Improved protein separation and identification by use of 2D liquid protein fractionation and ion mobility mass spectrometry
1
1
1
1
1
2
Susan E. Slade ; Konstantinos Thalassinos ; Sarah J. Nicholson ; Jonathan P. Williams ; James H. Scrivens ; Kevin Giles and Robert H. Bateman
2
1
Biological Mass Spectrometry and Proteomics, University of Warwick, Coventry, United Kingdom
2
Waters Micromass MS Technologies, Floats Road, Wythenshawe, Manchester, United Kingdom
OVERVIEW
MATERIALS AND METHODS
Chromatofocusing
Column (CF)
Sample
Reverse Phase
Column (RP)
Purpose
To utilise and evaluate the combination of a commercial two-dimensional liquid protein
separation system with mass spectrometry-based protein identification for proteomics
applications.
Two-dimensional liquid chromatography protein separation and mass spectrometry-based
identification
Intact Mass
Digest
Methods
E. coli cell lysate proteins were resolved by chromatofocusing followed by hydrophobicity
chromatographic separations. A number of protein fractions were concentrated and
tryptically digested prior to analysis by means of LC-ESI-MS/MS and protein identification.
MALDI-PMF
Bioinformatic approaches were developed to extract biologically relevant information from the
generated dataset.
Results
Preliminary results indicate that we have a demonstrated a significant improvement in our
protein identification confidence due to increased sequence coverage for proteins from a
widerange of molecular weight and isoelectric point, compared to samples from gel-based
sources.
Interesting observations were noted including one protein that eluted at a number of pI
intervals, with differing peptide sequences observed in each fraction.
In addition, the solution-based protein separation system allows characterisation of both
tryptically digested proteins and the intact species thus enabling further protein
characterisation.
INTRODUCTION
Gel-based proteomics experiments have proved highly successful in analysing relatively
complex biological systems but have a number of limitations. These include narrow sample
loading capacity reducing the quantity of protein available for downstream mass
spectrometry-based identification. When the capacity is exceeded, poor resolution of protein
species is evident. Other limitations include gel-to-gel variation, narrow dynamic range,
difficulties in generating a homogeneous sample for analysis due to problems with protein
solubility and the low efficiency of peptide extraction post-tryptic digestion compounding the
problems encountered in protein identification. Prefractionation of complex proteomes can
be required prior to analysis particularly when the sample source contains a number of highly
abundant proteins. The characterisation of low abundance species continues to be a
challenge in the field of proteomics.
UV Detector
280 nm
LC-ESI-MS/MS
UV Detector
214 nm
Database Search
Figure 1. Schematic representation of the separation mechanism employed by the PF2D
protein separation system (Beckman Coulter) and subsequent potential MS analysis
strategies.
Typically a single 2D-LC separation combining the collection of protein fractions from the CF
and RP columns could generate approximately 800 samples for digestion and MS analysis.
Clearly a more focused approach to the analysis of biological systems is required with the
necessity to derive biologically relevant information from the vast quantity of mass
spectrometry data generated.
Each CF fraction was then applied to a non-porous silica RP column (Beckman Coulter) and
further resolved by an organic solvent gradient using acetonitrile with trifluroacetic acid as
ion pairing agent. Fixed volume fractions were collected during each separation and
o
aliquotted into smaller volumes prior to storage at -70 C
A number of fractions were selected for MS-based protein identification across a number of
pH intervals and apparent protein concentrations. A 175 µL aliquot of each fraction was dried
and reconstituted in 10 µL of ammonium bicarbonate buffer. and processed using a
MassPrep robotic protein handling system (Waters Micromass MS Technologies) using the
maufacturer’s in-solution digest protocol. In brief, protein samples were reduced, alkylated
with iodoacetamide, digested with trypsin and the resultant peptides acidified.
Figure 3. ProteoView visualisation of E. coli fractionated proteins using a 2D-LC approach.
Vertical columns represent CF fractions and horizontal bands protein absorbance. Shown on
the left is the absorbance measured at 214 nm during the reverse phase elution of one
chromatofocusing fraction.
!
We have identified 183 proteins with a minimum of one observed peptide, of which 92
were identified with two peptides and 51 with a minimum of 3 peptides.
!
In addition, many CF fractions contain multiple proteins often identified with a number of
peptides observed, see Figure 4.
We have developed extensive in-house bioinformatics resources which have facilitated the
interpretation of the 2D-LC data, elements of which are presented on poster TP28
(Thalassinos, Slade et al.).
The tryptic extracts were analysed by means of nano-LC-ESI-MS/MS on a Q-Tof Ultima
Global with in-line CapLC system (Waters Micromass MS Technologies). The tryptic extract
was desalted using an in-line C18 precolumn cartridge (Dionex, U.S.A.) and the peptides
further resolved on a 75 µm C18 PepMap column (Dionex, U.S.A.) using an increasing
acetonitrile concentration gradient.
Due to the highly complex nature of many biological systems, it may not be possible to resolve
each of the proteins in a sample into individual fractions using the 2D-LC separation.
Therefore we have explored the potential of ion mobility to resolve peptides of identical m/z,
which may result from a tryptic digest of a 2D-LC fraction containing a number of proteins, see
Figure 2.
Ion mobility study of isomeric peptides
Doubly charged precursor ions at m/z 246 were selected from two peptides having the
sequences GRGDS and SDGRG, for ion mobility study using the Synapt HDMS (Waters
Micromass MS Technologies), a hybrid MS-IMS-MS instrument with a Quadrupole / IMS /
Orthogonal-TOF configuration. The ion mobility separation device (IMS) comprises three
consecutive travelling wave RF ion guides (Triwave) incorporating a repeating sequence of
transient DC pulses to propel ions through the guide in the presence of a background gas.
Ions are accumulated in the Trap T-Wave and periodically released into the IMS T-Wave
where they separate according to their mobility.
RESULTS
Protein resolution is achieved in the first dimension using chromatofocusing (CF) by
generating a pH gradient on an ion exchange resin. The column is equilibrated at basic pH
and the solublised proteins are applied to the column. Proteins that have an isoelectric point
(pI) equivalent to the column pH have a net zero charge, thus do not bind and elute
immediately from the column and are collected. Over a period of a few hours, the pH of the
column is reduced and sequentially the proteins elute from the column according to their pI
and are collected for further analysis. Any proteins still bound to the column at acidic pH are
eluted using a high ionic strength buffer and the fractions collected, see Figure 1.
RESULTS
Protein identification
To date, we have obtained significant numbers of protein identifications from seven of the
possible 32 reverse phase-separated CF fractions, spanning the pH region 6.4 to 4.0.
Figure 2. Overlay of the arrival time mobility distributions for the doubly charged ions of
two isomeric peptides using the Synapt HDMS (Waters Micromass MS Technologies,
U.K.). Collisional cross-section measurements indicate a 5 % difference in physical size
and shape.
!
Each of the CF fractions contains multiple proteins as depicted by a one-dimensional
representation generated by the ProteoView software (Beckman Coulter), see Figure 3.
Each vertical “track” depicts a single CF fraction, with the horizontal bands indicating the
presence of protein detected by absorbance at 214 nm.
CONCLUSIONS AND FUTURE WORK
The combination of 2D-LC chromatofocusing and hydrophobicity fractionation steps prior to
MS-based protein identification has proved highly successful in the analysis of our model
system.
We have demonstrated that proteins identified with one peptide in one RP fraction may
subsequently be observed in other fractions from the 2D-LC separation. The biological
implications of this project will only be evident when the complete dataset of RP fractions has
been analysed and the proteins identified.
ProteinLynx Global Server 2.1 (Waters Micromass MS Technologies) was used to
interrogate the data obtained from the LC-ESI-MS/MS experiments against an in-house
database containing sequences from E. coli W3110, trypsin and keratin contaminants.
The database chosen to store the experimental and protein identification results was MySQL
5.0 (http://www.mysql.com/). A program, written in the Java programming language (v
1.5.0_04), allows the user to enter a variety of experimental 2D-LC and MS parameters
used, including (but not restricted to) methods, locations of CF and RP fractions (tray and
well numbers), processing parameters, datafile locations, processed spectra etc. The
program also parses the GS protein identification results and links identifications to each
fraction. The database can then be queried, for example listing all fractions from CF and RP
separations in which a protein, by accession number, was identified.
Figure 5. Interpreted product ion spectrum of a doubly charged tryptic peptide with amino
acid residue assignments.
The ability to load milligram quantities results in protein identifications with substantially
improved sequence coverage and thus confidence in protein assignments. Where protein
sequences are not available for interrogation, the high quality of the tandem MS data in
combination with the higher m/z peptides observed, allow longer regions of peptides to be de
novo sequenced.
Relational database model
We have combined a two-dimensional liquid chromatography (2D-LC) protein separation
system with mass spectrometry-based identification using a well characterised commercially
available cell lysate.
Further resolution of the proteins is achieved by a second dimension of protein separation
according to hydrophobicity.
Each fraction from the chromatofocusing column is
sequentially applied to a reverse phase (RP) column under aqueous conditions. An organic
solvent gradient is used to elute the proteins from the column with the hydrophilic proteins
emerging initially from the column followed by those having a greater hydrophobic nature.
A commercially available Escherichia coli (strain W3110) cell lysate containing 3 mg of
protein was applied to a chromatofocusing HPCF column equilibrated in CF start buffer,
using the PF2D protein fractionation system (Beckman Coulter, U.S.A.). Proteins were
resolved over the pH range 8.5 to 4.0 using the proprietary elution buffer followed by a 1 M
sodium chloride eluate. Fractions were collected either by pH interval or volume.
The vast quantity of data generated for each CF and RP fraction ensures the absolute
requirement for bioinformatic handling of the biological information.
Proteins that elute from the CF column at differing pH values than their predicted pI have been
selected for further characterisation.
Figure 4. Global Server protein identifications from a single reverse phase fraction from the
2D-LC separation. The upper left pane indicates that multiple protein species are present.
The ability to produce intact proteins after the RP separation ensures that post-translational
modification mapping can be undertaken on both the digested and full length protein. The
latter approach is currently being explored in a high throughput manner.
!
We propose to introduce real-time database searching and exclusion of identified proteins onthe-fly to encourage the observation of the lower abundance protein species.
The RMMs of proteins identified to date range from 8 KDa to 100 KDa
!
Abundant peptides have been observed containing more than 3 charges, a number having
theoretical masses greater than 3.5 Kda.
!
The product ion spectra (MS/MS) of higher m/z peptides can be interpreted to yield good
candidate amino acid sequences (de novo), see Figure 5.
!
The majority of proteins elute near to their expected pI with reproducible RP retention
times, but some proteins exhibit differing properties to those expected (see poster TP28 for
further details).
!
Abundant proteins may elute across adjacent fractions from both the CF and RP columns.
Frequently a protein may be identified in one fraction with only one peptide, to be
subsequently identified in the next fraction by significantly more peptides.
!
!
We will incorporate more directed mass spectrometry approaches to the identification of
peptides e.g. neutral loss trigger for data directed acquisition on phosphorylated peptides.
We plan to extend our studies to include biological systems that are incompatible with gelbased separation methods.
Due to the complexity of the fractions analysed, protein quantitation may be an issue. We plan
to assess the suitability of iTRAQ technology (Applied BioSystems, U.S.A.) for the quantitation
of protein expression levels in proteomic studies in combination with 2D-LC protein
fractionation.
REFERENCES
Sequence coverage can approach 50 % on proteins having RMMs from 9 to 35 K Da.
Using the database, we have identified two doubly charged isobaric peptides of m/z 965
Da with identical protein elution profiles from the RP column, but differing in sequence
(AFTSEEFTHFLEELTK and LVDKVIGITNEEAISTAR) for ion mobility separation and
MS/MS identification. This would be achieved in the first and/or third T-Wave ion guides of the
Synapt HDMS system generating first and second generation fragment ion spectra.
Lubman, D.M. et al. (2002). Journal of Chromatography B. 782 (1-2) 183 -196.
Zheng, S. et al. (2003). Biotechniques 35 (6) 1202-1211.
Zhu, K. et al. (2004). Journal of Chromatography A.1053 (1-2) 133 -142.
Giles K, Pringle SD, Worthington KR, Little D, Wildgoose JL and Bateman RH, Rapid
Commun. Mass Spectrom. 18 (2004) 2401.
Levreri, I. et al. (2005). Clinical Chemistry and Laboratory Medicine 43 1327-1333.