Chapter 1: Introduction Free

Enzyme-catalyzed reversible covalent protein modification is a widespread cellular event, which effects that the number of individually distinct human protein species is at least one order of magnitude in excess of the number of gene-encoded proteins. Reversible protein phosphorylation catalysed by kinases and phosphatases is probably the most abundant and most important protein modification principle. In addition, several other protein modifying/demodifying systems are found (e.g. the pairs acetylase/deacetylase, N-methyltransferase/N-demethylase). Side-by-side to the principle of covalent modification, the functional diversity of the proteome is further expanded by the formation of noncovalent complexes exhibiting a unique functionality, activity or cellular localization. Proteins may form homo- and heteromeric complexes as well as noncovalent complexes with metabolites, drugs or metals. The lifetime of a noncovalent protein complex, its localization and its function may be influenced by covalent modifications. Thus, the interplay of covalent modification and noncovalent interaction forms the basis for the eclectic abilities of the proteome as sensor and regulator element in cells. Figure 1.1 gives a graphic summary of this situation.

Annotation of the human genome resulted in the discovery 518 protein kinase and 214 protein phosphatase genes.^1,2  This means that about 3% of our genome-encoded proteins are functionally specialized for adding a phosphate group to or removing it from proteins including the kinases themselves. As a result, it is estimated that about 30% of the human proteome is phosphorylated in some way.³  In addition experimental data allow the estimation that phosphoproteins are on average phosphorylated at three sites.⁴  Therefore, protein phosphorylation is probably the most abundant reversible protein modification and it is probably the functionally most diverse and influential one.

Phosphorus is a relatively rare element, which occurs in the biosphere mainly as phosphate. Due to its restricted availability and its multiple biochemical functions, phosphate is often a growth-limiting factor. A eukaryotic cell on average contains 0.5 pmol of phosphorus, mainly in the form of organic phosphate. Its distribution over the most important compound classes of a cell is schematically displayed in Figure 1.2.

Lipids contain about one third of the phosphorus content of a cell followed by roughly equal portions (of around 20%) located in metabolites, RNA, and proteins. ATP/ADP and DNA each contain about 5% of total cellular phosphorus, whereas inorganic phosphate (Pi) represents only about 2%.

The phosphate content of proteins first became evident in the characterization of acidic nutritional proteins such as casein and vitellin,⁶  which at that time were named conjugated proteins. The next progress occurred when phosphoserine was identified as component of vitellin,⁷  a protein isolated from egg yolk. The nutritional protein casein is highly phosphorylated, which confers calcium-binding properties to it. In this way both calcium and phosphate are provided for bone formation. The main constituent of bone is calcium hydroxyapatite, which is a mixed calcium hydroxide phosphate. The recognition of nutritional phosphoproteins as sources for phosphate and calcium describes the first known function of phosphorylated proteins.

The efforts of Carl and Gerti Cori in the 1940's, and the subsequent studies of Earl Sutherland, Edwin Krebs and Edmond Fischer paved the way for the revolutionary insight, that phosphate attachment was not only an add-on to nutritional proteins, but may control the function of a key enzyme of energy metabolism: glycogen phosphorylase. This enzyme cuts α-D-glucose from glycogen liberating it as α-D-glucose-1-phosphate by concomitant attachment of inorganic phosphate. The Coris unravelled that glycogen phosphorylase is regulated in its activity by hormones and that it exists in two interconvertible forms, an active and an inactive form. Studying the action of hormones on glycogen metabolism, Sutherland discovered the role of cAMP as the first ‘second messenger’ in transmembrane signal transduction. Roughly in parallel to these investigations, Krebs and Fischer discovered the principle of reversible phosphorylation as the metabolic control mechanism underlying the activation/deactivation of glycogen phosphorylase. They found that phosphorylation at Ser15 activates glycogen phosphorylase,⁸  whereas dephosphorylation led to inactivation. An important analytical step in this discovery was tryptic digestion of glycogen phosphorylase labeled with ³²P and isolation of the radiolabeled peptides. The main radiolabeled peptide found was sequenced as KQI-pS-VR. In these years it was also discovered that kinase-catalyzed phosphorylation requires ATP and Mg²⁺ ions and that kinase-catalyzed reactions can be performed in vitro.⁹  Numerous historical and scientific details of this development are summarized in the corresponding Nobel lectures [http://nobelprize.org], for instance in those of Krebs¹⁰  and Fischer.¹¹ 

In the structural characterization of kinases two milestones were achieved using cAMP-dependent protein kinase A as model enzyme. It was the first kinase, to be completely sequenced¹²  and for which the complete 3D structure was unravelled.¹³  As an extension of phosphorylation at serine and threonine, phosphorylation at tyrosine was discovered¹⁴  and viral oncogene products were identified as tyrosine kinases. In the following, a class of receptors with intracellular tyrosine kinase activity was found, laying the ground for unravelling a widespread mechanism for transmembrane signal transduction. In prokaryotes a signal transduction pathway was discovered based on phosphorylation at histidine (phosphoric acid amide) and at aspartate (phosphoric acid ester) (for reviews see e.g.^15,16 ) which represent two chemically more labile forms of phosphate addition compared to the phosphoric acid esters at serine, threonine and tyrosine.

The effect of phosphorylation is amplified, in case a kinase is modified so that its activity is switched on or off. For instance, in a multistep kinase cascades, one kinase activates the next kinase by phosphorylation, so that an exponential amplification of the start event is achieved. Since kinase cascades are the central principle of many signalling cascades,¹⁷  phosphorylation is one of the keys for transformation of extracellular signals into intracellular responses, as mediated for example by the class of receptor tyrosine kinases. Quenching of these signals is achieved by the action of tyrosine phosphatases,¹⁸  which are a large subgroup of phosphatases.¹⁹ 

As evident from the discovery of protein phosphorylation in the context of glycogen phosphorylase regulation, addition or removal of a single phosphate group may switch the activity of a metabolic enzyme on or off. Later, this concept was extended to kinases and phosphatases, and in general to phosphorylation cascades as discovered, for instance, in signal transduction. In addition, the linkage between phosphorylation and noncovalent interaction was recognized. Mechanisms for this linkage are, for instance, that phosphorylation disrupts existing structural motifs²⁰  or creates new structures. The latter effect can be proven by the observation that phosphosites are occurring frequently in natively disordered protein regions.^21–24  Upon phosphorylation, an unstructured region may form a stable structure which enables specific noncovalent interactions with specific target proteins. Figure 1.3 schematically summarizes the functions of protein phosphorylation in metabolic regulation, signal amplification and structural reorganization.

In addition to the reversible covalent modification of serine, threonine and tyrosine, chemically more labile modifications in the form of phosphate amides at His or Lys or of phosphate anhydrides with Asp have also been identified, mostly as intermediates in phosphate transfer processes. Finally, the regulation of nucleotide phosphate binding proteins, occurring e.g. in Ras proteins or membrane-associated G-proteins can be regarded as a kind of ‘noncovalent’ protein phosphorylation. In these systems, the guanine nucleotide exchange factors introducing GTP exhibit a kinase-related function, whereas the GTPase activity corresponds to the protein phosphatase function.

One of the molecular mechanisms of cancer development is that genomic mutations hijack a signal transduction pathway, which then provokes a dysregulated overshooting cellular proliferation. The recognition of viral oncogene products as tyrosine kinases or as products interfering directly with intracellular signal transduction opened the way for a broad acceptance of the involvement of kinases and phosphatases in cellular growth control. The most impressive recent example is the recognition that the chromosomal translocation leading to the Philadelphia chromosome is the molecular cause of chronic myelogenous leukemia (CML).²⁵  The resulting mutated BCR-ABL kinase is constitutively active and its activity is sufficient to drive the overshooting proliferation of white blood cells.

The direct consequence of the involvement of kinases in cancer is the development of kinase inhibitors as anticancer drugs. These inhibitors may act by different mechanisms, such as by interfering competitively with ATP binding or by stabilizing the inactive conformation of a kinase.²⁶  Currently, a set of strategies is applied for the development of new kinase inhibitors for cancer treatment.²⁷  One of the most successful achievements in this field is the introduction of imatinib as highly specific inhibitor of the BCR-ABL kinase for the treatment of CML.^28,29  This success further stimulated the worldwide search for drugs targeted to aberrant signal transduction processes. It is estimated that currently about 1/3 of pharmaceutical research investments is related to the development of kinase inhibitors for targeted therapies of cancer and other diseases.

In the last century numerous technological innovations occurred in originally separated fields, such as mass spectrometry, liquid chromatography, protein sequencing and computer technology. At a certain level of performance and refinement, these technologies developed a synergistic potential which lifted their analytical results to a new level of productivity and significance giving rise to the field of analytical proteomics. The field of phosphoproteomics exemplifies this development, since here mass spectrometric results may directly unravel biological functions. The dynamics of the research on protein phosphorylation is shown in Figure 1.4, displaying the number of publications retrieved from the publicly accessible scientific literature database PubMed (http://www.ncbi.nlm.nih.gov/pubmed/) using the keywords ‘mass spectrometry’ and ‘phosphorylation’. It can be recognized that this field began to flourish already about 5 years after the introduction of ESI (1984) and MALDI (1987/1988).

Some milestones of the development of mass spectrometry into a key technology of protein and protein phosphorylation analysis are summarized below.

About a century ago J. J. Thomson³⁰  and F. W. Aston developed in Cambridge, UK, the principle of mass spectrometry (MS). They generated beams of electrically-charged particles in evacuated glass tubes. By exposing them to electric or magnetic fields, they recognized that besides negatively charged electron beams, there were also ‘rays of positive electricity’, which as we now know were atomic or molecular ions. By mass spectrometry the isotopic nature of the elements was discovered, which enabled the development and refinement of the atomic model. Numerous accurate masses of elements were determined by mass spectrometry using high resolution double focusing mass analysers, data which were highly useful for the advancement of the field of nuclear physics. Ionization of gaseous molecules by a beam of electrons (electron ionisation, EI) was the standard ionisation method in mass spectrometry up to the 1960's. EI is very efficient, but generates ions with high internal energy and a radical site (one electron is lost during ionization), so that for organic molecules, ionisation is in general connected with extensive fragmentation. Structural analysis of mammalian steroid hormones was one of the first successful applications of mass spectrometry in the life sciences. Through their common cholesterol-derived steran system of four annealed rings, steroid hormones show a reasonable stability under electron impact and small structural differences can be read from their fragment ions, so that steroid hormone identification can be performed by EI-MS. From the late 1930's to the 1950's the influence of hormones on mammalian reproduction was a central theme in the life sciences. In this context the first orally-active steroid hormone was synthesized³¹  and later the birth control pill was introduced.^32–34  In the late 1950's and early 1960's the first mass spectral databases were published as books, containing a large number of EI spectra of low-molecular weight natural compounds.³⁵  As mentioned above, EI spectra are fingerprint-like mass spectra composed mainly of fragment ions. This is the combined result (i) of the transfer of electronic excitation energy during the ionisation by an electron beam, (ii) of the labile character of the radical molecular ions initially formed, and (iii) of the thermal energy the analytes have acquired during vaporization. Major analytical limitations of EI are that the molecular ion signals of organic compounds are often missing and the necessity of analyte evaporation, so that analytes with poor volatility show EI spectra containing both thermal and ionization-generated fragments. Derivatization sometimes alleviates the situation but numerous organic analytes, such as biopolymers, cannot be tackled at all by EI-MS. These limitations initiated several attempts to develop softer ionisation techniques.

The first soft ionisation method, chemical ionisation (CI),³⁶  was developed in an American Oil Company, which at that time made use of EI-MS for characterization of their raw materials and products, both being extremely complex mixtures of hydrocarbons. The aim was to develop MS ionization methods with less fragmentation for a better molecular fingerprinting. The developed CI method solved this task by EI of a reactant gas followed by secondary gas-phase reactions of reactant ions with analyte molecules. CI effects a more soft ionization than EI, but it still requires a transfer of the analyte into the gaseous state prior to ionisation.

The invention of CI marks the start for two vivid decades in the development of soft ionisation methods. Field ionisation (FI) was the next innovation.³⁷  Here very soft ionisation is achieved through an extremely high electric field of a positively charged wire with activated surface, which detaches electrons from gaseous analytes. However, analytes have to be evaporated before analysis and the molecular ions formed are radicals. Separate evaporation is avoided in the related technique field desorption (FD),^38,39  where the sample is adsorbed directly on the emitter surface. Heating of the emitter makes the analytes mobile so that they move to the ionizing centres. By FD a very soft ionization is achieved and mostly ‘even electron ions’ of the type [M+H]⁺ or [M+Cat]⁺ (Cat=singly charged metal ion) are produced, which exhibit much less fragmentation than ‘odd electron ions’ carrying a radical site. However, in FD the continuous sample supply to the ionizing centres emerged as a major limitation in particular for polar compounds. From then on the development proceeded into two main directions. One was based on discontinuous (sputtering) techniques applied to solid or liquid samples and in the other line; continuously operating ionization techniques were created.

Atmospheric pressure ionisation (API) was developed, which was a variant of CI that operated at ambient pressure. The primary ionising agent in API is a corona discharge. Impressive detection limits, e.g. for detection of trace constituents in ambient air were reported.⁴⁰  The next step was the direct coupling of a liquid chromatograph (LC) to an API source.^41–43  Several instrumental variants were used, where primary ions were generated from ion species in solution (thermospray)^44,45  or primary ions were produced separately, a technique denominated APCI. At that time HPLC instruments were generally operated at flow rates around 1 ml/min, of which–as today–a large portion was water. The technical challenge of interfacing LC to a mass spectrometer was to efficiently vaporize high volumes of an aqueous/organic solvent. This was achieved by nebulizers supported by heated gas flows and/or by vacuum pumps, and by spraying the aerosol into a heated spray chamber. Using APCI, the first efficient LC-MS coupling could be set up. Primary ions were produced either by a corona discharge or by a radioactive source (⁶²Ni) placed in the spray chamber. The next milestone was achieved by the introduction of electrospray ionisation (ESI),⁴⁶  where nebulizing and ionization were performed in a combined way by applying to the spray needle a positive or negative electric potential relative to a ‘counter electrode’ which is the inlet orifice to the mass analyser system. In addition, the interface had to be modified to achieve an effective desolvation of the highly charged microdroplets on their way into the mass spectrometer. Initially, electrospray ion sources were operated at flow rates of 1–10 μl/min. In the following this set-up was superseded by nanoESI working at flow rates of 20–200 nl/min without loss of sensitivity compared to higher flow rates.⁴⁷  This progress favoured the development of miniaturized LC equipment, leading from microLC to capillary LC and finally to nanoLC. The latter is designed for performing a splitless gradient LC separation at flow rates of around 200 nl/min and column diameters of 75 to 200 μm i.d. The currently achieved nanoLC-ESI coupling is technically a very good match of two different analytical principles. NanoLC separations are characterized by high sensitivity and separation power, and the flow rates of nanoLC are similar to those of nanoESI. The invention of electrospray ionisation has been honoured by the Nobel Prize in chemistry in 2002 to John Fenn.

The basic sputtering technology is secondary ion mass spectrometry (SIMS), where a beam of (noble gas) ions is targeted to a surface to achieve ionisation of atoms or small ions present on the surface. As a sputtering technique for the analysis of organic analytes, plasma desorption (PD),⁴⁸  was introduced. PD utilizes the high energy nuclear fission products of the transuranium element Cf-252 to sputter and ionise analytes from a surface. Separate evaporation of the analyte is avoided, since evaporation and ionization occur in one step from the solid state. This technique shifted the upper molecular weight limit of biopolymers detected by MS to several 10 kDa, but typically only low absolute ion currents and poor signal-to-noise ratios could be achieved. Significantly higher ion currents could be generated by fast atom bombardment (FAB),⁴⁹  where a beam of argon atoms and ions was directed to a glycerol droplet containing the dissolved analyte. Many organic compounds show protonated molecular ions in FAB. At the time of introduction, FAB was the most powerful technique for analysing polar metabolites, peptides, small proteins or small oligonucleotides. However, the liquid matrix used in FAB also generates a high level of non-specific background ions.

Overlapping with the introduction of FAB, ionisation of organic molecules by UV lasers was tested. As an outcome of numerous experimental variations matrix-assisted laser desorption/ionisation (MALDI) was developed.^50,51  It was observed that embedding organic analytes into an excess of a matrix compound with high absorbance of the laser radiation, a ‘soft’ ionisation of the embedded analytes could be realized. In a solitary publication⁵²  intact molecular weight analyses of small to medium-size proteins using a UV laser and a variant of the MALDI method using graphite nanoparticles as matrix was described. K. Tanaka was honoured for this study in 2002 with the Nobel Prize in chemistry. Irritatingly, the pioneering and influential work of Karas and Hillenkamp, which led to the currently widely applied MALDI technique, was not highly valued.

MALDI and ESI were quickly accepted in the scientific community and these two techniques are currently the standard mass spectrometric ionisation techniques in the life sciences. After their introduction, both techniques were miniaturized, so that better results could be achieved with lower sample amounts. In nano ESI, spraying from capillaries with μm diameters and several 10 nl/min flow rate⁴⁷  increased sensitivity. In addition, coupling of nanoUPLC to ESI resulted in very high sensitivity due to the extremely small LC peak width providing a high analyte concentration. In MALDI, sample preparation from picoliter sample volumes⁵³  or production of very small sample spots on pre-structured targets⁵⁴  increased sensitivity. Recently, a class of modified MALDI matrices was described with improved ionization efficiency.⁵⁵  MALDI and ESI led to a boom of applications of mass spectrometry in the life sciences, with a primary focus on protein and peptide analytics.

In the first years of the soft ionisation techniques the glamour of ions with high molecular weight induced many studies chasing for higher and higher molecular weight ions. As proof of principle it was been demonstrated that ESI is capable of producing ions of a virus or viral complexes in the molecular weight range of Megadaltons.^56,57  Knowledge of the molecular weight of an analyte is an important point. However, detailed structural information is usually mandatory for a specific analytical result. Recently, MS/MS analyses of intact small to medium size proteins have been demonstrated using in particular electron transfer dissociation (see below). These attempts have demonstrated the ability to read out relatively long amino acid sequences from the corresponding fragment ion spectra. The bottleneck of this ‘top-down’ approach is obviously that powerful separation technologies for liquid chromatographic separation of intact proteins are still lacking.

Once the enthusiasm over high molecular weight ions had settled, the interest focused more on the detailed characterization of small ions. The ideal tool for this purpose is tandem mass spectrometry (MS/MS), which had also undergone an impressive development roughly in parallel to that of the ionization techniques. MS/MS started as tool of physico-chemists for the study of gas phase reactions of small ions and ended up as key technique in bioanalytics.⁵⁸  In MS/MS, a single molecular ion species is filtered from a mixture of ions and then is fragmented individually. In this way, a selective structural analysis of mixture components became feasible. In the form of LC-MS/MS the method becomes even more specific, since a chromatographic and a mass spectrometric purification of an analyte are combined. Using a triple quadrupole instrument and multiple reaction monitoring (MRM), highly selective quantifications of drugs, drug metabolites and environmental pollutants could be realized by LC-MS/MS. Recently, two innovative activation methods for ESI ions were developed as an alternative to collision-induced dissociation. These are electron capture dissociation (ECD) and electron transfer dissociation (ETD).⁵⁹  Both ECD and ETD make use of electron attachment to a highly charged ESI ion to effect a radical-induced fragmentation, but they differ in the mechanism of electron attachment. In ECD, a thermal electron is transferred, whereas in ETD the electron is supplied by a large aromatic anion. The latter technique is more common and requires both a CI and an ESI source.

Nowadays, almost all MS instruments used in the life sciences are tandem instruments of different kinds. In a 2 hour analytical run of a complex protein digest sample, a LC-MS/MS system can generate thousands of MS/MS spectra. Generation, annotation and documentation of these large amounts of data are only possible by the development of powerful computers, protein sequence databases and search algorithms. Thus, current mass spectrometric analysis in the life sciences requires a high-end MS instrumentation and expertise combining separation technologies, mass spectrometry and bioinformatics.

Inductively coupled plasma-MS, (ICP-MS) ^60,61  also named element mass spectrometry was developed in the 1980's. The domain of this destructive ionization technique is quantitative trace analysis. The connection to proteomics is more in the capability to analyze so-called heteroelements in proteins and drugs, which are sulfur, phosphorus, arsenic, halogens, and selenium.⁶²  In the field of phosphoproteomics, direct and quantitative phosphorus detection is a valuable additional feature for quantitative measurements. In particular the combined coupling of ICP-MS and ESI-MS to LC analysis of phosphopeptides can give simultaneous access to structural and quantitative data.⁶³  In addition, introduction of selenomethionine into recombinant proteins can serve as quantification tag in combination with ICP MS and selenium detection.⁶⁴ 

The milestones of the development of mass spectrometry into a leading technology in the life sciences are summarized in Figure 1.5.

MALDI fingerprinting of protein digests generated sets of MS data which could not be annotated manually in an efficient way. As a solution, a search tool was introduced for automated annotation of MS data in connection to a protein sequence database.⁶⁵  Shortly later, a search tool for interpretation of ESI MS/MS data⁶⁶  was introduced. As MALDI MS fingerprinting of protein digests was automated and later on a similar development led to data-directed acquisition of LC-MS/MS data, the number and size of MS and MS/MS datasets increased dramatically. Currently, a large MS/MS laboratory may record 10⁵–10⁶ peptide MS/MS spectra per day. This situation has initiated the development of a multitude of search engines. The abilities of these search engines comprise database supported protein identification, and covalent modification analysis. Proteomic mass spectrometry and mass spectrometry-based bioinformatics have undergone a highly fruitful symbiotic development in the last decade. Instrumental innovations in mass spectrometry and related software (including databases) are currently of about equal importance for the advancement of analytical proteomics.