Category Archives: Analysis

Analysis: Markov clustering and the case of the unsupported protein complexes

In 2006, Krogan and coworkers published a paper in Nature describing a global analysis of protein complexes in budding yeast. This resulted in a network of 7,123 protein-protein interactions involving 2,708 proteins, which was organized into 547 protein complexes using the Markov clustering algorithm.

Considering my previous two posts, it probably comes as a surprise to nobody that I wanted to check if the issue of unnatural clusters also affected this study. Albert Palleja, a postdoc in my group, thus extracted the 547 sub-networks corresponding the protein complexes and applied single-linkage clustering to check if all clusters corresponded to connected sub-networks.

It turned out that 9 of the 547 protein complexes do not correspond to connected sub-networks in the protein interaction network that formed the basis for the clustering. Two complexes each contain two additional subunits that have no interactions with any of the other subunits of the proposed complex, five complexes contain one additional subunit with no interactions to other subunits, and two complexes are proposed hetero-dimers made up of subunits that do not interact according to the interaction network. These complexes are visualized in the figure below with the erroneous subunits highlighted in red:

To check if these additional subunits are in any way supported by the experimental data presented in the paper, I downloaded the set of raw purification from the Krogan Lab Interactome Database. For 4 of the 9 complexes, the additional subunits are weakly supported by at least one purification. It should be noted, however, that this evidence was not judged to be sufficiently reliable by the authors themselves to include the interaction in the core network based on which the complexes were derived.

To make a long story short, this analysis shows that 9 of the 547 protein complexes published by Krogan and coworkers contain one or more subunits that are not supported by the interaction network from which the complexes were derived. Of these, 5 complexes contain subunits that have no support in the underlying experimental data, and which are purely artifacts of using the MCL algorithm without without enforcing that clusters must correspond to connected sub-networks.

Analysis: Markov clustering and the case of the nonhomologous orthologs

In the previous blog post I described how the MCL algorithm can sometimes produce unnatural clusters with disconnected parts. The C implementation of MCL has an option to suppress this behavior (--force-connected=y), but I suspect that it is rarely used. I have thus taken a closer look at some notable applications of MCL in bioinformatics to see if unnatural clusters arise in real data sets.

Here I will focus on OrthoMCL-DB, which is a database of orthologous groups of protein sequences. These were constructed by applying the MCL algorithm to the normalized results of an all-against-all BLAST search of the protein sequences.

To check the connectivity of the resulting orthologous groups, I downloaded OrthoMCL version 4 including the 13+ GB of gzipped BLAST results that formed the basis for the MCL clustering. I wish to thanks to the OrthoMCL-DB team for being very helpful and making this large data set available to me.

A few Perl scripts and CPU hours later, Albert Palleja and I had extracted the BLAST network for each of the 116,536 orthologous groups and performed single-linkage clustering to check if any of them contained disconnected parts. We found that this was the case for the following 28 orthologous groups:

Orthologous group Protein
OG4_10123 tcru|Tc00.1047053448329.10
OG4_10133 cmer|CMS291C
OG4_11608 bmor|BGIBMGA011561
OG4_13082 lbic|eu2.Lbscf0004g03370
OG4_17434 cint|ENSCINP00000028818
OG4_20715 mbre|fgenesh2_pg.scaffold_4000474
OG4_20953 tpal|NP_218832
OG4_21182 tvag|TVAG_333570
OG4_24433 tmar|NP_229533
OG4_29163 tcru|Tc00.1047053508221.76
OG4_32884 gzea|FGST_11535
OG4_36484 cbri|WBGene00088730
OG4_39391 ddis|DDB_G0279421
OG4_43780 cpar|cgd3_1080
OG4_44179 atha|NP_177880
OG4_44684 bmal|YP_104794
OG4_45409 rcom|29647.m002000
OG4_50671 pram|C_scaffold_62000023
OG4_50712 bpse|YP_331887.1
OG4_52326 bmaa|14961.m05365
OG4_52455 bmal|YP_338428
OG4_55725 apis|XP_001952076
OG4_57272 bbov|XP_001610684.1
OG4_58797 hwal|YP_659316
OG4_61264 crei|122343
OG4_68577 bmor|BGIBMGA000864
OG4_71107 cbur|NP_819756
OG4_84041 tcru|Tc00.1047053479883.10

For convenience, the orthologous groups are linked to the corresponding web pages in OrthoMCL-DB, which enable viewing of Pfam domain architectures and multiple sequence alignments. Cursory inspection suggests that the majority of the of the sequences listed in the table do not belong to the orthologous groups in question.

Of the 28 orthologous groups, 24 groups contain a single protein with no BLAST hits to other group members, 2 groups each contain 2 such singletons, and the remaining 2 groups each contain 2 proteins that show weak similarity to each other but not to any other group members. The latter proteins are highlighted in red.

In summary, this analysis shows that the unnatural clustering by MCL reported for a toy example in the previous post also affects the results of real-world bioinformatics applications of the algorithm.

Analysis: Markov clustering and the case of the unnatural clusters

The MCL (Markov CLustering) algorithm was invented/discovered by Stijn van Dongen and was published in 2000. It has since become highly popular in bioinformatics and has proven to perform well on a variety of different problems.

It was also the method of choice when my postdoc Albert Palleja needed to cluster the human interaction network from the STRING database. However, we got strange results. More specifically, we observed that some clusters contained proteins that had no interactions with any other proteins within the same cluster. I call these unnatural clusters; this should be seen as a contrast to natural clusters, which are characterized by the presence of many edges between the members of a cluster.

After we had spent a week unsuccessfully trying to find out what we were doing wrong, I finally asked myself if it could be that we were not doing anything wrong. Might it be that applying the MCL algorithm to a protein interaction network can result in clusters of non-interacting proteins?

To test this, I constructed the following toy network consisting of only 10 nodes and 12 edges:

Assigning a weight of 1 to all edges and running this network through MCL using an inflation factor (the key parameter in the MCL algorithm) between 1.734 and 3.418 yields five clusters. In the figure below, the nodes are colored according to which cluster they belong to:

Note the black cluster which consists of two proteins, X and Y, despite the two nodes only being connected via nodes that are not part of the same cluster. This example clearly shows that the MCL algorithm is indeed capable of producing unnatural clusters containing nodes with no direct edges to any other members in the cluster.

In my view this is not as such a error in the the MCL algorithm. The algorithm is based on simulation of flow in the graph. The nodes X and Y are clustered due to the strong flow between them via nodes A, C, E, and G. However, I think it is fair to say that this behavior will catch many users by surprise and that it can give rise to misleading results when applying MCL to certain types of networks.

Edit: I suspect that this is the same issue that was reported on the Mcl-users mailing list by Sungwon Jung. Using the --force-connected=y option prevents the undesirable clustering of X and Y.

Analysis: Correlating the PLoS article level metrics

A few months ago, the Public Library of Science (PLoS) made available a spreadsheet with article level metrics. Although others have already analyzed these data (see posts by Mike Chelen), I decided to take a closer look at the PLoS article level metrics.

The data set consists of 20 different article level metrics. However, some of these are very sparse and some are partially redundant. I thus decided to filter/merge these to create a reduced set of only 6 metrics:

  1. Blog posts. This value is the sum of Blog Coverage – Postgenomic, Blog Coverage – Nature Blogs, and Blog Coverage – Bloglines. A single blog post may obviously be picked up by multiple of these resources and hence be counted more than once. Being unable to count unique blog posts referring to a publication, I decided to aim for maximal coverage by using the sum rather than using data for only a single resource.
  2. Bookmarks. This value is the sum of Social Bookmarking – CiteULike and Social Bookmarking – Connotea. One cannot rule out that a single user bookmarks the same publication in both CiteULike and Connotea, but I would assume that most people use one or the other for bookmarking.
  3. Citations. This value is the sum of Citations – CrossRef, Citations – PubMed Central, and Citations – Scopus. I decided to use the sum to be consistent with the other metrics, but a single citation may obviously be picked up by more than one of these resources.
  4. Downloads. This value is called Combined Usage (HTML + PDF + XML) in the original data set and is the sum of Total HTML Page Views, Total PDF Downloads, and Total XML Downloads. Again the sum is used to be consistent.
  5. Ratings. This value is called Number of Ratings in the original data set. Because of the small number of articles with rating, notes, and comments, I decided to discard the related values Average Rating, Number of Note threads, Number of replies to Notes, Number of Comment threads, Number of replies to Comments, and Number of ‘Star Ratings’ that also include a text comment.
  6. Trackbacks. This value is called Number of Trackbacks in the original data set. I was greatly in doubt whether to merge this into the blog post metric, but in the end decided against doing so because trackbacks do not necessarily originate from blog posts.

Calculating all pairwise correlations among these metrics is obviously trivial. However, one has to be careful when interpreting the correlations as there are at least two major confounding factors. First, it is important to keep in mind that the PLoS article level metrics have been collected across several journals. Some of these journals are high impact journals such as PLoS Biology and PLoS Medicine, whereas others are lower impact journals such as PLoS ONE. One would expect that papers published in the former two journals will on average have higher values for most metrics than the latter journal. Papers published in journals with a web-savvy readership, e.g. PLoS Computational Biology, are more likely to receive blog posts and social bookmarks. Second, the age of a paper matters. Both downloads and in particular citations accumulate over time. To correct for these confounding factors, I constructed a normalized set of article level metrics, in which each metric for a given article was divided by the average for articles published the same year in the same journal.

I next calculated all pairwise Pearson correlation coefficients among the reduced set of article level metrics. To see the effect of the normalization, I did this for both the raw and the normalized metrics. I visualized the correlation coefficients as a heat map, showing the results for the raw metrics above the diagonal and the results for the normalized metrics below the diagonal.

There are a several interesting observations to be made from this figure:

  • Downloads correlate strongly with all the other metrics. This is hardly surprising, but it is reassuring to see that these correlations are not trivially explained by age and journal effects.
  • Bookmarks is the metric that apart from number of downloads correlates most strongly with Citations. This makes good sense since CiteULike and Connotea are commonly used as reference managers. If you add a paper to you bibliography database, you will likely cite it at some point.
  • Blog posts and Trackbacks correlate well with Downloads but poorly with citations. This may reflect that blog posts about research papers are often targeted towards a broad audience; if most of the readers of the blog posts are laymen or researchers from other fields, they will be unlikely to cite the papers covered in the blog posts.
  • Ratings correlates fairly poorly with every other metric. Combined with the low number of ratings, this makes me wonder if the option to rate papers on the journal web sites is all that useful.

Finally, I will point out one additional metrics that I would very much like to see added in future versions of this data set, namely microblogging. I personally discover many papers through others mentioning them on Twitter or FriendFeed. Because of the much smaller the effort involved in microblogging a paper as opposed to writing a full blog post about it, I suspect that the number of tweets that link to a paper would be a very informative metric.

Edit: I made a mistake in the normalization program, which I have now corrected. I have updated the figure and the conclusions to reflect the changes. It should be noted that some comments to this post were made prior to this correction.

Analysis: Limited agreement among lists of Cdc28p substrates

A collaboration between the Morgan lab at UCSF and the Gygi lab at Harvard has resulted in a paper by Holt et al. in Science, which reports the identification of several hundred substrates of the central cell-cycle kinase Cdc28p (also known as Cdk1) in the budding yeast Saccharomyces cerevisiae:

Global analysis of Cdk1 substrate phosphorylation sites provides insights into evolution.

To explore the mechanisms and evolution of cell-cycle control, we analyzed the position and conservation of large numbers of phosphorylation sites for the cyclin-dependent kinase Cdk1 in the budding yeast Saccharomyces cerevisiae. We combined specific chemical inhibition of Cdk1 with quantitative mass spectrometry to identify the positions of 547 phosphorylation sites on 308 Cdk1 substrates in vivo. Comparisons of these substrates with orthologs throughout the ascomycete lineage revealed that the position of most phosphorylation sites is not conserved in evolution; instead, clusters of sites shift position in rapidly evolving disordered regions. We propose that the regulation of protein function by phosphorylation often depends on simple nonspecific mechanisms that disrupt or enhance protein-protein interactions. The gain or loss of phosphorylation sites in rapidly evolving regions could facilitate the evolution of kinase-signaling circuits.

The paper makes several interested in analyses and observations. However, I found the comparison to the previous study of Cdc28p substrates by Ubersax et al. from the Morgan lab to be less detailed than I had hoped for:

Phosphorylation of Cdk1 consensus sites was observed on 67% (122 of 181) of proteins previously identified as Cdk1 substrates in vitro (4). Sixty-six percent (80 of 122) of these proteins contained sites at which phosphorylation decreased (log2 H/L < –1) after inhibition of Cdk1 (only 45 of 122 are expected if there is no correlation between the experiments in vitro and in vivo; χ2 test, P < 10-10).

In other words, 44% (80 of 181) of Cdc28p substrates identified in the old study were confirmed by the new study, and only 26% (80 of 308) of the Cdc28p substrates identified in the new study are supported by the old study. There are many possible explanations for this discrepancy

Depth of the mass spectrometry

It is notoriously difficult to identify peptides from low-abundance proteins in mass spectrometry. In the new mass spectrometry study, the authors were able to map 8710 precise phosphorylation sites on 1957 proteins. However, budding yeast is estimated to express in the order of 4500 distinct proteins during exponential growth (Gavin et al., 2006). Assuming that the majority of these proteins contain sites that are phosphorylated during at least part of the mitotic cell cycle, it is likely that a considerable number of low-abundance Cdc28p substrates identified in the old study have been missed in the new study.

Biases in phosphopeptide enrichment

When doing phosphoproteomics, it is necessary to first enrich for phosphopeptides to improve the coverage. To this end, Holt et al. used immobilized metal affinity chromatography (IMAC). In 2007, the Aebersold group at ETH published a paper showing that different purification methods lead to isolation of different, partially overlapping segments of the phosphoproteome. Specifically, they showed that IMAC enrichment biases the data towards isolation of multiply phosphorylated peptides. Given that only a single purification method was used, it is likely that in vivo Cdc28p substrates may have been missed in the new study, in particular if the peptides contain only a single phosphorylation site.

In vitro vs. in vivo conditions

The old study by Ubersax et al. was done performed on cell lysate, which is an in vitro strategy (although all other proteins expressed during the cell cycle are present). It is thus likely that some of the proteins that are phosphorylated by Cdc28p under these conditions are nonetheless not in vivo Cdc28p substrates.

Can we do better?

As always, it is easy to point out potential flaws in other people’s data sets; however, it is much more constructive to do something about the problems. The challenge is thus to construct a larger and more reliable set of Cdc28p substrates by combining the data from the two studies.

To check the feasibility of assigning confidence scores to different putative Cdc28p substrates, I tested if the fold change observed in the new study correlates with the chance that the substrate was also identified in the old study. To this end, I divided the 308 Cdc28p substrates from the new studies into two groups and constructed histograms of the fold changes for each group:

Phosphorylation ratios from Holt et al.

The fold changes are clearly skewed towards larger negative values for the Cdc28p substrates also identified by the old study relative to the proteins that were not previously identified as Cdc28p substrates. This difference is statistically significant at P < 1% according to the Kolmogorov-Smirnov test. This suggests that the observed fold changes in the new mass spectrometry study correlates with the likelihood that the proteins are true Cdc28p substrates.

The old study gave rise to so-called P-score for the individual proteins (not to be confused with P-values). I decided to test if these too can be used as quality scores, I constructed an equivalent histogram in which the Cdc28p substrates found in the old study were divided into two groups based on whether or not they were also found in the new study:

P-scores from Ubersax et al.

In this case, no obvious trend is seen and a Kolmogorov-Smirnov test indeed reveals no statistically significant difference between the two distributions. Surprisingly, the P-scores do thus not appear to be useful quality scores for the putative Cdc28p substrates.

Given the two sets of putative Cdc28 substrates, only one of which can be ranked by reliability, how can we create a better combined set? If one aims for the high accuracy at the price of low coverage, one could obviously choose to trust only the substrates identified by both screens. However, given the caveats regarding depth of mass spectrometry and biases arising from the enrichment procedure, I would be hesitant to use this approach. Alternatively, one could aim for maximal coverage at the price of accuracy by trusting all sites identified by either study. However, seeing the large fraction of novel substrates identified by Holt et al. with a log2-ratio only slightly below -1, I would personally tend to apply a more stringent threshold to the data from the new study by Holt et al., for example requiring log2-ratio below -2, before merging the sets of substrates from the two studies.

WebCiteCite this post

Analysis: Results from thermal stability shift and competition binding assays correlate well

Several large kinase inhibitor screens have been published in recent years. Two of the largest come from Stefan Knapp’s lab and Ambit, respectively. The former group used a temperature shift assay to measure the change in thermal stability of 60 human serine/threonine kinases that is caused by the binding of each of 156 kinase inhibitors (Fedorov et al., 2007). The latter group used a competition a competition binding assay to measure the dissociation constants (Kd) for 38 kinase inhibitors and 290 distinct kinases (Karaman et al., 2008).

The two screens are not directly comparable because one measures temperature shifts whereas the other measures dissociation constants. To see if it possible to convert temperature shift values to Kd values, I asked Damian Szklarczyk (who is a Ph.D. student in my group) to map all data from both screens onto a common set of chemical and protein identifiers, extract all inhibitor-kinase pairs that were measured in both assays, and make a scatter plot of -log(Kd) as function of temperature shift. The result was a set of 704 pairs of temperature shift and Kd values. In the plot below, inhibitor-kinase pairs for which binding was not observed in the competition binding assay were defined to have a Kd of 10 microM, and negative values from the temperature shift assay were treated as zero temperature shift.

Correlation between temperature shift and -log(Kd)

The plot shows that the two assays are in very good agreement, which is surprising considering that the assays are fundamentally very different and were run using different expression constructs for several of the kinases. The linear Pearson correlation coefficient is 0.92 when excluding the one obvious outlier shown in red (BIRB796 vs. MAPK11; this appears to be a false negative in the competition binding assay).

The linear fit gives an intercept with the y-axis of 4.9223, which implies that a temperature shift of zero (i.e. no binding according to the temperature shift assay) does not translate precisely into a Kd of 10 microM (i.e. no binding according to the competition binding assay). We thus did a second linear regression in which we forced the intercept with the y-axis to 5 (red regression line in the plot). We thereby at the calibration function -log(Kd) = 5+0.244*Ts, which allows us to to convert temperature shifts to Kd values. We have thereby managed to put the measurements from the two kinase inhibitor screens onto a common basis that facilitates direct comparison and integration.

Full disclosure: I have an on-going collaboration with Stefan Knapp’s lab related to screening of kinase inhibitor.

WebCiteCite this post

Analysis: On the evolution of protein length and phosphorylation sites

It has been much too long since I have last written a blog post. Part of the reason has been that I have been busy moving back to Denmark, starting up a research group, and co-founding a company. More on that in other blog posts. The main reason, however, has been a lack of papers that inspired me to do the simple follow-up analyses that I usually blog about.

This has thankfully changed now. Pedro Beltrao and coworkers recently published an interesting paper in PLoS Biology on the evolution of regulation through protein phosphorylation. The paper presents several interesting analyses and comparisoins of phosphoproteomics data from three yeast species; the abstract summarizes the findings better than I can do:

Evolution of Phosphoregulation: Comparison of Phosphorylation Patterns across Yeast Species
The extent by which different cellular components generate phenotypic diversity is an ongoing debate in evolutionary biology that is yet to be addressed by quantitative comparative studies. We conducted an in vivo mass-spectrometry study of the phosphoproteomes of three yeast species (Saccharomyces cerevisiae, Candida albicans, and Schizosaccharomyces pombe) in order to quantify the evolutionary rate of change of phosphorylation. We estimate that kinase–substrate interactions change, at most, two orders of magnitude more slowly than transcription factor (TF)–promoter interactions. Our computational analysis linking kinases to putative substrates recapitulates known phosphoregulation events and provides putative evolutionary histories for the kinase regulation of protein complexes across 11 yeast species. To validate these trends, we used the E-MAP approach to analyze over 2,000 quantitative genetic interactions in S. cerevisiae and Sc. pombe, which demonstrated that protein kinases, and to a greater extent TFs, show lower than average conservation of genetic interactions. We propose therefore that protein kinases are an important source of phenotypic diversity.

Figure 1a in the paper shows the intriguing observation that, despite rapid evolution of individual phosphorylation sites, the relative number of phosphorylation sites within proteins from different functional classes (Gene Ontology categories) remains remarkably constant between species:

Beltrao et al., PLoS Biology, 2009, Figure 1a

However, it occurred to me that this could potentially be a consequence of longer proteins having more phosphorylation sites, and protein length being conserved through evolution. I thus counted the number of unique phosphorylation sites identified in each protein (thanks to Pedro Beltrao for providing the data) and correlated it with the length of the proteins. In the two plots below, I have pooled the proteins so that each dot corresponds to 100 proteins. The upper and lower panels show the results for S. cerevisiae and S. pombe, respectively:

Number of phosphorylation sites vs. protein lengh for S. cerevisiae

Number of phosphorylation sites vs. protein length for S. pombe

As should be evident from the plots, the average number of phosphorylation sites in a protein correlates strongly with its length, which is by no means surprisings. It is unclear to me why the intercept with the y-axis appears to differ from zero in both plots; suggestions are welcome.

The next question was whether the Gene Ontology terms that correspond to proteins with many phosphorylation sites are indeed assigned to proteins that are longer than average. I thus examined the terms “Cell budding”, “Morphogenesis”, and “Signal transduction”.

The average S. cerevisiae protein is 450 aa long. Proteins annotated with “Cell budding”, “Morphogenesis”, and “Signal transduction” are on average 1.6 (739 aa), 2.1 (945 aa), and 1.5 (679 aa) times longer, respectively. By comparison, the corresponding ratios observed for phosphorylation sites are approximately 2.3, 2.6, and 2.4. It would thus appear that differences in protein length between functional classes of proteins account for much, but not all, of the signal that was observed by Beltrao et al. when comparing the number phosphorylation sites.

Edit: Make sure to read Pedro Beltrao’s follow-up blog post, which nicely confirms that whereas protein length does play a role, it is not the full story.

WebCiteCite this post

Analysis: Four complementary yeast interactomes

The latest issue of Science features a paper by Yu et al. in which they report the results of a comprehensive yeast two-hybrid (Y2H) screen for interactions between budding yeast proteins. Just a few months earlier, Science published a paper by Tarassov et al. that describes a similar screen performed using a novel protein fragment complementation assay (PCA). Peer Bork and I wrote a Perspectives piece on these two papers, showing that the different assays for detecting protein interactions are complementary in the sense that they capture interactions for different subsets of the proteome. For example, PCA detects many interactions for membrane proteins whereas Y2H detects many interactions for nuclear proteins.

As part of writing the Perspectives piece, I performed numerous analyses that were not included in the final publication, because they were either too technical for a broad audience, not interesting enough to spend valuable space on, or would involve additional figures. Thankfully, my blog imposes no limitations on the number of words or figures (nor is it required that the content is interesting, although that is desirable).

The comparison included, in addition to the two interactomes introduced above, a third interactome that consists of all the high-confidence interactions identified by Gavin et al. and Krogan et al. using the tandem affinity purification (TAP) method. Also included in the comparison (but not in the Perspectives piece) was the literature-curated (LC) set of interactions published by Reguly et al. in 2006.

The Venn diagram below shows the overlap of the four interactomes in terms of proteins, that is a protein is considered to belong to an interactome if the method in question suggested at least one interaction partner:

The numbers outside the ellipses specify the total number of proteins for which a given method identified interactions. Notably, the PCA, Y2H, and TAP interactomes cover only approximately one sixth, one third, and half of the yeast proteome, respectively, despite all three assays having been tested on all yeast ORFs. This suggests that only a fraction of proteins can be targeted with a given assay.

A second way to compare the four interactomes is to count their overlaps in terms of pairs of interacting proteins. To provide additional detail, I distinguished between interactions that are not found in a given interactome because one or both proteins are not covered by the interactome in question (dashed lines in the diagrams), and interactions that were not found despite both proteins being covered (full lines in the diagrams). The Venn diagrams below show all twelve pairwise comparisions of the four interactomes:

As expected, the largest overlap is observed when comparing the two largest interactomes (LC and TAP), whereas the smallest overlap is observed when comparing the smallest interactomes (PCA and Y2H). Even if taking into account the differences in terms of protein coverage, however, the the overlaps between the interactomes leave a lot to be desired.

There are several reasons for the poor overlap at the level of pairwise interactions. One is that false positive interactions are unlikely to be reproducible by a different assay. A second is that the assays measure fundamentally different types of interactions: PCA and Y2H measure direct binary interactions between proteins, whereas TAP measures co-complex interactions, that is whether two proteins are part of the same complex or not. This is illustrated in the figure below, which shows the binary and co-complex networks for three different scenarios:

The two types of assays have different strengths and weaknesses. Binary interaction assays can in principle distinguish between the two first complexes, which only differ in that the subunits B and C are in direct contact in first complex but not in the second. However, binary assays are not able to distinguish between the second and the third scenario, that is whether A, B, and C form a single complex (ABC) or two complexes (AB and AC). Conversely, data from co-complex assays are able to answer the latter question but are unable to distinguish between the two first scenarios. The different assays thus complement each other, not only because they are able to interrogate different subsets of the proteome, but also because they provide us with complementary information about the composition and topology of protein complexes.

WebCiteCite this post

Analysis: Cell-cycle-regulated proteins are more abundant in haploid relative to diploid cells

Two days ago, Matthias Mann’s group published a paper in Nature in which they compare the level of individual proteins in haploid relative to diploid budding yeast cells:

Comprehensive mass-spectrometry-based proteome quantification of haploid versus diploid yeast

Mass spectrometry is a powerful technology for the analysis of large numbers of endogenous proteins. However, the analytical challenges associated with comprehensive identification and relative quantification of cellular proteomes have so far appeared to be insurmountable. Here, using advances in computational proteomics, instrument performance and sample preparation strategies, we compare protein levels of essentially all endogenous proteins in haploid yeast cells to their diploid counterparts. Our analysis spans more than four orders of magnitude in protein abundance with no discrimination against membrane or low level regulatory proteins. Stable-isotope labelling by amino acids in cell culture (SILAC) quantification was very accurate across the proteome, as demonstrated by one-to-one ratios of most yeast proteins. Key members of the pheromone pathway were specific to haploid yeast but others were unaltered, suggesting an efficient control mechanism of the mating response. Several retrotransposon-associated proteins were specific to haploid yeast. Gene ontology analysis pinpointed a significant change for cell wall components in agreement with geometrical considerations: diploid cells have twice the volume but not twice the surface area of haploid cells. Transcriptome levels agreed poorly with proteome changes overall. However, after filtering out low confidence microarray measurements, messenger RNA changes and SILAC ratios correlated very well for pheromone pathway components. Systems-wide, precise quantification directly at the protein level opens up new perspectives in post-genomics and systems biology.

Although the paper focuses on the larger amount of cell-wall proteins and proteins involved in pheromone response in haploid cells, the supplementary tables reveal similar biases for many other functional classes, including nucleosomes and cyclin-dependent kinase inhibitors. As many of these proteins are regulated during the cell cycle, I suspected that cell-cycle-regulated proteins might be more abundant in haploid cells relative to diploid cells.

To test this hypothesis, I divided the proteins quantified by the Mann group into two classes: dynamic proteins, which are encoded by genes that are periodically expressed during the cell cycle, and static proteins, which are encoded by genes that are expressed at a constant level (de Lichtenberg et al., 2005). For each class, I plotted the log2-ratios of the protein levels in haploid and diploid cells:

The plot reeals a quite strong shift of dynamic proteins toward higher log-ratios; this difference is highly significant according to the Mann-Whitney U test (P < 10-12). Proteins encoded by cell-cycle-regulated genes are thus in general more abundant in haploid budding yeast cells than in diploid cells.

Full disclosure: I currently collaborate with Matthias Mann and members of his group, and we will soon be colleagues a the Novo Nordisk Foundation Center for Protein Research.

WebCiteCite this post

Analysis: Transcriptional and posttranslational regulation of cell-cycle kinases

Daub and coworkers from Matthias Mann’s group recently published a paper in Molecular Cell, describing a phosphoproteomics study of kinases during S and M phase of the mitotic cell cycle:

Kinase-selective enrichment enables quantitative phosphoproteomics of the kinome across the cell cycle.

Protein kinases are pivotal regulators of cell signaling that modulate each other’s functions and activities through site-specific phosphorylation events. These key regulatory modifications have not been studied comprehensively, because low cellular abundance of kinases has resulted in their underrepresentation in previous phosphoproteome studies. Here, we combine kinase-selective affinity purification with quantitative mass spectrometry to analyze the cell-cycle regulation of protein kinases. This proteomics approach enabled us to quantify 219 protein kinases from S and M phase-arrested human cancer cells. We identified more than 1000 phosphorylation sites on protein kinases. Intriguingly, half of all kinase phosphopeptides were upregulated in mitosis. Our data reveal numerous unknown M phase-induced phosphorylation sites on kinases with established mitotic functions. We also find potential phosphorylation networks involving many protein kinases not previously implicated in mitotic progression. These results provide a vastly extended knowledge base for functional studies on kinases and their regulation through site-specific phosphorylation.

In the study, they identified phosphorylation sites for 219 protein kinases, of which 159 showed differential phosphorylation (at least two-fold induction for at least one site) in S and/or M phase.

My collaborators at CBS and I have previously shown that transcriptional and posttranslational regulation (for example, phosphorylation by cyclin-dependent kinases) tend to target the same proteins (de Lichtenberg et al., 2005; Jensen et al., 2006). One should thus expect that the differentially regulated kinases have a tendency to be encoded by periodically expressed genes.

To test this hypothesis, I compared the phosphoproteomics data of Daub et al. to the cell-cycle microarray expression study by Whitfield et al. (2002). I was able to map 132 of the 159 kinases to the microarrays and found that 17 of them are encoded by the top-600 cycling genes. This corresponds to a significant (P < 0.001) two-fold overrepresentation of transcriptional cell-cycle regulation among the genes encoding kinases that are differentially phosphorylated during S and/or M phase.

One could imagine that this trend is not specific to kinases that are differentially phosphorylated during the cell cycle, but that it instead applies to kinases in general. To test this, I also mapped the 60 non-modulated kinases found by Daub et al. to the microarrays (Whitfield et al., 2002). Of the 54 kinases that could be mapped, only 3 are encoded by periodically expressed genes, which is almost exactly what is expected by random chance.

I next examined if timing of phosphorylation correlates with the timing of expression of the 17 kinases mentioned above. The kinases can be divided into three classes: phosphorylated in S phase, phosphorylated in M phase, and phosphorylated in both S and M phase. Notably, 13 of the 17 kinases fall in to the M phase class. Looking at the peak times of expression for these (that is when in the cell-cycle the corresponding mRNAs are most highly expressed) reveals that 8 of the 13 kinases are presumably synthesized in M phase only shortly before they become phosphorylated.

In summary, comparison of the phosphoproteomics data from Daub et al. (2008) and the microarray expression data from Whitfield et al. (2002) supports the view that transcriptional and posttranslational regulation tend to target the same proteins during the mitotic cell cycle. Moreover, it shows that for most of the kinases that are subject to such dual cell-cycle control, both expression and phosphorylation takes place during M phase when the cyclin-dependent kinase activity is maximal.

Full disclosure: I currently collaborate with Matthias Mann and members of his group, and we will soon be colleagues a the Novo Nordisk Foundation Center for Protein Research.

WebCiteCite this post