October 09, 2017
A Swiss Army knife is a marvelous thing. It is a bunch of useful tools all wrapped up in something small enough to put in your pocket.
Need to cut a string? It has the scissors to do that. How about cutting that little piece of wood? Yup, the tiny saw can do that. It has lots of tools that can do lots of things.
Now of course, it isn’t perfect. You would not want to use those scissors for cutting fabric or to use that saw to cut a 2 by 4. And there really isn’t any easy way to make a better pair of scissors or a better saw in the Swiss Army knife without wrecking the other tools in it.
If you make a better pair of scissors, you’ll have a worse saw. And vice versa.
That is why if you want to improve the saw, you need to isolate it away from the other tools. This holds true for the scissors and yes, even the knife.
Multifunctional proteins can have similar constraints. Tweak one function and it can mess up the other functions. A better saw makes a worse pair of scissors.
Nature doesn’t usually improve on a protein’s function by pulling out the part of the protein that does a specific job and making it better. No, instead, she duplicates the gene with the instructions for the protein. Now Nature can optimize each function of each gene copy—both improved functions can still happen albeit with separate proteins.
It is like having two Swiss Army knives and improving the saw on one and scissors on the other. You still have scissors even when your improvements on your saw have wrecked the attached scissors. They are just on the other knife.
In a new study in GENETICS, Hanner and Rusche explored this idea of “subfunctionalization” using the Saccharomyces cerevisiae proteins Sir3p and Orc1p. The genes that encode these proteins are thought to have arisen from a single ancestral gene after a whole-genome duplication around 100 million years ago.
The ORC1 gene is highly conserved in eukaryotes and serves an important role in DNA replication. In addition to this role, in S. cerevisiae, it also works to silence the mating type loci by recruiting the SIR complex to either HMRa or HMRα. Sir3p is also involved in silencing genes including HMRa or HMRα, but it does its job by recognizing deacetylated histones.
The basic strategy of this study is to compare Sir3p from S. cerevisiae with KlOrc1p, an Orc1p homolog from the yeast Kluyveromyces lactis that is involved in both DNA replication and gene silencing.
Since K. lactis did not undergo the whole genome duplication that S. cerevisiae did, the idea is that KlOrc1p represents the multifunctional ancestral protein (or at least what it has evolved into over the past 100 million years or so). Sir3p is the saw while KlOrc1p is the saw as part of the Swiss Army knife.
In the first experiment, Hanner and Rusche wanted to see if Orc1p from yeasts that did not undergo the whole genome duplication could rescue strains of S. cerevisiae lacking Sir3p. In other words, do these non-duplicated proteins have Sir3p functions or did Sir3p evolve new functions once freed from its DNA replication constraints?
These authors showed that KlOrc1p could not rescue mating in a strain of S. cerevisiae lacking Sir3p (but still containing its own Orc1p). They took this even further and showed that Orc1p from two other non-duplicated yeasts, Lachancea kluyverii and Zygosaccharomyces rouxii, could not rescue the S. cerevisiae strain either. Sir3p in S. cerevisiae has functions that none of these “ancestral” Orc1p’s have.
The next step was to create chimeras between Sir3p and KlOrc1p to identify the parts of Sir3p that developed new functions. Both proteins consist of four broad subdomains: AAA+, winged helix, N-terminal bromo-adjacent homology (BAH), and a rapidly evolving linker.
The researchers swapped domains between the two proteins and found that a Sir3p with an AAA+ subdomain from the KlOrc1p did not allow a S. cerevisiae strain lacking Sir3p to successfully mate. So Sir3p has gained some new function in AAA+ that the ancestral protein lacked. The authors were able to narrow down this region a bit more—the AAA+ subdomain lacking its first two alpha helices worked too.
Hanner and Rusche used chromatin immunoprecipitation assays to show that this chimeric protein did not make it to the HMRa or HMRα loci or to other repressed genes. They also looked at mRNA levels and showed that genes that should have been repressed weren’t. This Sir3p with the AAA+ subdomain of KlOrc1p clearly cannot do the job of Sir3p.
And it isn’t because of the loss ATPase function of the AAA+ domain in Sir3p. While Sir3p has an ATPase domain, it does not have any discernible ATPase activity. When Hanner and Rusche replaced the KlOrc1p ATPase domain with the ATPase domain of Sir3p, this protein still couldn’t rescue the S. cerevisiae strain deleted for SIR3. The ATPase activity of KlOrc1p does not explain its inability to rescue the strain.
It is known that the AAA+ domain of Sir3p interacts with Sir4p. Another possible explanation for the failure of KlOrc1p to complement the loss of Sir3p may be that KlOrc1p has over the last 100 million or so years lost the ability to interact with the S. cerevisiae version of Sir4p. This does not seem to be the reason as adding in the K. lactis version of Sir4p, KlSir4p, to the S. cerevisiae strain deleted for Sir3p, did not allow the strain to mate. Inability to interact with Sir4p does not seem to explain the failure of KlOrc1p to rescue the strain.
The authors hypothesize that once freed of the constraints of the multifunctional Orc1p, the AAA+ region of Sir3p gained the ability to interact with deacetylated histones. This will need to be tested in future studies.
There you have it. Once freed of having to focus on two things, Sir3p was able to evolve new functions like recognizing deacetylated histones. Like a saw being freed from a Swiss Army knife so that it can sing.
Don’t try this with the saw from a Swiss Army knife.
by Barry Starr, Ph.D., Director of Outreach Activities, Stanford Department of Genetics
Categories: Research Spotlight
May 04, 2016
As HBO’s Silicon Valley scathingly relates, the mapping app from Apple was truly terrible when it was first launched. There are all kinds of funny (scary?) stories in which people following the directions ended up in the wrong place. (Click here for a few more of the epic fails.)
And sometimes it would show impossible ways to get from one location to the other. For example, to get to a certain place, my iPhone would recommend two different routes. After choosing the seemingly easiest route I quickly realized that it would take me through a building. Sometimes, even though it seems there are a couple of different ways to get somewhere, there is actually only one.
It turns out that this can be true in gene expression too. While you might increase expression by either mutating the promoter or duplicating the whole gene, sometimes only duplication is enough. There is just one route from here to there.
This point is driven home in a new study out in GENETICS by Rich and coworkers. Here they show that, under sulfate-limiting conditions, the only way that yeast can boost the expression of the SUL1 high affinity sulfate transporter enough to thrive is by duplicating it.
In many previous experiments, whenever yeast is starved for sulfate, after 100 generations or so the population almost always ends up selecting for a SUL1 duplication rather than increasing expression with a promoter mutation. This duplication results in a fitness advantage of 35% or more, which makes sense – when there isn’t much sulfate available, those cells that can get more of what’s there will outcompete their neighbors.
There are a couple of possible reasons things go down like this. It could be that the duplication is simply the most likely way to increase expression enough to survive, meaning that promoter mutations are possible but rare. Alternatively there may be no way to mutate the promoter enough to adequately increase the activity in such a short window of time. You simply can’t get to enough increased fitness by this route.
The first step in figuring out how to get somewhere is to map the roads in the area. This is also what Rich and coworkers needed to do – they needed to figure out how SUL1 is regulated. They did this in the standard way by nibbling away bits of the sequences upstream of the gene until there was a significant impact of gene expression. This is how they identified the 493 base pairs upstream of SUL1 as its promoter.
The next step was to generate a library of mostly single point mutations in this promoter using error-prone PCR. As might be expected, most of the mutants had no effect or decreased gene expression but a few did increase activity. However, none of these last set increased expression as much as duplicating the gene.
They found 8 mutants that gave a 5% or better increase in fitness with the best being a 9.4% increase. Even when they combined these point mutations they could not increase the fitness much beyond 11%, not even half of the increase in fitness that an extra SUL1 gene gives. It seems that to get the most bang for its buck, yeast needs to duplicate SUL1. At least in the time frame of the experiment, that is.
Using what is essentially the scanning mutagenesis of this promoter, they were able to identify three sites that were important for SUL1 regulation. One site at -465 to -448 corresponded to a Cbf1-Met28-Met4 regulatory site, the second site at around -407 was most similar to either a Met31 or Met32 site and the third site at around -350 matched a Met32 site.
Mutations that resulted in increased activity tended to bring one of these three sites closer to the consensus transcription factor binding site. For example, the strongest point mutation, -353T>G, did this with the Met32 site at -350.
Even with these stronger consensus sequences for sulfate-regulatory transcription factors, none of these point mutations could get the yeast to where it needed to be in the fitness landscape in order to be able to thrive under sulfur limitation. Point mutations just can’t hold a candle to simply duplicating the SUL1 gene. Sometimes there really is just one way to get from point A to point B.
by Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics
Categories: Research Spotlight
May 16, 2013
When a gene is duplicated, one copy usually dies. It is battered by harmful mutations until it eventually just fades into background DNA.
But this isn’t the fate of all duplicated genes. Sometimes they can survive by gaining new, useful functions. The genes responsible for snake venom proteins are a great example of this.
Another way for a duplicated gene to live on is when both copies get different mutations that confer different functions, so that a cell needs both to survive. Two examples of this type of codependent gene survival are highlighted in a new study by Marshall and coworkers. They compared various fungal species and identified cases where two functions were carried out by either one gene or by two separate genes. Surprisingly, these cases involve alternative mRNA splicing, which is a rare process in fungi.
The first gene pair they focused on was SKI7 and HBS1 from Saccharomyces cerevisiae. In this yeast these two genes exist as separate entities, but in other yeasts like Lachancea kluyveri they exist as a single gene which the authors have called SKI7/HBS1.
The SKI7/HBS1 gene makes two differently spliced mRNAs, each of which encodes a protein that matches up with either Ski7p or Hbs1p. In addition, the SKI7/HBS1 gene can rescue a S. cerevisiae strain missing either or both the SKI7 and HBS1 genes. Taken together, this is compelling evidence that SKI7 and HBS1 existed as a single gene in the ancestor of these two fungal species. In S. cerevisiae, after this gene was duplicated each copy lost the ability to produce one spliced form.
The second gene Marshall and coworkers looked at experienced the reverse situation during evolution. PTC7 exists as a single gene that makes two mRNA isoforms in S. cerevisiae: an unspliced form that generates a nuclear-localized protein, and a spliced form that produces a mitochondrial protein.
But in Tetrapisispara blattae, these two forms exist as separate genes. The PTC7a gene is similar to the unspliced form in S. cerevisiae and the protein ends up in the nucleus, while the PTC7b gene is similar to the spliced S. cerevisiae version and its product is mitochondrial.
Because an ancestor of S. cerevisiae had every one of its genes duplicated about 100 million years ago, yeasts have been a great system to study the fate of duplicated genes. This study shows that even though gene duplication is widespread in fungi and alternative splicing is rare, these mechanisms are actually interrelated and each can increase the diversity of the proteins produced by a species.
Fun fact: 544 genes survived duplication in S. cerevisiae. That is around 10%.
by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics
Categories: Research Spotlight
February 06, 2012
As life evolves there is a tendency for increased complexity. Up until now, scientists have mostly focused on gain of function mutations as the motor for this change. This has proven fertile ground for evolution deniers who have claimed that life’s complexity could not have arisen from these rare, gain of function mutations alone.
A new study by Finnigan and coworkers provides an important counterpunch to this argument. These authors resurrected ancient proteins and showed that an increase in complexity can come from much more common, loss of function mutations. Time for the deniers to find a new argument…
Making Molecular Machines
Finnigan and coworkers focused on the evolution of a protein called vacuolar H+ -ATPase or V-ATPase for short. Like other molecular machines, this proton pump consists of many different proteins all working together in a coordinated fashion. One key part of this machine is a rotary ring called V0. (This would be the ring of C proteins in the image to the right.)
In most eukaryotes, V0 is made up of five identical subunits (called Vma3) and one subunit called Vma16. In fungi, a third protein, Vma11, has replaced one of the Vma3 subunits. In other words, the fungal version is a bit more complex than other eukaryotic versions.
Current theories are that these three proteins all arose through gene duplication. Duplication of the Vma3 gene first led to the Vma16 gene and then later in fungi, Vma3 duplicated again this time becoming Vma11. Yeast V-ATPase absolutely requires Vma11 to function and other eukaryotic Vma3 family members cannot replace Vma11.
Using the 139 family members of the Vma family available in GenBank, members of the Thornton and Stevens lab recreated the ancestral proteins that existed before and after the Vma11 gene duplication event. Before the arrival of Vma11, there were only two proteins which the authors have named Anc.3-11 and Anc.16. Anc.3-11 presumably has functions of both Vma3 and Vma11. After the gene duplication event, there were three ancient proteins: Anc.3, Anc.11, and Anc.16.
Using these ancient proteins, the authors first showed that Anc.3-11 could substitute for either Vma3 or Vma11 in yeast. It could even partially rescue a yeast strain that lacked both of the other genes. They then showed Anc.16 could replace Vma16 and most importantly, that the two ancient proteins could replace the three modern ones. They reconstructed an ancient molecular machine that works.
The next step was to figure out what happened after Anc.3-11 duplicated again and the two genes began to evolve into the separate proteins, Anc.3 and Anc.11. Again using the GenBank sequences, the authors predicted that two single mutations were an initial step on the way to the separation of Anc.3-11 activities into the Anc3 and the Anc11 proteins.
The authors engineered each mutation independently into the Anc.3-11 protein and found that one mutation made Anc.3-11 more like Anc.3 and the other made Anc.3-11 more like Anc.11. The complex now required all three Anc proteins instead of just the two for maximal activity. The authors had recapitulated the first evolutionary steps that led to the formation of the three subunit V0 rotary ring.
Finally the authors showed that each of these mutations were loss of function mutations. The Anc.3-11 protein has two different interfaces that interact with Anc.16. The first mutation weakened one interface on Anc.3 and the second mutation weakened the other interface on Anc.11 causing both proteins to now be required to reconstitute the ring. The added complexity arose from a combination of gene duplication and relatively common loss of function mutations.
by D. Barry Starr, Ph.D., Director of Outreach Activities, Stanford Genetics
Categories: Research Spotlight