Rapid advancements in the areas of genomics, proteomics, biomedical, and chemical analysis have increased the need for efficient analytical tools. Arrays of optical chemical sensors based on bundles of fused optical fibers can measure parallel optical signals from fluorescent molecules at the face of the bundle with high spatial sensitivity, providing the ability to sense a number of gaseous chemicals, as well as fluorescent tagged DNA sequences and entire cells. These sensor arrays may contain thousands of individual sensing regions on the end of a fiber bundle with a diameter of only a few hundred microns. In addition to their small size, these sensors can also offer the ability to be multiplexed and used for remote monitoring. These functionalized optical-fiber sensors have emerged as alternatives to other conventional methods of analysis. etched wells
Figure 1. Tiny wells etched into the cores of fibers in a fused fiber bundle are filled with beads, which are sized so that one bead fits in a well. Fluorescent dyes attached to each bead produce optical emission that changes when a specific chemical is sensed. The optical emission passes through the fiber under the well to the detection system. (TUFTS UNIVERSITY)
Coherent imaging fiber bundles, or imaging fibers, are fabricated by fusing together the cladding of thousands of optical fibers so that the individual fibers retain their position throughout the fiber length. We have made high-density sensing arrays containing thousands of micrometer- and nanometer-sized sensors by etching wells into the individual fiber cores at the face in the imaging fiber. The wells can be filled with complementary-sized microspheres (beads) such that one bead is incorporated into each microwell (see figure 1). Using fluorescence techniques, we then optically interrogate the sensory materials placed in the wells. This scheme allows us to make sensors with spatially discrete sensing sites for multianalyte determinations.
Figure 2. A modified epifluorescence microscope uses a xenon arc lamp to excite fluorescence from the material at the face of the imaging fiber. The resultant fluorescence signals return through the individual fibers and are magnified and filtered before arriving at a CCD camera.
In our system, a modified epifluorescence microscope collects the fluorescence signals (see figure 2). The excitation source, a xenon arc lamp, provides a white light beam that is sent through a bandpass filter and reflected into a microscope objective through a dichroic filter. The objective focuses light onto the proximal end of the imaging fiber, overfilling its endface to simultaneously illuminate all the individual fibers and propagate out to the wells on the ends.
Fluorescence emitted from the sensing indicators on the beads returns through the individual fibers to the microscope objective, where it is magnified before passing through the filters on its way to the charge-coupled-device (CCD) camera. Because the CCD chip contains many more pixels than there are fibers in the array, the system provides ample resolution to look at each bead/fiber individually. Filter wheels allow the microscope to change the exitation and emission filters dozens of times per second. Image-processing software analyzes the images.
Figure 3. Microspheres designed to sense different chemicals randomly fill the wells in the imaging fiber face. Each type of microsphere fluoresces with an identifiable spectrum. Because each well is individually addressable, the positions of each sensor type can be registered.
A combination of beads, each with a specific chemical receptor, can be used to fill the wells in a random fashion (see figure 3). Each type of bead is treated with a unique combination of fluorescent dyes that acts as an optical barcode; the fluorescence spectrum identifies the bead at each specific location in the array. Consequently, the random assembly process must be followed by a positional registration step in which the location of the beads is determined by their optical signature.
An important consequence of the beads-in-wells process is that multiples of each bead type will be present in every array. This inherent redundancy provides two significant advantages: First, because repeat sensors must agree, their signals can be used to virtually eliminate both false positives and false negatives. Second, signal-to-noise ratios scale as the square root of the number of identical sensing elements, so sensitivity is enhanced by summing the signals from all the beads of a particular type in the array. Summing large numbers of sensors improves the precision of the measurements, thereby enabling the detection of lower analyte concentrations. the nose knows
The nose is an enormously sophisticated sensing system. Based on principles derived from the olfactory system, we have created optical sensors. We demonstrated a vapor-sensing system using this approach.
Figure 4. Pattern-recognition computational networks use video images of the spectral and temporal response from an array of multiple fluorescent sensors in the imaging fiber. The black bar (at top) indicates the duration of a 1.8-s exposure to a number of 50% saturated volatile organic compound vapors.
We patterned the design of the multianalyte sensor array after the mammalian olfactory system such that complex, time-dependent signals from multiple sensors provide a "fingerprint" of each analyte. Bead-immobilized dye molecules on the distal end of the fiber generate distinct fluorescent response patterns upon exposure to organic vapors. Each of the different beads incorporates solvatochromic dyes (fluorescent dyes sensitive to their environment) either through direct adsorption onto the substrate surface or by solvent swelling. The polymer matrices of the beads vary in polarity, hydrophobicity, porosity, elasticity, and swelling tendency, creating unique sensing regions that interact differently with vapor molecules. The intrinsic chemical and physical nature of the various bead types, in conjunction with the solvatochromatic dye, gives rise to unique responses to vapors. These different response patterns include spectral shifts and intensity changes resulting in temporal responses (see figure 4) that are influenced by the physical and chemical nature (polarity, shape, and size) of the vapor and the polymer.
Video images of the different temporal responses are captured as the input signals for a computational network vapor-recognition program. With this data, we "trained" the computational network to solve vapor-recognition problems. The system can recognize a wide variety of volatile organic compounds and mixtures of compounds and can also identify complex materials including various beverages, foods, perfumes, and even explosives. This "artificial nose" maintains the advantages of fiber optics: It is compact and can provide remote sensing. Several other approaches are under investigation for detecting analytes in aqueous samples.
The advantage of using cross-reactive arrays for sensing is that they possess a broadband capability; their sensitivity does not rely on a particular ligand-receptor interaction but is a function of a response pattern obtained over the entire array. Furthermore, computational algorithms can be employed to detect the presence of particular analytes even in dynamic, complex background environments. We have demonstrated our ability to detect explosive vapors at the low-part-per-billion level using the signal summing approaches described above. analyzing DNA
DNA sequence identification is another application area for imaging-fiber sensor arrays. There is tremendous interest in exploiting the genetic information obtained from sequencing the human genome. DNA chips offer a high-density format for conducting many experiments simultaneously. Imaging fibers, however, offer an alternative platform that yields arrays with unprecedented densities.
As with the chemical sensors mentioned above, multiple types of sensors are mixed together and randomly distributed on the etched face of an optical imaging fiber to create a randomly ordered, addressable, high-density bead array. In this case, however, each optically encoded 3-µm-diameter bead is attached to a single-stranded oligonucleotide probe.
One such study incorporated 25 different oligonucleotide probes and was employed to detect multiple DNA sequences in parallel by monitoring hybridization (i.e., the attachment of complementary DNA to the probes) of fluorescently labeled target oligonucleotides. The 25 sequences were from disease states (for example, lymphocyte and cytokine expression and genes related to diseases such as oncogenesis or cystic fibrosis). When the fluorescently labeled target DNA (the sample) hybridizes to its complementary strand immobilized onto the surface of the bead, we observe an enhancement in fluorescence. Hybridization occurs within seconds, and nonspecific binding does not contribute to the fluorescence signal.
Due to the randomized nature of the sensor platform, new sensors can be incorporated easily into the array as they are developed. The array takes minutes to fabricate, and some arrays can be stored for many months. The burden of producing identical sensing elements is minimized because billions of microbead sensors are fabricated at one time (for example, 1 mL of a bead suspension contains about 6 * 109 beads with identical chemistries). In addition, sensor stocks can be reproducibly fabricated from one batch to the next; sensor shelf life has been shown to be at least 10 months. The detection limit for specific DNA sequences has been demonstrated to be in the femtomolar range, which translates to the ability to detect a few hundred DNA molecules.
The arrays can also be used for other types of biomedical analysis, for example, living cells. The diameter of the wells can range from 3 µm to 25 µm, which allows the array to be tailored to accommodate different types of cells. Each well, containing a single living cell, can be used to monitor several physiological and genetic responses simultaneously.
Fiber arrays with bead-based fluorescent sensors provide a powerful technique for a wide range of analysis. We are investigating the limits of our ability to create high-density sensing arrays containing thousands to potentially millions of nanosensors. Nanometer-sized wells have been fabricated by etching the cores of tapered optical imaging fibers. Well arrays can be prepared with individual wells as small as 200 nm. These high-density sensor arrays, in which each sensor is connected to its own optical channel, offer the potential for highly multiplexed analytical systems with the ability to address a myriad of application areas. oe
David R. Walt
David R. Walt is professor of chemistry at Tufts University, Medford, MA.