Researchers Advance Understanding of Biological Noise
New technique allows scientists to measure the abundance of thousands of proteins with unprecedented resolution.
Researchers have developed a new technique that allows them to measure the abundance of thousands of proteins with unprecedented resolution. The approach involves making measurements one cell at a time, thereby permitting researchers to measure the random fluctuations in protein levels that can make one cell different from its genetically identical neighbors.
The new method, which was developed in yeast cells, will help researchers understand variations in protein expression between cells, which biologists call “noise,” and how those variations can be either problematic or beneficial for cells.
Cells produce thousands of different proteins, and modulating the abundance of those proteins is a key to the cell's ability to control their growth and respond to their environments. Indeed, so fundamental is the relationship between a protein's abundance and function that scientists can often gain critical insight into the function of a protein by determining under what conditions and for how long a protein is present in a cell, said Howard Hughes Medical Institute (HHMI) investigator Jonathan S. Weissman, who participated in the development of the new technique.
Weissman and his colleagues at the University of California, San Francisco, reported on the technique in an advance online publication on May 15, 2006, in the journal Nature.
For a variety of technical reasons, however, it has been extremely difficult to make high-precision measurements of the amount of each protein-largely because each protein is only produced in very small quantities. Thus, it is generally necessary to extract proteins from tens of millions of cells to get enough to measure each protein's abundance, and this process can introduce artifacts into measurements.
“The difficulty in accurately measuring changes in protein levels is very frustrating, particularly given the desire of biologists to move increasingly from qualitative to quantitative descriptions of how cells work,” said Weissman.
“It is as if you were trying to assemble a watch from a set of parts, but had no idea how many copies of each part you needed to use,” Weissman said. “Our technique uses intact, living cells and therefore allows us to measure how much of a particular protein is inside a cell in real-time. This provides a wealth of new data for our laboratory and other laboratories interested in understanding and ultimately modifying cellular behavior.”
In fact, the ability to monitor protein abundance on a cell-by-cell basis opens up new insights into how cells function. “When you have to extract proteins from tens of millions of cells in order to make measurements, it is easy to forget that even in genetically identical cells experiencing the same conditions, the concentration of a particular protein can be different enough to affect how the two cells function,” said Weissman.
According to Weissman, this difference in function can be both beneficial and harmful to cells. “When cells are faced with an ever-changing environment, it can be advantageous to have some cells producing a set of proteins that permit them to grow more quickly if one set of conditions prevails, while other cells will grow more quickly under different conditions. In essence, it is possible for a population to hedge its bets on what conditions might arise, without having to make long-term genetic changes,” he said.
In the studies reported in Nature, the research team used a library of yeast cells that Weissman, fellow HHMI investigator Erin O’Shea, who recently moved to Harvard University from UCSF, and their colleagues had previously created of more than 4,100 yeast strains. The library, which was developed for protein localization studies, consisted of strains of yeast where each protein is tagged with a sequence that causes that protein to glow green when illuminated by blue light.
Weissman and his colleagues used a technique known as flow cytometry to rapidly compare the protein content in vast numbers of these tagged cells, allowing them to determine the variability of levels of many proteins. In flow cytometry, a laser beam illuminates cells traveling in a narrow stream and the resulting fluorescence is collected by a series of detectors. The strength of the fluorescent signal is directly related to how many copies of a tagged protein are present in a cell.
To determine the variability of each of the proteins they had tagged, the researchers measured its level in 50,000 different cells. To make the task less daunting, Weissman, post-doctoral researcher John Newman, and HHMI investigator Joseph L. DeRisi at UCSF automated the growth and handling of the yeast cultures and developed software to manage delivery of cell samples to the flow cytometer, as well as to manage and analyze the mass of data on the cells.
“We were fortunate enough to be in an environment where there is a tremendous breadth of technical expertise. This allowed us to integrate different technologies so that we could collect high-quality data very rapidly,” said Weissman. The technique is so rapid that it can measure seven samples per minute, with more than 50,000 cells per sample. Thus, the entire complement of 4,100 yeast strains could be analyzed in a day.
The researchers found the technique to be quite precise and able to distinguish very small differences in protein levels. The results were consistent and agreed with those derived from other analytical techniques to quantify protein differences among cells.
They also found that the tinkering they had done to introduce the fluorescence into the cells did not interfere with the normal recycling and destroying of unwanted proteins that constantly takes place.
To explore how yeast cells adapted to different environments by altering their protein levels, the researchers compared cells grown on nutrient-rich media to those supplied only with minimal nutrients. Similar studies had been done in the past using DNA microarrays, also known as gene chips. This approach, widely used by scientists who wish to analyze the activity of a large number of genes under varying conditions, measures and compares the level of mRNA produced by particular genes.
Many factors influence how cells use those mRNA molecules to produce the corresponding proteins, and Weissman suggests that measuring protein levels directly may provide a more accurate picture of how cells respond to various conditions. “Microarray techniques use mRNA changes as a proxy for protein changes, but it's the protein changes that actually cause physiologic changes in the cell,” he said. In fact, he and his colleagues found a substantial number of cases in which the levels of protein they measured between the two types of cell cultures did not reflect the known differences in mRNA levels under the same conditions.
Their analysis also revealed that the level of noise for various proteins seemed to depend in part on their function. “Proteins that were involved in housekeeping and were central to many functions tended to be very uniform across cells,” said Weissman. “But proteins that were involved in environmental responses tended to vary from cell to cell.” This built-in variability would allow a cell to survive diverse conditions, giving robustness to the population that wouldn't be possible otherwise. Other factors that influenced the level of noise were a protein's location within the cell and the mechanisms used to control its production.
“Overall, we believe that these experiments give us a first overview of the major source of biological noise,” said Weissman. “There have been no other data like this that enable a focus on the cell-to-cell variability in protein levels. I think these data will serve as a defining guide for many future studies that aim to understand specific sources of biological noise.”
Similar fluorescent-tagged libraries of cells from species other than yeast are now under construction in other laboratories and could also be analyzed by the new proteome-wide analysis technique, said Weissman. “However, we think there is a great deal of useful analysis possible with yeast,” he said. “It's a huge research opportunity, and we have only just begun skimming the surface of these data to understand the mechanism and implications of noise.”