MICROFLUIDIC SYSTEM AND METHOD FOR

ANALYSIS OF GENE EXPRESSION IN CELL-CONTAINING

SAMPLES AND DETECTION OF DISEASE

This application claims priority from U.S. Provisional Application No.

60/764,390, filed 02 February 2006. The disclosure of the aforesaid application is incorporated by reference in its entirety in the present application.

FIELD OF THE INVENTION The present invention relates to the identification of biomarkers that distinguish pre-cancerous and cancerous cells from normal cells and a microfluidic "lab-on-a-chip" system and its components, subassemblies, modes of operation, and methods of use for identifying and separating precancerous and cancerous cells from normal cells, analyzing gene expression of the separated cells, detecting disease, and monitoring disease progression and treatment outcomes.

BACKGROUND OF THE INVENTION

Head and neck cancers are the sixth most common cancer worldwide and are associated with low survival and high morbidity (1). Cancers of the oral cavity account for 40% of head and neck cancers and include squamous cell carcinomas of the tongue, floor of mouth, buccal mucosa, lips, hard and soft palate and gingiva (2, 3). Despite therapeutic and diagnostic advances, the five-year survival rate for oral squamous cell carcinoma (OSCC) has not improved significantly over the past twenty years, and remains at about 50% (2-4). In addition, aggressive treatment of OSCC cancer is controversial since it can lead to severe disfigurement and morbidity (5). As a result, many patients with OSCC cancers are either over- or under-treated with significant personal and socio-economic detriment.

Currently, clinical examination and histopathological studies are the standard diagnostic methods used to ascertain whether biopsied material is from precancerous or cancerous lesions (7). Biopsies are invasive procedures typically involving

surgical techniques. Furthermore, biopsies are limited when it comes to lesion size. For example, small lesions may go undetected or may not provide enough material for accurate diagnosis while biopsies taken from large lesions may not accurately reflect every histopatholgical aspect of the lesion. As a result, a flawed diagnosis may occur, leading to an erroneous therapeutic approach. Finally, because of its reliance on the subjectivity of the pathologist, the biopsy, as a diagnostic tool, has limited sensitivity. Thus, additional methodologies are necessary to detect pre-malignant and malignant oral cancer lesions.

Non-invasive detection of pre-malignant and malignant oral cancer cells requires easy access to the site where such cancers typically arise and a readily available source of cells. Obtaining clinical samples from saliva in the oral cavity meets both of these criteria. Collection of saliva from within the oral cavity provides a convenient and non-invasive way of sampling exfoliated cells. Saliva is a complex exocrine secretion that contains serum components, bacteria and cells. Recently, flow cytometry studies of human saliva have demonstrated that saliva is composed of alive and dead erythrocytes, leukocytes, and epithelial cells (8). Saliva, because of its cellular composition, accessibility, inexpensive and non-invasive methods of collection, is ideal as a diagnostic medium for oral cancer detection. However, in order to distinguish precancerous and cancerous cells from the cellular population in saliva, it will be necessary to identify unique biomarkers that can distinguish such pre-cancerous and cancerous cells from normal oral mucosa.

Typically, genetic changes in cancer cells lead to altered gene expression patterns that can be identified long before the cancer phenotype has manifested itself. When compared to normal mucosa, those changes that occur in the cancer cell can be used as biomarkers. Attempts to find biomarkers that identify OSCC pre-malignant and cancerous lesions resulted in several candidate genes associated with OSCC tumor progression including p53, cyclin Dl, and EGFR (9, 10). However, to date, no single gene or small group of genes has shown sufficient diagnostic utility in OSCC. Thus, as in many other cancers, clinical diagnosis will require a determination of the combined influence of many genes. Not surprisingly, expression patterns of many genes have shown dramatic correlations with tumor behavior and patient outcome.

As such, microarray analysis of several tumor types has demonstrated that global expression profiling can distinguish tumor from normal biological samples, as well as the class and subtype of cancer far superior to current histopathological diagnostic systems (11-13). In order to carry out gene profiling reliably, it is desirable to separate the pre-cancer and cancer cells from the normal cell population.

A few recent studies, although preliminary, have used microarrays to identify biomarkers associated with OSSC development and progression (14). These studies were limited and did not involve detection of precancerous/cancerous cells in the saliva using the identified biomarkers. However, saliva has recently been shown to harbor a large panel of mRNAs suggesting that novel clinical approaches using saliva for disease diagnostics are possible (15). Furthermore, these results indicate that molecular signatures in saliva that distinguish normal from pre-malignant and malignant OSCC could be identified.

Recent studies also indicate that neck and head pre-cancer (dysplastic) and cancer cells' membranes express two glycoproteins: EpCAM and HSP-47, which are not expressed or minimally expressed by normal cells.

Many cancer diagnostic protocols involve a cell sorting step to enrich cell- containing samples with cancerous and pre-cancerous cells and thus facilitate or enhance their sensitive and specific detection. Cell sorting techniques are commonly based on tagging the cell with antibody against the cell membrane antigen specific to the target subpopulation of cells. The antibody is conjugated to a magnetic bead and/or fluorophore or other label to enable cell sorting and detection. See, for example, U.S. Patents Nos. 6,190,870, "Efficient Enrichment And Detection Of Disseminated Tumor Cells," and 6,365,362, "Methods And Reagents For The Rapid And Efficient Isolation Of Circulating Cancer Cells". Cancer diagnostic protocols such as these use relatively large sample volumes, i.e., on the order of 10-20 mis. of peripheral blood. As such, these protocols are not contemplated for use in a microfluidic format.

"Lab-on-a-chip" modules and systems for detecting HIV and bacterial pathogens in oral fluids have been reported (43). The technology is based on an inexpensive, disposable, polycarbonate cassette that contains either

pneumatically-driven or hydraulically-driven microfluidic circuits with conduits, valves, mixers, reaction and incubation chambers including PCR thermal cyclers, separation columns for isolation of nucleic acids, lateral flow membranes for immunochromatography, and reagent and waste storage. The cassette is used with a reader/analyzer that supports the more expensive components such as a detector, pressure and power supply, and storage and dispensing for some of the reagents. In operation, a sample of oral fluid collected from the patient with an absorbing sponge is introduced into the cassette. The cassette is next inserted into the analyzer to make quick connections with hydraulic and/or pneumatic lines. The sample is then mixed with buffer, distributed into multiple analysis paths and metered. The cassette features four separate, parallel analysis pathways, each for detecting one of target DNA, RNA, antibodies, or antigens. See US Provisional Patent Applications Nos. 60/679,797, 60/679,798, and 60/679,816, all of which were filed on May 11, 2005. One of the fundamental factors accounting for the poor outcome of patients with OSCC, as noted above, is that a great proportion of oral cancers are diagnosed at advanced stages and, therefore, are treated late. Detecting pre-malignant or oral cancer lesions at an early stage will greatly reduce morbidity associated with late disease treatment and improve overall patient survival. For example, early detection could lead to frequent patient monitoring, dietary changes, counseling on and cessation of smoking and drinking, preventative drug administration, and/or lesion removal. Indeed, early diagnosis and treatment of OSCC has been shown to lead to mean survival of over 80% and a good life quality after treatment (6). However, no reliable screening method to detect OSCC cancers has yet been developed.

Thus, an accurate and economical means for identification of individuals with pre-malignant or early cancerous lesions is needed to improve treatment planning and to decrease patient morbidity and mortality. Furthermore, it would be desirable to have a non-invasive, rapid test that can be conducted at the point of care (i.e., a practitioner's or dentist's office) by minimally trained personnel.

The methods described herein use oral cancer as an example but are applicable for the detection of various types of cancers such as colon, stomach, urinary track and bladder, and the detection of other diseases that modify the gene

profile of cells. Samples may include oral fluids, lung aspirations, stool, urine, lymph node fluids, blood and others.

SUMMARY OF THE INVENTION The ability to non-invasively monitor cancer onset, progression, and treatment outcomes requires the identification of specific biomarkers for cancers, as well as non-invasive access to and monitoring of these biomarkers. These prerequisites are met by the present invention which provides an integrated microfluidic "lab-on-a-chip" device for low-cost, rapid identification and quantification of pre-cancerous and cancerous cells and the determination of gene signatures and expression/transcription profiles of sorted, enriched cell fractions derived from samples collected from saliva, lung aspirations, stool, urine, lymph node fluids, blood and others, or, alternatively from biopsies, cell cultures, and other clinical media, as well as analyzing samples encountered in various environmental, forensic, agriculture, food and chemical processing applications.

The present invention identifies proteins specifically expressed in pre-cancer and cancer cells, provides methods and means for identifying and quantifying precancer and cancer cells, provides means for cell separation and sample enrichment, with the added options of cell lysis, and nucleic acid isolation, amplification, and labeling to facilitate the assaying of genes (DNA) and gene transcription levels

(messenger RNA). The present invention provides for two modes of diagnostics. In the first mode, pre-cancer and cancer cells expressing certain membrane-associated cell proteins (MACP), such as EpCAM and HSP-47, will be labeled and identified. In the second mode, these cells will be separated from the normal cell population and their gene expression will be profiled. More particularly, the present invention may be embodied in one or more modules or cassettes which include microfluidic circuits, valves and chambers for performing mixing, dilution, incubation, enzymatic reactions, various separation processes of liquid samples and reagents, and DNA amplification and detection. For example, cell separation and enrichment can be performed in one module, cell lysis and isolation of target nucleic acids in another module, and nucleic acid amplification in yet another module. The system may be

modified to use available specific antibodies to measure the protein levels expressed by these genes as a redundant and accurate diagnostic test. Cell-containing samples are introduced into the device, and the cells in the sample are incubated with magnetic beads functionalized with antibodies specific to certain cell subpopulations of interest in the sample, such that the cell subpopulation of interest is selectively tagged with the magnetic beads and other labels. The cellular content of the sample is then fractionated by application of a magnetic field. The sorted target cells may then be observed and/or counted. The presence of aberrant cells may provide, in some cases, sufficient information to diagnose disease. If so desired, the sorted cells may be lysed, and the nucleic acids isolated for amplification and detection by various known methods such as microarrays. In this way, the gene content and gene expression levels of the sorted cell populations from several sites of cancers including but not limited to colon, lung, bladder, ovary, breast, prostate, blood etc, can be readily identified by including the known gene profiles from these organ suites to compare with the known normal gene patterns to identify the presence and type of disease.

Aberrant gene expression can serve as a diagnostic signature for cancer, pre-cancer, hereditary and infectious diseases, and other ailments. Gene expression levels are also informative in the monitoring of disease progression, drug customization, drug studies, and other biomedical research and clinical procedures. Gene and gene expression assessments are facilitated by first enriching the sample in certain cell types using the aforementioned cell sorting techniques. Additional information is obtainable by counting the number of cells in the fractions that contain specific proteins on their surfaces. It has also been discovered in accordance with this invention that certain

MACP, such as EpCAM and HSP-47, are particularly useful biomarkers for cancers of the oral cavity. These are transmembrane proteins expressed on the cell surface of pre-cancerous and cancerous cells, but are not expressed on the surface of normal cells. The present inventors detected EpCAM and HSP-47 expression on all cancer and pre-cancer cell lines that they tested. They have also demonstrated that oral cancer cells can specifically be identified at ratios of 1 cancer cell in 5,000 normal

cells with an accuracy of 99% using MACP antibodies conjugated to magnetic beads.

This invention provides a reliable, non-invasive diagnostic technology, based on newly-identified protein that is singularly expressed on pre-cancer and cancer cells and known gene signatures that distinguishes precancerous and tumor cells from normal cells. In one specific embodiment, this technology is adapted for the detection of pre-cancerous and cancerous cells in oral fluids. As such, the system will accept and meter a sample of oral fluid, label and capture the target cells, and, provide the means for detecting the presence of pre-cancer and cancer cells. If desired, the system will determine the number of target cells in the sample through optical, impedance measurements, magnetic measurements, or alternatively, Coulter- type cell counting. The sorted target cells can be lysed and their genetic materials isolated and analyzed. The system can incorporate on-chip RT-PCR or nonenzymatic amplification and a detection system such as fiber optic-based hybridization detection systems and electrochemical detection system. Thus, the oral cancer laboratory on a chip device disclosed herein will lead to individualized treatment and increased patient survivability. The system and method described here are applicable for the detection of other diseases and conditions that alter the gene profile (result in over and under expression of certain genes) of cells.

BRIEF DESCRIPTION OF THE FIGURES

Fig. Ia: A schematic illustration of a cell capture and detection system, utilizing a suitable antibody immobilized on a lateral flow membrane for cell capture and up-converting phosphor reporters for detection. Fig. Ib: A schematic representation of the operation of the device of Fig. Ia.

Fig 2: A diagrammatic presentation of an analyzer/reader used with the present invention.

Fig. 3: A schematic illustration of a microfluidic analysis system embodying the present invention, including means for tagging target cells with magnetic particles, magnetic cell sorting, as well as cell counting, cell lysis, and isolation of RNA in a lab-on-a-chip format.

Fig. 4: A microscopic image of (a) cells separated from a mixed cell population using magnetic beads conjugated with MACP antibodies; and (b) dyed cells (the dye specifically identifies cancer cells) as seen under fluorescence microscopy. The populations in (a) and (b) are virtually identical, indicating insignificant percentage of false positives.

Fig. 5 (A): A prototype of a magnetic, continuous flow cell-sorter.

Fig. 5 (B): The transverse deflection of the free beads, unlabeled cells, and magnetically labeled cells some distance downstream of the sample introduction point. Fig. 5 (C): An image of the junction between the sample flow channel and the "focusing buffer" flow channel.

Fig. 6: A schematic illustration of an alternative embodiment of the microfluidic analysis system of the invention, including means for cell sorting, as well as cell counting (via Coulter-type counter), lysis, and nucleic acid (mRNA) isolation in a lab-on-a-chip format.

Fig. 7: A microfluidic cassette for cell lysis and nucleic acid isolation. Two- stage cell lysis, by incubation with enzymatic reagents and chaotropic salts, is integrated with a solid-phase extraction column incorporating a porous silica 'membrane'. Fig. 8: A microfluidic integration of solid-phase extraction column with 10- μ\ PCR chamber. Both PCR inlet and outlet are valved with electrically-actuated, temperature-sensitive hydrogel phase change valves.

Fig. 9: Lateral flow assay detection of phosphor-particle labeled PCR product blotted on a nitrocellulose test strip and detected with a laser scanner.

DETAILED DESCRIPTION OF THE INVENTION

For purposes of detailed discussion, the system of the invention is described herein with specific application to oral cancer screening using saliva-based samples, but it will be understood that the systems and methods embodying this invention have wider applicability to other fields including clinical diagnosis, screening and monitoring of infectious and hereditary diseases, as well as other types of cancer and

other diseases that lead to abnormal expression of cell membrane proteins and/or gene profile; sensing for environmental, biotechnology, agriculture, and bioterrorism applications; and biomedical research and drug development. Further, the systems and methods of this invention can utilize a wide assortment of other sample types, including cell and tissue cultures, environmental samples, biopsies, and clinical samples including blood, urine, CSF, and the like.

In one embodiment of the invention, pre-cancer and/or cancer cells can be captured to a surface, such as a lateral flow membrane and a surface of a capillary, function alized with one or more suitable binding agents, for example antibodies or antibody fragment which bind specifically to at least one protein expressed in the cancer and/or pre-cancer cells sought to be identified. The binding agent can be immobilized on the capture surface using coupling chemistries well known in the art. Detection of the captured cells can be accomplished by means such as up-converting phosphor reporters (44, 64), quantum dots, and fluorophores. An example of this mode of cell capture is illustrated in Figure 1. In this particular embodiment, a cell- containing sample potentially including pre-cancer and/or cancer cells is obtained from a subject, which may be conveniently done using a device for collection and assay of oral fluids, such as those described in U.S. Patent No. 6,303,081. The lateral flow membrane and associated elements of such a device are shown in Figure Ia and include a lateral flow membrane material 11 , in the form of a strip, which functions to transport the cell-containing sample from receiving area or load pad 15, which may be composed of a capillary matrix material, to capture zone 17, under the influence of an electric field established by imposing a voltage drop across electrodes 19a and 19b. As illustrated schematically in Figure Ib, the cell-containing sample is mixed with a specific binding agent coupled to a detectable label. Preferably, the biospecific reagent is an antibody, which is coupled to an up- converting phosphor reporter 21 to form conjugate 23 with the cancer cell(s) 25 sought to be detected. Suitable capture agents 27, e.g. antibody (or antibody fragments) which bind specifically to the cancer cell biomarker are immobilized at capture zone 17 to effect capture of the labeled cancer cells which are detectable by means of a laser bean 28, which induces fluorescence in reporter 21.

An absorbent pad 29 may also be associated with lateral flow membrane 11 if desired, as shown in Figure Ia, to serve as a reservoir, which receives excess cell- containing sample and prevents its backflow into lateral flow membrane 11. The device may also have a conjugate strip as an optional feature to facilitate fabrication of the device, as disclosed in U.S. Patent 6,303,081.

The device also beneficially includes a control area 31, in which predetermined, non-target cell types are bound and detected using conventional means, such as light or fluorescence microscopy, to confirm that the device is working properly. In order to improve material transfer to the capture zone, electrophoresis effects can be employed (53). The lateral flow strip may consist of conduits fabricated in the substrate material. The cells can also be collected on a membrane. Alternatively, the cells can be mixed with magnetic particles coupled to a suitable binding agent, labeled, separated from the sample in magnetic field, labeled with such as quantum dots, gold particles, and fluorophores, washed, and detected either by eye or with an appropriate reader.

The binding of cells to a specific binding agent by means of selective interaction between a cell-associated determinant, such as membrane-bound protein or glycoprotein (e.g. cell-surface antigen, histocompatibility antigen or membrane receptor), and binding sites of the specific binding agent (e.g. antibody complementary to the cell-associated determinant) is referred to herein a "selective binding".

The term "antibody" as used herein, includes immunoglobulins, monoclonal or polyclonal antibodies, immunoreactive immunoglobulin fragments, and single chain antibodies.

The term "detectable label" is used to herein to refer to any substance whose detection or measurement, either directly or indirectly, by physical or chemical means, is indicative of the presence of the target analyte in the test sample. Representative examples of useful detectable labels, include , but are not limited to the following: molecules or ions directly or indirectly detectable based on light absorbance, fluorescence, reflectance, light scatter, phosphorescence, or

luminescence properties; molecules or ions detectable by their radioactive properties; molecules or ions detectably by their nuclear magnetic resonance or paramagnetic properties.

In another embodiment, the lab-on-a-chip cancer detection and gene signature analysis system described herein comprises an inexpensive, disposable cassette

('chip') that is used in conjunction with an analyzer/reader. All the sample processing can be done in the microfluidic conduits and chambers within the cassette. Some of the reagents are pre-stored on the cassette. A diagrammatic illustration of a representative form of analyzer/reader is shown in Figure 2. The analyzer may include, without limitation, hardware and software for control and data analysis; a pressure source, e.g., a programmable pump for fluidic propulsion; valving for the control of the air pressure in the pneumatic lines; storage and means for dispensing reagents; electric power; a detector; readout electronics; and a computer interface. The cassette is docked or nested in the analyzer/reader to make fast hydraulic, pneumatic and electrical connections and optical coupling with the analyzer. The connections are made automatically upon docking without a need for any manual intervention.

Subsequent to the sample introduction into the system, under programmed control and without human operator intervention, the sample is subjected to the following processing steps:

• Metering and mixing with buffer and functionalized magnetic beads and with labels (optional)

• Separation of interfering cells such as lymphocytes

• Separation of pre-cancerous and cancerous cells from the clinical sample

• Optional cell detection/counting to determine the presence and number of pre-cancerous and cancerous cells in the sample

• Cell lysis to produce soluble nucleic acids from the pre-cancerous and cancerous cell fraction • Separation and purification of RNA

• RT-PCR amplification or nonenzymatic amplification

• Quantitative assay of specific mRNAs in order to evaluate expression profiles

• Comparison of the detected profile with library profiles

A schematic view of an embodiment the invention and its operation for application as a comprehensive cancer detection system are illustrated in Fig. 3.

The system 110 shown in Fig. 3 comprises a cassette which is adapted to be inserted into an apparatus or processing unit (not shown) and which includes means for causing mixing of the cell-containing sample in a receiving chamber 111 and causing flow through a main flow path 112 for transporting a primary flow of material through a metering device or valve 113 into the cassette. The sample is mixed with magnetic beads coupled to a suitable specific binding agent 114 and transported into a separation chamber 118 where the cells are sorted under the influence of magnetic field 119 which is applied to the separation chamber 118. The magnetic field causes the unbound magnetic beads 114 and the magnetically labeled target cells 115 to be attracted toward and adhered to their respective collection sites 120 and 124 on a surface of the separation chamber 118. The remaining constituents, including non-target cells 116, are allowed to flow through the path to a discharge port. The collected target cells may then be counted, if desired, and subjected to lysis. Nucleic acid molecules in the cell lysate can be isolated and detected as described below.

Various magnetic means may be used to apply a magnetic field is the sample flow path, including permanent magnets of various geometries and electromagnets. The magnetic means may be incorporated in the cassette, the analyzer/reader unit, or both. The system typically includes means to cause the sample stream to be enveloped by a "focusing" buffer at the start of the flow path 112, as indicated by the arrows, 117, in Fig. 3 of the drawing. The cassette also includes one or more secondary flow paths, e.g. 121 with a flow controller 122 for introducing additional components into the separation chamber, such as an RNA stabilizing solution and chemical agents for cell lysis and binding of lysate components. The processing unit may also include means for temperature control such as a heating element 123 for

heating the collected target cells to facilitate cell lysis at downstream collection site 124. Cell lysis may also be accomplished by using an electric field to rupture the cell membrane. See, for example, U.S. Patent No. 6,783,647. Electrodes may be disposed along the sample flow path for this purpose. Cell lysis occurs under the influence of an electric field produced by the application of a source of electrical potential to the electrodes. The unbound cells and undesired sample components may flow along channel 125a through flow control device 126a to waste chamber or reservoir 127.

The cell lysate is transported to a microcolumn 130 or similar device for isolation of nucleic acids of interest that may be present therein. A lateral flow chromatographic matrix, various forms of which are known in the art, may also be used for this purpose. Additional solutions or reagents for facilitating nucleic acid isolation can be introduced into microcolumn 130 or the like through conduit 128. Such materials may include a wash and/or an elution buffer solution. Conduit 129 has a flow controller 128 and connects with the flow path 112 at the entrance to microcolumn 130. The spent lysate is discharged to the waste chamber 127 through channel 125b having a flow controller 126b. The target cell collection site 124 may include an electrode array 131 operably coupled with an impedance cell 132, for counting the collected target cells. The isolated nucleic-acids from the target calls may be amplified e.g. by PCR, in a chamber 133 provided for that purpose in the microfluidic system (or in a bench-top PCR unit) and the amplified nucleic acid can be detected by a suitable detector 134. A fiber-optic based detection array is suitable for this purpose (16,17). See also, for example, U.S. Patents Nos. 5,244,636, 5,320,814, 6,023,540, 6,210,910, 6,327,410, 6,406,845 and 6,991,939. In the system shown in Figure 3, each of the cell separation or sorting operations, the cell counting operation, the cell lysis operation and the nucleic acid amplification operation may be performed using a separate module. Alternatively, two or more of these operations may be integrated into a single module.

The term "chamber", as used herein, is not intended to imply that any element so designated necessarily has dimensions that are different from any other structural elements in the microfluidic system. Thus, the separation of target cells,

cell lysis and isolation of lysate components desired for analysis may occur at different sites along the length of a microchannel structure of substantially uniform cross-sectional dimensions, e.g. of width, height or diameter, as the case may be. Thus, the term "chamber" should be understood in terms of the operation or process occurring therein, and may alternatively be referred to as a site, zone, area or the like. In a preferred embodiment of the invention, the epithelial component of saliva and a highly specific gene signature for OSCC are used to provide a simple, reliable, inexpensive, and noninvasive methodology to accurately diagnose precancerous or cancerous oral lesions that has routine clinical utility. Clinical samples can be collected with commercially available,

FDA-approved, single-use, saliva collectors such as the OraSure UPlink™ collector. This collector and samples of oral fluids were found to be compatible with PCR amplification and capable of collecting cells (45). As the collector is pushed into a sample inlet port of the cassette and locked into place, a pre- determined volume (~1 ml) of sample is squeezed out, mixed with PBS buffer, and incubated in a sample receiving chamber with magnetic beads (~4μm in diameter or smaller, superparamagnetic, polymer-coated, functionalized with an antibody to a cell surface antigen of the target cell, and commercially available from suppliers such as Dynal Biotech). The magnetic beads bind specifically to the antigen (e.g. MACP) expressed on the pre-cancerous and cancerous cells' membranes. The desired number of beads added to the sample depends on the number of target cells and may be determined through numerical simulations of the interactions between the target cells and the magnetic beads similar to the simulations carried out by Qian and Bau (46-48). Since in the microfluidic system, the fluid motion is laminar and well organized, it may in some situations be advantageous to accelerate the rate of interactions between the beads and the cells by stirring the solution. Stirring can optionally be provided by oscillating the cell-containing sample slug back and forth a few times in the sample receiving chamber, and/or passing the contents of the sample receiving chamber through a serpentine conduit that induces secondary flows (49). Furthermore, if necessary, the presence of the magnetic beads can be used to effect agitation of the sample under the influence of a rotating magnetic field. The

objective is to alternate between two or more different flow patterns to induce chaotic advection (50, 51). The sample receiving chamber's walls can be coated with either Teflon or BASF (52) to minimize the adhesion of cells to the walls. The preferred size of the beads depends on both the interaction kinetics, as well as the separation efficiency as discussed herein below.

The transport of the material in the cassette is achieved with air pressure supplied by syringe pumps located outside the cassette, preferably in the analyzer. The flow control is achieved by mechanical valves located in the analyzer that are operable to control air pressure lines and, if desired, ice and hydrogel valves (58, 65) integrated into the cassette. The ice valves are actuated by thermoelectric units located within the analyzer's docking chamber. The valve location is pre-cooled to below the freezing temperature of the solution. This type of valve allows free displacement of air and it freezes the liquid slug once it arrives at the valve location. In this sense, the valve is self actuated and it does not require a sensor for feedback control. We have successfully operated similar valves in our present work (58) and have demonstrated that these valves can be used to seal a PCR reactor to suppress bubble formation during the thermal cycling process (66). In yet another embodiment, the liquids such as buffers and wash solutions may be stored on chip in pouches that can also be used for actuation, pumping, mixing, and valving, Many cancer diagnostics protocols rely on a cell sorting step to enrich cell- containing samples with cancerous and precancerous cells and thus facilitate or enhance their sensitive and specific detection. In order to be able to determine the genetic content of the relatively small number of precancerous and cancerous cells in the sample, it is necessary to separate and enrich these cells from among the normal cells and the other genetic material that may be contained in the sample. The epithelial cell adhesion molecules (MCAP), also known as EpCAM, may be used to effect such cell separation and enrichment. MCAP is a cell surface glycoprotein that is not expressed by hapatocytes, thymic cortical epithelial cells, gastric parietal cells, myoepithelial cells as well as many non-epithelial cells. More importantly, human epidermal keratinocytes and normal squamous epithelial cells do not express MCAP on the cell surface (39). Furthermore, MCAP is found to correlate with progression

of OSCC by being expressed on the cell surface of dysplastic oral cancer cells and highly expressed in most if not all OSCCs (39-41). Indeed, in a more recent, study MCAP was found expressed in .89% of head and neck SCC tumor specimens on more than 50% of all epithelial cells (42). During isolation of specific cells, i.e cancer cells, cell sorting techniques are commonly used and are based on tagging the cell with antibody against a cell membrane antigen specific to the target subpopulation of cells. The antibody is conjugated to a magnetic bead or fluorophore to enable cell sorting and detection. For cancers of the oral cavity the present inventors have determined that precancerous and cancerous cells express a transmembrane protein MACP on the cell surface. MACP is not expressed or minimally expressed on the surface of normal cells. MACP expression has been detected on all oral cancer and pre-cancer cell lines tested so far. Moreover, the present inventors have demonstrated that oral cancer cells can specifically be identified at ratios of 1 cancer cell in 5000 and 500 normal cells with an accuracy of 99% using MACP antibodies conjugated to magnetic beads. An example of the use of MACP for identification of cancer cells is shown m Figure 4. A mixed cell population was prepared comprising 10,000 tumor cells (dyed with green fluorescent dye) mixed with 500,000 normal keratinocytes (10,000/500,000 x 100=2.0%). Magnetic beads conjugated with MACP antibodies were incubated with the mixed cell population and the tumor cells were magnetically separated. The separated cells were visualized by microscopy. As can be seen in Figure 4, the subpopulation in (a) and (b) are virtually identical, indicating an insignificant percentage of false positives. That was evident from the fact that all of the tumor cells appeared green when viewed under fluorescent light. Similar results were obtained when separating 1,000 tumor cells mixed with 500,000 normal cells. In addition, in recent experiments, the present inventors also discovered that MCAP antibodies failed to bind to normal oral epithelial cells, but did bind to MCAP in dysplastic and OSCC tissue sections. Furthermore, when conjugated to magnetic beads the MCAP failed to pull down oral keratinocytes, but did pull down OSCC tumor cell line SCC-I .

A primary flow of material containing the sample mixed with buffer, magnetic beads, unlabelled (non-cancerous) cells, and cell-bead complexes is transported from the sample receiving chamber 111, preferably enveloped with buffer solution, and into the cell-separation chamber 118. Conducting the material flow in this way should prevent non-cancerous cells and other proteins from adhering to the chamber's walls. In the absence of an applied magnetic field, the sample and buffer streamlined flows will persist through the chamber due to the laminar flow regime conditions and the small diffusion coefficients of the cells and beads. In a particular embodiment of the invention, the separation chamber includes nickel structures, e.g. in thick film form, patterned on or embedded in the chamber's walls. When exposed to an applied magnetic field, the nickel film become magnetized and produces a nonuniform magnetic field. As a result of this magnetic field, the magnetic beads acquire magnetic dipoles and the beads and bead-cell complexes migrate (undergo magneto-phoresis) in the magnetic field gradient towards the location of the maximum field intensity (53-55). Since the beads are smaller than the cells, the beads are subjected to smaller viscous drag, and migrate faster than the cell-bead complexes towards the surface along distinct and predictable trajectories or deflection paths (see Fig. 3 and experimental results described with reference to Fig. 5). This effect can be utilized to separate the unbound magnetic beads from the cell-bead complexes. Such a separation is desirable for improving the sensitivity of the cell counting technique described herein below. In contrast, under typical operating conditions, the unbound (normal) cells, continue unimpeded with the primary material flow, generally along the axis of the separation chamber and will be washed away together with the other constituents of the saliva. This arrangement assures that only pre-cancerous and cancerous cells become adhered to the chamber's wall.

To demonstrate the operability of the invention with respect to cell sorting, a simple prototype was fabricated (Fig. 5A) and an experiment was carried out to show the feasibility of magnetic-bead assisted, continuous flow, microfluidic cell sorting. A microfluidic chamber was fabricated in a polycarbonate chip with inlet ports for

the sample and 'focusing' buffer, which constitute the primary flow stream. The buffer provides a sheath around a central stream of unlabeled cells, labeled cells, and free magnetic beads. Both streams were injected concurrently with a programmable, external syringe pump (Fig. 5A). The cassette was mounted on a microscope stage and the flow was viewed with a CCD camera. To allow flow visualization, the fluids were dyed. As shown in Fig. 5A, the sample stream (dark region) remains confined within the buffer (gray region) along the chamber's entire length. Fig. 5B shows a histogram of the transverse distances (y) traveled by the labeled cells, unlabeled cells, and free beads. The measurements were taken a short distance downstream of the sample injection point (dashed circle in Fig. 5A) in either the presence or absence of the magnetic force. In the absence of a magnetic field, the cells and beads remained in the core of the sample stream and separated from the solid walls by a blanket of buffer. When a magnetic field was applied, the magnetic beads and labeled cells were deflected from the sample stream with the free or unbound beads moving faster and at a sharper angle with respect to the axis than the magnetically labeled cells (Fig. 5B). The unlabeled cells maintained their axial trajectory. These results demonstrate the feasibility of magnetic cell sorting and enrichment in a lab- on-a-chip format.

In order to design the magnetic field to achieve the desired separation, numerical simulations can be used to estimate the magnetophoretic trajectories of the unbound beads and cell-bead complexes as functions of the applied magnetic field. The magnetic field can be calculated with the multi-physics finite element program Femlab™ for various patterns of the nickel thick films. Once the magnetic field has been calculated, the magetophoretic velocity can be obtained by equating the magnetic force acting on the bead F _M ~ 2πηi« ³ if(ηl, η ₂ ) V H ² (53) with the Stokes drag force. In the foregoing expression, T] ₁ and η ₂ are, respectively, the permeability's of the suspending medium and the bead, K is the Clausius-Mossoti function, a is the bead's radius, and H is the magnetic field intensity. The Stokes drag force for the bead-cell complex will be based on the cell's dimensions and shape. The particle's trajectory can be determined by solving the kinematic equation dx_= u, dt

where x is the position vector and u is the velocity vector. Several cases can be considered as to when one or two beads are bound to a single cell. In this simple treatment, inter-particle interactions and the effect of the particles' on the fluid motion are neglected. Although not essential for the operation of the device, an estimate of the number of pre-cancerous and cancerous cells in the sample can often yield additional useful information. In the particular embodiment of the invention shown in Fig. 3, an impedance technique for cell counting is used. This option is appealing on account of its simplicity and the fact that it can be readily integrated into the cassette at the target cell collection site. The cell counting is carried out subsequent to a wash step in which all loose material is removed from the cell separation chamber.

The impedance cell counter can be implemented as described herein, although other design variations and methods of fabrication will be apparent to those skilled in the art. Briefly, using photolithographic techniques, a patterned gold electrode is formed on one of the separation chamber's walls. The cell-bead complexes descend on the electrode surfaces and cover part of the electrodes' areas, blocking the ionic current path between opposing electrodes and increasing the system's impedance. The change in the impedance will be correlated with the number of cells (56, 57). For instance, Xiao et al. (57) describes a simple equivalent circuit model, corroborated by experimental studies, demonstrating a linear relationship between impedance change and the number of cells.

In another particular embodiment of the invention, and as an alternative to the optimal use of impedance based cell counting, a Coulter-type cell counter can be used as shown in Fig. 6. Specifically, after the target cells are separated, the collected cells will proceed through a cytometry stage where the number of magnetic bead bound cells will be counted. The cell counter can be constructed out of a plate that intersects the flow path and that includes one or more small orifices permitting the passage of suspended cells - one at a time. Two electrodes that straddle the orifice will establish an ionic current across the orifice. The current will be disrupted when a cell passes through the orifice. This event will be registered as a (negative) current spike on an ammeter. The number of spikes will indicate the number of

translocating cells. In principle, the cell counter can discriminate between unbound magnetic beads and labeled target (precancerous or cancerous) cells with one or more bound magnetic beads. If the free beads induce undesired signals, they will be removed from the flow stream as explained above. In recent years, a number of highly sensitive magnetic field detection devices have been developed, such as giant magnetoresistance (GMR) (55, 59). These sensors are capable of measuring extremely weak magnetic fields such as fields generated by individual beads. Indeed, various systems for the capture and detection of micron-size beads have been developed (60-61). A separate Coulter counter can be inserted to count the unlabeled cells (not shown in Fig. 6). For optimization of the Coulter-type cell counter, a finite element model can be used to predict the ionic current in the absence and presence of cells as a function of electrodes ¹ location and pattern.

Alternatively, optical methods may be used to detect the presence and quantify the number of isolated cells. In this embodiment, one MACP (i.e.,

EpCAM) may be used to bind with the functionalized magnetic bead while the other MACP (i.e., HSP-47) may be used to attach a quantum dot, fluorophore, gold particle or another label known in the art.

In prior work, the present inventors have successfully adapted the materials and methods used in commercial benchtop silica/chaotrope nucleic acid extraction and isolation kits (e.g., Qiagen RNEasy™ kit) for incorporation into microfluidic devices. For example, Figure 7 shows a module with lysis chambers and a microcolumn packed with a porous silica matrix. When this module is utilized, cells collected from the separation step are mixed with an RNA stabilizing reagent such as KNAlater™, which is an aqueous solution that rapidly permeates cells to protect cellular RNA from degradation. This process inactivates RNAses and prevents alteration of transcription levels. The RN Alater™ treatment can also be done at an earlier stage of the process, upstream from the cell separation stage (i.e., in the mixer/incubation chamber). The sample is next mixed with a chaotropic salt, such as guanidium HCI (6M), that denatures cellular proteins and causes the cells to rupture. The resulting cell lysate is then mixed with ethanol and forced through the

microcolumn packed with porous silica Nucleic acids selectively bind to the silica m the presence of the chao trope. Contaminants are rinsed from the bound nucleic acid using several wash steps, and the nucleic acids will then be eluted using a low-salt, neutral pH buffer To provide some measure of the performance of the microfluidic extraction column, a few details on such devices can be mentioned The microfluidic nucleic acid isolation devices operate in a continuous-flow mode (100 to 400 microliters per minute), with loading, wash, and elution buffer volumes between 50 and 500 μ\, depending on sample size, and typically achieve nucleic acid yields of 50-80% with gram-positive bacteria samples.

The isolated nucleic acid (mRNA) can be assessed with a bench top PCR (or RT-PCR) using primers specific for target genes, followed by gel electrophoresis, or by usmg customized small DNA microarrays, prepared using known methodology, for the genes of interest Alternatively, the PCR reactor and/or the array detector can be integrated into a separate microfluidic module The mRNA may also be amplified using non-enzymatic techniques, such as biobarcode-based amplifications (67 and 68) Figure 8 shows a microfluidic cassette integrating a microcolumn and a PCR thermal cycler chamber with controlled heating and cooling provided by a thermoelectric element The 10-ul PCR chamber is sealed with electrically-actuated, hydrogel valves The PCR products are labeled with up-converting phosphor reporter particles conjugated to pπmers specific to the target pathogen, and are detected in a lateral flow immunoassay using laser-induced fluorescence (Fig 9) The up-converting phosphors shift up the frequency of the emitted signal, avoid background interference from auto fluorescence effects, do not bleach, and provide higher sensitivity than conventional fluorescent detection techniques (44).

Expression profiles of tumor specimens and normal adjacent tissue from OSCC patients have been established by microarray analysis. Briefly, a fluorescent target derived from mRNA of interest is hybridized to Affymetnx Human Genome Ul 33 Plus 2.0 Array Each chip has a set of human maintenance genes used to facilitate the normalization and scaling of the array experiments This allows the analyst to determine which genes are uniquely expressed and at what level they are

expressed. The integrity of the RNA extracts is confirmed by Northern blots, gel electrophoresis, and Agilent Bioanalyzer.

To identify a gene expression signature for OSCC, the present inventors have analyzed 55 OSCC primary tumors and 18 normal adjacent samples. In the initial analysis, the genes differentially expressed between normal and tumor samples were ranked using t-test/ANOVA genes with an adjusted p-valve using a method by Benjamini and Hochberg (BH) to correct for multiple testing (18). This analysis revealed 686 highly significant genes down-or-up-regulated in OSCC tumors with p<0.0001. Several of these genes are un-characterized or only minimally understood. However, many of the genes are suggested to play a role in tumor development, being involved in the extracellular matrix (ECM), cytoskeleton, cell- ECM adhesion, cell-cell adhesion, cell motility, proteolysis, and cell signaling.

Several strategies have been developed to distinguish the significant genes from the noise generated when large sets of genes are assessed. To determine which genes identified by the t-test and ANOVA were the best at discriminating between normal and tumor samples, Significance Analysis of Microarrays (S AM, downloadable software from the University of Stanford; 19) was used. Using SAM and a false positive discovery set at zero (0), 227 genes with a highly significant p- value <0.0001 were identified, as calculated by the t-test within the SAM program. Of these 227 genes, 52 were up-regulated in OSCC primary tumors, while the remaining 175 genes were down-regulated. These genes were then ranked using CART-based approach according to whether they were expressed in all tumors and all normals (20). Because of space limitations, the ranking of 25 up-regulated genes is shown in Table 1. Table I.

Gene Tide Probe Set ID D value collagen, type IV, alpha 1 211980_at 3.01E-07 myosin EB 212365_at 5.91E-O5

KDEL (Lys-Asp-Glu-Leu) 219479_at 0.00015 containing 1 secernin 1 201462 at 0.000192 solute carrier family 20 (phosphate 201920_at 0.000192 transporter), member 1

matrix metalloproteinase 1 204475_jιt 0.000192

(interstitial collagenase) myosin IB 212364_at 0.000192 snail homolog 2 (Drosophila) 213139_at 0.000192

Ras-induced senescence 1 213338_at 0.000192 signal transducer and activator of AFFX- 0.000192 transcription 1, 9IkDa HUMISGF fibronectin 1 211719_x_at 0.000359 kinesin family member 20A 218755_at 0.000401 collagen, type IV, alpha 2 211964_at 0.000417 proteasome.(prosome, macropain) 2OOO39_s_at 0.00043 subunit, beta type, 2

WNTl inducible signaling pathway 206796_at 0.000479 protein 1 high-mobility group box 3 203744 at 0.000482 tenascin C (hexabrachion) 201645_at 0.00049 bone morphogenetic protein 1 20270 l_at 0.000501 fibronectin 1 216442_x_at 0.000541 myosin X 201976_s_at 0.000553 collagen, type III, alpha 1 (Ehlers- 21507C _^ s_at 0.000553

Danlos syndrome type IV, autosomal dominant) ubiquitin-conjugating enzyme E2C 202954_at 0.000564 fibronectin 1 210495_x_at 0.000639 procollagen-lysine, 2-oxoglutarate 202185_at 0.000679

5-dioxygenase 3

TPX2, microtubule-associated 210052_s_at 0.000829

Protein homolog (Xenopus laevis)

Although several genes identified by SAM are relatively unknown, many have been implicated in OSCC tumor invasion or other cancers. These include molecules associated with the ECM, matrix proteolysis, signal transduction, angiogenesis, differentiation, cell adhesion, migration, proliferation, and carcinogenesis. For example, genes identified by SAM include the laminin-5 (Ln-5) 2 chain and 3 chain, protein kinase C, urokinase, insulin-like 3 ankyrin, MMP-I, HSP-47, tenascin, and Rho GTPase activating protein (21-30). Cofirming results described herein, several of these genes have been identified in recent microarray studies using OSCC as well as single gene biomarkers for OSCC (31-33). Furthermore, it is apparent that some genes identified by SAM are associated with stroma and lymphoid cells. It is becoming increasingly clear that host stroma cells

and infiltrating cells contribute to tumor invasion by release of several factors, including growth factors and chemokines (34). Together, these results demonstrate that the present inventors have identified a highly significant gene signature that can be used to distinguish OSCC tumors composed of complex cell and tissue types from normal tissue.

To identify genes capable of discriminating between normal and tumor tissue individuals, we applied a supervised classification approach used previously with success (12). The 73 samples described above were used as a validation set and 10 OSCC tumor/normal specimens received from the University of Pittsburgh (in a collaboration project with Dr. Jennifer Hunt) as a test set. Using k-nearest neighbor/Fisher's exact test method and the 686 genes identified by ANOVA/t- test/BH correct at p≤O.OOOl as the gene pool, all samples from the University of Pittsburgh were correctly identified as either normal or tumor, using a gene prediction signature of only 20 out of the 686 genes in the gene pool. Finally, using artificial neural networks (ANN, alyuda Neurointelligence software) the present inventors have confirmed their prediction signature. Thus, using standard prediction/classification algorithms, OSCC tumor status can be correctly predicted using a gene signature of only 20 genes.

The terms "expression profile" and "gene profile" and "gene expression signature," are used interchangeably herein, refer to those subsets of genes which are differentially expressed in cancerous cells when compared to their expression levels in cells from normal individuals (e.g., expression of a given gene may be increased or decreased in a cancer cell relative to the expression level observed in the normal or non-cancerous state). The identification of differences in gene profile between subjects with cancer and normal individuals facilitates clinical discrimination between members of those respective groups. Microarrays (e.g., cDNA microarrays) may be used to determine the gene profile of cells from subjects with cancer and/or normal individuals. Standard linear discriminate analysis can be used to assess the differences in gene profiles between subjects with cancer and normal individuals. Gene profiles may be established for any cancer as well as other diseases.

The present invention is unique in several respects, most notably the provision of at least several different modules for sorting, counting, and lysing cells and RNA isolation within the lab-on-a-chip. Such a Iab-on-a-chip will be useful for many applications in oral cancer diagnosis and research detection of metastatic cells in blood or lymph, determining treatment response, cancer protein expression arrays, cancer DNA cgh-arrays, identification of gene changes associated with treatment, drug resistance, smoking and drinking induced gene changes, etc. Furthermore, the chip can easily be modified for other studies, for example by changing antibodies the chip could be used to identify stem cells/RNA from mixed populations of cells, detection of bacterial infections, lymphocyte detection, genetic predisposition to disease, detection and analysis of DNA mutations, DNA adduct detection, etc.

In addition, the present inventors have recently determined that the gene signature for OSCC as opposed to laryngeal cancers is quite distinct. Thus, applying different gene signatures in the detection phase it may be possible to identify cancer cells originating from other sites in the body, for example lung, esophageal, salivary, laryngeal, etc. Finally, the combination of these processes in one chip will greatly reduce the time to perform these standard laboratory assays from 12-16 hrs to -2-3 hrs., thus increasing efficiency and reducing cost.

A number of literature and patent references are cited throughout the foregoing specification in order to describe the state of the art to which this invention pertains. The entire disclosure of each of these citations is incorporated by reference herein.

While certain embodiments of the present invention have been described and/or exemplified above, various other embodiments will be apparent to those skilled in the art from the foregoing disclosure. For example, the system described herein could be used to determine initial treatment regimens for patients discovered to have precancerous or cancerous lesions. Furthermore, the device could be used to monitor therapeutic responses, for example drug resistance and/or tumor progression, and indicate when treatment regimens should be changed or stopped, or if surgery is necessary. The present invention is, therefore, not limited to the particular

embodiments described and/or exemplified, but is capable of considerable variation and modification without departure from the scope of the appended claims.

REFERENCES CITED

I. Shingaki, S. et al. Impact of lymph node metastasis on the pattern of failure and survival in oral carcinomas. Am. J. Surg. 2003. 185:278-284. 2. Funk, G.F. et al. Presentation, treatment, and outcome of oral cavity cancer: a national cancer data base report. Head Neck. 2002. 24:165-180.

3. Weinberg, M. A. and Estefan, DJ. Assessing Oral Malignancies. Amer. Fam.

Phys. 2002. 65(7): 1379-1384.

4. Okamoto, M. et al. Prediction of delayed neck metastasis in patients with stage I/II squamous cell carcinoma of the tongue. J. Oral Pathol. Med. 2002.

31:227-233.

5. Ensley, J.F., Gutkind, J.S., Jacobs, J.R. and Lippman, S. M., eds. Head and Neck

Cancer: Emerging Perspectives. Academic Press, 2003.

6. Epstein, J.B., Zhang, L., and Rosin, M. Advances in the diagnosis of oral premaglinant and malignant lesions. J. Can. Dent. Assoc. 2000. 68:617-621.

7. Sobin, L.H. and Wittekind, CH. Head and neck tumors. In: Sobin, LH, Wittekind,

CH, eds. TNM Classification of Malignant Tumors. 5 ^th ed. Berlin, Germany: Springer- Verlag; 1997:17-32.

8. Aps, J.K., Van den Maagdenberg, K., Delanghe, J.R., and Martens, L.C. Flow cytometry as a new method to quantify the cellular content of human saliva and its relation to gingivitis. Clin. Chim Act. 2002. 321:35-41.

9. Greenman, J., Homer, JJ. and Stafford, N.D. Markers in cancer of the larynx and pharynx. Clin. Otolaryngol. 2000. 25(1):9-18.

10. Vielba, R, et al. p53 and cyclin Dl as prognostic factors in squamous cell carcinoma of the larynx. Laryngoscope. 2003. 113:167-172.

II. Alizadeh, A.A. et al. Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature. 2000. 403:503-511. 12. Golub, T, R. et al. Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science. 1999. 286:531-537. 13. Singh, D. et al. Gene expression correlates of clinical prostate cancer behavior. Cancer Cell. 2002. 1:203-209.

14. Beldin, TJ., Singh, B., Smith, R-V., Socci, N.D., Wreesmann, V.B.,

Sanchez-Carbayo, M., Masterson, J., Patel, S., Cordon-Cardo, C, Prystowsky, M.B., and Childs, G. Molecular profiling of tumor progression in head and neck cancer. Arch. Otolaryngol. Head Neck Surg. 2005. 131:10-18.

15. Li, Y., St. John, M.A.R., Zhou, X., Kim, Y., Sinha, U., Jordon, R.C.K., Eisele,

D., Abemayor, E., Elashoff, D., Rark, N.H., and Wong, D.T. Salivary transcriptome diagnostics for oral cancer detection, Clin. Cancer Research 2004. 10:8442-8450.

16. Kuang Y., Biran L, Walt D., R., Simultaneously monitoring gene expression kinetics and genetic noise in single cells by optical well arrays, ANALYTICAL CHEMISTRY 76 (21): 6282-6286 NOV 1 2004.

17. Monk, D. J., Walt D., R., Optical fiber-based biosensors, ANALYTICAL AND BIOANALYTICAL CHEMISTRY 379 (7-8): 931-945 AUG 2004

18. Benjamini, Y. and Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. Roy. Stat. Soc. B. 1995. 57:289-300.

19. Tusher, V.G., Tibshirani, R., and Chu, G. Significance analysis of microarrays applied to the ionizing radiation response. Proc. Natl. Acad. Sci. U.S.A.

2001. 98(18): 10515.

20. Boulesteix, A.L., Tutz, G., and Strimmer, K. A CART-based approach to discover emerging patterns in microarray data. Bioinformatic 2003.

19:2465-2475. 21. Bourguignon, L. Y., Zhu., Shao, L., and Chen, Y.W. Ankyrin-Tiaml interaction promote Racl signaling and metastatic breast tumor cell invasion and migration. J. Cell Biol. 2000. 150:177192. 22. Ziober B. L., et al Type I collagen degradation by invasive oral squamous cell carcinoma. Oral Oncol. 2000. 36:365-72. 23. Natarajan, E., et a Co-expression of pl6(INK4A) and laminin 5 gamma2 by microinvasive and superficial squamous cell carcinomas in vivo and by

migrating wound and senescent keratinocytes in culture. AmJ. Pathol. 2003. 163(2):477-491.

24. Yang, L.C., Ng, D. C, and BiMe, D.D. Role of protein kinase C alpha in calcium induced keratinocyte differentiation: defective regulation in squamous cell carcinoma. J. Cell Physiol. 2003. 195(2):249-259.

25. Hawighorst, T., et al Activation of the tie2 receptor by angiopoietin-1 enhances tumor vessel maturation and impairs squamous cell carcinoma growth. Am. J. Pathol. 2002 160(4): 13811392.

26. Le, Q., Soprano, D. R., and Soprano, KJ. Profiling of retinoid mediated gene expression in synchronized human SCC cells using Atlas human cDNA expression arrays. J. Cell Physiol. 2002. 190(3):345-355.

27. Suzuki, M., et al. Inhibition of tumor invasion by genomic down-regulation of matriptase through suppression of activation of receptor-bound pro-urokinase. J. Biol. Chem. 2004. 279:1489914908. 28. Horiguchi, A., et al. Impact of caveolin-1 expression on clinincopathological parameters in renal cell carcinoma. J. Urol. 2004. 172:718-722. 29. Atula, T., et al Tenasicn-c expression and its prognostic significance in oral and pharyngeal squamous cell carcinoma. Anticancer Research. 2003.

23:3051-3056. 30. Aldo, D. and Tuszynski, G.P. Thrombospondin-1 up-regulates tumor cell invasion through the urokinase plasminogen activator receptor in head and neck cancer cells. J. Surg. Res. 2004. 120:21-26.

31. Dooley, T.P., Reddy, S.P., Wilborn, T.W., and Davis, R.L. Biomarkers of human cutaneous squamous cell carcinoma from tissues and cell lines identified by DNA microarrays and qRTPCR. BBRC. 2003. 11(4): 1026-1036.

32. Chung, C.H., et al. Molecular classification of head and neck squamous cell carcinomas using patterns of gene expression. Cancer Cell. 2004, 5(5):489-500.

33. Cromer, A., et al. Identification of genes associated with tumorigenesis and metastatic potential of hypopharyngeal cancer by microarray analysis.

Oncogene. 2004. 23(14):2482-2498.

34. Uchinda, D., et al. Possible role of stromal-cell-derived factor-I/CXCR4 signaling on lymph node metastasis of oral squamous cell carcinoma. Exp Cell Res. 2003. 290(2):289-302.

35. O'Donnell, RX, et al. Gene expression signature predicts lymphatic metastasis in squamous cell carcinoma of the oral cavity. Oncogene. 2004, Nov 22; [Epub ahead of print].

36. Anderson, T.D., et al Tumor deposition of laminin-5 and the relationship with perineural invasion. Laryngoscope. 2001. 111:21403.

37. Park, N.H., et al. Immortalization of normal human oral keratinocytes with type 16 human papillomavirus. Carcinogenesis. 1991. 12:1627-31.

38. Simon, C, et al. An orthotopic floor-of -mouth cancer model allows quantification of tumor invasion. Laryngoscope. 1994. 108:1686-1691.

39. Armstrong, A. and Eck, S. L. MCAP: A new therapeutic target for an old cancer antigen. Cancer Biol. Ther. 2003. 2:320-326. 40. Schon, M.P., and Orfanos, CE. Transformation of human keratinocytes is characterized by quantitative and qualitive alterations of the T- 16 antigen (Trop-2, Mov-16). Int. J. Cancer 60:88-92.

41. Quak, JJ., Van Dongen, G., Brakkee, J.G., Hayashida, DJ., Balm, AJ., Snow,

G,B., and Meijer, CJ. Production of a monoclonal antibody (K 932) to squamous cell carcinoma associated antigen identified as the 17-1A antigen.

Hybridoma 1990. 9:377-387.

42. Gronau, S.S., Schmitt, M., Thess, B., Reinhardt, P., Wiesneth, M., Schmitt, A., and Riechelmann, H. Trifunctional bi specific anti body-induced tumor cell lysis of squamous cell carcinomas of the upper aerodigestive tract. Head Neck 2005. 27:376-82.

43. Malamud D., Bau H., Niedbala S., and Corstjens, P. Point Detection of Pathogens in Oral Samples, Adv. in Dental Res. 18: 12-16 (2005).

44. Corstjens, PLAM, Li, S, Zuiderwijk, M, Kardos, K, Abrams, WR, Niedbala, RS, and Tanke, H. J. Infrared up-converting phosphors for bioassays. IEE Proc. -Nano biotech nol. 152:64-72 (2005).

45. Carol Holm-Hansen, Gary Tong, Cheryl Davis, William R. Abrams, and

Daniel Malamud. Comparison of Oral Fluid Collectors for Use in a Rapid

Point-of-Care Diagnostic Device. Clinical and Diagnostic Laboratory

Immunology 11:909-912 (2004). 46. Qian, S. and Bau, H., H., 2003, A Mathematical Model of Lateral Flow

Bio-Reactions Applied to Sandwich Assays, Analytical Biochemist , 322,

89-98. 47. Qian S., and Bau, H., H., 2004, Analysis of Lateral Flow Bio-detectors:

Competitive Format, Analytical Biochemist 326, 211-224. 48. Qian, S. Burger, R., and Bau, H., H., 2005, Analysis of Sedimentation Biodetectors, Chemical

Engineering Science, 60, 2585 - 2598.

49. Yi, M., and Bau, H., H., The Kinematics of Bend-Induced Mixing in

Micro-Conduits, International Journal of Heat and Fluid Flow, 2003, 24, 645-656,

50. Ottino, J. M., The Kinematics of Mixing: Stretching, Chaos, and Transport, Cambridge 1989.

51. Suzuki, H., Kasagi, N., Ho, CM., Chaotic Mixing of Magnetic Beads in Micro

Cell Separator, Proc. 3rd Symp. Turbulence and Shear Flow Phenomena, pp 817-822, Sendai, Japan, June 14-27, 2003.

52. McPherson, T., Kidane, A., Szleifer, I, and Park K., Prevention of Protein

Adsorption by Tethered Poly(ethylene oxide) Layers: Experiments and Single-Chain Mean-Field Analysis, Langmuir, 14, 176-186 (1998).

53. Jones, T. B., Electromechanics of Particles, Cambridge, 1995. 54. Inglis, D. W., Riehn, R., Austin, R., H., and Sturm, J., C, Continuous Microfluidic Immunomagnetic Cell Separation, Applied Physics Letters, 85 (21) 5093-5095 (2004).

55. Gijs, M. A. M., Magnetic Bead handling On-Chip: New Opportunities for Analytical Applications, Microfluid Nanofluid 1 , 22-40, (2004).

56. Giaever, I., Keese, CR. Micromotion of Mammalian Cells Measured

Electrically, PNAS 88: 7896-7900 (1991)

57. Xiao et al. [2002] Analytical Chemistry 74 5748-5753.

58. Wang, J., Chen, Z., Mauk, M., Hong, K., Li, M., Yang, S., and Bau, H. H., SeIf- Activated, Thermo-responsive Hydrogel Valves for Lab in a Chip, Biomedical

Microdevices, 7(4): 313-322 (2005)

59. Baibich, M., N., Broto, J., M., Fert, A., Vandau, F., N., Petroff, F., Eitenne, P.,

Greuzet, G., Friederich, A., Chazelas, J., Giant Magntoresistance of (001)Fe/(001), Physical review Letters 61 (21), 2472-2475 (1988). 60. Rife, J. C, Miller, M., M., Sheehan, P., E., Tamanaha, C, R., Tondra, M., and Whitman, L., J., Design and Performance of GIVIR Sensors for the Detection of Magnetic Microbeads in Biosensors, Senosrs and Actuators A-Phys. 107 (3) 209-218 (2003).

61. Graham D., L., Ferreira, H., A., Freitas, P., P., Cabral, J., M., S., High Sensitivity Detection of Molecular Recognition Using Magnetically Labelled

Biomolecules and Magnetores si stive Sensors, Biosens Bioelectron 18(4), 483-488 (2003).

62. Jordan, R. C, et al. Overexpression of matrix metal loproteinase-1 and -9 mRNA is associated with progression of oral dysplasia to cancer. Clin Cancer Res. 2004. (19):6460-5.

63. Livak, KJ. and Schmittgen, T.D. Analysis of relative gene expression data using real-time quantitative PCR and the 2(-Delta Delta C(T)) Method. Methods. 2001. 25:402-8.

64. Corstjens, PLAM, et al. Anal. Biochem. 2003. 312:/91-200 Wang, J., Chen, Z., Qian S., and Bau, H. H., 2005, Thermally- Actuated, Phase-Change Flow

Control for Microfluidic Systems, Lab on Chip. 5, 1277 - 1285.

65. Wang, J., Chen, Z., Mauk, M., Hong, K-S, Li, M., Yang, S., and Bau, H., H.,

2005, Self-Actuated, Thermo-Responsive Hydrogel Valves for Lab on a Chip, Biomedical Microdevices 7:4

66. Wang, J., Chen, Z., Corstjens, P. L. A. M., Mauk, M., and Bau, H. H., 2005, A

Disposable Microfluidic Cassette for DNA Amplification and Detection, accepted for publication in Lab on a Chip, DOI: 10.1039/b511494b

67. Nam, J.M., Park, SJ. and Mirkin, CA. , "Biobarcodes based on oligonucleotide-modified nanoparticles", J. American Chemical Society,

124(15): 3820-3821 (2002)

68. Muller, V.R., "Protein detection using biobarcodes", MoI. Biosystems, 2(10): 470-476 (2006)