New & Noteworthy

Budding Yeast, a Caffeine Wimp No More

December 16, 2016

---

Some people get the jitters from a single espresso while others need a triple shot just to get started in the morning. Some of this is due to caffeine tolerance—a buildup of resistance to the marvelous effects of that wonderfully addictive substance, caffeine. But the rest has to do with genetic differences that affect how well each of us processes caffeine—our caffeine sensitivity.

Our best buddy Saccharomyces cerevisiae is a real wimp when it comes to caffeine. In fact, like a lot of other microorganisms, caffeine actually kills this yeast. S. cerevisiae is indeed a sensitive soul when it comes to caffeine.

In a new study in the Journal of Agricultural and Food Chemistry, Wang and coworkers were able to toughen up budding yeast against caffeine by adding bfr1, a gene from Schizosaccharomyces pombe that encodes the ABC transporter that shunts caffeine out of the cell. And then, using random mutagenesis, they were able to make bfr1 even better at its caffeine-exporting job. Although the yeast don’t get any of the pleasurable effects of caffeine, at least they can now happily grow in cultures that have more caffeine than a strong cup of coffee.

This new attribute could prove to be incredibly useful if caffeine producers ever want to start making caffeine biologically instead of synthetically. You can imagine adding the caffeine pathway from coffee to yeast and having the yeast merrily exporting caffeine to the culture medium where it can be harvested. And who knows, maybe they can have the yeast make caffeine and alcohol at the same time creating the equivalent of a vodka and Red Bull in a single step!

Previous research had shown that bfr1 was an important player in helping S. pombe deal with caffeine. When Wang and coworkers added the gene to S. cerevisiae, this newly engineered yeast could now better tolerate caffeine. For example, whereas wild type yeast barely grew with 8 mg/ml caffeine, the engineered yeast did OK.

These authors next turned to random mutagenesis of the bfr1 gene to screen for mutants that could tolerate even more caffeine. And boy did they win the lottery on this one! A mutant that they named bfr1-B did great even at concentrations of 25 mg/ml caffeine. Now they were getting somewhere.

Bfr1 doesn’t just export caffeine; it actually exports many different compounds. The authors found that bfr1-B was fairly specific for increased resistance to caffeine. For example, when they tested the bfr1-B mutant with theophylline, a structurally similar compound, and atropine, a structurally distinct compound, they found that S. cerevisiae expressing the mutant were, if anything, more sensitive to these compounds. They found what looked like a caffeine-specific mutant. 

When they looked at the mutant, Wang and coworkers found that there were 11 amino acid substitutions scattered across the protein. The next step was to figure out which ones mattered and which ones didn’t.

Using a bit of modeling with the 3-D structure of other ABC transporters, they settled on testing three mutations individually. Two of the mutations, S36 and D340 were in the nucleotide binding domain (NBD) and the third, Y497, was in the transmembrane domain (TMD). The NBD is where ATP binds to the transporter to supply the energy to move caffeine across the membrane. 

Of the three, only D340 in the nucleotide binding domain conferred caffeine resistance. While not as robust as bfr1-B, this mutant allowed yeast expressing it to tolerate caffeine concentrations up to 15 mg/ml, conditions under which cells with wild type bfr1 failed to grow. 

So while this mutation explains a lot of why bfr1-B is so good at dealing with caffeine, it is not the whole story. At least some of those other 10 mutations contribute to how well bfr1-B does with caffeine.

In the end we have a bullet-proof yeast when it comes to caffeine that should prove useful for anyone who wants yeast to synthesize caffeine for them. Of course unlike even the most grizzled 30 year coffee drinker with ideal genetics, the yeast almost certainly gets no joy from its morning Joe. But at least that cup of coffee won’t kill it!

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

Categories: Research Spotlight

Tags: ABC transporter, BFR1, caffeine, Schizosaccharomyces pombe

Move Along, Transcription Factor

December 07, 2016

---

For an election to go smoothly, people cannot stay too long in the voting booth. If a lot of people stayed in the booth and answered emails, sent texts, etc., after they finished voting, then the whole process would grind to a halt.

There is some evidence that activating genes may work similarly. The transcription factors (TFs) that bind DNA and turn up the expression of nearby genes can’t stay too long. If they do, the activation starts to peter out.

What is thought to happen with these sorts of TFs is that they bind their preferred DNA, and then once they have attracted the cellular machinery needed to read the gene, they are targeted for destruction. Then a new TF can bind and repeat the process.

In a new study in GENETICS, Akhter and Rosonina set out to investigate the process by which the yeast transcription activator Gcn4p is removed after it has bound DNA and done its job. Gcn4p activates a number of genes in response to amino acid starvation.

They found that a key step in the process is the addition of SUMO proteins to DNA-bound Gcn4p, which gets the ball rolling on the destruction of Gcn4p. Imagine a sumo wrestler settling in next to a voter once he enters the booth and then throwing him out if he tarries too long.

Their model is that once Gcn4p binds DNA, it is sumoylated. Then the DNA-bound, sumoylated GCN4 is further modified by kinases like Cdk8p, a component of the mediator complex which acts as a bridge between TFs and the cellular machinery responsible for reading a gene. This modified TF is then sent off to the 26S proteasome where it is degraded making room for an unmodified Gcn4p.

Previous research had shown that sumoylation of GCN4 required DNA binding. The first thing these authors did in this study was to determine if Gcn4p had to bind to its target DNA sequence in order to be sumoylated. It did not.

When they fused a mutant Gcn4p that could not bind DNA to the DNA binding domain of Gal4p, they found that this molecule was sumoylated at the correct places on the Gcn4p part of the fusion protein, lysines 50 and 58, when bound to a Gal4p binding site. Therefore, Gcn4p does not need to occupy its own DNA binding site in order to be sumoylated.

Another set of experiments showed that while DNA binding was required for sumoylation, interaction with RNA polymerase II (RNAP II), the enzyme that reads the genes that Gcn4p activates, does not appear to be necessary. For one of these experiments they used a temperature sensitive mutant of the largest subunit of RNAP II, Rpb1p, and showed that even at higher temperatures when RNAP II is inactive in these cells, DNA-bound Gcn4p is still sumoylated. In the other experiment they showed that DNA-bound
Gcn4p was still sumoylated when they used the “anchor away” technique to drag Rpb1p out of the nucleus and into the cytoplasm.

So DNA binding is sufficient, and the specific site is not important. And Gcn4p doesn’t have to be activated in order to be sumoylated.

Of course, turnover like this is a delicate thing. If Gcn4p is pulled off too soon, then it can’t activate as much as it might otherwise be able to do. This might affect the cell’s response to starvation just as much as Gcn4p staying put too long. Sort of like the sumo wrestler throwing a voter out of the voting booth before they could finish their voting can muck up the election.

Akhter and Rosonina created a fusion protein of Gcn4p and the yeast SUMO peptide Smt3p. Unlike Gcn4p, this protein is sumoylated before it binds DNA.

They found that yeast expressing this fusion protein fared less well under starvation conditions compared to yeast cells that expressed the wild type version of GCN4. And using chromatin immunoprecipitation (ChIP) analysis they showed that at least at the ARG1 gene, this was because there was less of the fusion protein bound under activating conditions.

So cells need for TFs to stay at the right place for the right amount of time. If they are pulled off too early or stay too long, the levels of activation can fall below what is best for the cells.

Unfortunately, we don’t have time to go over other experiments that tease out which kinases are important and when, but I urge you to read about them for yourselves. They take full advantage of the genetic tools available in yeast to make this sort of study possible…#APOYG!

Integrating all of this gives the following model:

Gcn4p is only a dimer when bound to DNA and this dimerization may be the signal for sumoylation by Ubc9p. A preinitiation complex forms through its interaction with the DNA-bound, sumoylated Gcn4p which brings in the enzyme RNAP II to transcribe the gene. Once the polymerase has left the nest, the kinase Cdk8p comes in and phosphorylates Gcn4p which signals Cdc4p/Cdc34p to ubiquitinate Gcn4p. The ubiquitinated Gcn4p is then degraded by the 26S proteasome opening the upstream activator sequence (UAS) up to a fresh, new Gcn4p.

Here, with the help of our super hero Saccharomyces cerevisiae, Akhter and Rosonina have dissected out what happens to a transcription factor once it binds to DNA (at least ones that bind for short times). It will be fascinating to see if this translates to other TFs in other beasts. While I love yeast for all it can do for us for bread, wine, beer, human health, helping solve world problems like climate change, and so on, I think my favorite use is still that it allows us to better understand the basic biology of how our cells work.

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

Categories: Research Spotlight

Tags: Cdk8, Gcn4, gene activation, sumoylation, transcription

Yeast Tackles Climate Change

November 29, 2016

---

One of the best parts about doing outreach with a museum is creating a successful hands on activity for the visitors. This is not an easy thing to do.

You first create something that you think will appeal to and educate visitors. When that falls flat on its face, you then do a series of tweaks until it is working smoothly. In the end you have smiling kids who understand DNA better (or at all)!

Genetic engineering can be similar. You can import the genes for a complex pathway into your beast of choice but it may not work first try. A bit of tinkering, evolution style, is often needed to get the engineering working well enough to be useful.

This is exactly what happened when a group of researchers tried to get our favorite beast, the yeast Saccharomyces cerevisiae, to turn the sugar xylose into ethanol. They added the right genes from either fungi or bacteria, but S. cerevisiae couldn’t convert enough xylose into ethanol to be useful.

And it is important for all of us that some beast be able to do this well. A yeast that can turn xylose into ethanol means a yeast that can turn a higher percent of agricultural waste into a biofuel. Which, of course, means lots of low carbon fuel to run our cars so that we have a better shot of limiting the Earth’s heating up by 1.5-2 degrees Celsius.

To get their engineered yeast to better utilize xylose, Sato and coworkers forced it to grow with xylose as its only carbon source. Ten months and hundreds of generations later, this yeast had evolved into two new strains that were much better at turning xylose into ethanol. One strain did its magic with oxygen, the other without it.

In a new study in PLOS Genetics, Sato and coworkers set out to figure out which of the mutations that came up in their evolution experiments mattered and why.

Two genes were common in both the aerobic and anaerobic strains – HOG1 and ISU1. Both needed to be nonfunctional in order to maximize ethanol yields from xylose. They confirmed this by deleting each individually and together from the parental strain.

HOG1 encodes a MAP kinase, and ISU1 encodes a mitochondrial iron-sulfur cluster chaperone. These probably would not have been the first genes to go after with a more biased approach. The benefits of evolution and natural selection!

Further experiments showed deleting each gene individually was not as good as deleting both at once when oxygen was around. In fact, while deleting only ISU1 had a small effect on the ability of this yeast to convert xylose into ethanol, deleting HOG1 alone had no effect at all. Its deletion can only help a strain already deleted for ISU1.

In the absence of oxygen, yeast needs a couple of additional genes mutated – GRE3 and IRA2. GRE3 is an aldose reductase and IRA2 is an inhibitor of RAS. Again, not very obvious genes!

Still, once you find the genes you can come up with reasonable hypotheses for why they are important.

Some are easier than others. Hog1p, for example, is known to enhance a cell’s ability to turn xylose into xylitol, which shunts the xylose away from the ethanol conversion pathway. GRE3 is involved in this as well. Deleting either should make more xylose available to the yeast.

This doesn’t mean this is Hog1p’s only role in boosting this yeast’s ability to turn xylose into ethanol of course. It also probably “…relieves growth inhibition and restores glycolytic activity in response to non-glucose carbon sources.” Consistent with this, the authors found that the xylose-utilizing strain deleted for HOG1 was also better at using glycerol and acetate as carbon sources.

Other genes were less obvious. For example, perhaps mutating ISU1 frees up some iron so extra heme can be made. Or alternatively, it may have increased the mass of mitochondria available. Again, probably would not have been the first gene to go after to improve yeast’s ability to convert xylose to ethanol.

Which again underlines the importance of letting natural selection improve an engineered organism as opposed to only trying to pick and tweak the genes you think are important. Biology is simply too complicated and our understanding too limited to be able to know which are the best genes to go after. This is reminiscent of prototyping museum activities.

Some tweaks are obvious but others you would never have guessed would be needed. For example, we had visitors spreading bacteria on a plate and found that if they labeled their plate first, they almost always put their transformation mixture on the lid instead of on the LB agar. This problem was solved by having them add their mixture first and then labeling their plate.

It would be very hard to predict something like this from the get-go. The activity needed to evolve on the museum floor to work optimally. Much like the yeast engineered to utilize xylose needed to evolve in the presence of xylose to work optimally. And to perhaps take a big step towards saving the Earth from warming up too much.

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

Categories: Research Spotlight

Tags: biofuel, ethanol, evolution, fermentation, glucose metabolism, xylose

Too Much of a Good Thing

November 09, 2016

---

Like a ruined cookie with too much salt, a cell can go haywire when it has too many copies of certain genes. And of course, cells can deal perfectly well with too many copies of other genes. Just like adding too many chocolate chips to your cookies might make an even better cookie!

Finding out which genes are like salt and which ones are like chocolate chips is of more than just general biological interest. It might help us to explain why cancer happens and to possibly find better treatments.

As you probably know, cancer cells are pretty messed up genetically. Their DNA is littered with mutations, rearrangements and somatic copy number amplifications (SCNAs).

A big reason for this genetic jumble is early DNA changes that increase the rate of mutations in a cell. This “mutator” trait makes a cell more likely to stumble on the mutations it needs to grow out of control or refuse to die.

In their new study in GENETICS, Ang and coworkers set out to find genes that can cause a mutator phenotype when they are part of a SCNA. In other words, which genes lead to an increased mutation rate when expressed at a higher level.

This is important because there are so many SCNAs in a typical cancer cell that it can be hard to figure out which ones matter and which ones don’t (or to put it into cancer parlance, to tell the drivers from the passengers). And despite all of the CRISPR hoopla and other mammalian resources, it would still be a very long process to find “dosage mutator” genes in cell culture and/or living animals.

Which is why Ang and coworkers used our favorite workhorse, the yeast Saccharomyces cerevisiae, to find genes that may cause an increased mutation rate when overexpressed.

The assay is conceptually simple. Yeast that have a functioning CAN1 gene do not survive in the presence of the drug canavanine. So these researchers looked for cells that did better in the presence canavanine when overexpressing a single gene. Presumably, they are surviving because that extra gene resulted in the CAN1 gene being mutated more often because of an increased mutation rate.

They found 37 genes that fit the bill, 18 of which that were involved in biological pathways known to affect genome stability. Combining this with previous studies that looked at gene deletions, this brings the grand total of suspected yeast mutator genes to 210.

Most of these 210 were identified because of mutations that made them stop working which can make figuring out why they cause the mutator phenotype relatively simple. For example, if a mutation kills a gene responsible for fixing DNA mistakes, then you are going to get more DNA mistakes in that cell. It is a little trickier to understand how extra copies of a gene might cause an increased mutation rate.

Ang and coworkers focused on trying to figure out the mechanism behind their top 5 dosage mutator genes: PIF1, MPH1, UBP12, RRM3, and DNA2. Since 4/5 of these code for helicases, they first checked to see if just being a helicase is enough to be a dosage mutator gene. It isn’t.

They retested 48 DNA helicases in their assay and found that none of them caused an increased mutation rate when mutated. There is more to a dosage mutator than being a helicase!

In the next set of experiments, they wanted to determine if the five strains, each overexpressing one of these five genes, had a higher mutation rate by the same mechanism. They tested this by determining the sensitivity of these 5 strains to 3 different DNA damaging agents. The idea is that if they share the same mechanism, they should have the same sensitivity profiles to each of these agents. They did not.

For example, overexpressing MPH1 resulted in a higher sensitivity to all three agents while overexpressing UBP12 only increased sensitivity to two of them. So each strain probably has an increased mutation rate for a different reason.

They next wanted to see if the increased mutation rate was due to a loss or gain of function. They did this by comparing the profiles of strains either deleted for or overexpressing the dosage mutator genes. The idea is that if overexpression leads to a loss of function, then deleting and overexpressing the genes should have the same profile. The three they could test like this did not.

The authors conclude from this that the increased mutation rate for MPH1, UBP12, and RRM3 is most likely due to the gain of an inappropriate function as opposed to a loss of function. In a final set of experiments, Ang and coworkers focused on what that new function might be in their strongest mutant, MPH1.

First they showed that of the three activities associated with Mph1p, only DNA binding and not its ATPase or helicase activities were important for it causing an increased mutation rate when overexpressed. From this they reasoned that perhaps Mph1p was displacing some other important DNA binding protein and that it was this displacement that was causing the increased mutation rate.

Through a set of experiments we don’t have time to go into here, they provided evidence that Mph1p was outcompeting the flap endonuclease Rad27p for DNA binding. This makes some sense as previous work had shown that deleting RAD27 causes mutation rates to go way up. So too much Mph1p keeps Rad27p from getting to where it needs to be with the end result being an increased mutation rate.

All this MPH1 work may have important implications in some human cancers. Nonsense or missense mutations in FANCM, the human homolog of MPH1, are known to make people more likely to get cancer. And there are examples of cancers where FANCM is overexpressed. Perhaps that overexpression results in an increased mutation rate in these cancers.

Yet again yeast is giving researchers new targets for, and new ways to think about, human disease. Thanks, yeast, for finding all of these mutator genes for us to investigate further! #APOYG!

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

Categories: Research Spotlight, Yeast and Human Disease

Tags: cancer, forward mutation, genome-wide, MPH1, mutator, overexpression

Personalized Essential Genes

November 02, 2016

---

Every smartphone has a few, key functions. For example, they all let you make calls, send texts, take pictures, and search the web. But smartphones also have their own specialized functions depending on who owns them.

Maybe you’re an avid Candy Crusher and so you have every iteration of that game on your phone. Or maybe you have an ESPN app that lets you watch football highlights. There are pretty much an infinite number of possible combinations to personalize your phone.

A new study out in PLOS Pathogens finds something similar for the XRN1 gene in yeast. Rowley and coworkers found that in terms of basic function, they could swap one XRN1 gene for another across 4 different Saccharomyces species. All these different versions of the XRN1 did their main job of degrading RNA just fine no matter which yeast species they were in.

But a closer look revealed that each version of this key gene had a personalized function that did not swap as well. And this specialization wasn’t something trivial like Apple Music vs. Spotify. The personalized XRN1 genes protected their own species of yeast against species-specific viruses better than the XRN1 genes from other species.

It will be interesting to see if something similar happens in people. Perhaps the human version of XRN1, which also plays a role in taming RNA viruses, works better with human-specific RNA viruses too. This might help to explain our poor response to viruses that move from one species to ours.

Rowley and coworkers used four different assays to show that XRN1 from Saccharomyces species cerevisiae, mikatae, kudriavzevii and bayanus were all interchangeable for rescuing a variety of growth defects present in a S. cerevisiae strain deleted for XRN1. These XRN1 genes are indistinguishable in terms of the basic function of degrading damaged or old RNA in the cell.

The same was not true of each gene’s ability to deal with RNA viruses.

The main virus that infects Saccharomyces goes by the name of the L-A virus. These RNA viruses are a little different from many other viruses in that they don’t spread from one yeast cell to another. Instead, they stay within their host cell and spread only when the infected yeast cell buds off a new daughter.

These researchers used three different assays to show that the XRN1 genes from different species worked less well than the cerevisiae XRN1 to reduce the viral load in Saccharomyces cerevisiae. All three assays were consistent with the S. cerevisiae version working best.

First, they used an assay that looked directly at the dsRNA of the L-A virus. In a cerevisiae strain deleted for XRN1, they saw a fat, juicy band on their gel. This band was unaffected when they added back a dead version of XRN1 (either E176G or Δ1206-1528) and severely reduced when they added back the complete XRN1 gene from S. cerevisiae.

The effect of XRN1 from other species depended on how closely related they were to S. cerevisiae. For example, the XRN1 from the more closely related S. mikatae was able to reduce the band a bit while the genes from the more distantly related S. kudriavzevii and S. bayanus had no discernible effect.

The second assay took advantage of a very cool RNA virus known as “killer”. It basically has the instructions for making a secreted toxin that kills any yeast around the host cell while sparing the host. It is completely dependent on the L-A virus.

Previous research showed that XRN1 affects how well yeast carrying both viruses kill off surrounding yeast. They measured this with something called a kill zone. The idea was to put a spot containing 6×10⁵ killer S. cerevisiae on a lawn of S. cerevisiae lacking the killer virus and to measure how big the “death” circle was that surrounded the spot.

Consistent with the results looking directly at the double stranded RNA of the L-A virus, Rowley and coworkers found that XRN1 from other Saccharomyces species were less able to negatively affect the ability of the killer virus to kill. This is presumably because they are less likely to degrade the virus meaning there is more of it around.

When they used S. cerevisiae XRN1, the kill zone averaged about 0.68 cm². S. mikatae, S. kudriavzevii, and S. bayanus average 0.92 cm², 0.96 cm², and 0.97 cm². S. cerevisiae harboring XRN1 from a different species were better killers.

The final experiment tested the ability of overexpressed XRN1 to cure S. cerevisiae of the L-A virus. In the absence of XRN1, 0/103 yeast managed to get rid of their virus. This number rose to 49% (78/159) when the XRN1 from S. cerevisiae was overexpressed. The XRN1 genes from other species fared much worse: only 12% (20/129) were cured with S. mikatae, 9% (11/123) with S. kudriavzevii, and 8% (10/120) with S. bayanus.

So it looks like the S. cerevisiae XRN1 gene has evolved to combat the L-A viruses that infect S. cerevisiae best. But is the reverse true? Are the XRN1 genes from the other species also specialized in their viral attacks? Looks like the answer is yes, at least for S. kudriavzevii.

The tricky part of answering this question is the only well characterized L-A virus is from S. cerevisiae. So their first experiment was to find a virus in another Saccharomyces species. With a little work they found one in S. kudriavzevii FM1138. The experiments they did with this strain weren’t as clean as the ones they did with S. cerevisiae because S. kudriavzevii is a trickier yeast to work with, but they found that XRN1 from S. kudriavzevii did best at reducing the amount of S. kudriavzevii-specific virus compared to the XRN1 genes from the other species.

So XRN1 does some important basic things in the cell like clearing out old and damaged RNA and these functions are pretty similar no matter which Saccharomyces species they come from. However, the same is not true for its role in keeping viruses in check. At least in S. cerevisae, its XRN1 does a way better job at keeping its endemic viruses manageable than do the XRN1 genes from three other Saccharomyces species.

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

Categories: Research Spotlight

Tags: pathogen, ribonuclease, virus

Winter is Coming (for Cancer)

October 17, 2016

---

I may be a little late to the game, but over the last few weeks I have started consuming episodes of Game of Thrones voraciously. It is such a fun show to watch! And this isn’t the only HBO show I enjoy. Veep, Silicon Valley, and Last Week Tonight with John Oliver all have my attention as well.

You might say that I need HBO, because without it I can’t get my fill of these shows. (Well, there are other routes, but HBO or HBO GO are the easiest). My over watching of these shows has made me dependent on HBO.

Something similar can happen in cancers. Sometimes a key player in keeping a cell cancerous is an overexpressed gene. And just like my binging of Game of Thrones makes me dependent on HBO, so this overexpressed gene (the TV show) makes the cancer cell dependent on another gene (HBO).

A real life example might make this clearer. Most folks have heard of BRCA1 and BRCA2 especially since the Angelina Jolie story. When either of these genes is damaged, you can end up with cancer.

Both of these genes are involved in DNA repair and damaging them means the cell builds up mutations. Making lots of DNA mistakes is a good thing for cancers but only up to a point. Too much damage and the cancer cell dies.

What this means is that these cancer cells are now more dependent on other DNA repair genes. Which means these other DNA repair genes are now targets to go after to selectively kill the cancer cells.

For cancer cells lacking BRCA1 or BRCA2 function, research has shown that these cells are now dependent on a second gene, PARP1. If PARP1 expression is turned down, normal cells survive but BRCA1/BRCA2-dependent cancers die. So, we can kill cancer cells, or end their TV show watching, by going after PARP1, their HBO.

Finding these sorts of genes is not easy unless, of course, you turn to our favorite lab workhorse, the yeast Saccharomyces cerevisiae. Given all of the genetic techniques and tools available with this yeast, it is possible to quickly do a synthetic dosage lethality assay – to look for genes that are lethal only in combination with deleting your gene of interest.

This is just what Reid and coworkers did in a new study just out in GENETICS for CKS1B, a gene that is amplified and overexpressed in many cases of breast, lung, and liver cancers. And they found a more “druggable” target to go after, the kinase PLK1 (the human homolog of yeast CDC5). PLK1 even comes with its own kinase inhibitor, Volasertib.

Reid and coworkers transformed a low copy plasmid containing the CKS1 gene, the yeast homolog of CKS1B, under the control of the galactose promoter into two different yeast strain libraries. The first screen used 9600 yeast deletion strains, each with a single gene deleted in either a MATa or MATα strain. The second screen used strains with temperature sensitive mutants of essential genes. They now looked to see which yeast strains did poorly or couldn’t survive when they were overexpressing CSK1 in the presence of galactose.

In the end they came up with 44 different genes that, when deleted or weakened, had a severe effect on the growth of yeast that overexpressed CKS1. Given that CKS1 plays an important role in cell cycle progression, they focused on the 15 genes that affect mitotic progression. Eventually, through a set of experiments that I don’t have time to go into here, they settled in on CDC5, a polo-like kinase involved in both mitotic entry and exit.

The next step was to see if what they learned about in yeast has any bearing on cancer. It did.

First Reid and coworkers looked at a variety of cancer cells in The Cancer Genome Atlas (TCGA) and found that it was very rare for both PLK1 and CKS1B to be overexpressed in the same cancer at the same time. Next they looked at a data set of short hairpin RNA (shRNA) knockdowns of ~16,000 human genes and found that knocking down PLK1 had negative effects on cancers overexpressing CKS1B. These are consistent with the two genes having a synthetic lethal relationship.

They then took eight breast cancer lines where the shRNA against PLK1 had a negative effect on growth and tested the effects of targeting PLK1 on apoptosis. Did decreasing expression of PLK1 in cells that overexpress CKS1B cause an increase in apoptosis in their hands?

The short answer is yes. They repeated the experiments with the shRNA and also tested the PLK1-specific kinase inhibitor Volasertib and found that both treatments increased apoptosis in CKS1B overexpressing cancer cells. It looks like they may have uncovered a way to go after a subset of cancers using yeast!

Which shouldn’t surprise us. Yeast and other model organisms have been teaching us about cancer at least since the days when Hartwell, Hunt and Nurse first identified cyclins and CDKs (for which they got the 2001 Nobel Prize in Physiology or Medicine), and will continue to school us for years to come.

Hopefully researchers will continue to turn to yeast to continue to better understand and find new treatments for cancer. Yeast has so much more to teach us! #APOYG!

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

Categories: Research Spotlight, Yeast and Human Disease

Tags: CKS1, cyclin-dependent kinase, polo-like kinase, synthetic dosage lethal

A Biological Tour de Force Reveals the Complexity of a Yeast Cell

October 11, 2016

---

As anyone who has tinkered with the inside of a living cell or looked at one of those daunting and overwhelming biochemistry poster knows, life is complicated. Like thousands of gears all interconnected in some vast steampunk machine, a cell has thousands of genes making thousands of proteins that come in a variety of flavors all interacting in overlapping, complicated ways to keep the cell alive. Toss in those important RNAs and a few other bells and whistles and you begin to understand why scientists have had to just focus in on individual parts of the machine to study.

Until now that is. In a new study out in Science led by University of Toronto Professors Brenda Andrews and Charles Boone, and Professor Chad Myers of the University of Minnesota-Twin Cities, Costanzo and coworkers were able to get a first approximation of how all those genes are connected in a yeast cell. And it is as complicated as you imagined.

This was a daunting undertaking that took over 15 years to pull off but it looks like it was well worth it. They have already learned an incredible amount and people will continue to learn more and more as scientists mine this treasure trove of data for a long time to come (Stay tuned for the data to be available here at SGD). This is the sort of study that will provide an invaluable reference guide that will keep giving year after year after year.

And it isn’t just a tour de force of basic biology either. This study will, among other things, reveal what genes of unknown function are actually doing in cells, identify new targets to go after in diseases like cancer and may even help geneticists find that “missing heritability” in their genome-wide association studies (GWAS). Given that these three examples are just a subset of what we’ll get out of this study, I think you can see why it is such a game changer.

It would not have been possible without all of the yeast work that has been done and carefully curated over the years.  And there is no way it could have been done in another organism, like humans. Human cells are simply too complicated, we don’t yet have the necessary tools, and we don’t have as deep a knowledge as we do with yeast. It is only in yeast that we could begin to get this amazing peek at life happening inside of a cell.

Yeast as a model organism is alive and well and helping us better understand life and ourselves. This sort of study drives that point home in a way that benefits many, many scientists and, eventually, patients the world over.

When 1+1 Does Not Equal 2

So how did these researchers pull this off? They basically looked at what happens when they knock out two genes at once in yeast. They did this for every possible gene pair for 5416 out of the 6,000 or so yeast genes.

Of course this strategy won’t work with the 1000 or so yeast genes that are essential—the yeast dies when these are knocked out. For these genes, the researchers used temperature sensitive mutations under conditions where the gene is tweaked but not dead.

What they were looking for were combinations that did either better or worse than you might predict based upon each individual deletion on its own. And they found plenty.

They generated 23 million of these double knockout yeast (Yes, you read that right!) and found ~550,000 negative interactions and ~350,000 positive interactions—a huge amount of data about the inner workings of a cell. The next step was to take all of this data and try to make sense of it.

They did this by clustering the genes together based on who was interacting with whom. If deleting gene A with either gene B or gene C had a larger negative effect than expected, then these three genes were clustered together. And if deleting gene A had an effect on gene D, but deleting gene B or C did not have an effect on D, then A was involved in both a cluster with B and C as well as a separate cluster with D. Now repeat this over and over thousands and thousands of times.

When they did this, many things about living cells quickly came into focus. I only have the space to touch on a few.

First off, while negative interactions usually happened between genes in the same biological processes or in the same subcellular compartments, the same was not true for the positive interactions. These were much more spread out and often genes like chaperones.

They also found about 1000 genes that weren’t involved with many other genes at all. These are most likely genes that would show more interactions under different growth conditions. For example, deleting a gene critical for utilizing galactose might not show a lot of interactions if a cell is grown in glucose. Repeating this experiment under different conditions will probably uncover where these genes fit in.

This is just a taste of what they found and of what is to come in the next few years. Here is a bit more:

Finding Out What a Gene Does

You’d think having a genome sequence for over 20 years might mean you’d pretty much know what all those genes are doing, but you’d be wrong.

There are still plenty of genes in yeast with unknown function. This kind of study can help uncover what a gene might be doing by seeing which genes it clusters together with. If gene X is clustered with known DNA repair genes, it is probably involved in something to do with DNA.

One such gene they looked at in this study was YJR141W. Not a lot was known about this gene other than it was essential.

What they found was that this gene clustered with genes in the cleavage polyadenylation factor (CPF) and cleavage factor 1A (CF1A) protein complexes suggesting this gene might have a role in mRNA 3’-end processing and polyadenylation. After showing experimentally that the protein from this gene physically interacted with CPF complex members Mpe1p and Ysh1p and that the temperature sensitive mutant had trouble processing mRNA in vitro, the researchers felt confident enough to give this gene a name based on its function. They named it IPA1 (Important for cleavage and PolyAdenylation 1).

They did this with a second gene as well (YPR153W was rechristened MAY24 because of its interactions with Mtc2p and Mtc4p). Stay tuned for lots more findings like this one in the near future.

Helping Find New Cancer Treatments

This study also has important implications for human health in lots of different ways. For example, it might help scientists find better cancer treatments.

Cancer happens when mutations in key genes cause cells to grow uncontrollably or to refuse to die. In an ideal world we’d treat these cancers by going after these genes directly. Unfortunately many of them are not particularly great targets to go after with pharmaceuticals.

This is where synthetic lethals can help. A synthetic lethal is when mutations in two separate genes do not kill a cell on their own, but they do when a single cell has both mutant genes.

The idea then is to find a pair of mutant genes that kills a cell with one of the genes being involved in cancer and the second one some other gene. You can go after the second gene and by affecting its activity kill the cancer cells that have the other gene in this pair mutated.

And of course noncancerous cells are fine with you targeting that second gene—mutating only that gene isn’t lethal. Chemotherapy with many fewer side effects!

If you think this sounds like something this study might uncover, you’d be right. This study found around 10,000 or so double knockouts that were lethal which might provide a lot of possibilities for cancer researchers.

Cancer is almost certainly just one of many possible applications to human health. And we will learn so much about the basic biology of a cell from such an exhaustive and detailed analysis of the inner workings of a cell.

This holistic portrait of the genetic interactions of a yeast cell is elegant, beautiful and useful. Not bad for the beast that gives us bread and booze.

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

Categories: Research Spotlight

Tags: double mutant, interactions, network, synthetic lethal

Yeast on the Red Carpet

September 28, 2016

---

Awards season just kicked off with the Emmys a couple of weeks ago. Julia Louis-Dreyfus won her fifth Emmy in a row for her work on Veep, and Tatiana Maslany finally got an Emmy for her incredible work on Orphan Black. (If you haven’t seen Orphan Black, it’s a fascinating look at the ramifications of human cloning and genetic engineering.)

For a lot of people, the awards themselves are secondary to what everyone is wearing and who shows up with whom. You can learn a lot about the evolution of a performer’s career through these subtle (and not so subtle) cues.

Turns out that PLOS just had their own awards. Not quite as glamorous as the Emmys or the Oscars but almost certainly more significant.

The PLOS Genetics Research Prize for 2016 went to a fascinating paper from December of 2015 that uses yeast to explore evolution and provide ways to get at the underlying mechanisms of polygenic inheritance. And it does this by studying the subtle cues of what happens when two different yeast species show up on a yeast plate together (and mate).

The study uses a technique called a sign test. This is a way to find a set of genes that have been up- or down-regulated in response to some sort of selection pressure.

The first step is to mate two different strains or species. In this case Naranjo, Smith and coworkers used two different yeasts – Saccharomyces cerevisiae and Saccharomyces paradoxus.

The next step is to take the result of this mating, the F1 hybrid, and to compare the expression from each species’ alleles to the other’s. In other words, how does the expression of gene A compare in the two species? And gene B? And so on, through all of the genes.

What they are looking for in this allele specific expression (ASE) is a set of genes in the same pathway that are all affected in the same way. In this case they found a set of eleven genes linked to resistance to the toxin citrinin that was upregulated in S. paradoxus, but not in S. cerevisiae, in the absence of citrinin. This suggested that there was some sort of evolutionary pressure on S. paradoxus to become resistant to citrinin.

An obvious prediction from this is that their strain of S. paradoxus, CBS432, is more resistant to citrinin than is their strain of S. cerevisiae, S288C. They tested this and their S. paradoxus strain did indeed do better than S288C when citrinin was around.

They next did RNA-seq on the F1 hybrid yeast in the presence and absence of citrinin to find the up-regulated genes responsible for the ability of S. paradoxus to better tolerate citrinin. They ultimately settled on five genes that were both more highly expressed in the absence of citrinin, and more strongly induced by citrinin.

To figure out which of these genes is critical for resistance, they next deleted each gene individually and tested each deletion strain for its ability to grow in citrinin. Four of the five genes – GPX2, FRM2, RTA1, and CIS1 – made it through this test.

They next checked to see if making more product from all of these genes at once increased the strain’s resistance to citrinin. To pull this off they turned to everyone’s favorite genetic tool, CRISPR/Cas9.

Unless you’ve been hiding under a rock, you already know that CRISPR/Cas9 uses a guide RNA to get the protein Cas9 to the specific spot in the genome you want to edit. But in this case they aren’t editing a gene. Instead they are activating genes by using a version of Cas9 with two important changes: it can’t cut DNA anymore and it has a transcription activation domain added to it.

The idea is to activate all four genes at once by providing the yeast with guide RNAs that can lead this Cas9 to each of the four genes. What a powerful and simple way to easily activate all four genes at once.

They found that this overexpression strain was able to better tolerate citrinin but that it came at a cost – the strain grew more poorly in the absence of citrinin.

They next set out to see if the mutations that distinguish S. paradoxus from S. cerevisiae were in the promoters of these four genes. First, they replaced the S. cerevisiae promoter with the S. paradoxus promoter for each gene in the citrinin-sensitive S. cerevisiae strain. This created four new strains, each with one of the promoters swapped.

They found that in all cases the S. paradoxus promoter led to increased gene activity. Expression from these genes increased by anywhere from 1.6-6.9 fold.

Their final experiment was a competition between the original S. cerevisiae parent and the four strains in which the native S. cerevisiae promoter had been swapped out with the S. paradoxus one. They found that except for the strain overexpressing RTA1, these strains did better than the original S. cerevisiae strain in the presence of citrinin, but worse in its absence. Each of the three strains alone did not provide as much advantage as all three together did.

This is pretty powerful stuff! They used the sign test and CRISPR/Cas9 to nail down the three differences between S. paradoxus and S. cerevisiae that help to explain the polygenic trait of citrinin resistance.

And this isn’t just some cool yeast experiment either (although it is definitely that – #APOYG!). Sign tests may provide a new way for all those geneticists dutifully doing genome-wide association studies (GWAS) to find the set of genes responsible for polygenic traits that have so far eluded them. This is the kind of work that definitely deserves an award!

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

Categories: Research Spotlight

Tags: adaptation, allele-specific expression, cis-regulation, natural selection

More Than Just a Brick

September 07, 2016

---

Cell phones have evolved in amazing ways. What started as a “dumbphone” that only let you make calls and, somewhat awkwardly, text, has now become a smartphone – a marvelous mini-computer.

What this means is that even if a smartphone’s SIM card goes bad so that you can’t make a call or send a text, you still have something that can do a lot. In fact, since it can still connect to Wi-Fi, you can even text or call! This is obviously different than a “dumbphone” which, if it can’t be used as a phone, is pretty much only useful as a hurled weapon. (Which is just what Dr. Drakken threatened Ron Stoppable with at 8:27 of this video.)

In a new study out in GENETICS, Thurtle-Schmidt and coworkers show that the histone deacetylase Sir2p is more like a smartphone than a dumb one. Even when you knock out its ability to deacetylate histones, Sir2p can, in the right background, still silence the genes it is supposed to.

Sir2p is the founding member (#APOYG!) of the important sirtuin enzyme family. It silences genes by first deacetylating acetylated lysines on histone tails (H3 and H4 specifically). This then allows the Sir-protein complex (which includes Sir2p, Sir3p, and Sir4p) to bind the nucleosome Sir2p just deacetylated. Now the Sir-protein complex deacetylates nearby histone tails and so on until the Sir-protein complex has spread across a gene, silencing it.

Obviously the ability of Sir2p to deacetylate is important in this scenario! But these researchers found that like a smartphone without a SIM card, Sir2p can sometimes do its job even without its deacetylase powers.

But instead of going around a carrier and using Wi-Fi, Sir2p needs for a second histone deacetylase, Rpd3p, to be gone. Without Rpd3p, Sir2p can now silence genes. Not as well as it could before, but some.

To find this out the researchers set up a suppressor screen. They used a reporter that replaced the a1 open reading frame (ORF) at HMR with the URA3 gene. The a1 ORF is normally silenced by Sir2p. Basically, if this gene is silenced, the yeast can grow in the presence of 5-fluoroorotic acid (5-FOA).

Next they added a mutant Sir2p that lacked its catalytic ability to this reporter strain. Finally they mutagenized this strain and looked for mutants that could again silence genes. They got 1500 5-FOA resistant mutants.

Of course many of these may have been the result of mutations in the URA3 gene. They did a secondary screen that used mating as a way to rule out this possibility. In the end they had four mutants.

One of these four was a mutation that had been found in previous screens – SUM1-1. This mutant appears to bypass the need for any of the SIR genes by setting up a different kind of silenced chromatin.

The other gene that came out of the screen was RPD3. They found three different mutations in this deacetylase that all partly restored Sir2p’s ability to silence genes and found that deleting the gene had the same effect. Follow-up work showed that this effect was indeed Sir2p-dependent. Now they had to figure out how eliminating a second deacetylase frees this mutant Sir2p to do its job.

In some ways it isn’t surprising to get RPD3 out of a screen like this. It seems to be important in keeping the Sir-protein complex from spreading too far (who wants the whole chromosome shut down?) and deleting it affects various silenced genes.

Rpd3p is found in two different complexes imaginatively named large (Rpd3L) and small (Rpd3S). When the authors deleted genes specific to either complex, their sir2 mutant did not regain its ability to silence genes. Only when they deleted a gene involved in both complexes, SIN3, were they able to mimic the effects of deleting RPD3 (“phenocopied the rpd3Δ”).

One possible idea is that since Rpd3p keeps the Sir-protein complex from spreading, its deletion might allow for increased spreading even in the absence of Sir2p’s histone deacetylase activity. This is what they found.

Using Chromatin ImmunoPrecipitation (ChIP) against Sir4p, one of the proteins in the Sir-protein complex, the authors repeated the result that in the absence of Sir2p’s histone deacetylase activity, there is less Sir4p at silenced regions. When they looked at the same strain deleted for RPD3, they found an increase of Sir4p at silenced genes. Not to the levels seen in the wild type strain, but enough to probably explain the partial silencing seen in the strain.

Makes sense so far but it isn’t the whole story. Nicotinamide (NAM) is competitive inhibitor of sirtuins like Sir2p. As such, we might predict that it should have no effect on the silencing of a gene by a catalytically inert Sir2p. We would be wrong.

Turns out NAM does affect silencing in this strain which suggests that some other sirtuin might be playing a role. There are four homologs of SIR2: HST1, HST2, HST3, and HST4. A bit of work including creating a triple mutant strain deleted for RPD3 and HST3, and containing the mutant Sir2p, showed that Hst3p is involved in this silencing.

Whew, that was a lot! So mutating away the deacetylase activity of Sir2 unsilences genes. And deleting RPD-3 from this strain restores some of that silencing. And the restored silencing in this strain is at least partly dependent on Hst3p.

So there you have it. Like a cell phone without a SIM card using Wi-Fi, Sir2p can still do its job if Rpd3p isn’t around to interfere. As long as Hst3p, like turning the phone’s Wi-Fi on, is there to help.

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

Categories: Research Spotlight

Tags: H4K16, heterochromatin, HST3, RPD3, silencing, SIR2, sirtuin

When Running Out of Steam is a Good Thing

August 24, 2016

---

Usain Bolt sprinting is a thing of beauty. It is just amazing how he can kick in the afterburners at the end of a race and just dominate the thing. I am sure Justin Gatlin of the U.S. would love for Bolt to lose this extra burst of speed so Gatlin could beat him at the Olympics.

Turns out that transcription elongation has an afterburner a bit like Bolt’s too. It goes by the name SPT4 in yeast and SUPT4H1 in you and me. The protein from this gene is needed to push through long transcripts.

A new study in Science by Kramer and coworkers suggests that like Gatlin, some people would like to see their cells lose the burst of speed that SUPT4H1 gives their polymerases. But instead of helping these folks win a race, this loss might help them deal with their amyotrophic lateral sclerosis (ALS) or frontotemporal dementia (FTD).

ALS is a progressive neurodegenerative disease that is always fatal. It was first made famous by Lou Gehrig and later with the bucket challenge. After Alzheimer’s, FTD is the second most common form of dementia.

Back in 2011 two groups found that a significant number of cases of FTD and ALS were associated with a gene called C9orf72 (chromosome 9 open reading frame 72). These people had hundreds or even thousands of copies of the hexanucleotide repeat GGGGCC in the first intron of their gene instead of the 30 or so that is more typical.

Later studies showed that these repeats caused two very specific problems in cells. First, the RNA (and antisense RNA) from this allele tended to build up in small bundles called foci. Some researchers think that these foci trap some of the important RNA binding proteins that the cell needs.

The second phenotype is a strange one. These RNAs get translated by a process called repeat-associated non-ATG, or RAN, which causes a buildup of dipeptide repeat proteins. Apparently there is something about the secondary structure of the RNA that allows it to get translated without a typical AUG start codon.

The idea is that these foci and weird dipeptide proteins are at least part of the reason why these folks have their ALS symptoms. Ideally you’d want to get at all three issues (the sense and antisense RNA-laden foci, and those newly translated proteins) with a single approach.

Kramer and coworkers reasoned that they might get such a result if they could get the cell to make a whole lot less of SUPT4H1 (or Spt4p in yeast). They reasoned correctly.

Previous research had shown that its deletion didn’t affect too many genes except for those involved in diseases like Huntington’s – those with long CAG repeats. Perhaps, then, deleting it might also just affect the copies of the C9orf72 gene with those hexanucleotide repeats without affecting too many other genes.

When they forced yeast, nematode, fruit fly and human cells to make less Spt4p or SUPT4H1, the number of RNA foci went down or even disappeared in all of these different cells. There was also much less of those dipeptide repeat proteins lurking about the cell as well.

They first set out to do some experiments in everyone’s favorite workhorse, Saccharomyces cerevisiae. They found that expressing either the sense or antisense RNA with the 66 hexanucleotide repeats caused both the RNA foci and the dipeptide repeat proteins seen in the cells of ALS patients to form in yeast too. Neither happened with the sense or antisense 2 repeat constructs.

Next they showed that deleting SPT4 greatly reduced the level of 66 repeat RNA but had little effect on the 2 repeat RNA. These researchers also saw no RNA foci and much less dipeptide repeat proteins in the deletion strain expressing the 66 hexanucleotide repeats. All without much affecting any other genes.

This yeast work suggests that targeting SUPT4H1 might reduce the effects of the ALS version of the C9orf72 gene without affecting the more typical version. Now Kramer and coworkers were ready to see what happens in bigger beasts.

When they expressed the 66 repeat in Caenorhabditis elegans neurons, these nematodes lived for a shorter time and their neurons had RNA foci and the dipeptide repeat proteins. Expressing human SUPT4H1 in these worms’ neurons worsened their condition while feeding them RNAi against nematode SPT4 helped.

The RNAi let these worms live longer and it decreased the number of RNAi foci and the amount of dipeptide repeat proteins. They saw similar results with a Drosophila system.

Finally they moved to the main stage—human cells from ALS patients who had the C9orf72 protein with too many hexanucleotide repeats. RNAi against either SUPT4H1 or its partner in crime, SUPT5H, reduced the number of RNA foci and reduced the amount of dipeptide repeat proteins with no “overt toxicity.” RNA-seq showed that only a small subset of genes was affected with the RNAi treatment.

So it looks like targeting SUPT4H1 may be a good strategy for dealing with ALS if the RNA foci and dipeptide repeat proteins are a big part of the problem. This is a big if.

But if it all does work out, we can thank yeast yet again (#APOYG!) for showing us the way to a new treatment for a devastating disease. Of course, though, yeast can’t do everything. It is unlikely to show sprinters the best way to beat Usain Bolt in a race!

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

Categories: Research Spotlight, Yeast and Human Disease

Next