Multi-dimensional Mapping of Human Blood Proteins

MULTI-DIMENSIONAL MAPPING OF HUMAN BLOOD PEPTIDES BY MASS SPECTROMETRY

Michael Jürgens*, Michael Schrader, Manfred Raida, Wolf-Georg Forssmann & Peter Schulz-Knappe
Lower Saxony Institute for Peptide Research (IPF),
Feodor-Lynen-Strasse 31,
D-30625 Hannover, Germany

*Corresponding author:
Michael Jürgens
Lower Saxony Institute for Peptide Research (IPF)
Feodor-Lynen-Strae 31
D-30625 Hannover
Germany
Phone: (49)-511-5466-214 Fax: (49)-511-5466-102
e-mail: michael_juergens@t-online.de

KEYWORDS
MALDI mass spectrometry; human blood; peptide hormones; databases; two-dimensional chromatography, proteome, 2D-gel electrophoresis

The progress in genome sequencing results in an increasing demand for sequence data of processed gene products. In human blood especially, manyfunctional important peptides are present of which only a fraction has been identified in their primary structure. We established a peptide bank from large amounts of human blood ultrafiltrate (hemofiltrate, HF) bytwo-dimensional chromatographic separation. Here, we describe the mapping of this peptide bank by mass spectrometry (MS).

All 340 fractions of the peptide bank are analysed by MALDI-MS using optimised conditions for sample preparation. In each fraction, up to 60substances are detected, resulting in a database of several thousand components characterised by elution position and molecular mass. In addition, LC/ESI-MS is used to achieve an even more comprehensive analysis including further chromatographic separation.

Up to now, several hundred peptides have been selected from the database,subsequently isolated from the peptide bank and analysed by chemical sequencing and MS methods. Many of them are fragments of plasma proteins but some 3-5 % turned out to be unknown in protein or DNA databases. Already to date, 40 % of the latter are nevertheless present in EST databases. This percentage will increase parallel to the progress of the genome project.

The comprehensive peptide mapping using a multidimensional chromatographic separation in combination with MS methods is a strong tool in linking genomic data with amino acid sequences of processed peptides in one of the most relevant biological compartments, human blood plasma. It thus complements efforts towards human proteome analysis by 2D-gel electrophoresis.

Systematic investigations to characterise human plasma proteins have been carried out for many years [1, 2]. In contrast, peptides in blood have not been investigated using systematic approaches. This is mainly due to the unfavourable protein-to-peptide ratio and the short half-life of peptide hormones in the blood. To create a peptide map from human blood, we have introduced the use of human HF [3, 4].

Human HF is generated during blood ultrafiltration treatment in patients with chronic renal failure. Since only molecules with a molecular mass smaller than about 20 ku pass the filter, HF is used as a source for the isolation of regulatory peptides [4] from blood plasma, the most comprehensive source of peptide hormones.

Compared to plasma, HF has some interesting attributes. Investigations show that HF contains peptide hormones in plasma-like concentrations [5]. Due to the ultrafilters used in hemofiltration, HF has a much lower total protein

content of about 70 mg/L compared to plasma (70 g/L), where the large molecules constitute a major portion of the protein-peptide fraction. In addition, proteases are retained by the filters and therefore the peptides

are not subjected to further proteolytic cleavage after filtration. For purification and isolation of peptide hormones, the continuous availability of the source is important. As a waste product from nephrological centres, about 2,000 L/week hemofiltrate are accessible to our laboratory.

2D-gel electrophoresis is the main tool for the characterisation of complex sources. Although very successful in many cases, this technique has some general disadvantages. Firstly, there is no exact parameter in gel-chromatographic separation like the molecular mass measured by MS. Secondly, the loading capacity is comparatively low, leading to only small amounts of substance for further analytical steps and access to the compounds in the gel is difficult. Moreover, the biological activity is usually lost during this separation. Finally, analysis of peptides with a

molecular mass smaller than about 3-5 ku cannot usually be done successfully, resulting in the loss of information on peptides in the lower molecular mass range.

Matrix-assisted laser desorption/ionisation mass spectrometry (MALDI-MS) [6, 7] is an excellent tool to achieve a fast and sensitive analysis of peptides and proteins. Even in complex mixtures, it is possible to detect a

single peptide specifically by its molecular mass. In combination with a two-dimensional liquid chromatography procedure (cation-exchange/ reverse phase chromatography) [8, 9] MALDI-MS is a powerful tool to characterise the components of a complex biological source.

The aim of this investigation was to establish a MALDI-MS-generated peptide mass database representing our peptide bank. On the basis of this database,a systematic characterisation of human blood peptides has been initiated

recently. Techniques such as collision-induced dissociation (CID) MS/MS [7, 10] with or without further purification as well as Edman sequencing or proteolytic digest pattern evaluation after purification are used. Inaddition, preparation of previously isolated and selected peptide hormones in quantities up to milligram amounts is performed using MALDI-MS as detection system [11].

Human peptide bank:
Human hemofiltrate is obtained continuously from a local nephrological centre (Nephrologisches Zentrum Niedersachsen, Hannoversch-M_nden, Germany) and processed in a standardised procedure (Fig. 1) as described in detail [8]. The peptides from batches of 800 L HF are extracted by cation-exchange chromatography (Fractogel SP 650(M), 40-90 micrometer particle size, 100 nm pore size, Merck, Darmstadt, Germany).

The collected extracts from five batches are separated in eight pools of the same cation-exchange material by increasing the pH of the eluting buffers. The resulting pools ("pH pools"), are further separated by reverse phase chromatography (Source RPC, 15 micrometer particle size, Pharmacia, Freiburg, Germany). About 340 salt free fractions are thus produced and lyophilised for further use.

Matrices for MALDI-MS:
We use up to six different matrices for MALDI-MS. Alpha-cyano-4-hydroxycinnamic acid (CHC), trans-3-indoleacrylic acid (TIA), 3-methoxy-4-hydroxycinnamic acid ("ferulic acid", FER), 2,5-dihydroxybenzoic acid ("gentisic acid", DHB) and 3,4-dihydroxycinnamic acid ("caffeic acid", DHC) are all obtained commercially from Sigma-Aldrich, Deisenhofen, Germany. 3,5-dimethoxy-4-hydroxycinnamic acid ("sinapic acid", SIN) is also obtained commercially (Fluka , Neu-Ulm, Germany). 6-desoxy-L-galactose (L(-)fucose, Fluka , Neu-Ulm, Germany) is used as co-matrix with all matrices. Currently all fractions used for the generation of the peptide mass database are measured using CHC and SIN as matrices. If necessary, the other matrices are used to complement and optimise procedures for the detection of specific peptides, especially in purification approaches.

MALDI-TOF-MS:
Aliquots of the lyophilised fractions of the peptide bank are dissolved in 0.1-4 mL acetonitrile-0.1% aqueous trifluoreacetic acid (TFA) (1:1, v/v) according to their absorption profile in reverse phase chromatography (optical density at 214 nm). The resulting analyte solutions contain an equivalent of 0.25-10 mL HF/microliter, which represents a low microgram to low nanogram range peptide content.

The inner 8 x 8 spots of a 10 x 10 spot stainless steel sample plate are loaded. The outer spots are not used because of field inhomogeneity and thus lesser mass accuracy. 1 microliter sample solution and 1 microliter matrix solution are applied by mixing on the sample plate with a pipette,followed by accelerated ambient temperature air-drying using a microventilator. The crystallisation process is thus improved and reliable automatic measurement is possible. SIN and CHC are applied as matrices.Acetonitrile-0.1% aqueous TFA (1:1, v/v) is used as solvent for the matrices and an admixture of fucose (5 mg/mL each of matrix and fucose).

Measurements are performed in linear mode with a LaserTec RBT II (PerSeptive Biosystems/Vestec, Houston, TX, USA). The instrument isequipped with a 1.2 m flight tube and a 337 nm nitrogen laser. Positive ions are accelerated at 25 kV and 30 laser shots are accumulated in an automated measurement. The 500 MHz digitizer TDS 520A (Tektronix Inc., Beaverton, Oregon, USA) is used for signal processing. The Voyager RP BioSpectrometry Workstation Version 3.07.1 (PerSeptive Biosystems , Framingham, MA, USA) is used as control software.

The automatic measurement includes a search pattern of 18 spots per sample position. The laser intensity is adjusted to signal intensity in a preset mass range. From the 18 spots per sample position, only the best measurement, i.e. that with the highest signal intensity, is saved to harddisk. Values for laser intensity, signal intensity and preset mass range have to be differently adjusted for CHC and SIN according to their specific properties [12,13]: CHC requires lower laser energies, is more sensitive and results therefore in higher overall intensities. SIN has a lower susceptibility for smaller peptides of approximately < 3 ku. Especially for complex mixtures, SIN often generates strong matrix background resulting in more difficult detection of low molecular weight compounds [13].

The time-of-flight data are externally calibrated (two-point calibrationalgorithm) for each sample plate and different sample preparations. Calibration and further data processing are done with the Voyager RP BioSpectrometry Workstation processing software (Version 3.07.1; PerSeptive Biosystems , Framingham, MA, USA) based on Grams/386 Version 3.0 (Galactic Industries Corp. , Salem, NH, USA).

LC-ESI/MS:
Routine LC-ESI/MS is carried out on a API 100 LC/MS System (Perkin-Elmer Sciex Instruments , Toronto, Canada). The HPLC system consists of a pump 140B and UV detector 785A (PE Applied Biosystems, Foster City, CA, USA).

For chromatographic separation a RP C18 column YMC ODS-AQ, 3 micrometer, 12 nm pore size 100x1 mm (Dr.A.Maisch, Ammerbuch, Germany; YMC, Schermbeck, Germany) is used. The system is combined with the series 200 autosampler (The Perkin-Elmer Corp. , Norwalk, CT, USA) for automation of large numbers of LC-ESI/MS experiments. Linear gradients are developed at 20 microliter/min using acetonitrile/TFA solvent systems.

The peptide bank from human HF consists of around 340 fractions, generated by cation-exchange and reverse phase chromatography (Fig. 1). Each fraction is analysed by MALDI-MS using CHC and SIN as matrices (Fig. 2, top left). Fig. 2 schematically summarises the construction and use of our peptide mass database. Details are given in the following.

Figure 2 Construction and use of a peptide mass database generated from human hemofiltrate representing the peptide bank. Exemplary for the approximately 340 fractions, the original MALDI-MS spectrum of fraction 22 pH pool 1 (with SIN as matrix) is shown on top left. A strong signal at m/z = 10,350 indicates the peptide hormone guanylin 22-115. The 2D map in the centre represents pH pool 1 with fractions in one dimension and related molecular masses in the other dimension. Black spots indicate molecular masses found by CHC, red spots those found by SIN and blue spots those found by both matrices. As an example of screening the peptide mass database for the mass of a known regulatory peptide (here: guanylin 22-115), an enlarged detail of this map (molecular mass range 10.0-10.5 ku) is shown on bottom left. The corresponding molecular mass is found in fractions 22 and 23 only.

Peaks are checked for multiply charged and multimer ions. CHC has a tendency to produce more multiply charged ions than SIN. However, in these complex mixtures only for larger (approx. > 7 ku) and/or abundant peptides, doubly and very rarely triply charged species are observed. The most abundant peptides sometimes form multimeric species. Peak tables (Table 1) are processed in spreadsheet programs (Microsoft Excel and Microcal Origin) to result in a 2D peptide map for every pH pool (Fig. 2, centre). For an even more comprehensive analysis of selected fractions, further chromatographic separation in LC/ESI-MS is performed (Fig. 3). These measurements are time-consuming and interpretation is more difficult because of the occurrence of multiply charged ion series and the generation of 3-dimensional datasets. We use it for validation of the peak tables since molecular masses are more accurate than in linear MALDI-MS without delayed-extraction technique.

	[M+H]+	fraction	Found with matrix		[M+H]+	fraction	Found with matrix
1	958.7	7	SIN	901	...	...	...
2	962.9	12	CHC	902	...	...	...
3	987.8	12	CHC	903	...	...	...
4	1,007.40	12	CHC	904	10,275	15	SIN
5	1,086.70	12	CHC	905	10,346	23	SIN
6	1,104.60	7	CHC	906	10,350	22	CHC + SIN
7	1,105.70	13	CHC	907	10,438	10	SIN
8	1,106.40	12	CHC	908	10,442	11	SIN
9	1,123.20	12	CHC	909	10,470	9	SIN
10	1,138.30	5	CHC	910	10,696	9	SIN
11	1,142.80	13	CHC	911	10,763	9	SIN
12	1,144.50	12	CHC	912	10,883	9	SIN
13	1,144.60	17	SIN	913	10,979	8	CHC
14	1,160.60	7	CHC	914	11,004	8	SIN
15	1,162.70	12	CHC	915	11,023	9	SIN
16	1,184.10	8	CHC	916	11,045	15	SIN
17	1,214.80	21	SIN	917	11,249	11	SIN
18	1,221.00	7	CHC + SIN	918	11,269	10	SIN
19	1,287.50	12	CHC	919	11,445	11	SIN
20	1,298.90	11	CHC	920	11,627	11	CHC
21	1,302.60	13	CHC	921	12,275	19	CHC
22	1,304.70	12	CHC	922	12,309	11	CHC
23	1,322.00	12	CHC	923	12,338	10	CHC
24	1,323.80	19	SIN	924	12,956	15	SIN
25	1,327.40	18	CHC + SIN	925	13,964	15	SIN
26	1,345.60	16	SIN	926	14,891	15	SIN
27	1,353.50	12	CHC	927	15,795	20	CHC
28	1,354.60	15	SIN	928	15,814	19	SIN
29	1,365.60	12	CHC	929	16,112	19	CHC
30	...	...	...	930	18,795	21	SIN
31	...	...	...	931	23,503	22	SIN
32	...	...	...	932	35,317	23	SIN

These pH pool peak tables indicate the reverse phase fraction of the peptide bank for each molecular mass detected. The matrix (CHC, SIN or both) used for detection of the molecular masses is given. The original Excel spreadsheets also give information about absolute and relative intensities of the peaks (left out for simplification here). Although MALDI-MS is not a quantitative method, intensities help to estimate whether a molecular mass represents a main or minor component. A 2D map for each pH pool is created using the peak table data. The maps show fractions in one dimension and the related molecular masses in the other dimension. Different colours indicate the matrix by which the molecular masses were found (see 2D map of pH pool 1 in Fig. 2, centre).

Figure 3 LC/ESI-MS from fraction 22, pH pool 1 (PE Sciex API 100, column: YMC ODS-AQ, 1x100 mm). Top trace: TIC (total ion current) for run time 34 - 45 minutes. Bottom trace: ESI-MS spectrum at 41.5 minutes (2 scans), showing the ion series for guanylin 22-115 and its calculated molecular mass M = (10,338 +/- 2) u.

As we use MALDI-MS in linear mode without delayed-extraction technique, the mass accuracy ranges between 0.05-0.4 %. This includes an observed systematic shift (up to 0.2 %) towards higher molecular masses with increasing complexity of samples. The resolution measured as ratio molecular mass/FWHM (Full Width at Half Maximum) is between 50 and 1,000. Because of insufficient suppression of matrix-related signals, only peptides with a molecular mass higher than about 600 u are taken into account. Our detection limit in routine analysis of a single peptide in a crude mixture is in the femtomole range (absolute), that means picomolar concentration in human HF in respect to our sample preparation. Most peptides detected are present in high picomolar to nanomolar concentrations. One has to be aware of these parameters when using the peptide mass database, looking for masses of specific peptides or comparing it with information from other databases.

The maps representing our peptide bank from human HF contain at least 5,000 human plasma peptides and proteins, whereas the number is limited by the actual sensitivity of determination of the methods used. Earlier investigations showed that HF contains peptide hormones in plasma-like concentrations [5]. Molecular masses found are from about 0.6 up to 67 ku (human serum albumin), the majority are < 10 ku. These maps give an overview of the number of localised blood peptides and even to some degree the quantity of single peptides. These data can be compared with other blood plasma databases such as those from 2D-gels like the 2D PAGE of human plasma in the SWISS-2DPAGE database [16]. The corresponding analysis of the peptide spectrum of other sources such as urine using multidimensional chromatography and mass spectrometry is in preparation.

Another approach to use and to complete the peptide bank in combination with mass spectrometry is peptide trapping [4], meaning systematic characterisation of human blood peptides where their specific molecular masses serve as the peptide trap. Small peptides can be characterised by collision-induced dissociation and MS/MS analysis even directly from the peptide bank fractions (Fig. 4) or at least after one subsequent purification step. Fast, computer-assisted identification [14-16] ispossible for peptides which are included in databases (first choice: SWISS-PROT ). Molecules of higher molecular mass still have to be purified to investigate them by Edman sequencing and/or protease digestion and subsequent comparison with databases.

Figure 4 CID spectrum of a parent ion (single charged species) from fraction 12 pH pool 1 (PE Sciex API III, 15 eV collision energy, collision gas: Argon). The parent ion was identified as alpha1-antitrypsin 385-394 (MGKVVNPTQK) by MS-Tag .

Furthermore, the peptide maps can also be used to search for known plasma peptides and to obtain native, bioactive material in sufficient quantities. In theses cases, purifications are monitored by MALDI-MS with respect to the molecular mass of the target [11]. Examples for this purification strategy are the isolation of the first human beta-defensin (hBD-1) [17], a new member of the chemokine family, the CC-chemokine HCC-1 [18] or the gastrointestinal peptide hormone guanylin 22-115 [11]. The enlarged detail of the 2D peptide map (Fig. 2, bottom left) represents the region where we expect the above mentioned hormone guanylin 22-115. In pH pool 1, itsaverage molecular mass of 10,337 u is found in fractions 22 and 23.

Applying the method of peptide trapping, we have been able to identify several hundred circulating peptides up to now. Most of them are fragments of known plasma proteins, but hormones, cytokines, growth factors and defensins have also been found (Table 2). Moreover, about 3-5 % are peptides with unknown primary structure. They are found to an increasing degree in EST databases (dbEST or TIGR ). In the near future, information about all precursor proteins will be available as a result of the human genome project and the important task will be to identify the processed forms present in blood.

peptide hormones	angiotensin I, guanylin (22-115), uroguanylin(89-112), atrial natriuretic factor (CDD/ANP99-126), GLP-1
cytokines, growth factors	HCC-1, IGF-1, IGF-2, osteoinductive factor, PDGF, CTAP-III, pigment endothelium derived factor
defensins	beta-defensin 1, propeptides of neutrophil defensins 1 to 3
plasma proteins	albumin, fibrinogen A, fibrinogen B, beta-2-microglobulin, zinc-alpha-2-glycoprotein, alpha-2-HS-glycoprotein (fetuin), serum amyloid protein A, haptoglobin, profilin, desmocollin, thymosin beta-4 and -beta-10, apolipoprotein C-III, uteroglobin, ubiquitin, gelsolin
transport proteins	retinol binding protein, alpha-1-microglobulin, transferrin, transthyretin, TGF-beta binding protein, IGF binding protein 2 and 3
complement factors	factor C3, factor D, factor C4A (anaphylatoxin)
enzymes, inhibitors	lysozyme, cystatin C, alpha-1-antitrypsin, pancreatic trypsin inhibitor, plasminogen, alpha-2-antiplasmin, carboxypeptidase N, inter-alpha-trypsin inhibitor component II, somatomedin B, vitronectin
matrix proteins	collagens alpha-1-(I), alpha-2-(I), alpha-3-(IV), alpha-1-(XVIII) and osteopontin

A map of more than 5,000 human plasma peptides and proteins in the molecular mass range from 0.6 - 67 ku, the majority < 10 ku, is established using a peptide bank from hemofiltrate. These peptides are characterised by molecular mass and elution position in cation-exchange and reverse phase chromatography. The primary structure of a growing number of them is characterised. This peptide bank in combination with our peptide map is a source for the isolation of functionally or diagnostically relevant peptides from human blood.

Our approach to characterise peptides from a complex biological source by the combination of two-dimensional chromatography and mass spectrometric methods is supplementary to the well established 2D-gel based analysis (cf.: 2D PAGE of human plasma in the SWISS-2DPAGE database). A precise molecular mass serves as a key parameter in our database and represents a relevant advantage compared to relative retention time in chromatography or migration indices in electrophoresis. Even when using linear MALDI-MS without state-of-the-art delayed extraction, mass accuracy is significantly better than in any gel-based technique. In addition, the reproducibility of mass spectrometric data should be very high from one laboratory to another. The robustness of MALDI-MS allows the investigation of fractions from the powerful chromatographic separation techniques for peptides, reverse-phase and ion-exchange chromatography. To obtain an extensive analysis of complex mixtures, it is generally advisable to use a combination of at least two matrices. The combination of two-dimensional chromatography and mass spectrometric methods is not restricted to the origin of the analyte. Any source such as urine [19], cerebrospinal fluid, saliva, tears or ascites is accessible to this approach.

We show in this study that it is possible to identify thousands of peptides from our peptide bank if modern, state of the art chromatographic and mass spectrometric methods, Edman sequencing and database research are used in combination. As a result, a comprehensive database of human circulating peptides will be established in the future.

This work was supported by the German government, BMBF grant No. 0311139. This paper is in partial fulfilment of the requirements of Michael Jürgens for the degree of Dr. rer. nat. at Westf·lische Wilhelms-Universtät, Münster. We thank D. Pape-Lange and H.D. Walouch for their expert technical assistance in generating the peptide bank and B. Kemper for performing the MALDI-MS analysis.

1. Hochstrasser, D.F.; Tissot, J.-D. Advances in Electrophoresis, Vol. 6. VCH, Weinheim, 1993.

2. Bauw, G.; Rasmussen, H.H.; Van Den Bulcke, M.; Van Damme, J.; Puype, M.; Gesser, B.; Celis, J.E.; Vandekerckhove, J. Electrophoresis. 11, 1990, 528-536.

3. Forssmann, W.-G; Schulz-Knappe, P.; Meyer, M.; Adermann, K.; Forssmann, K.; Hock, D.; Aoki, A. Peptide Chemistry, ed. Aanaihara, N. Escom, Leiden, 1992, 553-557.

4. Schulz-Knappe, P.; Raida, M.; Meyer, M.; Quellhorst, E.A.; Forssmann, W.-G. Eur. J. Med. Res. 1, 1996, 223-236.

5. Schepky, A.G.; Bensch, K.W.; Schulz-Knappe, P.; Forssmann, W.-G. Biomedical Chromatography 8, 1994, 90-94.

6. Hillenkamp, F.; Karas, M.; Beavis, R.C.; Chait, B.T. Anal. Chem. 63, 1991, 1193A-1203A.

7. Lehmann, W.D. Massenspektrometrie in der Biochemie. Spektrum Akademischer Verlag, Heidelberg, 1996.

8. Schulz-Knappe, P.; Schrader, M.; Ständker, L.; Richter, R.; Hess, R.; Jürgens, M.; Forssmann,W.-G. J. Chromatogr. A, 776, 1997, 125-132.

11. Schrader, M.; Jürgens, M.; Hess, R.; Schulz-Knappe, P.; Raida, M.; Forssmann,W.-G. J. Chrom. A, 776, 1997,139-145.

12. Juhasz, P.; Costello, C.E.; Biemann, K. J. Am. Soc. Mass Spectrom. 4, 1993, 399-409.

14. Taylor, J.A.; Walsh, K.A.; Johnson, S. Rapid Commun. Mass Spectrom. 10, 1996, 679-687.

17. Bensch, K.W.; Raida, M.; Mägert, H.J.; Schulz-Knappe, P.; Forssmann,W.-G. FEBS Lett. 368, 1995, 331

18. Schulz-Knappe, P.; M·gert, H.J.; Dewald, B.; Meyer, M.; Cetin, Y.; Kubbies, M.; Tomeczkowski, J.; Kirchhoff, K.; Raida, M.; Adermann, K.; Kist, A.; Reinecke, M.; Sillard, R.; Pardigol, A.; Uguccioni, M.; Baggiolini, M.; Forssmann,W.-G. J. Exp. Med. 183, 1996, 295.