WO2009039284A1 - Systems and methods for high-throughput detection and sorting - Google Patents

Abstract

The various embodiments of the present invention relate generally to devices, systems and methods of high-throughput detection and sorting. More particularly, the various embodiments of the present invention are directed to microfluidic systems and methods of high-resolution imaging and high-throughput sorting. Some embodiments of the devices, systems and methods of the present invention are directed toward a single pass microfluidic device for individually detecting and sorting a plurality of multicellular organisms, such as Caenorhabditis elegans, having at least one phenotype, wherein the system has an accuracy greater than about 95%.

Description

SYSTEMS AND METHODS FOR HIGH-THROUGHPUT DETECTION AND SORTING

RELATED APPLICATIONS

This application claims, under 35 U. S. C. § 119(e), the benefit of U.S. Provisional Application Serial No. 60/973,191, filed 18 September 2007, the entire contents and substance of which are hereby incorporated by reference as if fully set forth below.

STATEMENT OF FEDERALLY SPONSORED RESEARCH

This invention was made with U.S. Government support under Grant No. NS058465 awarded by the National Institutes of Health and Grant No. DBI-0649833 awarded by the National Science Foundation. The U.S. Government has certain rights in the invention.

TECHNICAL FIELD

The various embodiments of the present disclosure relate generally to devices, systems, and methods of high-throughput detection and sorting. More particularly, the various embodiments of the present invention are directed to microfluidic systems and methods of high-resolution imaging and high-throughput sorting.

BACKGROUND OF THE INVENTION

Small multicellular organisms, such as Caenorhabditis elegans, Danio rerio, or Drosophila melanogaster are often used as model systems for providing new insights into genetics, developmental biology, disease, and drug discovery. For example, human gene homologs have been identified in these model organisms. Mutations in these homologs often result in observable phenotypic changes in the model organism and provide new biological insights into understanding disease states, cell lineage, neurobiology, cell death, and cancer, among others. The modeling of diseases in multicellular organisms involves the generation of morphological or behavioral mutants with observable phenotypes. In many cases, these morphological or behavioral mutants are created to replicate human disease states. Researchers then observe these model organisms and their interaction with candidate therapeutics for in vivo screens of libraries of pharmacological compounds. Researchers can utilize these organisms and mutants thereof to identify novel compounds and cellular and molecular targets for drug intervention.

The soil nematode, C. elegans, has become a particularly important multicellular organism for this type of research. C. elegans is a small roundworm that has a generation time of about three days, which permits the rapid accumulation of large quantities of individual worms. Among multicellular organisms, C. elegans has unique advantages. C. elegans is extremely amenable to genetic approaches because its genome, anatomy, development, and behavior have been extensively studied, and a large collection of mutants have been isolated that are defective in embryonic development, behavior, morphology, and neurobiology, among others. Many knockout mutants are available for about 19,000 predicted genes, and the discovery and availability of RNA interference ("RNAi") gene knockdowns has facilitated a broad range of genetic studies. Not only is its genome fully sequenced and well-studied, but its small size (length of about 1 mm and a width of about 60 μm) also allows for complete anatomical description of the animal, including a complete synaptic wiring diagram of the 302 neurons. From sensory stimulus to behavior, C. elegans provides a unique opportunity to study and define neuronal mechanisms. Its transparent body and nearly invariant cell lineage coupled with fluorescent protein technology, enable precise cell-by-cell analysis of biological phenomena throughout the development of C. elegans. For example, the cell-lineage of C. elegans is fixed, allowing identification of each cell, which has the same position and developmental potential in each individual animal (e.g., muscle, gut, neuron, etc.).

Although there are important physiological differences between nematodes and mammals, the conservation of many genes and fundamental cellular processes between nematodes and mammals make C. elegans an attractive organism for use in drug screening studies. Many biotechnology companies, as well as pharmaceutical companies, now employ C. elegans in their drug discovery processes. The above unique advantages of C. elegans combined with high-throughput genetic tools make C. elegans readily adaptable for automation and high-throughput experiments, such as pharmaceutical compound screens useful in the identification and development of potential candidate drugs.

Despite the potential power of using multicellular organisms for rapid pharmaceutical drug discovery, conventional experimental methods, such as in vivo microscopy, visual phenotyping, and laser ablation techniques, require labor intensive and time-consuming manual handling of the organisms, which significantly limits the experimental throughput and in some cases precludes experiments all together.

In vivo microscopy is an important tool for biological studies. It is becoming more important to image large samples with high resolution as large-scale genetic studies are becoming feasible due to recent progress in genetic technologies. For example, the completed genome sequence, coupled with the discovery of RNAi and the development of fluorescent proteins enables global studies of gene function and expression in C. elegans. This kind of genome-wide analysis would benefit greatly from single-cell resolution microscopy. To elucidate the role of individual genes, gene screen methods, such as mutagenesis and RNAi, are used and often require phenotype identification of tens of thousands of individuals with high magnification microscopy. These screens require animals to be immobilized by anesthetics, which can disrupt neuronal signaling and induce undesirable physiological changes. In addition, current microscopy requires a technician to mount samples, record the orientation and position of the animals on the slide, examine each individual, and rescue the mutant of interest by sliding the coverslip off and transferring the animal back into a culture dish. This painstaking manual handling of organisms not only significantly limits the experimental throughput, but also increases noise in the experiments due to variation in sample-to-sample handling. The manual handling of samples also remains as a bottleneck in the laser ablation of cells, which uses a similar microscopy technique, but the technician uses a focused laser to ablate cells as opposed to perform phenotype identification. Therefore, for both high resolution microscopy and laser ablation, automated sample handling and digital image analysis will greatly improve experimental throughput and data quality. Accordingly, there is a need for automated systems and methods of high- throughput detection and sorting. It is to the provision of such automated systems and methods of high-throughput detection and sorting that the various embodiments of the present invention are directed.

SUMMARY The various embodiments of the present disclosure relate generally to devices, systems, and methods of high-throughput detection and sorting. More particularly, the various embodiments of the present invention are directed to microfluidic devices, systems, and methods of high-resolution imaging and high-throughput sorting.

Broadly described, an aspect of the present invention comprises a system, comprising a device for individually detecting and sorting a plurality of multicellular organisms having at least one phenotype at cellular resolution. In an embodiment of the present invention, the device can be a single pass device or a multipass device. In another embodiment of the present invention, the system further comprises an immobilization system. The system can be configured for the individual detection and sorting a plurality of multicellular organisms, wherein the multicellular organism is Caenorhabditis elegans.

In an embodiment of the present invention, the system can further comprise a cooling system.

An aspect of the present invention comprises a detection system, comprising: a detection environment comprising at least one inlet and at least one outlet; a loading element, wherein the loading element in fluid communication with at least one inlet of the detection environment, the loading element adapted to a load one sample object into the detection environment; an immobilization system, wherein the immobilization system is in operational communication with the detection environment; and a detector, wherein the detector can detect a phenotype of a sample object located in the detection environment. In an embodiment of the present invention, the detection system cam further comprise a container comprising a fluid and a plurality of sample objects having at least one phenotype, wherein the container is in fluid communication with at least one inlet of the at least one inlet of the detection environment. In another embodiment of the present invention, the detection system can further comprise at least one unloading element, wherein the unloading element is in fluid communication with the at least one outlet of the detection chamber, wherein the unloading element in operational communication with a detector. In yet another embodiment of the present invention, the detection system can further comprise a control system, wherein the control system receives a signal from the detector and controls the loading element, the unloading element, and the immobilization system. In an embodiment of the present invention, the immobilization system can comprise a cooling system or at least one restraining element. In some embodiment of the present invention, the sample object can comprise a unicellular or multicellular object. In an embodiment of the present invention, the sample object can comprise Caenorhabditis elegans. Another feature of the present invention is a system comprising a microfluidic system. The system can comprise at least one restraining element, wherein the at least one restraining element is pressure-based restraining element, such as a valve or a suction element.

An aspect of the present invention comprises a microfluidic device, the device comprising: a detection environment comprising at least one inlet and at least one outlet; a loading element, wherein the loading element is in fluid communication with at least one inlet of the at least one inlet of the detection chamber, the loading element adapted to a load a sample object into the detection chamber; at least one immobilization element, wherein the at least one immobilization element is in operational communication with the detection environment; an unloading element, wherein the unloading element is in fluid communication with the at least one outlet of the detection environment. In an embodiment of the present invention, at least one of the at least one immobilization element comprises a cooling element or at least one restraining element. In some embodiments of the present invention, the sample object can comprise a unicellular or multicellular object. In an embodiment of the present invention, the multicellular object can comprise Caenorhabditis elegans. The device can comprise at least one restraining element, wherein the at least one restraining element is pressure-based restraining element, such as a valve or a suction element.

An aspect of the present invention comprises a method for detecting at least one phenotype of an object, the method comprising: loading a microfluid comprising a single object from a fluid comprising a plurality of objects into a microfluidic device; immobilizing the object; and detecting the phenotype of the object. In an embodiment of the present invention, immobilizing the object can comprise restraining the object or cooling the microfluid. In an embodiment of the present invention, the method can further comprise unloading the object from the microfluidic device. In an embodiment of the present invention, the method can further comprise repeating the loading, the immobilizing, the detecting, and the releasing at least once. In another embodiment of the present invention, unloading the object from the microfluidic device can comprise sorting the object. In some embodiments of the present invention, the object can comprise a unicellular or multicellular object. In an exemplary embodiment of the present invention, the multicellular object can comprise Caenorhabditis elegans. In an embodiment of the present invention, the method is automated. Other aspects and features of embodiments of the present invention will become apparent to those of ordinary skill in the art, upon reviewing the following description of specific, exemplary embodiments of the present invention in conjunction with the accompanying figures.

BRIEF DESCIPTION OF DRAWINGS

Figure 1 is a schematic of the microsystem functioning in rapid imaging, phenotyping, and sorting of a mixed population of animals based on cellular and subcellular phenotypes.

Figures 2A-B are optical micrographs of the central region of the microfluidic device showing device components.

Figure 3A is a schematic illustration the process of fabricating a PDMS microfluidic device.

Figure 3B is an exploded view of a schematic illustration showing the individual lithography layers for a PDMS microfluidic device. Figure 3C is a schematic of a cross- sectional view of a valve.

Figure 4 illustrates the pressure profile in the detection environment.

Figures 5A-D are schematic diagrams demonstrating the valve control sequence in the worm sorting process.

Figures 5E-G are micrographs showing automated imaging and sorting sequence. Figure 6 is a diagram showing the on-chip and off-chip components and features of the microfluidic system.

Figure 7 is a schematic of a microfluidic device having two detection environments. Figure 8A is a schematic of a fluorescent Caenorhabditis elegans expressing AQR and PQR.

Figures 8B-M show automated gene expression pattern analysis in the microchip.

Figure 8N is a graph of the percentage of animals with each of the possible expression patterns in > 1,000 animals. Figure 8O is a histogram of animal loading time into the detection environment.

Figures 9A-D is a schematic representation of automated image processing and a decision-making process to sort animals at a cellular resolution.

Figures 10A-H illustrate automated high-throughput imaging and sorting based on synaptic phenotypes. Figures 101 -J compares the images of a worm imaged in the presence and absence of cooling.

Figure 1OK demonstrates the puncta structure of the nerve cord in a mutant animal expressing punc-25-YFP::RAB-5 before significant photobleaching (top) and quantification of puncta fluorescence from line scans with different amounts of photobleaching (bottom).

Figures HA-B show a computer-control, computer-enhanced image processing for a fast screen of C. elegans.

Figures 12A-L demonstrates computer-assisted phenotyping to identify mutants of interest.

DETAILED DESCRIPTION

Multicellular organisms, such as C. elegans and D. melanogaster, are important genetic models for studying developmental biology, physiology, and disease. In recent years, fully sequenced genomes and techniques that interrogate the functions of genes on large scales, such as protein microarrays, nucleic acid microarrays, and RNAi knockdowns, have become prevalent and important in these model organisms because of the high- throughput nature of these methods. Yet, important techniques for phenotyping, such as microscopy, are still largely limited in their manual modes of operation, making them not only low in throughput, but also prone to human biases and errors. For example, in vivo microscopy can be used for characterizing morphology and gene expression patterns of cells and tissues and for visualizing expression patterns, localization, synthesis, and degradation of molecules. This type of manual microscopy screen usually takes many months to perform, is very labor-intensive, and the phenotypes are usually qualitative. This creates a bottleneck for performing genetic analysis as phenotyping limits the speed of discovering new biological mechanisms and pharmacological compounds..

Microfluidics lends itself to solving some of these problems. As used herein, the term "microfluidic" and derivatives thereof refer to systems and methods of manipulation of small amounts of fluids (about 10^~9 to about 10^~18 liters) using channels with dimensions of a few to thousands of micrometers (about 1 μm to about 2000 μm). Microfluidic systems have distinctive physical characteristics compared to macroscopic systems. In a microchannel, when two fluid streams come together, the flow is laminar (usually at very low Reynolds number, for example, less than about 10) and the dominant mixing mechanism is the result of diffusion of molecules across the interface between the fluids. This unique behavior of liquids at the microscale allows for greater control of the concentration of chemicals, culturing environment of cells, and even multicellular organisms. A large surface-area-to-volume ratio and small thermal mass facilitate rapid heat transfer in microfluidic systems and enable precise spatial-temporal temperature control. Additionally, many electrical components can be integrated on a chip having a microfluidic device and allow the microsystems to perform complex functions. Thus, the scale of microfluidic systems matches that of small organisms, cells, and macromolecules.

The term "fluid" is used herein for convenience and refers generally to many fluids, liquids, gases, solutions, suspensions, gels, dispersions, emulsions, vapors, flowable materials, multiphase materials, or combinations thereof. A fluid can comprise a mixture of a plurality of fluids. The term "plurality" as used herein refers to more than one. In an exemplary embodiment, a fluid is a culture medium, a biologically buffered solution, a salt solution, or the like. Various embodiments of the present invention are directed to automated, microfluidic systems and methods of high-throughput detection of cells and multicellular organisms. Some embodiments of the present invention permit high-throughput detection of a plurality of multicellular organisms through a single-pass through the system of the present invention. Some embodiments of the present invention are directed to automated, microfluidic systems and methods of high-throughput detection and sorting of cells and multicellular organisms. Some embodiments of the present invention permit the high throughput analysis and high-resolution 3D imaging of multicellular organisms.

An embodiment of an automated microfluidic system comprises image processing capabilities for performing high -resolution, high-throughput imaging, phenotyping, and sorting of cells and multicellular organisms. From single-cell assays to cell sorting, many microdevices have been developed, which have revolutionized the throughput of experiments in the area of single cell studies. The impact of such microdevices, however, has not yet been realized for multicellular organisms, which is attributable in part to the difficulties of handling live, moving multicellular organisms. To date, manual microscopy is the only way to obtain 3D images of multicellular organisms at subcellular resolution.

Microscopy applications are currently limited as they can be performed on only a limited number of animals, and the results obtained can be strongly affected by stochastic variations among individuals. Embodiments of the system and methods of the present invention can automate sample handling and image analysis, reducing experimental time and human intervention. This greatly improves throughput and data quality of experimentation by increasing the number of individuals that can be examined and decreasing the effect of environmental noise.

Some embodiments of the present invention contemplate a microsystem that can automatically process a population of cells or multicellular organisms by imaging the cells or multicellular organisms one at a time at cellular or subcellular optical resolution in two- dimensions or three-dimensions, processing the images, determining the phenotype of the cells or multicellular organisms, and sorting the cells or multicellular organisms according to the identified phenotype, without human intervention (Figure 1). According to some embodiments of the present invention, advantages may include but are not limited to: (1) automation; (2) resolution at single-cell and/or subcellular levels; (3) a throughput that is at least an order of magnitude faster than that by manual operation; (4) reduction of human bias and errors (5) reduction in photobleaching; (6) compatibility with many microscope and camera systems; (7) relative inexpensive nature of the setup; (8) applicability and scalability to many model organisms and cell types; (9) as self-regulated single organism loading scheme; (10) reduction in the use of anesthetics; and (11) on-line quantitative analysis for screens, sorting, and gene expression profiling

Referring now to the figures, wherein like reference numerals represent like parts throughout the several views, exemplary embodiments of the present invention will be described in detail. Throughout this description, various components may be identified as having specific values or parameters, however, these items are provided as exemplary embodiments. Indeed, the exemplary embodiments do not limit the various aspects and concepts of the present invention as many comparable parameters, sizes, ranges, and/or values may be implemented.

Various embodiments of the present invention are directed to automated systems, devices and methods for high-throughput detection and sorting of cells and multicellular organisms. An aspect of the present invention comprises a microfluidic device 100, the device comprising: a detection environment 105 comprising at least one inlet 110 and at least one outlet 115; a loading element 130 located upstream from the detection environment 105, wherein the loading element 130 is in fluid communication with at least one inlet 110 of the detection environment 105, the loading element 130 adapted to a load a sample object into the detection environment 105; at least one immobilization element, wherein the at least one immobilization element is in operational communication with the detection environment 105; and a unloading element 140 located downstream from the detection environment 105, wherein the unloading element 140 is in fluid communication with the detection environment 105 and the at least one outlet 115 of the detection environment 105.

Referring now to Figure 2A, an exemplary embodiment of the present invention comprises a microfluidic device 100, the device comprising: a detection environment 105 comprising at least one inlet 110 at least one outlet 115, and at least one restraining element 125; a loading element 130 located upstream from the detection environment 105, wherein the loading element 130 is in fluid communication with at least one inlet 110 of the detection environment 105, the loading element 130 adapted to a load a sample object into the detection environment 105; a cooling element 135, wherein the cooling element is in thermal communication with the detection environment 105; and a unloading element 140 located downstream from the detection environment 105, wherein the unloading element 140 is in fluid communication with the detection environment 105 and the at least one outlet 115 of the detection environment 105. A number of different cell types and multicellular organisms may be employed as sample objects in the systems and methods of the present invention. For example, the systems and methods of the present invention are applicable to many cell types (e.g., mammalian cells, tissue culture cells, among others) and multicellular organisms know in the art, including animals, such as insects, amphibians, and fish, plants, fungi, seeds, and the like. Specific organisms of interest include, but are not limited to the genera of Xenopus, Danio, Caenorhabditis, Drosophila, and the like. In addition, the embryos of many animals and plants can be employed in the systems and methods of the present invention. The multicellular organisms employed in the present invention may be at many stages of their life (e.g., in the larval stage, the adult stage, etc.). Throughout the present application, specific reference may be made to analysis of specific organisms or cells; however, such references are not intended to limit the scope of the invention as the systems and methods of the present invention are suitable and can be configured for the analysis of many types of cells and multicellular organisms.

The a micro fluidic device 100 can be made of many materials, including but not limited to poly(dimethylsiloxane) (PDMS), polyurethanes, polyimides, polysilanes, polysiloxanes, polysilazanes, and other elastomers known in the art. In an exemplary embodiment of the present invention, a microfluidic device 100 is made of PDMS. PDMS is optically transparent, so it can be used with conventional optical methods of detection. In addition, its biocompatibility makes this elastomer particularly suitable for work with living cells and organisms. PDMS is a uniquely suitable material for handling living organisms because of its low toxicity and high permeability to oxygen and carbon dioxide. Furthermore, fabrication of a multilayer device made of PDMS is relatively inexpensive, making this device disposable and allowing avoidance of contamination issues in the culture of organisms and cells. The microdevice can be made by many lithography techniques know in the art, including but not limited, to multilayer soft lithography. In an embodiment of the present invention, the microfluidic device 100 comprises at least two layers, the at least two layers comprising a worm loading layer and a valve control layer. In an exemplary embodiment of the present invention, the microfluidic device 100 comprises at least three layers, the at least three layers comprising a worm loading layer, a valve control layer, and a membrane layer. In some embodiments of the present invention the channels containing the cooling fluid can be integrated into the worm flow layer, the valve control layer, a separate cooling layer, or combinations thereof.

As illustrated in Figures 3A-B, two different molds (i.e., masters) can be fabricated by photolithographic processes to create a worm loading layer and a control layer. The mold for the worm loading layer can be made by a two-step photolithographic process. In an exemplary embodiment, a negative photoresist of about 1-100 μm can be spin-coated onto a substrate (e.g., silicon wafer) for the worm loading chamber and the detection environment. The loading element, side channels (i.e., restraining element), inlet, and outlet can then be fabricated with a layer of positive photoresist of about 10-100 μm on the same substrate. After the positive photoresist is developed, the substrate can be heated at about 125 ⁰C for about 5 min to allow the positive photoresist to reflow so that the channels form a substantially smooth and substantially rounded shape. The master for the control layer can be made of a layer of negative photoresist of about 10-100 μm on a substrate. The two molds and a blank substrate can be treated with tridecafluoro-1,1,2,2- tetrahydrooctyl-1-trichlorosilane vapor or the like to prevent adhesion of PDMS during the molding process.

For fabricating the control layer, PDMS can be poured onto the control-layer master to obtain a layer of about 5 mm in thickness. Mixture of PDMS and tetrahydrofuran (THF) in a 2:1 ratio can be spin-coated on a substrate to produce a thin layer having a thickness of about 20 μm to form a membrane. Both can be partially cured at about 70 ⁰C for about 20 min. The thick control layer can then be peeled off from the master and holes can be punched for access to the control and cooling channels. The control layer can then be bonded to the thin PDMS membrane on the substrate. This assembled control layer can be cured at about 70 ⁰C for about 2 hours. For the worm- loading layer, PDMS was spin-coated onto the master to give a layer having a thickness of about 60 μm. The worm-loading layer can be cured at about 70 ⁰C for about 2 hours and then can be peeled off from the master. The layer can be then turned up side down and bonded to the control layer using an oxygen plasma treatment or the like. Another set of holes can then be punched for access to the worm loading channel. These assembled layers can then bonded onto a suitable substrate, such as a glass, to form the microdevice.

PDMS is a soft material with Young's modulus of approximately 750 kPa, which can be deflected with small forces. Structures with a high aspect ratio, such as the worm loading chamber, are especially prone to deformation and storing energy when pressure is applied. Once the pressure is removed, the deformed PDMS slowly returns to the former state and releases the stored energy. This mechanical compliance of the device causes flow fluctuation in the detection environment and thereby disturbs a loaded worm. To reduce flow fluctuation in the detection environment, on-chip valves (e.g., loading elements, at least one restraining element, and unloading elements) using multilayer soft lithography were fabricated. The PDMS membrane between the worm loading layer and the control layer is engineered to be relatively thin (about 10 μm to about 20 μm). When pressure is applied to the channel in the valve actuation layer of the control layer, the membrane deflects toward the worm loading layer and obstructs or closes the channel. Embodiments of the microfluidic device 100 of the present invention can comprise a detection environment 105 comprising at least one inlet 110, at least one outlet 115, and at least one restraining element 125. In an embodiment of the present invention, a detection environment 105 can comprise one inlet 110. In an embodiment of the present invention, a detection environment 105 can comprise two, three, four, or more inlets 110. One of ordinary skill in the art would realize that the number of inlets into the detection environment is likely directly related to the number of cultures of cells or microorganisms desired to be analyzed. In an embodiment of the present invention, a detection environment 105 can comprise one outlet 115. In an embodiment of the present invention, a detection environment 105 can comprise two outlets 115. In an embodiment of the present invention, a detection environment 105 can comprise three, four or more outlets 115. One of ordinary skill in the art would realize that the number of outlets from the detection environment is likely directly related to the number of detectable characteristics or phenotypes desired to be sorted. One of ordinary skill in the art would also realize that a receptacle can be associated with the at least one outlet to capture sorted objects. The detection environment 105 comprises the portion of the microfluidic device

100 where the sample is detected. Thus, the detection environment 105 comprises a channel designed to accommodate a sample object. A person of ordinary skill in the art would realize that the features of the lithography masters determine the features and parameters of the microfluidic device. A person of ordinary skill in the art would also realize that the features and parameters of the microfluidic device can vary depending upon the specific application and the sample objects (e.g., cells or multicellular organisms) of interest. With regards to the systems and methods directed to multicellular organisms, a person of ordinary skill in the art would realize that the parameters of the channels can be customized to accommodate the shape and size of the sample. For example, in the case of C. elegans, the detection environment can comprise a substantially longitudinal channel having a diameter on about the same order as the width of a worm (e.g., about 25-30 μm for an L4 worm) to physically constrain the worm in the channel and restrict its mobility. In an embodiment of the present invention, the substantially longitudinal channel can have a diameter of about 5 μm to about 60 μm. In an exemplary embodiment of the present invention, the substantially longitudinal channel can have a diameter of about 15 μm to about 20 μm. Various embodiments of the present invention comprise a microfluidic device 100 comprising at least one immobilizing element. In an embodiment of the present invention, at least one of the at least one immobilizing element comprises at least one pressure-based restraining element 125. In an embodiment of the present invention, at least one of the at least one immobilizing element comprises a cooling element 135. The at least one restraining element 125 functions to physically restrain the sample in the detection environment and minimize movement of the sample (e.g., C. elegans). In some embodiments of the present invention, the at least one restraining element 125 comprises a pressure-based restraining element. In an embodiment of the present invention, the at least one restraining element 125 can comprise a suction element. In an embodiment of the present invention, the at least one restraining element 125 can comprise a plurality of suction elements. In some embodiments of the present invention, a plurality of suction elements can comprise a series of parallel channels forming a pillar array (also referred to as "side channels") with each channel separated by about 20 μm to about 50 μm. In an embodiment of the present invention, the at least one restraining element can comprise a valve (Figures 2B and 3C). As used herein, a "valve" is a device that regulates the flow of fluids and sample objects within fluids by opening, closing, or partially obstructing various passageways. In an embodiment of the present invention, the microfluidic device 100 comprises a cooling element 135, wherein the cooling element is in thermal communication with the detection environment 105. The cooling element 135 is capable of locally cooling the detection environment 105. Cooling the detection environment 105 thermally reduces the mobility of the sample and permits imaging at cellular and subcellular resolutions. In embodiments of the present invention, the detection environment 105 is cooled to about 4 ⁰C. In an exemplary embodiment, the cooling element 135 comprises a channel in thermal communication with the detection environment 105. The use of a cooling element 135 eliminates the need to use anesthetics to immobilize worms, which often disrupt neuronal signaling, induce undesirable physiological changes, and may be toxic to the organisms of interest.

In an embodiment of the present invention, the microfluidic device 100 comprises a loading element 130 located upstream from the detection environment 105, wherein the loading element 130 is in fluid communication with the at least one inlet 110 and the detection environment 105, the loading element 130 adapted to a load a sample object into the detection environment 105. The microfluidic device also comprises an unloading element 140 located downstream from the detection environment 105, wherein the unloading element 140 is in fluid communication with the detection environment 105 and the at least one outlet 115 of the detection environment 105. In an embodiment of the present invention, the loading element is a valve, and the unloading element is a valve.

The loading element 130 automatically self-regulates the loading of a single nematode into the detection environment 105 by the design of the loading scheme through operational communication with the unloading element 140. Multiple worms in the detection environment 105 can cause many problems, including but not limited to: (1) significant distortion of the shape and orientation of a worm of interest, which can affect visualization of the native morphology of the sample microorganism and can cause errors in image processing; (2) mistaken identification of samples of interest as fluorescence signals from other worms in the field of view can be mistakenly identified as the worm of interest, causing significant errors in sorting and laser ablation; and (3) aggregation of worms causes clogging of the channel. Once the device is clogged, the system must be manually stopped to apply high pressure to the device to flush out aggregated worms. In many cases, high pressures cause break-down of the devices. To avoid these problems, some embodiments of the present invention employ a self- regulated loading scheme. This scheme to load one worm at a time takes advantage of the squeezable body of the nematode and pressure drops created by a loaded worm (Figure 4). C. elegans is enclosed by an elastic cuticle layer. This layer is pushed outward by a high internal hydrostatic pressure relative to the ambient environment, which results in the nematode adopting a cylindrical structure. For a worm to pass into and through the detection environment that has a narrower channel width than the diameter of a worm, the elastic cuticle layer must be deformed against the hydrostatic pressure by the force generated by the pressure drop across the worm. When a worm is not loaded in the detection environment, the entire pressure drop occurs over the single worm in the detection environment, and this is great enough to deform the worm and push it into the detection environment. In contrast, once a worm is loaded into the detection environment, it drastically increases the hydrodynamic resistance of the loading channel. Now the pressure drop across a second worm in the channel upstream and outside the detection environment is too small to push the animal into the detection zone. Once a loaded worm leaves the detection environment, however, the pressure drop across the loading element then becomes large enough to push the second worm into the detection environment.

In an exemplary embodiment of the present invention, to load an animal into the detection environment, the at least one unloading element is closed while the side positioning channels remain open (Figure 5A), and a constant pressure source is used to drive the flow of a fluid containing the sample object into the microfluidic device. Self- regulation of loading (one at a time into the detection environment) plays a role in high- resolution imaging and accurate sorting. When an animal is present in the detection environment, the flow resistance is increased. The reduced flow rate lowers the pressure on a second animal at the loading element located at the entrance of the detection environment to a point where it is not sufficient to push the second animal into the detection environment (Figure 4). Upon releasing the first animal (by opening the unloading element), the pressure drop across the second animal becomes sufficient to push it into the detection environment. The sample-loading element design was implemented by controlling the pressure on a partially closed valve (Figure 3C). Thus, great flexibility is achieved depending on the size and the deformability of the animals for each application because of the ability to fine-tune the system pressure as well as the sample-loading regulator pressure. In an exemplary embodiment of the present invention, to position an animal inside the detection environment, pressure differences between the side channels and the entrance of the main channel can be utilized. Similar to the loading element, the side channels can also be controlled by the partially closed positioning of the valve. Once the animal's nose or tail is positioned at the end of the detection environment, the hydrodynamic resistance of the positioning channels self-equalizes. As a result, the animal stops moving in the direction of the flow (Figure 5B). The channels of the pillar array can be opened to generate a pressure gradient to guide an animal into the detection environment. This distribution of the pressure force minimizes mechanical stress on the animal. The animal is cooled, immobilized, and imaged (Figure 5C) before being phenotyped and sorted accordingly (Figure 5D). Figures 5E-G are optical micrograph showing automated imaging and sorting sequence: (E) loading nematode into the detection zone; (F) a loaded animal preventing a second animal from entering; and (G) the second animal is automatically moved into the detection zone after the previous animal exits the detection zone.

One advantage of the design of both the at least one restraining element and loading element in the main channel is that there are no permanent small features (i.e., < 20 μm). The dimension of the at least one restraining element and loading element is not from the mask design but instead from the partially closed valves, and therefore is tunable and can be expanded if necessary. Because pieces of debris smaller than the size of the nematodes cannot be easily filtered out, this design feature prevents clogging of the channels as the valves can be opened to remove the debris when necessary.

An aspect of the present invention comprises a system for high-throughput detection of a characteristic of a sample object, the system 200 comprising a detection environment 105 comprising at least one inlet 110 and at least one outlet 115; a loading element 130 located upstream from the detection environment 105, wherein the loading element 130 is in fluid communication with the at least one inlet 110 and the detection environment 105, the loading element 130 adapted to a load a sample object into the detection environment 105; at least one immobilization element, wherein the at least one immobilization element is in operational communication with the detection environment 105; an unloading element 140 located downstream from the detection environment 105, wherein the unloading element 140 is in fluid communication with the detection environment 105 and the at least one outlet 115 of the detection environment 105; a container 205 comprising a fluid 210 and a plurality of sample objects 215, wherein the container 205 is in fluid communication with at least one inlet of the at least one inlet 115 of the detection environment 105; a drive system 220 that drives the fluids 210 of the system 200; a detector 230 capable of detecting a characteristic of a sample object; and a control system 235 which receives at least one signal from the detection element 120 and controls the loading element 110, at least one immobilizing element, and unloading element 140.

Referring now to Figure 6, an exemplary embodiment of the present invention comprises a system for high-throughput detection of a characteristic of a sample object as illustrated in Figure 6, the system 200 comprising a detection environment 105 comprising at least one inlet 110 and at least one outlet 115; a loading element 130 located upstream from the detection environment 105, wherein the loading element 130 is in fluid communication with the at least one inlet 110 and the detection environment 105, the loading element 130 adapted to a load a sample object into the detection environment 105; a cooling element 135, wherein the cooling element is in thermal communication with the detection environment 105; a unloading element 140 located downstream from the detection environment 105, wherein the unloading element 140 is in fluid communication with the detection environment 105 and the at least one outlet 115 of the detection environment 105; a container 205 comprising a fluid 210 and a plurality of sample objects 215, wherein the container 205 is in fluid communication with at least one inlet of the at least one inlet 115 of the detection environment 105; a drive system 220 that drives the fluids 210A, 2210B, and 210C of the system 200; a cooling system 225, wherein the cooling system 225 is in thermal communication with the detection environment 105 via the cooling element 135; a detector 230 capable of detecting a characteristic of a sample object; and a control system 235 which receives at least one signal from the detection element 120 and controls the loading element 110, at least one restraining element 125, and unloading element 140.

In an embodiment of the present invention, the system 200 can comprise a container 205 comprising a fluid 210 and a plurality of sample objects 215. The container can be in fluid communication with the drive system 220 and at least one inlet of the at least one inlet 115 of the detection environment 105. The container 205 can comprise many containers suitable for the culture or dispensation of sample objects 215. In addition, the container 205 is adapted to withstand forces provided by the drive system 220. In an exemplary embodiment of the present invention, the container can be a flask, a test tube, a microtube, a bottle, or the like. Some embodiments of the present invention may comprise more than one containers 205. In an embodiment of the present invention, a single container 205 may be in fluid communication with an inlet 110 of the detection environment 105. In another embodiment of the present invention, more than one container 205 may be in fluid communication with an inlet 110 of the detection environment 105. The container contains a fluid 210A. In an exemplary embodiment of the present invention, the fluid comprises a medium or buffer solution that is compatible with the sample objects of interest. For instance, for detection and sorting of C. elegans, the fluid 210A may be M9 buffer solution.

Various embodiments of the present invention comprise a microfluidic system 200 comprising at least one immobilizing element. In an embodiment of the present invention, at least one of the at least one immobilizing element comprises at least one pressure-based restraining element 125. In an embodiment of the present invention, at least one of the at least one immobilizing element comprises a cooling element 135. In an embodiment of the present invention, the at least one immobilizing element comprises at least one pressure -based restraining element 125 and a cooling element 135. Embodiments of the present invention comprise a cooling system 225, wherein the cooling system 225 is in thermal communication with the detection environment 105 via the cooling element 135. The cooling system comprises a cooling fluid 210B. The cooling fluid can comprise many suitable cooling fluids, including but not limited to, a salt solution, or many coolants that has a low freezing point, such as a high salt solution or a glycerol solution, among others. In some embodiments of the present invention, cooling is achieved by flowing a fluid 210B having a temperature of about 0 ⁰C to about -10 ⁰C through a cooling element 135 fabricated in the control layer of the device 150 and integrated beneath the detection environment where the sample object is restrained. One of ordinary skill in the art would realize that the cooling fluid temperature varies depending on the thickness of the microfluidic device (e.g., thickness of the PDMS) and substrates supporting the microfluidic device (e.g. glass layer). Thickness variation results in differing degrees of insulation and subsequent temperature gradients. In an embodiment of the present invention, the fluid 210B can be flowed off-chip and chilled through small metal tubes adjacent to a Peltier cooler or a refrigerated fluid bath. In an embodiment of the present invention, the cooling system (e.g., Peltier cooler) can be microfabricated on the chip comprising the microfluidic device. By varying the voltage applied to the Peltier cooler, the temperature of the fluid 210B can be precisely controlled.

Embodiments of the present system 200 also comprise a drive system 220 that drives the fluids 210A, 210B, and 210C of the system 200. In an exemplary embodiment of the present invention, a drive system 220 can comprise a compressed gas cylinder in fluid communication with a plurality of gas regulators. In an exemplary embodiment of the present invention, the drive system comprises a compressed gas cylinder of a fluid 210C. In embodiments of the present invention, the fluid 210C can comprise many fluids known in at art, including but not limited to N₂ or compressed air at a pressure of about 10 psi to about 100 psi. In another embodiment of the present invention, the drive system may comprise a pump. Examples of suitable pumps include but are not limited to a pulsatile pump (e.g., a peristaltic roller pump), a rotodynamic pump (e.g. centrifugal pump), a positive displacement pump (e.g., root-type pumps, reciprocating-type pumps, or compressed air-powered double-diaphragm pumps), a kinetic pump, or a gear pump, among others. The drive system can comprise a compressed gas cylinder, a pump, or combinations thereof. In an embodiment of the present invention, the system 200 can be configured to be driven by suction.

In an embodiment of the present invention, the drive system 220 provides a pressure by urging a fluid 210C into the container 205, which in turn urges the flow of a fluid 210A comprising a plurality of sample objects 215 into the at least one inlet 110 of the detection environment 105. In some embodiments of the present invention, cooling is achieved by providing a pressure by the drive system 220 that urges a fluid 210C into the cooling system 225, which in turn urges the flow of a fluid 210B having a temperature of about 10 ⁰C to about 0 ⁰C to the cooling element 135 fabricated in the control layer of the device 100 and integrated beneath the detection environment 105 where the animal is restrained. In some embodiments of the present invention, the drive system 220 provides a pressure by urging a fluid 210C into voids within the control layer. The pressure created by the fluid 210C is sufficient to actuate the PDMS membrane causing the membrane to deflect creating a valve. The drive system may comprise a plurality of control valves 240 to control specific actuation and the degree of actuation of the valves comprising the loading element 120, the at least one restraining element 125, and the unloading element 140.

Embodiments of the present system 200 also comprise a detector 230 capable of detecting a phenotype or characteristic of a sample object. In an embodiment of the present invention, the detector 230 can be many detectors known in the art, including but not limited to, an optical detector, a radiation detector, a magnetism detector, or other detectors capable of recognizing phenotypes or characteristics of a sample object. An optical detector can comprise a microscope, a lens system, a CCD device, a camera, a video recorder, or a photomultiplier tube, among others. In an exemplary embodiment of the present invention, a microscope can be an optical microscope, including but not limited to, a transmitted light microscope (including operations in bright field mode, differential interference contrast mode, and phase contrast mode, among others), a fluorescent microscope, a confocal microscope, or a multiphoton microscope. A sample object can be visulaized in two dimesions or three dimensions.

In embodiments of the present invention, the systems and methods of the present invention permit detection of at least one phenotypic marker in a population of cells or multicellular organisms. For example, the systems and methods of the present invention can detect and sort cells or multicellular organisms expressing a fluorophore, for example a fluorescent molecule, dye, or protein, including but not limited to, green fluorescent protein (e.g., GFP, EGFP), red fluorescent protein (RFP), blue fluorescent protein (EBFP), cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP), and derivatives thereof.

Some embodiments of the systems and methods of the present invention allows for the detection and analysis of not only the intensity of fluorescence of an organism, but also the location of fluorescence within the multicellular organism at cellular and subcellular resolutions. Some embodiments of the systems and methods of the present invention permit the differentiation of multicellular organisms based on the expression (e.g., intensity), morphology, and/or localization of at least one fluorophore to sort and separate an organism having a first phenotype from an organism having a second phenotype, or a third phenotype, and so forth. In some embodiments of the present invention, the phenotypic trait can result from exposure of an organism to a compound, for example, a pharmaceutical compound. In some embodiments, a phenotype can result from genetically crossing two genotypically different multicellular organisms. Alternatively, the automated system of the present invention can be used to separate multicellular organisms that are at a particular stage of development. Examples of applicable multicellular organisms are all stages developmental of C. elegans, D. melanogaster larvae and embryos, or Xenopus or D. rerio embryos.

In some embodiments of the present invention, the system 200 comprises a control system 235 which receives at least one signal from the detection element 120 and controls the loading element 110, at least one restraining element 125, and unloading element 140 via the drive system 220 in response to that signal. The control system 235 comprises an image acquisition component, an image processing component, and an image recognition component, as well as components to automate pressure control, control of the detector (e.g., the stage of the microscope), and feedback control of the valves. It is a self- contained and closed-loop system that needs minimal human intervention. The system and methods of the present invention are capable of a single pass detection and sorting of multicellular organisms with accuracy greater than about 50%. The system is capable of performing a single pass phenotyping and sorting of multicellular organisms with accuracy greater than about 75%. The system and methods of the present invention are capable of a single pass detection and sorting of multicellular organisms with accuracy greater than about 85%. The system is capable of performing a single pass phenotyping and sorting of multicellular organisms with accuracy greater than about 90%. The system and methods of the present invention are capable of a single pass detection and sorting of multicellular organisms with accuracy greater than about 95%. The system is capable of performing a single pass phenotyping and sorting of multicellular organisms with accuracy greater than about 97%. The system and methods of the present invention are capable of a single pass detection and sorting of multicellular organisms with accuracy greater than about 99%. In an embodiment of the present invention, the systems and methods are capable of recirculation of sample object. In an embodiment of the present invention, systems and methods of the present invention are capable of multipass phenotyping and sorting of multicellular organisms. As used herein, the term "accuracy" indicates the ability of the system to correctly identify of a phenotype of interest. The term "single pass" as used herein refers to the detection and sorting of a sample object without having to re-circulate a sample object. As used herein, the term "multipass" refers to the detection and sorting of a sample object wherein at least one sample object must be re-circulated.

In an exemplary embodiment of the present invention, a cycle of the system comprises loading a sample object into the detection environment, the animal is detected (e.g., by its auto-fluorescence) and the valves surrounding the chamber are closed. The camera and stage of the microscope are then controlled to grab a series of images that cover the three-dimensional volume that the animal occupies; the images are then stored and, if sorting is desired, are processed in real time to determine the phenotype of the animal and select the proper exit channel by triggering the corresponding outlet valve to open. The program waits until the animal leaves the observation chamber. The exit channel is then closed, and this completes a cycle. When the next animal is loaded into the chamber by the pressure-controlled self-regulated mechanism, the processing cycle starts over again. This sequence of events can occur in less than about twenty seconds, or less than about fifteen seconds, or less than about ten second, or less than about five seconds. In an embodiment of the present invention, this sequence of events can occur in less than about one second. In an embodiment of the present invention, this sequence of events can occur in about 0.1 seconds.

The systems and methods of the present invention can be automated and may require minimal human intervention. Automation of system reduces the processing time of such experiments, reduces the incidence of photobleaching of fluorescent markers, and reduces or eliminates some of the biases introduced by manual handling. The systems and methods of the present invention can be easily adapted for a wide variety of microscopy-based techniques, including but not limited to fluorescence recovery after photobleaching (FRAP), laser ablation of cells, and laser cutting of neuronal processes. In an embodiment of the present invention, devices, systems and methods of the present invention may comprise one or more detection environments. (Figure 7). For example, a microfluidic device 300 may comprise two detection environments 105. In an embodiment of the present invention systems and methods of the present invention may comprise a plurality of detection environments or an array of detection environments. Despite the plurality of detection environments, at least one loading element remains configured to load one sample object at a time into a detection environment. It must be noted that, as used in this specification and the appended claims, the singular forms "a", "an", and "the" include plural referents unless the context clearly dictates otherwise.

All patents, patent applications and references included herein are specifically incorporated by reference in their entireties.

It should be understood, of course, that the foregoing relates only to exemplary embodiments of the present invention and that numerous modifications or alterations may be made therein without departing from the spirit and the scope of the invention as set forth in this disclosure. Although the exemplary embodiments of the present invention are provided herein, the present invention is not limited to these embodiments. There are numerous modifications or alterations that may suggest themselves to those skilled in the art.

The present invention is further illustrated by way of the examples contained herein, which are provided for clarity of understanding. The exemplary embodiments should not to be construed in any way as imposing limitations upon the scope thereof. On the contrary, it is to be clearly understood that resort may be had to various other embodiments, modifications, and equivalents thereof which, after reading the description herein, may suggest themselves to those skilled in the art without departing from the spirit of the present invention and/or the scope of the appended claims. Therefore, while embodiments of this invention have been described in detail with particular reference to exemplary embodiments, those skilled in the art will understand that variations and modifications can be effected within the scope of the invention as defined in the appended claims. Accordingly, the scope of the various embodiments of the present invention should not be limited to the above discussed embodiments, and should only be defined by the following claims and all equivalents.

EXAMPLES Example 1: Materials and Methods

Microfluidic device fabrication. The microfluidic device was fabricated using multi-layer soft lithography. Two different molds were first fabricated by photolithographic processes to create worm loading layer and the control layer. The mold for the worm loading layer was made by a two-step photolithographic process. In the first step, a 30μm thick negative photoresist (SU8-2025, Microchem) was spin-coated onto a silicon wafer for the worm loading chamber and the detection channel. The loading regulator, side channels, and outlets were then fabricated with a 25μm layer of positive photoresist (AZ 50XT, AZ Electronic Materials) on the same wafer. After the positive photoresist was developed, the wafer was heated at 125 ⁰C for 5 min to allow the positive photoresist to reflow so that the channels form a smooth and rounded shape. The master for the control layer was made of a 50 μm layer of negative photoresist (SU8-2050, Microchem) on a silicon wafer. The two molds and a blank wafer were treated with tridecafluoro-lj^^-tetrahydrooctyl-l-trichlorosilane vapor (United Chemical Technologies ) in a vacuum desiccator to prevent adhesion of PDMS during the molding process.

For fabricating the control layer Polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning A and B in 5:1 ratio) was poured onto the control-layer master to obtain a 5 mm- thick layer. Mixture of PDMS (Sylgard 184 A and B in 5:1 ratio) and THF in 2:1 ratio was spin-coated on a blank wafer to give a 20 μm-thick layer. Both were partially cured at

70 ⁰C for 20 min. The thick control layer was then peeled off from the master and holes were punched for access to the control and cooling channels. The control layer was then bonded to the thin PDMS membrane on the blank wafer. This assembled control layer was fully cured at 70 C for 2 hours. For the worm-loading layer PDMS (Sylgard 184 A and B in 5:1 ratio) was spin-coated onto the master to give a 60 μm-thick layer. The worm-loading layer was fully cured in a convection oven at 70 ⁰C for 2 hours and then was peeled off from the master. The layer was then turned up side down and bonded to the control layer using oxygen plasma treatment. Another set of holes were punched for access to the worm loading channel. These assembled layers were then bonded onto the cover glass to form the micro device.

Strains. Strains used in this work include tax-4(ks28); kyls342 [pgcy-32::tax- 4::GFP, punc-122::GFP], kylsl4θ [str-2::GFP + lin-15(+)], kylsl4θ; rol-6(el87); slo- I(ky399), julsl98 [punc-25-YFP::rab-5], and a mutant that also carries julsl98.

C. elegans culture and sample preparation. Animals were cultured according to established methods. Synchronized L4 worms were prepared as follows: eggs were obtained by bleaching adults using a solution containing about 1% NaOCl and 0.1N KOH, washed and let hatch overnight in M9 buffer, and cultured on Nematode Growth Medium (NGM) plates seeded with E. coli OP50. Animals were washed and suspended in M9 solution containing 0.5 wt% Bovine Serum Albumin (BSA) for each experiment. Filtering device fabrication. To get rid of dust particles and debris in the worm suspension, a filtering device was fabricated using single-mold soft lithography. The mold was fabricated to obtain 40μm thick structures using SU8-2025 on a silicon wafer, and was treated with tridecafluoro-lj^^-tetrahydrooctyl-l-trichlorosilane vapor. PDMS (Sylgard 184 A and B in 5:1 ratio) was poured onto the mold to obtain a 5 mm- thick layer. The PDMS was cured at 70 ⁰C for 2 hours and peeled off from the master. Holes were punched for access to the channels. The PDMS layer was then bonded onto the slide glass by oxygen plasma.

The device comprises a parallel channels with a pillar array -25-30 μm apart. Deformable worms can path through the gap between pillars, but non-deformable debris bigger than the gap are filtered out.

System Control, Image Acquisition, and Image Processing. The code for the worm sorting contains three basic elements: waiting for worm's entrance to detection zone, grabbing images and performing the image processing, and allowing the worm to exit before returning to the initial state. The code for the entering and exiting is essentially the same for all the sorting experiments, and will be described first. The procedure for identifying and sorting the individual mutants is discussed separately and in greater detail.

Waiting for Worm to Enter: The procedure for trapping a worm is identical regardless of the genotype and whether the screen is done at high or low magnification. The valve that controls the side channels is opened to allow flow through the channel, while all the other valves are closed. Frames from the camera are continually grabbed and analyzed to determine the presence of an animal by the average pixel intensity over a threshold.

Waiting for a Worm to Exit: Once the decision has been made as to where the animal should exit, the valve that controls that channel is opened as well as the L-shaped positioning valve to expedite exiting. During this time, frames from the camera are continually acquired, and once no animal is detected to be present in the channel, the exit channels and the L-shaped positioning valves are closed immediately while the side channels are opened. For high magnification (e.g. 10Ox) sorting, only a fraction of the field of view is visible. In order to ensure that an animal has completely exited the imaging zone, a 300 ms delay is added to the routine before closing the exit channels. Image Processing and Decision Making. The image acquisition process for

CX6858 was as described below. To analyze the images, out-of-focus frames were discarded and the images are convolved with [1 1 -1] to accentuate small bright regions. A threshold was applied to determine the fluorescence from the intestine as well as AQR and PQR. Different thresholds are then applied to the left and right nematode centroid to identify AQR and PQR and to distinguish PQR from the intestine auto-fluorescence. Groups of remaining pixels are then compared based on a number of features (size, position, etc) to determine whether AQR and PQR are present, and if so, where they are located. For each animal, the most in-focus of the pictures is used to determine the intensity of AQR and PQR. The correctness of the output (the location and presence of AQR and PQR) for each animal was independently verified and corrected if necessary. The algorithm was found to have >95% accuracy.

Cellular Imaging of AWC Neurons. To locate the AWC neurons, sparse z-stacks (5 μm steps) are gathered along the A-P axis of the worm. These z-stacks are then flattened by computing the standard deviation of pixel in the x-y plane along the z- direction. To ensure that intestine autofluorescence is not mistakenly identified as a neuron, intestine fluorescence is located and removed from the picture. A threshold is applied to this newly flattened image and the neuron(s) are located. The xyz stage then moves to the location of the neuron(s) to grab a more detailed z-stack (1 μm step size) which is necessary to determine whether one or two neurons are present. This z-stack is similarly flattened and a threshold is applied to determine the number of neurons.

Subcellular Imaging of CZ5261 and CZ5264. The sorting of strains CZ5261 and CZ5264 relies on determining the locations of GFP along the ventral nerve cord. This is done using the same methods as mentioned earlier, namely compressing the z-stack to the x-y plane and then convolving it with a matrix [2 0 0 0 -1.5] to accentuate the puncta. A threshold is subsequently applied to the image to locate the puncta and depending on the number of puncta present, the animals are determined to be either wild type or have a mutant background. Example 2: Automated Rapid Imaging, Phenotyping, and Sorting of C. elegans in an Integrated Microsystem

The Automation and the Self-Contained Microsystem. The integrated microsystem comprises both hardware and software, many features of which play a role for the automation and robust operation for screens descried below. Figure 1 is a schematic of the microsystem functions in rapid imaging, phenotyping, and sorting a mixed population of animals based on cellular and sub-cellular phenotypes. The hardware is comprised of a microfluidic device (fabricated in-house), a microscope and camera system, a motorized stage, valves, pressure controller, and a Peltier cooling system (Figure 6). Figure 6 is a system block diagram showing the on-chip and off-chip components and features. The integrated system is controlled by in-house programs coded in Matlab®. The functionality of the software includes image acquisition, writing, processing, and recognition, as well as components to automate pressure control, stage control, and feedback. It is a self-contained and closed-loop system that needs minimal human intervention; in our experiments, the system was repeatedly left running unattended for hours.

Figure 3B is an exploded view of a schematic illustration showing the individual lithography layers of the microsystem. Figure 2A is an optical micrograph of the central region of the microchip. Animals are freely moving in a pre-imaging chamber (not included and to the left of the shown portion of the chip in Figure 2A-B). A gentle pressure gradient along the microchannel can load an animal into the detection zone. Temperature control channel below and around the detection zone carries a working fluid, and in the same layer are the control valves and the sample-loading regulator valve. Between the control layer and the sample layer is a thin membrane that can deflect.

Figure 3C provides schematics of cross-sectional view of the sample-loading regulator valve. The microchip has four design features that ensure its robust operation for an extended period of time. First, it automatically self-regulates the loading of nematodes by the sample-loading regulator design (Figure 3C). Second, it automatically positions the nematodes in an identical position in the chip (so as to minimize the travel of the motorized stage and thereby reduce the processing time and increase the throughput).

Third, it has no small features (< 20 μm), and therefore is not easily clogged by debris and can operate very robustly. Fourth, it has an integrated local temperature control system whereby animals are completely immobilized briefly, for imaging and manipulation without anesthetics.

To load an animal into the imaging chamber, both outlet channels are closed while the side positioning channels remain open (Figure 5A), and a constant pressure source is used to drive the flow into the microchip. Self-regulation of loading (one at a time into the imaging chamber) plays a role in high-resolution imaging and accurate sorting. When an animal is present in the imaging chamber, the flow resistance is increased. The reduced flow rate lowers the pressure on a second animal at the sample-loading regulator located at the entrance of the imaging chamber to a point where it is not sufficient to push the second animal into the chamber (Figure 4). Upon releasing the first animal (by opening one of the exit valves), the pressure drop across the second animal becomes sufficient to push it into the imaging chamber. The sample-loading regulator design was implemented by controlling the pressure on a partially closed valve (Figure 3C). Thus, great flexibility is achieved depending on the size and the deformability of the animals for each application because of the ability to fine-tune the system pressure as well as the sample-loading regulator pressure.

To position the animal precisely inside the imaging chamber, pressure differences between the side channels and the entrance of the main channel are utilized. The side channels are also controlled by the partially closed positioning valve, similar to the loading regulator valve. Once the animal's nose or tail is positioned at the end of the channel, the hydrodynamic resistance of the positioning channels self-equalizes. As a result, the animal stops moving in the direction of the flow (Figure 5B). The valve on the positioning channel is opened to generate a pressure gradient to guide an animal into the observation chamber. This distribution of the pressure force minimizes mechanical stress on the animal.

One advantage of the design of both the positioning valve and the sample-loading regulator in the main channel is that there are no permanent small features (< 20 μm). The dimension of the channels is not from the mask design but from the partially closed valves, and therefore is tunable and can be expanded if necessary. Because pieces of debris smaller than the size of the nematodes cannot be easily filtered out, this design feature prevents clogging of the channels - the valves can be opened to remove the debris when necessary. During experimentation, in order to further minimize interruptions of the imaging and sorting (and not to lose samples in conventional macro filters), a coarse microfluidic filter chip upstream from the imaging and sorting chip was employed.

Anesthetics in some cases may alter animals' metabolisms, growth, and phenotypes of interest. At the same time immobilization plays a role for imaging at cellular and more specifically at subcellular resolutions. To achieve this without using anesthetics, an on-chip temperature control scheme is used in conjunction with a pressure gradient through the side channels. The animal is cooled, immobilized, and imaged (Figure 5C) before being phenotyped and sorted accordingly (Figure 5D). We found that when a pressure gradient alone is applied to the side channels, imaging at high magnification (10Ox oil NA= 1.4) is difficult because the animal still retains enough mobility to blur the images. In contrast, when the animal is cooled to 4 ⁰C, the animal remains still for the duration of image acquisition and processing. Cooling is achieved by flowing salt solution of -8 ⁰C to -3 ⁰C on-chip through a large heat-exchanging channel (fabricated in the control layer of the device) beneath the observation chamber where the animal is positioned (Figure 2A). The salt solution is flown off-chip and chilled through small metal tubes adjacent to a Peltier cooler (Figure 6); by varying the voltage applied to the Peltier cooler, the salt solution temperature can be precisely controlled. The temperature in the observation chamber on-chip has been calculated to be about 4 ⁰C at the experimental conditions. Animals were observed to become still in the chamber almost instantaneously due to their small thermal mass, and immediately regained their typical thrashing motion upon exiting the cooled observation chamber. Once the animal is positioned in the observation chamber all the valves for fluid exiting the observation chamber are closed to eliminate flow fluctuation. By opening one of the exit valves the imaged worm is released. Once the worm leaves the observation chamber the valves are returned to the worm entering state.

The chip is microfabricated using well-established multilayer soft lithography techniques with some modifications. The device is made of silicone elastomer polydimethylsiloxane (PDMS), which is optically transparent, exhibits negligible auto- fluorescence, and is elastic so micro on-chip valves can be built into the structure. The device is capped with a standard microscope coverglass to ensure compatibility with all types of microscopes and objectives. Devices built with a conventional multilayer process have the control layer (where gas is pressurized to actuate the valve membrane) between the sample-handling layer and a glass substrate in order to have a fully closed channel. Light has to pass through both the coverglass and the control layer when microscopy is performed, which may pose a limitation for the sample thickness at high magnification. In the present device, the layer with microchannels for sample-handling is next to the coverglass while still having a rounded shape (Figure 3C) in order to allow use of high- resolution and high-numerical aperture objectives (e.g. 10Ox oil objective with NA=I.4 and a focal depth of 90 microns beyond the thickness of a coverglass).

The present microsystem is capable of interfacing with a wide variety of microscope and camera systems, and thus would be a relatively inexpensive addition to the experimental facilities typically present in a biology laboratory. The microscope (Leica DM4500), camera (Hamamatsu C9100-13), and the motorized stage (Applied Scientific Instrumentation MS -4000 XYZ) used in this study directly reflects the needs of our applications. Dim fluorescent reporters and subcellular features require high magnification and high numerical aperture lenses and a sensitive camera, while bright reporters and relatively larger features do not need such expensive equipment. In general, the end-user of our system will pick the microfluidic chip of the appropriate geometry and the software modules of appropriate capabilities with the microscope, camera, and stage of choice. To automate the operation of the microsystem, a series of software modules were developed. The specific module to be used depends on the imaging and sorting applications at hand. For all applications, the software controls image acquisition and processing, stage movement, and opening and closing of the on-chip valves through off- chip macro valves. An automated operation cycle of the microchip is demonstrated in Figures 5E-G, which are optical micrographs showing automated imaging and sorting sequence. Once animals are suspended in M9 buffer solution (1, 000-5, 000/ml) and loaded, the closed-loop automated system will process the animals with no human intervention. When one animal is loaded into the observation chamber of the microchip, the animal is detected (usually by its auto-fluorescence) and the valves surrounding the chamber are closed. The camera and stage are then controlled to grab a series of images that cover the three-dimensional volume the animal occupies; the images are then stored and, if sorting is desired, are processed in real time to determine the animal phenotype and select the proper exit channel by triggering the corresponding outlet valve to open. The program waits until the animal leaves the observation chamber. The exit channel is then closed, and this completes a cycle. When the next animal is loaded into the chamber by the pressure - controlled self-regulated mechanism, the processing cycle starts over again. This sequence of events usually happens within a few seconds, depending on the sophistication of the image processing algorithm. The individualized image processing modules take advantage of a priori knowledge of the phenotypes of the strains; in some cases the software is further fine-tuned in real time by examining the animals in the device at the beginning of each application to achieve high-speed processing.

This automated phenotyping and sorting process is gentle, and in our experiments, >99.8% of animals were viable, crawling on agar and thrashing in buffer normally immediately after the processing. The power of the microsystem in its throughput, resolution, and sophistication of automation are shown in three applications. Fast 2-D Phenotyping. Gene expression pattern analysis is a common technique in genomic studies, as well as mosaic and genetic analysis, and gross phenotyping. Typically the morphology, intensity, or location of a (fluorescent) marker is observed. Conventional manual inspections by microscopy yield high-quality 3-D diffraction-limited images but can only examine a few animals mounted onto a microscope slide at a time. Conversely, modified FACS has large throughput, but the images are only 1-D (e.g., average intensity in the dorsal-ventral left-right plane) and the resolution is on tissue scale. In many genetic experiments, finding the expression pattern accurately as well as quantifying the expression level is important. To demonstrate the utility of our high-throughput and high- resolution methodology, an experiment was performed to quantify gene expression profile in a population of animals. C. elegans strain CX6858 contains an integrated transgene kyh342 [pgcy-32::tax-4::gfp, punc-122::gfp]. Green fluorescence protein (GFP) is expressed in at least some of the following sensory neurons that normally express a soluble guanylyl cyclase gene gcy-32: AQR and URXL/R in the head and PQR in the tail. The expression pattern and levels in AQR and PQR vary from individual to individual. In this strain as in many other transgenic strains, GFP is also present in other cells through the expression of a coinjection marker (punc-122::gfp in coelomocytes). The present example shows that neurons can be identified and distinguished, gene expression levels can be quantified, and the animals can be phenotyped.

Greater than 400 adult animals per hour were processed through the system. Images for each animal that went through the system were automatically obtained and stored for further processing. Roughly half of the time processing each animal was spent writing the images to the computer hard disk. This experiment was performed using a 1Ox air objective without cooling, and the animals were essentially free-moving throughout the process. To ensure that the head and the tail would be in focus in at least one of the frames, several images were acquired at different focal planes. During the image analysis, all frames were processed and the frames with the most in-focus head and in-focus tail regions were used for quantification. Figure 8B-E show representative raw images for each of the four possible expression patterns (in URXs only, in AQR and URXs, in PQR and URXs, and in all four cells) in the microdevice (Scale bar: 100 μm). Figure 8B shows GFP expressed in URXL/R only. Figure 8C shows GFP expressed in AQR and URXL/R. Figure 8D shows GFP expressed in PQR and URXL/R. Figure 8E shows GFP expressed in AQR, POR, and URXL/R.

The automated image analysis was able to distinguish the neurons from each other and from the coelomocytes along the body that also express GFP. URX expression level of GFP is consistently high in all animals, which can be ignored by the image processing module. In contrast, GFP expressions in the AQR and PQR neurons are stochastic and at lower levels. Figures 8F-I show the processed images where gut auto-fluorescence, coelomocyte GFP, and GFP in URXL/R were filtered out by the software, only leaving AQR and PQR fluorescence. Figures 8J-M show the overlay of the raw images B-E and the processed images F-I. By setting a threshold, all animals were phenotyped into four categories by a binary definition of gene expression in AQR and PQR (Figure 8N). Figure 8N graphically depicts the percentage of animals with each of the possible expression patterns in >l,000 animals. In addition, quantification of the GFP expression level was performed on AQR and PQR. Interestingly, because AQR and PQR are putative oxygen- sensing neurons in the head and in the tail respectively, their distinct expression levels may explain the individual variations in behavior in an oxygen gradient. Because animals are loaded stochastically from a suspension (and the density decreases as more animals are processed), there is variation in the processing time. Figure 80 shows the histogram of the loading time for the individual animals and over 58% of animals were loaded in the observation chamber within 1 sec. The loading scheme of the present example is passive and therefore very simple; furthermore, this experiment demonstrated that the loading is also fast and efficient. The technique of phenotyping of the present example, compared to manual methods, is high-throughput, it minimizes photobleaching and ensures uniformity of treatment on the samples, and therefore it is able to produce imaging data that are quantitative.

3-D Imaging and Sorting with Cellular Resolution. The ability to distinguish and identify tissue types and specific cells in a multicellular organism and the ability to quantify gene expression profiles are important in many genetic studies. Currently databases such as the Wormbase (wormbase.org) curate experiments that individual laboratories perform on reporter genes. A systematic approach to detailing such information would be significantly beneficial for studies of gene, cell, and tissue interactions. By performing image analysis and using heuristics, one can infer the tissues that express the reporter genes. A compound epifluorescent microscope was used to obtain z-stack images that yield information of the reporter genes in the other two dimensions (left-right and dorsal- ventral).

To demonstrate the microsystem's ability to perform sorting according to expressions of reporter genes in different cells at similar locations, a small fraction of animals of a mutant genotype mixed in a background of animals of wild-type genotype were sorted. C. elegans has many anatomically bilaterally symmetric cells; interestingly, some pairs exhibit functional asymmetry. Bargmann and colleagues have shown that many genes disrupt such normal development in the functional asymmetry. In wild-type animals, stochastically one but not both of AWC chemosensory neurons express the gene str-2. The str-2 expressing AWC cell is called the AWC-ON, and the other cell AWC- OFF. Mutations in slo-1 gene produce 2-AWC-ON phenotype. In the present example, slo-1 animals (which also carry a rol-6 transgene) were sorted from animals of wild-type background. The experiment was performed at 10Ox magnification using an oil objective (NA= 1.4) with cooling to immobilize the animals. Figure 9 illustrates automated three- dimensional imaging and sorting with cellular resolution in the microchip. Figure 9 A-D are representations of an image processing and decision-making process to sort animals based on the number of AWC neurons expressing pstr-2::GFP (the AWC-ON cells). Several series of sparse z-stack images along the body of each animal were obtained and analyzed to determine the location of the head where fluorescence is most intense (Figure 9A). The stage then centered the neuron(s) within the field of view and a denser z-stack was acquired (Figure 9B). This z-stack was used to determine whether the animal is 2-ON (slo-1, Figure 9C-D). The z-stack images at 10Ox are necessary because the AWC cells are laterally symmetric, less than 30 microns apart, and impossible to distinguish at low magnifications.

A mixed population of the two genotypes was successfully sorted based on the GFP patterns. Age-synchronized adult animals of both strains were mixed at a ratio of -1.5 % slo-1 mutant in wild-type background and processed in the microsystem. A total of about 300 animals were sorted in about 6 hours continuously. Collected animals were 100% viable and behaved normally on agar plates with bacterial food. The accuracy was verified by both scoring the recorded images and also examining the collected animals by behavior for roller phenotype since the strain that carries slo-1 mutation also has rol-6 as a coinjection marker. All but one 2-ON animals were sorted correctly, and the 2-ON animals were enriched by > 25 fold. The false positive rate (1 -ON-I -OFF animals being sorted as 2-ON animals) is < 2%. The images in the experiments were recorded and can be retrieved if further analysis is required. For example, it is trivial to quantify the distribution of the absolute gene expression level (i.e. intensity of GFP) in the AWC-I-ON animals. Moreover, since the microsystem is compatible with any microscope system, one can use deconvolution or confocal techniques if higher image quality is required for a particular application.

High-Throughput Imaging and Sorting Based on Subcellular Phenotypes. Many genetic screens are difficult to perform because the screens require experts to mount, examine, and subjectively phenotype the animals, and most of the time "rescue" the anesthetized animals of interest from the microscope slides. This operation is slow and can be damaging to the animals (from the anesthetics and/or the physical handling). More importantly, the screens are becoming increasingly difficult to perform because the phenotypes of interest are finer and subtler. For example, many of the synaptic or other subcellular reporters are sub-micron in size, faint, and easily photobleached. In addition, manual phenotyping is subjective and cannot detect quantitative changes. Therefore, one of the target applications of the present microsystem is for genetic screens using such reporters. The advantage of using an automated system are multifold: photobleaching can be minimized because the images are automatically obtained; automatic positioning the animals in the microdevice and the single-pass image acquisition eliminate the need to search for features (which is necessary if such operation is performed manually); there is much reduced bias and noise in the data acquisition since all animals are treated identically in the device; and lastly, brief cooling to immobilize animals provides an alternative to using drugs.

To demonstrate the utility of the microsystem in this type of applications where subcellular changes are used as the criteria for screening, a sorting experiment was performed using strains CZ5261 and CZ5264. Strain CZ5261 is of wild-type background and carries an integrated reporter transgene julsl98 [punc-25-YFP::rab-5], which expresses YFP-RAB -5 in the cell bodies of the GABAergic motorneurons in C. elegans (Figure 10A-D). Strain CZ5264 also carries the marker transgene julsl98 but is mutant in its genetic background. CZ5264 has an altered phenotype in the marker intensity in the cell bodies and along the nerve cord (Figure 1 OE-H). Age- synchronized animals were cultured and mixed at a ratio of about 30 % mutants to about 70% wild-type background. Greater than 1,300 animals were sorted in 7 hours continuously without interruptions, showing the robustness of the device and the approach.

The software program was able to identify cell bodies as well as the puncta phenotype along the ventral nerve cord of the animals. To perform the identification, we flattened each z-stack by computing the standard deviation of each pixel in the x-y plane along the z direction. Positions where cell bodies or puncta are present will have a large standard deviation due to the significant variation in intensity along the z-axis. Abrupt changes in the standard deviation are enhanced to result in images shown in Figure 1OC and 1OG. Finally, a threshold is applied and groups of pixels above the threshold are counted to determine the number of cell bodies and puncta present. Over 99.9% of the animals were viable and behaved normally on culture plates after sorting. To verify the sorting results, the animals were collected and examined behaviorally, in addition to verifying the recorded image sets. The overall sorting accuracy was 97.7%. If desired, the present system can be easily set up as a multiple-pass sorting scheme to further increase the sorting efficiency or to sub-categorize the previously sorted animals. In addition, the sorting speed can be even further improved by improving the algorithm as well as upgrading the computer hardware to improve the speed for data writing.

In the present system, by applying cooling, the use of an anesthetic to immobilize the animals during imaging at high magnification that manual microscopy normally requires is not necessary. Without cooling, the animals still exhibit significant movement in very brief imaging period, such as shown in an overlay of two images 270 msec apart (Figure 101). On the other hand, with cooling two images 10 seconds apart are completely overlapping (Figure 10J). To show that cooling and the time spent in the device did not alter the phenotype based on which the animals were sorted, the images of mutant animals CZ5264 treated with sodium azide and by cooling in the device were compared. Images in showed no discernible differences in the patterns of the puncta. This also demonstrate that in scenarios where cooling is undesirable, anesthetics can still be used to immobilize the animals for imaging in our system.

Quantitative analysis of reporter intensity requires that photobleaching be both minimal and consistent between samples. Figure 1OK shows the effect of photobleaching on samples, which can be minimized by using the present automated system. After 20 seconds of exposure to the excitation light, the samples are much dimmer. If the same thresholding criteria are used, the number of puncta can be easily miscounted (or puncta miscategorized), and quantification of the puncta brightness can be noisy. The two sets of arrows in Figure 1OK point to two small puncta structures that may not be identified had the sample been photobleached for 20 seconds. In the present experimental approach, samples are only exposed to light once while the images are obtained, eliminating the need to focus and find targets by a skilled expert. Therefore the treatment of all samples is equal, and the bias from the operator is minimal. For screens on many synaptic markers, photobleaching is also a concern. This microsystem and the automated approach can be used to enable faster discovery of molecules and pathways at such subcellular resolutions.

Example 3: Automation and Computer Control

The automation and computer control of the microsystem was created using the Mathworks software Matlab™. Using the Matlab™ environment, custom programs and algorithms were created to control the three primary systems: (1) a camera, (2) the XYZ stage, and (3) the off-chip solenoid valves. The program controls the various camera functions and settings such as exposure time, sensitivity, gain, image grabbing, and logging. The XYZ stage is controlled via the COM port. It is actuated in the z-axis to acquire images at multiple focal planes, and in the xy-plane to acquire images along the anterior/posterior and the ventral/dorsal axes of the worm. With regards to the off-chip solenoid valves, a simple digital I/O board is initialized and used to selectively turn on/off the individual valves that expose the on-chip micro-valves to either an ambient air pressure, or high pressure. In this manner the on-chip valves are turned on or off.

A simple description of the program and algorithm that we designed to acquire images and sort a particular animal follows:

1. The digital I/O board actuates the valves to open the suction valve and close all others on-chip.

2. Images from the camera are constantly acquired during this time. The program waits until a certain number of pixels are above a threshold. Once 10 pixels are above this threshold, a worm is assumed to be within the field of view of the camera.

3. At this point, the suction valve is closed to prevent movement of the fluid and the L-shaped positioning valve is opened to prevent the animal from being compressed.

4. The xyz- stage is actuated in the z-axis to acquire images at multiple focal planes (this is called a z- stack). Images are acquired based on the step size in the z direction and the total distance desired to be covered by the z- stack.

5. Depending on the particular strain of worm to be sorted, the xyz-stage may be moved in the xy-plane so that additional z-stacks are acquired at different points along the worm. This step may or may not be necessary. 6. The z-stacks are processed to find the individual neurons or puncta where GFP is expressed. Due to the ability to of the system to completely stop the movement of the sample, animals can be sorted based on GFP expression patterns at the cellular and sub-cellular level. This process is extremely flexible, and numerous algorithms have been created to sort samples depending on their expression patterns. Examples of sorting criteria include but are not limited to number of GFP expressing neurons/puncta, intensity of the GFP expression, size of the neurons/puncta, and distance between adjacent neurons/puncta.

7. Depending on the results of the image processing, the worm is sent through either the left or right exit channels by using the digital I/O board to open the on-chip valve. During this time the camera continues acquiring images and once the images drop below a certain threshold (see step 2), the worm is assumed to have exited the field of view. The program then waits half a second to allow the worm to fully exit before closing the exit and returning to step 1.

Image Processing Examples. In order to provide increased insight into the mechanisms underlying the automated image processing that we designed and implemented, two examples are provided.

Localizations of GFP in Cellular Features. In this example, a few neurons are expressing GFP along the ventral nerve cord. The neurons are all in essentially the same focal plane, so a measure of focus (the variance within a single plane) is used to determine the correct image within the z-stack to use for further image processing. To automatically determine the thresholds for the conversion to a binary image, the mean and standard deviation of the worm are used. To find the mean fluorescence of the worm, a low threshold is used and several image processing steps are taken, including but not limited to the filling of holes or the application of a look up table._Finally, bounding boxes are found along each edge of the worm (used as the neurons are along the ventral nerve cord) and a threshold based on the mean plus one standard deviation is applied to get the following image.

Localization of GFP to Subcellular Features. In a second example, the worms have numerous puncta along the nerve cord, and thus, this example is interested in imaging and locating these subcellular features. Because the features are extremely small, have limited fluorescence and are not necessarily localized to a single focal plane, a different method of locating the puncta was selected. Instead of selecting a single image out of the z-stack as with the previous method, the information in the z-stack was compressed. This was done by creating a matrix in the same size as a single plane of the z-stack. The value of the matrix at each x-y coordinate was set equal to the standard deviation in the z-axis at that x- y coordinate in the z-stack. For x-y positions where puncta or neurons are located, there is a significant variation in the intensity between planes where they are in-focus, and planes where they are out of focus. This leads to a high standard deviation at that point. This matrix is then convolved with [2 0 1 0 -2]. This yields a matrix where the individual values are the difference between nearby coordinates of the standard deviation matrix. By performing thresholding and applying look up tables similar to the first example, the x-y location of the individual puncta can be found. After this, the sample is sorted based on the number, size, or intensity of the puncta.

Example 4: Computer-enhanced high-throughput genetic screens of C. elegans in a microfluidic system Visual screens based on fluorescent markers are commonly used in genetics and drug discovery but when applied to multicellular organisms are currently limited in throughput and accuracy. The present example provides a genetic screen of C. elegans on-chip, performed using computer-enhanced human decision-making. Animal handling streamlined by microfluidic devices and the control software enabled the identification of novel mutants and a large screening speed.

In forward, reverse, and chemical genetics for multicellular organisms, fluorescent reporter-based screens are common techniques. In the nematode C. elegans, often one is interested in changes to a specific phenotype based on morphology, including but not limted to reporter intensity, location, or patterns. Current standard approaches to these screens include manual microscopy, which is slow and a commercial system with high throughput capacity but limited resolution. Recent work has shown that microfluidics can greatly assist animal handling, and that it is possible to sort animals based on well-defined phenotypes. In many applications, however, there is a need yet to be met for high-throughput screening of phenotypes not determined a priori or not easily defined. To meet this challenge, the present example provides a computer-enhanced microfluidic screening system for complex phenotypical screens of C. elegans.

In the previous example, a fully automated system utilizing machine learning and classification was demonstrated to sort mixed animal populations of predetermined phenotypes. While computers are accurate in quantifiable traits, human decision-making is far superior in pattern-recognition and classification in complex situations, as those in many screens of previously unidentified mutants. Additionally, programming for machine learning is outside the purview of many potential users. To take advantage of the human flexibility and the computer quantitation and accuracy, a computer-assisted methodology was developed to allow an expert to determine in real-time, whether animals being screened are of interest. Using this software control interface (Figure HA), preconfigured image processing modules can be selected if needed to help clarify and accentuate phenotypical characteristics. In Figure HA, the video feed is shown in the top left box, and image processing steps can be selected, applied and displayed in the boxes on the right. Animals are sorted as either wild-type or mutant by selecting the appropriate button. If an image is unclear, pictures can be acquired at multiple focal planes and processed using selected image processing modules. While animals clearly exhibiting no interesting phenotypes can be dismissed quickly, potential mutants can be examined in greater detail using the image processing modules on the same user-interface. When a worm is in the field of view, one of over forty combinations of image processing options can be selected and subtle phenotypes emphasized. For markers that are out of focus, one option is to acquire a small z- stack of images at different focal planes (with user-determined step size and number), and either autofocus or flatten the z-stack before further processing the images. This significantly reduces the time relative to manual focusing of the microscope and searching for the reporter, and potentially avoiding photobleaching of the markers. There are eleven image-filtering options to accentuate features of interest, which tend to be dim or low in contrast to human eyes, but the phenotypes become more obvious with image enhancement.

Animals possessing interesting phenotypes usually take longer than four seconds to examine, but because these animals are rare, the throughput of this system is much higher than any pre-existing technology of similar resolution (Figure HB). Figure HB is a representative sequence of total processing time per animal, showing robust and easy animal handling and processing in the device. Animals of potentially interesting phenotypes are examined in detail, typically taking more than 4 seconds each (shaded in pink), while the majority of animals are processed in <2 seconds.

The hardware system of the present example takes advantage of the higher magnification and higher numerical aperture of a compound microscope and the simple and streamlined animal handling of a novel microfluidic device. The chip comprises a two-layer polydimethylsiloxane device with a positioning control valve and two outlets (Figure 2B). Two different molds were fabricated by photolithographic processes to create worm loading layer and the control layer: a 30-μm-thick negative photoresist (SU8- 2025, Microchem) for the worm loading chamber and the detection channel and a 15-μm layer of negative photoresist (SU8-2010, Microchem) for the control layer. Figure 2A provides an optical micrograph of the device active region.

More specifically, the microfluidic device was fabricated using multi-layer soft lithography. Two different molds were fabricated by photolithographic processes to create worm loading layer and the control layer as follows: a 30-μm-thick negative photoresist (SU8-2025, Microchem) was spin-coated onto a silicon wafer for the worm loading chamber and the detection channel. The master for the control layer was made of a 15-μm layer of negative photoresist (SU8-2010, Microchem) on a silicon wafer. The two molds and a blank wafer were treated with tridecafluoro-lj^^-tetrahydrooctyl-l-trichlorosilane vapor (United Chemical Technologies) in a vacuum desiccator to prevent adhesion of PDMS during the molding process. For fabricating the flow layer, polydimethylsiloxane (PDMS, Sylgard 184, Dow Corning A and B in 10:1 ratio) was dispensed onto the flow- layer master to obtain a 5 mm-thick layer. A mixture of 10:1 was then spin-coated onto the control layer to create a 50-μm-thick membrane. The control layer was then allowed to relax at room temperature for 30 min. The flow layer was partially cured at 70 ⁰C for 20 min, and the control layer at 65 ⁰C for 9 min. The thick flow layer was then peeled off from the master, cut into small rectangles and individually aligned and bonded to the thin PDMS membrane on the control layer. This assembled device was fully cured at 70 ⁰C for 2 hours. Once cured, the devices were removed from the wafer, and holes were punched to provide access to the various layers. Devices were then treated with oxygen plasma and irreversibly bonded to glass slides. In the present example, C. elegans were cultured according to established methods.

Mutagenesis was performed on age- synchronized L4 animals using EMS according to standard protocols. F2 eggs were obtained by bleaching Fl adults using a solution containing about 1% NaOCl and 0.1 M NaOH, washed in M9 buffer, and cultured on Nematode Growth Medium (NGM) plates seeded with E. coli OP50 until L4 stage. Animals were washed and suspended in M9 solution containing 0.02 wt% Bovine Serum Albumin (BSA) for each experiment. Animals were screened under a compound microscope at 20X based on differences in the reporter expression pattern or intensity; potential animals of interest were sorted into the mutant outlet and were collected directly from tubing connected to the mutant outlet with M9 solution containing 0.02 wt% BSA. Animals were subsequently transferred to individual plates for culture and further examination. The fluid flow and worm-loading are pressure driven. Animals are reliably positioned within the field of view of the camera: when the circular valve is pressurized, it only closes partially allowing fluid flow to continue but creating a cross-section too small for the animals to pass; at the same time, the valve controlling the wild-type outlet is left open to allow fluid to exit in order to load animals for rapid processing (Figure HB). To reduce the likelihood of another worm entering the field of view when one is already loaded, the device geometry was designed such that the presence of a worm in the imaging channel increases the resistance of flow significantly and thus reducing the flow. This design allows efficient and well-controlled loading of animals, even with size variations resulted from the mutagenesis. The device has a minimum feature size of 30 μm to reduce the likelihood of clogging. Because of the overall simplicity and large tolerance built into the design to minimize the consequences of poor feature registration (either rotational or translational), the device can be easily duplicated by users unfamiliar with microfluidics.

The software interface allows users to control various camera settings such as sensitivity and gain, and to control the exit time of animals. By selecting the appropriate buttons, an animal is sorted as either mutant or wild-type. If the image in the streaming video window is unclear to the user, selecting "stack" will acquire images at multiple focal planes (number of and spacing of images as specified by the user). Images can be flattened and processed according to the user selection.

The various options for flattening the z-stack are: (1) summation (this flattens the stack by making each x-y point equal to the summation of the values at that point over the z-direction); (2) maximum (this flattens the stack by making each x-y point equal to the maximum of the values at that point over the z-direction); (3) standard deviation (this flattens the stack by making each x-y point equal to the standard deviation of the values at that point over the z-direction); and (4) in-focus (this assumes that the slice with the highest standard deviation is the most in-focus and uses it for the subsequent image processing steps). The various options for image processing the flattened image are: (1)

Gaussian (applies a rotationally symmetric Gaussian low-pass filter to the flattened image); (2) Laplacian (applies a filter approximating the Laplacian operator to the flattened image); (3) Laplacian of Gaussian (applies a rotationally symmetric Laplacian of Gaussian filter to the flattened image); (4) Prewitt H (applies the prewitt filter for emphasizing horizontal edges to the flattened image); (5) Prewitt V (applies the prewitt filter for emphasizing vertical edges to the flattened image); (6) Sobel H (applies the sobel filter for emphasizing horizontal edges to the flattened image); (7) Sobel V (applies the sobel filter for emphasizing vertical edges to the flattened image); (8) unsharp (applies an unsharp filter for contrast enhancement created by the negative of a Laplacian filter and applies it the flattened image); (9) range (filters the image provided by the flattening step using the local range of the image); (10) entropy (filters the image provided by the flattening step using the local entropy of the image); (11) standard deviation (filters the image provided by the flattening step using the local standard deviation).

This phenotyping and sorting process is gentle, and in our experiments, 100% of animals were viable, crawling on agar and thrashing in buffer normally immediately after the processing. To load an animal into the imaging chamber, the outlet channel for wild- type animals is left open, while the positioning control valve and mutant channel valves are closed. This allows fluid flow to continue and carry a worm into the field of view of the camera, until its head pushes against the positioning control valve. The presence of the worm significantly increases the fluid resistance of the channel, so the flow is dramatically reduced, preventing another worm from entering. If the worm is clearly wild- type or mutant, the user selects the "Wild-Type" or "Mutant" buttons on the control interface. If wild-type, the positioning valve opens and allows the worm to be released. If mutant, the wild-type outlet is closed and then the mutant channel and position valve are opened. The mutant channel and positioning valve are then reset to the closed position for the next animal.

Next, the power of this computer-enhanced approach was demonstrated in a successful screen of an EMS-mutagenized C. elegans population carrying a synaptic reporter punc-25-YFP::RAB-5. According to the systems and methods of the present example, most animals can be processed in < 2 seconds, and sorting was achieved at rates up to 2500 animals per hour and at a sustained rate of at least 1500 per hour. From a screen of -15,000 mutagenized worms, a number of novel mutants that appeared different from wild-type were identified (Figure 12). Figures 12A, D, G, and J illustrate the wild type, and Figures 12B, E, H, and K illustrate an apparent synaptic mutant showing altered reporter expression along the nerve cord and puncta structures. Figures 12 C, F, I, and L depict a mutant showing reduced YFP expression. Figures 12A-C are images of animals that entered, not necessarily in focus and potentially rotated, resulting in an unclear image of the region of interest. Figures 12D-F are images determined to be in-focus by computer after a series of images at different focal planes was acquired. Figures 12G-I are selected alternative methods of viewing z-stack by flattening the matrix of images. Specifically Figure 12G demonstrates flattening by taking the standard deviation of the z-stack at each x-y location. Figure 12H demonstrates flattening using the maximum value at each x-y location. Figure 121 demonstrates flattening by taking the summation in the z-direction at each x-y location. Figures 12J-L depict application a few of the image processing features to the flattened image to accentuate different features: Laplacian filter (12J); Unsharp filter (12K); and Laplacian of Gaussian filter (12L). In Figures 12A-L, scale bars are 30 μm.

The identification of these mutants was greatly facilitated by applying the auto-z- stack and autofocusing (Figures 12D-F) and the many filters (Figures 12G-L). One class of mutants involved changes in the localization of the reporter yellow fluorescent protein (YFP) localization along the nerve cord and significant numbers of punctated structures (Figures 12B, E, H, K), and another class showed a dramatically reduced expression of YFP (Figures 12C, F, I, L). The computer-enhanced microfluidic approach demonstrated in the present example has many advantages: (1) computer-assisted screening to accentuate phentoypical characteristics, which may be missed by manual screens; (2) human decision-making to allow flexibility if presented with a novel phenotype; (3) preconfigured image processing modules for minimal algorithm-development time; (4) at least one or two orders of magnitude greater throughput than current manual screening; (5) higher magnification, higher numerical aperture optics than commercial or some manual screening systems; (6) almost three orders of magnitude less expensive than commercial systems; (7) simple assembly and operation for use by technicians with little or no familiarity with microfluidic s, among others. These advantages should enable new types of screens in the near future.

Claims

CLAIMSWhat is claimed is:

1. A system, comprising a device for individually detecting and sorting a plurality of multicellular organisms having at least one phenotype at least at cellular or subcellular resolution.

2. The system of Claim 1, wherein device is a single pass device or a multipass device.

3. The system of Clam 1, wherein the multicellular organism is Caenorhabditis elegans.

4. The system of Claim 1, further comprising an immobilization system.

5. A detection system, comprising: a detection environment comprising at least one inlet and at least one outlet; a loading element, wherein the loading element in fluid communication with at least one inlet of the detection environment, the loading element adapted to a load one sample object into the detection environment; an immobilization system, wherein the immobilization system is in operational communication with the detection environment; and a detector, wherein the detector can detect a phenotype of a sample object located in the detection environment.

6. The detection system of Claim 5, further comprising a container comprising a fluid and a plurality of sample objects having at least one phenotype, wherein the container is in fluid communication with at least one inlet of the at least one inlet of the detection environment.

7. The detection system of Claim 5, further comprising at least one unloading element, wherein the unloading element is in fluid communication with the at least one outlet of the detection chamber, wherein the unloading element in operational communication with a detector.

8. The detection system of Claim 7, further comprising a control system, wherein the control system receives a signal from the detector and controls the loading element, the unloading element, and immobilization system.

9. The detection system of Claim 5, wherein the immobilization system comprises a cooling system or at least one restraining element

10. The detection system of Claim 5, wherein the sample object comprises a unicellular or multicellular object.

11. The detection system of Claim 10, wherein the sample object comprises Caenorhabditis elegans.

12. The detection system of Claim 5, wherein the system comprises a microfluidic system.

13. The detection system of Claim 5, wherein the at least one restraining element is pressure -based restraining element.

14. A microfluidic device, the device comprising: a detection environment comprising at least one inlet and at least one outlet; a loading element, wherein the loading element is in fluid communication with at least one inlet of the at least one inlet of the detection environment, the loading element adapted to a load a sample object into the detection environment; at least one immobilization element, wherein the at least one immobilization element is in operational communication with the detection environment; an unloading element, wherein the unloading element is in fluid communication with the at least one outlet of the detection environment.

15. The device of Claim 14, wherein at least one of the at least one immobilization element comprises a cooling element or at least one restraining element

16. The device of Claim 14, wherein the sample object comprises a unicellular or multicellular object.

17. The device of Claim 16, wherein the multicellular object comprises Caenorhabditis elegans.

18. The device of Claim 14, wherein the at least one restraining element is pressure- based restraining element.

19. A method for detecting at least one phenotype of an object, the method comprising loading a microfluid comprising a single object from a fluid comprising a plurality of objects into a microfluidic device; immobilizing the object; and detecting the phenotype of the object.

20. The method of Claim 19, wherein immobilizing the object comprises restraining the object or cooling the microfluid.

21. The method of Claim 19, further comprising unloading the object from the microfluidic device.

22. The method of Claim 21, further comprising repeating the loading, the immobilizing, the detecting, and the releasing at least once.

23. The method of Claim 21, wherein unloading the object from the microfluidic device comprises sorting the object.

24. The method of Claim 21, wherein the object comprises a unicellular or multicellular object.

25. The method of Claim 21, wherein the multicellular object comprises Caenorhabditis elegans.

26. The method of Claim 21, wherein the method is automated.