Comparison of the Proteomes of Three Yeast Wild Type Strains: CEN.PK2, FY1679 and W303

Yeast deletion strains created during gene function analysis projects very often show drastic phenotypic differences depending on the genetic background used. These results indicate the existence of important molecular differences between the CEN.PK2, FY1679 and W303 wild type strains. To characterise these differences we have compared the protein expression levels between CEN.PK2, FY1679 and W303 strains using twodimensional gel electrophoresis and identified selected proteins by mass spectrometric analysis. We have found that FY1679 and W303 strains are more similar to each other than to the CEN.PK2 strain. This study identifies 62 proteins that are differentially expressed between the strains and provides a valuable source of data for the interpretation of yeast mutant phenotypes observed in CEN.PK2, FY1679 and W303 strains.


Introduction
The availability of the complete genomic sequence of the yeast Saccharomyces cerevisiae opened the possibility to systematically study gene function and gave a unique insight into the molecular basis of function and growth of a single eukariotic cell. An international effort was initiated with the aim of creating a fundamental tool for such functional analysis, a collection of yeast genomic disruptants and plasmids. By August 2000, 22 472 yeast deletion strains covering 5867 different genes plus 1309 plasmids had been collected in the EURO-SCARF collection generated by the German functional analysis project, Eurofan I and Eurofan II projects (http://www.rz.uni-frankfurt.de/FB/fb16/mikro/ euroscarf/index.html). During the functional analysis projects, the deletions have been carried out in four different genetic backgrounds: FY1679 (isogenic to S288C whose DNA was sequenced), CEN.PK2 (generated in the German functional analysis project), W303 used in EUROFAN I and in the BYseries of strains (also isogenic to S288C strain).
Since yeast deletion strains can show drastic phenotypic differences depending on their genetic background, for example the null mutants of SSH1 gene exhibit slow growth at 30uC and 37uC in the FY1679 and W303 strains, but have normal growth in the CEN.PK2 strain (Duenas et al., 1999), it is important to be aware of strain-related differences (for further examples see Table 1).
The surprising fact is that although many laboratories all around the world use these yeast strains, very little is known about how they compare to each other.
Genealogy and characteristics of CEN.PK2, FY1679 and W303 yeast strains FY1679 strain was constructed by crossing FY23 and FY73 haploid strains (23r73=1679), (Thierry et al., 1990). Strains FY73 (MATa ura3-52, his3-D200, GAL2) and FY23 (MATa ura3-52, trp1-D63, leu2-D1 GAL2) are isogenic derivatives of strain S288C constructed by gene replacement (Winston et al., 1995). FY1679 was used as a source of DNA for a library that has been used for the European Union Yeast Genome Sequencing Programme. The origin of S288C strain has been described by Mortimer and Johnston (1986). They have determined that the principal progenitor strain of S288C was the strain EM93, which contributes approximately 88% of the gene pool in S288C. EM93 was originally isolated by E. Mark from rotting figs near Merced, California (Lindegarden, 1949). There were other strains that contributed to the genetic pool of S288C: EM126 (Saccharomyces carlsbergiensis); NRRL-210 isolated from rotting bananas from Costa Rica in 1942 (C.Kurtzman); FLD-commercial baking yeast and LK (Lindegarden, 1949) and yeast foam (Ephrussi, Hottingner, Tavlitzki, 1949) -both baking yeast (Figure 1). For details on the genealogy of S288C strain see (Mortimer and Johnston, 1986). In fact S288C was constructed by RK Mortimer by genetic crosses as a parental strain for biochemical mutants (Mortimer and Johnston, 1986). Requirements for the strain were that it must be non-clumpy -dispersing into single cells in liquid culture and have a minimal number of nutritional requirements. S288C requires only biotin, nitrogen source, glucose, salts and trace elements.
Several studies have characterised different genetic properties of S288C strain. Here are just some examples. The S288C strain is unable to grow pseudohyphally because it carries a nonsense mutation in FLO8 gene that is necessary for Table 1. Examples of genes exhibiting strain-dependent mutant phenotypes. Gene name and function are cited from the YPD protein database (Costanzo et al., 2000) Gene name  (Liu et al., 1996 andKron et al., 1997). The HAP1 gene coding for a haemdependent transcription factor in S288C and strains derived from it carries a Ty1 insertion which results in replacement of 13 amino acids (Gaisne et al., 1999). The strain S288C has only one copy of the NHA1 gene (putative Na+/H+ antiporter), whereas W303 appears to have more than one copy (Prior et al., 1996). Two genes encoding killer toxins, KHR1 and KHS1 are modified in S288C. The KHR1 gene is absent in strain S288C and KHS1 is interrupted by multiple frameshifts (Goto et al., 1991). Strain S288C has a null mutation in the KSS1 gene, which is involved in the filamentous and invasive growth pathway (Elion et al., 1991). Most laboratory strains, including S288C have substitutions in AQY1 gene and are interrupted in AQY2, which cause reduced water transport activity (Bonhivers et al., 1998). MEL1, MEL2, MEL5 and MEL6, which are involved in melibiose utilisation are not found in S288C (Lieberman et al., 1991). FY1679 has a poor sporulation frequency compared to CEN.PK2 and W303 strains (Entian and Kotter, 1998). The diploid strain W303 was constructed by transforming haploid strain W301-18A with a plasmid containing the HO gene and screening for diploids after loss of the plasmid (Thomas and Rothstein, 1989;Wallis et al., 1989). The diploid strain was dissected to obtain the isogenic MATa and MATa strains, W303-1A and W303-1B. Unfortunately, the genealogy of strain W301-18A has not been described before (Rothstein, 1983). Here we can conclude that W303 and S288C are very similar (as stated by Rodney Rothstein, personal communication). In fact, W303-18A was constructed by many crosses of W87 strains (Rothstein, 1977;Rothstein et al., 1977), which are mainly, but not exclusively, descendants of the strain X2180, itself derived from S288C by self -diploidisation (Mortimer and Johnston, 1986). Part of the genetic background of W301-18A was also obtained from the D311-3A strain constructed by Fred Sherman (Rothstein and Sherman 1980a,b). Additionally, one of the grandparents of W301-18A was the D190-9C strain that was obtained from Jack Szostak and about which is very little known (personal communication, Rodney Rothstein) (Figure 1).
The strain CEN.PK2 was developed specially for functional analysis by K-D Entian et al. (1999) but its origin and progenitor strain have not been published (Entian and Kotter, 1998). The sporulation efficiency of CEN.PK2 is as good as that of the W303 strain, and it has a faster growth rate, with doubling times of about 80 min for haploid strains (Entian and Kotter, 1998).

Background dependent yeast deletant phenotypes
The large number of examples of genetic background-dependent mutant phenotypes and differences between wild-type strains shows the importance of understanding the molecular characteristics of these wild-type strains. Although, various basic phenotypic analyses of specific yeast deletants revealed significant differences between the background strains (Table 1), only one study has been carried out to characterise some of these differences. Gü nter Daum and co workers (Daum et al., 1999) have shown that CEN.PK2, FY1679-28C and W303 strains exhibit different levels of sterols and triacylglycerols. Thus there is an urgent need to thoroughly characterise these three laboratory wild type strains.
Proteome analysis by two dimensional gel electrophoresis (2DGE) and mass spectrometry The recent developments of 2DGE and associated tools, such as mass spectrometry and software programs dedicated to image analysis, offer novel opportunities for studying the genetic background of different yeast strains (Joubert et al., 2000). Twodimensional electrophoresis separates proteins in terms of their isoelectric point and molecular mass. When applied to yeast whole cell extracts it can resolve several thousand proteins (Fey et al., 1997;Nawrocki et al., 1998). Hence 2DGE provides an opportunity to analyse a global picture of a given yeast strain under given environmental conditions. Several protein maps of Saccharomyces cerevisiae strains have been made (Boucherie et al., 1996;Shevchenko et al., 1996;Nawrocki et al., 1998;Garrels et al., 1997;Norbeck and Blomberg, 1997;Perrot et al., 1999) and several hundreds of proteins have been identified on these maps, proving that 2DGE is a very powerful tool for analysing the yeast proteome.
In this study, we have compared the protein map of three wild-type strains, CEN.PK2, FY1679-28C and W303, which are widely used for functional analysis of yeast genes. We demonstrate that there are 64 pronounced differences in the expression patterns between these strains, some of which have previously been found at the transcriptome level. The results show that FY1679 and W303 strains are more closely related to each other than to the CEN.PK2 strain.

Labelling of proteins and protein extraction
The cell culture was incubated on a rotary shaker shaking at 200 rpm at 28uC. Growth was monitored by following the optical density (OD) of the culture measured as light scattering at 600 nm wavelength (GENESYS 2, Spectronic Instruments). As the culture reached the OD=0.35 (approx. 5r10 6 cells per ml), 3 ml of culture was transferred to a 10ml flask and labelled for 30 min by adding 100 mCi of [ 35 S]-methionine. The cells were centrifuged, washed with 1.5 ml of distilled water and the yeast pellet was resuspended in 120 ml of lysis buffer (7 M urea, 2 M thiourea, 2% CHAPS, 0.4% DTT, 1% v/v pharmalytes 3-10). Cells were sonicated on a
melting-ice ethanol bath, at an amplitude of 6 microns, 2 times for 5s with 30s break, as described before (Nawrocki et al., 1998) and left shaking at room temperature overnight. Unlabelled samples (for preparative gels) were prepared in the same way with the exception that 50 ml of culture were used and [ 35 S]-methionine was not added.

Protein and CPM determination
The protein concentration in the samples was determined using the Bradford method (Bradford, 1976), which was adapted for use with lysis buffer as previously described (Fey et al., 1997). Determination of isotope incorporation into proteins was performed using trichloroacetic acid precipitation as described before (Fey et al., 1997).

2D gel electrophoresis
1 st dimension gel electrophoresis was performed on 18 cm long IPG4-7 and IPG6-9 gradient gels. The rehydration buffer for IPG4-7 strips was identical to the lysis buffer used for sample preparation and the sample was applied by in-gel rehydration. IPG6-9 gels were rehydrated in similar buffer except that 0.5% v/v pharmalytes 3-10 and 0.5% v/v pharmalytes 6-11 were added and the sample was applied by cup-loading. 2r10 6 CPM were loaded on each gel. Focusing was performed on Multiphor II at 20uC using a voltage/time profile linearly increasing from 0 V to 600 V for 2 : 15 h, from 600 V to 3500 V for 1 h and 3500 V for 13 : 30 h for IPG4-7 or for 3 h for IPG6-9. After focusing, strips were equilibrated twice for 15 min in equilibration buffer (6 M urea, 2% SDS, 30% Glycerol, 50 mM Tris-HCl pH 8.8, 1% DTT). For convenience, the gels were kept frozen at x80uC between the equilibration steps.
SDS PAGE second dimension gel electrophoresis was performed using a vertical Investigator 2 2-D Electrophoresis System (Millipore) and laboratorymade 12.5% (w/v) acrylamide gels (acrylamide: N, Nk-ethylene-bis-acrylamide ratio 200 : 1). The gels were run overnight at 20uC at a constant current setting. The running buffer was recirculated using an aquarium pump (flow rate nominally 4 l per min).

Pattern visualisation and computer analysis
After the second dimension separation, gels were dried directly on 3 mm Whatmann paper, exposed to Phosphoimager plates (AGFA) for 120 hours, and read in an AGFA ADC70 reader. The 2D gel patterns obtained have been compared to our existing database (Nawrocki et al., 1998) and other yeast 2D databases (Norbeck and Blomberg, 1997;Perrot et al., 1999;Hoogland et al., 2000) and verified in relation to known identified proteins.
Three images of each strain were matched and compared with images of other strains using Image Master (Amersham Pharmacia Biotech) computer program. Protein expression is measured as the sum of all the pixel grey values within the spot boundary (integrated optical density, IOD). This is then given as a percentage of the total of all the spots (%IOD). Boundaries, for spots that are present in one strain and missing in another have been added into the correct position in the latter images so that the background values can be used for statistical purposes. On each gel image we analysed the same number of spots (1222 on IPG4-7 gels and 389 on IPG6-9 gels).
The average %IOD and standard deviation was calculated from the expression data from the 3 corresponding images for each strain. These were then compared for each spot between each pair of strains using the Student's t-test, to reveal proteins whose expression was statistically different ( p>99%).
In order to estimate the reproducibility of the 2D gel system, the average percentage standard deviation for all the spots of each strain and each gel system used was calculated. This ranged from 35% to 44% for the IPG4-7 gels and from 28% to 35% for IPG6-9 gels. Thus the high reproducibility of the protein patterns allowed a reliable selection of spots differing by a 40% change in spot intensity (factor 1.4 or more and 0.71 or less). These two criteria were used together to select differences that were significant at the level of 99% and which differed by at least factor 1.4 or less than factor 0.71 between at least two of the analysed strains. Each selected difference was visually evaluated to make sure that the spot detection and matching were correct.

Preparative and zoom gels
For protein identification by mass spectrometry, preparative IPG4-7 and IPG6-9 gels were loaded with 200mg of cold proteins in addition to the [ 35 S]methionine labelled proteins and were run using the same gradients and procedures. Three preparative gels were made for each strain. In addition to the standard gradients, two one-pH unit zoom gradient gels, IPG4.5-5.5 and IPG5.5-6.7, were used for Yeast wild type strains 211 protein identification. Zoom gels were loaded with 300 mg of cold protein and 4r10 6 CPM of [ 35 S]methionine labelled proteins. The 1 st dimension was performed using the same procedures as for IPG4-7 and 6-9 gradients with the following running profile: voltage/time profile linearly increasing from 0 V to 600 V for 2 : 15 h, from 600 V to 3500 V for 1 h and 3500 V for 21 : 45 h. After the 2 nd dimension, preparative and zoom gels were dried and exposed to X-ray film for 10 days.

Mass spectrometric protein identification
Proteins of interest were manually excised from preparative and zoom gels, and after in-gel digestion they were analysed by MALDI mass spectrometry (Jensen et al., 1998). The mass spectra obtained were internally calibrated using trypsin autodigestion peptides and then used to search the NCBI database using the MASCOT (http:// www.matrixscience.com), MS-Fit (http://prospector. ucsf.edu) and ProFound (http://www.proteometrics. com) search programs. Database searches were performed using the following parameters with minor modifications needed for each program: all species, no restrictions for molecular weight and protein pI were used, trypsin digest, one missed cleavage allowed, cysteines modified by acrylamide allowed, oxidation of methionines possible, mass tolerance between 0.1-0.5 Da. An identification was considered positive when at least 5 peptides were matching with no sequence overlap.

Results and discussion
This study was performed in triplicate: each strain was grown from three independent cultures, each derived from a single cell colony. These were grown, labelled and after protein extraction independently analysed by 2DGE. The replicates were used as independent experiments during the statistical analysis of the spot intensity data. Consequently, the proteome of each strain was analysed and quantitated using three IPG4-7 and three IPG6-9 gels ( Figure 2A, B and C). The overlapping region between the gels was analysed on both gradients. During the computer assisted image analysis we have detected and matched 1222 spots on IPG4-7 and 389 spots on IPG6-9 gradient gels. Comparison and statistical analysis of the nine IPG4-7 and 9 IPG6-9 [ 35 S]-methionine labelled protein 2DGE patterns from analysed yeast strains revealed: 73 protein spots significantly changed between CEN.PK2-1B and FY1679-1D; 67 spots significantly changed between CEN.PK2-1B and W303-1B; and only 39 spots changed between FY1679-1D and W303-1B. In total, 122 protein spots were significantly changed between at least two of the three studied strains. The presented data and the genealogy of the strains both indicate that FY1679-1D and W303-1B are more similar to each other than to the CEN.PK2-1B strain.
The selected protein spots have been divided into four categories (A, B, C and ABC) depending on the relative intensity of selected protein expression ( Table 2). These correspond to proteins that are either up (+) or down (x) regulated in respectively CEN.PK2-1B (45 spots), FY1679-1D (25 spots) or W303-1B (17 spots) strains compared to the other two strains. The last category (ABC) contains 35 spots that show different expression levels in all three strains. The last group has been divided into subgroups where the order of the letters indicates the order of decreasing abundance of the protein. The large group of proteins that exhibit changes in all the strains indicates that the protein expression of some genes is not tightly regulated between the strains.

Mass spectrometric protein identification
All protein spots that exhibited significant intensity differences (122) between at least two of the analysed strains were subjected to MALDI mass spectrometric analysis irrespective of their abundance. In the first attempt spots were cut out from preparative IPG4-7 and IPG6-9 gels. Each spot was cut from the gel of the strain where the protein was most abundant.
For some of the spots, with relatively weak spot intensity, we could obtain only a few peptide mass peaks, which were not sufficient to unambiguously identify the protein. There were also other spots, which contained more than one protein. To identify some of these spots we have run narrow range IPG4.5-5.5 and IPG5.5-6.7 preparative gels and cut the remaining spots from them.
In total, in this work we have identified proteins in 101 of the spots (14 of which contained more than one protein), which represents an 82.7% rate of identification. The spots selected for cutting can be divided into three intensity groups. Spots of low intensity (<0.066 of %IOD) comprising 48 spots; spots of medium intensity (between 0.066 and 0.198 %IOD) comprising 50 spots and high

212
A. Rogowska-Wrzesinska et al.   Yeast wild type strains 215 intensity spots (>0.198 %IOD) comprising 24 spots. The identification rate in these groups was 73.3% (33), 90% (55) and 95.8% (23 spots) respectively. These results clearly show that most of the unidentified spots belong to the group of weak spots. To confirm this, the radioactively labelled gels were compared with silver stained gels of the same strains loaded with 150 mg of protein.
This comparison has shown that the spots from group of low intensity spots were in most cases very small or hardly visible on silver stained gels, confirming the high sensitivity of the mass spectrometric methods used.

Truncated proteins
Some of the spots were identified as a mixture of two or three protein components. Usually one of the two components was identified by a higher number of matching peptides and a better sequence coverage. Sometimes the other protein was a breakdown product of a larger protein with peptide masses covering only a fragment of the protein sequence. Some other protein spots contained only a fragment of a larger protein. Whether these protein fragments are generated by specific protein degradation involved in the control of protein turnover or some post translation processing, or are induced during sample preparation remains to be determined. It is, however, very unlikely that the observed protein fragments are generated randomly and introduced by sample preparation and handling, because we observed reproducible, statistically significant differences between the intensities in the strains we analysed.
There is unfortunately no space in this article to discuss all the protein groups and the possible implications for the phenotypes of the analysed strains and of deletion mutants made in these background strains. Therefore we have selected only some of the proteins for detailed analysis and discussion.

Major coat protein from Saccharomyces cerevisiae virus ScV-L-A
In this study 3 protein spots that lay in a row very close to each other were selected. All three spots were present in the CEN.PK2-1B strain and were not visible in gels from the FY1679-1D and W303-1B strains (Figure 3).
MALDI mass spectrometric analysis revealed that all three spots contain Gagp -the major coat protein from Saccharomyces cerevisiae virus L-A (ScV-L-A) indicating that the CEN.PK2-1B strain contains active ScV-L-A virus.
It has been reported previously that N-terminal acetylation of the Gagp is necessary for viral assembly and that the yeast Mak3p N-acetyltransferase is responsible for that modification (Tercero et al., 1993). In our study the presence of three spots representing the Gagp protein indicates a post-translational modification of this protein. To confirm this we re-examined the MALDI mass spectrometry results. Unfortunately the N-terminal peptide resulting from digestion of Gagp with trypsin consists of three amino acids and has a theoretical mass of only 418.2 Da. In this range, the background noise created by matrix ions in the MALDI mass spectrum is too high to distinguish protein peaks. Therefore we were not able to -1B strain total cell lysate. c) W303-1B strain total cell lysate. Proteins were separated using IPG4-7 and IPG6-9 gradient in the 1 st dimension (horizontal) and by SDS-PAGE electrophoresis in 12.5% polyacrylamide gels in the 2 nd dimension (vertical). Names indicate all proteins listed in Table 2. Star (*) following protein name indicates observed protein isoform

216
A. Rogowska-Wrzesinska et al. Table 2. Functional classification of proteins differently expressed between yeast wild type strains CEN.PKC-1B, FY1679-1D and W303-1B. The table lists all the proteins that have been identified by mass spectrometry, and neither their spot position, nor the mass spectrometry peptide coverage map indicated protein truncation. The proteins were identified in spots that exhibited intensity differences at significance level of 99% and differed by at least factor 1.4 or less than factor 0.71 between at least two of the analysed strains. The expression groups divide proteins into categories based on the spots percentage integrated optical density (%IOD). The 4 categories (A, B, C and ABC) depend on the relative intensity of selected protein expression. These correspond to proteins that are either up (+) or down (x) regulated in CEN.PK2-1B, FY1679-1D or W303-1B strains respectively, compared to the other two strains. The last category (ABC) contains spots that show different expression levels in all 3 strains. The last group has been divided into subgroups where the order of the letters indicates the order of decreasing abundance of the protein. Cellular localisation, protein name and biochemical function and cellular role are cited from the YPD protein database (Costanzo et al., 2001). Protein names marked with stars (*) indicate protein isoforms shown in Figure 2 Protein

Sterol metabolism
Mvd1p was expressed at a higher level in the CEN.PK2-1B strain. This enzyme decarboxylates mevalonate diphosphate and produces isopentenyl diphosphate (IPP) that is used for synthesis of sterols and for protein farnesylation or geranylgeranylation. Interestingly, this protein is less abundant in FY1679-1D and W303-1B compared to CEN.PK2-1B by about 63%. Differences in lipid metabolism between these strains have been reported before (Daum et al., 1999) where it has been shown that for example the FY1679 strain has lower levels of triacylglycerols and sterol compared to the other two strains.
The family of aminoacyl-tRNA synthetases play a key role in the readout of the genetic code catalysing the attachment of a given amino acid to the corresponding tRNA and indirectly take part in protein synthesis. The elevated expression of four aminoacyl-tRNA synthetases in CEN.PK2-1B strain could indicate a more intensive protein synthesis. This is, however, not confirmed as all the strains exhibited the same growth rate (about 90 min per generation) in logarithmic growth phase conditions used here.

Ribosomal proteins Rps0
Rps0Ap and Rps0Bp are two nearly identical ribosomal proteins (95% identity) (Demianova et al., 1996) that are required for assembly and stability of the 40S ribosomal subunit.
Demianova and co-workers have observed that disruption of the RPS0A gene in W303 results in a slow growth phenotype, and that disruption of the RPS0B gene results in an even slower growth rate and that the steady state levels of RPS0B mRNA are
higher than the RPS0A mRNA. Based on the ability of plasmid-borne copies of RPS0A and RPS0B genes to complement the growth defects associated with disruptions in either gene they have postulated that these two proteins are functionally equivalent but Rps0Bp makes a greater contribution to the pool of Rps0 molecules (Demianova et al., 1996).
Comparing the relative expression of Rps0Bp and Rps0Ap in each strain we have observed that the expression ratio between Rps0Bp and Rps0Ap in the CEN.PK2-1B and FY1679-1D is 3.08 and 3.05 whereas in the W303-1B strain it is 81.6. In this study we have observed that the expression level of the Rps0Ap in W303-1B is significantly lower than in the CEN.PK2-1B and FY1679-1D (the protein spot was less intense by a factor of 21 and 19 respectively) (compare Figure 4 and Table 2). On the contrary, the expression of the Rps0Bp was higher in W303-1B than in the other two strains by factor of 1.2 (CEN.PK2-1B) and 1.4 (FY1679-1D). Thus our results confirm the finding of Demianova and co-workers and additionally indicate that the difference in the differences in the relative amounts of Rps0Ap and Rps0Bp is most pronounced in the W303 strain.

Possible post-translational modifications detected by 2DGE
Five of the proteins selected and identified in this study (Leu1p, Hsp78p, Ssb1p, Dsk2p and Yhb1p) were found in two or three spots. All these proteins have been identified as two spots placed in a row very close to each other (Figure 2A, spots marked with star (*)). This kind of spot pattern indicates a post-translational modification that changes the pI of the protein, without major changes of its molecular mass (change of mass up to 1% will not be distinguishable on a 2DGE of this kind).
Leu1p, 3-Isopropylmalate dehydratase, is involved in the second step of the leucine biosynthesis pathway and no protein modification site has been identified or predicted. Both protein spots are significantly less expressed in the FY1679-1D indicating that the observed difference results from overall decreased expression of Leu1p and not only of one of its isoforms.
Hsp78p has been identified in two spots in a row, separated by a smaller spot containing a C-terminal truncated form of Met6p. The more neutral spot is significantly more expressed in CEN.PK2-1B and the more acidic protein was most abundant in W303-1B, moderately in CEN.PK2-1B and least in FY1679-1D ( Figure 2A and Table 2). Based on the protein spot position and the mass spectrometry peptide coverage, we can deduce that the N-terminal putative mitochondrial leader sequence has been cleaved off. Ssb1p is a heat shock protein of HSP70 family, involved with the translational machinery. The two spots show similar expression patterns ( Figure 2A, Table 2). It is most abundantly expressed in the W303-1B strain, less in the FY1679-1D strain and at the lowest level in the CEN.PK2-1B strain. Ssb1p has been shown to be N-acetylated by the Nat1p-Ard1p N-terminal acetyltransferase (Polevoda et al., 1999), which could account for the observed shift in the position.
Dsk2p is a protein that is required, with Rad23p, for duplication of the spindle pole body. This protein was found to be up-regulated in the more acidic spot and down regulated in the more neutral spot in CEN.PK2-1B (Table 2). This would indicate a post-translational modification seen in CEN.PK2-1B, that is less pronounced in the FY1679-1D and W303-1B strains. Sequence-based methods predict possible N-terminal acetylation (Huang et al., 1987) and a mass peak of 824.43 Da that corresponds to the N-terminal sequence (SLNIHIK) of Dsk2p (after trypsin digestion with the N-terminal methionine cleaved off) was detected in the more neutral spot. In the spectrum obtained from the more acidic spot we have found a mass peak of 866.42 Da that corresponds to the same N-terminal peptide of Dsk2p (after trypsin digestion). The increased mass of the peak by 42.01 Da confirms the prediction that the sequence is modified by N-terminal acetylation but the results demonstrate that both the modified and unmodified proteins exist. Yeast flavohemoglobin (Yhb1p) is related to globins and a reductase family (Zhu et al., 1992) and is involved in protecting the cell from nitrosative stress (Liu et al., 2000). (Figure 2 and Table 2). Both spots identified here, especially the less abundant and more acidic spot, showed the lowest expression level in the FY1679-1D strain. It has been reported that the N-terminus of the Yhb1p protein is unmodified (Zhu et al., 1992). Unfortunately the N-terminus was not detected in these studies and none of the other detected peptides gave any indication as to what type of post-translational modification could be related to the observed shift in the spot position.

Conclusions
This study represents the first comparison of protein expression levels between the haploid strains derived from yeast wild type CEN.PK2, FY1679 and W303 strains. It reports differences between the strains shown by 73 protein spots different between CEN.PK2-1B and FY1679-1D, 67 between CEN.PK2-1B and W303-1B and 39 spots different between FY1679-1D and W303-1B. These data show that the FY1679-1D and W303-1B strains are more similar to each other than to the CEN.PK2-1B strain, in agreement with their genealogy (FY1679 and W303 have a common ancestor strain S288C). Undoubtedly the observed differences in protein expression and post-translational modification influence the molecular and biochemical characteristics of the cells and possibly result in different phenotypes of yeast mutants observed in these strains. Therefore it is very important to identify and understand these differences prior to functional interpretation of phenotypic characteristics of yeast mutants obtained in functional analysis studies. This study identifies 62 proteins that are changed between the strains and provides a valuable source of data for the interpretation of differences in yeast mutant phenotypes observed in CEN.PK2, FY1679 and W303 derived strains.