identifier (or group of AGIs) or a protein
sequence
via
a
web
browser
interface,
accessible
at
http://www.plantenergy.uwa.edu.au/APP/. A series of screen shots and basic
instructions are available in Supplemental Material, Figure S2. In addition to
providing a proteotypic prediction of the tryptic peptides from the AGI or sequence,
the output from the APP also provides details of all resulting peptides, their mass, the
number of potential missed cleavage sites, the presence and mass change of possible
methionine oxidation and the number of times a resulting peptide is observed in the
entire Arabidopsis proteome as either tryptic or non-tryptic fragments.
Selection and optimization of candidate SRM peptides for aconitase and malate
dehydrogenase.
Downloaded from on June 17, 201710
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
To test the value of APP to SRM development, we selected unique peptides for
mACO1, mACO2, mMDH1 and mMDH2 that we had previously shown to be
detected by mass spectrometry. Their predicted collision energies (CE) were
calculated using the Skyline software package (MacLean et al., 2010). Of the 12
peptides chosen, 9 were predicted by APP as being proteotypic while three were
predicted to be non-proteotypic. QqQ mass spectrometry then assessed candidate
peptides for detection and their collision energy was optimized using a sample of
isolated Arabidopsis mitochondria. The validity of the transitions was also confirmed
by examining the MSMS spectra of the precursors ions (Supplemental Material,
Figure S3). From these results, a list of SRM transitions was finalized with three
transitions per peptide, one quantifier and two qualifiers and a total of three peptides
per protein (Table I.). An overview of the SRM workflow can be found in
supplemental material, Figure S4.
SRM analysis of protein abundance in mitochondria of aconitase and malate
dehydrogenase knockout lines.
A single maco1 knockout line of mACO1 (At2g05710.1) has previously been
characterized by Moeder et al. (2007), where they showed it had no morphological
phenotype but an absence of mACO1 transcript and a 70% loss in total ACO activity.
No protein measurements for mACO in maco1 have been reported. The maco1 plants
showed an enhance tolerance to oxidative stress induce by the superoxide generating
agent paraquat (Moeder et al., 2007). A single maco2 knockout line (mACO2,
At4g26970.1) has been characterized here (Supplemental Material, Figure S5.),
briefly it showed no morphological phenotype, a ~30% loss in mACO activity and a
significant decrease in mACO2 protein abundance based on 2D gel analysis. Single
knockout lines (mmdh1, mmdh2), a double knockout (mmdh1mmdh2) and a
complemented knockout line (mmdh1mmdh2-35s:MDH1) for mitochondrial malate
dehydrogenase (mMDH) have been previously characterized by Tomaz et al. (2010).
We isolated mitochondrial samples from each of these plant lines and protein samples
were digested with trypsin for analysis.
To examine the abundance of mACO1 we first examined the abundance of the
peptide VVNFSFDGQPAELK in mitochondria isolated from WT, maco1 and maco2
plants (Figure 4.). The SRM 775.9
557.3 transition was used to quantify the
Downloaded from on June 17, 201711
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
abundance and its elution occurred in WT mitochondria between 10.5 - 10.9 minutes
(Figure 4Ai.), between 10.3 - 10.5 minutes from maco1 mitochondria (Figure 4Aii.)
and between 10.0 - 10.3 minutes from maco2 mitochondria (Figure 4Aiii.). The area
under the SRM 775.9
557.3 transitions was then used to calculate the abundance
of VVNFSFDGQPAELK in each of the isolated mitochondrial samples. In addition to
this quantifier transition two qualifier ions were used to confirm the peptide identity,
with an example of the MS/MS spectrum of VVNFSFDGQPAELK provided in
Figure 4B to highlight the quantifier and qualifier ions. We then quantified the
abundance of two other peptides of mACO1 SSGEDTIILAGAEYGSGSSR (979.4
970.4) LSVFDAAMR (505.3
710.3), Figure 5A., Supplemental Material, Table
S4.) along with three peptides from mACO2 (GVISEDFNSYGSR (715.8
FSYNGQPAEIK (627.3
557.3) ILDWENTSTK (603.8
1161.5)
980.4), Figure 5B.,
Supplemental Material, Table S5.). These SRM transitions were designed to allow us
to estimate the protein abundance of mACO1 in maco1 and confirm the protein
knockdown of mACO2 in maco2 observed by 2D-PAGE analysis (Supplemental
Material, Figure S1.). Examining mACO1 we saw that the peptides from this protein
decreased in abundance to between 0.5% to 6.7% with an average of ~4.63% and an
average error of ~0.12% in the maco1 when compared to WT (Figure 5A.,
Supplemental Material, Table S4.). We also saw a decrease in the abundance of
mACO1 in maco2 mitochondria to ~75% (Average Err = 2.7%) of WT levels. A
similar result was obtained for mACO2 where peptides from this protein decrease in
abundance to between 0.02% to 0.1% with an average of ~0.05% and an average error
of ~0.02% in maco2 when compared to WT (Figure 5B., Supplemental Material,
Table S4.). We also saw a slight decrease in the abundance of mACO2 in maco1
mitochondria to ~82% (Average Err = ~3.9%) of WT levels. To further verify these
results heavy labeled peptides (heavy labeled K or R) corresponding to all mACO
peptides were used to construct standard concentration curves in a WT background
(Supplemental Material, Figure S6.). These standard curves were then used to
calculate the absolute abundance of SRM transitions of mACO1 and mACO2 in
mitochondria isolated from WT, maco1 and maco2 plants (Supplemental Material,
Figure S7., Table S7). The absolute concentrations of the peptides confirmed previous
observations that mACO1 was twice as abundant as mACO2 (Taylor et al., 2011) and
corresponded well with the relative abundance results reported in Figure 5 confirming
the knockout of ACO1 and ACO2 in their respective mutant lines. This also showed
Downloaded from on June 17, 201712
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
that for analysis of specific protein abundance in mutants, both relative abundance
and absolute abundance provided very similar insights.
To examine the protein knockdown of mMDH1 in mmdh1 and mMDH2 in mmdh2 we
carried
out
SRM
analysis
of
SEVVGYMGDDNLAK (749.0
(415.3
peptide
transitions
1083.5) EGLEALKPELK (409.6
VAILGAAGGIGQPLALLMK (897.0
(736.3
the
(mMDH1;
486.3)
970.6), mMDH2 SQVSGYMGDDDLGK
1157.5) VVILGAAGGIGQPLSLLMK (613.0
801.5) NLSIAIAK
602.4)) to determine the protein abundance of each isoform in WT, mmdh1
and mmdh2 plants (Figure 6, Supplemental Material, Table S5). In each case, the
similarity of the proteins was such that methionine containing peptides had to be
chosen in order to ensure three specific peptides could be assessed. We also examined
the abundance of mMDH1 and mMDH2 in mmdh1mmdh2 and in mMDH1
complemented mmdh1mmdh2 (mmdh1mmdh2-35s:MDH1) (Figure 6.). We saw a
significant reduction in protein abundance in each of the knockout lines for their
respective proteins with mMDH1 reduced to ~0.1% (Average Err = ~0.09%) of WT
and mMDH2 reduced to ~0.5% (Average Err = ~0.03%) of WT. In mmdh1mmdh2
we saw the loss of both the mMDH isoforms to a level similar to those observed in
the single knockouts. In the complemented line we saw a dramatic increase in the
abundance of mMDH1 to levels much greater to those observed in the WT (~400.0%
(Average Err = ~19%)).
SRM analysis of protein abundance of mMDH1 in leaf extracts from WT mmdh1
and mmdh2 plants.
To assess the sensitivity of SRM mass spectrometry for the detection of this protein in
leaf extracts from 5 week old WT, mmdh1 and mmdh2 plants. SRM analysis of the
peptide
transitions
(mMDH1;
EGLEALKPELK (409.6
SEVVGYMGDDNLAK
486.3) VAILGAAGGIGQPLALLMK (897.0
mMDH2; SQVSGYMGDDDLGK (736.3
(613.0
(749.0
801.5) NLSIAIAK (415.3
1083.5)
970.6),
1157.5) VVILGAAGGIGQPLSLLMK
602.4)) were used to determine the protein
abundance in leaf extracts (Figure 7, Supplemental Material, Table S6). Similarly to
the result we observed in isolated mitochondria we were able to confirm the knockout
of mMDH1 in the leaf extracts from mmdh1 plants and show the detection of this
Downloaded from on June 17, 201713
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
protein in the leaf extracts from WT and mmdh2 plants. Attempts to quantify the
abundance of mMDH2, mACO1 and mACO2 in whole leaf extracts were
unsuccessful due to insufficient signal-to-noise ratio (data not shown).
Downloaded from on June 17, 201714
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
DISCUSSION
The emergence of targeted proteomics now allows researchers to focus on using
proteomics to address specific biological questions in a way that is fundamentally
unlike traditional discovery proteomics pipelines (Marx, 2013). This led to the
recognition of targeted proteomics as the 2012 Method of the Year in Nature Methods
(Aebersold et al., 2013; Editorial, 2013). Here we show the utility of targeted
proteomics approach in Arabidopsis to enable researchers to quantify the abundance
of proteins of interest in T-DNA insertion lines, and to overcome the time and
expense, and lack of specificity when relying on Western blotting to measure protein
abundances in plant samples. In principle, SRM mass spectrometry also enables the
development of specific assays for the quantitation of selected alternative forms of a
protein such as, isoforms, splice variants, mutated versions or those containing PTMs
(that cannot be distinguished by antibodies) as long as these different forms can be
characterized as a mass difference (Picotti et al., 2013).
At this time, where
antibodies are available, they often retain the upper hand in terms of sensitivity for
detection of low abundance proteins, but often lack the specificity required for some
questions such as those outlined above. However, as equipment and approaches
improve, targeted SRM mass spectrometry is likely to surpass Western blotting as the
preferred method for protein quantitation in Arabidopsis. The term ‘Mass Western’
has previously been coined to describe the potential of SRM mass spectrometry to
complement and perhaps replace the use of antibodies for protein quantitation
(Lehmann et al., 2008).
Further advancing the use of SRM mass spectrometry in Arabidopsis will require a
series of underpinning datasets to help design assays and to interpret the results. The
theoretical considerations, outlined in the development of APP, must be combined
with practice and often compromise choices will need to be made between peptide
quality, peak abundance, uniqueness and peptide/transition number in order to obtain
datasets for analysis. Guidelines for analysis are still being developed, but the strong
genetic resources of Arabidopsis will be very useful for testing and assessing SRM
mass spectrometry as a tool for determining the abundance of specific proteins of
interest. SRM mass spectrometry can be immediately helpful to explore the impact of
genetic insertions and deletions on protein abundance. The data provided here allow
Downloaded from on June 17, 201715
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
us to consider five key questions about the use of peptide SRM mass spectrometry in
Arabidopsis and these are each discussed in turn below.
Are there sufficient specific peptides to enable SRM mass spectrometry to
quantify the entire Arabidopsis proteome?
Gene duplication and large-scale expansion of specific protein families in Arabidopsis
has resulted in a large number of similar proteins derived from distinct gene loci. This
means that many tryptic peptides that can be observed in mass spectrometry can be
mapped to multiple locations in the genome. To assess how significant the impact of
this problem is on using SRM mass spectrometry as a measure of protein abundance,
we considered the entire in silico trypic digest of Tair10pep and applied a series of
filters to assess the impact of size, amino acid composition and uniqueness on the
availability of peptides for SRM development. This showed ~90% of proteins
contained at least one unique peptide sequence, and thus that ~24000 Arabidopsis
proteins are theoretically quantifiable by SRM mass spectrometry. Typically, the
longer the peptide the more likely it is to be unique. However, in some cases, certain
biological questions can be addressed by, or even enhanced by, the use of redundant
peptides that can combine the abundance of a group of defined or closely related
proteins.
Which peptides should be prioritized for assessing specific proteins of interest in
Arabidopsis?
The choice of peptides for assessment by SRM mass spectrometry of a given protein
is often not intuitive and requires a series of considerations including uniqueness,
size, detectability and the absence of commonly modified amino acids. To date only
20% (~70000 peptides) of the total number of theoretical trypsin peptides have been
experimentally observed and reported, which limits the number of Arabidopsis
proteins theoretically quantifiable by SRM mass spectrometry to only 12104 proteins
(~44% of the proteome). To investigate whether the low number of peptides/proteins
observed was due to a low number of collected spectra or a limitation in the detection
of Arabidopsis peptides/proteins we developed the Arabidopsis proteotypic predictor
(APP, http://www.plantenergy.uwa.edu.au/APP/) to predict which Arabidopsis
proteins should be observable by mass spectrometry. This analysis highlighted over
168494 peptides that can be predicted to be proteotypic, which would increase the
Downloaded from on June 17, 201716
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
number of proteins that could be quantified by SRM mass spectrometry by over
10,000 to nearly 80% of the Arabidopsis proteome.
Together with the predictions of being proteotypic the APP provides researchers with
key information for designing SRM mass spectrometry experiments including:
theoretical trypsin fragments of a query protein, their mass, the number of potential
missed cleavage sites, the presence and mass change of possible methionine oxidation
and the number of times a resulting peptide is observed in the entire Arabidopsis
proteome as either tryptic or non-tryptic fragments. The APP and MRMaid can serve
as the first stops for Arabidopsis researchers embarking on designing an SRM mass
spectrometry assay by providing them with complementary knowledge required to
select candidate peptides. We also envisaged that the ability to interrogate the APP
and MRMaid, select candidate peptides and collect the results as excel spreadsheets,
will provide researchers with a rapid and data-rich starting point to approach mass
spectrometry labs or services for custom SRM assay development. Interestingly,
despite concerns and limitations we had when selecting candidate peptides for
mMDH and mACO proteins, we were able to correlate the abundance of both
peptides containing an oxidisable Met residue and a trypsin miss cleavage site to other
peptides that did not contain them (Figures 5, 6 and 7; Supplemental Material, Tables
S4, S5 and S6). We also used one peptide for mMDH1 which was miss-cleaved
adjacent to an internal Pro residue (EGLEALKPELK). While there is evidence that
trypsin can cut before Pro in some cases, cleavage at Pro in this case was rare (based
on our MSMS spectral libraries) and the miss-cleaved peptide abundance correlated
with changes in the abundance of other mMDH1 peptides (Figure 6). This suggests
that despite the presence of possible PTM sites or trypsin miss cleavage sites on
peptides, at least some such peptides may be suitable for the establishment of SRM
assays when they can be confirmed by other peptides of the same proteins without
these characteristics.
How can SRM mass spectrometry assays be optimized and confirmed in
Arabidopsis?
After peptide selection, the optimization of transitions is essential to define quantifiers
and qualifiers. We optimized assays for one quantifier and two qualifier ions per
peptide and three peptides per protein. The use of qualifier ions validates that the
Downloaded from on June 17, 201717
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
optimized assay is specific to the peptide for which is designed by confirming the
generation of two further fragmentation ions from the peptide of interest with the
elution time of the peptide during RP-HPLC and reduces the need to produce MSMS
spectra of each precursor ion. The quantifier ions, as the name implies, are the only
ions quantified and used to determine the abundance of the protein in the sample and
typically are the optimized peptides with the highest signal to noise ratio. Undertaking
this optimization work is greatly aided by using partially purified protein (e.g.
bacterial overexpression), an enriched protein fraction (eg. we used a mitochondrial
fraction for mitochondrial proteins) or a 35s-driven complemented plant line where
high levels of the proteins are present. This increased peak signal intensity and helps
ensure that the signal from the transitions dominate the spectra, allowing rapid
definition of the optimal collision energy, retention time and optimization of the
assay. Once assays are optimized they can be deployed to more complex mixtures and
whole tissue assays carried out with greater confidence. The availability of confirmed
knockout mutants in Arabidopsis is also a great asset to test if SRM mass
spectrometry assay signals are removed when single genes are deleted to
independently confirm SRM assay target specificity.
What is the value of SRM mass spectrometry in assessing protein abundance in
Arabidopsis mutant lines?
T-DNA insertion transformation studies in Arabidopsis can provide rapid insight into
gene function, but knowing if a mutant is a knockout or a knockdown or if the protein
is made but truncated, can have a major bearing on the explanation of phenotype. We
tested SRM mass spectrometry in two cases where pairs of highly related proteins
were genetically manipulated leading to different impacts on their respective enzyme
activities. Such differences are commonly attributed to knockdown rather than
knockout, with compensatory responses or with evidence of the different abundance
of isoforms fulfilling the same function. However, often such datasets are
incompletely validated or largely conjecture due to the absence of specific and highly
quantitative means of measuring the different protein products involved. We
examined the relative abundance of two isoforms of mitochondrial aconitase and two
isoforms of mitochondrial malate dehydrogenase in WT, KO, dKO and
complemented plants that have been previously characterized (mACO1, mMDH1 and
mMDH2) or characterized here (mACO2). The isoforms of mACO and mMDH have
Downloaded from on June 17, 201718
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
a high sequence homology and mACO isoforms are indistinguishable using
antibodies (Bernard et al., 2009), whilst the only commercially available mMDH
antibody cross reacts with both isoforms (mMDH1/2, Agisera AS12 2371). We were
able to confirm the protein knockout of all the proteins investigated in mitochondria
isolated from their respective single knockout lines. In comparison, DIGE-based
quantitative proteomics indicated a 3-20 fold difference in abundance but was not
able to confirm the absence of proteins in the T-DNAs lines (Tomaz et al 2010, Supp
Fig 1). Using the optimized SRM assays we were clearly able to distinguish between
and to quantify both pairs of very similar proteins. Further we were able to confirm
the protein knockout of both isoforms of mMDH in the double knockout to below
0.5% of WT levels, and to accurately measure the degree of overexpression of
mMDH1 in the complemented line. Absolute quantitation of proteins is possible using
heavy labelled synthetic peptides that can be spiked into biological samples.
Undertaking this analysis for mACO1 and mACO2 provided similar data from the
relative quantitation in terms of assessing knockout plants (Supplemental Figure 7).
However, there will be circumstances where reproducibility of extraction,
chromatography or ionization, or comparisons between proteins and stoichiometry
within protein complexes would be aided by absolute abundance information.
What are the limits of detection for SRM in assessing Arabidopsis proteins?
The use of the APP for the selection of candidate SRM peptides and the
implementation of this workflow is applicable for the assessment of a wide range of
proteins in mutant lines of Arabidopsis, however, abundance will be a critical issue in
its deployment for many proteins. Very low abundance proteins such as transcription
factors would require very significant enrichment strategies until increases in MS
sensitivity of ~2 to ~3 orders of magnitude are achieved. However, with increasingly
more sensitive QqQ mass spectrometers being released each year this time is
approaching. To determine the sensitivity of the detection of proteins by SRM mass
spectrometry, we attempted to assess the abundance of all four proteins in
Arabidopsis leaf total protein extracts. We were able to quantify the abundance of the
most abundant of the 4 proteins, mMDH1, however we were not able to quantify the
other three. We have previously shown that mACO1 and mACO2 together account
for ~1.5% of mitochondrial protein with mACO1 twice as abundant as mACO2 and
that mMDH1 and mMDH2 account for ~2% of mitochondrial protein with mMDH2
Downloaded from on June 17, 201719
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
representing only ~20% of the abundance of mMDH1 (Taylor et al., 2011). While
mMDH1 is the most abundant protein measured in this study, it is likely it accounts
for only <0.05% of total cellular protein (TCP), as mitochondrial protein represents
only approximately 3% of TCP. Attempts to quantify the abundance of mMDH2
(~0.01% TCP), mACO1 (~0.03% TCP) and mACO2 (~0.015% TCP) in whole leaf
extracts were unsuccessful, showing that this approach currently requires proteins of
at least ~0.05% or 5 pg/g TCP, which represents approximately 0.5 ng in the 1 µg of
protein on column in liquid chromatography. However, we have shown here that with
some subcellular fractionation lower concentration proteins can be readily quantified.
With the increasing adoption of instrumentation and software to enable SRM mass
spectrometry approaches in a large number of labs around the world, scientists will
increasingly be able to access different types of equipment for SRM assays. As a
result quantitation of protein abundance in T-DNA insertion lines will soon become
the norm rather than the exception.
Downloaded from on June 17, 201720
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
MATERIALS AND METHODS
Plant material.
Seeds of Arabidopsis (Arabidopsis thaliana) ecotype Columbia wild type and
homozygous T-DNA insertion lines, double knockout and complemented lines
mmdh1 (mMDH1 KO, At1g53240.1, GABI_540F04), mmdh2 (mMDH2 KO,
At3g15020.1, Salk_126994), mmdh1mmdh2 (mMDH double KO), mmdh1mmdh235s:mMDH1 (MDH double KO-1:35s MDH1) were obtained from ABRC or
generated in-house and characterized as in Tomaz et al. (2010). maco1 (mACO1 KO,
At2g05710.1, Salk_014661) was obtained from Moeder et al. (2007) and maco2
(mACO2 KO, At4g26970.1, Salk_090200) was obtained from ABRC and is
characterised in Supplemental Material, Figure S1.
Theoretical analysis of Arabidopsis thaliana proteome to identify candidate SRM
peptides.
The non-redundant Arabidopsis protein set was obtained from The Arabidopsis
Information Resource (TAIR, release 10, TAIR10pep.fasta) (Lamesch et al., 2012)
and a theoretical digestion was carried out on the whole genome using Expasy Peptide
Mass (http://web.expasy.org/peptide_mass/) using trypsin as enzyme (C-terminal side
of K or R except when P is C-terminal to K or R), maximum number of missed
cleavages = 1, all cysteines in reduced form, variable methionine oxidation to form
methionine sulfoxide. Monoisotopic masses of the occurring amino acid residues
were used and peptide masses were recorded as [M+H]+. Resulting peptides with a
Mr of 500-3000 Daltons were collected. All resulting peptide data (Arabidopsis
Theoretical Trypsin Digest, ATTD) were then standard protein BLAST (blastp)
searched against the TAIR10pep and the results were used to determine the number of
times a peptide occurred in TAIR10pep. ATTD was then further interrogated as
required for analysis. Dataset from Baerenfaller et al (2008) was obtained from
http://fgcz-pep2pro.uzh.ch and dataset from Castellana et al (2008) was obtained from
http://proteomics.ucsd.edu/Software/ArabidopsisProteogenomics.html.
Downloaded from on June 17, 201721
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Theoretical prediction of Arabidopsis thaliana proteotypic peptides.
PeptideSieve
software
was
obtained
from
http://tools.proteomecenter.org/wiki/index.php?title=Software:PeptideSieve (Mallick
et al., 2007). PeptideSieve was run using a 2000 peptide training list or with ATTD
and minMass=500, maxMass=3000, numAllowedMisCleavages=1 and pValue=0.
CONSeQuence
predictions
were
carried
out
using
the
website
http://king.smith.man.ac.uk/CONSeQuence/ using the “Rank score” prediction type
based on a linear SVM using a 2000 peptide training list or with ATTD (Eyers et al.,
2011).
STEPP
software
was
downloaded
from
http://omics.pnl.gov/software/STEPP.php (Webb-Robertson et al., 2010). STEPP was
run using a 2000 peptide training list in “Peptides in Excel” mode or with ATTD in
“Proteins in FASTA” mode using “minimum sequence length” = 6, “max missed
cleavages” = 1 and “max daltons” = 3000. The resulting peptides and proteotypic
predictions were then further interrogated as required for analysis.
The Arabidopsis Proteotypic Predictor (APP).
The
APP
utilizes
the
database
programming
language
SQL
(Structured
QueryLanguage) and is housed on a Linux server running Ubuntu 10.04 LTS. The
APP web browser-based graphical user interface is written in Dynamic Hyper Text
Markup Language that makes use of Asynchronous JavaScript and XML (AJAX) to
interact with the APP server. The back-end of the APP utilizes a number of PHP
scripts that interact with the MySQL tables housing the APP data. Making use of
complex JavaScript, the interface works best via the Mozilla Firefox, Google Chrome
or Safari web browsers but will work on Microsoft Internet Explorer (6 and above).
The APP leverages open-source technologies in order to provide a freely available
platform at http://www.plantenergy.uwa.edu.au/APP/. The APP was developed using
an assessment of three previously published proteotypic predictors (PeptideSieve,
(Mallick et al., 2007); CONSeQuence, (Eyers et al., 2011); STEPP, (Webb-Robertson
et al., 2010)) on the ATTD and a 2000 peptide training list. The APP prediction is
based on Bayesian-based classification of probabilities calculated from these
predictions and the training list for each peptide.
Downloaded from on June 17, 201722
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Purification of mitochondria from hydroponic shoot tissue
All relevant Arabidopsis genotypes were grown and maintained in exact accordance
with a previously published protocol (Lee et al., 2008). Shoot mitochondria were
isolated from 3-week-old hydroponically grown Arabidopsis using the method from
(Tomaz et al., 2010).
Isolation of protein extracts from leaf tissue
Five week old leaf tissue was snap frozen and ground in liquid nitrogen. Proteins were
extracted in 2 mL grinding buffer (300 mM sucrose, 50 mM Tris (pH 7.5), 1 mM
DTT, 0.5 % (w/v) PVP and complete protease cocktail inhibitor (Roche, 1 tablet per
50 mL) per mg fresh tissue while rocking on ice for 20 min. Cell debris was removed
by filtration through disposable, fritted syringe barrels containing Miracloth
(Calbiochem). Soluble proteins were separated from the filtrate by a 20 minute
centrifugation at 20000 x g and 4 °C. The protein concentration was then determined
by Bradford assay and spectrophotometric measurement at a wavelength of 595 nm
using BSA as a standard. Samples of 200 µg soluble protein were precipitated in 5
times volume of chilled (-20 °C) acetone overnight followed by 20 min centrifugation
at 20000 x g and 4 °C. After two further acetone washes the samples were resuspended in 20 µL buffer (8 M urea, 50 mM NH4HCO3, 5 mM DTT) and incubated
at room temperature for an hour. Iodoacetamide was added to a final concentration of
10 mM followed by 30 min incubation at room temperature in the dark. The sample
solutions were then diluted to below 1 M urea with 50 mM NH4HCO3. For protein
digestion 10 µg trypsin (dissolved in 0.01 % TFA to a concentration of 1 mg mL-1)
were added to each sample and incubated at 37 °C overnight. The samples were
acidified to 1 % (v/v) with formic acid and SPE cleaned using Silica C18 Macrospin
columns (The Nest Group). After each of the following steps SPE columns were
centrifuged for 3 min at 150 x g at room temperature. Before loading the samples
columns were washed with 750 µL of 70 % (v/v) acetonitrile, 0.1 % (v/v) formic acid
and charged with 750 µL of 5 % (v/v) acetonitrile, 0.1 % (v/v) formic acid. After
loading the samples onto the columns two washes with 750 µL of 5 % (v/v)
acetonitrile, 0.1 % (v/v) formic acid were carried out, followed by two elution steps
with 750 µL 70 % (v/v) acetonitrile, 0.1 % (v/v) formic acid. The eluate was dried
under vacuum and re-suspended in 5 % (v/v) acetonitrile, 0.01 % (v/v) formic acid to
a final concentration of 1 µg µL-1 for mass spectrometry.
Downloaded from on June 17, 201723
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Optimization of SRM transitions
SRM transitions were optimized using trypsin digested isolated mitochondrial extracts
run on an Agilent 6430 QqQ mass spectrometer with an HPLC Chip Cube source
(Agilent Technologies). The chip consisted of a 160 nL enrichment column (Zorbax
300SB-C18, 5-mm pore size) and a 150 mm separation column (Zorbax 300SB-C18,
5 mm pore size) driven by Agilent Technologies 1200 series nano/capillary LC
system. Both systems were controlled by MassHunter Workstation Data Acquisition
for QqQ (B.03.01 (B2065), version 5.1.2600, SP3, build 2600; Agilent Technologies).
Peptides were loaded onto the trapping column at 3 mL/min in 5% (v/v) acetonitrile
and 0.1% (v/v) formic acid with the chip switched to enrichment and using the
capillary pump. The chip was then switched to separation, and peptides were eluted
during a 15.5-min gradient (5% [v/v] acetonitrile to 100% [v/v] acetonitrile) directly
into the mass spectrometer. The mass spectrometer was run in positive ion mode, with
a drying gas temperature of 365 °C and flow rate of 5L/min, for each transition the
fragmentor was set to 130 and dwell time was 5 ms. Based on a historical ‘in-house’
dataset and a theoretically digested background proteome of each of the proteins of
interest (mACO1, mACO2, mMDH1 and mMDH2) possible peptide transitions were
determined using Skyline (version 1.1.0.2905) (MacLean et al., 2010). These were
then optimized for collision energy (CE) based on predicted values by Skyline
following an algorithm specific for Agilent instruments. For each transition a total of
five CEs were analyzed, including the predicted CE ± 4 V and 8 V. Having optimized
all available SRM transitions optimized data was used to select candidates for
quantitative data analysis (Table 1.).
Selection of candidate SRM transitions
Selection of candidate SRM transitions was carried out in two steps. First peptides
were assessed by their uniqueness in the Arabidopsis proteome, number of miss
cleavages and composition. Peptides were selected based on preference for
uniqueness > number of miss cleavages > presence of M. Optimized SRM transitions
of these peptides were then reviewed in Skyline for signal intensity, signal-to-noise
ratio, dot product and y-ion ranking. For each peptide three transitions, one quantifier
and two qualifiers were selected to validate it. A total of three peptides per protein
were analyzed.
Downloaded from on June 17, 201724
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Protein analysis using SRM mass spectrometry
Protein extracts from isolated mitochondrial and whole leaf digests were analyzed on
an Agilent 6430 QqQ mass spectrometer as described above for method optimization.
Heavy
labeled
standards
for
3
mACO1
(VVNFSFDGQPAELK13C615N2,
SSGEDTIILAGAEYGSGSSR13C615N4, LSVFDAAMR13C615N4) and 3 mACO2
(GVISEDFNSYGSR13C615N4, FSYNGQPAEIK13C615N2, ILDWENTSTK13C615N2)
peptides were obtained from JPT Peptide Technologies and prepared following
manufacturer’s instructions. Standard curves of heavy labeled peptides were
constructed using various concentrations of peptides spiked into isolated WT
mitochondria.
Data Analysis
Transitions that had an intensity greater than 1000 and s/n >50 were then further
analyzed. Resulting total ion chromatograms were opened in MassHunter
Workstation Qualitative Analysis (version B.01.04, build 1.4.126.0; Agilent
Technologies), and SRM chromatograms were obtained using the Extract
Chromatogram feature using default settings. Each SRM chromatogram was then
integrated, and the area under the peak within 30 s of the expected retention time was
calculated. The area under the curve for each replicate was then averaged to obtain an
abundance value for each peptide. The abundance of a peptide in each of the mutant
lines was then compared to WT to calculate a relative abundance.
Downloaded from on June 17, 201725
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Supplemental Material
Figure S1. Anaysis of prediction of peptide flying by PeptideSieve, CONSeQuence
and STEPP of 1000 proteotypic & non-proteotypic peptides from Arabidopsis
thaliana.
Figure S2. Screenshots of the Arabidopsis Proteotypic Predictor (APP) web interface.
Figure S3. MSMS Validation of mACO1 (standard and heavy labeled), mACO2
(standard and heavy labeled), mMDH1 and mMDH2 peptides.
Figure S4. An overview of SRM workflow for relative and absolute quantitation
Figure S5. mACO2 (At4g26970.1) T-DNA mutant characterization.
Figure S6. Standard curves of mACO1 and mACO2 heavy labelled peptides.
Figure S7. SRM analysis of absolute protein abundance of mACO1 and mACO2 in
WT, maco1 and maco2 mitochondria A.
Table S1. 1000 proteotypic peptides used to rate proteotypic predictors
Table S2. 1000 non-proteotypic peptides used to rate proteotypic predictors
Table S3. Amino acid composition of proteotypic peptides
Table S4. Relative abundance of mACO1 and mACO2 peptides in mitochondria from
genotypes
Table S5. Relative abundance of mMDH1 and mMDH2 peptides in mitochondria
from genotypes
Table S6. Relative abundance of mMDH1 peptides in whole tissue extracts from
genotypes
Table S7. Absolute abundance of mACO1 and mACO2 peptides in mitochondria
from genotypes.
Downloaded from on June 17, 201726
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
LITERATURE CITED
Aebersold R, Burlingame AL, Bradshaw RA (2013) Western blots versus selected
reaction monitoring assays: time to turn the tables? Mol Cell Proteomics 12:
2381-2382
AGI (2000) Analysis of the genome sequence of the flowering plant Arabidopsis
thaliana. Nature 408: 796-815
Alonso JM, Ecker JR (2006) Moving forward in reverse: genetic technologies to
enable genome-wide phenomic screens in Arabidopsis. Nature Rev Genet 7:
524-536
Baerenfaller K, Grossmann J, Grobei MA, Hull R, Hirsch-Hoffmann M,
Yalovsky S, Zimmermann P, Grossniklaus U, Gruissem W, Baginsky S
(2008) Genome-scale proteomics reveals Arabidopsis thaliana gene models
and proteome dynamics. Science 320: 938-941
Bernard DG, Cheng Y, Zhao Y, Balk J (2009) An allelic mutant series of ATM3
reveals its key role in the biogenesis of cytosolic iron-sulfur proteins in
Arabidopsis. Plant Physiol 151: 590-602
Castellana NE, Payne SH, Shen Z, Stanke M, Bafna V, Briggs SP (2008)
Discovery and revision of Arabidopsis genes by proteogenomics. Proc Natl
Acad Sci USA 105: 21034-21038
Eyers CE, Lawless C, Wedge DC, Lau KW, Gaskell SJ, Hubbard SJ (2011)
CONSeQuence: prediction of reference peptides for absolute quantitative
proteomics using consensus machine learning approaches. Mol Cell
Proteomics 10: M110 003384
Fan J, Mohareb F, Jones AM, Bessant C (2012) MRMaid: The SRM assay design
tool for Arabidopsis and other species. Front Plant Sci 3: 164
Ito J, Batth TS, Petzold CJ, Redding-Johanson AM, Mukhopadhyay A,
Verboom R, Meyer EH, Millar AH, Heazlewood JL (2011) Analysis of the
Arabidopsis cytosolic proteome highlights subcellular partitioning of central
plant metabolism. J Prot Res 10: 1571-1582
Lamesch P, Berardini TZ, Li D, Swarbreck D, Wilks C, Sasidharan R, Muller R,
Dreher K, Alexander DL, Garcia-Hernandez M, Karthikeyan AS, Lee
CH, Nelson WD, Ploetz L, Singh S, Wensel A, Huala E (2012) The
Arabidopsis Information Resource (TAIR): improved gene annotation and
new tools. Nuc Acids Res 40: D1202-1210
Lange V, Picotti P, Domon B, Aebersold R (2008) Selected reaction monitoring for
quantitative proteomics: a tutorial. Mol Syst Biol. 2008; 4: 222
Lee CP, Eubel H, O'Toole N, Millar AH (2008) Heterogeneity of the mitochondrial
proteome for photosynthetic and non-photosynthetic Arabidopsis metabolism.
Mol Cell Proteomics 7: 1297-1316
Lehmann U, Wienkoop S, Tschoep H, Weckwerth W (2008) If the antibody fails a Mass western approach. Plant J 55: 1039-1046
MacLean B, Tomazela DM, Shulman N, Chambers M, Finney GL, Frewen B,
Kern R, Tabb DL, Liebler DC, MacCoss MJ (2010) Skyline: an open
source document editor for creating and analyzing targeted proteomics
experiments. Bioinformatics 26: 966-968
Mallick P, Schirle M, Chen SS, Flory MR, Lee H, Martin D, Ranish J, Raught B,
Schmitt R, Werner T, Kuster B, Aebersold R (2007) Computational
prediction of proteotypic peptides for quantitative proteomics. Nature Biotech
25: 125-131
Marx V (2013) Targetted proteomics. Nature Meth 10: 19-22
Downloaded from on June 17, 201727
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Moeder W, Del Pozo O, Navarre DA, Martin GB, Klessig DF (2007) Aconitase
plays a role in regulating resistance to oxidative stress and cell death in
Arabidopsis and Nicotiana benthamiana. Plant Mol Biol 63: 273-287
Monneuse JM, Sugano M, Becue T, Santoni V, Hem S, Rossignol M (2011)
Towards the profiling of the Arabidopsis thaliana plasma membrane
transportome by targeted proteomics. Proteomics 11: 1789-1797
O'Malley RC, Ecker JR (2010) Linking genotype to phenotype using the
Arabidopsis unimutant collection. Plant J 61: 928-940
Picotti P, Aebersold R (2012) Selected reaction monitoring-based proteomics:
workflows, potential, pitfalls and future directions. Nature Meth 9: 555-566
Picotti P, Bodenmiller B, Aebersold R (2013) Proteomics meets the scientific
method. Nature Meth 10: 24-27
Taylor NL, Heazlewood JL, Millar AH (2011) The Arabidopsis thaliana 2-D gel
mitochondrial proteome: Refining the value of reference maps for assessing
protein abundance, contaminants and post-translational modifications.
Proteomics 11: 1720-1733
Taylor NL, Howell KA, Heazlewood JL, Tan TY, Narsai R, Huang S, Whelan J,
Millar AH (2010) Analysis of the rice mitochondrial carrier family reveals
anaerobic accumulation of a basic amino acid carrier involved in arginine
metabolism during seed germination. Plant Physiol 154: 691-704
Tomaz T, Bagard M, Pracharoenwattana I, Linden P, Lee CP, Carroll AJ,
Stroher E, Smith SM, Gardestrom P, Millar AH (2010) Mitochondrial
malate dehydrogenase lowers leaf respiration and alters photorespiration and
plant growth in Arabidopsis. Plant Physiol 154: 1143-1157
Wang YH (2008) How effective is T-DNA insertional mutagenesis in Arabidopsis ? J
Biochem Tech 1: 11-20
Webb-Robertson BJ, Cannon WR, Oehmen CS, Shah AR, Gurumoorthi V,
Lipton MS, Waters KM (2010) A support vector machine model for the
prediction of proteotypic peptides for accurate mass and time proteomics.
Bioinformatics 26: 1677-1683
Wienkoop S, Larrainzar E, Glinski M, Gonzalez EM, Arrese-Igor C, Weckwerth
W (2008) Absolute quantification of Medicago truncatula sucrose synthase
isoforms and N-metabolism enzymes in symbiotic root nodules and the
detection of novel nodule phosphoproteins by mass spectrometry. J Exp Bot
59: 3307-3315
Downloaded from on June 17, 201728
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
FIGURE LEGENDS
Figure 1. The number of peptides per protein and the number of times a peptide
occurs in a theoretical trypsin digest of the Arabidopsis thaliana proteome. A. The
number of peptides per protein of all theoretical peptides resulting from in silico
trypsin digestion. B. The number of times a peptide occurs in the TAIR10pep
following an in silico trypsin digestion.
Figure 2. Analysis of theoretically observed unique peptides and predicted
peptides of the Arabidopsis thaliana proteome. A. All theoretical peptides per
protein resulting from in silico trypsin digestion that contain no missed cleavage and
are unique. B. Number of peptides per protein in the 12104 proteins observed in
Baerenfaller et al (2008), Castellana et al (2008) and our own data. C. Number of
peptides per protein of the predicted proteotypic peptides calculated by APP. D.
Number of peptides per protein of the predicted proteotypic peptides calculated by
APP and proteins observed in Baerenfaller et al (2008), Castellana et al (2008) and
our own data.
Figure 3. Number of peptides from an in silico trypsin digest of the Arabidopsis
proteome that have been observed experimentally by mass spectrometry.
Peptides observed in any of the three data sets (black) as well as peptides unique to
Baerenfaller et al (2008)(magenta), Castellana et al (2008)(yellow) and our own work
(cyan).
Figure 4. The elution of VVNFSFDGQPAELK of mACO1 in WT and knockout
lines and the MSMS spectra showing quantifier and qualifier ions. A. SRM 775.9
557.3 transition of VVNFSFDGQPAELK peptide of mACO1 in WT, maco1 and
maco2 mitochondria. i. WT, ii. maco1, iii. maco2. B. The MS/MS spectrum of
VVNFSFDGQPAELK showing the y-series ions and the selected quantifier ion (y5)
and the two qualifier ions (y8 and y10).
Figure 5. SRM analysis of protein abundance of mACO1 and mACO2 in WT,
maco1 and maco2 mitochondria A. SRM analysis of unique peptides mACO1
Downloaded from on June 17, 201729
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
abundance using the quantifier ion transitions VVNKSFDGQPAELK (SRM 775.9
557.3) SSGEDTIILAGAEYGSGSSR (SRM 979.4
505.3
quantifier
970.4) LSVFDAAMR (SRM
710.3). B. SRM analysis of unique peptides mACO2 abundance using the
ion
transitions
GVISEDFNSYGSR
FSYNGQPAEIK (SRM 627.3
(SRM
715.8
1161.5)
557.3) ILDWENTSTK (SRM 603.8
980.4).
Data presented is averages ± SE (n=3).
Figure 6. SRM analysis of protein abundance of mMDH1 and mMDH2 in WT,
mmdh1, mmdh2, mmdh1mmdh2 and mmdh1mmdh2 complemented with mMDH1
cDNA mitochondria. A. SRM analysis of unique peptides mMDH1 abundance using
the quantifier ion transitions SEVVGYMGDDNLAK (SRM 749.0
EGLEALKPELK (SRM 409.6
1083.5)
486.3) VAILGAAGGIGQPLALLMK (SRM 897.0
970.6). B. SRM analysis of unique peptides mMDH2 abundance using the
quantifier ion transitions SQVSGYMGDDDLGK (SRM 736.3
VVILGAAGGIGQPLSLLMK (SRM 613.0
1157.5)
801.5) NLSIAIAK (SRM 415.3
602.4). Data presented is averages ± SE (n=3).
Figure 7. SRM analysis of protein abundance of mMDH1 in WT, mmdh1 and
mmdh2 leaf extracts. SRM analysis of mMDH1 abundance using the quantifier ion
transitions SEVVGYMGDDNLAK (SRM 749.0
409.6
1083.5) EGLEALKPELK (SRM
486.3) VAILGAAGGIGQPLALLMK (SRM 897.0
presented is averages ± SE (n=3)
Downloaded from on June 17, 201730
- Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
970.6). Data
Table I. Optimized SRM transitions for mACO1, mACO2, mMDH1, mMDH2.
Downloaded from on June 17, 2017 - Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
AGI
Protein
Sequence
SRM
Pre
m/z
At2g05710.1
mACO1
VVNFSFDGQPAELK
775.9
2
At2g05710.1
mACO1
SSGEDTIILAGAEYGSGSSR
979.0
2
At2g05710.1
mACO1
LSVFDAAMR
505.3
2
710.30
y6
809.40
y7
At4g26970.1
mACO2
GVISEDFNSYGSR
715.8
2
1161.50
y10
945.40
y8
At4g26970.1
mACO2
FSYNGQPAEIK
627.3
2
557.30
y5
724.40
y7
At4g26970.1
mACO2
ILDWENTSTK
603.8
2
980.40
y8
679.30
y6
At1g53240.1
mMDH1
SEVVGYMGDDNLAK
749.4
2
1083.50
y10
732.40
y7
963.40
At1g53240.1
mMDH1
EGLEALKPELK
409.6
3
486.30
y4
614.40
y5
727.50
SRM
Pre
z
SRM Pro 1
m/z
(Quantifier)
SRM Pro 1
ion
(Quantifier)
SRM Pro
2 m/z
(Qualifier)
SRM Pro
2 ion
(Qualifier)
557.33
y5
1094.54
y10
970.40
y10
1041.50
y11
SRM Pro
3 m/z
(Qualifier)
SRM Pro
3 ion
(Qualifier)
RT
(minutes)
Predicted
CE (V)
Optimized
CE (V)
Area
Ratio
P1/P2
Area
Ratio
P1/P3
857.44
y8
11.0
24.0
20.0
0.60
0.35
1154.50
y12
10.1
34.4
34.4
0.75
0.65
563.30
y5
10.0
10.2
14.2
0.60
0.60
830.40
y7
9.1
20.9
20.9
0.60
0.35
856.50
y8
8.2
16.4
16.4
0.45
0.40
865.40
y7
9.1
15.2
15.2
0.20
0.20
y8
8.9
22.7
18.7
0.60
0.60
y6
9.3
5.4
9.4
0.20
0.05
At1g53240.1
mMDH1
VAILGAAGGIGQPLALLMK
897.0
2
970.60
y9
1197.70
y12
785.50
y7
15.4
30.2
26.2
0.75
0.70
At3g15020.1
mMDH2
SQVSGYMGDDDLGK
736.3
2
1157.50
y11
1256.50
y12
850.40
y8
8.3
22.0
22.0
0.55
0.50
At3g15020.1
mMDH2
VVILGAAGGIGQPLSLLMK
613.0
3
801.50
y7
591.40
y5
986.60
y9
15.8
12.9
12.9
0.10
0.20
At3g15020.1
mMDH2
NLSIAIAK
415.3
2
602.40
y6
515.40
y5
402.30
y4
9.3
5.6
5.6
0.10
0.05
AGI, Arabidopsis Genome Initiative identifier; Protein, protein name; Sequence, peptide sequence; SRM Pre m/z, peptide precursor ion
mass/charge ratio; SRM Pre z, peptide precursor ion mass; SRM Pro 1 m/z (Quantifier), peptide product ion 1 (Quantifier) mass/charge ratio;
SRM Pro 1 ion (Quantifier), peptide product ion 1 (Quantifier) fragmentation series location; SRM Pro 2 m/z (Qualifier), peptide product ion 2
(Qualifier) mass/charge ratio; SRM Pro 2 ion (Qualifier), peptide product ion 2 (Qualifier) fragmentation series location; SRM Pro 3 m/z
(Qualifier), peptide product ion 3 (Qualifier) mass/charge ratio; SRM Pro 3 ion (Qualifier), peptide product ion 3 (Qualifier) fragmentation
series location; RT, retention time of peptide on column; Predicted CE, predicted collision energy from Skyline (MacLean et al., 2010);
Optimized CE, optimized collision energy; Area Ratio P1/P2, ratio of the area of XIC of production ion 1/product ion 2; Area Ratio P1/P3, ratio
of the area of XIC of production ion 1/product ion 3.
31
32
Downloaded from on June 17, 2017 - Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Downloaded fro
Copyright © 2013
Downloaded fr
Copyright © 20
Downloaded from on June 17, 2017 - Published by www.plantphysiol
Copyright © 2013 American Society of Plant Biologists. All rights reser
Downloaded from on June 17, 2017 - Published by www.plantphysiol.org
Copyright © 2013 American Society of Plant Biologists. All rights reserved.
Downloaded from on June 17, 2017 - Published by www
Copyright © 2013 American Society of Plant Biologists. A
Downloaded from on June 17, 2017 - Published by www
Copyright © 2013 American Society of Plant Biologists. A
Downloaded from on June 17, 2017 - Published by w
Copyright © 2013 American Society of Plant Biologists