Top Down Proteomics: Facts and Perspectives

Abstract

The rise of the “Top Down” method in the field of mass spectrometry-based proteomics has ushered in a new age of promise and challenge for the characterization and identification of proteins. Injecting intact proteins into the mass spectrometer allows for better characterization of post-translational modifications and avoids several of the serious “inference” problems associated with peptide-based proteomics. However, successful implementation of a Top Down approach to endogenous or other biologically relevant samples often requires the use of one or more forms of separation prior to mass spectrometric analysis, which have only begun to mature for whole protein MS. Recent advances in instrumentation have been used in conjunction with new ion fragmentation using photons and electrons that allow for better (and often complete) protein characterization on cases simply not tractable even just a few years ago. Finally, the use of native electrospray mass spectrometry has shown great promise for the identification and characterization of whole protein complexes in the 100 kDa to 1 MDa regime, with prospects for complete compositional analysis for endogenous protein assemblies a viable goal over the coming few years.

Proteomics in a Post-Genomics World

The rise in genome sequencing has greatly propelled the understanding of the living world, but alone is insufficient for full description of a biological system [1]. Focusing on the protein level, proteomics has emerged as another large-scale platform for improving the understanding of biology. Proteomic experiments can be used for the annotation and correction of genome sequences, quantitation of protein abundance, detection of post-translational modifications (PTMs), and identification of protein-protein interactions [2]. In many ways proteomics can serve as an important complement to genomics and transcriptomics [1]. For example, while mRNA abundance differences between cellular states can be routinely monitored, these levels may not be indicative of protein levels due to controls over protein translation and degradation. In certain systems, including extracellular fluids or subcellular organelles, transcript levels are of significantly less interest than protein abundance. Additionally, protein activity, perhaps the most important factor in understanding biological pathways, may be precisely regulated by post-translational modifications [1].

Top Down Proteomics

While a variety of methods, including cell imaging and protein arrays, are capable of large-scale protein study, mass spectrometry-based approaches are uniquely well suited in terms of throughput and sensitivity to handle proteome-wide investigations [2]. Mass spectrometry-based proteomics has traditionally been carried out in a Bottom Up approach [3,4]. This entails the chemical or enzymatic digestion of proteins prior to their introduction to the mass spectrometer. The detection and typically fragmentation-based identification of the peptides allows for the inferred identification of the original protein. Immediately several disadvantages to this approach become clear: a peptide or even several peptides may not be specific to an individual protein or protein form, large regions of the protein may not be identified which can leave behind important information regarding PTMs or sequence variants, and modifications or sequence variations may occur on disparate peptides, causing their relation to one another to be lost following digestion. Top Down mass spectrometry seeks to eliminate these problems by introducing the intact protein into the mass spectrometer where both its intact and fragment ions masses are measured (Fig. 1). This approach routinely allows for 100% sequence coverage and full characterization of proteoforms, the specific molecular form of the protein resulting from combinations of genetic variation, alternative splicing, and post-translational modifications [5].

Fig. 1.

Comparison of Top Down and Bottom Up mass spectrometry [3]. In the traditional Bottom Up approach, intact proteins are digested into peptides prior to introduction into the mass spectrometer where they are then detected and fragmented. In Top Down mass spectrometry, the protein is ionized directly, allowing for improved sequence coverage and detection of PTMs.

The potential for achieving full protein characterization has made the Top Down mass spectrometry approach extremely useful for analysis of single proteins or simple mixtures of significant biological interest [6,7,8,9,10,11]. However, the technical difficulty of proteome-wide analysis at the intact protein level has caused Top Down proteomics to lag behind Bottom Up in terms of proteome coverage, sensitivity, and throughput. However, recent advances in separations, mass spectrometry instrumentation, and tailored bioinformatic tools have propelled the Top Down approach towards becoming a powerful complement and perhaps a viable alternative to digestion-based approaches.

Intact Protein Separation Methods

The great complexity within most proteomic samples requires that they be fractionated prior to introduction to the mass spectrometer [12]. Many separation strategies can be applied off-line, or independent of the mass spectrometer [13]. This entails collection of the eluted fractions followed by their infusion into the mass spectrometer. Using this approach, more instrument time can be spent collecting data on a single protein or simple mixture. Additionally, off-line separations are more flexible as the separation conditions do not need to be mass-spectrometry compatible. In comparison, on-line separations couple directly to mass spectrometry, allowing for increased throughput and reduced sample handling but with limitations to data acquisition and separation conditions. Given the complexity of most proteomics samples, multiple separations are often required to achieve sufficient separation, often using an off-line approach coupled to an on-line separation.

Liquid Chromatography

One of the most common methods for the separation of intact proteins, peptides, and small molecules is liquid chromatography (LC). This general separation approach relies on differential partitioning of analytes between a liquid mobile phase and a stationary phase. In many cases, liquid chromatography can often be coupled to electrospray ionization (ESI), proving an effective method for on-line analysis [14]. While a variety of liquid chromatography methods have been developed, reversed-phase liquid chromatography (RPLC), hydrophobic interaction liquid chromatography (HILIC), and ion exchange chromatography (IEX) are three of the most common liquid chromatography approaches applied to intact proteins [13].

Reversed-Phase Liquid Chromatography

RPLC uses a non-polar stationary phase and a polar mobile phase, allowing the most hydrophilic analytes to elute first. Alkyl chains (C4, C5, C8, C18) linked to porous silica particles are common stationary phases, where shorter chains are generally preferred for intact proteins as these phases are less retentive and offer higher recovery [13]. Additionally, many reports have been published using derivatized nonporous silica (NPS) particles, which offer increased speed and protein recovery, but suffer from limited loading capacity and high back pressure [15,16,17,18,19,20]. Use of superficially porous particles, which contain a nonporous silica core and a porous shell, has been reported for protein separations, offering similar efficiency to that of nonporous columns but with reduced back pressure and improved loading capacity [21]. Polymeric reversed-phase materials, offering increased mechanical strength, uniform hydrophobicity, and high recovery, have also been utilized for the separation of intact proteins [13,22,23,24].

For intact proteins, RPLC is typically employed as a second dimension of separation, but it has often been successfully applied as the sole separation strategy [13,15,22,25]. For example, a 4.6 mm inner diameter derivatized nonporous silica column was used to fractionate Methanococcus jannaschii lysates prior to off-line analysis to facilitate protein detection and automated fragmentation [15]. Another study utilized a 1.0 mm polymeric column to separate histone extracts from HeLa cells [22]. The column effluent was split such that 300 nL/min was electrosprayed directly in the mass spectrometer and the remaining 99.7 μL/min was collected for later automated off-line analysis.

Hydrophobic Interaction Liquid Chromatography

In contrast with RPLC, HILIC utilizes a polar stationary phase and gradients of increasing water content, resulting in the elution of more hydrophobic species first [26,27]. Analytes partition between the mobile phase and water-enriched region surrounding the stationary phase, differing from traditional normal phase chromatography where analytes are actually adsorbed to the hydrophilic stationary phase. Membrane proteins extracted from bovine heart mitochondria have been fractionated using HILIC.[28,29] HILIC has also been extensively applied for the separation of modified histone forms prior to Top Down mass spectrometry [30,31,32].

Ion Exchange

While separation in RPLC and HILIC rely primarily on differences in hydrophobicity to achieve separation, ion-exchange chromatography (IEX) uses differences in the charge of the analyte. Increasing the ionic strength of the mobile phase is used to elute analytes from the charged stationary phase. Opiteck et al. reported the use of cation exchange coupled to on-line RPLC for the two dimensional separation of the Escherichia coli proteome [33]. Besides increasing the fractionation power of the system, the use of RPLC following IEX allows for efficient desalting prior to electrospray ionization. Strong anion exchange (SAX) coupled to RPLC-MS for intact protein separations has also been reported for the study of E. coli [34]. In this work, the authors combined Top Down and Bottom Up proteomics by collecting intact protein spectra and also digesting a portion of the protein effluent, allowing for identification. Anion exchange-RPLC has been utilized for the Top Down study of Shewanella oneidensis [35], yeast [36], and human leukocytes [37]. Chromatofocusing, a variant of IEX which uses a change in pH rather than ionic strength to achieve elution, has been coupled to off-line RPLC for fractionation of intact proteins from breast cancer [17] and Methanosarcina acetivorans [38]. In contrast to traditional IEX, chromatofocusing elutes proteins as a function of their isoelectric point (pI), a physical parameter which can be useful for protein identification.

Electrophoresis

In addition to chromatography, electrophoresis, which relies on the differential migration of proteins in an applied electric field, is an extremely popular general approach for separating intact proteins [2,13,39]. The most common electrophoretic method is SDS-PAGE, in which SDS-coated protein molecules migrate through a polyacrylamide gel matrix in an electric field achieving separation based largely on molecular weight [40]. This is commonly utilized in Bottom Up proteomics by digesting the proteins out of the gel, then performing on-line LC-MS [41,42]. This approach can be extended to two-dimensional gel electrophoresis (2-DE) in which isoelectric focusing (IEF) is used as the first dimension of separation to separate proteins according to their pI, which can be useful for resolving modified proteins, and then SDS-PAGE is performed as the second dimension [39,43]. While these gel-based approaches separate intact proteins, recovery of the protein from the gel is often very difficult and offers poor recovery [39]. Proteins can be electroeluted from SDS-PAGE gels using an orthogonally applied electric field to the gel; however few studies have been reported, likely due to poor recovery [44].

Tube Gel Electrophoresis

While traditional gel-based approaches are generally not applicable to Top Down proteomics, similar separation strategies have been applied. Continuous-elution gel electrophoresis utilizes a tube gel column to separate proteins which are then collected as they elute from the end of the gel column [16]. This approach was applied to the fractionation of the S. cerevisiae proteome using an acid-labile surfactant (ALS) rather than SDS, as it could be degraded upon acidification, limiting downstream interferences. The fractions were further separated using off-line RPLC on a 4.6 mm C4 column before MS analysis. An improvement to this original separation used a mini prep-cell, featuring a 7 mm gel column, rather than a 37 mm column, coupled to a 320 μm C4 column allowing for 15 to 300-fold less sample loading as well some on-line analysis [45].

The use of tube gel electrophoresis for protein separation was further expanded with the invention of gel-eluted liquid fraction entrapment electrophoresis (GELFrEE), shown in Fig. 2 [46]. This device differs from the prep cell in its use of a sample collection chamber, in which fractions are manually collected. Fractionating this way ensures that higher molecular weight proteins are not continually diluted and dispersed across many fractions. Additionally, utilizing a short gel column, this device offers separation in ~75% less time. The GELFrEE device was first applied to a Top Down study in 2009, utilizing SDS which was then removed using methanol/chloroform/water precipitation [47] prior to online nano-LC-MS [23]. Using a Tris-tricine [48] variant of the original separation, the authors were able to identify 35 unique proteins from a single GELFrEE separation of HeLa cells. In the last few years, several papers have been published using GELFrEE for molecular-weight based fractionation for Top Down proteomics, allowing for increased throughput and number of identifications [24,49,50,51,52].

Fig. 2.

Diagram of the GELFrEE device [46]. A gel column is utilized to achieve electrophoretic separation of proteins, analogous to SDS-PAGE, which are then eluted into the liquid-phase for manual collection. The fractionation can then be visualized by running a portion of the fractions on a SDS-PAGE gel.

Isoelectric Focusing

Isoelectric focusing (IEF) for Top Down proteomics is generally considered more difficult as proteins tend to precipitate at their isoelectric point, significantly reducing their recovery from the gel media [18]. The Rotofor device uses an IEF separation but within an open channel, where the pH gradient is formed through the use of carrier ampholytes in solution between an acidic anode and a basic cathode [53]. While precipitation can still be problematic, especially for hydrophobic proteins, recovery can be increased by the use of 8 M urea and a nonionic detergent, such as CHAPS. A preparative-scale Rotofor, capable of separating ~1 g of protein within 55 mL of buffer, and the mini-Rotofor, a smaller version of the device which can hold ~15 mL of solution, have been used for the fractionation of the human erythroleukemia cell line proteome prior to NPS-RPLC-MS [18,19,20].

Another variant of solution isoelectric focusing (sIEF) for protein fractionation was reported in 2008 [54]. This device featured a separation channel divided into 8 chambers for fraction collection, where restriction channels and glass filter membranes were used to limit bulk flow between chambers. Recovery of precipitated proteins was achieved by washing the collection chambers with 0.1% TFA. It was shown that the device could be used as a first dimension of separation prior to a multiplexed version of GELFrEE, analogous to a two-dimensional gel separation [55]. This separation strategy was utilized for the MS²-based identification of over 1,000 human proteins using a Top Down approach, by far the most ever identified at that time [50].

Capillary Electrophoresis

Another electrophoretic technique used for the separation of intact proteins is capillary electrophoresis (CE). The small capillaries (<100 μm inner diameter) used within CE allow for high separation voltages (10–30 kV) without Joule heating, thereby reducing separation time and increasing peak capacity by limiting longitudinal diffusion [56,57,58]. Capillary zone electrophoresis (CZE), the simplest separation mode in CE, utilizes differences in the electrophoretic mobility of the analytes within an open capillary to achieve separation. CZE has been the most common CE mode applied to the mass spectrometry of intact proteins [59]. However, these studies have been largely limited to the analysis of a few target proteins. Notable examples include the detection of the α and β subunits of hemoglobin from a single erythrocyte [60] and the detection of various protein glycoforms of erythropoietin, fetuin, and α1-acid glycoprotein [61]. A larger scale study reported the detection of 55/56 ribosomal proteins from E. coli [62].

Capillary isoelectric focusing (cIEF), another common CE separation mode, provides a high resolution separation within an open capillary tube which has typically been coated with polyacrylamide to reduce electroosmotic flow and protein adsorption [63,64]. The use of cIEF-MS for the analysis of intact proteins was first reported in 1998, for the detection of protein standards, human hemoglobin variants, and E. coli cell lysate [65]. After focusing, proteins were mobilized into the mass spectrometer by applying potential to the inlet electrode and raising the inlet reservoir above the electrospray needle. Improvements to the cIEF separation and data acquisition, as well as the use of isotopically-depleted growth media for improved sensitivity, allowed for the detection of 1,000 polypeptides/proteins from ~300 ng of protein [66,67]. cIEF has also been coupled to RPLC-MS, which allowed for a second dimension of separation and the removal of ampholytes which can cause significant ionization suppression [68]. Using this platform, 1200 polypeptides/proteins from Chlorobium tepidum were detected over an eight hour separation [69].

Mass Spectrometry of Intact Proteins

The detection and identification of intact proteins, especially on a proteome-wide level, depends on high performance mass spectrometers [3]. High resolution and mass accuracy are critical to separate and accurately assign spectral peaks arising from complex precursor spectra containing multiple intact proteoforms or fragmentation spectra containing hundreds of fragment ions. Extremely high resolution may be required to distinguish disulfide bridges (Δm = 2 Da), deamidation (Δm = 1 Da), trimethylation versus acetylation (Δm = 39 mDa), and phosphorylation versus sulfation (Δm = 10 mDa) [70]. Sensitivity is also vital, as high molecular weight species such as proteins will have broad isotopic distributions, distributing the signal from a single protein across many peaks [71]. Additionally, in electrospray ionization (ESI) a population of protein ions will display a distribution of charge states. Combining these two effects, the signal arising from an individual proteoform may easily be split into hundreds of channels, reducing the signal at any mass-to-charge ratio (m/z). This becomes more significant at higher mass as the number of charge states and isotope peaks increase.

Before a protein can be detected or fragmented, it must first be ionized and desorbed into the gas phase. By far, the two most common techniques for protein ionization are matrix-assisted laser desorption ionization (MALDI) [72,73] and ESI [74,75]. MALDI primarily generates singly protonated protein ions, requiring a mass analyzer capable of detecting very high m/z species. This has largely limited MALDI analysis to the use of time-of-flight (TOF) instruments. One growing application of intact protein analysis using MALDI-TOF is tissue imaging [70,76]. However, the use MALDI-TOF in high-throughput proteomics has been limited due low resolution, poor fragmentation, the requirement of relatively purified samples, and difficulty coupling to separations [2,3]. ESI generates multiply charged ions and is the preferred method for the analysis of both peptides and intact proteins, especially on a proteomics wide-scale. Taking advantage of high resolution and mass accuracy, Top Down proteomics studies have largely been implemented using ESI coupled to either Fourier transform ion cyclotron resonance (FT-ICR) or Orbitrap mass analyzers and will be the focus of further discussion.

Fourier Transform Ion Cyclotron Resonance Mass Spectrometry

Fourier transform ion cyclotron resonance mass spectrometry relies on the excitation of an ion at its cyclotron frequency within a strong magnetic field [77,78,79]. This excitation creates a spatially coherent packet of ions, which orbit at an increased radius, allowing for detection by monitoring the image current on a detection plate. The detected signal, also termed a transient, is converted from the time domain to the frequency domain through a Fourier transform, and then to m/z through mass calibration.

ESI-FT-ICR was first used for the analysis of intact proteins in 1989, with the detection of multiple charge states on a single protein, allowing for intact mass determination [80]. Further studies demonstrated isotopic resolution on proteins using a 2.8 Tesla (T) instrument, allowing for accurate mass determination [81,82]. The same instrument was also utilized to perform collision-induced dissociation (CID) and nozzle-skimmer dissociation (NSD) of ubiquitin, using the high resolving power of the instrument to determine charge state and identity of the fragment ions [83]. These fragmentation modes utilize collisions with gas molecules to fragment the protein backbone typically at the amide linkage, resulting in b- and y- type fragment ions (Fig. 3) [84]. Another technique, electron capture dissociation (ECD), has also been utilized for protein fragmentation [85]. A heated metal filament was used to introduce electrons into the ICR cell, where they are then captured in the Rydberg orbital of one of the protonated sites on the backbone of the protein, forming a radical which results in cleavage of the N-C_α bond to form c-and z^•-type ions (Fig. 1.3) [84]. In the initial report ECD of several proteins resulted in an increase in the number of backbone cleavages compared with CID [85]. CID and other collisional methods are based upon vibrational excitation, where vibrational energy can be distributed throughout the molecule, often causing the fragmentation of the weakest bonds [86]. In ECD bond-cleavage occurs before energy can be redistributed throughout the molecule, often attributed to the process being “non-ergodic” and can be utilized to achieve significantly different fragmentation patterns than collisional methods [86,87].

Fig. 3.

Comparison of collisional dissociation (CID/HCD) and electron-based dissociation (ECD/ETD) of peptides and proteins. Collisional dissociation results in cleavage of the amide-bond, resulting in b- and y-type ions. Electron-based methods cleave of the N-C_α bond, resulting in c- and z^•- type ions.

Significant advances to FT-ICR technology, often focused on improving the analysis of intact proteins, included increases in magnetic field [88,89], addition of a resolving quadrupole for mass selection [90], and an accumulation octupole for ion storage before transmission to the ICR cell [91,92]. These modifications, allowing for increased sensitivity, dynamic range, and resolution were utilized on a 9.4 T instrument for the detection and identification of proteins from M. jannaschii and S. cerevisiae [93]. This instrument design has been utilized for a variety of Top Down proteomic studies [16,38,45,94].

The construction of a linear quadrupole ion trap/FT-ICR mass spectrometer was first reported in 2004 [95]. This instrument allowed for the storage and manipulation of ions from a continuous ion source in the linear trap before injecting them into the ICR cell. Mass accuracy was improved through the use of automatic gain control (AGC), accurately controlling the number of ions that are allowed to enter the ICR cell even from variable ion flux into the instrument which is typical of LC-MS. Fragmentation is also performed within the ion trap, with fragment ions able to be detected using the high resolution and mass accuracy of the ICR analyzer or the speed of the ion trap (mostly used for peptides). This instrument was commercialized utilizing a 7 T magnet, achieving 100,000 resolving power (m/z 400) with a one second transient as well as <2 ppm mass accuracy without internal calibration.

Use of 7 T LTQ-FT-ICR for Top Down proteomics includes the analysis of the S. cerevisiae proteome [36] and a variety of membrane proteins [96,97]. A 12 T version of the instrument has been utilized for increased throughput studies using the automated on-line/offline RPLC-MS platform for the study of HeLa nuclei [22], human leukocytes [37], and M. acetivorans [98] as well as the GELFrEE platform for the analysis of S. cerevisiae [49] and HeLa extracts [23,24,50]. A 14.5 T version of the LTQ-FT-ICR instrument has also been reported, featuring approximately 4-fold higher mass accuracy and twice the resolving power of the 7T instrument [99]. Besides the increase in field strength, a wired octupole following the ion trap allowed for the storage of an increased number of ions before detection in the ICR cell. This instrument was also utilized for analysis of GELFrEE fractions, using the improvement in sensitivity and resolution for the detection and identification of higher molecular weight proteins [23,100].

Orbitrap Mass Spectrometry

A new type of Fourier transform mass spectrometer was described in 2000, the Orbitrap mass analyzer [101]. This trap features a pair of axially symmetric electrodes: a central “spindle-like” electrode and an outer “barrel-like electrode”. In this electric field, ions rotate around the central electrode while oscillating down the length of the electrode. The frequency of these oscillations is proportional to (m/z)^−1/2. Image current on the outer electrodes is monitored and the resulting time domain signal is converted to frequency and then to m/z as in FT-ICR. Also similar to FT-ICR, the Orbitrap mass analyzer has been coupled to a LTQ allowing for use of a continuous ion source (e.g., ESI), increasing mass accuracy with AGC, and enabling efficient fragmentation [102]. Coupling of the two analyzers was achieved by the use of a transfer octupole following the ion trap into a curved rf-only quadrupole (C-trap) used to eject ions axially towards the Orbitrap analyzer. This new instrument was capable of obtaining 60,000 resolving power (m/z 400) using a one second transient, achieving isotopic resolution of myoglobin and carbonic anhydrase.

The use of the LTQ-Orbitrap for more extensive examination of intact proteins was first reported in 2006 [103]. The authors demonstrated the ability of the instrument to consistently achieve <10 ppm mass accuracy on intact proteins and confidently identify the proteins using CID fragmentation (MS² and MS³). The LTQ-Orbitrap was also utilized for the study of low molecular weight proteins from human blood including the quantitation of apolipoprotein proteoforms [104] and detection of transthyretin and hemoglobin variants [105]. The LTQ-Orbitrap was also used to distinguish several glycoforms from intact recombinant antibodies (~150 kDa) and fragment the reduced light and heavy chains using CID [106]. A similar study utilized higher-energy collisional dissociation (HCD) to obtain sequence information on subunits [107]. HCD is a collisional-based fragmentation approach similar CID, but allows for the detection of low m/z fragment ions and provides higher-energy collisions which can result in more informative fragmentation spectra [108]. In the original design, HCD was achieved by colliding oscillating ions within the C-trap with nitrogen gas, but an improved design relied on an additional higher pressure octupole collision cell following the C-trap.

Significant modifications were made to the design of the linear ion trap into a new instrument branded as the LTQ Velos [109]. Improved ion injection optics allowed for ~5-fold reduction in ion injection times and the use of two linear ion traps allowed for more efficient trapping and CID fragmentation in a higher pressure trap and higher resolution scanning in a lower pressure trap. While a stand-alone LTQ Velos was capable of achieving isotopic resolution of intact myoglobin, coupling the LTQ Velos to the Orbitrap (LTQ Orbitrap Velos) provided the routine high mass accuracy and resolving power needed to advance both Bottom Up and Top Down proteomics [110]. With similar coupling of the ion traps and Orbitrap as in the previous LTQ-Orbitrap, this instrument contained an improved integrated C-trap and collision cell enabling more efficient HCD. In the initial report, the LTQ Orbitrap Velos was able to achieve isotopic resolution of carbonic anhydrase along with its confident identification using HCD fragmentation. It should also be noted that the LTQ Velos has also been coupled to a 12 T FT-ICR mass spectrometer for Top Down analysis [52], however the instrument has not been commercialized and no other uses of such an instrument have been reported.

The LTQ Orbitrap Velos has been used for the analysis of disease causing hemoglobin variants from dried blood droplets for potential clinical use [111]. Another report utilized the instrument to identify 53 proteins, many of them containing modifications including pyroglutamates, disulfide bonds, and S-glutathiolation, from the periplasm of the bacterium Novosphingobium aromaticivorans [112]. Antibodies have also been analyzed using this instrument, allowing for improved sequence coverage through the use of electron transfer dissociation (ETD) of the disulfide intact species [113]. ETD is an electron-based fragmentation technique similar to ECD, but utilizes gaseous anions to transfer low-energy electrons to protonated analytes [114]. This approach allows for electron-based dissociation using a linear ion trap, which is not adequately capable of trapping free electrons used in ECD. ETD was originally coupled to a LTQ-Orbitrap through ESI of the precursor ETD reagent into the trap and then allowing for reaction [115], but an improved design transported the ions from a negative chemical ionization source to the ion trap [116].

A compact high-field Orbitrap, coupled with an improved Velos PRO dual ion trap mass spectrometer and advanced signal processing was recently reported, capable of a nearly four-fold increase in resolution [117]. This improved resolution was used to achieve isotopic resolution of carbonic anhydrase and enolase with a sub-second transient. The utility of this instrument for Top Down proteomics was demonstrated by the identification of 690 unique proteins from the H1299 human cancer cell line [51], followed by the identification of 1976 unique proteins from an H1299 cell line used to probe the proteome-wide difference in cellular senescence[118]. SID, CID, HCD, and ETD were each utilized to achieve complementary fragmentation and improved proteome coverage.

Data Processing for Top Down Proteomics

While powerful separation devices and mass spectrometers can be used together to generate data impressive in both quality and quantity, it must be adequately processed in order to identify and characterize proteoforms. As Top Down proteomics continues to increase in throughput and complexity of the samples analyzed, it is clear that a software platform must allow for fast, automated processing of raw data. ProSight PTM was the first search engine and web application designed for the identification of intact proteins [119,120]. In absolute mass searching (Fig. 1.4), the software uses the precursor mass and mass tolerance window to generate a possible list of candidates from a larger annotated database. The theoretical fragment ions from the candidates are then compared to the experimentally determined fragment ions within a fragment mass tolerance. A P-score is calculated for each hit, representing the probability that a random sequence could account for the matching ions [121]. Sequence tag searches can also be performed, allowing for identification of proteins based on amino acid mass differences from the fragmentation data (Fig. 1.4) [119,120]. An updated version, ProSight PTM 2.0, included the ability to include fixed modifications (e.g. alkylation of cysteine residues) and terminal modifications (e.g. N-terminal acetylation) [122]. ProsightPC, a desktop-based version of the software, has also been developed and commercialized. Additional features include error-tolerant searches, which applies the difference between observed and expected precursor masses to fragment masses, and biomarker searches, which searches against all possible protein fragments within the database (Fig. 4). A high-throughput mode allows for rapid processing of LC-MS data, utilizing the algorithms THRASH [123] or Extract to determine monoisotopic neutral precursor and fragment masses prior to database searching. ProSightPC has also been implemented on a computing cluster for processing large amount of Top Down data.[50] More recently, an online database search and retrieval has been paired with ProSight with a resulting 42 orders of magnitude gain in protein characterization[124].

Fig. 4.

Search modes utilized within ProSightPC. Absolute mass searching (a) uses an intact mass window to generate candidates, whose theoretical fragments are compared within experimental data. The hash marks indicate a matching fragment ion within a set fragment tolerance. In this case the experimental data was matched to a proteoform within the database featuring a cleaved initial methionine with 131.20 Da mass shift (outlined and boxed in orange). In sequence tag searching (b), a tag (green and boxed in orange) is generated from fragmentation data which is then used to search for candidates within the database displaying the same tag. Biomarker searching (c) is similar to absolute mass searching, but searches all possible sequences of proteins in the database, in this case resulting in identification of a proteoform containing a large truncation of the N-terminus (outlined and boxed in orange).

Alternatives to ProSight have been reported, but their uses have been limited.[125,126] BIG Mascot or MascotTD, utilizes the popular Bottom Up software platform Mascot,[127] but extends the precursor mass cutoff from the typical 16 kDa to 110 kDa.[125] The software was used for the identification of several standard proteins in addition to 13 variants of human superoxide dismutase. Additionally, the identification of a 669 kDa protein was achieved using funnel-skimmer dissociation without precursor detection. Another alternative platform, precursor ion independent Top Down algorithm (PIITA), searches tandem mass spectra against all possible tandem spectra from a sequence database [126]. This program was aimed primarily at protein identification, not full proteoform characterization. The use of the software for the analysis of Salmonella typhimurium outer membrane extracts resulted in 154 protein identifications.

Native Mass Spectrometry

Current high-throughput Top Down workflows have proven extremely successful at identifying a large number of the proteins present in human cells, yet the great majority of these studies have denatured the proteins prior to their introduction into the mass spectrometer[118]. While these conditions are gentle enough to preserve many covalent PTMs, the potentially biologically relevant non-covalent protein-protein and protein-ligand interactions are mostly destroyed. Native size-exclusion chromatography[128] and ion-exchange chromatography[129] have been reported for mapping out protein complexes from endogenous samples, however, these have mostly been followed by a bottom up approach; any stoichiometric information of subunit clustering or PTMs is lost in the digestion. While the top down study of intact protein complexes has been reported since the early 1990’s, [130,131] their characterization by gas-phase monomer ejection[132], and further by fragment ions have only recently been reported[133].

Interrogation of whole protein assemblies is a rapidly expanding sub-area of biological mass spectrometry. While most work has been on homomeric assemblies in the 60–700 kDa regime, the next major step will be to target heteromeric complexes like the proteasome, trying to find ways to eject each monomeric subunit for characterization by top down tandem MS. Many groups are active in this area [134,135,136,137,138] which really has shown that complexes present in the condensed phase can be transmitted into the gas phase with high fidelity. In addition to maintaining many of the non-covalent interactions crucial to biological processes, native mass spectrometry offers the additional benefit of a lower distribution of charge states. A 2011 study found that the large distribution of charge states observed for denatured proteins >30 kDa is the major cause of the decreased signal (and corresponding decrease in identification) for these large proteins[71]. However, ESI of these proteins under native conditions significantly reduces the number of charge states observed (coupled with a corresponding decrease in overall charge) in the mass spectrum; a recent study[139] of intact norovirus P particles (877 kDa) reported only 10 observable charge states, fewer than that for a denatured spectrum of cytochrome C (12 kDa)[140]. Fig. 5 shows the decrease in overall charge and number of different charge states for a protein when moving from denaturing to native solution conditions (by adjusting the solution pH).

Fig. 5.

The charge state distribution of electrosprayed ubiquitin ions from denaturing (top) and native (bottom) solution conditions.

Conclusions

Top Down proteomics offers an alternative to digestion-based approaches, with the promise of full protein characterization on a proteome-wide scale. While the measurement of intact proteins presents many technical challenges, the field has seen tremendous advances in separations tools, mass spectrometry instrumentation, and data processing. There has been a clear trend towards miniaturization of separations and increased use on-line and multidimensional separations. With increases in scanning speed, mass spectrometers have become more capable of handling on-line analysis and data acquisition has become focused largely on protein identification, rather than solely detection. Additionally, multiple modes of ion fragmentation have been exploited for increased sequence coverage. There has also been a shift towards commercial instrumentation, most recently the Orbitrap analyzer, which does not require expensive superconducting magnets. Software has become capable of performing various search types on complex databases while taking advantage of high mass accuracy. In general, there has been a trend towards analysis of more complex proteomes (e.g., human), with decreased sample amounts, resulting in both more identifications and improved proteoform characterization. Finally new advances in native separations and the direct detection of protein complexes offers many promising new directions of study for Top Down proteomics.

• Top Down versus Bottom Up proteomics analysis

• Separations methods for Top Down proteomics

• Developments in mass spectrometrical instrumentation and fragmentation

• Native mass spectrometry