Opening windows on new biology and disease mechanisms: development of real-time in vivo sensors

From X-ray radiology to blood panels, medicine is informed by the status of in vivo systems and their deviations from normal ranges. It is now widely acknowledged that the fundamental aetiology of many disease states can be understood at the molecular level. Over the past half a century there has been a push towards the development of novel technologies to probe the in vivo molecular status of biological systems for diagnosis, prognosis and response to therapies [1]. This Theme Issue will highlight some of the most recent exciting work in the fields of molecular-, nano-, and micro-scale devices for real-time in vivo sensing. Although this field is already large and growing, this Introduction will attempt to provide context and background for understanding this set of papers, and how they attempt to address the unique challenges that arise in the development of these novel and powerful devices.

At the most fundamental level, a sensor is simply a device capable of quantifying or detecting a specific, biologically relevant analyte. Practically, this consists of a detector (such as an electrode, antibody, aptamer, peptide or enzyme) paired with a detection method (which may be optical, electrochemical or magnetic, among others). In this Theme Issue, we will focus on biosensors capable of interrogating the status of biological systems in vivo and at high temporal resolution. These in vivo sensors promise to have great utility in the future, as they will allow continuous, rapid monitoring of biological systems in the context of disease, response to therapy, cell–drug interactions or understanding normal biology.

The first real-time in vivo sensors were electrochemical glucose detectors that took advantage of the catalytic activity of glucose oxidase to reliably and simply detect glucose levels over time [2]. These electrochemical detection modalities have since been improved upon and incorporated into devices capable of detecting numerous analytes, including xanthine and lactate [3]. Optical sensors have recently become increasingly popular owing to their high potential for multiplexing and ease of adaptation from existing in vitro assays. For high spatial resolution, these detectors require complex imaging modalities, such as two-photon confocal imaging, for reliable temporal and spatial resolution. Feasible use of such sensors in human patients will probably sacrifice spatial resolution for broad in vivo accuracy, as in embedded optical glucose sensors [4]. Positron emission tomography sensors are well established, with a growing list of increasingly validated substrates for in vivo sensing including glucose metabolism, oxygenation, neurotransmitters and proteases. Detection methods commonly used with in vitro sensors, such as those that take advantage of surface plasmon resonance or the piezoelectric effect have not yet found wide utility with in vivo sensors. Adaptation of these techniques for in vivo sensing will surely expand the breadth of in vivo sensors in the future. Yasun et al. [5] specifically review recent advances in the field of gold nanorods with exciting potential for in vivo applications, including the development of photoacoustic imaging techniques for cancer.

Although the range of in vitro biosensors is enormous, encompassing a huge variety of detection modalities with exquisite sensitivity, the range of in vivo sensors is significantly smaller owing to the unique challenges associated with in vivo sensing [1,6,7]. A particular concern is biocompatibility: any sensor must not only be biologically inert, but also not perturb the normal biology the sensor is interrogating. A variety of techniques have been adopted to address this problem, from incorporating polyethylene glycol linkers to embedding the device in a hydrogel [8]. The optimum solution for biocompatibility depends on the demands of the system and its composition. Continuously maintaining calibration is another difficulty: re-calibration of in vivo sensors is difficult, if not impossible in many cases. Ratiometric techniques, such as Föster resonance energy transfer (FRET)-based sensors, have major advantages in calibration and stability [9]. For some detector modalities, such as aptamer- and antibody-based sensors, reversibility of analyte binding is an inherent demand of the system that must be addressed in any type of sensor, requiring appropriate association and dissociation constants. Multiplexing is perhaps the most difficult problem in biosensing; all techniques are limited to a range of channels based on the detection modality. For increased multiplexing, combining detection techniques will probably be required.

Although the division of biosensors into discrete categories is always somewhat arbitrary, sensors are commonly classified by size. Molecular-scale devices are those in which the presence or absence of a substance acts on a molecule (such as a peptide, enzyme or DNA sequence) to induce a signal that is directly detected [1,10]. Nanoscale devices incorporate components of the order of 1–100 nanometres in size, such as graphene particles, nanotubes or nanoparticles, to detect, amplify or transduce information about a biological system to a detector [6]. Finally, micro-scale devices include a diverse range of devices ranging from embedded electrode-based sensors to implantable microfluidic devices. It is important to note that these nano- and micro-scale devices may still use molecular-level detectors for assessing an analyte; the microdevice in this case may act as an integrated signal transducer or amplifier [1,6,11].

The discovery and optimization of green fluorescent protein (GFP) revolutionized in vivo imaging research. In addition to well-characterized roles in detection of protein localization, dynamics, interactions and cell tracking, GFP and its derivatives have found wide use in a variety of sensing applications. The simplest GFP-based sensors take advantage of GFP's intrinsic response to analytes; the fluorescence of GFP is highly correlated to both the pH and halide concentration of its surroundings [12,13]. This technique may be improved upon and generalized by inclusion of binding domains in GFP for specific proteins. When coupled with bioluminescence, this has been elegantly used to visualize and quantify calcium signalling [14]. More complex sensors use multiple fluorescent moieties that act as FRET pairs. In this case, the presence of an analyte may lead to activation or quenching of FRET-based fluorescence; this technique, also used with aptamer-based sensors, is also referred to under the umbrella term of ‘switch-based’ biosensors [15]. This technique has currently been successfully used to investigate maltose and calcium ion concentrations, among others [16,17]. The ratiometric nature of FRET allows highly quantitative measurements to be made [15]. In addition, reporters specific to cancer-associated proteases such as matrix metalloproteinase 2 (MMP2) and MMP9 have been developed with potential applications in diagnosis, staging and surgery [18]. A major advantage of these reporter systems is that fluorescent protein-based biosensors can be genetically encoded in reporter mice, such as the cyclic adenosine 3′-5′-monophosphate (cyclic AMP) reporter mouse [19]. These techniques promise to greatly expand our understanding of normal and disease-associated processes in the future.

Despite the enormous potential of fluorescent protein-based technologies and their inherent simplicity, they are still fundamentally limited by the fact that detection must be made optically. This leads to real limits on the spatial and temporal resolution of these probes as well as the tissue depth obtainable with current technologies [7]. The recent development of fluorescent proteins with excitation and emission wavelengths approaching the infrared promises to expand the utility of these probes for in vivo use, owing to the enhanced tissue permeability of long-wavelength light [20]. In this Theme Issue, Progatzky et al. [21] provide a detailed summary of commonly used fluorescent proteins, molecular dyes and reporters for in vivo applications, with a particular emphasis on the advantages and caveats of each system. Development of improved detection devices and techniques to facilitate in vivo use of these reporters will be necessary for widespread adoption of these sensors for clinical use, however. Recent refinements of implantable optical detectors are a particularly promising development that will have major implications for the use of fluorescence-based biosensors [22]. These implantable optical sensors have recently been applied to great effect to subcutaneously assay glucose concentrations using a fluorescent hydrogel-based glucose sensor [4].

Just as we have harnessed fluorescent proteins to develop powerful in vivo sensors, recent work has begun to take advantage of the unique bioluminescent properties of luciferase. Commonly used for real-time quantification of transcription as well as imaging of transplanted cell fate in vivo, recent innovations have engineered these enzymes to act as bona fide real-time in vivo sensors [18]. In vivo reporters of Akt activity have been engineered by modifying luciferase with an Akt consensus substrate peptide [23]; similarly robust reporters of caspase-3 activity that incorporate a caspase-3 substrate peptide have been used to report apoptosis in tumours of living animals [24]. The identification and optimization of luciferases that use substrates other than luciferin has expanded their use and allowed true multiplex detection of multiple cell types in vivo [25]; it is easy to imagine how these alternative luciferases could be modified for multiplex sensor development.

Emerging in vivo sensor modalities include aptamers and related functional nucleotides sequences that have enormous promise in the creation of optical biosensors as well. With binding affinities and avidities similar to antibodies, aptamers are highly specific for their target proteins or molecule [26]. By taking advantage of the fact that aptamer binding to target molecules often induces conformational changes in the DNA sequence, several different techniques have been developed for reagent-less fluorescent biosensors. These include ATP and thrombin sensors that become fluorescent in the presence of their respective ligand with potential utility in multiple disease states [27–29]. In 2011, we demonstrated that a platelet-derived growth factor (PDGF)-responsive biosensor can be conjugated to mesenchymal stem cells for potential real-time in vivo imaging of PDGF ligand concentrations [30]. Despite the promise of these applications, the hurdles for application of aptamer-based biosensors are similar to those facing fluorescent protein-based sensors: real-time, high-resolution optical imaging is a difficult undertaking with fundamental current limitations. Modification of these aptamer-based techniques to respond to signals via electrochemical means has been promising and suggests that these techniques may be generalized to multiple detection modalities [31]. In addition, these electrochemical approaches benefit from the relative scarcity of electro-active contaminants in vivo [1].

The unique physicochemical properties of particles at the nanoscale, including interesting thermal, optical and magnetic properties, have made the union of nanotechnology and biosensors an emerging trend in biosensor development. In particular, the high quantum yields, photostability and long emission/excitation spectra of nanoparticles have made them natural components of next generation optical biosensors [6]. These properties of nanoparticles have been incorporated into both aptamer- and peptide-based biosensors to improve both the sensitivity and spectral characteristics of biosensors [6]. In 2011, the Bhatia laboratory [32] reported on the development of a novel means of signal amplification via ‘nanoparticle communication’: in this system, functionalized tumour-targeting nanoparticles activate the coagulation cascade to recruit additional nanoparticles to induce robust signal ampflication. In this Theme Issue, Lacroix et al. [33] fully harness the unique physical properties of the nanoscale to develop inorganic nanoparticles that may be used for both optical and magnetic in vivo detection. The recent explosion of interest in graphene has led to the development of graphene quantum dots, with spectral and biocompatibility profiles ideal for future incorporation into biosensors [34,35]. In addition, nanoparticles have been used as dynamic and specific contrast agents for cellular magnetic resonance imaging of cancer [36]. Excitingly, Shen et al. [37] report in this Theme Issue on the development of a functionalized nanocomplex to dynamically detect mRNA in vivo, a technique with implications for both normal biology and disease. Although nanoparticles will doubtlessly play a central role in the development of future biosensor technologies because of their unique properties and versatility, the long-term safety of these nanoparticles is still a controversial and emerging issue that must be addressed for effective translation of biosensors based on these technologies [38]. Browning et al. [39] examine both the properties and toxicity of nanoparticles in a zebrafish model and shine light on the effects of nanoparticle size and composition on toxicity and biocompatibility.

Embedded or implantable microdevices constitute an enormous and growing field that encompasses everything from brain sensors for monitoring of electrical activity to implanted lactic acid and oxygen sensors [6]. These include the first real-time in vivo biosensors, proposed by Clark and Lyons [40], for the monitoring of glucose levels in patients with diabetes [1]. These devices most often couple an enzymatic or electrochemical detector to an embedded signal transducer/amplifier. These devices have enormous and immediate clinical utility in the management of disease, but suffer from issues of biofouling and biocompatibility as well as sensor drift and calibration difficulties [1,6,11,35]. Development of novel surface functionalities, often based on the principles of nanotechnology, and improved miniaturization technologies will be essential to improve the generalizability and applicability of these devices. Despite these challenges, multiple implantable microdevices are entering the market, including the GlucoWatch, a wearable gluocose monitor that uses an electric current to sample small volumes of glucose from the peripheral circulation [41]. In addition, recent work has raised the interesting possibility of incorporating microfluidic devices into implanted biosensors for detection of cancer biomarkers to both reduce biocompatibility issues and increase throughput, sensitivity and temporal resolution of these devices [42]. As more research and refinement occur, the future will probably see an increase in the breadth and sensitivity of these technologies, allowing dynamic and continuous monitoring of biological states in target patient populations.

Each class of biosensor has its own set of strengths and weaknesses, making them more or less suitable for a particular application. For some applications, such as glucose sensing, high spatial resolution is less important than temporal resolution, dynamic range and sensitivity. In these cases, implantable microdevices have already found utility in monitoring of blood glucose levels for patients with diabetes [43]. In contrast, biosensors for detection of cancer-associated protease activity are primarily used for detection and localization. In these cases, high spatial resolution, specificity and selectivity become more important. There have been exciting advances in the application of cancer protease-activated biosensors based on fluorescent proteins for surgery and diagnostics [44]. In the future, we believe that biosensors of different classes will both complement and supplement one another to obtain the maximum useful amount of information for facilitating clinically relevant decisions.

Although much progress has been made in improving and diversifying the range of in vivo biosensors, more work is clearly needed to exploit the full potential of these powerful techniques. This will involve not only refinements of current devices, but also development of new, paradigm-shattering technologies that combine high temporal and spatial resolution with increased sensitivity and excellent biocompatibility. True progress towards this goal can only be achieved by uniting the efforts of exceptional basic and translational biologists with engineers, chemists, physicists and materials scientists. We hope that this Theme Issue serves to not only highlight exciting research in the field, but also act as an impetus for investigators to challenge themselves by entering this exciting field.