New & Noteworthy
August 21, 2014
Say you want to send a letter to your friend on the other side of the country. First off you’ll need to put the right address and postage on the envelope. Then you’ll need the U.S. Postal Service (USPS) to take your letter and deliver it to the right person. The stamp tells the USPS to deliver the letter, and the address indicates where it should be delivered (unimpeded by snow nor rain nor heat nor gloom of night, of course!).
It turns out something similar happens in human cells with aggregated proteins. Aggregated proteins are “stamped” by attachment of the small protein ubiquitin and “addressed” to the Atg8 protein. Atg8p triggers the aggregated proteins’ incorporation into autophagosomes for eventual degradation in the lysosome.
And just as it can be devastating if your mail doesn’t get to where it needs to go, so too can it be devastating for these aggregates to accumulate instead of being properly delivered. A buildup of these aggregates is a big factor in Alzheimer’s and Huntington’s diseases.
Enter the cellular USPS. Just as is the case for a prepared letter, the human cell has a service that delivers the ubiquinated proteins to the autophagosome, in the form of the protein adaptor p62 (SQSTM1) and its relative, NBR1.
These adaptor proteins can act as a postal service because they recognize both the aggregated proteins’ stamp (ubiquitin) and their addressee (Atg8p). Specifically, they each possess an ubiquitin-conjugate binding domain (UBA) and an Atg8-interacting motif (AIM). The protein p62 in particular has been shown to associate with protein aggregates linked to neurodegenerative diseases like Huntington’s disease.
In a new paper published in Cell, Lu et al. asked whether there is a link between the ubiquitin and autophagy systems in yeast. If so, yeast might provide some clues about diseases like Huntington’s. Proteins stamped with ubiquitin are known to be addressed to the proteasome for degradation in yeast, but no link between ubiquitination and autophagy had previously been seen, even though many central components of autophagy were actually first described in yeast.
Indeed, the authors showed that cells specifically deficient in the autophagy pathway (atg8∆, atg1∆, or atg7∆), accumulated ubiquitin conjugates under autophagy-inducing conditions. This suggests that the ubiquitin and autophagy pathways are connected in yeast, as is the case for humans.
Next, the researchers looked to see if there is an adaptor in yeast analogous to p62 in humans. When they pulled down proteins that bind yeast Atg8p under starvation conditions, they found ubiquitin conjugates and, using mass spectrometry, further identified peptides from a few other proteins – one of which was Cue5p.
Could Cue5p, like p62 in humans, be the postal service that recognizes both stamped ubiquitin conjugates and the addressee Atg8p in yeast? Strikingly, Cue5p had both a CUE domain that binds ubiquitin and an Atg8p-interacting motif (AIM). The authors confirmed in vivo that Cue5p binds ubiquitin conjugates and Atg8p using these domains, particularly under starvation conditions. They also showed that it acts specifically at the stage of ubiquitin-conjugate recognition and on aggregated proteins, without affecting the process of autophagy itself.
Given that Cue5p functions similarly to p62 and p62 is known to associate with protein aggregates involved in neurodegenerative disease, Lu et al. were quick to look for Cue5p substrates. Analyzing ubiquitin-conjugated proteins that accumulated in cue5 mutant cells, they identified 24 different proteins. Although these 24 Cue5p substrates had diverse functions, the common thread was that many had a tendency to aggregate under certain conditions such as high temperature.
Could Cue5p then actually facilitate removal of cytotoxic protein aggregates in neurodegenerative diseases? Indeed, the authors showed that CUE5 helped clear cytotoxic variants of the human huntingtin protein (Htt-96Q) when it was expressed in yeast, and that Htt-96Q is ubiquitinated in yeast.
These experiments started with an observation in human cells that prompted discovery of an analogous system and adaptor protein in yeast. Now the authors turned the tables and used yeast to look for new adaptor proteins in human cells. Using bioinformatics, they identified a human CUE-domain protein, Tollip, which, although different in its domain organization from Cue5p, contains 2 AIM motifs.
To make a long story (and a lot of work!) short, they showed that Tollip binds both human Atg8p and ubiquitin conjugates and clears cytotoxic variants of huntingtin in human cells. Expressed in yeast, it similarly binds ubiquitin conjugates and Atg8p and suppresses the hypersensitivity of cue5∆ cells to the variant huntingtin protein Htt-96Q. So Tollip is a newly defined adaptor protein and functional homolog of Cue5p!
Letter carriers of one sort or another have been around for as long as human civilization has existed, from homing pigeons to FedEx. Now we know that for even longer, cells from yeast to human have been using similar ways to recognize stamped proteins and deliver them to the right address. And once again, yeast has helped us understand the inner secrets of human cells.
August 14, 2014
One way to think about the cell is that organelles float around in it like those globs in a lava lamp. This is obviously a simplification, but it’s also true that organelles aren’t locked into place. As usual, the real picture lies somewhere in between these two extremes.
What we know about the architecture of the cell has mostly been discovered using classical cell biology and genetic techniques. But in a paper published in Molecular BioSystems, Cohen et al. uncovered some very interesting small details using a very large-scale approach.
The authors were interested in peroxisomes, where a lot of critical metabolic reactions happen (or fail to happen, in several human diseases). The researchers were able to see that peroxisomes not only interact with other organelles, but they contact the endoplasmic reticulum (ER) and mitochondria in a way that could be extremely important for cellular metabolism. And surprisingly, it was by combining a variety of different high-throughput techniques that Cohen and colleagues could uncover this fine structure.
The first step was to set up two reporter constructs to look for genes involved in two different peroxisomal processes.
One reporter was a red fluorescent protein, mCherry, modified to carry a peroxisomal targeting signal and show whether import into peroxisomes was normal. Another reporter, a peroxisomal membrane protein (Ant1p) tagged with green fluorescent protein (GFP), would show whether peroxisomal membranes were normal.
The reporters were crossed into mutant collections, creating one strain for each gene in the genome that had either a complete deletion (for nonessential genes) or a knock-down allele (for essential genes), plus both reporters. Now the researchers could systematically test for genes that, when mutated, affected one or both of these aspects of peroxisomal biogenesis.
To visualize the mutant phenotypes, they used a sophisticated technique termed “high-content screening.” This is an automated way to analyze micrographs that both pinpoints the intracellular location of a fluorescent reporter and measures its quantity. Screening the mutant collection in this way showed that 56 strains had altered distribution of the two different reporter proteins. Some had a reduction in peroxisomal protein import (mCherry fluorescence), while some had fewer or no peroxisomes and some had peroxisomes that were smaller than normal (GFP fluorescence).
One result that caught the researchers’ eyes was that one of the strains with smaller peroxisomes had a mutation in the MDM10 gene. Mdm10p is part of the ERMES (ER-Mitochondria Encounter Structure) complex that tethers mitochondria to the ER, and this wasn’t previously known to have any connection with peroxisomes. Strains that were mutant in other ERMES subunits had the same phenotype, confirming that the complex has something to do with peroxisome structure. Other results from the screens added weight to the idea of a three-way connection between peroxisomes, the ER, and mitochondria, and the authors went on to show that peroxisomes often sit at the ERMES complex where mitochondria contact the ER.
Next, to test whether mitochondria might have specific subdomains where peroxisomes interact, the authors used yet another large-scale screen. In the C-terminal GFP fusion library, where each yeast open reading frame is C-terminally tagged with GFP, 96 strains showed a punctate pattern of the fluorescent signal – meaning that the protein was concentrated in spots, rather than evenly distributed. They labeled the mitochondria with a red fluorescent marker protein in these strains and, again using the high-content screening system, identified protein spots that co-localized with mitochondria. The most intense hit was for Pda1p, a subunit of the mitochondrial enzyme pyruvate dehydrogenase (PDH), and a similar result was obtained for another PDH subunit. So PDH isn’t distributed uniformly in the mitochondrion, but is instead concentrated in clusters.
Looking more closely using the various reporter constructs in their collections, the authors found that peroxisomes and the ERMES complex most often co-localized with those mitochondrial globs of PDH. It would make metabolic sense for peroxisomes to hang out near PDH on mitochondria because this could increase the local concentration of metabolites that they both use.
Intriguingly, Cohen et al. also found that mitochondria and peroxisomes co-localized in mammalian cells. Given that many diseases are linked to peroxisomal metabolism, this is an important avenue to investigate.
So while organelles don’t float around in the cell quite as fluidly as the globs in a lava lamp, the data generated from large-scale approaches boiled down to learning some very fine-grained detail about cellular architecture. We think that’s, like, groovy.
July 15, 2014
In the art of rock balancing, the artist positions large rocks with exquisite precision. If he or she succeeds, the rocks counterbalance each other and stay in seemingly impossible positions to make a surprising and beautiful sculpture. But a little uneven pressure is enough to make the whole thing collapse.
It turns out that the cellular acetylation state is just as precisely balanced. In a new GENETICS paper, Torres-Machorro and Pillus identify Esa1p, an acetyltransferase, as the balancing artist in Saccharomyces cerevisiae cells.
Acetylation is an important type of protein modification. Histones, the proteins that interact with DNA to provide structure to chromosomes, are acetylated by histone acetyltransferases (HATs) and deacetylated by histone deacetylases (HDACs). Some HATs and HDACs also act on non-histone proteins.
The acetylation state in a cell is a dynamic process. All those HATs are adding acetyl groups at the same time that HDACs are removing them. The final level of acetylation depends on the activities of each of these classes of proteins.
Acetylation of histones has been associated with increases in gene expression and deacetylation with decreases. So to keep gene expression levels in balance, it is very important to keep acetylation balanced as well. Throwing acetylation patterns just a bit out of whack can have profound consequences on global gene expression that can ultimately lead to cell death.
The authors focused on one particular HAT, Esa1p, that acetylates histones H4 and H2A and also has non-histone targets. They were intrigued by the fact that yeast cells cannot survive without Esa1p, since no other HAT or HDAC subunit is essential in yeast.
An obvious explanation for lethality is that losing this protein leads to too low a level of acetylation. They reasoned that if they also knocked out an HDAC, then the overall acetylation levels might increase and so rescue the esa1 null mutant. And they were right.
Using a plasmid-shuffling method, they created various double mutant strains of esa1 and HDAC genes, and found that a strain that was mutant in esa1 and also in either the SDS3 or DEP1 genes was viable. SDS3 and DEP1 both encode subunits of the Rpd3L HDAC complex.
Torres-Machorro and Pillus next characterized the esa1 sds3 double mutant further. They found that although the sds3 mutation suppressed the inviability of the esa1 mutant, it did not suppress other phenotypes such as sensitivity to high temperature and DNA damaging agents.
The authors found that the sds3 mutation subtly increased histone H4 acetylation, which was low in the absence of Esa1p. However, acetylation levels of a different histone, H3, remained high even in the absence of Esa1p. This suggested that the fundamental problem in the esa1 null mutant was an imbalance in the global state of histone acetylation.
To test this hypothesis, the researchers used a variety of different genetic methods to tweak the balance of cellular acetylation in the esa1 sds3 mutant. They created mutations in histones H3 and H4 that made it seem as if acetylation was low or high, and they also mutated other genes for HDAC subunits. It is as if they were passers-by who decided to poke at a balanced rock sculpture to see what it took to bring the whole thing down.
Although the details are too numerous to report here, the results showed that by using these genetic methods to tweak the overall acetylation state of the cell, the fitness of the esa1 sds3 strain could be improved: phenotypes such as slow growth, sensitivity to high temperature or DNA damaging agents, or cell cycle defects were suppressed to some extent by the various manipulations. This lends support to the hypothesis that Esa1p is the master balancer of acetylation levels in the cell and that this is its essential function.
This balancing act may happen in human cells too. Esa1p has a human ortholog, TIP60, that has been implicated in cancer and other diseases. Like Esa1p, TIP60 is essential and is involved in the DNA damage response.
So yeast teaches us that the acetylation of proteins is balanced on a knife’s edge. Even the slightest changes can lead to a collapse in global gene regulation, which can have catastrophic effects like cancer. All that we learn about Esa1p, the acetylation balancing artist, may have much broader implications for human health.
June 19, 2014
In the Matrix Trilogy, the delicate balance of a virtual world is upset by a rogue computer program that goes by the name of Agent Smith. This program finds and touches other agent programs, converting them into copies of itself. Eventually, all the agent programs are copies of Agent Smith and only the hero Neo can save humanity in an epic battle within the virtual world of the Matrix.
A new study out in GENETICS by Li and Du provides additional evidence that prions in the yeast Saccharomyces cerevisiae work similarly to Agent Smith, in that they spread through a direct contact model. These prions are proteins that have entered a rogue conformation, and they end up converting all copies of the same protein into a similar rogue conformation. The proteins change from a hardworking Agent Smith trying to do its job into something that mucks up the working of a cell. And the results, at least in humans, can be as catastrophic for the cell as Agent Smith was for the Matrix.
Mad cow disease, for example, is caused by prions converting the prion protein (PrP) in the brain cells of people from a useful conformation to a dangerous one that spreads. As the conformation spreads throughout the cell, these prions form amyloid fibrils that eventually kill the cell. When enough brain cells are killed, the person dies.
The authors chose to work in yeast because unlike in people, there are multiple examples of proteins in yeast that can go prion. The list includes Sup35p, Ure2p, Rnq1p, Swi1p, Cyc8p, Mot3p, Sfp1p, Mod5p and Nup100p. As you might guess from the sheer number of these prion-ready proteins, prions actually do more than kill a cell in yeast; they can serve useful functions. Scientists have yet to identify any useful functions for the prion form of PrP in people.
Having multiple prions in a cell allowed Li and Du to perform some experiments to try to distinguish between two models of prion conformation spreading. In the first, called the cross-seeding model, the prion acts very much like Agent Smith in that it needs to contact a “healthy” protein to convert it into a prion. In the second model, the titration model, factors in the cell that prevent prion formation are titrated out when prions form. As the factors are taken out of commission, prions are free to form.
The main evidence in this study that supports the cross-seeding model has to do with the localization of pre-existing prions during the de novo formation of a new prion. Li and Du found that the prion [SWI+] localized to newly forming [PSI+] prions but not to already formed [PSI+] prions. This is not the result we would expect if prion formation were due to titrating out of inhibitors of prion formation. If that were the mechanism, then there would be no reason for [SWI+] to colocalize with newly forming [PSI+]. These experiments are like having a google map of the Matrix where we could see Smiths converting other agents by touch and then moving on and touching other agents.
Work like this is important for helping to find treatments for prion associated diseases and, perhaps, other amyloid fibril forming diseases like Huntington’s or Alzheimer’s. Scientists need to focus on the amyloid fiber forming proteins themselves instead of trying, for example, to ramp up the activity of factors that inhibit formation. Scientists probably need to eliminate Agent Smith to prevent the destruction of the Matrix and all of mankind.
This is how prions turn other proteins into copies of themselves: