MXPA00011492A - Microfabricated cell sorter - Google Patents

Abstract

The invention provides a microfabricated device for sorting cells based on a desired characteristic, for example, reporter-labeled cells can be sorted by the presence or level of reporter on the cells. The device includes a chip having a substrate into which is microfabricated at least one analysis unit. Each analysis unit includes a main channel, having a sample inlet channel, typically at one end, and a detection region along a portion of its length. Adjacent and downstream from the detection region, the main channel has a discrimination region or branch point leading to at least two branch channels. The analysis unit may further include additional inlet channels, detection points, branch points, and branch channels as desired. A stream containing cells is passed through the detection region, such that on average one cell occupies the detection region at a given time. The cells can be sorted into an appropriate branch channel based on the presence or amount of a detectable signal such as an optical signal, with or without stimulation, such as exposure to light in order to promote fluorescence.

Description

- .. MICROFABRICATED CLASSIFIER OF CELLS The US government may have certain rights in this invention in accordance with Assignment no. DAAH04-96-1 -0141 granted by 5 the Army. This application claims the priority benefit of the US patent application no. 08 / 932,774, filed on September 25, 1997, and is a continuation in part thereof; provisional US application No. 60 / 108,894, filed November 17, 1998; and provisional US application No. 60 / 086,394, filed May 22, 1998, each incorporated herein by reference in its entirety.

BACKGROUND OF THE INVENTION This invention relates to a method and microfabricated device for classifying cells or particles by size, charge or other identifying characteristics, for example, characteristics that can be detected optically. The invention includes a fluorescence activated cell sorter (FACS), and methods for analyzing and classifying cells by measuring a signal produced by an optically detectable reporter (e.g., color change, ultraviolet or fluorescent) associated with the cells. The methods and apparatuses of the invention allow FACS machines of high sensitivity, without cross contamination and lower cost than the conventional ones. In modalities Preferred, cell sorting is done on a microfabricated chip - ^ - - - - - «•• * -i-. **** * £ **, * with a detection volume of about 1 to 1,000,000 femtoliters (fl), preferably about 200 to 500 fl, and most preferably about 375 fl Classification occurs immediately after detection In one particular mode, the input and collection receptacles are incorporated in the same chip. The classifiers of the invention can function as devices that operate on their own, or as components of integrated microanalytical chips, and can be disposable. Living cells with a distinctive feature, such as E. coli cells expressing a fluorescent protein, can be efficiently separated from cells lacking this characteristic. Additionally, the cells remain viable after being removed from the classification device. An advantage of the invention is that it can be applied to various aspects of chemical and biological studies, for example, cell classification, enzymatic catalysis and molecular evolution (1). The references cited herein are referred to numerically, and are appended to the Bibliography below. All references are incorporated in their entirety. Harrison et al. (39) describe a microfluidic device, which manipulates and stops the flow of fluids through a microfabricated chip, so that a cell can be observed after it interacts with a chemical agent. The cells and the chemical agent are loaded into the device via two different input channels, which intersect with a main flow path. The flow of the fluid is controlled by a pressure pump or by electric fields (electrophoretic or electro-osmotic) and can be stopped, so that the cells can be observed, after they mix and interact with the chemical. The cells then pass through the main flow path, which ends in a simple common waste chamber. Harrison et al. they do not provide a device or method for classifying cells, nor do they suggest or motivate someone who has ordinary skill in the art to make and use any such device. On the contrary, the cells are mixed with chemicals, observed and discarded as waste. Conventional flow cell classifiers, such as FACS, are designed to have a flow chamber with a nozzle and use the principle of hydrodynamic focusing with case flow to separate or classify biological material, such as cells (2-7). In addition, most sorting instruments combine ink-jet writing technology and the effect of gravity to achieve high-speed droplet and electric charge generation (8 -10). Despite these advances, many failures of these instruments are due to problems in the flow chamber. For example, orifice clogging, particle adsorption and contamination in the pipeline can cause a turbulent flow in the jet stream. These problems contribute to the enormous variation in illumination and detection in conventional FACS devices. Another major problem is known as sample remanence, which occurs when remnants of previous specimens left in the channel retro-flow into the current ..... a, ..... J ^. .... «- -" .. - i- > i * m * to, 3 * .. * -. new sample during consecutive runs. A potentially more serious problem occurs when the dyes remain in the pipe and the chamber, which can give false signals for the detection of fluorescence or light scattering apparatus. Although such systems can be sterilized between runs, it is expensive, slow, and inefficient, and results in inactive machine hours for blanching and sterilization procedures. Similarly, each cell, as it passes through the hole, can generate a different disturbance in response to the formation of droplets. The larger cells can possibly fall into the droplet pattern, the non-spherical cells they tend to align with the long axis parallel to the flow axis, and the deformable cells may elongate in the direction of flow (9, 10). This may result in some variation in the analysis time to the actual classification event. Additionally, a number of technical problems can make it difficult to generate droplets loaded identically, which increases the error of deviation. A charged droplet can cause the next droplet of the opposite polarity to have a reduced charge. On the other hand, if consecutive droplets are charged identically, then the first droplet could have a smaller potential than the second droplets, and so on consecutively. Still, charged droplets will have a defined path only if they are loaded identically. In addition, increasing droplet charges can cause mutual electrostatic repulsion between adjacent droplets, which also increases the error of deviation. Other factors, such as, the very high cost even for a conventional FACS team «. t. i -jS ». -. modest (in the order of $ 250,000), the high cost of maintenance and the requirement for personnel trained to operate and maintain the equipment have been among the main considerations that obstruct this technology and its accessibility and use disseminated (10). Even though the field of flow cytometry has been extensively exploited in the development of cell sorting devices, significant problems remain unresolved. In this way, there is a need for improved methods and machines for classifying cells that are fast, efficient, cost effective and disposable.

BRIEF DESCRIPTION OF THE INVENTION The invention provides a microfabricated device for classifying cells based on a desired characteristic, for example, cells marked with reporter can be classified by the presence or reporter level in the cells. The device includes a chip having a substrate, in which at least one analysis unit is microfabricated. Each analysis unit includes a main channel, which has a sample input channel, usually at one end, and a detection region along its length. Adjacent and downstream of the detection region, the main channel has a discriminating region or branching point leading to at least two branching channels. The analysis unit may further include additional input channels, detection points, branch points and branch channels as desired. A stream containing the cells, for example, in a solution or mixture, is passed through the region of detection, so that on average only one cell occupies the detection region at any given time. The cells can be drawn based on their ability to emit a detectable signal, such as an optical signal, with or without stimulation, such as exposure to light in order to promote fluorescence. According to the invention, the presence or reporter level of each cell is measured within the detection region, and each cell is directed to a branching channel selected based on the detected or measured reporter level. In addition to classifying fluorescent and non-fluorescent cells, the invention can also provide multiparameter analysis, such as multi-color detection or a gate window detection. For example, beads of different colors, or cells marked with one or more chromophores, can be classified by the invention. Classification according to a window, or threshold, means that the cells or particles are selected by classification based on the presence of a signal above a certain value or threshold, and that is usually less than a certain upper limit. There may also be several analysis points on the same chip for multiple measurements over time. The invention offers several advantages over traditional cassette flow methods. Because the channels in the present device can be made with micron dimensions, the volume of the detection region is controlled precisely and there is no need for a hydrodynamic approach. The flat geometry of the device allows the use of optimal high numerical aperture, thereby increasing the sensitivity of the system. Because the fluid flows continuously through the ^ uUUH ^^ | f ^ || te system, there is no need for droplet formation, or droplets charged, and many challenging technical issues can be avoided. In addition, there is no aerosol formation because the system is completely self-contained, allowing a much safer classification of biohazardous material, compared to conventional FACS devices. The sorting device in the invention is also disposable, which obviates the need to clean and sterilize the instrument, and prevents cross-contamination between samples. Thus, a cellular classifier of the invention, like a disposable microfabricated FACS, employs a substrate that it integrates at least one inlet channel and at least two outlet channels, which meet a sorting or branching point. In a preferred embodiment, the substrate is planar and contains a microfluidic chip made from a silicone elastomer print of a glass disc. silicon engraved according to replication methods in soft lithography (11). In one embodiment, the channels meet to form a "T" (union T). You can also use a Y-shaped joint, and other shapes and geometries. A detection region is usually upstream of the branch point. The cells are diverted towards one or another output channel based on a predetermined characteristic that is evaluated as each cell passes through the detection region. The channels are sealed, preferably, to contain the flow, for example, by attaching a transparent cover strip, such as glass, on the chip, to cover the channels while allowing optical examination of one or more channels or regions. , particularly the detection region. In a preferred fashion, the cover strip is pyrex, anodically bonded to the chip. In a fashion, the cells are directed towards one or the other of a pair of output channels by electrodes that apply an electric field through the branch point, which effectively directs a particular cell to an outlet or channel. Default branching In another embodiment, a flow of cells is maintained through a device via a pump or pressure differential. A valve structure at the branch point allows each cell to enter the branch channels independently, depending on the measurement at the point of detection. In a similar embodiment, a valve structure can be provided for each ramification channel, downstream of the branch point, which allows or reduces the flow through a particular channel. Alternatively, the pressure can be adjusted within or at the outlet of each ramification channel, to allow or reduce flow through the channel. An apparatus, machine or device of the invention may include a plurality of analysis units, and in such embodiments, a plurality of manifolds may also be included (eg, an accessory or point with more than one lateral outlet to allow connection). of, or division to, branching channels). The number of collectors normally equals the number of branching channels in a unit of analysis, to facilitate the collection of cells from the corresponding branch channels of the different units of analysis. . -,? .t. > ..--. ....., ..- ...... 5 = ^ 1 -: t-te- - j ^ a_ ^ - ^ ¿^ aa ^ - -tiÉ aifaaS-.

The microfabricated device includes a transparent cover strip (for example, glass), attached to the base layer and which covers the channels to form a "roof" and / or "floor" for the channels. A silicon chip can be used with an anodically bonded pyrex cover strip. The channels in the device are preferably between about 1 and 500 microns wide and between about 1 and 500 microns deep, and the detection region has a volume between about 1 fl and 1 00 ni. Where you want, an external laser, an integrated semiconductor diode or laser, or a high intensity lamp (eg, a mercury lamp) can be used to stimulate the reporter to release a measurable or detectable signal (eg, energy of light). The measurements can be taken, for example, using a microscope in connection with an intensified charge coupling device (CCD) camera, photomultiplier tube, avalanche photodiode, an integrated photodiode or similar. In another as, the invention includes a method for isolating cells by having a threshold amount selected from an optically detectable reporter (e.g., color change, ultraviolet or fluorescent) attached or associated. The method includes, (a) flowing a stream of solution containing reporter-labeled cells through a channel comprising a detection region having a selected volume, where the concentration of the cells in the solution is such that they pass through. the detection region one by one, (b) determine the presence or quantity of reporter in each cell as it passes to IMIMI MM, through the detection region, (c) diverting cells having a selected threshold of reporter towards a first branching channel, and diverting the cells that do not have the selected threshold towards a second branching channel, and (d) ) collect the deviated cells to one or 5 more branching channels. The method can be applied to divert a cell having a selected reporter threshold towards the first branch channel, in such a way that the deviation action blocks the flow to the second branch channel. That is, the second channel is blocked and The current carries the cell having the selected reporter threshold towards the first branch channel. Alternatively or additionally, the method can be used to divert a cell that does not have the selected reporter threshold to the second branch channel, block the flow to the first branch channel. This can for example, using a valve or valves that are actuated by an electrical or mechanical switch in response to a reporter measurement. The method can be applied to any cell, including prokaryotic or eukaryotic, such as bacterial, plant, animal and the like.

The method is particularly useful for the classification of mammalian (eg, human) blood cells, such as peripheral blood mononuclear cells (PBMCs), based on the expression of several antigens, such as HLA DR, CD3, CD4 , CD8, CD11a, CD11c, CD14, CD16, CD20, CD45, CD45RA, CD62L, etc. The method also can be used to classify any cell based on whether it expresses or it produces a detectable protein, either directly or in cooperation with a reporter molecule. For example, cells that produce a fluorescent protein can be classified from those that do not. Alternatively, a fluorescent protein can be used as a reporter, for example, by co-expression with another protein (50). , 51). Alternatively, the cell can produce a detectable substance (e.g., a fluorescent compound) through its interaction with another substance added to the fluid medium. For example, cells that contain a gene for a monooxygenase enzyme can catalyze a reaction on an aromatic substrate (eg, benzene or naphthalene) with the net result that fluorescence, or other detectable property of the substrate, will change This change can be detected in the detection region, and the cells that have that change in fluorescence can be collected based on predetermined criteria. A second reagent or coupling enzyme can be used to enhance fl uorescence. See, Affholter and Arnold (50) and Joo et al. (51).

BRIEF DESCRIPTION OF THE DIAMETERS FIGS. 1 A to 1 D show the steps in the photolithographic microfabrication of a cell sorting device from a silicon disk, using photolithography and several etching steps. FIG. 2A shows one embodiment of a detection region used in a cell classification device, having an integrated photodiode detector; FIG. 2B shows another modality of a detection region, having an integrated photodiode detector, and which provides a greater detection volume than the mode of FIG.2A. FIGS. 3A and 3B show a mode of a valve within a branch channel of a cellular sorting device and steps 5 in the manufacture of the valve. FIG. 4A shows a modality of a discrimination region and associated channels used in a cell sorting device, having electrodes disposed within the channels for electrophoretic discrimination; FIG. 4B shows another embodiment having electrodes 10 arranged for electro-osmotic discrimination; FIGS. 4C and 4D show two additional modalities, having valves arranged for electrophoretic separation by pressure, where the valves are within the branching point, as shown in FIG. 4C, or within branching channels, as shown in FIG.4D. 15 FIG. 5 shows a device with analysis units containing a detection cascade and discrimination regions suitable for successive rounds of cell sorting. FIG. 6 is a photograph of an apparatus of the invention, showing a chip with an inlet and reservoir channel, a detection region, a branch point and two outlet channels with reservoirs. FIG. 7 shows a schematic representation of a process for obtaining a silicone elastomer impression of a silicon mold, to provide a microfabricated chip according to the invention.

FIG. 8 shows a schematic representation of an apparatus of the invention, in which a chip of silicone elastomer is mounted on an inverted optical detection of a reporter stimulated laser microscope. The electrodes are used to direct cells in response to the microscope detection FIG. 9 shows the results of the classification of blue and red fluorescent beads having an initial ratio of 10: 1, respectively, using a forward mode. The darker bar represents the proportion of red pearls over the total number of pearls classified and the lighter bar represents the proportion of blue pearls over the total number of pearls classified. FIG. 10 shows the results of classifying blue and red fluorescent beads having an initial ratio of 100: 1, respectively, using a reversible change mode. The darker bar represents the proportion of red pearls over the total number of pearls sorted and the lightest bar represents the proportion of blue pearls over the total number of pearls classified. FIG. 11 shows the results of classifying green and red fluorescent beads having an initial ratio of 100: 1, respectively, 20 using a reversible mode of change. The darker bar represents the proportion of pearls over the total number of pearls classified and the lightest bar represents the proportion of green pearls over the total number of pearls classified. FIG. 12 shows the results of classifying wild type E. coli 25 HB101 cells (non-fluorescent) and E. coli HB101 cells that M ?? ?? Lk l¡l¡ ^ ?? ^ '^ mTim itwai í- expressing green fluorescent protein (GFP) having an initial ratio of 100: 1, respectively, using a forward shift mode. The lightest bar represents the proportion of natural type E. coli cells on the total number (approximately 120,000) of sorted cells and 5 the darkest bar represents the proportion of E. coli cells expressing GFP on the total number of cells classified. FIG. 13 shows the results of classifying cells of wild-type E. coli HB101 (non-fluorescent) and E. coli HB101 cells expressing green fluorescent protein (GFP) having an initial ratio of 10: 3, respectively, using a mode of change forward. FIGS. 14A and B show a classification scheme according to the invention, in diagrammatic form. FIGS. 15A and B show a reversible classification scheme according to the invention. 15 DETAILED DESCRIPTION OF THE INVENTION Definitions The terms used in this specification generally have their ordinary meanings in the art, within the context of the invention, and 20 in the specific context where each term is used. Certain terms are discussed below, or elsewhere in the specification, to provide additional guidance to the practitioner to describe the devices and methods of the invention and how to make use of them. For convenience, certain terms are highlighted, for example, using italics 25 and / or quotes. The use of highlighting has no influence on the scope and I ^^^ j ^^^^ g ^^ g ^^ I ^ II ^^^^^^^^^^^^^^^^^^^ 1 ^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ H ^^^^^^^^^^^^^^ ^^^^^^^^^^^^^ g ^ meaning of a term; The scope and meaning of a term is the same, in the same context, whether it is highlighted or not. It will be appreciated that the same can be said in more than one way. Accordingly, alternative language and synonyms may be used for any one or more of the terms discussed herein, and neither should special importance be given to whether a term is elaborated or discussed or not herein. Synonyms are provided for certain terms. An exposure of one or more synonyms does not exclude the use of other synonyms. The use of examples anywhere in this specification, exemplifying examples of any term discussed herein, is uniquely illustrative and in no way limits the scope and meaning of the invention or any exemplified terms. In like manner, the invention is not limited to the preferred embodiments. As used herein, "cell" means any cell or cells, such as a virus or any other particle having a microscopic size, for example, a size that is similar to that of a biological cell, and includes any prokaryotic or eukaryotic cell, for example, cells of bacteria, fungi, plants and animals. The cells are usually spherical, but they can also be elongated, flattened, deformable and asymmetric, that is, not spherical. The size or diameter of a cell normally varies from about 0. 1 to 1 20 microns, and is usually from about 1 to 50 microns. A cell can be alive or dead. Because the microfabricated device of the invention is directed to classify materials having a size similar to a biological cell (for example, approximately 0.1 to 120 microns), any material having a size similar to a biological cell can be characterized and classified using the microfabricated device of the invention. Thus, the term cell should also include microscopic beads (such as chromatographic and fluorescent beads), liposomes, emulsions or any other encapsulating biomaterial and porous materials. Non-limiting examples include latex, glass or paramagnetic beads; and vesicles, such as emulsions and liposomes, and other porous materials, such as silica beads. Beads that vary in size from 0.1 microns to 1 mm may also be used, for example to classify a library of compounds produced by combination chemistry. As used herein, a cell may be charged or uncharged. For example, charged beads can be used to facilitate flow or detection, or as a reporter. Biological cells, alive or dead, can be loaded, for example, by using a surfactant, such as SDS (sodium dodecyl sulfate). A "reporter" is any molecule, or a portion thereof, that is detectable or measurable, for example, by optical detection. In addition, the reporter is associated with a cell or with a marker or particular characteristic of the cell, or is detectable by itself, to allow identification of the cell. Such a label includes antibodies, proteins and portions of sugars, receptors, polynucleotides and fragments thereof. The term "brand" can be used interchangeably with "reporter". The reporter is usually a dye, fluorescent, ultraviolet or chemiluminescent agent, ***?? A ** aiA? , - ^ ?? ^ ^ A ^, chromophore or radio-mark, any of which can be detected with or without some kind of stimulating event, for example, fluoresce with or without a reagent. In one embodiment, the reporter is a protein that is optically detectable without a device, eg, a laser, to stimulate the reporter, such as horseradish peroxidase (HRP). A protein reporter may be expressed in the cell to be detected, and such an expression may be indicative of the presence of the protein, or may indicate the presence of another protein that may or may not be co-expressed with the reporter. A reporter can also include Any substance on or in a cell, which causes a detectable reaction, for example, by acting as a starting material, reagent or a catalyst for a reaction that produces a detectable product. Cells can be classified, for example, based on the presence of the substance, or on the ability of the cell to produce the product detectable when the reporting substance is provided. A "marker" is a feature of the cell that is made detectable by the reporter, or it can be co-expressed with a reporter. The characteristics may include a protein, including enzyme, receptor and ligand proteins, saccharides, polynucleotides and combinations of the same, or any biological material associated with a cell. The product of an enzymatic reaction can also be used as a marker. The marker can be associated directly or indirectly with the reporter or it can be a reporter by itself. The term "flow" means any movement of liquid or solid to through a device or in a method of the invention, and encompasses limiting any fluid stream, and any material that moves with, in or against the current, whether or not the material is carried by the current. For example, the movement of cells through a device or in a method of the invention, for example, through channels of a microfluidic chip of the invention, comprises a flow. This is so, according to the invention, whether or not the cells are carried by a fluid stream further comprising a flow, or whether the cells are caused to move by some other direct or indirect force or motivation, and whether the nature of any force motivating be known or understood or not The application of any force can be used to provide a flow, including without limitation, pressure, capillary action, electro-osmosis, electrophoresis, dielectrophoresis, optical cursors and combinations thereof, without considering any theory or particular mechanism of action, always and when the cells are targeted for classification according to the invention. An "input region" is an area of a microfabricated chip that receives cells for classification. The entrance region can contain an entrance channel, a receptacle or deposit, an opening and other features that facilitate the entry of cells into the device. A chip can contain more than one input region if desired. The inlet region is in fluid communication with the main channel and is upstream thereof. An "exit region" is an area of a microfabricated chip that collects or dispenses cells after classification. A region of i rii i 'r? rt < "? 'rrr? | -1Ül? r W iif II 11 r. downstream of a discrimination region, and may contain branch channels or output channels A chip may contain more than one output region if desired.A "analysis unit" is a microfabricated substrate, for example, a microfabricated chip having at least one input region, at least one main channel, at least one detection region and one branch point, forming at least two branch channels and two output regions A device according to the invention may comprise a plurality of analysis units 10 A "main channel" is a channel of the chip of the invention, which allows the cell flow to pass a detection region and towards a discrimination region for classification. manufactured to the main channel, the main channel is normally in communication of fluid with an inlet channel or region of entry, which allows the flow of cells to the main channel. The main channel is also normally in fluid communication with branching channels, outlet channels or waste channels, each of which allows the flow of cells out of the main channel. 20 A "detection region" is a location within the chip, usually within the main channel where the cells to be classified are examined to classify based on a predetermined characteristic. In a preferred embodiment, the cells are reexamined one at a time, and the characteristic is detected or measured optically, for example, by testing the presence or quantity of a reporter For example, the detection region is in communication with one or more microscopes, diodes, light-stimulating devices (e.g., lasers), photomultiplier tubes, and processors (e.g., computers and computer programs), and combinations thereof. , which cooperate to detect a signal representative of a cell characteristic or reporter, and to determine and direct the sorting action to the discrimination region. The detection region is in fluid communication with a discrimination region and is in, near, or upstream of, the discrimination region. A "discrimination region" or "branch point" is a junction of a channel where the cell flow can change the direction to enter one or more other channels, for example, a branch channel, depending on a signal received at connection to an examination in the detection region. Normally, a discrimination region is monitored and / or is under the control of a detection region, and therefore, a discrimination region may "correspond" to that detection region. The discrimination region is in communication with and is influenced by one or more classification techniques or flow control systems, for example, electric, electro-osmotic, (micro-) valve, etc. A flow control system can employ a variety of classification techniques to change or direct the flow of cells to a predetermined branching channel. A "branching channel" is a channel that is in communication with a discrimination region and a main channel. Normally, a branching channel receives cells depending on the characteristic of interest i *,. . ,, .. . . ... j.-... *. .. . - _ "- r .., - ..... t. i »^. ~. ¿¿* ~," of the cell as detected by the detection region and classified in the discrimination region. A branch channel can be in communication with other channels to allow additional classification. Alternatively, a branching channel may also have an outlet region and / or terminate with a receptacle or reservoir to allow collection or disposal of the cells. The term "forward classification" describes a flow of a cell address, typically from an input region (upstream) to an output region (downstream) and preferably without a change in direction, for example, which is oppose the flow "forward". Preferably, the cells travel forward in a linear manner, i.e., single row. The preferred "forward" sorting algorithm consists of running the cells from the input channel to the waste channel, until a cell fluorescence is above a pre-set threshold, at which point the voltages are temporarily changed to divert it to the collection channel. The term "reversible classification" describes a movement or flow that can change, ie, inverse direction, for example, from a forward direction to an opposite backward direction. In other words, a reversible classification allows a change in the direction of flow from a downstream to an upstream direction. This can be useful for classifying more precisely, for example, by allowing confirmation of a classification decision, selection of the particular branch channel, or correcting an improperly selected channel. '' '"" * "* ...... i» ..- ... -.,. i, - «fe- Different algorithms to be used in the microfluidic device can be implemented by Different programs, for example, under the control of a personal computer Different algorithms can be implemented to classify the microfluidic device through different programs, for example, under the control of a personal computer. For example, a scheme changed by pressure in the electro-osmotic flux should be considered.The electro-osmotic change is virtual and instantaneous and the performance is limited by the voltage that can be applied to the classifier ( which also affects the running time through the effects of ion suppression.) A pressure changed system does not require high voltages and is more robust for longer runs.However, the mechanical deformation in the system is probably to cause the veil change of fluid becomes limiting of the proportion with the "forward" classification program. Because the flux is low Reynolds number and is completely reversible, when it comes to separating rare cells, one can implement a classification algorithm that is not limited by the intrinsic rate of change of the device. The cells flow at the highest speed static (no change) possible from the entrance to the waste. When an interesting cell is detected, the flow is stopped. As the flow stops, the cell may be beyond the junction and partly below the waste channel. The system is then returned at a slow (changeable) speed from the waste to the inlet, and the cell is changed to the collection channel when it passes through the region of detection. At that point, the cell is saved and the device can run at high speed in the forward direction again. This "reversible" sorting method is not possible with standard FACS machines and is particularly useful for collecting rare cells or making 5 multiple measurements over time of a single cell.

Cellular classifier architecture and method A cellular classifier according to the invention comprises at least one analysis unit having an input region in Communication with a main channel, a detection region within or coincident with a portion of the main channel, a detector associated with the detection region, a discrimination region or branch point in communication with the main channel and with branch channels, and a flow control that responds to the detector. The channels of branching can each lead to an outlet region and to a receptacle or reservoir. The entrance region can also communicate with a receptacle or deposit. As each cell passes into the detection region, it is examined by a predetermined characteristic (ie, using the detector), and a corresponding signal is produced, for example, that indicates "yes" if the characteristic is present, or "no" if it is not. In response to this signal, a flow control can be activated to divert a cell or cells to a branch or other channel. In this way, a cell or cells within the discrimination region can be classified into an appropriate branching channel according to a signal produced by the corresponding examination in the region of ^^^ ^ ^ ^ ^ ^ ^ ^^^^^^^ ri ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ detection. Optical detection of the features is preferred, for example, directly or through the use of a reporter associated with a feature chosen for classification. However, other detection techniques can also be employed. A variety of channels for sample flow and mixing can be microfabricated on a simple chip and can be placed at any location on the chip such as detection and discrimination or classification points, for example, for kinetic studies (12, 14). A plurality of analysis units of the invention can be combined in one device. Microfabrication applied according to the invention eliminates the dead time that occurs in conventional flow cytometric kinetic studies and achieves a better time resolution. Additionally, the linear arrays of the channels in a simple chip, that is, a multiplex system, can simultaneously detect and classify a sample by using an array of photomultiplier tubes (PMT) for parallel analysis of different channels (15). This arrangement can be used to improve performance or enrich the successive sample, and can be adapted to provide very high performance for microfluidic devices exceeding the capacity allowed by conventional flow classifiers. Moreover, microfabrication allows other technologies to be integrated or combined with flow cytometry on a simple chip, such as PCR (21), moving cells using optical cursors / cell traps (16-18), cell transformation by electroporation (19), μTAS (22), and DNA hybridization (6). The detectors , MM_ > _ ^ ^ »_ M * > _ ^ ÍMÍu ^ _ti_ * UM > MY > ? ti?? i and / or light filters that are used to detect cell characteristics or reporters can also be manufactured directly on the chip. A device of the invention can be microfabricated with a reservoir or sample solution receptacle in the inlet region, which is normally in fluid communication with an inlet channel. A deposit can facilitate the introduction of cells into the device and into the sample input channel of each analysis unit. An inlet region may have an opening, such as in the floor of the microfabricated chip, to allow entry of the cellar sample into the device. The inlet region may also contain a connector adapted to receive a suitable piece of tubing, such as HPLC tubing or liquid chromatography, through which a sample may be delivered. Such an arrangement facilitates introducing the sample solution under positive pressure, in order to achieve a desired flow rate through the channels. Receptacles and output channels can be similarly provided Substrate and flow channels A standard analysis unit of the invention comprises an input region that is part of, and is fed or communicated with, a main channel, which in turn communicates with two (or more) branching channels in a union or branch point, forming, for example, a T-shape or a Y-shape. Other shapes and channel geometries may be used as desired. The region in or around the union may also be referred to as a region of discrimination, however, it is not .. ....... .... ^ .......... ^. .... ^. i ^ > .. i - .. *., ". n ~?. * ", * ,, J,":, - ,,. ,,,. "-,., - í,. *., j,. ^ jtam * ¿k * require precise limits for the region of discrimination. A detection region is identified within or coincident with a portion of the main channel downstream of the entry region, and in or upstream of the discrimination region or branch point. 5 precise limits are not required for the detection region, but are preferred. The discrimination region may be located immediately downstream of the detection region, or it may be separated by an appropriate distance. Preferably, the distance between the detection and discrimination regions of single cells at high rates of change, in response to the examination of cells in the detection region. It will be appreciated that the channels can have any suitable shape or cross section, such as tubular or corrugated, and can be arranged in a suitable manner, provided that a cell flow can be directed from one channel to at least one of two or more channels of 15 branch. The channels of the invention are microfabricated, for example, when recording a silicon chip using conventional photolithography techniques, or using a microlabor technology called "soft lithography", developed in the late 1990s (11). These and other microfabrication methods 20 can be used to provide inexpensive miniaturized devices, and in the case of soft lithography, they can provide robust devices having beneficial properties, such as improved flexibility, stability and mechanical strength. When optical detection is employed, the invention also provides a minimum light scattering of the cell suspension and the material of the ^ »» .i? atm? üLt .J-i. . . ,. -. * .. . The devices according to the invention are relatively inexpensive and easy to establish, and can also be disposable, which alleviates enormously many of the concerns of sterilization and permanent adsorption of particles to the flow chambers and channels of conventional FACS machines Using these kinds of techniques, microfabricated fluidic devices can replace the conventional fluid flow chambers of the prior art. A microfabricated cell sorting device of the invention is preferably manufactured from a silicon microchip or silicon elastomer. The dimensions of the chip are those of the normal microchips, ranging from approximately 0.5 cm to approximately 5 cm per side and approximately 1 miera to approximately 1 cm in thickness. The device contains at least one analysis unit that contains a main channel having regions of detection and discrimination. Preferably, a device also contains at least one input region (which may contain an input channel) and two or more output regions (which have fluid communication with a branch channel in each region). It should be appreciated that the "regions" and "channels" are in communication They flow with each other, and therefore may overlap, that is, there may be a clear boundary where a region or channel begins or ends. A microfabricated device can be covered with a material having transparent properties, for example, a glass cover strip to allow the detection of a reporter, for example, by a optical device, such as an optical microscope.

The d imensions of the channels and in particular of the detection region are influenced by the size of the cells under study. These cells can be quite long by molecular standards. For example, mammalian cells may have a diameter of about 1 to 50 microns, more usually 1 to 30 meters, although fat cells may be larger than 1 20 feet, and plant cells are generally from 1 0 to 1 00 microns. Accordingly, the detection regions used to detect cells in this size range have a sufficiently large cross section to allow a desired cell to pass through if not substantially encouraged in relation to the flow of the sol ution that the portal. To avoid "bottlenecks" and / or turbulence, and promote single row flow, the dimensions of the channels, particularly in the detection region, should generally be at least about twice, preferably at least about five times. as large per side or in diameter as the diameter of the largest cell that will be passed through the la. A microfabricated device of the invention is adapted to handle particles on the scale of cell size, and is dependent on the dimensions of the microfabricated channels, detection and discrimination regions. Specifically, the channels in a device are typically between about 2 and 500 meters wide, and between about 2 and 500 microns deep, to allow an ordered cell flow in the channels. Similarly, the volume of the detection region in a cell sorting device • ~ A ~ - '^ "* • * is usually in the range of between about 1 femtoliter (fl) and 1 nanoliter (ni) .To prevent the cells from sticking to the sides of the channels, the channels (and of cover, if used) may have a coating, which minimizes cell adhesion Such a coating may be intrinsic to the material from which the device is manufactured, or may be applied after the structural aspects of the channels have been microfabricated "TEFLON" is an example of a coating that has suitable surface properties 10 A silicon substrate containing microfabricated flow channels and other components is preferably covered and sealed, most preferably with a transparent cover, eg thin glass or quartz, although other transparent or opaque cover materials may be used When the sources are used or external radiation detectors, the detection region is covered with a transparent cover material to allow optical access to the cells. For example, the anodic bond to a cover strip of "PYREX" can be achieved by washing both components in an aqueous H2SO4 / H2O2 bath by rinsing in water, and then, for example, by heating approximately 350 ° C, while a voltage of 450V is applied.

Change and flow control The electro-osmotic and pressure-driven flow are examples of methods or systems for flow control, that is, it manipulates the flow of cells, particles or reagents in one or more directions and / or in one or more »» .. ^ Aaa? & fc..i., * ... channels of a microfluidic device of the invention (8, 12, 13, 23). Other methods can also be used, for example, electrophoresis and dielectrophoresis. In certain embodiments of the invention, the flow moves in a "forward" direction, for example, from the entry region through the main and branch channels to an exit region. In other modalities, the direction of flow is reversible. The application of these techniques according to the invention provides faster and more precise devices and methods for classifying cells, for example, because the classification occurs in or within a region of discrimination that can be placed in or immediately after a detection region. This provides a shorter distance for cells to travel, cells can move more quickly and with less turbulence, and can move, be examined and more easily sorted in a single row, that is, one cell at a time. In a reversible mode, potential classification errors can be avoided, for example, by reversing and encouraging the flow to re-read or re-classify a cell or cells before irretrievably delivering the cell or cells to a particular branch channel. . Without being bound by any theory, it is believed that electro-osmosis 20 produces movement in a current containing ions, for example, a liquid such as a buffer, by application of a voltage differential or charge gradient between two or more electrodes. Neutral cells (uncharged) can be transported by the current. Electro-osmosis is particularly suitable for rapidly changing the course, direction or flow velocity. It is believed that electrophoresis ^ > ^. ". ^, Sah3Mi > J-.j? Fascia £ ^ .... ..... ... ...... - ..... -_.... J_ ^ .. ..._ .. ». ._, ...... i. ^ i. ^. ^.

It produces the movement of objects charged in a fluid to one or more electrodes of opposite charge and away from one or more charge electrodes sim i la r It is believed that dielectrophoresis produces the movement of dielectric objects 5 They have no net charge, but they have reg ions that are charged positively or negatively in relation to each other. Alternatively, inhomogeneous electric fields in the presence of particles, such as cells or beads, cause them to become electrically polarized and thus experience electrodephoretic forces.

Depending the dielectric polarizability of the particles and the suspension medium, the dielectric particles will move either to the regions of high field strength or low field strength. For example, the polarizability of living cells depends on their composition, morphology and phenotype, and is highly dependent on the frequency of the electric field. applied. Thus, cells of different types and in different physiological states generally have distinctly different dielectric properties, which can provide a basis for cell separation, for example, by differential dielectrophoretic forces. According to formulas provided in Fiedler et al. (1 3), 0 simple individual particle manipulation requires field differences (inhomogeneities) with dimensions close to the particles. The manipulation is also dependent on the permittivity (a dielectric property) of the particles with the suspension medium. In this way, polymer particles and living cells show negative electrophoresis at high field frequencies in water. For example, the forces j? g ^^? ^^ j * ^^^^^ dielectroforéticas experienced by a sphere of latex in a field of 0.5 MV / m (10V for an opening of electrode of 20 microns) in water, is predicted to be approximately 0.2 piconewtons (pN) for a latex sphere of 3.4 microns at 15 pN for a 15 micron latex sphere (13). 5 These values are mostly greater than the hydrodynamic forces experienced by the sphere in a current (approximately 0.3 pN for a sphere of 3.4 microns and 1.5 pN for a sphere of 15 microns). Therefore, the manipulation of individual cells or particles can be achieved in a flow fluid, such as in a sorting device of cells, using dielectrophoresis Using conventional semiconductor technologies, the electrodes can be microfabricated on a substrate to control the force fields in a microfabricated classification device of the invention. Dielectrophoresis is particularly suitable for moving objects that are electrical conductors. The use of AC current is preferred to prevent permanent ion alignment. The megahertz frequencies are suitable to provide a net alignment, attractive force and movement over relatively long distances. See, for example, Benecke (49). Optical tongs can also be used in the invention to trap and move objects, e.g., cells, with focused light beams, such as lasers. The flow can also be obtained and controlled by providing a pressure differential or gradient between one or more channels of a device or in a method of the invention. 25 Detection and discrimination for classification The detector can be any device or method for interrogating a cell as it passes through the detection region. Normally, the cells are to be classified according to a predetermined characteristic that is directly or indirectly detectable, and the detector is selected or adapted to detect that characteristic. A preferred detector is an optical detector, such as a microscope., which can be coupled with a computer and / or other image intensification or processing devices, to process images or information produced by the microscope using known techniques. For example, cells can be classified either by containing or producing a particular protein, by using an optical detector to examine each cell for an optical indication of the presence or quantity of that protein. The protein may be detectable by itself, for example, by a characteristic fluorescence, or it may be labeled or associated with a reporter that produces a detectable signal when the desired protein is present, or is present at least in a quantity of um bral. . There is no limit to the number of cell characteristics that can be identified or measured using the techniques of the invention, which includes, without limitation, cell surface characteristics and intracellular characteristics, provided that the characteristic or characteristics of interest for classification, they can be identified and detected or measured enough to distinguish the cells having the desired characteristics from those that do not. For example, any brand or reporter as described herein, may be used dtfßMIÜ? MÉi as the basis for classifying cells, that is, detecting cells to be collected. In a preferred fashion, the cells are separated based on the intensity of a signal from an optically detectable reporter linked to, or associated with, the cells as they pass through a detection window or "detection region" in the device. Cells that have a quantity or level of the reporter at a selected threshold or within a selected range are diverted into a predefined branch or output channel of the device. The reporter signal is collected by a m icroscope and med a by a photom u ti onist (PMT). A com puter d ig italizes the signal of PMT and controls the flow via valve action or osmotic potentials. In one embodiment, the chip is mounted in an inverted optical microscope. The fluorescence produced by a reporter is cut using a laser beam focused on cells passing through a detection region. Fluorescent reporters include, for example, rodam ina, fluorescein, red Texas, Cy 3, Cy 5, and phycobiliprotein. Thus, in one aspect of the invention, the classification device can classify cells based on the level of expression of cell markers selected, such as cell surface markers, which have a detectable reporter bound thereto, in a manner similar to that currently employed using fluorescence activated cell sorting machines (FACS). Proteins or other characteristics within a cell, and which do not necessarily appear on the cell surface, they can also be identified and used as a basis for classification. The sorted cells can be collected from the output channels and can be used as needed. To determine whether a cell has a desired characteristic, the detection region may include an apparatus for stimulating a reporter for that characteristic to emit measurable light energy, for example, a light source, such as a laser, laser diode, lamp high intensity (for example, mercury lamp) and the like. In modalities where the lamp is used, the channels are preferably protected from light in all regions except the detection region. In modalities where a laser is used, the laser can be set to scan through a set of detection regions from different units of analysis. In addition, laser diodes can be microfabricated on the same chip that contains the analysis units. Alternatively, the laser diodes can be incorporated into a second chip (ie, a laser diode chip) which is placed adjacent to the microfabricated classifier chip, so that the laser light of the diodes shines on the or Detection regions. In preferred embodiments, an integrated semiconductor laser and / or a photodiode detector integrated in the silicon disk in the vicinity of the detection region are included. This design provides the advantages of solidity and a shorter optical path for excitation and / or emitted radiation, thus minimizing distortion.

Classification schemes According to the invention, the cells are classified dynamically in a flow stream of microscopic dimensions, based on the detection or measurement of a characteristic, marker or reporter that is associated with the cells. The current is normally, but not necessarily, continuous, and can be stopped and initiated, inverted or changed in speed. Prior to sorting, a liquid that does not contain cells can be introduced into an input region of the chip (eg, from a receptacle or input channel) and is directed through the device by capillary action, to hydrate and prepare the device for classification. If desired, the pressure can be adjusted or equalized, for example, by adding shock absorber to an outlet region. The liquid is usually an aqueous buffer solution, such as ultrapure water (for example, resistivity of 18 mega ohms, obtained for example, by column chromatography), water ullrapure, Tris HCl 10 mM and 1 mM EDTA (TE), phosphate buffer saline (PBS) and acetate buffer. Any liquid or buffer that is physiologically compatible with the population of cells to be classified can be used. A sample solution containing a mixture or population of cells in a suitable carrier fluid (such as a liquid or buffer described above), is delivered to the inlet region. Capillary force causes the sample to enter the device. The force and direction of flow can be controlled by any desired method to control the flow, for example, by differential pressure, by valve action or by electro-osmotic flow, for example, produced by electrodes in the , - ",.-..., ah» ri "a?» »To channels of entry and exit. This allows the movement of the cells towards one or more branching channels or desired output regions. A "forward" classification algorithm, according to the invention, includes modes where cells from a 5-way channel flow through the device to a predetermined output or ramification channel (which can be called a "waste channel"), until the measurable reporter level of a cell is above a pre-set threshold. At that time, the flow is diverted to deliver the cell to another channel. For example, in an electro-osmotic mode, where the change is instantaneously instantaneous and the performance is limited by the higher voltage, the voltages are temporarily changed to divert the chosen cell to another predetermined output channel (the which can be called a "collection channel"). Classification, including detection of a reporter's synchronization and deviation from The flow can be controlled by several methods including control by computer or microprocessor. Different algorithms can be implemented for classification in the microfluidic device through different computer programs, such as programs used in conventional FACS devices. For example, a programmable card can be used to control the change, such as a La b PC 1 200 Card, available from National I nstruments, Austin, TX. Algorithms as classification procedures can be programmed using C ++, LABVI EW, or any suitable computation program. A "reversible" classification algorithm can be used instead of a "forward" mode, for example in modalities where it may be ^ - .. «A.A .: limited change in speed. For example, a shifted scheme may be used instead of electro-osmotic flow and does not require high voltages, and may be more robust for longer runs. However, mechanical constraints can cause the rate of fluid change 5 to become rate limiting. In a scheme changed by pressure, the flow is stopped when a cell of interest is detected. While the flow stops, the cell may be beyond the junction or ramification point and may be part of the downstream path of the waste channel. In this situation, a reversible mode can be used. The system can run backwards at a lower speed (changeable) (for example, from waste to entry), and the cell is then switched to a different branching or collection channel. At that point, a potentially misclassified cell is "saved", and the device can run again at high speed in the forward direction. This "reversible" sorting method is not possible with standard FACS machines. The FACS machines classify mostly aerosol droplets, which can not be returned to the camera, in order to be redirected. The aerosol droplet classifier is virtually irreversible. Reversible classification is particularly useful for identifying rare cells (eg, in molecular evolution and cytological identification of cancer), or cells in small quantities, which can be misdirected due to a margin of error inherent in any fluid device. . The reversible nature of the device of the invention allows a reduction in this possible error.

- "».-A *** ^ ** "- ** -. In addition, a "reversible" classification method allows multiple measurements over the course of a cell's time if this is allowed. This allows observations or measurements of the same cell at different times, due to that the flow returns the cell to the detection channel 5 again before redeploying the cell to a different channel. Thus, the measurements can be compared or confirmed, and changes in the properties of the cells over time, for example, in kinetic studies, can be examined. When it comes to separating cells in a sample at a ratio m and low for the total number of cells, a classification algorithm may be implemented that is not immedi- ated by the intrinsic rate of change of the device. Consequently, the cells flow at the highest possible static (no change) speed from the input channel to the waste channel. Unwanted cells can is directed towards the waste channel at the highest possible speed, and when a desired cell is detected, the flow can be encouraged and then inverted, to direct the cell back to the detection region, from where it can be redirected (i.e. , to achieve an efficient change). Hence, the cells can flow at the most static speed high possible. Preferably, the fluid that carries the cells has a relatively low Reynolds number, for example 1 0"2. The Reynolds number represents an inverse relationship between the density and velocity of a fluid and its viscosity in a length channel. given, more viscous fluids, less dense, which move slower over a shorter distance, tá ^^ MÉ at ¡• Ü? MíkCikltíaíU will have a lower Reynolds number, and will be easier to divert, stop, start or invest without turbulence. Due to small sizes and slow speeds, microfabricated fluid systems are often in a low Reynolds number regime (Re < < 5 1). In this regime, the effects of inertia, which cause turbulence and secondary flows, are negligible; viscous effects dominate the dynamics. These conditions are advantageous for classification, and are provided by microfabricated devices of the invention. Accordingly, the microfabricated devices of the invention are preferably operated, if not exclusively, at a low or very low Reynolds number. Exemplary classification schemes are shown diagrammatically in FIGS. 14A and B, and FIGS. 15A and B.

EXAMPLES EXAMPLE 1 - Microfabrication of a silicon device Analytical devices that have flow channels, valves and other microscale elements can be designed and manufactured from a solid substrate material. Silicon is a substrate material Preferred due to a well-developed technology that allows its accurate and efficient manufacture, but other materials can be used, including polymers such as polytetrafluoroethylenes. Microlabeling methods well known in the art include film deposition processes, such as spin coating and vapor deposition chemical, laser manufacturing or photolithographic techniques, or methods of i Tilín r i JlMli? ^ - "" 8 engraving, which can be made either by wet chemical processes or plasma. See, for example (37) and (38). FIGS. 1A-1D illustrate the initial steps in microfabricating channels and discriminating region of a 5 cell sorting device of the invention by photolithographic techniques. As shown, the structure includes a silicon substrate 160. The silicon disc that forms the substrate, is usually washed in a 4: 1 bath of H2SO / H2O2, rinsed with water and dried by spinning. A layer 162 of silicon dioxide is formed, preferably of approximately 0.5 μm of I0 thickness, in silicon, usually by heating the silicon disk to 800 to 1200 ° C in a vapor atmosphere. The oxide layer is then coated with a photoresist layer 164, preferably about 1 μm thick. Suitable negative or positive resistant materials are well known. The resistant materials common negatives include two-component bisarilazide / rubber protective layers. Common positive resistant materials include polymethyl methacrylate (PMMA) and two-component diazoquinone / phenolic resin materials. See, for example, (36). The coated laminate is irradiated through a photomask 166, which has been printed with a pattern that corresponds in size and disposition to the desired pattern of the microchannels. Methods for forming photomasks having desired photomask patterns are well known. For example, the mask can be prepared by printing the desired arrangement on a higher transparency using a high resolution printer (3000 dpi). The exhibition is made on ~. ** ui ** ai *? u¡ * ~. ** ** •• * -aab • * - • - ** - '- • * - a standard equipment, such as a Karl Suss contact litography machine. In the il method used in FI GS. 3A-3D, the photoresist layer is a negative protective layer. Thus, exposure of the protective layer to a selected wavelength, for example, UV light, produces a chemical change that makes the exposed protective layer material resistant to passage. of g rabado his bsequent. The treatment with a suitable engraver removes the unexposed areas of the protective layer, leaving a silicon oxide pattern uncovered and covered with a protective layer on the surface of the disc, corresponding to the display and dimensions of the desired microstructures. In this embodiment, because a negative protective layer was used, the uncovered areas correspond to the printed arrangement in the photomask. The disc is then treated with a second recording material, such as a reactive ion etch (RI E), which effectively dissolves the exposed areas of silicon dioxide. The remaining protective layer is removed, usually with hot aqueous H2SO4. The remaining pattern of silicon dioxide (1 62) now serves as a mask for silicon (1 60). The channels are recorded in the unmasked areas of your silicon substrate when dealing with a KOH recording solution. The engraving depth is controlled by the treatment time. Additional microcomponents can also be formed within the channels by additional steps of photolithography and etching, as discussed below. Depending on the method to be used to direct the flow of cells through the device, for example, electro-osmotic or microvalve, electrodes and / or valves are manufactured in the flow channels. A variety of different techniques are available to apply thin metal coatings to a substrate in a desired pattern. These are reviewed, for example, in (32). A convenient and common technique used in the manufacture of microelectronic circuits is vacuum deposition. For example, metal contacts or electrodes can be evaporated on a substrate using vacuum deposition and a contact mask made of, for example, a "MYLAR" sheet. Various targets, such as indium / tin oxide, silver, gold or platinum can be used for the electrodes. The deposition techniques that allow precise control of the deposition area are preferred when applying electrodes to the side walls of the channels in the device of the invention. Such techniques are described, for example, in (32) and the references cited therein. These techniques include plasma atomization, where a plasma gun accelerates molten metal particles into a carrier gas to the substrate, and physical vapor deposition using an electron beam to deliver atoms in the line of sight to its surface. bstrato from a virtual point source. In the laser coating, a laser is focused on the target point on the substrate, and a carrier gas projects powder coating material into the beam, so that the molten particles are accelerated towards the substrate. Another technique that allows precise focusing uses an electron beam to induce the selective decomposition of a previously deposited substance, such as a metal salt, to a metal. This technique has been used to produce submicron circuit trajectories, for example (26).

EJ EM PLO 2 - Photodiode detectors In one embodiment of the invention, shown in F I G. 2A, each detection region is formed from a portion of a channel 74 of an analysis unit and includes a photodiode 72, preferably located on the floor of the main channel. The detection region covers a field receiving the photodiode in the channel, where the receiving field has a circular shape. The volume of the detection region is the volume of a cylinder with a diameter equal to the field receiving the photodiode and a height equal to the depth of the channel above the photodiode. The signals of the photodiodes 72 can be brought to a processor via one or more lines 76, representing any form of electrical communication (including, for example, wires, conductive lines recorded on the substrate, etc.). The processor acts on the signals, for example, by processing them into values for comparison with a predetermined set of values to classify the cells. In one embodiment, the values correspond to the amount of optically detectable signal emitted from a cell, which is indicative of a particular cell type or characteristic that gives rise to the signal. The processor uses this information (i.e., the values) to control the active elements in the discrimination region to determine how to classify the cells (eg, electro-osmotic change or valve action).

When using more than one detection region, the photodiodes in the laser diode chi p are preferably separated, relative to the separation of the detection regions in the detection unit. analysis. That is, for more accurate detection, the photodiodes are placed apart in the same separation as the separation of the detection region. The processor can be integrated in the same chip that contains the analytical unit (s), or it can be separated, for example, an independent microchip connected to the chip containing the analysis unit via electronic controls that are connected to the region (s) of detection and / or the discriminant region (s), such as by a photodiode. The processor may be a computer or microprocessor, and is usually connected to a data storage unit, such as a computer memory, hard disk or the like, and / or a data output unit, such as a computer. display monitor, printer and / or raficador. The types and numbers of cells, based on the detection of a reporter associated with or going to the cells that pass through the detection region, can be calculated or determined, and the data obtained can be stored in the data storage This information may be processed or additionally routed to the data output unit for presentation, eg, histograms, of the cell types or levels of a protein, saccharide or some other characteristic on the cell surface in the sample. The data can also be presented in real time as the sample is flowing through the device. tt ^^ n ^ t ^ t ^ t ^^ at? kaüMaiat In the mode of FIG. 1B, the photodiode 78 is larger in diameter than the width of the channel 82, forming a detection region 80 that is longer (along the length of the channel 82) than its width. The volume of such detection region is approximately equal to the cross-sectional area of the channel above the diode multiplied by the diameter of the diode. If desired, the device may contain a plurality of analysis units, ie, more than one detection and discrimination region, and a plurality of branch channels, which are in fluid communication with and branch out from the discrimination regions. It will be appreciated that the position and destination of the cells in the discrimination region can be monitored by additional detection regions installed, for example, immediately upstream of the discrimination region and / or within the branch channels immediately downstream of the branch point. The information obtained by the additional detection regions can be used by a processor to check continuously estimates of the speed of the cells in the channels and to confirm the cells having a selected characteristic enter the desired branch channel. A group of collectors (a region consisting of several channels, which lead to, or from a common channel) can be included to facilitate the movement of the cell sample of the different units of analysis, through the plurality of branching channels and the appropriate solution output. The collectors of preference are microfabricados'!,, .......... .... *.; .., ^.:., .. .... _. ., ..... ........ . ... ~ ........ j. ^. ^ ...... ........ "... X. , .. .. ... - on the chip at different depth levels. In this way, the devices of the invention having a plurality of analysis units can collect the solution from associated branch channels of each unit in a manifold, which directs the flow of solution to an outlet. The outlet can be adapted to receive, for example, a pipe segment or a sample tube, such as a standard 1.5 ml centrifuge tube. Collection can also be done using micropipettes.

EXAMPLE 3 - Valve Structures In a mode where pressure separation is used for cell discrimination, the valves can be used to block or unblock the pressurized flow of cells through selected channels. A thin cantilever, for example, may be included within a branch point, as shown in FIGS. 3A and 3B, so that it can be displaced towards one or another wall of the main channel, usually by electrostatic attraction, thus closing a selected branch channel. The electrodes are on the walls of the channel adjacent to the end of the cantilever. Appropriate electrical contacts are also provided to apply a potential to the cantilever in a similar manner as the electrodes. A valve within a channel in the form of an electrostatically operated cantilever or diaphragm can be microfabricated, if desired. Techniques for forming such elements are well known in the art (eg, 24, 29, 35, 36, 37). Normal processes include the use of , ..., * ». i. Í ^^ K ^ SJ *.,. sacrificial layers selectively etched in a multilayer structure or, for example, the undercutting of a layer of silicon dioxide via anisotropic etching. For example, to form a cantilever within a channel, as illustrated in FIGS. 3A and 3B, a sacrificial layer 168 can be formed adjacent to a small section of a non-recordable material 170, using known photolithography methods, on the floor of a channel, as shown in FIG. 3A. Both layers may then be coated, for example, silicon dioxide or another non-recordable layer, as shown in 172. The sacrificial layer etching the deposited member of cantilever 174 into the channel, as shown in FIG. 3B. Suitable materials for the sacrificial layer, non-recordable layers and etchant include undisturbed silicon, p-altered silicon and silicon dioxide, and the EDP (ethylenediamine / pyrocatechol) recorder, respectively. Because the cantilever in FIG. 3B is parallel to the engraving direction, it can be formed from a thin layer of silicon by incorporating the element into the original photoresist pattern. The cantilever is preferably coated with a dielectric material, such as silicon nitride, as described in (35), for example, to prevent short circuits between the conductive surfaces. It will be apparent to someone of skill in the field that other types of valves or switches can be designed and manufactured, using photolithographic techniques or other well-known microfabrication techniques, to control the flow within the channels of the device. Multiple channel layers can also be prepared. f * •• - - - - -. . . .. . . . . »EXAMPLE 4 - Classification Techniques As illustrated with respect to FIGS. 4A-4D, there are a variety of ways in which cells can be routed or sorted to a selected branching channel. FIG. 4A shows a discrimination region 102, which is suitable for electrophoretic discrimination as the classification technique. The discrimination region is preceded by a main channel 104. A junction divides the main channel into two branch channels 106 and 108. The discrimination region 102 includes electrodes 110 and 112, placed on the outer side walls of branch channels 106 and 108, and which are connected to knobs 114 and 116. The knobs are connected to a voltage source (not shown) incorporated in, or controlled by, a processor ( not shown), as described below. The distance (D) between the electrodes is preferably less than the average distance separating the cells during flow through the main channel. The dimensions of the electrodes are usually the same as the dimensions of the channels in which they are placed, so that the electrodes are as tall and wide as the channel. The region of discrimination shown in FIG. 4B is suitable for use in a device employing electro-osmotic flow, to move the cells and solution in bulk through the device. FIG. 4B shows a discrimination region 122, which is preceded by a main channel 124. The main channel contains a junction that divides the . < . < ».....« m * < . . -. . . . .. .. .. i. t main channel in two branch channels 126 and 128. An electrode 130 is placed downstream of the junction of the main channel, for example, near the sample inlet of the main channel. The electrodes are also placed in the branch channel (electrodes 5 132 and 134). The electrode 130 can be negative and the electrodes 132 and 134 can be positive (or vice versa) to establish the flow of bulk solution according to the well-established principles of the electro-osmotic flow (25). After a cell passes the detection region (not shown) and enters the discrimination region 122 (eg, between the main channel and the two branch channels), the voltage for one of the electrodes 132 and 134 may be closed, leaving a simple attractive force acting on the solution and the cell to influence the selected branching channel. As before, the appropriate electrodes 15 are activated after the cell has been consigned to the selected branch channel, in order to continue bulk flow through both channels. In one embodiment, the electrodes are charged to divert the bulk flow of cells in a branch channel, for example, channel 126, which may be called a scrap channel. In response to a signal indicating that a cell has been identified or selected for collection, the charge on the electrodes can be changed to divert the selected cell to the other channel (channel 128), which can be called a collection channel. In another embodiment of the invention, shown in FIG. 4C, the 25 cells are directed to a predetermined branching channel via "- '• -. ¡* * &&~ ~ ~ I ~ _ £« £. "- • > A valve 140 in the region of discrimination The valve 140 comprises a thin expanse of material to which a charge can be plumped via an electrode command 142 The valve 140 is shown with both can open them, and can be diverted to close any branching channel by applying a voltage across the electrodes 1 44 and 1 46. A cell is detected and selected to be classified in the detection region (not shown), and can be directed to the appropriate channel by closing the other channel, for example, by applying, removing or changing a voltage applied to the electrodes.The valve can also be configured to to close a channel in the presence of a voltage, and to close the other channel in the absence of a voltage. FIG. 4D shows another embodiment of a discrimination region of the invention, which uses the flow stop in one or more branch channels as the discrimination means. The sample solution is m ueve through the device by applying positive pressure at one end, where the solution input is located The discrimination or routing of the cells is affected by simply blocking a branch channel (145 or 148) or an output of sample of branching channel using the valves in a flow operated under pressure (147 or 149). Because the small scale of the channels and the inability to compress the liquids, blocking the solution flow creates an effective "plug" in the unselected branch channel, thereby temporarily routing the cell along with the flow of bulk solution to the selected channel. The valve structures can be incorporated downstream from the discrimination region, which are controlled by the detection region as described herein. Alternatively, the discrimination function shown in FIG. 4D can be controlled by changing the hydrostatic pressure in the 5 sample outputs of one or both branch channels 145 or 148. If the branch channels in a particular analysis unit have the same resistance to fluid flow, and the pressure at the sample inlet of the main channel of an analysis unit is P, then the fluid flow out of any selected branch channel can be stopped by applying a pressure P / n on the sample output of the channel of desired branch, where n is the number of branch channels in the unit of analysis. According to this, in an analysis unit having two branching channels, the pressure applied at the outlet of the branch to be blocked is P / 2. 15 As shown in FIG. 4D, a valve is located within each branch channel, rather than at the branch point, to close and terminate the pressurized flow through the selected channels. Because the valves are located at a point downstream of the discrimination region, the channels in this region 20 can be formed having a greater width than in the discrimination region, in order to simplify the formation of valves. The width of the cantilever or diaphragm should be approximately equal to the width of the channel, allowing movement within the channel. If desired, the element can be coated with a more malleable material, such as a metal, to allow a better seal. Such a coating can also be - - *** "" "• * '- • - • • •" * * • - **' - • * > "- used to make a non-conductive material, such as silicon dioxide, conductor As before, suitable electrical contacts are provided to move the cantilever or diaphragm toward the opposite surface of the channel When the top surface is a glass cover plate, the electrodes and contacts can be deposited on the glass FIG 5 shows a device with analysis units containing a cascade of detection and discrimination regions suitable for successive turns of cellular sorting For example, such a cascade configuration can be used to sequentially test the cells for at least three different reporters, for example, fluorescent dyes, corresponding to the expression of at least three different cellular characteristics (markers) .The samples collected in the exit region of the different branch channels contain cell deposits that express levels defined of each u not the three markers. The number of reporters employed, and therefore the number of markers expressed of interest, can be varied to meet the needs of the practitioner.

EXAMPLE 5 - Reporters and labels for cell sorting To classify cells of the invention, the cells are marked with an optically detectable reporter, which is analyzed and interpreted to determine if the cell that has the reporter should be classified. The reporter can work in a variety of ways to issue or . ,. .- .. i i. «MG .. ^ .- ^». display effectively a readable signal that can be detected by the detection region. In one embodiment, the signal is in the form of a marker that associates within, or binds to, a particular cell type. Therefore, the signal acts to identify the cell by having a particular characteristic, for example, a protein (receptor) or saccharide, so that the reporter signal of a given cell is proportional to the amount of a particular characteristic. . For example, the reporter may be an antibody, a receptor or a ligand for a receptor (which binds to a protein or sugar), or a fragment thereof, each having a detectable portion, such as a dye that fluoresces. The reporter can bind to a structure on the surface or inside the cell of interest, and because the antibody contains a detectable reporter, any cell to which the reporter joins would be detectable by the detection region of the device as the cell it flows beyond that region. It should be appreciated by those of ordinary skill in the art, that the antibody, receptor, ligand or other agent that can act as a marker, can be modified to meet the needs of the practitioner, for example, such as using fragments. or make chimeras. Fluorescent dyes are examples of optically detectable reporters. There is a variety of known dyes, which bind selectively to nucleic acids, proteins and sugars. For DNA and RNA studies, these include, but are not limited to, Hoechst 33258, Hoechst 33342, DAPI (4 'HCl, 6-diam idino-2-phenylindole), propidium iodide, dih idroetidium, acridine orange na, ethidium bromide, ethidium homodimers (eg, EthD-1, EthD-2), acdd ma-ethidium heterodimer (AEth D), and thiazole orange derivatives PO-PRO, BOP RO, YO-PRO , To-PRO, as well as its dimeric analogues, POPO, BOBO, YOYO and TOTO. All these compounds can be obtained from Molecular Probes (Eugene, OR). Extensive information on their spectral, use and similar properties is provided in Haugland (30). Each dye is bound to a known maximum density or determined empirically. Thus, by measuring the fluorescence intensity of a reporter molecule, the presence, concentration or relative amount of the desired cell characteristic can be determined., for example by comparison with an empirically determined reference standard. For example, it has been found that a YOYO-1 molecule binds to 4-5 base pairs of DNA, and this ratio can be used to evaluate the length of an unknown DNA sequence, or to classify DNA based on a range or corresponding colorant signal window with a desired classification length. You can also use reporters or ltravioleta. Examples include green fluorescent protein and cascade blue. Two applications of the invention are for the quantification of cell surface and intracellular antigens, and of nucleic acid contents in cells, for the study of cell differentiation and function, for example, in the field of cancer immunology and cytology. . For studies of cell surface antigens, phycobiliproteins, phycoerythrin, Texas red and allophycocyanin can be used as fluorescent labels for monoclonal antibodies for the identification of blood cells and cancer cells. For cell DNA / RNA analysis the dyes mentioned above can be used. For the study of cellular functions, chromogenic or fluorogenic substrates were first used in flow cytometry to detect and quantify the activities of intracellular enzymes (e.g., 4-nitrophenyl, 5-bromo-5-chloro-3-indolyl, digalactoside of fluorescein, fluorescein diglucuronide, fluorescein diphosphate and creatine phosphate). These reporters can be used in the invention. Fluorescent dyes and substrates can also be used for the detection of other cellular functions, such as protein contents (dyes, for example, fluorescein isothiocyanate, sulfurodamine, sulfosuccinimidyl esters, fluorescein-5-maleimide), intracellular pH (such as, carboxyfluorescein). and its derived esters, and acidSulphonic fluorescein and its derived diacetates), for signal transduction (eg, fluorescent bisindolylmaleimides, hypericin, hypocrelin, forscolma), mitochondrial and cytoplasmic membrane potentials developed for analysis and cellular activation processes. Other suitable applications for use in the invention include probes chromogenic or fluorogenic for analysis of other cellular environments or encapsulants, such as, detection of organelles (for example, mitochondria, lysosomes), cell morphology, cell viability and proliferation, receptors and ion channels, and for measurements of certain ions ( for example, metal ions, Ca2 +, Mg2 +, Zn2 +) in the cells or in the ^ t? tSt¿m¡jilííím¿á environment. Probes of the classes described herein can be obtained, for example, from Molecular Probes (Eugene, OR). In another fashion, cells can produce a reporter in vivo (eg, a fluorescent compound) through interaction with a reagent adhered to the fluid medium. For example, cells that contain a gene for an oxygenase enzyme can catalyze a reaction in an aromatic substrate (eg, benzene or naphthalene), with the net result that the fluorescence, or other detectable property of the substrate will change. This change can be detected in the detection region, and the cells that have that change in fluorescence can be collected based on the predetermined criteria. For example, cells that produce a desired monooxygenase enzyme (such as a P450 enzyme) can be detected in the presence of a suitable substrate (such as naphthalene), and can be collected according to the invention, based on the capacity of the enzyme to catalyze a reaction in which a detectable (eg, fluorescent) product is produced from the substrate. Classification can also be done based on a threshold concentration or window of the reaction product, which in turn can be correlated with the level of fluorescence. See, Affholter and Arnold (50) and Joo et al. (51). Any mechanism of this kind, including any reporter or combinations of substrate, enzyme and product, can be used for detection and classification in a similar manner, as long as there is at least one way to detect or measure the presence or degree of the reaction of interest.

The invention can be used to classify any prokaryotic (e.g., bacterial) or eukaryotic (e.g., mammalian cell, including human blood cell, such as human peripheral blood mononuclear cells (PBMCs)) cells, which have a characteristic or detectable label, or which can be labeled with a detectable reporter, for example, an optically detectable label. For example, antibodies or fragments thereof that recognize a receptor or antigen of interest, and which are linked to a fluorescent dye can be used to label cells. Examples of antigens that can be labeled with antibodies to classify cells include, without limitation, HLA DR, CD3, CD4, CD8, CD11a, CD11c, CD14, CD16, CD45, CD45RA and CD62L. Antibodies can be detected, in turn, by using an optically detectable reporter (either via directly conjugated reporters or via secondary antibodies). marked) according to methods known in the art. Alternatively, a ligand that is linked with a fluorescent dye and has affinity for a particular antigen or receptor of interest can be used in the same manner. It will be appreciated that the cell sorting device and method As described above, they can be used simultaneously with multiple optically detectable reporters having different optical properties. For example, fluorescein fluorescent dyes (FITC), phycoerythrin (PE) and "CYCHROME" (Cy5-PE) can be used simultaneously because of their different excitation and emission spectra. The different dyes can be tested, for example, in regions of detection and discrimination ^ iMá t tÍÍÍ tÉi successive. Such regions may fall in cascades as shown in FIG. 5 to provide samples of cells having a selected amount of signal from each dye. Optical reporters, such as fluorescent portions, can be excited to emit light of characteristic wavelengths by an excitation light source. The fluorescent portions have an advantage since each molecule can emit a large number of photons at a distance of 3,014 m in response to radiation stimuli. Other brands of optically detectable reporter include chemiluminescent and radioactive portions, which can be used without an excitation light source. In another embodiment, the absorbance at a particular length of time, or measuring the refractive index of a particle, such as a cell, can be used to detect a characteristic. For example, if a refractive index is used, different types of cells can be distinguished by comparing differences in their retraction properties as a light source passes.

EXAMPLE 6 - Operation of a microfabricated cell sorting device 20 In an operation of the microfabricated device of the invention, it is advantageous and preferred to "hydrate" the device (i.e. fill the channels of the device with a solvent, such as water). or the buffer solution, in which the cells will be suspended), before introducing the solution containing cells. Hydration of the device can be achieved by supplying the solvent to the device reservoir and applying a ~ A * í.- ~ »MI¡ill *? L ~. ~ .xt •. -.- ... .... .. ... "._t ..«. !, ... ... .. _ ... ¿ ,. »,.! I t, > The hydrostatic pressure to force the fluid through the analytical units. Following the hydration step, the solution containing cells is introduced into the sample inputs of the analysis unit (s) of the device. The solution can conveniently be introduced in a variety of ways, including by an opening in the floor of the inlet channel, reservoir (receptacle) or via a connector. As a stream of cells to be clasified by a detectable characteristic or reporter moves through the detection region, a signal for each cell is detected or measured and compared to a fixed range of values to determine if the cell has the desired characteristic based on the amount of the detected reporter. Preferably, the cells move in a single row. In the modality of this example, the level of the reporting signal is measured in the detection region using an optical detector, which may include one or more than one photodiode (e.g., avalanche photodiode), a fiber optic light guide leading, for example, to a photomultiplier tube, a microscope with a high numerical display objective and an intensified video camera, such as a CCD camera or similar. The optical detector can be microfabricated or placed on a cell analysis chip (e.g., a photodiode as illustrated in FIGS 2A and 2B), or it can be a separate element, such as a microscope objective. If the optical detector is used as a separate element, It is generally advantageous to restrict the collection of signal from the region ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ It can also be advantageous to provide an automated means for scanning the laser beam in relation to the cell analysis chip, by scanning the light emitted on the detector or by using a multichannel detector. For example, the cell analysis chip can be secured to a movable assembly (eg, a motorized / computer controlled micromanipulator) and scanned under the lens. A fluorescence microscope, which has the advantage of an excitation light source built in (epifluorescence), is preferably used for the detection of a fluorescent reporter. The signal collected from the optical detector is routed, for example, via electrical traces and secure in the chip, to a processor, which interprets the signals in values that correspond to the characteristic of cell type that gives rise to the signal. These values are then compared, by the processor, with pre-loaded instructions containing information on which branching channel the cells having the desired characteristic will be routed. In some embodiments, there is a period of signal delay (ie, long enough to allow the cell's reporter signal to reach the discrimination region), after which the processor sends a signal to drive the active elements in the discrimination region to route the cell to the appropriate branch channel. In other embodiments, there is a small period of signal delay or none, because the detection region is immediately adjacent to the branch point, and the change may be immediate. There may be a period of delay of classification, just how long it is necessary to ensure that a selected cell is sorted to the correct branching channel, that is, before switching back to normal flow (not selected). This period can be determined empirically. Any period of necessary or desired delay can easily be determined according to the ratio at which the cells are moved through the channel, ie, its velocity, and the length of the channel between the detection region and the discrimination region. In addition, depending on the flow mechanism, the cell size can also affect the movement (velocity) through the device. In cases where the sample solution is moved through the device using hydrostatic pressure (for example, as a pressure in the inlet region and / or suction in the outlet region), velocity is usually the flow rate of the the solution . If the cells are directed through the device using some other force, such as electro-osmotic flow (for example, using an electric gradient or field between the input region and the output region), then the delay period is a function of speed and cell size, and can be determined empirically by running standards including different sizes or types of known cells. Thus, the device can be calibrated appropriately for the intended use. The time required to isolate a desired number of cells depends on a variety of factors, including the size of the cells, the proportion at which each analysis unit can process the individual fragments and the number of analysis units per chip. He ll.

The required time can be calculated using known formulas. For example, a chip containing 1000 analysis units, each of which can classify 1,000 cells per second, could isolate approximately 100 μg of approximately 3 μm cells in 1 hour. The concentration of cells in the sample solution can influence the classification efficiency and can be optimized. The concentration of cells should be sufficiently diluted so that most of the cells pass through the detection region one by one (in a single row). ), with only a small statistical opportunity for two or more cells to pass through the region simultaneously. This is to ensure that for the vast majority of measurements, the level of reporter measured in the detected region corresponds to a single cell and not to two or more cells. The parameters that govern this relationship are the volume of the detection region and the concentration of cells in the sample solution. The probability that the detection region will contain two or more cells (P> 2) can be expressed as P > 2 = 1 -. { 1 + [cell] V.}. x e - [cell] x V where [cell] is the concentration of cells in units of cells per μm3 and V is the volume of the detection region in units of μm3. It will be appreciated that P > 2 can be minimized by decreasing the concentration of cells in the sample solution. However, decrease the concentration of cells in the sample solution as well > kiM_ .MW results in an increased volume of solution processed through the device and may result in longer run times. Accordingly, it is desirable to minimize the presence of multiple cells in the detection chamber (thereby increasing the accuracy of the classification) and to reduce the volume of sample flow, thereby allowing a classified sample. in a reasonable time in a reasonable volume containing an acceptable concentration of cells. The P > Tolerable maximum depends on the desired "purity" of the classified sample. The "purity" in this case refers to the fraction of cells classified that are in a specified size range, and is inversely provided by P > 2. For example, in applications where a high purity is not necessary or desired, a P > 2 relatively high (for example, P> 2 = 0.2). For most applications, keep the P > 2 to or below approximately 0. 1, preferably at or below about 0.01, provides satisfactory results. For example, where P > 2 is 0. 1, it is expected that in approximately 10% of the measurements, the signal for the detection region is a result of the presence of two or more cells. If the total signal of these If the cell is in the range corresponding to the value set for a desired cell type, those cells will be classified in the predetermined channel or tube for the type of cell desired. The cell concentration necessary to achieve a value of P > 2 particular in a particular detection volume can be calculated from the previous equation. For example, a detection region in the form of a cube of 10 micras per side, has a volume of 1 pl. A concentration of cells having a diameter of 1 miera, resulting in an average in one cell per pl, is approximately 1.7 pM Using a value of P > 2 of about 0.01, the concentration of cells in a sample analyzed or processed using the volume of detection region of 1 pl is approximately 10 pM, or scarcely one cell per 3 detection volumes ([cell] x V = -0.3). If the concentration of cells is such that [cell] x V is 0.1, then P > 2 is less than 0.005; that is, it is less than one-half of a one-percent chance (0.5%) that the detection region will contain, at any given time, two or more cells. As discussed before, the sample mixture introduced into a device of the invention should be sufficiently diluted, so that there is a high probability that only a single cell will occupy the detection region at any given time. This will allow the cells to be in "single row", separated by stretches of cell-free solution as the solution flows through the device between the detection and discrimination regions. Consequently, the length of the channel, discussed above, between the detection and discrimination region should not be too long, so that the random thermal diffusion does not substantially alter the separation between the cells. In particular, the length of the channel should be sufficiently short, so that a cell can traverse it in a sufficiently short time, so that even the smallest cells being analyzed, will normally be unable to diffuse and change position or order in the line of the cells. The channel should also be long enough, so *. ", ^ .." * ¡¡^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ that the flow control can be changed in time to properly divert a cell selected in response to detection or measurement of a signal produced for the examination of the cell as it passes through the detection region. 5 The diffusion constant of a sphere of 0.5 m is approximately 5x1 0"9 cm2 / s The diffusion equation gives the distance (x) that the sphere diffuses in time (t) as: <x2 > = Dt, where D is the diffusion constant given by D = kBT / 6p? R0 In this equation, kb is the Boltzmann constant, T is the temperature,? Is the viscosity of the fluid and R0 is the diameter of the sphere. Using this relationship, it will be appreciated that a 0.5 μm cell takes approximately 50 seconds to diffuse 500 μm. The average cell separation in the channel is a function of the cross-sectional area of the channel and the cell concentration, the latter usually determined in view of acceptable values of P > 2, discussed before. From these relationships, it is then easy to calculate the maximum channel length between the region of detection and discrimination, which would ensure that the cells do not change order or position in the cell line. In practice, the channel length between the detection and discrimination regions is between approximately 1 μm and approximately 1 00 μm. Cutting forces can affect the rate at which cells move through the microfluidic device, particularly when living cells are going to be sorted and collected. Experiments have shown that high electric fields, in the range of 2 - 4 kV / cm for human erythrocytes and 5 - 1 0 kV / cm for yeast cells, can be used to introduce D NA and other substances into cells using electroporation. At these voltages there was no cell lysis, although membrane permeation was possible. To avoid membrane permeation and cell lysis, it is preferred that the electric fields applied to move cells in any of the described flow techniques be less than about 600 V / cm and most preferably less than about 1 00 V / cm.

EJ EM PLO 7 - FACS microfabricated elastomer 10 This example demonstrates the fabrication and operation of a disposable microfabricated FACS device, which can function as a device that operates alone or as a component of an integrated microanalytical chip, to classify cells or biological materials. The device allows a high sensitivity, without cross-contamination, lower cost to operate and manufacture than conventional FACS machines and multiple hour run times. In this example, the microfabricated chip had a detection volume of approximately 250 fl and a single channel yield of approximately 20 cells / sec undo. The device obtained a substantial enrichment of populations of fluorescent pearls of microporous size of different colors. In addition, populations of E. coli cells expressing green fluorescent protein were separated, and enriched, from a non-fluorescent E. coli cell carrier (wild type). It was also found that the bacteria were viable after the extraction of the sorting device.

---- * • * "Preparation of the microfabricated device A silicon disk was recorded and manufactured as described above and in (15) After the standard contact photolithography techniques 5 to form the oxide surface patterns From the silicon disc, a mixture of C2F2 / CHF3 gas was used to burn the disc by RI.The silicon disc was subjected to additional etching with KOH to expose the silicon below the oxide layer, thereby forming a mold for the silicone elastomer.The silicon mold forms a "T" arrangement of channels The dimensions of the channels can vary widely, having approximately 5x4 μm in dimensions. A representative device of the invention is shown in FIG. 6. The engraving process is shown schematically in FIG. 7. Standard microlabor techniques were used to create a mold negative master out of a silicon disk. The disposable silicone elastomer chip was made by mixing General Electric RTV 615 (20) components together and emptying onto the etched silicon disc. After being cured in an oven for two hours at 80 ° C, the elastomer was stripped from the disc and sealed to a glass cover strip for classification. To make the elastomer hydrophilic, the elastomer chip was immersed in HCl (pH = 2.7) at 60 ° C for 40 to 60 min. Alternatively, the surface could have been coated with polyurethane (3% w / v in 95% ethanol and diluted 10X in ethanol). It is noted that the master disk can be reused indefinitely. The device shown had channels that are 100 μm wide in the receptacles, narrowing to • BiriÉiliÉÉl? IhilÉÉi 3 μm in the classification junction (discrimination region). The channel depth is 4 μm, and the receptacles are 2 mm in diameter. In this embodiment, the cell sorting device was mounted on an inverted optical microscope (Zeiss Axiovert 35) as shown in FIG. 8. In this system, flow control can be provided by voltage electrodes for electro-osmotic or capillary control for pressure-controlled control. The detection system can be photomultiplier tubes or photodiodes, depending on the application The input receptacle and two collection receptacles were incorporated into the elastomer chip on three sides of the "T" forming three channels (FIGS 6 and 7) . The chip was attached to a strip of glass cover and mounted on the microscope. Three platinum electrodes were inserted into separate receptacles. A water-cooled argon laser (for cells) or a 100W mercury lamp (for beads) focused through an oil immersion objective (Olympus Plan Apo 60 X 1.4NA) was used to excite the fluorescence and a Charge-coupled device camera (CCD) took the picture. To select the red fluorescence emission at 630 nm ± 30, an emission filter (Chroma) is used. The detection region was approximately 5 to 10 μm below the T junction and has a window of approximately 15x5 μm in dimension. The window is implemented with an adjustable Zeiss slit. Using one or two Hammatzu R928 photomultiplier tubes (bias -850V) with customary current-to-voltage amplifier, or using photodiodes, as detectors, and Using different emission filters (depending on the fluorescence), the photocurrent (s) of the detector (s) was converted to voltage using a Burr-Brown optical amplifier OP128 (107 V / A), digitized by a National Board. Instrument PC1200 and processed on a computer. The voltages on the electrodes are provided by a pair of Apex PA42 HV optical amplifiers energized by Acopian power supplies. The third electrode was earth. Adjusting the voltage settings on the PC1200 board analog outputs and its amplification to the platinum electrodes, can control the change of the directions 10 of the fluids. Thus, the cells can be directed to either side of the "T" channels, depending on the voltage potential settings. Additionally, different ways of classifying in the microfluidic device can be achieved by different computer programs, for example, different computer controlled procedures using known programming techniques.

Classification experiments This microfabricated FACS system modality was used to classify fluorescent beads of different emission wavelengths in different proportions up to a yield of 33,000 beads per hour (see FIGS 9-11). Extra reservoir receptacles were incorporated in the outer side of the three receptacles of the chip, in order to avoid the suppression of ions, and platinum electrodes (with the ground electrode in the input receptacle) were inserted in the reservoir receptacles . 25 Pearls of one diameter miera were suspended in PBS (137 mM NaCl, t ^? ^ -tMMMagaiMjH 2.7 mM KCl, Na2HP04-7H20, 1.4 mM KH2PO4) with 10% BSA (1 g / l) and 0.5% Tween 20 in various proportions and dilutions. Samples of the different colored fluorescent beads were injected, having proportions as indicated below, in the input receptacle 5 in aliquots of 10 to 30 μl. The collection receptacles were filled with the same shock absorber. To classify the beads, the optical filter in front of the PMT passed only the color fluorescence corresponding to the color of the bead of interest, for example, red fluorescent light to classify red beads. The I0 voltages at the electrodes were set for change purposes, either for classification or reversible change. The duration of the classification can be as long as 3 hours, although the voltage settings may have to be readjusted from time to time. The coefficient of variation of bead intensity was measured as approximately 1 to 3% depending on the depth of the channel and the surface treatment of the elastomer. After classification, the enrichment of the beads was determined by the processor that recorded the data collected by the detection region and verified by counting. In the following experiments, the channels of the microfabricated device 20 were 3x4 with pearl classification and 10x4 with cell classification. A. Classification of fluorescent green beads of red fluorescent beads As shown in FIG. 7, the classification of green fluorescent pearls 25 to red fluorescent beads in a proportion of nii'iiiiiipl'iilifiíw- - '... .... i l JMJ? ^^^ 100: 1. A 0.375% mixture of beads resuspended in 137 mM NaCI PBS with 10% BSA + 05% Tw20 was launched through the 3x4 μm silicone elastomer device of the invention for approximately 22 minutes Using a mercury lamp as the source of light, the bias R928 Ha matzu PMT was -850 V with an emission filter of 630 nm ± 30 S. Classification of red fluorescent beads of fluorescent blue forward and reverse. FIG 9 shows the classification of blue fluorescent beads to red fluorescent beads in a ratio of 101 using a forward mode. A 1.5% mix of resuspended beads was classified in NaCl PBS 137 M with 10% BSA + 0.5% Tw20 using a 3x4 μm device for approximately 24 minutes. The red pearls were enriched 8.4 times. The darker and lighter bars represent the proportion of red or blue pearls over the total number of classified pearls, respectively. FIG. 10 shows the classification of red fluorescent beads of blue fluorescent beads using a reversible mode. The beads were prepared in the buffer as described in a ratio of 100: 1 (azuhrojo). After 6 min, the collection channel had a sample of red beads that had been enriched 96 times. The darker and lighter bars represent the proportion of red or blue pearls over the total number of pearls classified, respectively. The yield was approximately 10 beads / s.

C. Classification of fluorescent green fluorescent red beads in reversible mode. FIG. 11 shows the results of classifying, by reversible change, green fluorescent beads to red fluorescent beads in a ratio of 100: 1. A mixture of 0.375% of beads resuspended in NaCl PBS 137mM with 10% BSA + 0.5% Tw20 was classified using a 3x4 um device for approximately 12 minutes. The reversible change provides fast and high performance of pearls or unwanted cells, with a rapid reversal of fluid flow once detects a desired bead or cell. This allows a high performance and a reliable capture of cells or rare cases, with a quick analysis of results. The data represented in FIG. 11 show that the red beads were enriched by approximately 36 times. The darker and lighter bars represent the proportion of pearls15 red or green on the total number of pearls classified, respectively. D. Classification of E. coli cells expressing green fluorescent protein from wild-type cells. Classification results using E. coli cells demonstrated the microfractance capacity of FACS enrichment in living cells. E. coli (HB101) cells expressing green fluorescent protein (GFP) were grown at 30 ° C for 12 hours in LB + amp (a colony inoculated in 3 ml of medium). The preparation of E. coli cells expressing GPF is described, for example, in Sambrook et al. (48). The 25 wild-type E. coli HB101 cells were also incubated for 13 days. hours in between only LB. After incubation, E coli HB101 and GFP HB101 cells were resuspended in PBS (l = 0021) three times and stored at 4 ° C for classification. Immediately prior to classification, the cells were resupended again in PB (43 M Na2HPO47H2O, 4.3 mM KH2PO4) containing 10 ~ 5 to 10"4 M SDS and diluted 10 to 100 times depending on the absorbance (1 to 1.5) and the concentration of the cells The cells were filtered through a 5 mm syringe filter (Millipore) for removal of any elongated cells.The fluorescence was excited using an argon ion laser Coheren Innova 70 of 488 nm (light source from 30 to 50 mW, 6 mW outside the target) the RAS BMP50 Hammatzu PMT was -850V (Chroma) and the emitted fluorescence was filtered using a 535 ± 20 filter. Different proportions of E. coli type cells were mixed and introduced natural to E coli cells expressing GFP (described below) in the input receptacle of the device (volume varies from 10 to 30 μl of sample), collection wells were also filled with 10 to 30 μl PB with SDS 10 ~ 5 to 10"4. After inserting the three platinum electrodes into the receptacles (with the ground electrode in the input receptacle), the voltages were set for forward or reversible sorting modes. The omission voltages here were set at -80V and -56V for the waste and collection channels, respectively. After sorting for approximately two hours, the cells were harvested using a pipette and scratched on antibiotic-containing plates (LB plates) and incubated overnight at 37 ° C for colony counting. In a first experiment, the initial ratio of natural type of E. coli cells expressing GFP was 100: 1 (results in FIG 12). After 2 hours of sorting, cells of E. coli recovered from GFP collection receptacle were enriched 30-fold, with yields of 20%. In FIG. 12, the dark and light bars represent the proportions of E. coli wild type nonfluorescent and E. coli expressing GFP to the total number of sorted cells (approximately 120,000 cells), respectively. The classified cells show a relatively constant viability in electric fields up to approximately 100 V / cm, corresponding to speeds of approximately 1 to 3 mm / s. The yield was about 20 cells / sec, which can be improved, for example, by adding a pressure-controlled change scheme or parallel device fabrication. FIG. 13 shows the results of classifying E. coli cells expressing GFP in an initial ratio of 3: 2. The cells of E. coli expressing GFP were enriched for approximately 1.75 times.

EXAMPLE 8 - Classification by molecular evolution A preferred embodiment is to classify microbial cells (bacteria, yeast, fungi) and particles containing genetic material for molecular evolution, also known as "directed evolution" (41). Directed evolution involves creating a set of acid sequences ---- - - - * • t ^ im IIMM Mistress nucleic mutated, followed by classification for those that alter or enhance a predetermined characteristic (eg, activity of an enzyme, productivity of a certain metabolite or an activity associated with a particular gene or polynucleotide). The gene products are then created, either by inserting the genetic material into microbial cells and relying on the cellular machinery to create the gene products (proteins), or by transcribing the genetic material into protein using cell-free synthesis methods. (42-45). The gene products are then classified to identify altered or improved sequences.

If desired, these sequences can be isolated and further mutated in extra cycles of evolution. After one or more cycles of evolution, gene products exhibiting the desired characteristics can be generated. It is also possible that the activity or property detected, for example, the reporter, is associated with a sequence of Polynucleotide itself, and that the polynucleotide is a desired end product for which classification is desired. In this case, evolution can be performed in vitro (without cells) and without translating the polynucleotides into gene products. The cellular classifier of this invention can be used for classify the microbial cells to determine if they exhibit the desired characteristics, and therefore contain altered or improved genetic material. If the synthesis of cell-free protein has been used, ie, without living biological cells, this synthesis apparatus, for example, RNA-fusion protein such as that described by Roberts & Szostak (45) or a cell-free synthesis system as in Jermutus ? It can be encapsulated in particles, such as liposomes or a water-in-oil emulsion, Tawfik and Griffiths (47), for generation of the gene product and classification. . Polynucleotide libraries (D NA or RNA) can also be classified in an encapsulated form. The size of the particles for cell-free synthesis or evolution of polynucleotide can be smaller than normal microbial cells: tan particles can be used. small as 0. 1 meters, provided that sufficient signal is obtained in the detection region to allow detection of the desired particles (ie, those that exhibit the desired characteristic) and their discrimination of unwanted particles. For example, the objective of a directed evolution experiment may be to improve the activity of an enzyme to catalyze a reaction on a particular substrate, or even to create a new enzyme that catalyzes this desired reaction. If the enzymatic reaction generates a fluorescent product, either directly through its interaction with the substrate, or indirectly (for example, via other enzymes), the microbial host cells or particles will become fluorescent. Affholter (50) and Joo (51). The classifier can be used to classify those cells or particles that exhibit higher levels of fluorescence of those with lower levels, placing those with high levels of fluorescence in the collection receptacle. The genetic material can be isolated from these cells or particles, and can be subjected to PCR amplification and further mutation, for example, by PCR and / or in vitro recombination (DNA intermixing) as in Stemmer (46). This new material can be inserted into the host cells or particles to create the i «...,. AMU .. . ,. ,; ,,. . ... " .. . ,, ^ ¡. t i: .A ^ to &. new gene products and can be classified again for as many generations as required to achieve the desired properties. Alternatively, the objective of a directed evolution experiment may be to discover a protein that binds to a particular target molecule. In this case, the target molecule can be labeled with a reporter, whose fluorescence properties change in a detectable way over the junction. Thus, if the cells contain the desired protein, they will be detected in the detection region of the invention, and will be collected according to this. The cells can be biological cells that produce the protein, or particles that encapsulate the genes and cell-free protein synthesis apparatus. The final product or desired result may be a polynucleotide, and not a protein, in which case there is no need for a protein synthesis machinery. A) Yes , In appropriate applications, the binding or catalysis can be detected or measured directly, using appropriate reporters as described.

EXAMPLE 9 - Exemplary mode and additional parameters Microfluidic chip fabrication 20 In a preferred embodiment, the invention provides a series of "T" or "Y" shaped channels, molded in optically transparent silicone rubber or polydimethylsiloxane (PDMS), preferably PDMS. This is emptied of a mold made by recording the negative image of these channels on the same type of crystalline silicon disc used in the semiconductor manufacturing. As described above, the same «». ^ * »«. d »^. ,. . ... .. ,. " .. , ,,, _., techniques for forming semiconductor feature patterns are used to form the pattern of the channels. The uncured liquid silicone rubber is emptied onto these molds placed in the bottom of a Petri dish. To accelerate the curing, these cast molds are baked. After the PDMS has cured, it is removed from the top of the mold and cut. On a chip with a set of channels forming a "T", three holes are cut in the silicone rubber at the ends of the "T", for example, using an orifice cutter similar to that used to cut holes in cork, and sometimes called cork borers. These holes form the sample, waste and collection receptacles in the finished device. In this example, the hole at the lower end of the T is used to load the sample. The hole in one arm of the T is used to collect the classified sample, while the opposite arm is treated as waste. Before use, the PDMS device is placed in a hot HCl bath to make the surface hydrophilic. The device is then placed on a strip of square microscope cover (25x25 mm) No. 1 (150 μm thick). The cover strip forms the floor (or ceiling) for the three channels and receptacles. The chip has a detection region as described above. Note that any or all of these manufacturing and preparation steps can be done by hand, or they can be automated, such as the operation and use of the device. The above assembly is placed in an inverted Zeiss microscope. A carrier holds the cover strip, so that it can be manipulated by the x-y positioning mechanism of the microscope.

This carrier also has mounting surfaces, which support three electrodes, which im- plement the electro-osmotic and / or electrophoretic manipulation of the cells or particles to be analyzed and classified. The electrodes are stretches of platinum wire secured with tape on a small piece of glass cut from a microscope slide. The wire is bent into a wide g-shape, which allows it to reach one of the receptacles from above. The cut glass acts as a supporting plate for each of the electrodes. They are attached to the usual carrier with glue tape from both sides. This allows the flexible positioning of the electrodes. Platinum wire is preferred because of its low consumption ratio (long life) in electrophoretic and electro-osmotic applications, although other metals such as gold wire may also be used.

Device loading To operate the device for sorting, one of the receptacles, for example the collection or waste receptacle, is first filled with a shock absorber. The three channels, starting with the channel connected to the filled receptacle, are filled with shock absorber via capillary action and gravity. Preferably, no other receptacle is loaded until the channels are filled with cushion to prevent the formation of air pockets. After the channels are filled, the remaining receptacles can be loaded with dampener, as needed, to fill or balance the device. The inlet or sample receptacle is normally charged to the latter, so that the flow of liquid in the channels is * - '•' • - - - "» * • - '' - directs initially to the same In general, equal volumes of buffer or sample are loaded in each cavity, this is done in order to prevent a flow net of liquid in any direction, once all the receptacles are charged, including loading the sample receptacle with sample.In this mode, it is preferred that the material flow through the device (i.e., the sample flow) be driven only by the electrodes, for example, using electro-osmotic and / or electrophoretic forces.The electrodes may be in place during charging, or they may be placed in the receptacles after charging, to contact the damper.

Electrodes Two of the above electrodes are driven by op-amps capable of supplying voltages of + -150V. The third electrode is connected to the electrical ground (or zero volts) of the high-voltage op-amp electronics. To classify the operation, the handled electrodes are placed in the collection and disposal receptacles. The ground electrode is placed in the sample receptacle. The op-amps amplify, by a factor of 30, a control voltage generated by two digital-to-analog converters (DACs). The maximum voltage that these DACs generate is + -5V, which determines the amplification factor of 30. The limit of 150 V is determined by the power supply to the amplifiers, which are estimated at + -175 V. These DACs reside on a card (a Lab PC 1200 Card, available from National Instruments, Austin, TX), mounted on b ^ eUiuMlt ^ kUi ^^ ilÉÜIiÉÉ a personal computer. The card also contains multiple channels of digital analogue converters (ADC's), one of which is used to measure the signal generated by photomultiplier tubes (PMTs). This card contains two DACs. A third DAC can be used to drive the third electrode with an additional high-voltage op-amp. This would provide a higher voltage gradient, if desired, and some additional operational flexibility. Without being bound by any theory, it is believed that the electrodes handle the flow of the sample through the device using electro-osmotic or electrophoretic forces, or both. To initiate the movement of cells or particles to be classified, a radiant voltage of the channels is established. This is done by generating a voltage difference between the electrodes. In this example, the voltage of the two driven electrodes rises or decreases with respect to the ground electrode. The polarity of the voltage depends on the charge of the cells or particles to be classified (if they are charged), the ions in the buffer, and the desired flow direction. Because the electrode in the sample receptacle in the examples is always at zero volts with respect to the other two electrodes, the voltage at the "T" intersection or branch point will be at a voltage above or below zero. volts, as long as the voltage of the other two electrodes is high or decreased. Normally, the voltage is set or optimized, usually empirically, to produce a flow from the sample receptacle to a junction or ram down point where two or more channels meet. In this example, where two channels are used, a channel is usually a waste channel and ends up in a waste receptacle; the other channel is a collection channel and ends in a collection receptacle. To direct the items or cells to a particular channel or arm of the "T" (eg, pick-up or waste), the voltage at the electrode in a receptacle (or multiple receptacles, in multiple-channel modes) becomes the more than the voltage at the junction or branch point, where the channels are located. Electrode voltage in a receptacle with two or more receptacles is high or low., to produce a radius between the receptacle and the ramification point This causes the flow to move down the channel towards the receptacle, in the direction produced by the gradient. Normally, the voltage of the electrode in the waste receptacle is high or decreased with respect to the voltage in the collection receptacle, to direct the flow to the waste channel and the waste receptacle, until a particle or cell is discharged. is identified for collection. The flow is diverted to the collection channel and the collection receptacle by adjusting the voltages on the electrodes to eliminate or reduce the gradient to the waste receptacle, and to provide or increase the gradient to the collection receptacle. For example, in response to a signal indicating that a cell has been detected for classification, by examination in a detection region upstream of the branch point, the voltage at the waste and collection points can be changed, to divert the flow from one channel and receptacle to the other.

The voltage at the point of ramification (the intersection voltage) is determined by the desired voltage g (for example, Volts / mm) by the length of the sample receptacle electrode to the branch point (g). radiant x distance), which in this example, is placed where all the channels of the "T" intersect. The gradient also determines the voltage at the waste or collection electrode (gradient x distance from sample receptacle to collection receptacle). Conceptually, the channels and receptacles of the "T" can be treated as a network of three resistors. Each segment of the "T" is carried as a resistor, whose resistance is determined by the cond uctivity of the damper and the dimensions of the channel. A voltage difference is provided through two of the resistors, but not the third. If the electrodes in each of the three receptacles are equidistant from the branch point, then each channel will have the same resistance. For example, assume that each section of the "T" has 1 00 K ohms of resistance. If 100 volts are applied through two of the resistors and the third resistor is left unconnected, the current at the junction of the two resistors would be 50 volts. If a 50 volt voltage source is 20 connected to the end of the third resistor, the voltage at the junction will not change. That is, a net of zero volts is established through the third resistor, there is no voltage gradient and no flow is started or changed. If a different voltage is applied, a gradient can be established to start or direct the flow to one channel or another. For example, to change the flow direction from one arm of the "T" to the other, the voltage values of r.?ié**e,i > ? U? -jMl.m, J 1 '* »*' '- * - fc- ~ the two electrodes handled are changed. The voltage of the ion remains igual. If the electrode distances of the "T" intersection are not equal, then the voltages can be adjusted to accommodate the resulting differences in effective channel resistance. The final result 5 is still the same. The electrode in the channel receptacle, which is temporarily designated as not receiving particles or cells, conforms to the voltage of the "T" intersection. The voltage at the other lead electrode is adjusted to provide a gradient that directs the flow of cells or particles to that receptacle. Thus, cells or particles can be sent to one channel or another, and finally to one cavity or another, by effectively opening a channel with a net voltage or relative voltage while keeping the other channel or channels closed by a net voltage gradient. or relative of zero. In a preferred embodiment to classify according to the In the invention, a slight downward flow is desired in the channel that is "turned off". This keeps the particles or cells moving away from the point of ramification (the one "T" ion), particularly those that have already been directed to that channel. Thus, a small non-zero gradient is established in the "off" or unselected channel. The channel The selected one is provided with a significantly greater gradient, in order to quickly and effectively divert the desired cells or particles towards that channel. The placement of the receptacles and their electrodes with respect to the branch point, and in particular their distance from the point of branching is an important factor in driving the flow of particles or - ^ a ^^ aaafeaaatota. cells as desired. As the receptacles and electrodes are brought closer to the branch point, it becomes more important to accurately position the electrodes, or to adjust the voltages precisely.

Detection optics In this example, a Ziess Axiovert 35 inverted microscope is used for the detection of cells or particles for classification. The objective lens of this microscope looks up, and is directed to the detection region of the described microfluidic chip, through the cover strip, which is, in this example, the floor of the device. This microscope contains all the components for epifluorescence microscopy. See, Inoue pp 67-70, 91-97 (52). In this mode, a mercury arc lamp or argon ion laser is used as the light source. The mercury lamp provides a broad spectrum of light that can excite many different fluorophores. The argon ion laser has greater intensity, which improves the sensitivity of detection, but is generally restricted to fluorophores that excite at 488 or 514 nm. The mercury lamp is used, for example, to classify beads as described herein elsewhere. The laser is used to classify E. coli bacterial cells of GFP as described herein elsewhere. The high-energy argon ion beam is expanded to fill the illumination port of the microscope, which matches the optical characteristics of the mercury arc lamp and provides fairly uniform illumination of the entire image area in one . ¿. -? ,. . ,. . ,, • ... . . ... . , - > ... , > ^ _- ,, ...,, similar to the mercury lamp. However, it is a bit wasteful of laser light. If a laser of lower energy ed is used, the laser light is focused down to coincid go with the reg ion detection chi p, to log rar the same intensity and uniformity of ilum ination or 5 similar with less consum or energy a. The objective used in the example is an Olympus PlanApo oil immersion lens 60x 1 .4N .A. The lenses are of the infinite corrected type. An oil immersion lens allows to collect a substantial percentage of the hemisphere of 1 80 g of light from the sample. This allows some of the highest possible sensitivity in fluorescence detection. This microscope has 4 optical ports and includes the ocular vision port. Each port, except the eyepiece, takes advantage of -20% of the available light collected from the sample when changing in the optical path. Only the eye port can see 1 00% of the light collected by the objective. In this mode, a color video camera is mounted on one port, another has an adjustable slot. Zeiss, whose total light output is measured with a photomultiplier tube (PMT). The fourth port is not used. The microscope focuses the image of the sample on the plane of the adjustable slit. An achromatic lens to light the light from the slit image on the active area of the PMT. Immediately in front of the PMT window, a specific bandpass optical filter is placed for the fluorescence to be detected. The PMT is a kind of lit side and does not have a highly uniform sensitivity through its active area. When transmitting the image to the PMT with the achromatic, it does not Uniformity is averaged and its effect on the measured signal is greatly minimized. This also allows an almost ideal performance of the bandpass filter. A beam splitter of 20% has been placed in the light path between the achromatic and the filter. An eyepiece with a reticle 5 re-forms the image of this portion of the aligned light. This allows to see the adjustable slit directly, to ensure that the detection area measuring the PMT is in focus and aligned. This adjustable slit allows to provide windows with a specific area of the channel for detection. Its width, height and position x, and are adjustable, and conceptually define a detection region on the chip. In this mode, the microscope is usually adjusted to see a length of 5 μm (micrometers) of the channel directly below the "T" inserts. The PMT is a current output device. The current is proportional to the amount of light incident light in the photocathode. 15 transimpedance amplifier converts this photo-current at a voltage which is digitized by the Lab PC 1200. Card This allows interpret the image to select cells or particles having an optical reporter for classification, as they pass through the detection region, based for example on the amount of light or fluorescence measured as an indication of whether a cell or particle has a predetermined level of reporter and should be chosen for collection. The voltages on the electrodes of the chip can be adjusted or changed in accordance with this determination, for example, by signals initiated by or under the control of a personal computer acting in agreement with the Lab PC card 25 1200.

-. ~ ¿T - *** a * ¿*? * ± ~? B ....?.-....-- > - i. . . * ---.... - -. .. ... ^. ^ .. ^ .. ^^^^. ,.?.? . ^^. ^ ^ ^ Rn ^ s absorbance detection to detect In another embodiment cells or particles, absorbance detection is used, which typically uses relatively larger wave lengths of light 5, for example, ultraviolet (UV ). Thus, for example, a UV light source can be used. Additional objective lenses can be used to form an image in the detection region, so that the lenses are preferably placed from the upper surface, if the PDMS device is made reasonably thin. The measurement of transmitted light, ie, not absorbed by the particle or cell, using an adjustable slit, eg, an adjustable Zeiss slit as described above, is similar to that used in fluorescence detection. A spectrophotometer can also be used. As the particles or cells pass through the window of Detection attenuates light, allowing the detection of particles or cells having a desired characteristic, and particles or cells lacking it. This can improve the accuracy of particle classification, for example, when classifying based on a quantity of a characteristic, rather than the presence of the characteristic alone, or for confirm a signal. It is noted that in some cases, absorbance detection may be detrimental to certain wavelengths of light for some biological material, for example, E. coli cells at shorter wavelengths (UV). Therefore, the biological material to be classified in this way should be tested first under various wavelengths of light using routine methods in the art. Preferably, a longer wavelength can be selected, which does not damage the biological material of interest, but which is sufficiently absorbed for detection.

Optical entrapment In another embodiment, an optical trap, or laser tongs, can be used to classify or direct cells in a PDMS device of the invention. An exemplary method to achieve this is to prepare an optical trap, methods for which are well known in the art, which are focuses on the "T" intersection next to, and preferably downstream of, the detection region. Different pressure gradients are established in each branch. A larger gradient in a branching channel creates a dominant flow of particles or cells, which should be directed towards the waste channel. A second smaller gradient in another Branch channel should be established to create a slower flow of fluid from the "T" intersection to another channel for collection. The optical trap remains in a "off" mode until a desired particle is detected in the detection region. After the detection of a desired characteristic, the particle or cell is "trapped", and with it directed or moved to the predetermined branching channel for collection. The particle or cell is released after it is delivered to the collection channel when the trap laser is turned off. The movement of the cell or particle is further controlled by the flow in the collection receptacle. The optical trap retains its focus on intersection "T" until the detection region detects the next cell or particle. Optical entrapment flow control allows a similar flexi bility in damper selection as a pressure driven system. In addition, pressure gages can be easily established by adjusting the volume of liquid added to the receptacles. However, it is noted that the flow velocity will not be as fast when the pressure in one channel is above the ambient pressure and the pressure in another is below.

Forward Classification In an electrode driven mode, before charging the sample and buffer receptacles and placing the electrodes, the electrode voltages are set to zero. Once the sample is When charged and the electrodes are placed, the voltages for the operated electrodes are adjusted, for example, by using a computer control with a commutating prog ram that incites the desired voltages, for example, the voltage differential between the electrodes of sample and scrap. If the three receptacles are equidistant from the "T" intersection, a voltage will be slightly more than half of the other. In a normal run, the voltages are adjusted by the program to start with the direction of particles or cells to the waste channel. The user is prompted by the threshold voltage of the PMT signal to identify a cell by classification, ie, deviation to the channel and collection receptacle. I also know sets a delay time. If the PMT voltage exceeds the threshold .. ..to*. j *? A, m? e fixed, the electrode voltages handled are changed and then, after the specified delay time, the voltages are changed again. The delay allows the party or cell to select enough time to travel through the collection channel, so that it will not be redirected or lost when the voltages are changed again. As described above, a slight radiant is maintained in the waste channel, when the voltages are changed, to provide continuity in the flow. It is not strong enough to keep the particle or cell moving towards the other can or channel "off", if it is too close to, or is still in, the ramification point. The value of this delay depends mainly on the velocity of the particles or cells, which is usually linearly dependent on the voltage gradients. It is believed that the effects of momentum do not influence the time of delay or the classification process. The particles or cells change the direction almost instantaneously with changes in the direction of the voltage gradients. Unexpectedly, experiments have so far failed to vary the voltages faster than the particles or cells can respond. By way of Similarly, experiments so far have shown that the dimensions of the channels do not affect the delay, in the time and distance scales described, and using the described electronics. In addition, the speed with which the cells change direction even at high voltage gradients is significantly less than that required to move them downward in the appropriate channel, when using a forward classification algorithm. Once the voltage and delay value are entered into the program, a sorting circuit enters, in which the ADC of the Lab PC 1200 card is scrutinized until the threshold value is exceeded. During that time, the flow of particles or cells is directed towards one of the channels, usually a waste channel. Once the threshold is detected, the previous voltage change sequence is initiated. This directs a selected cell or particle (usual and most preferably one at a time) to the other channel, usually a collection channel. It will be appreciated that the cells or particles are being classified and separated according to the threshold criteria, regardless of which channel or receptacle is considered "waste" or "collection". In this way, the cells can be removed from the sample for further use, or they can be discarded as impurities in the sample. After the change cycle is complete (that is, after the delay), the program returns to the ADC scrutiny circuit. A counter has also been implemented in the change sequence, which keeps track of the number of times the change sequence is executed during a program run. This should represent the number of cells or particles detected and classified. However, there is a statistical possibility that two cells or particles can pass through simultaneously and be counted as one. In this mode, the program continues in this scrutiny circuit indefinitely until the user leaves the circuit, for example, at - - - - ••• - * 'Write a password on the computer keyboard. This adjusts the DACs (and the electrodes) to zero volts, and the sorting process stops.

Reverse Sorting The reverse sort program is similar to the forward sort program, and provides additional flexibility and an error correction resource. In the case of a significant delay in changing the direction of flow in response to a signal to bypass a selected cell or particle, for example, due to momentum effects, the reversible classification may change the overall direction of flow to overlay. and redirect a cell or particle that is initially diverted to the wrong channel. Experiments using the described electrode array show a problem of delay and a rate of error that are sufficiently low (ie, virtually non-existent), so that the reversible classification does not seem necessary. The algorithm and method may be beneficial, however, for other modalities, such as those that use pressure-driven flow, which although benefiting from the prevention of high voltages, may be more susceptible to momentum effects. If a cell is detected for flow separation, and the change is not fast enough, the cell will end up going to the waste channel with all the other undifferentiated cells. However, if the flow stops as soon as possible after detection, the cell will not get too far. A smaller driving force can then be used to slowly drive the particle in the direction iulkik ^^^ iMiMiAaiItlil Reverse back to the detection window. Once it is detected for a second time, the flow can be changed again, this time directing the cell to the collection channel. Having captured the desired cell, the flow at a higher speed can be resumed until the next cell is detected for classification. . This is achieved by altering the voltages at the electrodes, or altering the analog pressure gradient, according to the principles described above. To move cells at higher speeds, for a faster and more efficient classification, higher voltages may be required, it could be harmful to cells, and it can be fatal to living cells. Preliminary experiments indicate that there may be a limit to the exchange of voltage and speed in a system driven by electrodes. Accordingly, a pressure-driven flow can be advantageous for certain embodiments and applications of the invention. The reversible classification can be advantageous or preferred in a pressure driven system, since the hydraulic flow change may not be done as quickly as the voltage change. However, if a main or waste stream can move fast enough, there may be a net gain in speed or efficiency over the voltage change, even when the flow is reversed and temporarily lowered to provide an accurate classification. Pressure-driven applications can also offer greater flexibility in the use of shock absorbers or carriers for sample flow, for example, because a response to electrodes is not necessary. "^^ * ^ ** ~ ^ at-A?., JA". . ,. ^ ..,., _. »,,,«. ^^. u ... ,,., ....,. . _. ...... "Aj ...? .Í j & * ¿- People of ordinary skill in the art will appreciate that the examples and preferred embodiments herein are illustrative, and that the invention can be practiced in a variety of modalities, which they share the same inventive concept Itftt ^ i '^ a-a ^ A-fe-¿- • - 1Ü.

BIBLIOGRAPHY 1 J. P Nolan, L.A. Sklar, Nature Biotechnology 16, 633 (1998). 2. P. J. Crosland-Taylor, Nature (London) 171, 37 (1953). 3. U.S. Patent No. 2,656,508 issued to Coulter (1949) 4. L. A. Kamensky, M.R. Melamed, H. Derman, Science 150, 630 (1965). 5 A. Moldavan, Science 80, 188 (1934). 6 M. A. Van Villa, T. T. Trujillo, P. F. Mullaney, Science 163, 1213 (1969). 7. M. A. Van Villa, et al., A fluorescent cell photometer: a new method for the rapid measurement of biological cells stained with fluorescent dyes. (A fluorescent cell photometer: a new method for rapid measurement of biological cells stained with fluorescent dyes) (Biological and Medical Research Group of the Health Division, LASL., 1997). 8. M. J. Fulwyer, Science 156, 910 (1974). 9. H. Shapiro, Practical Flow Cytometry (Wiley-Liss Inc., New York City, 1995). 10. M. R. Melamed, T. Líndmo, M. L. Mendelsohn, Flow Cytometry and Sorting, (Wiley-Liss Inc., New York City, 1990). 11. G. Whitesides, Y. Xia, Angewandte Chemie International Edition 37, 550 (1998). 12. P. H. Li, D.J. Harrison, Analytical Chemistry 69, 1564 (1997). 13. S. Fiedler, et al. Analytical Chemistry 70, 1909-1915 (1998). 14. L. A. Sklar, Proc. SPIE 3256, 144 (1998). 15. H.P. Chou, A. Scherer, C. Spence, S.R. Quake, Proc. Nati Acad. 5 Sci. USA 96: 11-13 (1998). 16. A. Ashkin, J. M. Dziedzic, Science 235, 1517 (1987). 17. A. Ashkin, J. M. Dziedzic, Nature 330, 769 (1987). 18. T. N. Buican, M. J. Smyth, H. A. Verissman, Applied Optics 26, 5311 (1987). 10 19. C. Spence, S. R, Quake, "Transformation of cells with DNA sorting on microchips." (Transformation of cells with DNA classification into microchips); personal communication, 1998. 20. R. V. Hare, "Polyvinylsiloxane impression material." (Polyvinylsiloxane printing material); US patent no. 5,661,222, 15 1997. 21. M.U. Kopp et al., Science, 280: 1046 (1998) 22. D.J. Harrison et al., Science, 261: 895 (1993) 23. J.P. Brody, "Valveless Microswitch" (Microchange without valves), U.S. Patent No. 5,656,155 (1998). 20 24. Aine, H.E., et al., U.S. Patent No. 4,585,209 (1986). 25. Baker, D.R., in Capillary Electrophoresis (capillary electrophoresis), John Wiley Sons, New York. 1995. 26. Ballantyne, J.P., et al., J. Vac. Sci. Technol. 10: 1094 (1973). 25. 27. Castro, A., et al., Anal. Chem.85: 849-852 (1993).

^ MiM < to | j | M ^ ^ M ^^^^ h ^? ^^^^^^^^^^ Mi ^ M ^^ _ ^^^ _? aMM ^ éaÉrflt ..... ....... | - F • V'giiVnir- '28 Goodwm, P.M., et: al., Nucleic Acids Research 21 (4): 803-806 (1993) 29 Gravesen, P, et al., U.S. Patent No. 5,452,878 (1995). 30 Haugland, R P, in Handbook of Fluorescent Probes and Research 5 Chemicals (Manual of Fluorescent and Chemical Research Probes), 5th ed, Molecular Probes, Inc., Eugene, OR (1992). 31 Keller, R A, et al., British patent No.2,264,296 (10/95). 32 Krutenat, R.C., Kirk-Othmer Concise Encyclopedia of Chemical Technology (Concise Encyclopedia of Chemical Technology of Kirk-Othmer), John Wiley & Sons, New York (1985) 33 O'Connor, J.M., US Patent No. 4,581,624 (1986). 34 van Lintel, H T.G., U.S. Patent No. 5,271,274 (1993). 35 Wise, K.D., et al., U.S. Patent No. 5,417,235 (1995). 36 Thompson, L F., "Introduction to Lithography" (Introduction to 15 lithography), ACS Symposium Series 219: 1-13, (1983). 37 Angelí ef al, Scientific American 248: 44-55 (1983). 38 Manz er a /., Trends in Analytical Chemistry 10: 144-149 (1991) 39. Harpson et al., International publication No. 98/52691, published on November 26, 1998. 20 40 Bein, Thomas, Efficient Assays for Combinatorial Methods for the Díscovery of Catalysts (Methods of combination for the discovery of catalysts), Angew ,. Chem. Int. Ed. 38: 3, 323-26 (1999). 41. F. H. Arnold, Acct. Chem. Research 31, 125-131 (1998). 42. Hanes, J. & Pluckthun A. Proc. Nati Acad. Sci., USA 94, 4937 25 (1997) üUMMüaüfia 43 Hoffmuller, U. & J. Schneider-Mergener, Angew Chemie. Int. Ed. 37, 3241-3243 (1998) 44. Jermutus, L., L. A. Ryabova & A. Pluckthun, Curr. Opin. Biotechnol. 9, 534-548 (1998). 5 45 Roberts, R. W. & Szostak, J. W. Proc. Nati Acad Sci USA 94, 12297-12302 (1997). 46. Stemmer, W. P. C. Nature, 370, 389 (1994). 47 Tawfik, D. and Griffiths, A. Nat. Biotechnol. 16, 656 (1998). 48. Sambrook et al., Molecular Cloning: A Laboratory Manual 10 (Molecular cloning, a laboratory manual) 2nd edition, Cold Spring Harbor Laboratory Press (1989). 49. Benecke et al., U.S. Patent No. 5,454,472 (1995). 50. J. Affholter and F. Arnold, "Engineering a Revolution," (Designing a Revolution), Chemistry in Britain, April 1999, p.48. 15 51. H. Joo, Z. Lin and F. Arnold, "Laboratory evolution of peroxide-mediated cytochrome P450 hydroxylation," (Laboratory evolution of P450 hydroxylation of peroxide-mediated cytochrome), Nature (1999), in press. "52. Inoue, Shinya and Spring, Kenneth R., Video Microscopy: The Fundamentals (Video microscopy: the fundamentals), 2nd ed., Plenum Press, New York, New York (1997). iil.MBMflliiarai l l

Claims (50)

1. CLAIMS 1. A device for classifying biological material comprising - a microfabricated substrate having at least one main channel and at least two branch channels which are in a junction, a detection region upstream and close to the junction, comprising an apparatus detection to evaluate the biological material according to at least one characteristic as the material passes through the detection region, a discrimination region downstream of the detection region, a flow control system that responds to the detection apparatus and is adapted to direct the biological material through the region of discrimination towards a branching channel. 2. A device of claim 1, wherein at least one of the main and output channels communicates with a reservoir. 3. A device of claim 1, wherein the substrate is comprised of silicon. 4. A device of claim 1, wherein the substrate comprises a silicone elastomer. 5. A device of claim 1, wherein the biological material comprises cells. 6. A device of claim 4, wherein the silicone elastomer substrate is made of a print of a recorded silicon disc 7 A device of claim 1, wherein the flow control system is electro-osmotic. 8. A device of claim 1, wherein the flow control system is electrophoretic. 9. A device of claim 1, wherein the flow control system is dielectrophoretic. 10 1 0 A device of claim 1. where the flow control system is driven by pressure 1 1. A device of claim 1, wherein the flow control system is a microvalve.
2. 2. A device of claim 1, wherein the flow control system is an optical trap.
3. 3. A device of claim 1, wherein the flow control is a control based on the flow stop. 14. A device of claim 1, wherein the flow control is provided by a voltage gradient between the branch channels 20 and the junction. 5. A device according to claim 14, wherein the voltage gradient is generated by electrodes in the branch channels. 16. A device of claim 1, wherein the flow control is by a pressure gradient between one or more channels and the joint. • ^^^^^^^^ - «> - > • > - * - - ~ - - - - - »" * * 1 7 A device of claim 1 6, wherein the flow control operated under pressure is provided by capillary action in one or more channels of the substrate. A device of claim 1, wherein the flow control 5 comprises one or more valves 9. A device of claim 17, wherein the flow control comprises one or more valves. of claim 1, wherein the flow control is reversible, and the device of claim 1, wherein the characteristic is optically detectable 22. A device of claim 1, wherein the characteristic is determined by a fluorescent reporter 23. A device of claim 1, wherein the characteristic is I5 determined by a chemiluminescent reporter 24. A device of claim 1, wherein the characteristic is determined by a radioactive reporter. ispositive of claim 1, wherein the characteristic is determined by a spectroscopically detectable reporter. 26. A microfabricated classifier according to claim 1, wherein the predetermined characteristic is size. 27. A device of claim 1, wherein the detection apparatus comprises a light scattering apparatus. * il. JA ^^?; - 28. A device of claim 1, wherein the detection apparatus comprises an apparatus for recognizing electromagnetic radiation. 29. A device of claim 28, wherein the sensing apparatus further comprises a source of electromagnetic excitation. 30. A device of claim 29, wherein the excitation source is a light source and the recognition apparatus is a load coupled device. 31 A device of claim 1, wherein the detection stop comprises at least u of photom tubes and photodiodes and photodiodes. 32. A device of claim 1, wherein the detection apparatus is positioned to focus biological materials within a predetermined detection region. 33. A device of claim 1, wherein the width and height of a channel of the device is at least about twice as long as the diameter of the largest material to be classified. 34. A device of claim 1, wherein a channel is from about 20 μm to 200 μm wide and about 20 μm 20 to 200 μm deep. 35. A device of claim 1, wherein the biological material is a cell having a predetermined characteristic, which is identified according to a reporter signal selected from a dye, fluorescent agent, chemiluminescent agent, chromophore, Radio-isotope and optically detectable protein. ? Mj ¡lUl ^^ - m ^^^^^^^^^^^^ - ^ - '^ - T' - 111 Ii '* --rf ri - 36. A device of the claim 35, wherein the flow control is selected from electro-osmotic, electrophoretic, electrodephoretic, pressure-driven, micro-valve, laser-trapping control and based on flow arrest. 37. A device of claim 36, wherein the flow control is reversible. 38. A method for classifying a fluid mixture of cells comprising: providing the cell mixture to a main channel of a microfabricated substrate, wherein the main channel is in flux communication with at least two channels of ramification downstream, which are in a union; produce a flow of fluid in the channels; interrogate each cell by a predetermined characteristic as it passes to a detection region associated with the main channel; generate a signal indicating the results of the interrogation; direct the flow of each cell to a branching channel selected according to the signal. 39. A method of claim 38, wherein the width and height of each channel is at least about twice as large as the diameter of the largest cell in the cell mixture. 40. A method of claim 38, wherein the feature is an optically detectable reporter in or on cells. 41 A method of claim 38, wherein the cells are interrogated by at least one device selected from the group of ^ ¿^ AfeiHlbÜÜHk. m icroscopes, diodes, light estimators, lasers, light scattering devices, electromagnetic excitation sources, electromagnetic radiation detector devices, your fotom ultiplicadores bos and processors. 42. A method of claim 38, wherein the reporter is selected from a dye, fluorescer, miniscent agent, chromophore, radioisotope, and optically detectable protein. 43. A method of claim 38, wherein the flow is controlled by electro-osmosis, electrophoresis, dielectrophoresis, pressure gradient, micro-valve, optical trap and flow stop. 44. A method of claim 43, wherein the flow control is provided by a voltage gradient between the branch channels and the one ion. 45. A method of claim 44, wherein the voltage g is generated by electrodes in the branch channels. 46. A method of claim 44, wherein the main channel comprises an electrode. 47. A method of claim 43, wherein the flow control is by a pressure gradient between one or more channels and the junction. 48. A method of claim 43, wherein the pressure gradient is provided by capillary action in one or more channels of the substrate. 49. A method of claim 38, wherein the flow control comprises one or more valves. 50. A device of claim 38, wherein the flow is reversible. ÉÉÉÉÉÉÉÉ