Optical biosensing: future possibilities

It is difficult to pinpoint the precise time at which optical biosensors came of age, but it is clear that the ‘omics’ era has been instrumental in the recent need, evolution and significant expansion of this area of technology. Research involving genomics, proteomics and metabolomics is reliant upon the evaluation of large numbers of different bioanalytes and the relative quantification of these within a set time frame, preferably in real time. Although some of the sensing and analysis approaches used for current ‘omics’ studies are not optical ones, optical biosensing and analysis methods have and will continue to contribute significantly to providing ways for bioscience and biomedical science questions to be answered. One can categorically state that research in optical engineering topics has not been driven by the biosciences – the major engineering advances have been made for far more commercially lucrative reasons, for instance telecommunications applications as exemplified by optical fibers, laser sources and integrated optical waveguides. However, these small-sized structures, which have been developed for massively parallel processing of optical signals, are beginning to provide some of the optical biosensors, devices and methodologies for the future. Most notably for the detection and quantification of small-sized bioanalytical samples – all in high throughput and in formats suitable for fast data collection.

It is unclear, to me at least, when the first optical biosensor device was invented. However, perhaps one of the first important optical biosensors was the oximeter for the evaluation of blood oxygenation. Indeed, optical oximetry is perhaps the most commonly used optical sensor technology in use. Such a system is based upon the differences in the light absorption spectra of hemoglobin and hemoglobin–oxygen complex – the difference between arterial and venous blood. Two sensors are used – one at approximately 600 nm (red) and the other in the near-infrared (805–1000 nm). The number of photons of 805 nm light absorbed per molecule of hemoglobin or oxygenated hemoglobin is the same, and the sensor at this wavelength is used as the reference one (providing a method to compensate for the variation of the blood volume). Thus, relative to this reference, the absorption of the red light (600 nm) provides a method to directly measure the level of oxygenation of the blood. This methodology has been effectively used and optimized, and pulsed oximetry is now one of the most reliable methods for noninvasive blood oxygenation sensing. The basis of the pulsed method is to measure the change in light transmitted through the skin that occurs as a result of the arterial pulsation. The light transmitted through the skin at the inflow of the cardiac cycle is arterial blood. Most pulse oximeters are of the transmission type, where forward-scattered light through the fingertip or earlobe is analyzed.

How the construction of oximeter sensors has evolved over the last decade is a reflection of the advancement in light sources and detectors from the optoelectronics and microelectronics research fields. These fields have delivered miniaturized glass and silicon structures for new light sources and devices, as well as detectors and the associated electronics. It is now clear that by evolving optical biosensing systems to be compatible with the telecommunications technologies, mainly bluetooth, mobile phones and the internet, will have obvious advantages for the future, not only for laboratory-based research, but for the care of patients in the home and local environment.

Although the physical operating parameter of oximeters is based upon optical absorption of light, this is not the physical parameter that is most commonly applied in most optical biosensor systems. The reason for this is that the volume of most bioanalytes is generally very small and the solutions or tissues often scatter light. However, more importantly, most biomolecules (of mammalian origin) absorb light in the same spectral range – in the UV wavelengths up to approximately 350 nm, thus, there is limited potential to discriminate particular components from the bulk mixture. In considering optical measurements of biological systems one always has to consider the transmission spectrum of water. Generally speaking, this limits most optical biosensing methods to studies using light of between approximately 200 and 1000 nm in wavelength; although there is a window in the terahertz range that has been effectively exploited for imaging applications [1], it will not be detailed further here.

Currently, methods based upon fluorescence are the most commonly used for biosensor applications. The majority of biomolecules, tissues and structures are only weakly fluorescent when exposed to light and this means that specific biomolecules can be ‘tagged’ with artificial fluorophores. These fluorophores can be synthetic molecules, such as the wide range of cyanine dyes, or a fluorescent protein, such as the green fluorescent protein. The relatively low background fluorescence of the biological system provides an ideal contrast for the low numbers or even single copies of fluorescently tagged biomolecules. The recent advancements in this field are a result of a number of factors, primarily the development of commercially available fluorescent molecules. Largely, this is thanks to an Oregon-based company, formerly known as Molecular Probes, as well as the large numbers of research chemistry groups and companies, for instance Pierce, providing approaches for the bioconjugation of these fluorescent molecules to biological molecules, most importantly within DNA synthesizer. A second important factor is the availability of very sensitive photodetectors and charge-coupled device cameras, as well as laser sources.

Optical biosensing is an extremely broad term, so it is perhaps best if this is now classified into two primary configurations that fluorescence-based optical biosensing is applied, either in the solution phase where the biomolecules have been partially purified or directly from fluid samples (blood, serum, urine, tear fluid, saliva), or directly within the cells or tissue structures. Both require slightly different approaches.

Presently, quantification of biomolecules must be carried out by utilizing solution phase analytes, which can be enzymatically digested cells or tissues, or other untreated body fluids. These biosensor systems are usually configured in parallel for a number of different analytes. Selection of particular analytes is provided by specific recognition and binding by fluorescently tagged natural or artificial molecules. Such biomolecules include single DNA strands that are complementary to a particular DNA sequence, forming a duplex – the double helix or, alternatively, the biomolecule may be an antibody raised towards a particular antigen (the bioanalyte, often a protein). The most important factor is the specificity of the ‘recognition biomolecule’ for the bioanalyte, and this is sometimes where the biosensing approach for some analytes can be poor. The method for detecting and quantifying different analytes can be achieved in a number of ways, which include:

•

Pretreating the analyte solution with the fluorescently labeled recognition molecule and then passing the mixture over the biosensor array. Any fluorescently labeled recognition molecule without a bioanalyte in its recognition site will bind to the specific sensor surface where a synthetic bioanalyte molecule is covalently attached. The highest number of fluorophore molecules will thus be present on sensor patches where there is the smallest number of corresponding bioanalyte molecules in the sample under investigation;

•

Configuring the assay such that a comparison is made between a ‘normal’ (perhaps healthy sample) and a sample under test. Two different fluorophores are used, thus, the difference in the fluorescence yield of the two fluorophores on the sensor (versus a calibrant) to which the recognition molecule is attached provides information as to whether a particular biomolecule is expressed in high levels or downregulated under the particular test conditions.

Microarray methods have provided the possibility of massive parallel analysis of small quantities of biomolecules. These systems were first reported in the DNA microarray format by Brown and his team at Stanford University [2]. These microarrays consisted of micron-sized spots of different DNA sequences printed or spotted in an array pattern onto a simple glass slide. An unknown mixture of fluorescently labeled DNA sequences is hybridized – such that the different spots are fluorescent if the sequence is present. Although DNA microarrays can be used for DNA sequencing, one of the most valuable applications of microarrays is for gene expression analysis. Gene expression analysis is an approach used to establish which genes are switched on or off at a particular time point (and the level) in a cellular or tissue sample. The mRNA molecules (synthesized from the gene sequence in the chromosomal DNA), isolated from cells, are used as templates for the synthesis of fluorescently labeled DNA probes for the assay. Thus, if a fluorescently labeled strand of DNA of a particular sequence is detected on the microarray spot – then the mRNA is present (and the gene is being expressed) in the sample under investigation. Since so many of the genes are expressed, it is necessary to make a comparison of them all; thus, the assay is performed with a ‘normal or reference’ sample. Two samples are tested concurrently, each with different fluorophores, and the yield of the fluorescence from each of the probes is measured and calibrated against a reference.

Microarrays have been instrumental in advancing the field of genomics, providing a methodology used routinely to provide a reliable method for gene expression analysis. While the early examples of microarrays consisted of simple glass microscope slides, a number of attempts have been made to improve this optical technology by introducing structures within the glass surfaces. The cost makes incorporation of such improvements unrealistic and indeed unnecessary for most applications.

The long-term future of the conventional microarray is unlikely, because the format of microarrays is not optimal for a number of reasons. The most significant problem is that a large surface area is covered by the spots of recognition molecules and a small volume of bioanalyte is dispensed over this area. Thus, diffusion of the analyte molecules to the various spots presents a significant problem. This has been remedied in part by the application of mechanical agitation of the system, that is, ultrasound. The fabrication and data analysis of microarrays is done by highly standardized protocols and it would be fair to say that much of the information on many thousands of the spots is redundant and, thus, wasteful. It is clear that in the longer term, the microarray platform will be superseded – but with exactly what remains unclear, possibly a bead or barcode system. However, a higher understanding of the underpinning science will mean that smaller numbers of bioanalytes will need to be probed for future pure research and other potential diagnostic applications.

The future of array technologies, particularly for proteins, lies in approaches where higher detection limits and ranges are possible. At present, integrated optical sensors provide some of the most sensitive devices for fluorescence-based biomolecular sensing [3]. For fluorescent molecules these have a number of advantages, most notably specific detection at the surface. The advantage of these systems is that the fluorescence from the bound fluorescently tagged antibody is rejected from the bulk as only the molecules in the evanescent field of the waveguide are detected. Thus, only fluorophores attached to recognition molecules associated with molecules on the sensor surface are detected. Integration of these waveguides with cheap laser sources and CCD cameras or fiber/photodiode configurations enables such systems to have potential to evolve from a research technology to the marketplace.

One of the major areas of research impacting upon the optical biosensor field is ‘lab-on-a-chip’ technology. Lab-on-a-chip structures are designed to provide facile processing methods for small volumes of analytes. Often included in the structure are microfluidic channels, reactor chambers and biosensors. These microstructured devices are a spin-off of the silicon fabrication approaches of the microelectronics field. In the early stages of development, the majority of the lab-on-a-chip technology was fabricated in silicon, these are being superseded with plastic structures. Plastic fluidic channels are commonly made by using silicon moulds and subsequent plastic hot embossing methods, plastic lamination methods or simply from moulds built up from photolithographic resists. Currently, most optical biosensors are fabricated separately from the fluidic structures used to ‘handle’ the bioanalyte. Better methods for integration are needed in the future, not just of the fluidic elements, but of all the components involved in the optical sensing. However, there is a wide application for this technology, which includes drug screening, environmental sensing (including for defense/bioterrorism) and remote sensing. Although, currently, there is much reported literature [4,101], the full potential of lab-on-a-chip technology has yet to be demonstrated.

Fluorescence-based optical biosensing, in cellular or tissue systems, is a routinely used microscopy-based technique. Cellular constituents can be fluorescently ‘stained’ with fluorescent tagged molecules that bind to specific biomolecules or classes of biomolecules, thus providing information about their localization. Alternatively, cells can be transfected (i.e., microinjected) with a DNA sequence that codes for the protein of interest linked to a fluorescent protein. Upon expression the fluorescently tagged protein will be localized or transported within the cell or tissue. Microscopy is conventionally used for this optical biosensing method, however, the field has advanced as a result of new laser technology – particularly by applying infrared femtosecond lasers. These lasers are now routinely incorporated with microscopes. The application of focused multiphoton beams to cells provides the potential for submicron-level excitation of fluorescent probes within a cell, thus providing high resolution imaging by scanning. Light in the near-infrared range is used, thus excitation is achieved only where the photon density is high enough for fluorescent molecules to be excited by two photons or more. A significant advantage of the multiphoton approach is that fluorophores outside the focal region do not become photodamaged.

The major problem with fluorescence-based optical biosensing is the complexity of the method. For instance, fluorescent molecules must be attached to recognition molecules, are prone to photobleaching and, when attached to biomolecules within cellular environments, can interfere in the cellular processes. It is clear that fluorescence-based biosensors offer significant advantages, but it is likely that these methods will not lead but complement others in the foreseeable future.

Providing optical biosensing strategies that do not need fluorescent probes has been an area of research that has been on-going for many years. Perhaps the most well-known biosensor system for the quantification of bioanalytes and for the measurement of biomolecular recognition processes, is the Biacore™ system [5]. These sensors are based upon surface plasmon resonance and, in this case, a gold layer is used. Biomolecules are first covalently attached to the gold surface; light is used to excite surface plasmons in the metal in a resonant manner, subsequent binding of bioanalytes to the covalently attached biomolecules results in a change in thickness of the layer and also the surface plasmon resonance frequency. There are, of course, a number of other optical sensor systems that function by refractive index changes; these include Mach–Zehnder Interferometers, which are integrated optical devices.

However, perhaps some of the most innovative nonfluorescence-based optical sensors are the holographic sensors of Professor Chris Lowe at Cambridge University (UK) [6] and the photonic crystal sensors of Professor Sanford Asher at Pittsburgh University (PA, USA) [7]. Both systems are based upon hydrogel polymers that swell or contract upon the presence of the analyte. The analytes are typically metabolites or nutrients, for instance lactate or glucose. The holographic sensors of Lowe et al. consist of periodic silver layers that are fabricated within the hydrogel polymer. Upon illumination with white light a narrow band of some wavelengths is reflected. The wavelengths reflected depend upon the separation of the silver layers – the amount the polymer has swollen and thus the extent of analyte in the polymer. This technology has been developed for incorporation into contact lenses for diabetics – the hologram yields a black spot over the pupil when the glucose concentration was lowered.

Sensing technologies on the nanoscale are starting to provide highly sensitive and multiplexed analysis methods. The optical properties of nanostructured, metals and semiconductor materials, are defined by their small size and are often different to those seen for the bulk. Colloidal semiconductor quantum dots have excellent properties for fluorescence-based biosensing, having narrow emission bands with a wavelength dependent upon the size of the nanocolloid and they tend to be photostable [8]. The attachment of these colloids to either protein or DNA molecules is possible, and application for biosensors valuable. However, application for cell biology is questionable, particularly as these materials are potentially toxic. Metal nanoparticles, such as gold, provide a very different approach for optical biosensing. Enhancement of Raman spectral signals are observed for molecules adjacent to the gold surface [9]. Incorporation of nanostructured gold on the surface of planar devices also provides an alternative approach for creating enhanced Raman spectral signals for analytes deposited on the surface. Many of these probes can be utilized within cells, however, it is evident that the introduction of nanoparticles covalently attached to recognition molecules will not provide a valuable approach for real-time imaging. The systems are consumptive – removing the biomolecules to which they bind from the system and thus, likely to perturb the cellular biochemistry.

It is clear that the future in optical biosensing is being driven by optical approaches for detection in situ without the need for fluorophores, nanoparticles or other tags. One of the potential methods of the future will be the application of coherent antistokes Raman spectroscopy (CARS) [101]. With the recent developments in laser technology, it has become possible to utilize femtosecond laser sources of significant value for optical biosensing inside cells. The CARS method provides excellent Raman signals directly from cellular systems (CARS should not be confused with conventional Raman microscopy.) This multiphoton method provides an approach that enables the direct chemical analysis of the molecular composites of cells. With significant further development it could be possible that a detailed knowledge of biochemical processes will be measured in real time within cellular systems.

Our understanding of biological systems has advanced significantly over the last 10 years as a result of new biomolecular, optical and electronic technologies becoming available. Bioscience research has evolved very quickly from a molecular approach to a systems approach. With the ability to obtain large datasets and process these using computational tools, information on various concurrent biochemical processes that occurr within a cell, tissue or organism is beginning to be understood.

Approaches to detect and quantify, in real time, some of these biomolecules involved in the complex network are now required. An increased understanding of these systems will provide better therapeutic approaches, but also better diagnostic and monitoring approaches for biomedicine. These will be achieved by integration of optoelectronic, microelectronic and nanoscience methods with intelligent electronic and computational tools for drug delivery and monitoring.

Financial & competing interests disclosure

Tracy Melvin is a reader in the Optoelectronics Research Centre, and teaches in the School of Electronics and Computer Science at the University of Southampton. Her research interests are in the area of optical biosensing. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.