By starting with intact mass measurement,Top Down mass spectrometry differs from the more traditional Bottom Up strategy of protein analysis. If the research objective is just the detection of a protein — or its simple identification from a database — then Bottom Up tends to be easier. However, if the aim of the study is the characterisation of the entire primary structure of a protein, including sequence and modifications, Top Down is, in many cases, more efficient. Today there are more hardware and software combinations available than ever before that enable the acquisition of Top Down data. For example,  Top Down proteomics — where hundreds of intact proteins are fragmented directly in a mass spectrometer (without prior protease treatment) — is being carried out more and more.
In this article we outline some recent developments for processing intact proteins in a highly automated and robust manner. Selected options are highlighted to give the reader a sense of this growing and increasingly important sub-field of protein analysis.

A common current application of Top Down mass spectrometry is in the measurement of endogenous proteins such as histones, biomarkers < 20 kDa, or in the development of protein-based therapeutics in drug development [1,2]. For example, Top Down approaches have been used for the characterisation of recombinant antibodies and endogenous secretory peptides [3, 4, 5]. As can be seen in Figure 1, in a targeted mode, Top Down analysis of single proteins containing multiple post-translational modifications (PTMs) has clear advantages. The approach is increasingly being used in protein-based therapeutics, where routine characterisation of deamidations, of synthetic modifications on Cys or Lys residues, and of simple glycosylation patterns is required. For example, in biopharmaceutical development, process engineers often like to change conditions or production hosts from prokaryotic to higher organisms such as yeast, plant, insect, or mammalian cells that have machinery for complex splice forms and PTMs.

Given these advantages, we envisage a steady shift from classical peptide mapping via Bottom Up mass spectrometry, to the use of 20 - 100 kDa peptides and intact proteins for precise isoform and PTM analysis of biopharmaceuticals with ~100% sequence coverage [6].

Beyond these relatively low throughput targeted applications of Top Down mass spectrometry, we also foresee a high throughput version of Top Down that could, one day, become a realistic option for many labs. For a systems approach, Top Down analysis can facilitate the understanding of protein-level complexity by dissecting isoform-relative abundances and profiling endogenous arrays of PTMs without the use of isotopic labelling for quantification [2, 7].
 

Separation science for intact proteins
Given the enormous complexity of proteomic mixtures, which can range from sub-organellar complexes to whole-cell proteomes, mass spectrometry alone is not sufficient to adequately characterise a proteome. Effective separations are critical to decrease this sample complexity and to increase the dynamic range of detection. Because of its unrivalled peak capacity, the most commonly used separation platform for intact proteins is two-dimensional polyacrylamide gel electrophoresis (2D-PAGE). However 2D-PAGE has several disadvantages, including the low recovery of intact proteins after extraction from gels. Therefore, prior to Top Down mass spectrometry, intact protein separations are generally carried out in a way that produces fractionated proteins in solution. Such separations are based on a wide range of protein-intrinsic parameters, which include charge, size, and hydrophobicity.
Solution-phase separations based on protein charge have historically been carried out by ion exchange chromatography principally  because of the widespread familiarity of the technique and its relatively high loading capacity. Although popular, the method does not yield high resolution separations nor does it predictably separate proteins according to isoelectric point. Despite this, by combining ion exchange chromatography with reversed-phase liquid chromatography and Top Down mass spectrometry, we have shown that as many as 133 protein forms could be identified from human white blood cells [8]. Alternative techniques are chromatofocusing (a derivative of ion exchange chromatography) and solution isoelectric focusing. Both these techniques use pH gradients and are capable of separating proteins with high isoelectric point correlations. They are thus promising alternatives to salt gradient ion exchange chromatography for Top Down proteomic separation [9].
The ability to separate proteins according to molecular weight (MW) is highly desirable.  Recently, a form of semi-preparative gel electrophoresis known as gel-eluted liquid fraction entrapment electrophoresis (GELFrEE) has shown highly predictable separations at high resolution [Figure 2A] [10]. Using this method, it was possible to achieve rapid partitioning of a proteome into about 20 well-resolved, discrete mass range fractions of complex proteomic mixtures in less than one hour, over mass ranges from less than 10 kDa to 250 kDa [Figure 2B]. The combination of this size-based separation technique with online reversed-phase liquid chromatography coupled to tandem mass spectrometry (RPLC-MS/MS) has promising potential for high throughput analysis of intact proteins. Indeed, comprehensive analysis using RPLC-MS/MS is becoming more feasible, even at low resolution using ion trap mass spectrometers [11, 12]. Recently, 22 proteins with MW values between 14 and 35 kDa were identified from a yeast whole cell lysate using a single online RPLC-MS/MS run [Figure 3]. In total, 231 metabolically labelled (14N/15N) protein pairs were detected [13]. In another study, 1000 bacterial proteins could be detected using cIEF and nano RPLC-MS [14].

 
Mass analysers for Top Down
There are four basic components in a typical mass spectrometer, namely sample inlet, ion source, mass analyser and detector.
Usually it is the mass analyser that defines the instrument type, with the most frequently used being Ion Trap, Time-of-Flight (TOF), Orbitrap, and Fourier Transform Ion Cyclotron Resonance (FTICR). In the context of Top Down mass spectrometry, these mass analysers vary in one or more fundamental parameters [Table 1].
Ion trap mass spectrometers are commonly used for Bottom Up protein analysis. Due to the lower resolving power and low mass accuracy of ion traps, Top Down analysis is limited to 5 - 20 kDa [11]. TOF analysers [Table 1] have been used for Top Down analysis of intact proteins between 10-30 kDa; however, often pure, concentrated samples are required [15, 16]. A hybrid quadrupole-TOF instrument is capable of unit resolution for fragment ions enabling Top Down MS/MS on proteins up to 50 kDa [4].

FTICR and Orbitrap mass analysers rely on Fourier transformation of the image current that is recorded at the detector. The Orbitrap mass analyser features electrostatic trapping and is utilised in direct infusion and RPLC-MS/MS modes, with MW limits up to 50 and 25 kDa, respectively [3]. The FTICR mass analyser relies on ion excitation under the influence of a superconducting magnet, and is capable of characterising proteins well over 100 kDa. However, these size limits decrease during RPLC-MS/MS time scales [Figure 3] [13].

Methods of ion fragmentation
Collision-induced dissociation (CID) is often used in lower resolution ion traps or quadrupoles; fragmentation products can be analysed at high resolution in FT mass analysers. Electron capture dissociation (ECD) occurs in the superconducting magnet portion of FTICR mass analysers and has proven to be a robust method for extensive characterisation of whole proteins in direct infusion analysis. More recently, the ETD technique has involved an electron transfer reaction in an ion trap, with fragmentation readout in an Orbitrap mass analyser [Table 1]
 
Computational proteomics for Top Down MS and MS/MS
There are three primary search modes that can be used in protein identification: absolute mass, biomarker, and sequence tag [Figure 4]. Absolute mass searches involve matching, within a user-specified tolerance, the observed mass to a database intact mass and then comparing observed fragment masses to those calculated from possible forms. Biomarker searching involves calculating all possible sequences that could match the observed intact mass, then comparing calculated fragments from each to the observed fragments. Sequence tag searching identifies a series of fragment ions which match amino acid additions and attempts to match the tag to warehouse sequences. Automated characterisation of protein modifications can be achieved in Top Down by creating an array of possibilities in a customised protein database. This can be done for a protein target or in a full proteomic mode. In contrast, protein identification by Bottom Up relies on matching small peptides. This often means that protein modifications remain uncharacterised because the pieces of the protein containing these modifications are simply not identified. The database construction strategy for Top Down (termed “Shotgun Annotation”) allows the construction of a database that can accurately represent protein complexity. Such custom databases can be made for protein targets or for entire proteomes.

Outlook
Top Down mass spectrometry has progressed substantially in the last decade and now provides answers for intact protein characterisation that Bottom Up simply cannot achieve, particularly when there are multiple mass shifts due to sequence differences and modifications. The targeting of abundant proteins in cells, of endogenous biomarkers, and of protein pharmaceuticals is now no longer a mere possibility but is a reality on modern plug-and-play commercial instruments.
In the next decade, optimised front end separations, sophisticated yet affordable bench top instrumentation, and robust data analysis will solidify Top Down not only as practical for proteins above 50 kDa but also viable in a high-throughput environment. Within a few years, Top Down will emerge as a method of choice for those wishing to definitively investigate proteins at the biochemical level in the context of protein therapeutics, as well as those wishing to pose targeted
questions in biomedical/clinical research, and to researchers involved in discovery-type systems biology.
When robust Bottom Up protein analysis became high-throughput it was adopted by many laboratories, no matter the difficulty. Alternative strategies must now be considered in terms of their ability to produce high value and definitive data.
Using some of these criteria, Top Down mass spectrometry can now compete with Bottom Up in some contexts. With the implementation of Top Down on a truly proteomic scale, more researchers in a whole range of laboratories will start to believe that a new world of protein analysis lies on the horizon.

References
1. Kelleher N L. et al. Journal of the American Chemical Society 1999; 121: 806-812.
2. Pesavento J J. et al. Analytical Chemistry 2006; 78: 4271-4280.
3. Zhang Z, Shah B. Anal Chem 2007; 79: 5723-5729.
4. Ren D. et al. H. S. Anal Biochem 2008.
5. Taylor S W. et al. J Proteome Res 2006; 5: 1776-1784.
6. Note that the term “sequence coverage” refers to the percent of the primary sequence represented in detected peptides. By definition, measurement of an intact protein mass means 100% sequence coverage (using the term as in classic peptide mapping). Some use this same term to indicate the percent of backbone bonds cleaved during an MS/MS sequencing experiment; used in this way “100% sequence coverage” means that fragment ions resulting from cleavage of each backbone bond were observed, and a primary sequence can be determined de novo.
7. Zabrouskov V. et al. Mol Cell Proteomics 2008; 7: 1838-1849.
8. Roth M J. et al. Analytical Chemistry 2008; 80: 2857-2866.
9. Lubman D M. et al. Journal of Chromatography B-Analytical Technologies in the Biomedical and Life Sciences 2002; 782: 183-196.
10. Tra J C, Doucette A A. Analytical Chemistry 2008; 80: 1568-1573.
11. Bunger M K. et al. An alytical Chemistry 2008; 80: 1459-1467.
12. Chi A. et al. International Journal of Mass Spectrometry 2007, 259, 197-203.
13. Parks B A. et al. Analytical Chemistry 2007; 79: 7984-7991.
14. Zhou F. et al. Analytical Chemistry 2007; 79: 7145-7153.
15. Suckau D, Resemann A. Anal Chem 2003; 75: 5817-5824.
16. Liu Z., Schey K L J. Am Soc Mass Spectrom 2008; 19: 231-238.

The authors
J F Kellie, Dr John C. Tran, Dr Ji Eun Lee, Dorothy R. Ahlf, Haylee M. Thomas, Dr Adaikkalam Vellaichamy, Prof. Neil L. Kelleher*
Technology Development Team • Center for Top Down Proteomics
University of Illinois at Urbana-Champaign
Illinois, USA

Acknowledgements
The authors thank the Packard Foundation, the Roy J. Carver Charitable Trust, the University of Illinois and the National Institutes of Health (GM 067193) which have supported the development of Top Down mass spectrometry and Top Down proteomics. Additionally, we thank Michael Senko, Bryan Parks and the many former members of the Kelleher Research Team that have all contributed to the development of tandem mass spectrometry above 10 kDa.150