New & Noteworthy

Mass Production in Yeast

May 11, 2017

Workers on first moving Ford assembly line

This approach changed everything when it came to manufacturing in factories. Perhaps the ideas in this new study will change things for manufacturing in cells.Image from Wikimedia Commons

After Henry Ford invented the moving assembly line, manufacturing was never the same. With it, his workers were able to push out a car every 2 ½ hours instead of the 12 it used to take. (Another website said it was reduced to 90 min!)  The technology quickly spread to every factory.

Now of course, an assembly line is only as fast as its slowest worker. If someone is taking extra time to bolt down that part, then everyone downstream will have to go slower too, resulting in fewer cars being made.

But, you also can’t go too fast. If you do, someone can get injured, shutting down the whole line. (Or the worker has to eat all the candy to keep up, like Lucy.)

And you want to make sure things happen in the right part of the factory. You don’t want the paint sprayer out in the open, poisoning factory workers. So, that needs to happen in a special room.

This also applies to cell processes where something complicated is built, step by enzymatic step. All the enzymes need to be at the right levels and in the right place to maximize the productivity of the whole process.

This all becomes very obvious when you try to move an enzymatic process from one beast to another. What worked perfectly before, now barely works at all.

One way to fix this is through trial and error, trying to optimize one part of the process at a time. This is incredibly time consuming!

In a new study out in Nature Communications, Awan and coworkers show one way to tweak all of the enzymatic steps involved in making penicillin at the same time in the yeast Saccharomyces cerevisiae. While this isn’t that useful for making this antibiotic (there are better ways available right now), it does show how researchers can apply the same techniques to perhaps identify and produce new antibiotics. And, it can also be applied to other unrelated enzymatic processes.

Penicillin is made in a five-step process in filamentous fungi. In the first part of the process, two enzymes create a tripeptide precursor using alpha-aminoadipic acid, cysteine, and valine, called ACV. This part of the process had been previously recapitulated in yeast, so Awan and coworkers used this as a starting point for their penicillin producing strain.

The next part of the process uses the last three enzymes and takes place in peroxisomes in filamentous fungi. These authors found that they only got penicillin when these enzymes were tagged to be sent to the peroxisome in yeast. Like a special room for spray painting cars, these enzymes need to be in the right place to make penicillin.

But this was by no stretch of the imagination an efficient penicillin-making machine. The thing managed only 90 pg/ml in the media. As Ursula from Little Mermaid might say, “Pathetic.”

Ursula from The Little Mermaid

From Tumblr

Still, it is a starting point. The next step is to get the yeast to crank out more penicillin. To do this, they used a combinatorial approach to optimize the process all at once. Well, not really all at once.

First, they set out to optimize how much of the precursor ACV the yeast made. Then, they optimized how much ACV was converted to penicillin.

Awan and coworkers created a library of low copy plasmids that had the genes for the first two enzymes, pcbAB and npgA, under the control of different pairs of promoters. One plasmid, with the pTDH3 promoter driving pcbAB expression, and the pPGK1 promoter driving npgA expression, outperformed all of the others. As measured by Liquid Chromatography-Mass Spectrometry (LCMS), the yield of ACV increased from 20 to ~280 ng/ml.

Next, the authors used this new strain as a starting point for optimizing the activity of the final three enzymes using a similar approach. They used a “…one-pot combinatorial DNA assembly using Golden Gate cloning…” to make a library of around 1000 high copy plasmids where each gene was under the control of one of ten different promoters of varying strength. Using LCMS they found strains that could make 3 ng/ml of penicillin, a significant improvement over the original 90 pg/ml.  

The 3 ng/ml of penicillin in the media should be high enough concentration to inhibit the growth of bacteria like Streptococcus pyogenes. So, they confirmed that their penicillin was active using growth inhibition assays.

After sequencing the plasmids, the authors saw that the best strains tended to have strong constitutive promoters driving one of the genes, pclA, and medium strength promoters driving another one of the genes, pcbC. They used a minION DNA sequencer to confirm that this was not the result of a biased library.

As a final step, they set out to optimize penicillin production and to increase the throughput of their assay. They created another library that swapped six different promoters that varied in strength from medium to high for each of the last three genes in the pathway, pclA, pcbC and penDE. Instead of using LCMS to screen for penicillin production, they used a 96 well plate-based assay that looked for inhibition of Streptococcus pyogenes growth for their 120 new strains.

They selected 12 of the highest performing strains and confirmed by LCMS that they made lots of penicillin. Five of the strains made more than 5 ng/ml, a more than 50-fold increase over their original strain.

As this concentration is still three orders of magnitude below what other organisms can currently do, this new yeast strain will not go into penicillin production any time soon. But this study gives us a way to quickly optimize antibiotic production using growth inhibition assays instead of the more cumbersome LCMS.

And it isn’t restricted to just antibiotic production. Similar combinatorial approaches can be used for almost any stepwise enzymatic process. Researchers can create libraries of plasmids where levels of enzyme vary and use the long reads of minION DNA sequencing technology to confirm that their results are not skewed by a biased library.

As usual, this is only possible as a simple, easy procedure because of the awesome power of yeast genetics (#APOYG). Researchers have the tools to use yeast to find new antibiotics and to manufacture them at a high rate, like inventing the car and the assembly line at the same time.

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

Meiotic Fail Safes

May 4, 2017

minuteman in silo

Launching into meiosis too soon is as dangerous (for a cell) as launching a nuclear missile. Luckily both have protocols to make sure each can only happen in the right circumstances. (Hopefully never for the nuclear missile.) Image from Wikimedia Commons.

If the movie WarGames is anything to go on, the US government does not make it easy to launch a nuclear missile. Two soldiers have to do many things simultaneously and in the right order before that missile can take flight.

This makes perfect sense as you do not want to launch a nuclear attack unless you absolutely have to. The continued existence of the human race depends on these fail safes being in place and working. The same goes for a cell that is heading into meiosis.

Meiotic fail safes are in place to ensure the survival of a cell during the dangerous, early part of meiosis, when there are lots of double-strand breaks in the DNA. These all need to be resolved before a cell is allowed to continue through meiosis to create gametes. If the cell moves on while the breaks are still there, gamete production will fail and the cells will die.

While the exact sequence of events needed to launch World War III is known (at least by a few people), the exact details of getting a cell safely through meiosis are a bit murkier. With the help of good old Saccharomyces cerevisiae, we have the broad outlines, but are still investigating the finer points.

A new study by Prugar and coworkers in GENETICS has helped clear up a bit of the murk in yeast. They have uncovered a connection between the meiosis-specific kinase Mek1p and the transcription factor Ndt80p that may explain how a cell “knows” when it is safe enough to emerge from prophase and keep progressing through Meiosis I.

Mek1p is known to be active when there are lots of these double-strand breaks around and to lose activity as these breaks are resolved. Ndt80p, on the other hand, is inactive when there are lots of these breaks and active when they are resolved. So it makes sense that their activities might be related to each other.

In this study, the authors show that once Mek1p activity falls below a certain level, it can no longer keep tamping down Ndt80p activity. Once unleashed, Ndt80p can go on to activate many genes, including the polo-like kinase CDC5 and the cyclin CLB1. This round of gene activation allows the cell to progress through meiosis.

The key to teasing this out was a set of experiments where Prugar and coworkers were able to control the activities of Mek1p and Ndt80p independent of the cell’s DNA state. It is like circumventing the set of protocols to get those missiles launched.

To independently control Ndt80p activity, they used a form of the protein that requires estradiol to be active. And they controlled the activity of Mek1p by using a mutant, mek1-as, that is sensitive to the purine analogue 1-NA-PP1. In the presence of this inhibitor, Mek1p stops working.

They looked at the targets of these two proteins to infer activity. For example, they determined if Ndt80p was active by looking for the presence of CDC5. And to see if Mek1p was active, they looked for phosphorylated Hed1p.

In the first experiment, they showed that in the absence of both estradiol and 1-NA-PP1, Hed1p stayed phosphorylated. Mek1p was constitutively active in the absence of Ndt80p even as double-strand breaks were resolved.  (They used phosphorylated Hop1p as an indirect measure of double-strand breaks.)

War Games (1983)

In the end, as yeast relies on MEK1 to prevent a meiotic disaster, humans, not computers, kept the world safe in the movie WarGames. Image from flickr

When Ndt80p was activated through the addition of estradiol, CDC5 was turned on and Hed1p lost its phosphorylation. This loss of Mek1p activity did not happen as quickly as with 1-NA-PP1.

These results suggest a negative interaction between Mek1p and Ndt80p. When Ndt80p is active, Mek1p is not and when Mek1p is active, Ndt80p is not. The resolution of the DNA breaks as indicated by the loss of phosphorylated Hop1p was not sufficient to shut off Mek1p activity. It took the activation of Ndt80p for this to happen.

Well, Ndt80p did not directly cause Mek1p’s inhibition. A second set of experiments suggested that a target of Ndt80p, CDC5, was responsible.

For this they made Cdc5p activity independent of Ndt80p induction by making it dependent on estradiol, similar to what they did with Ndt80p. Using a strain deleted for NDT80, they found that inducing Cdc5p activity was enough to eliminate Mek1p activity.

I don’t have the space to go into the rest of the experiments in this study, but I urge you to read it if you want to learn about more of the details of the cell’s protocol for know when it is OK to progress through meiosis.

With the help of the awesome power of yeast genetics (#APOYG), Prugar and coworkers have added to our knowledge about the safeguards that are in place to keep a cell from launching into meiosis too soon. Turns out they are even more complicated than the ones that prevent accidental thermonuclear war.

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

Knock out YME1, Luke

April 20, 2017

Death Star Explosion & Millenium Falcon

Instead of an exhaust port, one of a mitochondrion’s fatal flaws may be the YME1 gene. Image from Manoel Lemos, flickr.

In the original Star Wars, Luke destroys the Death Star with a precise strike of proton torpedoes down a small thermal exhaust port. For him it was as easy as bullseyeing “womp rats in my T-16 back home.”

Luke and the rest of the Rebel Alliance learned of this engineered fatal flaw from Jyn and her friends in the prequel Rogue One. With this information the Rebel Alliance was able to keep the rebellion alive long enough to finally bring down the Empire by the end of Return of the Jedi.

It turns out that our friend Saccharomyces cerevisiae has taught us about a fatal flaw in mitochondria. Like proton torpedoes in an exhaust port, when the gene YME1 is inactivated, mitochondria become unstable. But instead of bits of Death Star raining down on nearby planets, mitochondrial DNA (mtDNA) is released into the cytoplasm.

Sometimes this mtDNA can end up in the nucleus and find its way into nuclear DNA. And if the conclusions of a new study in Genome Medicine by Srinivasainagendra and coworkers turns out to be right, this numtogenesis (as the authors call this process) can have profound consequences when it happens in people. Their data suggests that it might lead to cancer or possibly cause cancers to spread.

These researchers searched through whole genomes of colon adenocarcinoma patients and found that these cancer cells had 4.2-fold more mtDNA insertions compared to noncancerous cells from the same patient. They also found that patients with more of these insertions tended to do worse (although the sample sizes were too small to say this definitively).

Why is this happening in the cancer cells? What has caused the mitochondria to give up their DNA?

Srinivasainagendra and coworkers turned to previous work that had been done on the YME1 gene in the yeast S. cerevisiae to find one possible reason. YME1 had been shown to be an important suppressor mtDNA migration to the nucleus. Perhaps this was true in mammalian cells as well.

A search through the genomes of cancers suggested that this seemed to be the case. Around 16% of the colorectal tumors they looked at had a mutated YME1L1 gene, the human homologue of YME1. And mutated YME1L1 genes were found in other tumors as well.

If only destroying gene function was as fun.

They used CRISPR/Cas9 to directly test the effects of knocking out YME1L1 in the breast cancer cell line MCF-7. The knock out cells had a 4-fold increase in the amount mtDNA in the nuclear fraction compared to cells that still had working YME1L1.

As a final experiment, they used a yeast strain, yme1-1, in which YME1 function was inactivated, to show that the human homologue, YME1L1, could suppress the migration of mtDNA to the nucleus.

This yme1-1 strain has a TRP1 gene encoded in the mtDNA instead of the nucleus. Since the gene cannot be read by the mitochondrial transcription machinery, the only way this yeast strain can survive in the absence of tryptophan is if the TRP1 gene moves from the mitochondrion to the nucleus. 

In their experiment, with vector alone, they got around 1000 TRP+ colonies with yme1-1. When they added back yeast YME1, this number dropped to less than 50 compared to the 100 or so they got when they added the human homologue, YME1L1. So YME1L1 can suppress mtDNA migration to the nucleus.  

Given that YME1L1 was mutated in just a subset of the cancers, it is unlikely that it is the only player in the mtDNA these authors found in the nuclei of cancer cells. But it does look like it is one way this can happen.

And it would have been very hard to fish out the human gene without the critical work that had been done in yeast previously. Yeast shows us the way again. #APOYG

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

Exploring the Global Yeast Genetic Interaction Network

April 17, 2017

Global yeast genetic interaction profile similarity network. Image from

With the construction of a global genetic interaction network in S. cerevisiae, it’s not hard to see why yeast genetic interactions remain a treasure trove for biological discovery. When combined with tools for visualization and analysis, these data can be used to draw powerful functional maps of the cell and infer potential functions for uncharacterized genes.

In a recent paper published in G3: Genes|Genomes|Genetics, Usaj et al. describe a web-based resource for exploring the global yeast genetic interaction network: is an online database and visualization tool for quantitative yeast genetic interaction data. It provides an interactive version of the global yeast genetic interaction similarity network described by Costanzo et al., enabling users to scroll through and zoom in on different clusters of functionally related genes within the network. Users can search for specific genes or alleles, extract and re-organize sub-networks for genes of interest, functionally annotate genetic interactions, and more. Further, if more details about a gene are needed, users can even double-click on the gene to be taken to its respective locus summary page at SGD!

For more information about this resource, see or access the publication at

Next Page »