**1. Introduction**

224 Biomarker

Suzuki, N., Yasui M., Geacintov N. E., Shafirovich V. and Shibutani S. (2005). Miscoding

Tanaka, H., Fujita N., Sugimoto R., Urawa N., Horiike S., Kobayashi Y., Iwasa M., Ma N.,

*Biochemistry,* Vol.44, No.25 (Jun 28), pp. (9238-9245), 0006-2960 (Print) Takahashi, T., Nau M. M., Chiba I., Birrer M. J., Rosenberg R. K., Vinocour M., Levitt M.,

268), 0027-5107 (Print)

(Oct), pp. (254-261), 0022-3565 (Print)

Vol.25, No.5 (Apr 30), pp. (418-420), 0890-9091 (Print)

8075 (Print)

(Print)

9232 (Print)

(Print)

suppression of NF-kappa B activation. *Mutat Res,* Vol.480-481, (Sep 1), pp. (243-

events during DNA synthesis past the nitration-damaged base 8-nitroguanine.

Pass H., Gazdar A. F. and Minna J. D. (1989). p53: a frequent target for genetic abnormalities in lung cancer. *Science,* Vol.246, No.4929 (Oct 27), pp. (491-494), 0036-

Kawanishi S., Watanabe S., Kaito M. and Takei Y. (2008). Hepatic oxidative DNA damage is associated with increased risk for hepatocellular carcinoma in chronic hepatitis C. *Br J Cancer,* Vol.98, No.3 (Feb 12), pp. (580-586), 0007-0920 (Print) Tao, Y., Song X., Deng X., Xie D., Lee L. M., Liu Y., Li W., Li L., Deng L., Wu Q., Gong J. and

Cao Y. (2005). Nuclear accumulation of epidermal growth factor receptor and acceleration of G1/S stage by Epstein-Barr-encoded oncoprotein latent membrane protein 1. *Exp Cell Res,* Vol.303, No.2 (Feb 15), pp. (240-251), 0014-4827 (Print) Tedeschi, E., Menegazzi M., Margotto D., Suzuki H., Forstermann U. and Kleinert H. (2003).

Anti-inflammatory actions of St. John's wort: inhibition of human inducible nitricoxide synthase expression by down-regulating signal transducer and activator of transcription-1alpha (STAT-1alpha) activation. *J Pharmacol Exp Ther,* Vol.307, No.1

epithelium in oral lichen planus. *J Oral Pathol,* Vol.2, No.1 pp. (46-57), 0300-9777

Formation of 8-nitroguanine by the reaction of guanine with peroxynitrite in vitro.

with peroxynitrite in vitro and its rapid removal from DNA by depurination. *FEBS* 

Induction of inducible nitric oxide synthase by Epstein-Barr virus B95-8-derived LMP1 in Balb/3T3 cells promotes stress-induced cell death and impairs LMP1 mediated transformation. *Oncogene,* Vol.21, No.52 (Nov 14), pp. (8047-8061), 0950-

microbial pathogenesis. *J Pharmacol Sci,* Vol.98, No.2 (Jun), pp. (117-129), 1347-8613

Trinchieri, G. (2011). Inflammation in cancer: a therapeutic target? *Oncology (Williston Park),*

Tyldesley, W. R. and Appleton J. (1973). Observations on the ultrastructure of the

Yermilov, V., Rubio J., Becchi M., Friesen M. D., Pignatelli B. and Ohshima H. (1995a).

Yermilov, V., Rubio J. and Ohshima H. (1995b). Formation of 8-nitroguanine in DNA treated

Yu, J. S., Tsai H. C., Wu C. C., Weng L. P., Li H. P., Chung P. J. and Chang Y. S. (2002).

Zaki, M. H., Akuta T. and Akaike T. (2005). Nitric oxide-induced nitrative stress involved in

*Carcinogenesis,* Vol.16, No.9 (Sep), pp. (2045-2050), 0143-3334 (Print)

*Lett,* Vol.376, No.3 (Dec 4), pp. (207-210), 0014-5793 (Print)

The state of the organism is reflected to the key process in the living body - protein metabolism. Proteomics is the large-scale study of gene expression at the protein level, which will ultimately provide direct measurement of protein expression levels and insight into the activity state of all relevant proteins (Pandey & Mann, 2000). The proteome analysis usually includes the following strategies: native protein pre-separation, then digestion followed by separation and identification, or alternatively straight digestion, separation and identification by mass spectrometry. Therefore, starting with one protein, after digestion we will end up with approximately 30 to 70 short peptide fragments. Identification of only very few of them will provide sufficient information which protein was present in the sample. The subproject of proteomics, namely the study of all peptides expressed by a certain cell, organ or organism, is termed peptidomics. The term was introduced in 2001 (Clynen et al., 2003). Peptides often have very specific functions as mediators and indicators of biological processes. They play important roles as messengers, *e.g.*, as hormones, growth factors, and cytokines, and thus have a high impact on health and disease. Peptidomics comprises not only peptides, originally synthesized by an organism to perform a certain task, but also degradation products of proteins (degradome). Therefore, proteolytic cleavage of proteins leads to peptides as indicators of protease activity, degradation, and degeneration therefore it is also reflects the organism state. The sensitivity of proteomics and peptidomics suffers from the lack of an amplification method, analogous to the polymerase chain reaction, to reveal and quantify the presence of low-abundance proteinaceous constituents therefore the display level is difficult. These challenges motivate the researches to develop reliable analytical platforms. Shortcomings in throughput are due to the absence of technologies that can deliver fast and parallel quantitative analysis of complex peptide distributions in an automated fashion. In the future, when peptidomics will be more analyzed and understood, and biomarkers identified straight capture step of biomarkers from complex bio-sample might be used. Peptidomics especially challenges the need for robust, automated, and sensitive high-throughput technologies. Most single-dimension separations lack sufficient resolution capability to resolve complex biological matrixes. For example, in human blood serum, 90 % of the protein content of serum is composed of 10 basic proteins. The remaining 10 % of serum consists of trace amounts of millions of different proteins. Thus, partial

Profiling of Endogenous Peptides by Multidimensional Liquid Chromatography 227

small peptides are not detected (Issaq, 2001). Additional methodologies and techniques in sample preparation, selective enrichment, high resolution separation, and detection need to be developed which would allow one to achieve even higher resolution than 2D-PAGE. Acceptable sensitivity to detect the low-abundant proteins is also still an issue. LC can address some of the above-mentioned problems. In comparison with gel-based separation methods, sample handling and preparation are simplified and automated. MD-LC has a number of advantages, such as a higher sensitivity, faster analysis time, variable sample size (preconcentration of the target substances is possible), possesses a large number of separating mechanisms, and, what is most important, and it is amenable to automation. However, because of the wide dynamic range, no single chromatographic or electrophoretic procedure is likely to resolve a complex mixture of cell or tissue proteins and peptides. Liquid chromatographic techniques are fast, quantitative, easy to automate, and can be coupled more readily to mass spectrometry than two-dimensional gel electrophoresis (Premstaller et al., 2001). The drawback of LC is the limited peak capacity of a single column. Thus, multidimensional LC is the choice, fractionating the eluent and transfer the fractions between different columns through automated valve switching (Cortes, 1990). Mass spectrometry has limitations with respect to sensitivity, therefore, a certain number of analyte molecules should be injected in order to be identified. Thus, higher amounts of the sample should be applied. Knowing the target analyte concentration in the sample provides the answer to the question: how much we should inject? In other words, the mass loadability of columns in the multidimensional column train plays a significant role, otherwise displacement phenomena and unwanted proteinprotein-interactions will take place, which may change the down-stream composition of the individual fractions in an irreproducible way (Willemsen et al., 2004). Another important prerequisite for the suitability of a separation system for proteomic analysis is the ability to handle very small amounts of biological material (Premstaller et al. 2001). These methods allow one to detect low concentrations of peptides from complex mixtures

Multidimensional (multistage, multicolumn) chromatography had been discovered early as a powerful tool to separate complex mixtures. Two of the protagonists were J.C. Giddings (Giddings, 1984, 1995) and J.F.K. Huber (Huber & Lamprecht, 1995). MD-LC is based on coupling columns in an on-line or off-line mode, which are operated in an orthogonal mode, i.e. separate the sample mixture by different separation mechanisms. The sample separated on the first column (first dimension) is separated into fractions which can then be further treated independently of each other. The practical consequence is an enormous gain in peak capacity (number of peaks resolved at a given resolution) and the potential of independent optimization of the separation conditions for each fraction. Simultaneously, there is the

Multidimensional LC separation typically relies on utilizing two or more independent physical properties of the peptides to fractionate the mixture into individual components. Physical properties commonly exploited include are size, shape, charge, hydrophobicity and biomimetic or affinity interactions. These processes are the underlying phenomena for peptide/protein separations using different chromatographic modes, such as size exclusion,

option of relative enrichment/depletion and peak compression by fractionation.

reversed phase, cation/anion exchange and hydrophobic interaction columns.

with a high degree of automation.

purification of proteins is necessary so that proteins in trace amounts can be identified and their exact structural analysis can be performed. Chromatographic separation techniques are well suited for the analysis of complex multi-components samples. To overcome the limited peak capacity and concentration diversities of the analytes utilizing chromatographic separation systems, multidimensional chromatography (MD-LC) has been realized by analyte transfer between different separation modes through automated valve switching (Link, 2002). Another important prerequisite for the suitability of separation systems for proteomic analysis is the ability to handle very big and very small amounts of biological material (Machtejevas et al., 2006). However, the application of several orthogonal LC separation systems also bears the danger of severe sample losses due to adsorption on the separation and capture column and sample transfer. The mass loadability of LC columns is much higher than for 2D-gelelectrophoresis systems and can be tuned to the requirements of a MD system. LC modes can be implemented into the sample clean-up which in return becomes more selective, robust and reproducible, thus enhancing the quality of the final data. The most important feature is however, that MD-LC can be automated with a high degree of robustness and reproducibility.

Most sample-preparation procedures are performed manually and are thus time-consuming and laborious. On-line sample clean-up and on-column concentrations avoid this disadvantage. There are a number of important features that is gained by having the liquid phase separation system to be operated on-line. Direct injection techniques are generally preferable, since problems involved in off-line sample pretreatments, such as time consuming procedures, errors and risk for low recoveries can be readily avoided. Introduction of the Restricted Access Materials (RAM) offers a unique and intelligent solution. It designates a support family that allows direct injection of biological fluids by limiting the accessibility of interaction sites within the pores to small molecules only. The term restricted access material is a general term for a packing material having a hydrophobic interior covered by a hydrophilic barrier. The hydrophilic barrier allows passage of small molecules to the hydrophobic part of the stationary phase, while sterically preventing large molecules, such as proteins, from interacting with this part of the stationary phase. Macromolecules are excluded and may interact only with the outer surface of the particle support coated with hydrophilic groups, which minimizes the adsorption of matrix proteins.

In a search for new stationary-phase configurations the concept of monolithic silica stationary phases was explored and investigated in depth (Unger et al., 2011). A monolith consists of a continuous rod, of a rigid, porous polymer, that has no interstitial volume but only internal porosity consisting of micro-and macropores. All of the mobile phase is forced to flow through the channels of the porous separation medium, resulting in enhanced mass transport also improved chromatographic efficiency (Meyers & Liapis, 1999) and simultaneous extension of column life time.

### **2. Multidimensional LC/MS approaches in proteomics and peptidomics**

Two-dimensional gel electrophoresis (2D-PAGE) and mass spectrometry are wellestablished and the most employed techniques in proteomics today. 2D-PAGE, however, provides limited information of the total amount of proteins. Low abundant proteins and

purification of proteins is necessary so that proteins in trace amounts can be identified and their exact structural analysis can be performed. Chromatographic separation techniques are well suited for the analysis of complex multi-components samples. To overcome the limited peak capacity and concentration diversities of the analytes utilizing chromatographic separation systems, multidimensional chromatography (MD-LC) has been realized by analyte transfer between different separation modes through automated valve switching (Link, 2002). Another important prerequisite for the suitability of separation systems for proteomic analysis is the ability to handle very big and very small amounts of biological material (Machtejevas et al., 2006). However, the application of several orthogonal LC separation systems also bears the danger of severe sample losses due to adsorption on the separation and capture column and sample transfer. The mass loadability of LC columns is much higher than for 2D-gelelectrophoresis systems and can be tuned to the requirements of a MD system. LC modes can be implemented into the sample clean-up which in return becomes more selective, robust and reproducible, thus enhancing the quality of the final data. The most important feature is however, that MD-LC can be automated with a high

Most sample-preparation procedures are performed manually and are thus time-consuming and laborious. On-line sample clean-up and on-column concentrations avoid this disadvantage. There are a number of important features that is gained by having the liquid phase separation system to be operated on-line. Direct injection techniques are generally preferable, since problems involved in off-line sample pretreatments, such as time consuming procedures, errors and risk for low recoveries can be readily avoided. Introduction of the Restricted Access Materials (RAM) offers a unique and intelligent solution. It designates a support family that allows direct injection of biological fluids by limiting the accessibility of interaction sites within the pores to small molecules only. The term restricted access material is a general term for a packing material having a hydrophobic interior covered by a hydrophilic barrier. The hydrophilic barrier allows passage of small molecules to the hydrophobic part of the stationary phase, while sterically preventing large molecules, such as proteins, from interacting with this part of the stationary phase. Macromolecules are excluded and may interact only with the outer surface of the particle support coated with hydrophilic groups, which minimizes the adsorption of

In a search for new stationary-phase configurations the concept of monolithic silica stationary phases was explored and investigated in depth (Unger et al., 2011). A monolith consists of a continuous rod, of a rigid, porous polymer, that has no interstitial volume but only internal porosity consisting of micro-and macropores. All of the mobile phase is forced to flow through the channels of the porous separation medium, resulting in enhanced mass transport also improved chromatographic efficiency (Meyers & Liapis, 1999) and

**2. Multidimensional LC/MS approaches in proteomics and peptidomics** 

Two-dimensional gel electrophoresis (2D-PAGE) and mass spectrometry are wellestablished and the most employed techniques in proteomics today. 2D-PAGE, however, provides limited information of the total amount of proteins. Low abundant proteins and

degree of robustness and reproducibility.

simultaneous extension of column life time.

matrix proteins.

small peptides are not detected (Issaq, 2001). Additional methodologies and techniques in sample preparation, selective enrichment, high resolution separation, and detection need to be developed which would allow one to achieve even higher resolution than 2D-PAGE. Acceptable sensitivity to detect the low-abundant proteins is also still an issue. LC can

address some of the above-mentioned problems. In comparison with gel-based separation methods, sample handling and preparation are simplified and automated. MD-LC has a number of advantages, such as a higher sensitivity, faster analysis time, variable sample size (preconcentration of the target substances is possible), possesses a large number of separating mechanisms, and, what is most important, and it is amenable to automation. However, because of the wide dynamic range, no single chromatographic or electrophoretic procedure is likely to resolve a complex mixture of cell or tissue proteins and peptides. Liquid chromatographic techniques are fast, quantitative, easy to automate, and can be coupled more readily to mass spectrometry than two-dimensional gel electrophoresis (Premstaller et al., 2001). The drawback of LC is the limited peak capacity of a single column. Thus, multidimensional LC is the choice, fractionating the eluent and transfer the fractions between different columns through automated valve switching (Cortes, 1990). Mass spectrometry has limitations with respect to sensitivity, therefore, a certain number of analyte molecules should be injected in order to be identified. Thus, higher amounts of the sample should be applied. Knowing the target analyte concentration in the sample provides the answer to the question: how much we should inject? In other words, the mass loadability of columns in the multidimensional column train plays a significant role, otherwise displacement phenomena and unwanted proteinprotein-interactions will take place, which may change the down-stream composition of the individual fractions in an irreproducible way (Willemsen et al., 2004). Another important prerequisite for the suitability of a separation system for proteomic analysis is the ability to handle very small amounts of biological material (Premstaller et al. 2001). These methods allow one to detect low concentrations of peptides from complex mixtures with a high degree of automation.

Multidimensional (multistage, multicolumn) chromatography had been discovered early as a powerful tool to separate complex mixtures. Two of the protagonists were J.C. Giddings (Giddings, 1984, 1995) and J.F.K. Huber (Huber & Lamprecht, 1995). MD-LC is based on coupling columns in an on-line or off-line mode, which are operated in an orthogonal mode, i.e. separate the sample mixture by different separation mechanisms. The sample separated on the first column (first dimension) is separated into fractions which can then be further treated independently of each other. The practical consequence is an enormous gain in peak capacity (number of peaks resolved at a given resolution) and the potential of independent optimization of the separation conditions for each fraction. Simultaneously, there is the option of relative enrichment/depletion and peak compression by fractionation.

Multidimensional LC separation typically relies on utilizing two or more independent physical properties of the peptides to fractionate the mixture into individual components. Physical properties commonly exploited include are size, shape, charge, hydrophobicity and biomimetic or affinity interactions. These processes are the underlying phenomena for peptide/protein separations using different chromatographic modes, such as size exclusion, reversed phase, cation/anion exchange and hydrophobic interaction columns.

Profiling of Endogenous Peptides by Multidimensional Liquid Chromatography 229

be approached by MD protein identification technologies (Gevaert et al., 2002, Griffin et al., 2002, Walters et al., 2001, Pang et al., 2002). The disadvantage of this approach is that one ends up with an extremely large number of peptides, which need to be resolved. However such an approach could be compared to the efforts of virtually to restore the forest look after it has been completely milled to the sawdust. No one would ague that this would require a

Another attractive approach is to separate proteins first by ion exchange chromatography (see Figure 1 b) according to charge and charge distribution under "soft" (biocompatible) conditions and collecting fractions. The fractions are subjected to digestion and consecutive re-injection on to a RP column is performed, whereby the separation is based on the hydrophobicity. This is particularly favorable since the mobile phase in the second dimension (RP) is compatible with the solvent requirements of mass spectrometry. The restrictions associated with this method lie in the limited size of proteins that can be investigated (MW < 20,000 Daltons) and the insolubility or incomplete separation of very hydrophobic peptides. All peptide-containing fractions are then investigated by mass spectrometry to generate a peptide map (Schulz-Knappe et al., 2001). This approach has already been found to be sufficient to deal with smaller subsets of the proteome (i.e. several hundred proteins) (Hille et al., 2001). These studies also clearly demonstrate that this methodology is not yet suitable for the analysis of a whole proteome due to its enormous complexity. Therefore, pre-selection of the protein from a given tissue or a pre-separation seems mandatory. For example, for the analysis of human urine solid-phase extraction (C-18 packings) to trap peptides, followed by IEX chromatography in the first dimension collecting 30 fractions and analysis of the collected fractions by RP LC (C-18) in the second dimension (Heine et al., 1997) was successfully employed. A similar procedure was used for the separation of proteins and peptides in human plasma filtrate and plasma (Richter et al.,

It is easy to be misguided by vast amount of publications usually dealing with standard protein digests. Separation of a few digested proteins peptides are shown in Figure 2a. Easy to recognize small differences in dynamic range, and even peak distribution, therefore the conclusion could be drawn, that all what we need for successful proteomics analysis is high peak capacity separation and one dimension then would be sufficient. However, using the same chromatographic conditions and column, also injecting eight times more of a real biosample (amniotic fluid) we are not observing nice and even separation any longer (Figure

Direct analysis of biofluids (Figure 1c and d) without prior digestion is a definitive option in biomarker discovery peptidomics. Those routes could be accomplished by employing restricted access materials. RAM columns possess a dual function: firstly, they operate as size-exclusion columns to remove high molecular weight proteins and other undesired constituents. The size characteristics of proteins in pure SEC are known to be highly dependant on eluent composition such as pH, ionic strength (I) of the buffer (which includes salt type and concentration) and on the flow-rate (Quaglia et al., 2006). Ionic strength and pH, however, can vary significantly among biofluids such as plasma and urine. The consequence will be that the sample cleanup procedures have to be adjusted individually with respect to each type of biological sample and standardised protocols have to be worked out. Secondly, the RAM column serves as trap or capture column to selectively enrich target

2b). This is a common situation with all real bio-samples.

lot of guessing and speculations, multiple details might be lost or misinterpreted.

1999).

Biological, individual, and variations between individuals (such as gender, age and nutrition) affect peptidomes and require careful consideration in order to find valid biomarkers. A few, equally important factors for successful proteomic biomarker research are high sample quality, high sensitivity, and reproducibility which depend on proper selection of the high quality samples.

While MD-LC MS has found widespread use in the analysis of peptides from natural sources or generated by proteolytic digestion of larger proteins, the method is not suitable for analyzing proteins directly. First, proteins tend to denature under reversed-phase conditions either by stationary phase or mobile phase induced effects (strongly hydrophobic surfaces, low pH and high organic solvent concentrations) making their quantitative elution rather difficult. Observed recoveries are also often low and life time of the columns is compromised. Also, measuring the molecular mass of a protein by MS is not sufficient for its unambiguous identification. To circumvent these obstacles the proteins are digested and the separation is performed at the peptide level. One can distinguish two approaches (i) proteins are separated and then digested ("top-down" proteomics (Wolters et al., 2001)); (ii) in "shotgun" proteomics a complex protein mixture is first digested (see Figure 1, a) and peptides are then chromatographically resolved ("bottom-up" proteomics (see Figure 1, b) (Regnier et al., 2001)). In both cases, separation technologies play a critical role in protein identification and analysis.

Fig. 1. Liquid chromatography workflow strategy options in proteomics. A – "bottom-up" approach, b – "top-down approach", c – selective sample clean-up directly combined with chromatographic separation ("digestion free" strategy), d – direct capture of target substances.

Even though in the "shotgun approach" sample complexity is vastly increased, there are an increasing number of reports on the comprehensive analysis of human proteomes using this strategy. Prior digestion gives access to the higher molecular weight proteins, however, at the expense of rendering the mixture much more complex. Assuming that a given biofluid contains 1,000 proteins and that each protein will generate approximately 50 proteolytic fragments, we are talking about 50,000 and more peptides to be resolved. This task can only

Biological, individual, and variations between individuals (such as gender, age and nutrition) affect peptidomes and require careful consideration in order to find valid biomarkers. A few, equally important factors for successful proteomic biomarker research are high sample quality, high sensitivity, and reproducibility which depend on proper

While MD-LC MS has found widespread use in the analysis of peptides from natural sources or generated by proteolytic digestion of larger proteins, the method is not suitable for analyzing proteins directly. First, proteins tend to denature under reversed-phase conditions either by stationary phase or mobile phase induced effects (strongly hydrophobic surfaces, low pH and high organic solvent concentrations) making their quantitative elution rather difficult. Observed recoveries are also often low and life time of the columns is compromised. Also, measuring the molecular mass of a protein by MS is not sufficient for its unambiguous identification. To circumvent these obstacles the proteins are digested and the separation is performed at the peptide level. One can distinguish two approaches (i) proteins are separated and then digested ("top-down" proteomics (Wolters et al., 2001)); (ii) in "shotgun" proteomics a complex protein mixture is first digested (see Figure 1, a) and peptides are then chromatographically resolved ("bottom-up" proteomics (see Figure 1, b) (Regnier et al., 2001)). In both cases, separation technologies play a critical role in protein

Digestion Digestion Sample prep

LC

Sample prep

**a) b) c)**

MS or MS/MS MS or MS/MS MS or MS/MS

Fig. 1. Liquid chromatography workflow strategy options in proteomics. A – "bottom-up" approach, b – "top-down approach", c – selective sample clean-up directly combined with chromatographic separation ("digestion free" strategy), d – direct capture of target

Even though in the "shotgun approach" sample complexity is vastly increased, there are an increasing number of reports on the comprehensive analysis of human proteomes using this strategy. Prior digestion gives access to the higher molecular weight proteins, however, at the expense of rendering the mixture much more complex. Assuming that a given biofluid contains 1,000 proteins and that each protein will generate approximately 50 proteolytic fragments, we are talking about 50,000 and more peptides to be resolved. This task can only

LC or MD-LC

MD-LC

Sample prep

LC

Sample prep AF-LC

**d)**

MS or MS/MS

LC or MD-LC

selection of the high quality samples.

identification and analysis.

Proteins

Peptides

substances.

be approached by MD protein identification technologies (Gevaert et al., 2002, Griffin et al., 2002, Walters et al., 2001, Pang et al., 2002). The disadvantage of this approach is that one ends up with an extremely large number of peptides, which need to be resolved. However such an approach could be compared to the efforts of virtually to restore the forest look after it has been completely milled to the sawdust. No one would ague that this would require a lot of guessing and speculations, multiple details might be lost or misinterpreted.

Another attractive approach is to separate proteins first by ion exchange chromatography (see Figure 1 b) according to charge and charge distribution under "soft" (biocompatible) conditions and collecting fractions. The fractions are subjected to digestion and consecutive re-injection on to a RP column is performed, whereby the separation is based on the hydrophobicity. This is particularly favorable since the mobile phase in the second dimension (RP) is compatible with the solvent requirements of mass spectrometry. The restrictions associated with this method lie in the limited size of proteins that can be investigated (MW < 20,000 Daltons) and the insolubility or incomplete separation of very hydrophobic peptides. All peptide-containing fractions are then investigated by mass spectrometry to generate a peptide map (Schulz-Knappe et al., 2001). This approach has already been found to be sufficient to deal with smaller subsets of the proteome (i.e. several hundred proteins) (Hille et al., 2001). These studies also clearly demonstrate that this methodology is not yet suitable for the analysis of a whole proteome due to its enormous complexity. Therefore, pre-selection of the protein from a given tissue or a pre-separation seems mandatory. For example, for the analysis of human urine solid-phase extraction (C-18 packings) to trap peptides, followed by IEX chromatography in the first dimension collecting 30 fractions and analysis of the collected fractions by RP LC (C-18) in the second dimension (Heine et al., 1997) was successfully employed. A similar procedure was used for the separation of proteins and peptides in human plasma filtrate and plasma (Richter et al., 1999).

It is easy to be misguided by vast amount of publications usually dealing with standard protein digests. Separation of a few digested proteins peptides are shown in Figure 2a. Easy to recognize small differences in dynamic range, and even peak distribution, therefore the conclusion could be drawn, that all what we need for successful proteomics analysis is high peak capacity separation and one dimension then would be sufficient. However, using the same chromatographic conditions and column, also injecting eight times more of a real biosample (amniotic fluid) we are not observing nice and even separation any longer (Figure 2b). This is a common situation with all real bio-samples.

Direct analysis of biofluids (Figure 1c and d) without prior digestion is a definitive option in biomarker discovery peptidomics. Those routes could be accomplished by employing restricted access materials. RAM columns possess a dual function: firstly, they operate as size-exclusion columns to remove high molecular weight proteins and other undesired constituents. The size characteristics of proteins in pure SEC are known to be highly dependant on eluent composition such as pH, ionic strength (I) of the buffer (which includes salt type and concentration) and on the flow-rate (Quaglia et al., 2006). Ionic strength and pH, however, can vary significantly among biofluids such as plasma and urine. The consequence will be that the sample cleanup procedures have to be adjusted individually with respect to each type of biological sample and standardised protocols have to be worked out. Secondly, the RAM column serves as trap or capture column to selectively enrich target

Profiling of Endogenous Peptides by Multidimensional Liquid Chromatography 231

The primary criteria for the choice of a separation phase system are selectivity and orthogonality, mass loadability, and biocompatibility (in case of quantitation). As a rule of thumb, the first dimension should possess a high mass loadability combined with sufficient selectivity and maintenance of bioactivity. Ion exchange chromatography (IEC) therefore is the method of choice offering charge selectivity. In principle, there are two options in IEC, either to employ a cation or anion exchanger, which in return influences the pH working range. Note that either cationic or anionic species are resolved, i.e. only a limited number of species from the whole spectrum. The IEC columns are operated via salt gradients with increasing ionic strength. Consequently, the salt load must be removed before the fractions

It is most common to use reversed phase chromatography as the second dimension. The term RP stands for a number of columns with different degrees of hydrophobicity. The most commonly applied phases are n-octadecyl bonded silicas (RP-18 columns). An intrinsic feature of RP columns is their desalting property. Salts are eluted at the front of the chromatogram, when running a gradient elution with an acidic buffer/acetonitrile mobile phase with increasing acetonitrile content. The hydrophobic surface of the RP packing and the hydrophobic eluent are not favorable with respect to providing a biocompatible environment for proteins: they may change their conformation or denature which may be seen by the appearance of broad peaks, splitting of peaks etc. RP columns possess a much lower mass loadability than IEC columns (10 mg of protein per gram of packing as compared to 100 mg in IEC). An advantage of RP is the fact that the eluents are compatible

In case of an on-line MD-LC system, the speed of analysis in the second dimension should be as high as possible (Wagner et al., 2002). This, however, conflicts with the requirement of high resolution or high peak capacity. The highest peak capacity in gradient elution RPC is obtained with a shallow gradient at relatively low flow-rate. Thus, a compromise between the desired peak capacity and the gradient time is inevitable. Often gradient times of several

A question often arising is: How many dimensions do we need in MD-LC? It becomes obvious that as the number of dimensions increases, the peak capacity will increase. In an ideal case the total peak capacity of the MD-LC system is equal to the product of the individual orthogonal dimensions. At the same time, above two dimensions an on-line MD-LC system becomes very sophisticated in its instrumental setup and may be difficult to control. The major goal in proteomics for the common user is to design a highly efficient, error minimizing and easy-to-handle system. Reduction of the system complexity is the major demand. It is essential to select a minimum number of dimensions to handle complex

Most sample-preparation procedures are performed manually and are thus time-consuming and laborious. On-line sample clean-up and on-column concentrations avoid this disadvantage. There are a number of important features that is gained by having the liquid phase separation system to be operated on-line. The overall yield in most cases is improved

separations which also should preferably include on-line sample clean-up steps.

to MS, provided volatile buffers such as ammonium acetate are employed.

hours are applied for the analysis of peptides from protein digests.

**2.2 Advantages of on-line sample clean-up approaches** 

**2.1 Designing a MD-LC system** 

are transferred to the second dimension column.

Fig. 2. Separation of standard protein digest and real biological sample analysis: upper chromatogram - 1 µl BSA digest (1mg/ml), lower chromatogram 8 µl of filtered amniotic fluid. Conditions: column - Chromolith Performance RP-18e 100 mm x 2 mm I.D.; eluents - A: 95% H2O/5% ACN/0.1% TFA (v/v/v), B: 5% H2O/95% ACN/0.085% TFA (v/v/v); gradient - from 5% B to 50% B in 20 min; flow rate - 0.3 ml/min, detection - UV 214 nm.

compounds in a reversed phase mode or in an ion-exchange mode. By regulating the pore size of the particles, the molecular weight exclusion can be varied as well as the molecular weight fractionation range, which allows certain analytes to be trapped at the internal surface. In this case, only proteins and peptides below a certain molecular shape and size have access to the inner pore surface of the RAM, are thus retained while the larger proteins encounter only the hydrophilic, non-adsorptive outer surface, and will be flushed out in the following washing step. Of the RAM, the strong cation exchanger with sulphonic ligands (RAM-SCX) was preferably employed in the sample clean up of proteins, which proved to show an acceptable capacity towards positively charged peptides and proteins. The features described above, when elegantly combined with column switching, become a powerful tool for direct analysis in the profiling of endogeneous peptides in a fully automated, multidimensional LC platform.
