We describe here the operation of a microfluidic device that allows continuous and high-resolution microscopic imaging of single budding yeast cells during their complete replicative and/or chronological lifespan.
We demonstrate the use of a simple microfluidic setup, in which single budding yeast cells can be tracked throughout their entire lifespan. The microfluidic chip exploits the size difference between mother and daughter cells using an array of micropads. Upon loading, cells are trapped underneath these micropads, because the distance between the micropad and cover glass is similar to the diameter of a yeast cell (3-4 μm). After the loading procedure, culture medium is continuously flushed through the chip, which not only creates a constant and defined environment throughout the entire experiment, but also flushes out the emerging daughter cells, which are not retained underneath the pads due to their smaller size. The setup retains mother cells so efficiently that in a single experiment up to 50 individual cells can be monitored in a fully automated manner for 5 days or, if necessary, longer. In addition, the excellent optical properties of the chip allow high-resolution imaging of cells during the entire aging process.
Budding yeast is an important model organism for aging research1. Until recently studying replicative aging in yeast cells was a laborious process requiring a dissection method, in which each bud was manually removed from the mother cell2,3. To solve this problem, we recently presented a novel microfluidic setup able to track individual mother cells throughout their entire lifespan4.
In our microfluidic chip, yeast cells are trapped under soft elastomer-based micropads (see Figure 1). A continuous flow of medium washes away newly formed daughter cells and provides the cells with fresh nutrients. In a single experiment, up to 50 mother cells can be monitored in a fully automated manner throughout their entire replicative lifespan. Due to the excellent optical properties of the microfluidic chip, it is possible to simultaneously monitor different aspects of yeast cell biology (e.g. by using fluorescent proteins).
Compared to the classical dissection method, the microfluidic setup provides substantial advantages. It ensures a defined and constant environment during the whole aging experiment. It requires no expensive specialized equipment and can be run on any microscope equipped with automated focus and time-lapse abilities as well as temperature-control for cell cultivation. The production and operation of the microfluidic chips can be learned within a few days. In addition, cells can be directly loaded from an exponentially growing culture, an advantage over another recently published microfluidic method5, which requires biotinylation of mother cells.a Combined with high-resolution imaging, the here described method can be used to measure gradual changes in cellular morphology, protein expression and localization during yeast aging in an unprecedented manner. The capability for long-term monitoring of single cells also provides unique possibilities for yeast cell cycle studies.
aThis method has recently been optimized to remove the biotinylation from the protocol16, which was published while this manuscript was in review.
1. Production and Preparation of a Silicon Wafer Mold
Microfluidic chips are created from a silicon wafer mold produced by soft lithography. These wafers can be reused many times to produce microfluidic chips. It is advisable that production of a respective wafer is performed by a group specialized in microfluidics6.
The wafer is made in a two-step photolithography process using two different layers of negative photoresist, SU-87. The bottom layer is used to generate the cell trapping area (SU-8 2002; height 3-4 μm), whereas the channels are made with the top layer (SU-8 2010; height 10 μm). An impression of the wafer production process can be found in Huang et al8. The drawings of the microfluidic chips can be obtained from the authors as well as advice on how to obtain a respective wafer.
2. Production of Microfluidic Chips
Make sure the PDMS chip is still lying upside down on the bench while punching the holes. This prevents PDMS from sticking to the inside of the channel, which could cause blockage of the channel.
3. Preparing the Chip for Cell Loading
Failure of a microfluidic device can potentially cause leakage of medium into the microscope. To avoid this, it is advisory to use a metal holder and/or seal the sides of the chip where the PDMS meets the glass with nail polish or epoxy glue. This prevents leakage of medium in case the PDMS chip is not bonded well enough to the cover glass. Also the tubing connections can be secured around the insertion points in a similar manner to avoid leakage of medium. A home-made water sensor that switches off the syringe pump in case of leakage can be used as an extra safety precaution. Drawings of the metal holder especially designed for the microfluidic chip can be obtained from the authors.
The use of oil immersion objectives is preferred over that of water immersion objectives because oil will not dry out over the time-course of the experiment.
Depending on the microscope there might be issues with retaining the focus during the initial hours of the experiment. It is therefore advisable to leave the chip for one to two hours in the microscope stage so it can settle down before loading cells and starting the experiment.
4. Loading of Yeast Cells into the Microfluidic Chip
The preferred cell count for loading is between 1-5 x 106 cells per ml. Cultures with higher cell counts need to be diluted before loading. When culturing cells on YPD, cell loading efficiency can be improved many-fold if cells are first washed with and resuspended in minimal medium containing a similar glucose percentage before loading. During operation of the chip, YPD medium can again be safely used.
In this protocol, cells are loaded into the microfluidic chip directly from mid-exponential culture. To ascertain whether the age distribution of cells trapped in the microfluidic chip is similar to that of the culture prior to loading, cells were stained with wheat agglutinin conjugated to FITC (WGA-FITC) to visualize bud scars. As can be seen in Figure 3, the entrapment of cells under the micropads of the microfluidic chip is not biased to cells of a certain age.
Replicative lifespan can simply be determined by counting the number of buds that are produced by a single mother cell. This data is transformed into a lifespan curve by plotting the percentage of viable cells against the number of buds produced. Statistical tests, such as those performed by Kaeberlein, et al.9, can be used to compare lifespan data from different experiments. Figure 4 shows an example of lifespan curves obtained for a WT BY4741 yeast strain and two mutants (sir2Δ, fob1Δ).
Movie 1 Time-lapse movie of a single WT BY4741 yeast cell growing on YPD medium. Images were taken every 10 min. The cell produces a total of 30 buds before it dies. Scale bar, 5 μm. Click here to view movie.
An important advantage of the microfluidic method is the ability to use continuous high-resolution imaging. To monitor changes in mitochondrial morphology as cells age, an aging experiment was performed with BY4742 yeast cells expressing ILV3-GFP, which is targeted to the mitochondria. Figure 4 gives an overview of mitochondrial morphology for the same set of cells at different replicative ages (0 buds, 10 buds and prior to death). The morphology changes seen in mid-aged cells resemble those reported in literature before10. Although in this particular representative experiment, the flow rate of the medium was not optimal and daughter cells were retained with a relatively high frequency in contrast to Movie 1, the quality of the experiment was still sufficient to analyze mitochondrial morphology in 49 individual cells over their entire lifespan.
Movie 2 Time-lapse movie of a single BY4742 yeast cell expressing ILV3-GFP, which is used to visualize the mitochondria. Images were taken every 30 min. Both the brightfield as well as the GFP images were background corrected before merging the two channels into a single image using ImageJ. Scale bar, 5 μm. Click here to view movie.
Figure 1. Schematic overview of the microfluidic setup. The height difference between the PDMS micropad and cover glass is similar to the diameter of a yeast cell (about 3-4 μm). Loading: Yeast cells are loaded under the PDMS micropads. Culturing: Fresh medium flows continuously over the trapped cells (blue arrow). Dissection: Emerging daughter cells are flushed away by the flow of medium (orange arrow).
Figure 2. Schematic illustration of the channel design of the microfluidic chip. A. Black circles indicate the location of the punched holes at the end of the inlet, outlet and side channel. B. Medium flows through the inlet channel (10 μl/min) and exits via the larger side channel (orange arrow). There is no flow through the outlet channel, because of the resistance difference between the side and outlet channel. C. Cells are loaded into the chip (black arrow) via the outlet channel. The flow through the side channel is reduced to 0.5 μl/min. D: After flushing the side channel, the side channel is blocked and medium starts to flow over the cells trapped underneath the micropads (1-5 μl/min).
Figure 3. Cell loading onto the chip is not biased towards a certain cell age. A mid-exponential YSBN6 WT culture (107 cells/ml) was labeled with WGA-FITC (2 μl of 5 mg/ml stock solution per 106 cells) for 1.5 hr. Cells were then put on a glass slide or loaded into the microfluidic chip. Z stacks were made (15 z-slices at 0.8 μm distance) at an excitation wavelength of 470 nm using a 63X magnification water immersion objective. The number of bud scars on each cell was counted manually. Bud scars were determined from 75 cells in the culture and 79 cells in the microfluidic chip.
Figure 4. Examples of life span curves obtained for BY4741 WT (n=76), sir2Δ (n=63) and fob1Δ (n=58) using the here described microfluidic device. Both deletions have a strong, yet opposite, effect on the measured median replicative lifespan of the yeast cells. For each experiment, the strains were grown overnight in YPD medium to stationary phase and then allowed to resume exponential growth by dilution into fresh YPD medium and incubation at 30 °C for 3 hr before starting the experiment. Data was obtained from Lee et al.4.
Figure 5. Mitochondrial morphology as a function of age. A: A representative example of age-associated changes in mitochondrial morphology in a single cell (white arrow). All images are scaled identically. Scale bar, 5 μm. B: Mitochondrial morphologies observed in a set of 49 cells before producing their first bud (0), after 10 buds (10) and prior to death (†). An example image of each morphology class is included. The presented data was obtained during a single experiment and the same set of cells was used for scoring morphology at each time point. Mitochondria were visualized in a BY4742 strain expressing a GFP-tagged version of the mitochondrial protein ILV3 (obtained from the GFP collection database15). Both the brightfield as well as the GFP images were background corrected before merging the two channels into a single image using ImageJ.
The microfluidic method described here is an important novel tool in aging research as it enables simple and automated generation of yeast replicative lifespan data in combination with continuous high resolution imaging. These attributes are major improvements over the experimental possibilities of the classical dissection method, yet there are a few limitations of the method that need to be taken into account.
Note that the determined replicative lifespan can be affected by the efficiency of the retention of mother cells: Every cell kept under a micropad has a certain probability to be washed out from the chip (e.g. the budding cell of a neighboring cell could push it away from the micropad). The integrated probability of a cell to be washed out increases with increasing lifespan. Thus, it is more likely that short-lived cells completed their life cycle before being washed away than long-lived cells. The fact that similar lifespan data was generated with the microfluidic setup4 as with the classical microdissection method11-13 indicates that this potential issue is only minor, if not irrelevant.
It is important to note that with the microfluidic setup it is not possible to select virgin daughter cells at the start of the experiment, in contrast to the classical dissection method. Nevertheless when loading cells from an exponentially growing liquid culture into the chip, the majority of the loaded cells are newly born or relatively young (i.e. about 54% of the cells never budded before and around 27% once)14. This means that lifespans are underestimated by 1 to 2 generations at most. In the exceptional case that the age structure of the cell population is significantly altered, it is possible to enrich the cell population for virgin daughters prior to loading, e.g. by gradient centrifugation or elutriation.
The microfluidic chip can in principle be used to study haploid yeast strains in different culture media. However, different yeast strains and/or media may require some optimization of flow rate to allow efficient retention of mother cells. As cells are retained based on their size, it is not possible to load yeast strains that are significantly larger or smaller than 3-4 μm in diameter, such as diploid yeast cells. However by producing a wafer with different photoresists, microfluidic chips could be generated with a different spacing between the cover glass and micropads to allow loading of these differently sized cells.
Although the protocol presented here describes how to make a chip with a single channel, it is possible to add multiple channels in a single chip and run parallel experiments simultaneously. Depending on the number of desired channels in a single chip, it may be necessary to create an adapted wafer in which the channels are spaced more closely together.
The authors have nothing to disclose.
We would like to thank Laura Schippers for writing the first versions of the cell loading protocol and Marcus de Goffau and Guille Zampar for scoring mitochondrial morphologies.
Name | Company | Catalogue number | Comments |
REAGENTS | |||
DC Sylgard 184 elastomer | Mavom bv | 1060040 | This package contains PDMS base and PDMS curing agent. |
Glass Petri dishes 120/20 mm | VWR International | 391-2850 | |
Cover glasses 22×40 mm | CBN Labsuppliers BV | 190002240 | |
Tough-Tags | Sigma-Aldrich | Z359106 | |
Aluminum foil | |||
Plastic disposable cup | |||
Serological pipette 5 ml | VWR International | 612-1245 | |
Scotch tape | VWR International | 819-1460 | |
Baysilone paste (GE Bayer silicones) | Sigma-Aldrich | 85403-1EA | |
PTFE microbore tubing, 0.012″ID x 0.030″OD | Cole Parmer | EW-06417-11 | Referred to as thin tubing |
Tygon microbore Tubing, 0.030″ID x 0.090″OD | Cole Parmer | EW-06418-03 | Referred to as thick tubing |
Scalpel | VWR International | 233-5334 | |
50 ml Luer-Lok syringes | BD | 300137 | |
5 ml syringes, Luer tip | VWR International | 613-1599 | |
Tweezers | VWR International | 232-2132 | |
20 Gauge Luer stubs | Instech Solomon | LS20 | |
Syringe filters (pore size 0.20 μm) | Sigma-Aldrich | 16534K | |
Stainless steel catheter Plug, 20 ga x12 mm | Instech Solomon | SP20/12 | |
Petri dishes | VWR International | 391-0892 | |
EQUIPMENT | |||
Benchtop UV-Ozone Cleaner | NOVA Scan | PSD-UVT | |
Harvard Pump 11 Elite | Harvard Apparatus | 70-4505 | |
SU-8 silicon master mold (wafer) | Self-made; For details contact corresponding author | ||
Balance | Sartorius corporation | ED4202S | |
Vacuum pump | KNF Neuberger | N022 AN.18 | |
Desiccator | VWR International | 467-2115 | |
Hot plate | VWR International | 460-3267 | |
Optional: Metal holder for cover glass (22×40 mm) | Self-made; For details contact corresponding author | ||
(Fluorescence) Microscope with 60x objective, autofocus, time-lapse abilities and preferably an automated (motorized XY control) stage | Nikon | Eclipse Ti-E |