The budding yeast Saccharomyces Cerevisiae, a simple single-celled organism, has served as an important model for aging research. In the past few decades, genetic studies have identified a number of conserved pathways that regulate lifespan across species 
. Such studies have helped establish the modern field of the molecular genetics of aging. Yeast is also one of the favorable model organisms for studying aging, due to its short lifespan and the relative ease of genetic manipulation. In addition, recent functional genomic studies have revealed a large number of regulatory interactions from which a global gene regulatory network is beginning to emerge. Knowledge of such a network makes it possible to study aging from a systems perspective.
The phenomenon of yeast replicative aging was discovered about half a century ago, when Mortimer and Johnston reported that single yeast cells have finite replicative lifespan (RLS), defined as the number of daughter cells a mother cell can produce throughout its life 
(). The original lifespan assay, as devised by Mortimer and Johnston, was to grow virgin mother cells on a agar plate and remove daughter cells from their mothers by micro-dissection using a micromanipulator (a microscope with a dissection needle and a movable stage). Removing daughter cells is absolutely necessary in order to track the lifespan of mother cells. Without the removal of the daughter cells, the cell population will quickly expand to a big clone in less than 10 generations, which is much shorter than the typical life span of a mother cell (25 generations on average).
The design of the microfluidic system for yeast aging analysis
50 years after the initial discovery by Mortimer and Johnston, manual micro-dissection remains the canonical method for yeast lifespan analysis. This has become a major bottleneck limiting the progress of the field. The traditional method is laborious and time-consuming, make it very difficult to perform large-scale screening for genetic mutations that extend the lifespan. More importantly, with the traditional assay, it is almost impossible to follow molecular markers throughout the lifespan of the mother cells. This pose a great challenge to phenotyping aging in single cells at the molecular level.
Due to its technological importance, several groups attempted to develop methods for retaining mother cells while removing daughter cells automatically 
. For example, exploring mother/daughter size difference (mother cell is in general larger than its daughters), a microfluidic device was developed that confines mother cells in micro-jails with open gates for daughter cells to escape 
. Daughter cells can then be separated by the flow. However, such device only works for the first few generations. As the size of mother and daughter grows with age, the daughter cells eventually jam the gates.
Recently we have developed a microfluidic system that is capable of retaining mother cells in microfluidic chambers while removing daughter cells automatically throughout the lifespan of the mother cells 
. To achieve stability, we explored two properties of budding yeast cell division: 1) usually the size of the mother cell is bigger than that of the daughter; 2) the cell wall of the daughter comes from de novo synthesis at the budding site of the mother 
, so that if the surface of the mother cell is labeled, the daughter would not inherit the label. In the device, mother cells are trapped by a combination of geometric confinement (the height of the chamber is comparable to the size of mother cells) and adhesion between biotin labeled mother cell surface and BSA-Avidin modified glass. Although effective, the requirement for surface labeling and glass modification makes the device fabrication and operation more demanding. We found that geometric confinement by itself alone is not stable and is sensitive to the height of the chamber: if it is too high, the mother cells will not be stably trapped; if it is low enough to stably trap mother cells, there is a certain probability that daughter cells will be trapped and jam the device.
Here we report the development of a new generation of microfluidic device for yeast aging study that eliminates the requirement for surface labeling (a similar device has been developed recently by Lee et al 
, see Note Added) and can be used to study problems not possible with the first generation device. Such system allows us to simplify the experimental procedure and to achieve higher success rate. The elimination of surface labeling makes it possible to generalize the study to species other than budding yeast, as the surface labeling is based on differential partitioning of the cell wall – a specific property of budding yeast. Furthermore, this design allows trapping of the daughter cell of a trapped mother cell and the probability of trapping daughter cells can be adjusted by changing the cross section of the columns. This allows us to analyze mother/daughter asymmetry as a function of age, which is not possible with the first generation device since trapping daughter cells will eventually lead to jamming of the whole device.
Using this device, we have reproduced the lifespan curves of long and short-lived mutants, discovered a surprising change of the expression of a translation elongation factor (generally believed to be constitutively expressed) in single mother cells during aging, and analyzed asymmetric partitioning of a stress response reporter between mother and daughter cells as a function of mother age.