A molecular biosensor [1] is a molecule composed of a biological recognition element (receptor) covalently associated with a transducer, generally a fluorophore, for signaling. The receptor is selected to specifically recognize a molecule of interest (analyte), while the fluorophore responds to the recognition event and transforms it into a measurable signal. The molecular biosensor can be distinguished from a chemosensor by the nature of the receptor, which is typically a biomacromolecule such as a nucleic acid (DNA or RNA), or a protein (often an enzyme or an antibody). Fluorescent molecular biosensors are mainly designed for the detection and quantification of analyte, with potential applications in the fields of medicine, agro-industry, defense or protection of the environment, but are also exploited to understand biomolecular events.

Compared to an immunoassay, which requires indirect labeling and multistep reactions, the detection of an analyte by a fluorescent biosensor is simple, direct and can be appreciated in real-time. The principle of sensing is illustrated in Figure 1. When an environmentally sensitive fluorophore is coupled near the binding site of a receptor, it will respond to changes in the physico-chemical environment induced by complex formation, by altering its fluorescence properties. Typically, polarity variations in the immediate vicinity of the fluorophore result in quenching or enhancement of the fluorescence and/or a wavelength shift in the emission spectrum. Recent reviews describe the structures, key physical parameters (extinction coefficients, excitation and emission wavelengths, quantum yields, size, hydrophobicity, stability) and advantages of some of the most common environmentally-sensitive fluorophores [2], or the design of biosensor molecules based on proteins [3]. While implemented with many different types of protein receptors, analytes and fluorophores, site-specific incorporation of fluorophores in proteins remains a challenge. The most widely used approach to covalently couple an extrinsic fluorophore to a recombinant protein involves the replacement of a less reactive amino acid residue with cysteine, followed by dye coupling to the free thiol group of the purified protein in buffer. This approach is straightforward, but difficult to implement on a general basis because mutations and coupling can lead to poor expression levels, deleterious effects on analyte binding, poor stability, aggregation or unfolding. Moreover, side reactions between thiol-reactive fluorophores and lysine side chains have been observed [4,5]. Other approaches to coupling a fluorophore at a specific site of proteins involve nonsense suppression [6], active-site-selective labeling [7–9], post-photoaffinity labeling modification [10–12], and, for chemically solid-phase synthesized proteins, the Suzuki-coupling reaction [13] and native-chemical ligation of peptides [14].

Protein-based environmentally-sensitive fluorescent biosensors have been successfully used for analyte detection and quantification with affinities varying from 0.2 nM to 150 mM, sometimes even in immobilized formats [14–17]. Still, synthetic peptides are interesting alternatives as receptors, because they provide obvious advantages over proteins in terms of production and stability. Peptide synthesis is straightforward and cost-effective and peptide-dye constructs can be manufactured at large scale. Moreover, synthetic peptides are robust, offer an even larger chemical versatility than recombinant proteins and are easily modified in a site-specific manner by means of orthogonal protection group strategies. Thus, peptides should have a great potential as components in molecular biosensors.

This review will specifically focus on biosensor constructs based on synthetic peptides and designed for the detection of biomolecular analytes by transduction via environmentally sensitive fluorescent dyes. Detection based on energy transfer (FRET) or changes in anisotropy will not be considered. The peptide may either serve as the recognition unit to provide ‘intrinsic recognition’ (Section 2) or as a scaffold onto which other recognition elements can be grafted (‘extrinsic recognition’, Section 3).

Half of the reported protein biosensors are based on bacterial periplasmic binding proteins and target sugars, the reason being that these proteins are well described and that sugar binding induces a pronounced conformational change that may result in local changes in the chemical environment in many parts of the protein. In contrast, most of the singly-labeled fluorescent peptide biosensors target proteic or peptidic analytes. Binding-related peptide sensors constructed by labeling of an amino acid residue and by incorporation of a fluorescent amino acid are listed in Tables 1 and 2, respectively, together with targeted analytes, sensor affinities and performances. Sensors for enzymes, where the fluorescence signal is changed in response to the enzymatic activity (discussed in Section 2.1.), are not included.

This section will describe how and for what purpose peptide biosensors are constructed.

In the examples given in Section 2, the sequence of the peptide was chosen in order to provide the recognition site for the analyte. A different role for the peptide has been presented in a series of papers by the groups of Akihiko Ueno and Lars Baltzer. Here, the folding propensity of polypeptides is a key property that makes them useful as scaffolds, onto which non-peptidic molecules for recognition can be grafted along with fluorophores for signaling. A well-defined structure provides directionality of incorporated groups, which is an important parameter in the design and optimization of biosensor performance. Also, the distance between recognition unit and fluorophore can be systematically varied by addressing different amino acid side chains in the scaffold.

A number of biosensor molecules based on a 17-residue, de novo designed α-helical peptide were presented, where different cyclodextrins were employed for recognition of bile acids and aliphatic alcohols [59–67]. The sensing principle was based on the disruption of an inclusion complex between cyclodextrin and a fluorophore upon binding of the analyte (guest). In the simplest design, dansyl or dimethylaminobenzoyl was attached in position i, i + 4 or i, i + 7 with respect to a β-cyclodextrin moiety, which corresponds to the two groups being separated by one or two turns of the helix scaffold while oriented similarly in space [59,65]. This arrangement allowed the fluorophore to interact with the hydrophobic interior of the cyclodextrin moiety. Upon introduction of a guest molecule, e.g., ursodeoxycholic acid, displacement of the fluorophore was reflected in quenching of its emission. In a different construct, α- or β-cyclodextrin was coupled to the scaffold along with pyrene and its quencher nitrobenzene [63,66]. The signaling principle was based on nitrobenzene and the analyte competing for the binding site inside cyclodextrin, affecting the distance between fluorophore and quencher and thus the fluorescence intensity (Figure 4). Yet another design approach involved functionalization with two molecules of naphthalene or pyrene, and analyte binding was monitored from excimer fluorescence [60–62,64]. Bile acids and ring-structured alcohols were sensed using constructs with β- [59,63–65,67] or γ-cyclodextrin [60–62] while α-cyclodextrin was used for short chain aliphatic alcohols [66].

In a similar approach, de novo designed helix-loop-helix polypeptides that dimerize into four-helix bundles upon folding were functionalized to become molecular biosensors for protein analytes. In the initial design, dansyl and a benzenesulfonamide derivative were attached to separate helices of the scaffold [68,69]. The presence of carbonic anhydrase, which binds specifically to benzenesulfonamide, was accompanied by a disruption of the dimer [70] and a significant intensity increase of the dansyl fluorescence.

The equilibrium dissociation constant (K_D) of interactions between peptide biosensors and their analytes can be determined from the hyperbolic binding curves obtained by titration of a constant concentration of peptide biosensor with increased concentrations of analyte [13,19,20,27,31,35]. Affinities determined in this way vary from low nM to 500 μM depending on the system under study. In two studies [30,36], competition binding assays with non-fluorescent peptides recognizing the same analyte as the peptide biosensor further complemented titration experiments and allowed for determination of the relative affinities of unlabeled peptides for their protein analytes. The binding affinities were also independently evaluated by using non-fluorescence-based methods; isothermal titration calorimetry (ITC, [36]) and surface plasmon resonance (SPR, [30]). In both cases, K_D values determined by ITC or SPR were in the same range as those determined by fluorescence-based procedures.

Based on the helix-loop-helix scaffolds designed by Baltzer and colleagues, biosensors with affinities towards carbonic anhydrase spanning three orders of magnitude were designed by varying the length of an aliphatic spacer attached to the benzenesulfonamide moiety [69,71]. Although interactions between the spacer and a hydrophobic pocket in the protein contributed to binding, the scaffold itself played an important role in providing both sterical constraints and chemical functionalities that affected the affinity [71].

Singly labeled reagentless biosensors based on proteins and used for analyte quantification displayed limits of detection (LOD) in the submillimolar range for maltose [54], and in the submicromolar range for glucose [72], glutamine [17,73], and β-lactam antibiotic [74].

The LOD of analyte by peptide biosensors was reported only in a few cases. The protein detection method established with folded peptides (see Section2.1) allowed for the detection of 5 μg/mL (# 90 nM) of α-amylase [24]. The LOD of the antibody fragment scFv1F4_Q34S by the pE6 peptide biosensor [30] was 15 nM in buffer, and 50–100 nM in a more complex medium (bovine serum albumin (BSA) 100 μg/mL). Normal concentrations of specific antibodies and BSA in serum are around 50 μg/ml (330 nM) and 35–50 mg/ml, respectively.

The performances of peptide and protein sensors were recently compared [30] based on various antigen-antibody interactions that display affinities in the 0.1–20 nM range. In these studies, the receptor was either an antibody [52,53], or a peptide antigen [29,30], but the analytes were large in all cases (between 14 and 54 kDa). A biosensor was defined as operational if it showed a >35% fluorescence intensity variation upon analyte binding. Out of the total number of sensors constructed in each study (maximum 10 for antibody receptors and 3 for peptide receptors), 20–50% behaved as biosensors. In these studies, the proportion of efficient fluorescent sensors was therefore in the same range whether the receptor was a protein or a peptide. This observation makes peptide biosensors promising tools for developing new applications. With the availability of automatedchemical synthesis of long peptide fragments at high scale (100 mg of a 100 amino acid peptide can be routinely synthesized at >99% purity) and low price, rational or combinatorial peptide libraries can be straightforwardly produced for the optimization of recognition sites and positions for fluorophore attachment. However, the main challenge in the future will be to make peptide-based fluorescent biosensors useful for sensitive detection of bio-analytes in complex media (e.g., serum). In order to accomplish this, problems with interference from non-analyte components in the sample have to be addressed. Specifically, the possibility of enzymatic peptide degradation and of non-specific binding of sample components to the peptide, causing irrelevant sensor signals, has to be taken into consideration. The frequent problem of autofluorescence can be avoided by employing long-wavelength dyes, which are usually very bright although they may be less sensitive to environmental changes compared to fluorophores excited in the UV region. Ratiometric sensor peptides, where the relative rather than the absolute intensity change upon analyte binding is monitored, should also be very useful in this respect.

In conclusion, fluorescent sensor peptides have already proven useful in a number of applications, ranging from analyte detection to elucidation of molecular details of protein-peptide and protein-protein interactions. It is expected that their general applicability with respect to analyte quantification – beyond proof-of-concept – will be expanded by combining combinatorial methods for peptide design with further improvement of fluorophores and fluorescent amino acids in terms of sensitivity to environmental changes. The most robust protein detection system will then probably rely on fingerprints obtained from arrayed sensor peptides.