New & Noteworthy

Old Genes, New Tricks

March 14, 2013

You can’t teach an old dog new tricks, or so the saying goes. But imagine you found that your old dog knew a complicated trick and had been doing it all her life, right under your nose, without your ever noticing it! You’d be surprised – about as surprised as the Hinnebusch group at NIH when they discovered that some long-studied S. cerevisiae genes had an unexpected trick of their own.

They were working on the VPS* (vacuolar protein sorting) genes. While known for a very long time to be important in protein trafficking within the cell, Gaur and coworkers found that two of these genes, VPS15 and VPS34, play an important role in RNA polymerase II (pol II) transcription elongation too. Now there is an unexpected new trick…like your dog learning to use a litter box!

There had been a few hints in recent years that the VPS genes, especially VPS15 and VPS34, might have something to do with transcription. Following up on these, the researchers tested whether vps15 and vps34 null mutants were sensitive to the drugs 6-azauracil and mycophenolic acid. Sensitivity to these drugs is a hallmark of known transcription elongation factors. Sure enough, they were as sensitive as a mutant in SPT4, encoding a known transcription elongation factor. Further experiments with reporter genes and pol II occupancy studies showed that pol II had trouble getting all the way to the end of its transcripts in the vps mutant strains.

There was a bit of genetic interaction evidence that had suggested that there might be a connection between VPS15, VPS34, and the NuA4 histone acetyltransferase complex. This is important, since NuA4 is known to modify chromatin to help transcription elongation. Looking more closely, the researchers found that Vps34p and Vps15p were needed for recruitment of NuA4 to an actively transcribing reporter gene.

Other lines of investigation all pointed to the conclusion that these VPS proteins have a role in transcription. They were required for positioning of several transcribing genes at the nuclear pore, could be cross-linked to the coding sequences of transcribing genes, and could be seen localizing at nucleus-vacuole junctions near nuclear pores.

One appealing hypothesis to explain this has to do with what both genes actually do. Vps34p synthesizes phosphatidylinositol 3-phosphate (PI(3)P) in membranes, while Vps15p is a protein kinase required for Vps34p function. The idea is that when Vps15p and Vps34p produce PI(3)P at the nuclear pore near transcribing genes, this recruits the NuA4 complex and other transcription cofactors that can bind phosphoinositides like PI(3)P. There are hints that this mechanism may also be at work in mammalian and plant cells.

There’s a lot more work to be done to nail down the exact role of these proteins in transcription. But this story is a good reminder to researchers that new and interesting discoveries may always be hiding in plain sight.

* These genes were also called VPL for Vacuolar Protein Localization and VPT for Vacuolar Protein Targeting

by Maria Costanzo, Ph.D., Senior Biocurator, SGD

Categories: Research Spotlight

Tags: RNA polymerase II, Saccharomyces cerevisiae, transcription, VPS genes

Sixty New Expression Analysis Datasets

March 05, 2013

Sixty new datasets have been added to our expression analysis tool at SGD, facilitating the rapid identification of co-expressed genes based on patterns of expression shared with query gene(s) across the entire collection. Expression data are now available at SGD from a comprehensive collection of 430 datasets representing 9190 microarrays from a total of 286 publications. The expression analysis tool can be accessed via the Expression tab and Expression Summary histogram located on Locus Summary pages, or using the ‘Expression’ option in the Function pulldown in the menu bar at the top of SGD pages. The new data will by default be included with the previous data when using the ‘New Search’, ‘Show Expression Levels’, or ‘Dataset Listing’ options. Alternatively, the new datasets can be specifically filtered using the dataset tag ‘not yet curated’. All of the RNA expression data are available for download in expression directory. Datasets are grouped by publication and are in PCL format.

Categories: New Data

Cancerous Avalanche

March 05, 2013

Cancer often gets going with chromosome instability.  Basically a cell gets a mutation that causes its chromosomes to mutate at a higher rate.  Now it and any cells that come from it build mutations faster and faster until they hit on the right combination to make the cell cancerous.  An accelerating avalanche of mutations has led to cancer.

There are plenty of obvious candidates for the genes that start these avalanches: genes like those involved in segregating chromosomes and repairing DNA, for example.  But there are undoubtedly sleeper genes that no one has really thought of.  In a new study out in GENETICS, Minaker and coworkers have used the yeast S. cerevisiae to identify three of these genes — GPN1 (previously named NPA3), GPN2, and GPN3.

A mutation in any one of these genes leads to chromosomal problems.  For example, mutations in GPN1 and GPN2 cause defects in sister chromatid cohesion and mutations in GPN3 confer a visible chromosome transmission defect.  All of the mutants also show increased sensitivity to hydroxyurea and ultraviolet light, two potent mutagens.  And if two of the genes are mutated at once, these defects become more severe.  Clearly, mutating GPN1, GPN2, and/or GPN3 leads to an increased risk for even more mutations!

What makes this surprising is what these genes actually do in a cell.  They are responsible for getting RNA polymerase II (RNAPII) and RNA polymerase III (RNAPIII) into the nucleus and assembled properly.  This was known before for GPN1, but here the authors show that in gpn2 and gpn3 mutants, RNAPII and RNAPIII subunits also fail to get into the nucleus. Genetic and physical interactions between all three GPN proteins suggest that they work together in overlapping ways to get enough RNAPII and RNAPIII chugging away in the nucleus.

So it looks like having too little RNAPII and RNAPIII in the nucleus causes chromosome instability. This is consistent with previous work that shows that mutations in many of the RNAPII subunits have similar effects.  Still, these genes would not be the first ones most scientists would look at when trying to find causes of chromosomal instability. Score another point for unbiased screens in yeast leading to a better understanding of human disease.

by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics

Categories: Research Spotlight, Yeast and Human Disease

Tags: cancer, chromosome instability, RNA polymerase II, RNA polymerase III, Saccharomyces cerevisiae

Crowdsourcing Genetic Disease

February 28, 2013

Remember when sequencing the human genome was going to help us better understand and treat complex diseases like Type 2 diabetes or Parkinson’s? Well, ten years later, we’re still waiting.

Sure we’ve made some progress. Using genome wide association studies (GWAS), scientists have uncovered markers here and there that explain a bit about how a genetic disease is inherited. But despite a seemingly never-ending stream of these assays, scientists simply can’t explain all of the genetics behind most of these diseases.

So now scientists need to try to explain this missing heritability. If they can find out why they aren’t getting the answers they need from GWAS, then maybe they can restructure these assays to give better results.

As usual, when things get dicey genetically, scientists turn to the yeast Saccharomyces cerevisiae to help sort things out. And in a new study out in Nature, Bloom and coworkers have done just that.

In this study, they mated a laboratory and a wine strain of yeast to get 1,008 test subjects from their progeny. They extensively genotyped each of these 1008 and came up with a colony size assay that allowed them to determine how well each strain grew under various conditions. They settled on 46 different traits to study genetically.

What they found was that none of these traits was determined by a single gene. In fact, they found that each of the 46 different traits had between 5 and 29 different loci associated with it, with a median of 12 loci. This tells us that at least in yeast, many genetic loci each contribute a bit to the final phenotype. And if this is true in people, it could be a major factor behind the missing heritability in GWAS.

If a trait is dependent on many genetic loci that each have a small effect, then researchers need large populations in order to tease them out. In fact, when Bloom and coworkers restricted their population to 100 strains, they could only detect a subset of the genetic loci. For example, the number of loci went from 16 to 2 when they looked at growth in E6 berbamine.

So it may be that scientists are missing loci in GWAS because there are simply too few participants in their assays. If true, then the obvious answer is to increase the size of the populations being studied. Thank goodness DNA technologies get cheaper every year!

Of course as the authors themselves remind us, we do need to keep in mind that humans are a bit more complex than yeast. There may be other reasons that we aren’t turning up the genetic loci involved in various traits. It may be that we can’t as accurately measure the phenotypes in humans or that human traits are more complicated than the yeast ones studied. Another possibility is that in humans, there are more rare alleles that can contribute to a given trait. These would be very hard to find in any population studies like GWAS.

Still, this study at the very least tells us that larger populations will undoubtedly uncover more loci involved in human disease. Thank you again yeast.

by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics

Categories: Research Spotlight, Yeast and Human Disease

Tags: GWAS, model organism, Saccharomyces cerevisiae

Puzzling Out Gene Expression

February 21, 2013

Have you ever put together a million piece puzzle that was all blue? That is sort of what it sometimes feels like figuring out how genes are turned on or off, up or down.

There are hundreds or even thousands of proteins called transcription factors (TFs) controlling gene expression. And there is a seemingly simple but frustratingly opaque string of DNA letters dictating which TFs are involved at a particular gene. Figuring out which sets of proteins bind where to control a gene’s expression can be a baffling ordeal.

Up until now most of the ways of identifying which TFs are bound at which genes have been incredibly labor intensive to do on a large scale. With all of the current techniques, researchers need to construct sets of reagents before they even get started. For example, to be able to immunoprecipitate TFs along with the DNA sequences they bind, you need to insert epitope tags in all the TF genes so an antibody can pull them down. Other techniques are just as involved.

What the field needs is a quick and dirty way to find where TFs bind in the genome. And now they just might have one.

In a new study, Mirzaei and coworkers used a modification of the well-known technique mass spectrometry (mass spec) to identify TFs that bind to a specific piece of DNA. With this technique, called selected reaction monitoring, the mass spec looks only for specific peptide sequences. This not only makes it much more sensitive and reproducible than ordinary mass spec, but it should also be relatively straightforward to do if a lab has access to the right sort of mass spec. They haven’t worked out all the bugs and it is definitely still a work in progress, but the technique looks promising.

Mirzaei and coworkers set up assays to detect 464 yeast proteins that are known or suspected to be involved in regulating RNA polymerase II transcription. Then they tested their assay on a 642 base pair piece of DNA known to contain signals that affect the levels of FLO11 transcription. They found fifteen proteins (out of the 222 they searched) that bound this piece of DNA. Of these, only one, Msn1p, had been previously identified as regulating the FLO11 gene. The other fourteen had not been found in any previous assays.

The authors next showed that two of these fourteen proteins, Mot3p and Azf1p, represented real regulators of the FLO11 gene. For example, deletion of MOT3 led to a threefold increase in FLO11 expression under certain conditions. And when AZF1 was deleted, FLO11 could not be activated under a different set of conditions. So Mot3p looks like a repressor of FLO11 and Azf1p looks like an activator.

This was a great proof of principle experiment, but much more work needs to be done before this will become a standard assay in the toolkit of scientists studying gene expression. They need to figure out why some known regulators of FLO11 (Flo8p, Ste12p, and Gcn4p) were missed in the assay and whether the other twelve proteins they discovered play a role in the regulation of the FLO11 gene.

Having said this, it is still important to note that even this early stage model of the assay identified two proteins that scientists did not know controlled FLO11 gene expression. At the very least this is a quick and easy way to quickly identify candidates for gene expression. We may not be able to use it to see the whole picture on the puzzle, but it will at least get us a good start on it.

by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics

Categories: Research Spotlight

Tags: RNA polymerase II, Saccharomyces cerevisiae, transcription

The Rhythm of Ribosomes

February 13, 2013

We all know that some people march to the beat of a different drummer. But now we’re finding out that mRNAs also have their own particular rhythms as they move along the ribosome.

It’s long been known that some codons just work better than others. They are translated faster and more accurately mostly because they interact more strongly with their tRNAs and because there are more of their specific tRNAs around. So why hasn’t evolution gotten rid of all the “slow” codons? With only optimal codons, translation could move at a marching beat all the time.

One idea has been that a few pauses every now and then are a good thing. For example, maybe slowing down translation at the end of a stretch coding for a discrete protein domain gives that domain time to fold properly. This would make it less likely for the polypeptide chain to end up tangled, or misfolded. Great thought, but even when researchers looked in multiple organisms, they couldn’t find a consistent correlation between codons used and protein structure. Until now, that is.

In a recent study published in Nature Structural and Molecular Biology, Pechmann and Frydman took a novel approach to this question. They derived a new formula to measure codon optimality. Using it they found that codon usage was highly conserved between even distantly related species, and that this conservation reflected the domain structure of the particular protein a ribosome was translating.

First, the authors came up with a more accurate way of classifying codons as optimal or non-optimal. They took advantage of the huge amount of data available for S. cerevisiae and included a lot more of it in the calculation, such as the abundance of hundreds of mRNAs and their level of ribosome association. They also took into account competition between tRNAs based on supply and demand, something that the previous studies had not done.

Once they developed this new translational efficiency scale, they applied it to ten other yeast species – from closely related budding yeasts all the way out to the evolutionarily distant Schizosaccharomyces pombe. The authors found that positions of optimal and non-optimal codons were indeed highly conserved across the yeasts. And codon optimality was highly correlated with protein structure.

One of the better examples of this is alpha helices. These protein domains form while still inside the ribosomal tunnel. The authors found that the mRNA regions coding alpha helices use a characteristic pattern of optimal and non-optimal codons to encode the first turn of the helix. They theorize that this sets the rhythm for folding the rest of the helix. Other structural elements are coded by distinct codon signatures too.

This isn’t just interesting basic research. It has some far-reaching practical implications too.

When using yeast to make some sort of industrial product, the thought has been to use as many optimal codons as possible. This has not always worked out, and now we may know why. A gene that tailors the codon usage to the rhythm of the protein structure is probably the best way to make a lot of correctly folded protein.

And the factory isn’t the only place where this kind of information will come in handy. Protein misfolding is the known or suspected culprit in a whole slew of human neurodegenerative diseases such as Alzheimer’s, ALS, Huntington’s chorea, and Parkinson’s disease. A better understanding of its causes might give us insights into managing those diseases.

Who knew in 1971 that translation actually is a rhythmic dance?

by Maria Costanzo, Ph.D., Senior Biocurator, SGD

Categories: Research Spotlight

Tags: evolution, protein folding, ribosome, Saccharomyces cerevisiae, translation

Giving the Keys Back to the Cell

February 06, 2013

When someone has a bit too much to drink, it is a good idea to take away their car keys. This keeps them safe until they can drive again. But the next morning, that hung over person needs to get their keys back so they can get to work.

Cells sometimes face a similar situation. Instead of being drunk though, cells have something go wrong while they are growing and dividing. When this happens, the cell stops the cell cycle at the next checkpoint, fixes what is wrong, and then starts the cell cycle back up again where it left off.

Scientists have learned a lot about how the keys are taken from cells, but not a whole lot about how they get them back. Fong and coworkers help to rectify this situation in a new study out in GENETICS. There they identified proteins key to releasing a yeast cell from its S-phase checkpoint.

If a cell’s DNA is damaged while it is growing and dividing, replication is slowed at the S-phase checkpoint. This gives the cell a chance to fix the DNA before it is copied. The authors found that in the absence of the DIA2 gene, yeast cells had trouble getting replication up and running again. This implies that this gene is required for yeast to overcome the S-phase checkpoint. The cell needs DIA2 to get its keys back.

Dia2p is an F-box protein involved in identifying certain proteins for destruction. It is one of several interchangeable subunits that provide specificity to the SCF ubiquitin ligase complex. The idea would be that Dia2p is important for degrading the “keeper of the keys,” the protein responsible for stopping the cell cycle in the S-phase.

To test whether Dia2p is important for checkpoint recovery, Fong and coworkers first activated the S-phase checkpoint by adding the DNA damaging agent MMS. Then they removed the MMS and measured how long it took the cells to finish copying their DNA. The dia2Δ mutant was significantly slower than wild type.

Given that Dia2p is involved in ubiquitin-mediated degradation, the authors reasoned that it may help a cell get out of S-phase arrest by degrading a protein that was keeping it there. To find this “keeper of the keys,” Fong and coworkers looked for mutations that rescued dia2Δ cells in the presence of high levels of MMS. The idea is that if they knock out the gene that is keeping the dia2Δ cells arrested, then the cells could overcome the block caused by the MMS.

One of the genes that came up in the screen was MRC1. To confirm that Dia2p and Mrc1p work together in releasing a yeast cell from the S-phase checkpoint, the authors constructed a double mutant carrying dia2Δ and a mutant version of MRC1, mrc1AQ, that they knew was checkpoint defective. Indeed, the double mutant behaved like wild type in their checkpoint recovery assay. Since the mutant Mrc1-AQp could not keep cells at the checkpoint, there was no need for Dia2p to target it for degradation. The double mutant cell never let go of its keys.

The simplest model to explain what happens in wild type is that when its DNA is damaged, a cell is prevented from progressing through S-phase by Mrc1p. Then when the DNA is repaired, Dia2p, providing specificity to the SCF ubiquitin ligase complex, targets Mrc1p for degradation. The cell is now released, allowing the cell cycle to continue.

The authors did a lot more work that we won’t go into here, but suffice it to say that Dia2p and Mrc1p are not the only players involved in releasing a cell from the S-phase checkpoint. There were other genes, both identified and unidentified, that came up in their screen. These will need to be studied as well.

And this isn’t all just interesting from a scientific standpoint. Many cancer treatments work by damaging the cancer cell’s DNA while it is growing and dividing. A better understanding of how cells are arrested and released may lead to better cancer treatments.

by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics

Categories: Research Spotlight, Yeast and Human Disease

Tags: DNA replication checkpoint, Saccharomyces cerevisiae

Ghosts of Centromeres Past

January 28, 2013

Every cell needs to correctly divvy up its chromosomes when it divides.  Otherwise one cell would end up with too many chromosomes, the other with too few and they’d both probably die.

Cells have developed elaborate machinery to make sure each daughter gets the right chromosomes.  One key part of the machinery is the centromere.  This is the part of the chromosome that attaches to the mitotic spindle so the chromosome gets dragged to the right place. 

Given how precise this dance is, it is surprising how sloppy the underlying centromeric DNA tends to be in most eukaryotes.  It is very long with lots of repeated sequences which make it very tricky to figure out which DNA sequences really matter.  An exception to this is the centromeres found in some budding yeasts like Saccharomyces cerevisiae.  These centromeres are around 125 base pairs long with easily identifiable important DNA sequences.

The current thought is that budding yeast used to have the usual diffuse, regional centromeres but that over time, they evolved these newer, more compact centromeres.  Work in a new study published in PLOS Genetics by Lefrançois and coworkers lends support to this idea.

These authors found that when they overexpressed a key centromeric protein, Cse4p (or CenH3 in humans), new centromere complexes formed on DNA sequences near the true centromeres. The authors termed these sequences CLR’s or Centromere-Like Regions.  And they showed that these complexes are functional.

When Lefrançois and coworkers kept the true centromere from functioning on chromosome 3 in cells overexpressing Cse4p, 82% of the cells were able to properly segregate chromosome 3.  This compares to the 62% of cells that pull this off with normal levels of Cse4p.  The advantage disappeared when the CLR on chromosome 3 was deleted.

A close look at the CLRs showed that they had a lot in common with both types of centromeres.  They had an AT-rich 90 base pair sequence that looked an awful lot like the kind of sequence that Cse4p prefers to bind and a lot like the repeats found within more traditional centromeres.  They also tended to be in areas of open chromatin and close to true centromeres. The obvious conclusion is that these are remnants of the regional centromeres budding yeast used to have. 

The hope is that the yeast CLRs might make studying regional centromeres easier.  They are so long and complicated that it is very difficult to pick out which sequences matter and which don’t, but the yeast CLRs could be a simpler model system.  Even better, the CLRs might shed some light on the process of neocentromerization – the formation of new centromeres outside of centromeric regions, which happens a lot in cancer cells. Once again, simple little S. cerevisiae may hold the key to understanding what’s going on in much larger organisms.

by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics

Categories: Research Spotlight, Yeast and Human Disease

Tags: centromeres, evolution, Saccharomyces cerevisiae, yeast model for human disease

Autophagy’s (Atg)9th Symphony

January 17, 2013

(Please click the musical note and listen to the music while reading.) The music you’re listening to starts off with a marimba. Then a flute joins in and as the marimba fades, in comes a shamisen. The piece progresses similarly with a harp, and then ends with the reappearance of the marimba. A nice, jaunty little piece of music.

This song is actually a tool for learning about autophagy in the yeast S. cerevisiae. Autophagy is a way to break down damaged or no longer useful proteins and recycle their components for later use. It is a very important pathway in keeping starving yeast alive. Many of the proteins involved in autophagy are highly conserved, and autophagy defects are implicated in several kinds of human disease.

As described in a recent paper, Takahashi and coworkers converted the sequences of four proteins involved in a step in autophagy – Atg9, Sso1, Sec9, and Sec22 – into pieces of music using UCLA’s Gene2Music program. Each protein was then assigned a musical instrument. Atg9 was played with the marimba, Sso1 with the flute, Sec9 with the shamisen, and Sec22 with the harp. The orchestrated piece of music reflects how each protein interacts with the others in the autophagy pathway.

Atg9 is a transmembrane protein that is key to making the vesicles that carry the damaged or unused proteins to the lysosome for destruction. But it, like the marimba, is not enough. Atg9 is recruited into service by at least three other proteins, Sso1, Sec9, and Sec22. These appear in succession in the musical piece as a flute, shamisen, and harp. Just like all four are needed for the orchestral piece, all four are also needed for successful autophagy.

Now listen to the music again. With this background, did you find the piece more illuminating? If you didn’t, it may simply be because it doesn’t fit your learning style, or match the type of intelligence that is your strength. Some people may respond to music better than they do to pictures of pathways or memorizing the steps involved. It may be that these people’s understanding of complicated pathways is enhanced with a musical component.

There will need to be more research on musical representation of complex pathways to see if they actually help students and even the public better understand science. If they do, I am looking forward to hearing the Krebs Cycle put to music. Or the assembly of the RNA polymerase II preinitiation complex. Which pathways do you want put to music?

by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics

Categories: Research Spotlight

Tags: autophagy, music, Saccharomyces cerevisiae, teaching

P is for Protection (not Processing)

January 11, 2013

Growing and dividing are dangerous work for a cell. Making all that energy throws off free radicals that mutate DNA and wreak havoc with delicate intracellular machinery. Given this it might seem surprising that just sitting there, not growing, is dangerous too. And yet it looks like it is.

When a cell runs out of food and goes into a quiescent state, it creates ribonucleoprotein (RNP) complexes called processing bodies (P bodies). In a new study out in GENETICS, Shah and colleagues were able to control how well yeast cells could make these P bodies. What they found was that cells that had trouble making P bodies didn’t survive the quiescent state as well as those cells that were great P body makers. It looked like P bodies were doing something to protect the cell when it wasn’t growing. In other words, being quiescent is dangerous too.

The key discovery made by the authors that allowed them to do these experiments was the fact that the Ras/PKA signaling system works specifically through the Pat1 protein to make P bodies. So by controlling the sensitivity of Pat1p to the signaling system, they could control the number of P bodies in the cell.

The Ras/PKA pathway phosphorylates two serine residues on Pat1p. When they are phosphorylated, P bodies are disrupted and/or are prevented from forming. The Pat1-EE mutation replaces the serine residues with glutamic acids, mimicking the phosphorylated state. The authors found that yeast cells carrying Pat1-EE produced fewer, smaller P bodies than did yeast carrying the wild type version of Pat1.

The authors then used this constitutively active mutant to ask whether P bodies helped cells survive the quiescent state. They compared the survival rate of cells carrying either the wild type version or the Pat1-EE protein and found that cells carrying the wild type version of Pat1 were more likely to survive after quiescence than were those cells carrying the constitutive form. More P bodies led to better survival.

The authors don’t yet know why this is, but one idea is that proteins and RNAs critical for survival after quiescence are stored in these particles. The idea would be that cells that have these key components squirreled away and protected survive better than those cells where these proteins and RNAs have degraded.

As a final point, it is important to mention why this matters (besides the excitement of figuring out how things work). Quiescent yeast cells are used as models for aging in higher eukaryotes like us. Perhaps by understanding how to make a yeast cell better survive this non-growing state, we can learn something about how to make people live longer too.

by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics

Categories: Research Spotlight

Tags: P bodies, quiescence, Saccharomyces cerevisiae

Next