Abstract
The vast selection of Drosophila mutants is an extraordinary resource for exploring molecular events underlying development and disease. We have designed and constructed an instrument that automatically separates Drosophila embryos of one genotype from a larger population of embryos, based on a fluorescent protein marker. This instrument can also sort embryos from other species, such as Caenorhabditis elegans. The machine sorts 15 living Drosophila embryos per second with more than 99% accuracy. Sorting living embryos will solve longstanding problems, including (1) the need for large quantities of RNA from homozygous mutant embryos to use in DNA microarray or gene-chip experiments, (2) the need for large amounts of protein extract from homozygous mutant embryos for biochemical studies, for example to determine whether a multiprotein complex forms or localizes correctly in vivo when one component is missing, and (3) the need for rapid genetic screening for gene expression changes in living embryos using a fluorescent protein reporter.
This is a preview of subscription content, access via your institution
Access options
Subscribe to this journal
Receive 12 print issues and online access
$209.00 per year
only $17.42 per issue
Buy this article
- Purchase on Springer Link
- Instant access to full article PDF
Prices may be subject to local taxes which are calculated during checkout
Similar content being viewed by others
References
Mendel, G. Versuche über Pflanzen-Hybriden. Vorgelegt in den Sitzungen 8 (1865).
Casso, D., Ramirez-Weber, F.A. & Kornberg, T.B. GFP-tagged balancer chromosomes for Drosophila melanogaster. Mech. Dev. 88, 229–232 (1999).
Melamed, M.R., Lindmo, T.L. & Mendelsohn, M.L. Flow cytometry and sorting. (Wiley-Liss, New York, NY; 1990).
Shapiro, H.M. Practical flow cytometry. (Wiley-Liss, New York, NY; 1995).
Crosland-Taylor, P.J. A device for counting small particles suspended in a fluid through a tube. Nature 171, 37–38 (1953).
Ashburner, M. Drosophila, a laboratory handbook. (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY; 1989).
Wood, W.B. The nematode Caenorhabditis elegans. (Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY; 1988).
Thisse, B., el Messal, M. & Perrin-Schmitt, F. The twist gene: isolation of a Drosophila zygotic gene necessary for the establishment of dorsoventral pattern. Nucleic Acids Res. 15, 3439–3453 (1987).
Reuter, R. & Leptin, M. Interacting functions of snail, twist and huckebein during the early development of germ layers in Drosophila. Development 120, 1137–1150 (1994).
Goldstein, L.S. & Fyrberg, E.A. Practical uses in cell and molecular biology. (Academic Press, San Diego, CA; 1994).
Ellenberg, J., Lippincott-Schwartz, J. & Presley, J.F. Dual-colour imaging with GFP variants. Trends Cell Biol. 9, 52–56 (1999).
Acknowledgements
We are grateful to Dr. Stephen Smith for his suggestions and discussions on the optics used in the machine. We thank Dr. Roel Nusse for his careful reading of the manuscript, and Dr. Allan Spradling for advice. The C. elegans larvae were kindly provided by Drs. Peter J. Roy and Stuart Kim. E.F. was supported by a European Molecular Biology Organization fellowship and a Stanford Berry Fellowship. The research was supported by the Howard Hughes Medical Institute and DARPA grant number N00014-98-1-0689.
Author information
Authors and Affiliations
Corresponding author
Supplementary information
Introduction: Embryo Sorter Website (Eileen Furlong, David Profitt, Matthew Scott)
Welcome to the Embryo Sorter Web Site!
The information presented here describes how we built a device that can sort Drosophila embryos based on their levels of fluorescence. A number of applications for this machine as well as a validation of its performance are described in the accompanying paper.
Overview
The machine functions by determining the levels of fluorescence of embryos that are suspended in a buffer. Solution is continuously pumped through the machine. Embryos randomly enter this fluid stream and pass through an optical cuvette (see below) where they are excited by an argon laser. The emitted light from the embryos is detected by two PhotoMultipler Tubes (PMT’s). The PMT’s send a signal to a microcontroller which signals to the computer to draw a peak for both the GFP and the autofluorescence. A microcontroller determines whether the peaks are above or below a defined threshold and sends a signal to an electromagnetic switch to divert the fluid flow to either a waste or save position.
The purpose of this supplemental web site is to provide information on how we built the machine in order to help other people who wish to build an embryo sorter. This project was initiated about 2.5 years ago and after 2 years we had a working prototype that could accurately sort Drosophila embryos. We then made several modifications and improvements to this prototype which resulted in the second generation embryo sorter that is described here. Because the construction was an on-going process incorporating many different adjustments we do not have blue-print plans for the machine. What this site provides is a parts list for the commercially available items and also both schematic drawings and photographs of the custom made parts. Navigation to different links on this site can be made by selecting from the bars on the top left hand side of the page.
There are a number of improvements that could be made in the next generation machine. For example making a lighter weight switch will increase the speed of sorting. Currently all of the embryo timing and sorting decisions are being made through hardware and we are sorting based upon the peak values for GFP. Doing these calculations through software would allow sorting to be done based on the integrated area under the curve. This may become necessary when sorting embryos that express GFP only in a small number of cells. We would appreciate if you could let us know of any modifications or improvements that you have made to the design of the instrument while making your own.
In order to build this device extensive experience is required in electronics (including microcontrollers), basic programming, optics, and machining.
This technology is exclusively licensed by Stanford Universityfor commercial purposes to Union BioMetrica Inc.
Embryo Chambers
In order to maintain a constant sorting rate (embryos/sec.), we used a two-chamber system, a high-density embryo chamber, and a low-density embryo chamber.
Collected embryos are initially added to the high-density chamber. Solution is pumped into the low-density chamber (blue arrows) and then moves through the exit tube to the optical cuvette. If the sorting rate drops below a defined threshold, the computer sends a signal to a fluid valve. This fluid valve then diverts the fluid flow to the high-density chamber (black arrow). Fluid will then leave this chamber and enter into the low-density chamber, resulting in the addition of embryos to the low-density chamber. When the rate of embryo sorting increases to the defined threshold, the computer will send a signal back to the valve and re-direct the fluid flow to the low-density embryo chamber. This will stop the addition of more embryos. The sorting switch that we are currently using can open and close in ~10 ms, giving a theoretical sorting speed of approximately 100 embryos per second. However, we have empirically determined the optimal sorting rate to be 15 embryos per second under present conditions.
The dimensions of the chambers are as follows:
All fluid flow tubing is micro-bore Polyvinylchloride with an inside diameter of 0.030".
Optics Cuvette
The cuvette serves several functions: 1) to isolate the fluid system from the optical system as much as possible, 2) to deliver embryos to the excitation light in a single file fashion, and 3) to emit embryos from the system in a fluid stream that is easily diverted. From the beginning it became apparent that occasional plugging would be unavoidable. For this reason having the cuvette easily removable is also of benefit. The dimensions of the optical cuvette are shown below.
In order to achieve the goal of delivering the embryos in a single file manner, the inside dimensions of the delivery tube must be such that two embryos cannot exist side by side. Tubing with such small dimensions causes high resistance to fluid flows, necessitating either high pressures or excessively slow flow rates. We overcame this problem by using relatively large diameter tubing for the collection and transfer of the embryos (0.030" I.D.), and transitioning gradually to a smaller dimension glass detection tube (0.4 mm I.D. square glass tube). This gradual transition allows the embryos to accelerate before entering the much faster fluid flow of the small glass tube. This approach dramatically reduced the frequency of clogging. The transition was fabricated by making an epoxy cast over a stylus tapered to the inside dimensions of the glass tube.
Panel A shows the mold used to make the epoxy junction between the tubing and the square glass tube. The fine end of the stylus is initially placed inside the square glass tube (panel B). The two halves of the mold are then assembled and epoxy resin is injected into the mold. Once hardened the stylus was removed, leaving the glass tube in place. Secondly, a tube was inserted in place of the stylus and sealed with an adhesive.
This wide diameter tubing-0.4 mm diameter glass tube-epoxy assembly was placed inside a brass ‘cuvette’ (panel C) and glued into place. (Note that we used brass for this purpose due to the ease of machining brass. Manufacturing this piece from a non-corrosive material would increase the life span of the cuvette). The cuvette has two windows perpendicular to each other. One window allows the laser light to excite the embryos as they pass through the square tube and the second window allows the emitted light from the embryos to be detected by the PMTs. The windows are covered by 12 mm #1 coverslips (panel D). The inside of the cuvette, surrounding the square glass tube, was filled with immersion oil using a small opening on the other side of the cuvette (panel E).
Jacketing the glass tube with immersion oil has several advantages. Because the indices of refraction at the excitation light and glass tube interface are nearly equal, excitation light "scatter" is reduced. Also, the oil has the effect of reducing the angle at which the emission light exits the glass tube. This allows the use of a smaller light collection lens, at a distance farther from the point of excitation.
When the fluid reaches the end of the glass tube (the glass air interface) it puddles until a drip is formed (panel F). The embryos entering this drip then mix, defeating our goal of sorting. To prevent this puddling, the flow rate is increased until the fluid exits the system in a continuous stream. Panel G is an end-on view of the optics cuvette, showing the central opening where the embryos exit. A metal ring (Fig 2 in paper, see also diagram above) is firmly attached onto the end of the cuvette via 4 screws. The ring injects fluid from a second peristaltic pump into the embryo stream as it exits the glass tube, resulting in a high velocity stream of fluid exiting its orifice. Increasing the fluid flow after the point of detection has two advantages: 1) it allows the embryos to travel at a much slower rate through the point of detection and 2) it allows the sorting switch to operates at moderate speeds, greatly simplifying its design.
Optics Chamber
The optics chamber (or box) contains the dichroic mirrors, filters (excluding the laser clean-up filter), and light detection elements. The optics cuvette mounts directly to the front of the box and is held in place by a clamping block. This block is light tight, allowing only light from the laser to enter the box via a telescoping tube. The schematic diagram shows the direction of the light path from the optics cuvette (drawn in red shading) to the PMTs. The relative distances between each of the components are indicated.
The optics box, cuvette clamp, filter holders, and dichroic mirror mounts are fabricated from black delren for ease of machining and opaqueness. Photographs of the optics box are shown below where red arrows indicate the direction of the light path (panel A) and the direction of fluid flow is shown with yellow arrows. The box assembly is mounted on standard ball slides, allowing alignment only in the horizontal direction. The laser (hidden behind a white plastic partition that acts as a splash guard) is mounted perpendicular to the box on adjustable legs, allowing alignment only in the vertical direction. The laser clean-up filter (488/10 nm) is housed in a telescoping tube assembly mounted directly on the front of the laser. No beam steering optics are required.
The alignment procedure begins by fixing the height of the laser to the optical axis of the box. Then, with the optics cuvette mounted to the box, the box assembly is adjusted until the collection tube is in the center of the laser beam. With the dichroic mirrors removed, the lenses are aligned until a projection of the collection tube is formed on the front surfaces of the PMT’s. In our present configuration, the projected image is about sixteen times larger than that of the tube I.D. Filling the collection tube with a diffusing liquid, such as diluted milk, and placing a piece of white paper in front of the PMTs, is helpful in observing the projected image. Note that a part number for the light collection lens has been omitted from our parts list. The lens we are using came from a colleague who unfortunately does not have that information. However, the lens has the physical appearance and properties of #01LAG000, Melles Griot Optics.
Some readers more familiar with optics have probably noticed the absence of any stop or slit in the image plane. A crude slit was fabricated for this project, but found to be difficult to align, and not very effective.
Sorting Switch
After detection, embryos leave the optics cuvette in a fluid stream. Embryo sorting is accomplished by diverting the fluid stream into a conduit leading to a collection vessel. Because the fluid stream is in physical contact with the optical system, it is easier to sort the embryos by moving the collection conduit, rather than the fluid stream. The conduit is composed of two tubes (a waste tube and a save tube) separated by a very thin central wall (membrane). Sorting is achieved by moving the appropriate tube under the fluid stream.
Contained within the conduit assembly are rare-earth neodymium super-magnets. This assembly is in turn suspended between two electromagnets. Applying electric current to the electromagnets exerts force on the suspended magnet in one direction, moving the conduit. Reversing the current produces a force in the opposite direction.
In fabricating this type of switch it is very important that the electromagnets be uniformly placed on both sides of the center rare-earth magnet. If one of the iron core electromagnets is closer to the center than its opposite, it will have a stronger attraction to the center magnet. This results in the switch having a preference to stay in one position, and/or operating faster in one direction than the other. To operate the switch, a current pulse of ~2A, for a duration of 5ms, is applied to the electromagnets causing the switch to move from the closed position to the open position. A similar pulse of the opposite polarity will return the switch to the closed position. A well-balanced switch should come to rest equally in the open or the closed position when no power is applied. The embryo and solution flow is always aligned to the waste tube, which is the default "closed" position.
The time it takes to move an object is related to its mass (F=ma), and to the distance it must travel (d=vt). A good switch should also have the following properties: 1) a lightweight moving element or conduit, 2) a small distance of travel to direct the fluid to either the waste or save tube and 3) little resistance to movement. The minimum distance the conduit must travel is the width of the fluid stream plus the thickness of the separating membrane. Therefore using a very thin membrane will reduce the distance that the switch has to travel and also reduces the amount of splatter produced as the membrane passes under the fluid stream. Another property to consider is the distance from where the fluid stream leaves the cuvette to the switch membrane. Very long distances will affect system timing. Very short distances may cause the switch to come into contact with the cuvette, causing friction and/or erratic action. Our design maintains a clearance of ~0.005" via a micrometer adjustment. The entire assembly is modular (including the switching conduits, collection vessels and fluid re-circulation tray) facilitating clean up and maintenance.
We fabricated the conduit from polystyrene (panel 1-4) and suspended it using a pair of bearings. The center membrane dividing the save and waste tubes was cut from 0.002" Mylar sheeting. Panel 3 shows a top view of the two tubes. As seen in panel 3, the disadvantage of a thin membrane is that it is easily deformed (although this has not had any effect on the sorting accuracy). Thin wall plastic tubing was inserted into the end of the conduit on either side of the thin membrane (see * panel 4). These tubes direct the fluid flow into the collection tubes.
Assembly
In order to protect the laser from accidental splashes or spills a splash guard was placed between the laser and the optics chamber. The white splash guard and base were made from Corian counter top material. Though somewhat unorthodox, this plastic is very dense and therefore stable. It also is machineable, bondable, and non-corrosive. Small quantities are usually obtainable from a local cabinet shop, sold as sink cutouts. A photograph of the assembled machine is shown below. The yellow arrow represents the direction of embryo and fluid flow.
The embryo chambers and magnetic stirrer are located to the front to allow easy access to the user. Note that the PMT’s are located at the back of the optics chamber, to the rear of the assembly, which protects them from accidental knocks that could disrupt their alignment.
Electronics
Two methods of measuring light intensity are used in this design. The first, a photo diode, is used to measure relatively high levels of light. This wavelength is primarily at the excitation wavelength, and is used to detect embryos as they pass through the detection optics. While this signal intensity may vary from embryo to embryo, the information contained in its amplitude is of little significance. This signal is amplified at a fixed gain. The other method of measuring light intensity uses photomultiplier tubes (PMTs). The sensitivity of a PMT is directly proportional to the voltage applied to its cathode. Therefore, the gain of these signals is adjusted by varying this voltage. Both detectors use a current to voltage conversion circuit similar to the one shown below.
Signal scaling for both the PMTs and the photo diode are achieved through attenuation. This is easily accomplished using a potentiometer as a voltage divider. This approach has two advantages. First, the relative signal amplitude is directly proportional to the position of the panel indicator or knob. If one uses a multi-turn potentiometer and a counting dial, this value is directly readable. Secondly, because the gain of each amplifier is fixed, and because the frequency characteristics of an amplifier are related to its gain, overall system linearity and reproducibility are improved. As in many optical measuring systems, there are several sources of DC errors in the output signals. The primary sources are unwanted light reaching the detectors, and leakage current from the detectors themselves. In the case of PMTs, this leakage current error can vary significantly with tube voltage. To compensate for these errors, a DC offset adjustment is included in the circuit - see diagram below.
In this embryo sorting system, the signal from one of the PMTs is used to detect a very narrow band of light that represents fluorescence from a known source. However, light from other sources (autofluorescence) may also be present at these wavelengths, and may be detected by the PMT. To compensate for this, light is measured at wavelengths close to those of the known source, and is used to represent the relative intensity of the autofluorescence. A percentage of this autofluorscence signal is then subtracted from the fluorescence signal. By proper scaling of these signals, a more accurate representation of fluorescence from the known source are obtained. The diagram below illustrates a simple circuit that accomplishes this. This circuit also AC couples the signals to eliminate any residual DC offsets.
Since the objective of this device is to sort embryos based on the presence or absence of the fluorescence signal, some means of determining the position of the embryo is necessary. This is accomplished by determining the peak intensity of the photo diode signal as the embryo passes through the optics. The photo diode signal is used because its amplitude is relatively constant, and is less affected by the presence or absence of fluorescence. Once photo diode detects a peak, the magnitude of the adjusted fluorescence signal from the PMTs is compared to that of a threshold signal. If the system is set to save embryos that emit fluorescence, and this signal is greater than the threshold signal, two signals are sent: a new embryo (New) signal and a save embryo (Save) signal. Conversely, if the system is set to save embryos that do not emit fluorescence and this signal is greater than the threshold signal, only the New signal is output.
A dedicated microcontroller is used to produce the New and Save signals. The controller continuously measures the diode signal at 40 micro-second intervals. When the controller determines that the signal is negative it measures the fluorescence signal and then the threshold signal. At this time it reads the save above/below threshold input and generates the appropriate outputs. This cycle is repeated continuously as long as power is applied to the controller. It should be noted that the fluorescence signal is measured about 40 micro-seconds after the peak has passed, and that the outputs are generated at least 80 micro-seconds later. Using a microcontroller with an eight-bit analog to digital converter limits the measurement resolution to ~ 20 mV, assuming a maximum signal of 5 V. Thus, the signal must be at least 20 mV below its peak before the controller recognizes it. Our system uses a PIC16F84 microcontroller, Microchip Technologies, with an internal 8-bit analog to digital converter, operating at 5 V.
System Timing
The New and Save signals are in turn used by another microcontroller, dedicated to timing the operation of the sorting switch. For any given flow rate through the system, there is a predictable time for the embryo to travel from its detection point to the outlet. Additionally, more than one embryo may occupy this space, requiring the switch to predict is operation. There are four operator-defined parameters input to the controller used in making this prediction.
-
a
Delay to open. This delays the switch from opening at the detection of a New & Save condition, allowing any potential unwanted embryos time to exit the switch.
-
b)
Open duration. This specifies how long the switch should stay open for a New & Save condition, allowing time for the embryo to pass through the switch into a collection conduit. This time begins at the end of the delay to open time.
-
c
Delay to close. This delays the switch from closing at the detection of a New & Save-not condition, allowing any potential wanted embryos to pass through the switch.
-
d
Close duration. This specifies the minimum time the switch must stay closed at the detection of a New & Save-not condition, and begins at the end of the delay to close time.
A delay to open or delay to close timer is initiated for every embryo detected. The switch position requests for each embryo are then polled, giving priority to non-save requests. At the conclusion of any Save condition the switch defaults back to the close (non-save) position.
Computer Display
The computer receives inputs from the electronics via an analog/digital computer board. We have used LabView to graphically display the following:
-
1
A graph showing the amplified signals from PMT1 and PMT2.
-
2
A graph showing the scaled signals for PMT1 (GFP) and PMT2. PMT2 signal indicates the amount of autofluorescence that is being subtracted from the PMT1 signal. (This subtraction is due to the overlap in the wavelengths of the autofluorescence to GFP).
-
3
An adjusted graph which shows a GFP peak (this is the scaled GFP signal minus some of the signal from PMT2 (graph 2), which for our conditions is usually about 25%), and the threshold amplitude at which sorting will take place.
This is a movie that shows the GFP peaks while sorting embryos. Each peak is the adjusted GFP signal for an embryo. The horizontal yellow line is the threshold level below which the machine is saving embryos. (We also have the option to save above this threshold.) To the right hand side of the graph there are a number of parameters being recorded; 1) the total number of embryos that have been detected, 2) the number of Save pulses that the microcontroller has sent to the switch, 3) the percentage of saved embryos, 4) the threshold level (i.e. a quantitative value for where the threshold bar is) and 5) the power output (mW) from the laser. During this movie the threshold bar is being raised which results in more embryos being saved. Note that the percentage of saved embryos subsequently increased as the threshold bar is being raised. The clicking sound is made by the switch. Each time the switch opens and closes it hits a stop barrier in each direction making a clicking sound. Therefore each time an embryo peak is below the threshold the switch will move and therefore click. Unfortunately the movie has lost some quality during the compression program to make it web compatible.
Parts List
Contacts
Eileen Furlong http://cmgm.stanford.edu/~furlong/
David Profitt http://www.stanford.edu/~profitt/
Matthew Scott http://scottlab.stanford.edu/
Rights and permissions
About this article
Cite this article
Furlong, E., Profitt, D. & Scott, M. Automated sorting of live transgenic embryos. Nat Biotechnol 19, 153–156 (2001). https://doi.org/10.1038/84422
Received:
Accepted:
Issue Date:
DOI: https://doi.org/10.1038/84422
This article is cited by
-
A high throughput method for egg size measurement in Drosophila
Scientific Reports (2023)
-
Microfluidics for understanding model organisms
Nature Communications (2022)
-
Tools to reverse-engineer multicellular systems: case studies using the fruit fly
Journal of Biological Engineering (2019)
-
Microsystems for controlled genetic perturbation of live Drosophila embryos: RNA interference, development robustness and drug screening
Microfluidics and Nanofluidics (2009)
-
Drosophila under the lens: imaging from chromosomes to whole embryos
Chromosome Research (2006)
