New & Noteworthy
September 25, 2014
One of the great joys of teaching can be found in the questions that students ask. Because they are unconstrained by previous knowledge, they can think outside of the box and ask questions that force the teacher to see a problem in a new light. Their unbiased questions often uncover aspects of a problem that a teacher didn’t think to look for or even consider.
The scientific enterprise can be very similar. Sometimes an unbiased search of a process will uncover hidden parts scientists were completely unaware of.
This is exactly what happened in a new study in Science by Foresti and coworkers. Using an unbiased proteomics approach they found a previously hidden part of the endoplasmic reticulum-associated degradation (ERAD) pathway in the inner nuclear membrane (INM) of the yeast Saccharomyces cerevisiae. No one knew it existed before and, frankly, no one even knew to look! By thinking outside of the box, these authors found that a novel protein complex in the INM targets certain proteins for degradation – both misfolded proteins, and some correctly folded proteins whose functions are no longer needed.
Scientists already knew that the ERAD pathway uses different protein complexes to target proteins for degradation, depending on where those proteins are located. For example, misfolded cytoplasmic proteins are targeted by a complex containing Doa10 (also known as Ssm4), while those in the membrane are targeted by the Hrd1 complex. However, degradation of both sets of proteins requires ubiquitination by the shared subunit Ubc7. In addition to targeting misfolded proteins, both of these complexes also target certain functional proteins in response to specific conditions.
In the first set of experiments, Foresti and coworkers looked at the proteomes of strains deleted individually for Doa10, Hrd1, or Ubc7. To their surprise, they found a set of proteins, including Erg11 and Nsg1, that are unaffected by the deletion of either Doa10 or Hrd1, but whose levels are increased in strains deleted for Ubc7. This suggested there is a branch of the ERAD pathway that involves Ubc7 but is independent of Doa10 and Hrd1. The authors set out to find this undiscovered third branch lurking somewhere within the yeast.
Some possible candidates for being part of the ERAD pathway were two paralog proteins Asi1 and Asi3, and their associated protein Asi2. Based on their sequences, Asi1 and Asi3 are putative ubiquitin-protein ligases like Doa10 and Hrd1. Interestingly, all three Asi proteins localize to the inner nuclear membrane, which connects to the ER at nuclear pores.
When Foresti and coworkers deleted any one of the three Asi proteins, degradation of Erg 11 and Nsg1, both involved in sterol synthesis, was blocked. However deletion of Asi1, Asi2, or Asi3 didn’t affect all proteins involved in sterol biosynthesis, since Erg1 was unaffected. Biochemical experiments confirmed that Erg11 binds to a complex composed of these three Asi proteins.
Since the ERAD pathway is important for degradation of misfolded proteins, the authors set out next to determine whether the Asi complex plays a role in this process as well. That would be a somewhat surprising finding, since misfolded proteins aren’t generally found near the INM. But through a complicated set of experiments summarized below, Foresti and coworkers confirmed that the Asi complex does also have a role in this process.
They first tested several proteins that are known ERAD substrates, but mutations in the ASI genes had no measurable effect on them. Because some misfolded proteins are targeted by more than one ERAD complex, the authors next looked to see whether the Asi pathway contributed to either the Hrd1 or the Doa10 pathways. Testing the accumulation of several substrates in strains with different combinations of asi, hrd1, and doa10 mutations, they found that one mutant protein that misfolds, Sec61-2, had high steady state levels in a hrd1 knockout, but even higher steady state levels in a double knockout of hrd1 and asi1 or hrd1 and asi3. So both the Asi and Hrd1 pathways appeared to work on this misfolded protein.
The researchers hypothesized that the Asi branch may target misfolded proteins for degradation as they travel through the inner nuclear membrane on the way to the ER. To test this idea, they compared the steady state levels and localizations of two differently mutated versions of the Sec61 protein – one that localized to the inner nuclear membrane and one that did not, in both wild-type cells and a variety of deletion strains.
The bottom line from these experiments was that the mutant protein that was located at the inner nuclear membrane was more dependent on the Asi complex than the mutant that wasn’t. Not only that, but the mutant Sec61 protein that was directed to the inner nuclear membrane changed its localization to the nuclear envelope in an asi1 deletion strain. Both of these results are consistent with a role for the Asi complex in targeting proteins for degradation while they are in the inner nuclear membrane.
The final set of experiments confirmed the importance of the Asi complex in ER protein quality control. Yeast responds to the presence of too many misfolded proteins in the ER with a signaling pathway called the unfolded protein response (UPR). Strains in which this pathway is compromised, for instance by deleting IRE1, need a functional ERAD to thrive. The authors found that deleting HRD1, IRE1, and ASI1 had a much more severe effect on viability than did just deleting HRD1 and IRE1. This supports the idea that the Asi complex is important in ER protein quality control.
Foresti and coworkers have thus uncovered a previously undiscovered branch of the ERAD pathway in yeast by doing a broad, unbiased proteomics study. The key proteins they identified, Asi1, Asi2, and Asi3, were originally discovered for their genetic effects on the transcriptional repression of amino acid permeases (hence their name, Amino acid Signaling Independent). Their detailed biochemical functions were unknown until now.
A lesson here is that just because a process looks like it is pretty well locked down, this doesn’t mean that there aren’t hidden parts yet to be discovered. And just because a gene is implicated in one process, don’t assume it isn’t also involved in other processes as well. Looking from a different angle can allow you to see things you had missed before.
by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics
September 16, 2014
Different people can respond to alcohol differently because of their genes. For example, many Asians flush or even become ill from alcohol because of a mutation in their ALDH2 gene. (This is not just a minor annoyance—these unpleasant side effects come with a significant increase in esophageal cancer rates.)
This is a simple example where one gene has a significant effect. But of course, not everything to do with people and alcohol is so simple at the genetic level!
For example, some people can drink you under the table while others are lightweights. Some of this has to do with their lifestyle (how often they drink, how much they weigh, etc.), but a lot undoubtedly has to do with the variations they carry in multiple genes.
Well, it turns out this is also the case for yeast (our friend in the alcohol business). A new paper in GENETICS by Lewis and coworkers confirms that different strains of the yeast Saccharomyces cerevisiae tolerate high levels of alcohol differently because of their specific genetics. And at first the response seems…shall we say…incapacitatingly complex.
The results are interesting in that they help parse out how yeast responds to ethanol, but the implications are more far-reaching than that. This analysis helps to form the framework for investigating how natural variation in gene expression can affect the traits of individuals and their responses to certain environmental stimuli.
Lewis and coworkers used three strains in their study: a lab strain that came from everyone’s favorite workhorse S288c, the strain M22 from a vineyard, and the oak soil strain YPS163. They had previously shown that thousands of genes in each strain responded differently to 5% ethanol. In this study they set out to find out what was behind these differences.
First off they wanted to confirm their previous results. Using six biological replicates, they found that 3,287 genes out of a total of 6,532 were affected in at least one strain when treated for 30 minutes with 5% ethanol. This is over half the genes in the genome!
To try to get a handle on what is causing such widespread effects, they next performed eQTL mapping in 45 F2 crosses of S288c X M22 and S288c X YPS163 (these particular matings were chosen because much of the variation they saw was in S288c). This analysis was designed to try to find “hotspots” in the genome: loci that affected many different transcripts, or that could account for all the variation they saw.
When they did this analysis they found 37 unique hotspots. Each hotspot represented 20-1,200 different transcripts, with a median of 37 transcripts. Of these, 15 were seen in both crosses, 12 in just the S288c X M22 and 10 in the S288c X YPS163 matings. No silver bullet, but 37 is certainly easier to work with than 3,287!
Lewis and coworkers next set out to find the key gene(s) in the hotspots responsible for affecting multiple transcripts in the presence of ethanol. Some were easy to find. For example, HAP1 in S288c and CYS4 in M22 X S288c. But the big prize in this analysis probably goes to MKT1, which affected over 1,000 transcripts in this study.
Now MKT1 is not one of the usual suspects, in that it is not a transcription factor. However, MKT1 has been implicated in lots of observed differences between strains, including alcohol tolerance in one Brazilian overproduction strain. Given this, the authors set out to explore whether there were any differences in Mkt1p activity in response to ethanol in the different strains.
This analysis revealed that Mkt1p localizes to P-bodies upon ethanol stress in S288C but not YPS163. And this wasn’t some general defect in Mkt1p, since it is known to colocalize with P-bodies in both strains in response to hypo-osmotic stress.
With this discovery, things were starting to make a bit more sense! Since P-bodies are involved in mRNA turnover, it follows that a P-body component might affect so many transcripts. One potential explanation might be that Mkt1p serves as a regulator by translationally silencing specific mRNAs at P-body loci. This would be consistent with its known role in translational regulation of the HO transcript.
This study reveals how difficult it is to get to the bottom of determining exactly how massive differences in gene expression lead to differences in traits. But it also shows that while daunting, it is doable. And perhaps yeast can show us how best to interrogate our own differences in gene expression to help figure out why we are the way we are—not only in terms of whether we dance on the tables or fall to the floor after a few drinks, but in many other respects as well.
by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics
September 15, 2014
Have you ever wondered about the role played by the homolog of a particular yeast gene in other fungal species? SGD’s advanced search tool, YeastMine, can now be used to find homologs of your favorite Saccharomyces cerevisiae genes in the pathogenic yeast, Candida glabrata. There are now 25 species of pathogenic and non-pathogenic fungi in YeastMine, including S. cerevisiae.
The fungal homologs of a given S. cerevisiae gene can be found using the template called “Gene –> Fungal Homologs.” Fungal homology data comes from various sources including FungiDB, the Candida Gene Order Browser (CGOB), the Yeast Gene Order Browser (YGOB), the Candida Genome Database (CGD), the Aspergillus Genome Database (AspGD) and PomBase, and the results link directly to the corresponding homolog gene pages in the relevant databases.
A results table is generated after each query and the identifiers and standard names for the fungal homologs are listed in the table. As with other YeastMine templates, results can be saved as lists for further analysis. You can also create a list of yeast gene names and/or identifiers using the updated Create Lists feature that allows you to specify the organism representing the genes in your list. The query for homologs can then be made against the custom gene list.
All of the new templates that query fungal homolog data can be found on the YeastMine Home page under the “Homology” tab. This template complements the template “Gene → Non-Fungal and S. cerevisiae Homologs” that retrieves homologs of S. cerevisiae genes in humans, rats, mice, worms, flies, mosquitos, and zebrafish.
September 11, 2014
If you think back really hard to your basic molecular biology classes you can probably remember that weird nucleotide pseudouridine (ψ). You probably learned that it is found in lots of tRNAs and rRNAs but never in mRNA. You also may remember that while its function is still a bit unclear, it may have something to do with RNA stability and/or helping aminoacyl transferases interact with tRNAs.
If a new paper in Nature holds up, one of those things we learned is almost certainly wrong. In this study, Carlile and coworkers show pretty convincingly that ψ is also found in mRNA. And not only that, but it may be doing something important there.
The authors used a sensitive high-throughput technique called Pseudo-seq to look for ψ in all the RNA in a yeast cell. The first step in this technique is to treat the RNA with a chemical called CMC.* This chemical reacts with ψ in such a way as to create a block to reverse transcriptase. In other words, reverse transcriptase can only convert RNA into DNA up to the point of a ψ. The next step is to analyze the products and to determine where reverse transcriptase has been halted.
Carlile and coworkers first validated their technique by looking at RNAs known to have ψ’s. They showed that their technique had an estimated false discovery rate of 5% for highly expressed genes and 12.5% for poorly expressed genes. They were now ready to tackle mRNA to see what they could find.
They first looked at the mRNA of the yeast Saccharomyces cerevisiae during post-diauxic growth (after log phase) and found 260 ψ’s in 238 protein coding transcripts. This is 260 more ψ’s than had been found before.
The next step was to try to get a feel for whether or not these changes were important. To do this, they decided to compare pseudouridylation (we promise not to use that word again!) in log phase and post-diauxic growth. They found that 42% of the sites modified after log phase were not modified during log phase. In other words, it looks like the level of mRNA modification is different depending upon the growth rate.
Uracils are modified to ψ by a surprisingly large number of enzymes. One enzyme, Cbf5p, uses snoRNA guide sequences to find the right uracils to modify. Cbf5p may not be that important for converting U’s to ψ’s in mRNA , however, since only 3/260 of the sites identified by the authors appeared to be targeted by this enzyme.
The other nine known enzymes in yeast all have the rather unfortunate acronym “PUS,” for PseudoUridine Synthase. Carlile and coworkers tested the effects of individually deleting eight of these on their newly identified ψ sites in mRNA and found that deleting PUS1 affected the highest number of mRNAs. Interestingly, many of the Pus1p target sites were modified more often during post-diauxic growth than during exponential growth. Deleting the other PUS genes had similar, if smaller, effects.
The authors next confirmed that something similar happens in human cells. Using very strict criteria, they identified 96 ψ’s in 89 human mRNAs and found that some of these were regulated by growth conditions (serum starvation), just as in yeast. So, modification of mRNAs with this interesting residue appears to happen in people too (or at least in HeLa cells).
Finding ψ’s in mRNA is a big contradiction to everything we’ve been taught! The next step is to figure out what they are doing there, and there are lots of possible answers.
One possibility is that the newly discovered mRNA modifications make possible a whole new set of translated proteins. Adding a ψ to mRNA changes codon usage at that position in vitro. For example, one study found that converting the stop codons UAA and UGA to ψAA and ψGA, respectively, changed them from stop codons into sense codons both in vitro and in vivo. So ψ’s in mRNAs could cause a whole slew of new alleles to appear under certain conditions – at the RNA level instead of the DNA level. A proteomics study should help determine whether this is happening or not.
Another possibility has to do with the fact that ψ’s make an RNA more stable. Making certain mRNAs more stable could increase the number of protein molecules they can produce: yet another way to affect gene expression post-transcriptionally. A stability study of mRNA and/or more proteomics might help determine whether this is the function of the unusual modifications.
Whatever the reason, it definitely looks like another bit of biological dogma has been overturned with the help of our faithful and reliable friend yeast. Yes Virginia, mRNA almost certainly has the modified nucleotide ψ. And, as usual, thanks to yeast for teaching us the fundamentals of our own basic biology.
* CMC stands for N-cyclohexyl-N′-(2-morpholinoethyl)carbodiimide metho-p-toluenesulphonate
by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics