Detection of phosphorylated proteins blotted onto

PAPER

www.rsc.org/jaas | Journal of Analytical Atomic Spectrometry

Detection of phosphorylated proteins blotted onto membranes using laser ablation inductively coupled plasma mass spectrometry Part 1: Optimisation of a calibration procedurew A. Venkatachalam,a C. U. Koehler,b I. Feldmann,a P. Lampen,a A. Manz,a P. H. Roosb and N. Jakubowski*a Received 28th March 2007, Accepted 27th June 2007 First published as an Advance Article on the web 17th July 2007 DOI: 10.1039/b704705n Inductively coupled plasma mass spectrometry (ICP-MS) has been used already for biological applications, especially for determination of the phosphorylation status of proteins. For this purpose we have coupled a laser ablation (LA) system to ICP-MS for detection of heteroelements on membranes after sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and electro-blotting. Our experiments show good reproducibility with a standard deviation of 6.1%. In single line scans nitrocellulose membrane material showed significantly higher signal intensities in comparison to polyvinylidene fluoride membranes. Quantification of phospho-proteins has been performed with (1) dotting and (2) SDS-PAGE separated and blotted standards to overcome protein losses during blotting. With this method, good linearity could be achieved in the range from 20 to 500 pmol phosphorus in proteins. We have estimated the limit of detection of phosphorus in b-casein, evaluating the whole protein spot to be 1.5 pmol using the 3s criterion measured in a medium resolution of the sector field instrument.

Introduction Phosphorylation is the most important post-translational modification of proteins, and therefore quantitative measurement of the phosphorylation state of proteins is a key issue in many life science applications. The significance is underlined by the fact that a new term and research field, namely phospho-proteomics, has been established.1–3 Dynamics and activity of the proteome are not merely defined by expression levels of proteins, because phosphorylation has a fundamental impact on their activity and life-span. Phosphorylation of specific amino acid residues in enzymes functions as a molecular switch either turning on or turning off the catalytic activity.4 Cellular signal transduction processes are largely a cascade of protein phosphorylations starting at the receptor level down to transcription factors and proteins controlling the function of transcription factors.5 At last, phosphorylation may result in accelerated or retarded degradation of proteins by the proteasome complex. In this case, the phosphorylation status determines the steady state level of a protein.6 This enumeration of protein phosphorylation effects clearly demonstrates the significance of the phosphorylation status of the proteome for assessing the biochemical activity status of cells or tissues. Thus, quantitative methods in phospho-proteomics are needed in order to identify characteristics of different physiological and pathological states. a

Institute for Analytical Sciences (ISAS), Bunsen-Kirchhoff-Str. 11, D-44139 Dortmund, Germany. E-mail: [email protected]; Fax: +49-231-1392-120; Tel: +49-231-1392-108 b Institute for Occupational Physiology at the University of Dortmund, Ardeystr. 67, D-44139 Dortmund, Germany. w Presented at the 2007 Winter Conference on Plasma Spectrochemistry, Taormina, Italy, February 18–23, 2007.

This journal is

c

The Royal Society of Chemistry 2007

Determination of the phosphorylation status of the proteome or of specific sub-proteomes requires (1) separation of proteins and (2) quantitation of the phosphorylation degree of discrete proteins. The most powerful tool for protein separation is two-dimensional gel electrophoresis which, however, does not work well for certain groups of closely related membrane proteins, such as cytochromes P450. Overall detection of phosphorylated forms of proteins is based on the reaction of blotted proteins with fluorescence dye labelled antibodies directed against phospho-serine, phospho-threonine and phospho-tyrosine residues. While anti-phosphotyrosine antibodies work well, antibodies directed against phospho-serine and phospho-threonine often show weak affinities for their antigens.5 Besides this, there are some more facts that aggravate the quantitative analysis of the phosphorylation status of proteins by means of available antibodies only. This concerns phosphorylation of amino acid residues other than serine, threonine and tyrosine such as histidine, lysine, hydroxylysine and hydroxyproline. The significance of histidine phosphorylation in signal transduction has been highlighted by Klumpp and Krieglstein.7,8 To our knowledge, antibodies specifically recognizing phospho-hydroxyproline are not available. Furthermore, steric hindrance could anticipate quantitative phospho-proteomics with regard to proteins with multiple phosphorylation sites. This is the case, for example, for the retinoblastoma protein and the tumor suppressor protein p53, which contain at least 5 and 12 phosphorylation sites, respectively.9,10 Some of these problems can be overcome by the use of molecular mass spectrometry in combination with electrophoretic separation of protein samples and techniques for enrichment of phospho-proteins.11 Molecular mass J. Anal. At. Spectrom., 2007, 22, 1023–1032 | 1023

spectrometry is most often applied for phospho-proteomics after an enzymatic digest of the protein spot in the gel, followed by analysis of the resulting polypeptides including scanning for phosphorylated forms.12 This can be laborious and time consuming. Additionally, the mass spectrometric response of a phospho-peptide may be suppressed by the presence of other peptides in a complex mixture. For sulfur, this was shown by Wind and co-workers, who analysed a mixture of a tryptic digest of two recombinant proteins after LC separation by electrospray ionisation (ESI) mass spectrometry and by ICP-MS.13 They could show that the S signal measured by ICP-MS was very well representing the expected molar S content of the observed peptides, and by that they could compare the ICP-MS sensitivities with those measured by ESI-MS using identical LC conditions. From this comparison they found that the ESI ionisation efficiency of tryptic peptides increases in proportion to the LC retention time and is decreased by the presence of an additional basic residue, which makes it difficult to use ESI-MS for quantification. This demonstrates that for quantification of phosphorylated polypeptides and proteins, ICP-MS looks very promising and it has already been applied for this purpose, as it is discussed in the following. An alternative approach was presented by Bandura and coworkers who have determined the degree of phosphorylation on the proteomic level by monitoring the signals of 31P16O+ and 32S16O+ by use of an ICP-MS with a dynamic reaction cell. With their LC approach they could detect sub-nanomolar amounts of phosphorylated polypeptides in a digest. In the investigation of total protein extracts from cultured malignant cells and from human malignant tissue, they found an increased global degree of phosphorylation compared to controls.14 Pro¨frock et al. have investigated various nebulizers for application of nano and capillary liquid chromatography and have measured the phosphorylation state of tryptic digests of b-casein using a reaction cell instrument in an energy discriminating mode and sub-pmol detection of the digests was possible.15 A sector field ICP-MS coupled to reversed phase high performance liquid chromatography was already used for the first time as an element specific detector for quantitative detection of phosphorylated polypeptides.16 By use of sulfur detection, being present in the amino acids methionine and cysteine of the peptides, an internal standard was available, which can be measured together with phosphorus in medium resolution (4000). After calibration by use of standards an easy and straightforward determination of the phosphorylation state was possible by measuring the P : S ratio.17 These examples show that LC-ICP-MS already plays an important role for detection of phosphorylated polypeptides in enzymatic digests, most often of protein standards only. For many applications, standards are not available and therefore techniques are needed which can measure, qualitatively and quantitatively, the phosphorylation status already in the gel or on a blot membrane directly. Sample introduction from a gel or a blot membrane into an ICP-MS is in principle possible by use of laser ablation (LA). This sample introduction technique has been used first for 1024 | J. Anal. At. Spectrom., 2007, 22, 1023–1032

metallo-proteins separated by immuno-electrophoresis by McLeod and co-workers.18 More recently, LA has also been applied for 1D and 2D electrophoretic gel separation to detect cadmium-binding proteins,19 metallo-proteins20 and selenoproteins,21 to mention a couple of applications. More details and a more complete overview is given in a review, which was published recently.22 The group of McLeod also applied whole gel elution of phosphorylated proteins for quantification by application of calibration strategy using liquid standards.23 An additional LC coupling by use of activated alumina was also applied for reduction of phosphate contaminations from the gel and from the buffers.23 LA-ICP-MS for qualitative detection of phosphorus in b-casein has first been presented by Marshall and co-workers in protein separation by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and blotted onto a membrane24 using a reaction cell instrument pressurized with hydrogen and helium. The limits of detection (LOD) for phosphorus were dominated by high blanks. Becker et al. applied LA to detect phosphorylated proteins qualitatively and quantitatively by drilling a hole into the gel just in the centre of the electrophoretic spot, which was made visible by silver staining.25 They applied this method for analysis of the human tau protein, which is a key protein in the formation of neurofibrillary tangles in Alzheimer disease. By this method, 17 phosphorylation sites of the tau protein had been quantified. More recently, Becker et al. had also applied LA for direct detection of metals, sulfur and phosphorus in human brain proteins using again the medium resolution of the sector field device.26 Semi-quantitative calibration was performed by generating aerosols from standards by use of a nebulizer and added to the LA aerosol. A limit of detection of 0.18 mg g1 for phosphorus has been determined in the gel blank.27 Phosphorylated proteins have not only been detected in gels but after blotting onto membranes too. This looks promising due to several reasons. Most importantly, a trace matrix separation can be achieved to reduce phosphate blanks from phosphate buffers used for sample preparation in SDS-PAGE. Additionally, the proteins are enriched in a thin surface layer which looks promising if LA is used for sample introduction. Nevertheless, losses during blotting are often mentioned as a limiting factor.28 In our previous work we have already applied LA-ICP-MS for detection of phosphorylated proteins separated by SDS-PAGE and blotted onto membranes.29 A linear relation between the measured intensity and the total amount of a-casein protein loaded onto the gel had been measured and could be used as a calibration for quantification of identical proteins. From the calibration curve a limit of detection of about 10 pmol had been estimated from the signal to noise ratio.29 More recently, Kru¨ger et al.30 had applied LA-ICP-MS for detection of phosphorylated proteins after separation by SDS-PAGE and blotting onto membranes. This technology has been used in the meantime to determine the phosphorylation state of the cytoplasmatic proteome of selected bacterial and eukaryotic cells.30 The finding in this case was that the eukaryotic cells exhibit a significantly higher phosphorylation degree compared to the bacterial proteome. Sulfur was used here as an internal standard. This journal is

c


In our previous investigation we have described a new laser ablation cell to analyze blot membranes with dimensions of more than 100 cm2 and the optimization strategy for the working conditions31 is already discussed. This cell is now applied for qualitative and quantitative detection of phosphorylated proteins separated by 1D PAGE after blotting onto membranes. For analysis of the membranes, a new calibration approach for quantitative detection of phosphorylated proteins blotted onto membranes will be presented, which is making use of a cross-calibration by proteins with well known phosphorus content.

storage, the filtrate and the retentate solutions were freeze dried at 20 1C for 24 h until it became a solid residue. The phosphorus content of the purified proteins was quantified by means of ICP-MS in diluted protein solutions, and in all cases the value measured was lower than the theoretical value. We found the following values: 7.4 (theoretical value 8) pmol phosphorus per pmol of a-casein, 3.8 (theoretical value 5) pmol phosphorus per pmol of b-casein and 1.4 (theoretical value 2) pmol phosphorus per pmol of pepsin. These proteins were used for calibration but instead of the theoretical, the measured phosphorus content was used. SDS-PAGE

Materials and methods Standard proteins and chemicals The following proteins have been used in this investigation: pepsin, bovine serum albumin (BSA), a-casein and b-casein (Sigma Aldrich Deisenhofen, Germany). Inorganic phosphorus was prepared from a stock of KH2PO4 (Merck, Germany) with the PO43 concentration of 1 g L1. The chemicals for casting of the gels are as follows: acrylamide and bis-acrylamide (Bio-Rad, Germany), ammonium peroxodisulfate, glycin (Roth, Germany), N,N,N 0 ,N 0 -tetramethylethylenediamine (TEMED) (Riedel de Hae¨n AG, Germany), tris(hydroxymethyl)aminomethane (Tris) (Applichem, Germany), and sodium dodecyl sulfate (SDS) (Biomol, Germany). The electrophoresis sample buffer includes tracking dye bromophenol blue; reducing agent dithiothreitol (Merck, Germany) and glycerol (Roth, Germany). All experiments were done with bi-distilled water. Electrode buffer was prepared with Tris, glycine and SDS and maintained at pH 8.3 for running the gel. For staining, the following preparations were made: staining solution (20 mL methanol, 60 mL water and 20 mL of Roti-Blue (5 conc.)), stabilizer solution (20 g of ammonium sulfate in 100 mL water) and drying solution (10% glycerol and 20% ethanol). For blotting the following buffers were used: anode buffer I (0.3 mol L1 Tris, 20% methanol made up to 500 mL with water), anode buffer II (25 mmol L1 Tris, 20% methanol, made up to 1000 mL with water) and cathode buffer (40 mmol L1 6-aminohexanoic acid, 20% methanol, made up to 500 mL with distilled water). Protein purification Most proteins from the stock, which were used as standards, contain inorganic phosphorus and therefore have to be purified if used for calibration. They were purified by ultrafiltration using the Centriprep ultrafiltration column (Millipore, Germany) with a cut-off of 3000 Da. The proteins in 0.125% ammonia solution were initially centrifuged at 3000 g for 20 min. The sample was washed with 0.125% ammonia solution various times. When the retentate volume in the Centriprep columns reached around 0.5 mL the centrifugation was stopped and the sample was decanted quantitatively into a separate flask. After purification the concentration of the proteins were measured with the Bradford assay. The concentration of the stock solution of b-casein, a-casein, pepsin and BSA were found to be 13.2, 11.8, 16.2 and 11.6 mg mL1. For This journal is

c


Proteins were separated charge-independently but according to their size in a one-dimensional, denaturing SDS-gel electrophoresis, which was performed on discontinuous and vertical 10% SDS-gels according to Laemmli.32 Stacking gel was prepared at 6% and the separating gel at 10%. Total protein concentrations used depend on the phosphorus content and 10 mL of sample diluted in 10 mL sample-buffer was applied in each well. Prior to electrophoresis, proteins were denatured in sample-buffer containing SDS and mercaptoethanol at 95 1C for 3 min. The gel was run at 200 V for 25 min using the miniprotean 3 dodeca cell from Bio-Rad. To focus the proteins in the stacking gel, the gel was started initially with 20 mA. For the resolving gel a current of 40 mA was chosen. Dual color marker (Bio-Rad, Germany) was used for calibrating the molecular weight of the proteins. Each time two sets of similar electrophoresis experiments were made. One is used for LA after blotting and the other was stained with colloidal coomassie staining (Roth, Germany) to check the protocol. Gels were stained with Roti-Blue (Carl Roth, Germany) staining solution and then washed with 25% methanol (Carl Roth, Germany). It was then stored in stabilizer solution (20 g ammonium sulfate in 100 mL water). These gels were scanned using the FLA-5100 imager (Fujifilm Life Sciences, Du¨sseldorf), which has a lateral resolution of 10 microns. They were preserved in cellophane sheets after equilibration in 20% ethanol–10% glycerol for 30 min. Blot procedure Electrophoretically separated proteins were immediately transferred to the blot membranes. A slightly modified protocol from Westermeier33 was used and is described as follows. Semi dry blotting was performed using the Amersham Pharmacia blotting unit. Proteins were transferred on ‘‘Protan’’ nitro cellulose membranes (NC) (Schleicher and Schuell, Dassel, Germany) or Immobilon-P (Millipore, Schwalbach, Germany) membranes. PVDF membranes, but not NC membranes, require activation with 100% methanol. Semi dry blotting was performed with a constant current density of 0.8 mA per cm2 blot area for 1 h for both PVDF and NC membranes. 18 filter papers were cut according to the size of the gel, 6 were soaked in anode I buffer, 3 were soaked in anode buffer II and 9 were soaked in cathode buffer just before blotting. PVDF and NC membranes were cut in the same dimension as the gel and soaked in anode buffer II for 5 min. 6 anode buffer I and 3 anode buffer II soaked filter papers were J. Anal. At. Spectrom., 2007, 22, 1023–1032 | 1025

placed on the anode graphite plate. Above that, the blot membrane was placed with the gel on top of it. Then, 9 filter papers soaked in cathode buffer were placed and stacked with the cathode graphite plate. For blotting as well as electrophoresis, the current was generated by the power supply EPS 3501 XL from Pharmacia (Amersham Biosciences, Germany). LA-ICP-MS For LA a flash lamp pumped Nd : YAG laser (Minilite II, Continuum, Santa Clara, USA) operated at the fourth harmonic wavelength was used (266 nm), achieving a sufficient laser energy of about 3 mJ and a pulse width of 3 to 5 ns. The laser was operated in a Q-switched mode and with its maximum repetition rate of 15 Hz. The homebuilt laser ablation chamber and its optimization were described already in a previous paper.31 The cell had been coupled to the ICP torch via polyethylene (PE) tubing with a length of 50 cm and an internal diameter of 4 mm. He was used as a carrier gas with a flow rate of 1.3 L min1. For daily tuning of the instrumental conditions, the Aridus (Cetac) was used as a sample introduction system producing a nearly dry aerosol, thus plasma conditions should be not equal but similar to the dry aerosol produced by the laser. The sample outlet of the nebulizer was mixed with the He flow behind the chamber by use of a T-connector. The argon flow was set to 1 L min1. To establish a good spatial resolution during line-scanning, a small crater diameter is advantageous, while for highest signal intensity a larger crater volume would be better to increase the amount of ablated material. To solve this problem, we used beam shaping to produce a laser beam of oval shape, with higher width than length. For beam shaping, two planar-convex cylindrical lenses were used, mounted one upon the other in the crossed direction. The length of the resulting crater was about 100 mm in scan direction and the width was adjustable in a limited range; finally, a width of 500 mm has been used throughout this investigation. The whole chamber was moved relative to the fixed laser beam with a forward speed of 1 mm s1. The blot membranes were subsequently rastered line by line with a distance of 1 mm inbetween. For ICP-MS, a sector field instrument type Element 2 (Thermo Fisher Scientific, Bremen, Germany) was used. Due to the spectral interferences being present at m/z 31 phosphorus was detected at a medium resolution of 4000 if not mentioned otherwise. For the phosphorus intensity measurement at medium resolution, 10 channels per peak were chosen with an integration window of 60% and mass window of 100%. Hence, 6 channels per peak from the centre of the maximum peak height were selected for a single data point. For the low resolution experiments, an integration window of 80% was chosen with 8 channels per peak taken into account for a single data point. The sample time taken for each channel is 10 ms. Hence, for 10 channels the total segment duration is 100 ms. The settling time of the magnet is 1 ms. The number of data points was chosen according to the length of the blot. For example, for a blot of 58 mm length, with a scan speed of 1026 | J. Anal. At. Spectrom., 2007, 22, 1023–1032

1 mm s1, with the above set of conditions for the medium resolution measurement, the total of calculated data points was 665. For imaging, all separated intensity data sets for each laser trace were copied together by means of a self developed Matlab routine (The Matworks, USA), which additionally enables a fast preview imaging of the elemental distribution of the measured blot membrane. Final surface plots as shown here were created using Origin 7 (OriginLabCorporation, Northhampton, USA). For quantitative evaluation of the results, the intensities for all data points in the time interval of all laser traces belonging to one protein spot were averaged—again by means of a Matlab routine—resulting in the intensity–time relation of one electrophoretic lane. In a second step, all peaks of interest were integrated over time (or related distance) using a commercially available chromatography software (EuroChrom 2000, Knauer, Berlin).

Results and discussion Scanning conditions In principle, the ablation cell, which was developed already in our previous paper31 and which is used for the imaging of 31P+ distribution after blotting onto large size membranes, should have the following features. It should provide a reasonable local resolution so that adjacent electrophoretic spots can be separated for which a sufficient lateral resolution is required. Additionally, it should provide an as high as possible but constant sensitivity, independent of the location of the electrophoretic spot in the cell. All these conditions had been successfully optimized in the previous work.31 Another important feature is that the total analysis time, even of a large area, should be as short as possible. From this point of view, back and forth scanning of the laser beam looks advantageous. Thus, at the beginning we checked this, using 13C+ as a marker element for LA conditions and signal dispersion. This experiment also answers the question if this element can be used as an internal standard. Two measurements were performed, first by scanning in direction from the gas inlet to the outlet (forward), the next in the reverse direction (Fig. 1). The 13C+ intensity measured in the forward direction shows a constant signal for a length of the scan line of about 6 cm. It was found that the 13C+ signal intensity did not change during ablation of the NC membrane. But when the membrane is scanned in a reverse direction, we could see a certain drift in the 13C+ intensities. This phenomenon can possibly be attributed to particles trapped in the dead volume in a turbulent zone. Although time can be saved for the whole measurement, scanning in reverse direction was not performed to avoid this drift effect. Concerning this measurement, it is straightforward to use 13 + C as an internal standard, but it was not applied here so as to spend the most time, of our relatively slow scanning instrument, to measure the analytical element. This will change in the future with an upgrade of the electronics of the Element 2, which allows faster peak jumping. This journal is

c


Fig. 1 13C+ trace on an NC membrane in the forward and the reverse direction.

31 +

P

intensity measured with low and medium mass resolution

Detection of 31P+ is hampered in plasma-based spectrochemistry, since it features a high first ionization potential (10.484 eV), and therefore low ionization efficiency in an argon-based plasma and various polyatomic interferences, in particular 15 16 + N O , are limiting as well in the case of mass spectrometry. However, the commercial availability of high resolution instrumentation and collision/reaction cell ICP-MS has essentially solved this interference problem.14,34,35 In principle a medium resolution of 4000, which is provided with the Element 2 used here, is sufficient to overcome the most prominent interferences by 15N16O+. However, due to the nitrocellulose matrix, also matrix related spectral interferences such as from 15N16O+, 13C18O+, 14N16O1H+ and 14 17 + N O can be expected and therefore the mass window of phosphorus at m/z 31 was detected by using a magnetic scan and the medium resolution mode. We have measured here the P intensities together with the predominant 15N16O+, 13 18 + C O and 14N16O1H+ intensities. The result is shown in Fig. 2 for a NC membrane on which b-casein with a total amount of 152 pmol of phosphorus was blotted after 1D SDS-PAGE separation. This spectrum was measured during ablation of the blank area. Three peaks are seen, the first one belongs to phosphorus at 30.973 Da, the second one can be attributed to 15N16O+ at mass 30.995 Da and the third is 14 16 1 + N O H at mass 31.0053 Da. From Fig. 2 it can be seen that the interfering signals of 15N16O+ and 14N16O1H+ are far below 1% of the intensity of phosphorus. This is surprising, because by ablation of the NC membrane, all critical elements like C, N and O are introduced into the plasma. Nevertheless, it seems that the amount of oxygen from the membrane is not high enough to form oxides. To compare the detection limits in both resolutions, in the next experiment two spots of b-casein with a total amount of 152 pmol of phosphorus were blotted after 1D SDS-PAGE separation and ablated. One was measured in the low resolution mode (R = 400) and the other in the medium resolution mode (R = 4000). Per spot, 8 line scans were summed up and the results are shown in Table 1. The peak height of the instrument for the 31P+ measurement at low resolution was 8 This journal is

c


Fig. 2 Mass window of 31P+, 15N16O+, 13C18O+ and 14N16O1H+ during measurement of the blank of a NC membrane using LA-ICPSFMS (medium resolution mode).

times higher than that of the medium resolution. Nevertheless, at medium resolution a limit of detection (LOD) of 1.5 pmol of phosphorus could be achieved, which is lower than that of 4 pmol measured in the low resolution mode. This is the reason why in this study, the medium resolution mode has been used further on. Nevertheless, this measurement also shows that phosphorylation of proteins can be studied with quadrupole instruments, even without a reaction cell if LA is used for sample introduction. Reproducibility It was not known at the beginning of our studies how reproducible PAGE separations and blotting onto membranes were, and how signal generation by a pulsed laser ablation might even contribute to the worsening of the standard deviation of signals. Therefore, the reproducibility of the method was measured with 7 parallel 1D separations. For this purpose, 20 pmol b-casein corresponding to a total phosphorus content of 76 pmol was separated in 7 parallel runs by 1D SDS-PAGE and blotted onto a NC membrane using conditions already described. The 31P+ intensity of the protein spot area was measured during a line scan with a length of 57 mm. The whole membrane area (57 53 mm) was covered by 53 line scans with a distance of 1 mm inbetween. In addition to the return back time of the chamber and dead time of around 1 min, a total instrumental analysis time of 65 min was required. From this measurement (not shown here) we have compared the peak area of the resulting 31P+ intensity distribution with single line scans, measured through the middle of the protein spot. The mean value of all 7 protein spot areas Table 1 Intensity and estimated limits of detection in medium resolution and low resolution; application of 152 pmol of phosphorus in b-casein Low resolution Peak height/cps S Background level/cps Limit of detection

6

38 10 0.33 106 2.6 106 4 pmol

Medium resolution 4.81 106 0.0153 106 2.21 105 1.5 pmol

J. Anal. At. Spectrom., 2007, 22, 1023–1032 | 1027

were found to be 2.5 106 counts per second (cps) compared to 4.5 105 cps of the single line scan. The relative standard deviation of the mean value of all 7 spots amounts to 6.1%, which is surprisingly low and demonstrates that blotting as well as laser ablation is quite reproducible. When only the mean values of single line scans are evaluated the relative standard deviation was increasing to 10%, and can increase even more for those protein spots in which proteins are more inhomogeneously distributed (see for instance Fig. 6). This is the reason why integration is always performed with respect to the whole protein spot in the following experiments. Comparison of PVDF and NC membrane characteristics for LA For blotting, different membrane materials are commercially available, but most often either PVDF or NC are applied because they provide a stable binding of proteins on the surface, so that losses of proteins penetrating the membrane material are reduced as much as possible. The binding capacity of both materials is similar, but dependent on the manufacturer. Typical values are 70–150 mg cm2 for NC and 125–300 mg cm2 for PVDF. Additionally, from the analytical point of view, the material should have a couple of features with respect to laser ablation, laser energy and detection by ICP-MS. For instance, it should show good ablation properties and should have low blanks for the elements of interest. Therefore, both materials have been investigated after separation and blotting of 40 pmol of b-casein (not shown here). After integration of the spot area from the phosphorus intensity distribution plot, we found that for the same laser energy the 31P+ intensity of the protein spot area is more than a factor of 13 higher for the NC in comparison to the PDVF material. This strong difference can only partially be explained by the different ablation behavior of the membrane material. After the first laser scan the depth of the crater in NC and PVDF membranes are found to be 65 mm and 31 mm, respectively (Fig. 3a and 3b). For PVDF, a second scan on the same laser trace reaches a depth of 54 mm, while for NC the membrane is drilled through. From the crater profile for a laser trace of 10 mm length, a volume of 0.19 mm3 for NC and only 0.074 mm3 for PVDF was calculated, thus 2.5 fold more volume of the material is ablated in the case of a NC membrane. This effect has not been investigated in detail, but shows that both materials behave quite differently as the laser energy is more suitable for NC membranes. One reason for this different behavior might be that NC decomposes easily after heating to temperatures above 185 1C, whereas PVDF decomposes at around 425 1C. In a similar experiment, the depth distributions of the protein spot were investigated by ablating every scan line a few times until the membrane material in the crater was completely ablated. This was done for both membrane materials, but because of the different ablation behavior, the depth resolution is strongly different for both materials because the same laser energy was used for both experiments. The resulting profile is shown only for the PVDF material in Fig. 4, because no protein related signal was detected for the NC material during the second scan. Considering the intensities of all the scans to be 100%, we found out that we are able to get only 1028 | J. Anal. At. Spectrom., 2007, 22, 1023–1032

Fig. 3 (a) Crater profile of a laser trace on a NC membrane measured by means of a white light interferometer. (b) Crater profile of a laser trace on PVDF membrane measured by means of a white light interferometer.

56% of the intensities in the first scan (depth of 31 mm), whereas with the second scan—reaching a depth of 54 mm—30% are detected. With the 3rd scan this value drops to 9% only, but the depth was not measured in this case. In total, five scans are needed to penetrate this material. From the results of this investigation we can conclude that on both membrane materials the proteins are enriched within a depth of the upper 70 mm. But as to give a fair assessment, it should be mentioned that our beam profile is not well suited to draw conclusions for which a better depth resolution is required. The discussion therefore is only suited to explain some differences but not in a quantitative manner. The value of 56% in the first scan is less than the value reported by Mu¨ller et al.,36 who had also used PVDF membranes and found recovery rates of gold-conjugated antibodies in the first scan line of 78.6%. However, the depth distribution

Fig. 4 Multiple line scans of a protein spot of 40 pmol b-casein on PVDF membrane.

This journal is

c


of proteins also depends on the applied protein and the amount, but has yet not been investigated in detail. From a pragmatic point of view, the selection of the material is quite straightforward. Although both membranes enrich the protein in a thin surface layer, NC membranes are more suitable, so that for not too high amounts of protein loaded onto the membrane, single scanning of the protein spot per sample is even sufficient and guarantees highest sensitivity, too. Additionally, it was shown in Fig. 2 that no additional spectral interferences are generated. This is the reason why this material was exclusively used further on. Calibration and quantification strategy For calibration, different concepts are described in literature. For instance, Becker et al. have used an ultrasonically generated aerosol simultaneously introduced with the ablated aerosol for quantification.25 For application of this approach it is a prerequisite that the laser aerosol size distribution is very well represented by the ultrasonically generated aerosol, as well as it does not change during ablation. Additionally, transport, vaporization and ionisation must be identical for both types of aerosols, which is hard to achieve. From this point of view, all those concepts are preferred in this work where the signal of the standard and of the target protein is coming from a LA process of the same matrix. For gel applications, a couple of papers have already shown the suitability of a calibration by matrix matched standards. For instance, Becker et al. have recently also used an alternative calibration approach for phosphorus determinations in proteins.27 For this purpose the P intensity of the standard protein, in this case ovalbumin, was dotted with total amounts ranging from 0.1 ng up to 500 ng on a gel after separation of the proteins. From the slope of the linear calibration graph and the gel blank they calculated a LOD for phosphorus of 0.18 mg g1. Marshall et al. have used the phosphorus containing protein b-casein for calibration and a LOD of 16 pmol has been estimated.24 The direct ablation of the gel was limited by a too high blank value of phosphorus in the gel. Bandura et al.14 have used the protein phosvitin in 1D separation for a calibration calculating LODs at 50–100 pmol level. Alternatively, Che´ry et al. have used hydrated gels with multielement standard solutions for a calibration and calculation of limits of detection under multielement conditions.21 This technique, using metal standards hydrated in a gel, cannot be used here because metals are lost during the blot procedure due to the fact that they are moving to the cathode, whereas SDS treated proteins move to the anode. Therefore, in our case two different calibration procedures were investigated here for comparison. The first procedure is based on the use of standards directly dotted by a pipette onto the surface of the membrane and in the second procedure a protein with well known phosphorus content is PAGE separated, blotted and then used to calibrate the samples which are present on the same blot membrane. For the dotting experiment, 5 mL of three solutions in 5 series, ranging from 19 to 152 pmol phosphorus amount in bcasein, 17.5 to 140 pmol phosphorus amount in pepsin and 25 This journal is

c


to 200 pmol phosphorus in an inorganic phosphorus solution, were dotted directly onto a NC membrane and were ablated after drying. Images of the phosphorus intensity distribution measured are shown in Fig. 5. It can be seen from the dots that each protein is inhomogeneously distributed in the protein spots. Nevertheless, after integration of the spot area the calibration graphs are linear and sensitivity for b-casein (y = 54 400x + 762 000), pepsin (y = 63 900x + 466 000) as well as inorganic phosphorus (y = 57 000x + 607 300) fits well with not more than 20% deviation. A correlation coefficient (R2) of 0.9915, 0.9823 and 0.9898 could be obtained for P043, b-casein and pepsin, respectively. As a test analysis, an SDS-PAGE separated a-casein (not shown here) with a total phosphorus amount of 93 pmol was used as an unknown sample and quantified after blotting and LA using the calibration curves. Using the calibration of PO43, pepsin and b-casein, we found 47, 44 and 47 pmol phosphorus amounts of a-casein, respectively. All results show that by dotting, the amount of phosphorus in the protein is underestimated and only a fraction is found (around 55%). As an alternative approach, the standards were separated and blotted together with the test sample. For this purpose, b-casein with increasing total amounts of phosphorus ranging from 38 to 304 pmol has been blotted already together with a-casein with a total phosphorus amount of 139 pmol. The phosphorus intensity distribution is shown in Fig. 6 for different protein spots on a NC membrane after 1D SDSPAGE separations. Using the calibration (R2 = 0.985), a total amount of 126 pmol of phosphorus is determined for the a-casein sample. Although the sensitivity (y = 15 000x 163 300) in this calibration is lower compared to the dotting experiment, the quantified a-casein amount is in better agreement with the amount chosen for this experiment. Concerning the linear dynamic range, we don’t observe here a deviation from linearity for phosphorus content above 100 pmol, as this was the case in our previous paper.31 One reason for this higher dynamic range might be that here semidry blotting is

Fig. 5 31P+ images of dotting experiments with different phosphorus amounts of (1) PO43, [a-25, b-50, c-100, d-150, e-200 pmol] (2) pepsin [a-17.5, b-35, c-70, d-105 and e-140 pmol] and (3) b-casein [a-19, b-38, c-76, d-114, e-152 pmol].


and 10 000 pmol were separated and blotted onto 2 NC membranes, one behind the other. The 31P+ intensity distributions are shown in Fig. 7a and b. The integrated intensities for the phosphorus of the protein spots are shown in a graph in Fig. 7c. From this Figure it clearly can be seen that the graph deviates from linearity already for total protein amounts of more than 100 pmol. In addition to that, quite significant intensities could also be observed in the second membrane.

Fig. 6 31P+ images of a- and b-casein spots on a NC membrane (1) 38, (2) 76, (3) 152, (4) 228, (5) 304 pmol phosphorus in b-casein and (6) 139 pmol phosphorus in a-casein.

used instead of tank blotting. We calculated the number of phosphate residues in a-casein using the two calibration strategies. The b-casein PAGE separated calibration strategy shows 7.3 phosphate residues in a-casein, which is in good agreement with the theoretical value of 8, whereas by external dot calibration this value would be only 4.4. From the better accuracy which is achieved using SDSPAGE and blotting, compared to the experiments where dotting has been applied for calibration, it can be concluded that not only the standards are ablated and transported in the same way as the test sample, but also that possible blotting losses are compensated if proteins are chosen as standards. This is the reason why this quantification strategy is preferred in our work. It is not clear yet, if this calibration is valid only for those proteins which have a similar molecular weight, because the proteins selected here cover only a small range of the molecular weight scale: b-casein (MW 23.6 kDa), pepsin (35 kDa) and a-casein (24.5 kDa). This has to be investigated in future experiments in more detail. Usually, one wants to determine the phosphorylation state of a given protein. This information can be achieved directly from the amount of phosphorus measured once the stoichiometry of the protein of interest is known or alternatively, one has to determine the total amount of protein being present in an electrophoretic spot as it is shown in the work of Wind et al. by rationing of the phosphorus signal to an internal standard of the protein, such as sulfur.17 This technique can not be applied here because the compound SDS used for charging and defolding of the protein contains already a large amount of sulfur, so that high blank values are becoming a limiting factor. Linear dynamic range It was already discussed in the previous experiment that we have observed a much higher linear dynamic range in comparison to our previous paper.31 Therefore, the next experiment was performed to see if the membrane becomes overloaded. For this purpose, b-casein with amounts of 1, 10, 100, 1000 1030 | J. Anal. At. Spectrom., 2007, 22, 1023–1032

Fig. 7 (a,b) 31P+ intensity distributions of increasing amounts of b-casein [1–1, 2–10, 3–100, 4–1000 and 5–10 000 pmol] in the (a) first membrane and (b) second membrane. (c) Integrated intensities of the spot in the first and the second membrane.

This journal is

c


The main spot area of spot numbers 4 and 5 are about 0.3 and 0.8 cm2 and they contain 1000 and 10 000 pmol of the protein, respectively (Fig. 7a). Thus, we can estimate a protein loading of about 80 mg cm2 and 300 mg cm2, where the latter is above the upper limit of the NC membranes capacity. For too high amounts of protein, they penetrate deeper—at least through the membrane into the second membrane and possibly also to the anode filter causing a severe loss. This means that one has to determine always the linear dynamic range for a specific membrane material, a given amount of protein and for the chosen laser conditions if quantitative results are needed.

Conclusion LA-ICP-MS was applied here for imaging of the intensity distribution of 31P+ in electrophoretic spots after 1D PAGE separation and blotting onto commercially available blot membranes. We have shown that the detection of phosphorus containing proteins is barely disturbed by molecular interferences if LA is used. Thus, detection in low mass resolution is possible, but nevertheless the LOD is lower if medium mass resolution is used. Reproducibility of the whole method—including electrophoretic separation, blotting and ablation—shows a standard deviation of 6% if the whole spot area is taken into account. With the laser setup used throughout this investigation, we found NC membrane to be advantageous compared to PVDF membrane. Two calibration procedures for quantification of unknown amounts of phosphorus containing proteins are compared. It is shown that to a certain extent a calibration procedure using a protein standard during the same electrophoretic run compensates blot losses and incomplete ablations. At the moment, this method can be applied only for one dimensional SDS-PAGE. As already mentioned S is an important marker element present in most proteins, but can not be used as an internal standard for calibration and determination of the phosphorylation state. In this investigation a very high S blank signal is caused from the SDS used for separation in PAGE. This problem can only be overcome if a native PAGE is applied. This is discussed in more detail in two recently published papers. In the work of Garijo et al.37 S detection was possible using native conditions in an online PAGE system coupled directly to ICP-MS for the quantification of Fe in proteins. In the work of Polatajko et al.38 1D SDS-PAGE is compared with 1D anodal native PAGE separations of Cd containing proteins, and it can be seen from the results that the resolving power of both methods is quite similar. Both examples demonstrate that native PAGE is becoming a powerful alternative, in particular for separation of metallo-proteins. In Part II we will apply the calibration procedure developed in this part for experiments in which cell cultures are stressed by chemicals and hormones and we will present first data demonstrating that LA-ICP-MS of blot membranes can be applied to measure changes in the phospho-proteome. This journal is

c


Acknowledgements We are indebted to Mr Messerschmidt and Mrs Bichbaümer for sample preparation. The work of ISAS and of IfADo was financially supported by the ‘‘Bundesministerium fu¨r Bildung und Forschung’’ and by the ‘‘Ministerium fu¨r Innovation, Wissenschaft, Forschung und Technologie des Landes Nordrhein-Westfalen’’.

References 1 M. R. Larsen, G. L. Sorensen, S. J. Fey, P. M. Larsen and P. Roepstorff, Proteomics, 2001, 1, 223–238. 2 Y. Oda, T. Nagasu and B. T. Chait, Nat. Biotechnol., 2001, 19, 379–382. 3 E. Salih, Mass Spectrom. Rev., 2005, 24, 828–846. 4 E. H. Fischer and E. G. Krebs, J. Biol. Chem., 1955, 216, 121–132. 5 M. de Graauw, P. Hensbergen and B. van de Water, Electrophoresis, 2006, 27, 2676–2686. 6 G. G. Mason, K. B. Hendil and A. J. Rivett, Eur. J. Biochem., 1996, 238, 453–462. 7 H. R. Matthews, Pharmacol. Ther., 1995, 67, 323–350. 8 S. Klumpp and J. Krieglstein, Biochim. Biophys. Acta-Proteins and Proteomics, 2005, 1754, 291–295. 9 J. A. Lees, K. J. Buchkovich, D. R. Marshak, C. W. Anderson and E. Harlow, EMBO J., 1991, 10, 4279–4290. 10 X. J. Yang, Oncogene, 2005, 24, 1653–1662. 11 D. E. Kalume, H. Molina and A. Pandey, Curr. Opin. Chem. Biol., 2003, 7, 64–69. 12 A. P. Navaza, J. R. Encinar and A. Sanz-Medel, J. Anal. At. Spectrom., DOI: 10.1039/b703555a. 13 M. Wind, A. Wegener, A. Eisenmenger, R. Kellner and W. D. Lehmann, Angew. Chem., Int. Ed. Engl., 2003, 42, 3425–3427. 14 D. R. Bandura, O. Ornatsky and L. Liao, J. Anal. At. Spectrom., 2004, 19, 96–100. 15 D. Pro¨frock, P. Leonhard, W. Ruck and A. Prange, Anal. Bioanal. Chem., 2005, 381, 194–204. 16 M. Wind, M. Edler, N. Jakubowski, M. Linscheid, H. Wesch and W. D. Lehmann, Anal. Chem., 2001, 73, 29–35. 17 M. Wind, H. Wesch and W. D. Lehmann, Anal. Chem., 2001, 73, 3006–3010. 18 J. L. Neilsen, A. Abildtrup, J. Christensen, P. Watson, A. Cox and C. W. McLeod, Spectrochim. Acta, Part B, 1998, 53, 339–345. 19 M. R. B. Binet, R. Ma, C. W. McLeod and R. K. Poole, Anal. Biochem., 2003, 318, 30–38. 20 R. D. Evans and J. Villeneuve, J. Anal. At. Spectrom., 2000, 15, 157–161. 21 C. C. Chery, D. Guenther, R. Cornelis, F. Vanhaecke and L. Moens, Electrophoresis, 2003, 24, 3305–3313. 22 R. Ma, C. W. McLeod, K. Tomlinson and R. K. Poole, Electrophoresis, 2004, 25, 2469–2477. 23 V. L. Elliott, C. McLeod and P. S. Marshall, Anal. Bioanal. Chem., 2005, 383, 416–423. 24 P. Marshall, O. Heudi, S. Bains, H. N. Freemann, F. Abou-Shakra and K. Reardon, Analyst, 2002, 127, 459–461. 25 J. S. Becker, S. F. Boulyga, J. S. Becker, jun, C. Pickhardt, E. Damoc and M. Przybylski, Int. J. Mass Spectrom., 2003, 228, 985–997. 26 J. S. Becker, M. Zoriy, J. S. Becker, C. Pickhardt and M. Przybylski, J. Anal. At. Spectrom., 2004, 19, 149–152. 27 J. S. Becker, M. Zoriy, C. Pickhardt, M. Przybylski and J. S. Becker, J. Anal. At. Spectrom., 2004, 19, 1236–1243. 28 V. J. Nesatyy and N. W. Ross, Analyst, 2002, 127, 1180–1187. 29 M. Wind, I. Feldmann, N. Jakubowski and W. D. Lehmann, Electrophoresis, 2003, 24, 1276–1280. 30 R. Kru¨ger, D. Ku¨bler, R. Pallisse´, A. Burkovski and W. D. Lehmann, Anal. Chem., 2006, 78, 1987–1994. 31 I. Feldmann, C. U. Koehler, P. H. Roos and N. Jakubowski, J. Anal. At. Spectrom., 2006, 21, 1006–1015. 32 U. K. Laemmli, Nature, 1970, 227, 680–685.


33 R. Westermeier, Electrophoresis in practice, Wiley-VCh Verlag, Weinheim, 2001, p. 65. 34 D. Pro¨frock, P. Leonhard and A. Prange, J. Anal. At. Spectrom., 2003, 18, 708–713. 35 J. S. Becker, S. F. Boulyga, C. Pickhardt, J. S. Becker, J. Buddrus and M. Przybylski, Anal. Bioanal. Chem., 2003, 375, 561–566.

1032 | J. Anal. At. Spectrom., 2007, 22, 1023–1032

36 S. D. Mu¨ller, R. A. Diaz-Bone, J. Felix and W. Goedecke, J. Anal. At. Spectrom., 2005, 20, 907–911. 37 M. Garijo Anorbe, J. Messerschmidt, I. Feldmann and N. Jakubowski, J. Anal. At. Spectrom., DOI: 10.1039/b704541g. 38 A. Polatajko, M. Azzolini, I. Feldmann, T. Stuetzel and N. Jakubowski, J. Anal. At. Spectrom., DOI: 10.1039/b703245e.

This journal is

c
