Conventional molecular imaging technologies for ion imaging have largely depended on synthetic organic probes. The pivotal strategy for designing organic probes is that a known fluorophore, such as fluorescein or rhodamine, is chemically modified with a known metal ion receptor. These organic probes enable to trace changes in intracellular metal ion concentrations and distributions, as well as mobilization of metal ion pools upon physiological or pathological stimulation. New efficient chemical design of the probes may offer the ability to perform new types of molecular imaging experiments that are currently unavailable.

For instance, a rhodamine derivative, RhodZin-3, was previously studied for tracing mitochondrial Zn²⁺ homeostasis [14] and in studies of the functions of the zinc transporter, Znt5B [15]. FuraZin and IndoZin are Zn²⁺ sensors derived from the classic Fura and Indo Ca²⁺ probes and operate via an internal charge transfer mechanism, in which metal ion binding to the probe modulates the electron-donating properties of the electron-rich receptor [16]. Fluorescein-based Zn²⁺ probes were also developed and named as ZnAF series indicators. In addition to fluorescein- and rhodamine-based indicators, cell-compatible Zn²⁺ sensors have been developed using the backbone of BODIPY [17], naphthalimide [18], cyanine [19], and coumarin [20] dyes.

Furthermore, as an example for peptide-sensing, Griffin et al. developed a synthetic chemical indicator that can recognize six natural amino acids (cysteines) that may be genetically incorporated into proteins of interest. The fluorescein-derived indicator, named “FLAsH”, recognizes four cysteines placed at the i, i + 1, i + 4, and i + 5 positions of a α-helical peptide [21,22].

As an inorganic nano-imaging agent, QDs are representative for molecular imaging. QDs are semiconductor nanoparticles of 1–10 nm in diameter. These luminescent nanocrystals are composed of atoms from the II–VI (CdS, CdSe, CdTe, ZnO, ZnSe), III–V (InP, InAs, GaN, GaP, GaAs), and IV–VI (PbS, PbSe, PbTe) groups of the periodic table [23]. CdX QDs (X = S, Se, Te) are especially attractive because they emit electromagnetic spectrum in the range of ultraviolet, visible, and near-infrared. The emission color of the QDs is size-dependent [6,7].

However, cytotoxicity is a big issue for QDs. Class A element (Cd, Pb, and Hg) based QDs show cytotoxicity in vitro under extreme conditions [24]. To avoid this problem, these elements can be replaced by other more benign elements, such as CuInS₂/ZnS core/shell QDs.

Near-infrared light (700–900 nm) efficiently penetrates skin and blood more than visible light because these tissues scatter and absorb less light and low autofluorescence at longer wavelengths [25]. Near-infrared light also has potentially high spatial resolution, sensitivity, and low risk to living subjects because it utilizes non-ionizing radiation [24]. Therefore, the desirable optical property of nano-imaging agents for tissue imaging is emission of near-infrared light.

A smart design of nano-imaging agents emitting near-infrared light was reported with name of “self-illuminating nanoprobes” [26,27]. The surface of QDs was covalently coupled to a donor of bioluminescence resonance energy transfer (BRET), RLuc8. Energy from a chemical reaction catalyzed by the donor luciferase is transferred to the linked QD emitting near-infrared light. This molecular design has enabled us to carry out self-illuminative imaging of cancers in the deep tissue of living subjects without the need for external excitation.

When compared with organic nano-agents, QDs possess higher photostability and brightness, but exhibit toxicity and poor in vivo biodistribution. Rational design of new QD imaging agents is deeply related to the surface labeling chemistry. The general principle for QD labeling is that first the QDs are made water-dispersible and then are bound to biomolecules. This labeling can be achieved electrostatically, via biotin–avidin interactions, by covalent cross-linking or by binding to polyhistidine tags [28].

As organic and inorganic nano-scale imaging agents have been well reviewed in previous papers [1,13,24], we have focused on genetically encoded nano-imaging agents.

Since green fluorescent protein (GFP) was first extracted from jellyfish Aequorea victoria by Dr. Shimomura [29], fluorescent proteins have been popularly utilized in a variety of molecular imaging studies.

This native fluorescent protein (FP) was then carefully modified with mutagenesis to improve its optical properties, such as red color-emitting spectra, fast folding properties, enhanced fluorescence intensity, low sensitivity to ions, and temperature stability. A crystallographic study on the structure of GFP revealed regular β-barrels with 11 strands on the outside of the cylinder. These cylinders were found to have a diameter of about 3 nm and a length of about 4 nm [30]. In addition to wild-type GFP, the discovery of novel GFP-like proteins from Anthozoans (coral animals) has significantly extended the range of colors from green to red.

The characteristic of FP-based probes is that they smartly make use of the optical properties of FPs. For instance, a photoconvertible fluorescent protein was first developed by Dickson et al. in the process of GFP mutation [31]. Since then, FPs with changing fluorescence intensity (photoactivation) or color (photoconversion or photoswitching) in response to light of a specific wavelength have been discovered. The photoconversion of FPs, such as Kaede, has been explained as the break or extension of π-conjugation in the chromophore of Kaede in response to specific light [32]. Similarly, illumination of an FP named Dronpa reversibly switches on and off by emission at 488 and 405 nm. These interesting optical characters of some FPs imply that they can act as excellent fluorescent probes even without any artificial modification to the FPs. In fact, in the early days’ design of probes, GFP was simply fused to a protein of interest.

The basic designs of fluorescent probes have been further elaborated to create more complex imaging probes, which include probes with circular permutation (CP) and FRET. CP is an interesting design for fabricating nano-imaging probes embedding GFP variants, where the β-barrels of GFP are rearranged. To exert an on–off switch to the structure of GFP, the original N- and C-terminals of GFP have been fused with a short peptide linker, whereas a new N- and C-terminals have been created at the hinges of the β-barrels. Subsequently, the new terminals have been extended with a pair of proteins of interest, respectively. This artificial crack to the cylindrical structure results in disruption of fluorescence because of solvent penetration to the hydrophobic chromophore within the cylinder. The fluorescence is resurrected by the closure of the crack through the binding of the protein pair [33,34].

Many of the fluorescent probes based on protein-fragment complementation and reconstitution are unfortunately irreversible and time-consuming for chromophore formation. This irreversible feature destroys the quantitative fidelity of the results and limits the analysis of temporal interactions of proteins of interest.

Majority of FP-based imaging probes make use of FRET, which is an excellent platform for imaging the molecular dynamics of proteins in cells. The FRET-based probes make use of the ability of a higher energy donor fluorophore to transfer energy directly to a lower energy acceptor. The energy donor and acceptor fluorophores are fused to a pair of proteins of interest. Upon interaction between the pair of proteins in the probe, the relative distance between the donor and acceptor fluorophores varies. On the other hand, upon dissociation of the pair of proteins, FRET disappears. Therefore, this method is based on reversible, intracellular protein–protein interactions. The working mechanism of FRET-based probes is explained in detail in Section 4.1.

Beetle and marine luciferases are attractive imaging agents for bioanalysis and molecular imaging. The light-emitting reaction is very simple, harmless, and does not require an external light source or cofactors [35]. Unlike other organic fluorescent agents or inorganic nanoparticles, which have to be delivered to the imaging target, bioluminescent probes can be genetically encoded and expressed in situ. The bioluminescent probes do not need external light source for luminescence and thus background intensities can be significantly reduced [13].

Recent engineering of luciferases with designed properties and functionalities represents an important direction for molecular imaging and bioassays. Many of the native luciferases are unstable and poor in optical intensity. These native luciferases have been modified through a random and site-directed mutagenesis for enhancing the optical intensities and prolonged emission half-life [36].

Random mutagenetic approach of luciferase engineering is normally slow and tedious, and enforces severe consumption of time and labor. In particular, in the absence of crystallographic data, no appropriate measures are available for luciferase engineering. Thus, several rational strategies to manipulate luciferases have been developed to relieve the efforts on luciferase engineering.

A semi-rational strategy for luciferase engineering, called “a consensus sequence-driven mutagenesis strategy”, was demonstrated in a study by Steipe et al.[37]. An alignment of relative protein sequences is known to reveal frequently occurring and less occurring amino acids. It is considered that frequently occurring amino acids at a given position have a larger thermostabilizing effect than the less frequent amino acids [38]. In this sense, mutation sites and kinds of amino acids are decided accordingly to increase the frequently occurring amino acids in the alignment.

Kim et al. also introduced a semi-rational strategy for effective luciferase engineering [10,39]. They interestingly focused on the chemical structural similarity between the fluorophore of GFP and the common substrate of marine luciferases, coelenterazine (Figure 2). Based on this structural comparison, they speculated that GFP variants embed the chromophore inside the molecular backbone, whereas marine luciferases recruit the choromophore (we call it “substrate”) from the outside. In this view, they speculated that a hydrophilic interface in the luciferases provide a favorable platform for substrate recruitment (we call this as “active site”).

Accordingly, Kim et al. suggested a hydrophilicity search of luciferases to assume a putative active site as an engineering target of the luciferases for mutation and dissection, prior to empirical engineering, because a hydrophilic interface of luciferases should be advantageous to recruit the substrate in the aqueous phase. As shown in Figure 3A, this view is supported by many precedent studies on luciferase-based assays [11,40–44]. This characteristic hydrophilic region is interestingly highly conserved and comprises a drastic interface between highly hydrophilic and hydrophobic amino acids in the sequence of beetle and marine luciferases (Figure 3B). This region has been frequently utilized in the engineering as an optimal dissection or mutation target. The firefly luciferase (FLuc) is fragmented into two parts at amino acid 415, which are sandwiched between the ligand binding domain of glucocorticoid receptor (GR LBD) and an LXXLL motif. In the presence of androgen, the activated AR LBD binds to an LXXLL motif. The binding of AR LBD–LXXLL motif further induces reconstitution of the adjacent split-FLuc and light emission (Figure 3C).

In this section, we list the prototypical procedure for fabricating a single-chain probe with split-FLuc.

This basic procedure may be modified according to the characters of luciferases.

When designing a single-chain probe, flexible linker is a difficult issue to examine. Although great efforts have been devoted to create excellent linkers for ideal integration of the components in a single entity, elaboration of an efficient linker still remains a challenging mission. In any case of the linkers, their intended control is practically difficult owing to unexpected features of single-chain probes, such as unpredictable steric hindrance. Therefore, researchers may consider the following practical strategies as a labor-saving choice of developing linker peptides: (i) to mimic the success stories of fusion protein probes that were previously well approved and (ii) to gain cumulative experience based on trial-and-errors.

The principle of FRET is based on energy transfer from a higher energy donor fluorophore to a lower energy acceptor fluorophore [53]. The energy donor and acceptor fluorophores are fused to a pair of proteins of interest. Upon interaction between the pair of proteins in the probe, the relative distance between the donor and acceptor fluorophores varies. The variance in the distance exerts a color change. Efficient FRET probes are designed to address the following concerns: (i) optimizing the brightness (quantum yield (QY) and extinction coefficient (ɛ)) between the donor and acceptor proteins; and (ii) minimizing spectral cross-over between the donor and acceptor. For instance, the acceptor can be directly excited with a light source, instead of the donor. This should exhibit false-positive fluorescence.

Molecular imaging with FRET is one of the most popular methods for observing the spatial and temporal dynamics between two distinct proteins of interest in living cells. FRET provides a very sensitive measure of small changes in intermolecular distances. Therefore, the presence of FRET is a good indicator of close proximity, implying protein–protein interactions.

A key strategy to fabricate an efficient FRET probe is to use potential fluorescent protein pairs exerting excellent photophysical properties, such as high extinction coefficient, QY, photostability, and greater Förster radius R₀.

An interesting design of a FRET probe was previously reported with the name of “Flip-Flop”-type indicator or simply “Fllip” [54,55] (Figure 4A). This probe makes use of two GFP variants and their energy transfer is similar to a common FRET probe. The distinctive feature of this probe is the rigid α-helical linkers inserted between the probe components. This probe is designed to be fixed on the plasma membrane (PM) and remain inactive in the basal condition. However, the probe dramatically transforms in the presence of a second lipid messenger, PIP₃. This intramolecular conformation change varies the relative distance between the incorporated GFP variants, which results in color change.

BRET is similar to FRET with respect to the resonance energy transfer (RET). In fact, BRET shares a similar working mechanism with FRET, except that the donor molecule is a luciferase, not a fluorescent protein. The resonance energy by a luciferase is transferred to an adjacent fluorophore.

BRET is a useful technique to access protein–protein interactions because the BRET-permissive distance of less than 10 nm is very similar to the dimensions of biological macromolecular protein complexes [57].

An interesting design of BRET probe was demonstrated by Hoshino et al.[58] (Figure 4B). The probe named “BRET-based Autoilluminated Fluorescent Protein on EYFP (BAF-Y)” exhibited an extraordinary efficiency in the RET between a variant of Renilla lucicferase (RLuc8) and EYFP. This excellent RET was achieved simply through an optimization of the linker length and selection of the luciferase and fluorescent protein.

Although BRET between luciferases and fluorescent proteins is well known, that between luciferases and QDs has also been represented previously. In the probe design, Cypridina luciferase (CLuc) and QD were conjugated to biotin and streptabidin, respectively. The biotin–streptavidin reaction resulted in BRET between CLuc and QD [59].

The above-mentioned bioluminescence is a member of chemiluminescence. Many chemical reactions emitting light have been utilized for bioassays. Although being less familiar than FRET and BRET, CRET has been utilized in the design of various nano-imaging agents [60] (Figure 4C). CRET is a nonradiative RET process, in which the excited donor is produced by a chemical reaction. A molecular design of CRET was typically elucidated by conjugation of an energy donor (e.g., horseradish peroxidase (HRP)) and an energy acceptor (e.g., QDs). For example, gold nanoparticle (AuNP) based CRET systems have been reported previously. In the probe design, AuNP and HRP were conjugated to two different antibodies. When the two antibodies bind to a same antigen, it indicates construction of a sandwich immunocomplex. This immunocomplex quenches CRET between HRP (energy donor) and AuNPs (energy acceptor) [61].

Nanospectroscopy with PRET is another example of advancement in molecular imaging in living cells. PRET imaging relies on the resonant plasmonic energy transfer from a gold nanoplasmic probe to conjugated target molecule.

The imaging mechanism is illustrated in Figure 4D. When fluorescent molecules are close to the surface of a metal nanoparticle, their fluorescence increases, which is called metal-enhanced fluorescence (MEF) where the optimal distance is about 10 nm. Quenching of the fluorescence occurs when the fluorophore is located much closer to the metal surface than that for MEF. PRET suppresses light scattering from nanoparticles when resonance-matched molecules, such as metal complexes or dyes, are in the vicinity of the nanoparticles [62]. The scattering spectrum of PRET creates quantized quenching dips within the Rayleigh scattering spectrum of the probe [63]. The spectrum change in response to the functionalized PRET probe to various environmentally and biologically relevant metal ions appears significant.

Protein-fragment complementation-based probes (PCP) are one of the major tools for determining protein–protein interactions. The basic design of PCP is as follows: (i) monomeric reporter proteins such as GFP and luciferase are first dissected into N- and C-terminal fragments for the temporal inactivation of the optical properties; (ii) a set of proteins of interest is linked to each of the fragments of the reporter protein via appropriate linker peptides for unity. Upon interaction between the pair of proteins, the adjacent fragments of the reporter protein are approximated. The restored optical intensity is determined with an appropriate detector. One challenge for the development of complementation strategies is the determination of the optimal dissection sites of luciferase in the PCP. Optimization of the split sites in luciferase are of critical importance for guaranteeing temporal inactivation and conditional reconstitution of the fluorescence or luciferase activity.

Many PCPs have been designed for complementation between two independent fusions. In fact, such analyses are validated on the premise that the two component fusions are expressed equally beforehand. Biased expression of the component fusions inevitably exerts their inefficient complementation. To overcome this limitation, Kim et al. demonstrated a series of single-chain bioluminescent probes, where all components for ligand sensing and light emission were integrated into a single backbone. In contrast to the precedent inter-molecular complementation, this type of probes provides an intra-molecular complementation [64].

The above-mentioned basic design of PCP has been applied in various directions. CP of split-luciferase in a single-chain probe is an interesting design for low S/N ratios [65,66] (Figure 1D). In this probe, the active site in the luciferase is first dissected into two fragments. The N- and C-terminal fragments are replaced in a C- and N-terminal order. This means that the active site fragments are placed at the opposite sides of each other. In this molecular design, the encountering chances between the active site fragments dramatically decrease, which consequently minimizes the background intensities.

The concept of PCP was further extended to visualize mRNAs in living cells (Figure 4E). PUMILIO is an RNA-binding protein that comprises an array of RNA-recognizing amino acids. In the unique probe design, two RNA-binding domains of human PUMILIO1 (e.g., mPUM3 and mPUM4) are fused with N- and C-terminal fragments of split-EGFP, respectively. When the mPUM3 and mPUM4 sense the corresponding mRNA sequences in living cells, the adjacent N- and C-terminal fragments of EGFP are complemented, resulting in resurrection of strong fluorescence. This strategy was successfully applied to determine mRNA of β-actin [67] and endogenous mitochondrial mRNA [68].

Another tendency in the molecular design of PCPs is to construct a set of multicolor imaging probes, where multiple luciferases emitting various colors are fragmented in the same probe set, and are conditionally reconstituted according to the signal that is activated; e.g., whether the signals are agonistic or antagonistic [11]; or whether Smad4 binds to Smad1 or Smad2 (Figure 4F) [49].

The basic concept of PCPs is also utilized for imaging molecular events on the PM. For instance, interaction of G-protein coupled receptors (GPCRs) with β-arrestin on the PM was imaged with a set of bioluminescent probes [69,70] (Figure 4G). Similarly, dimerization of extracellular signal-regulated kinase 2 (ERK2) was visualized with a pair of ERK2-linked split-RLuc [40].

Protein splicing is a naturally occurring, posttranslational processing event involving precise excision of an internal protein segment (intein) from a primary translation product, i.e., external protein (extein). At intein–extein junctions, conserved amino acid residues are directly involved in the protein-splicing reaction. An important feature of protein splicing is a self-catalyzed excision of the intein and ligation of the flanking exteins without any exogenous cofactor or energy source [71].

This natural occurrence for protein maturation was successfully utilized for designing a new series of molecular probes. In the classic probe design, N- and C-terminal fragments of split-VDE intein (1–184 amino acids and 389–454 amino acids) were respectively linked to N- and C-terminal domains of split-GFP (1–128 amino acids and 129–239 amino acids). This pair of fusions was further labeled with calmodulin (CaM) and its target peptide (M13), respectively [72].

Ca²⁺-bound CaM interacts with M13. This interaction provokes a protein splicing reaction between the approximated N- and C-terminal fragments of VDE, generating a full-length GFP.

Conditional protein splicing was well demonstrated by Tyszkiewicz et al., who developed a protein splicing system activated by red (660 nm) and far-red (750 nm) light in yeast [58] (Figure 4H).

An innovative design called “luciferase cyclisation” among reporter reconstitution technologies was previously reported [73,74] (Figure 4I). In the probe design, the N- and C-terminal fragments of FLuc were circularly permutated via a substrate peptide of caspase-3 (DEVD). The N- and C-terminal ends of the CP FLuc were further linked to the N- and C-terminal fragments of a DnaE intein. After translation into a single polypeptide, the N- and C-terminal ends were self-catalytically ligated by protein splicing, creating a closed circular fusion protein. As the molecular structure is distorted and strained by a tension, CP FLuc temporarily loses its enzymatic activities (closed circular form). In the presence of an apoptosis signal, active caspase-3 digests the DEVD sequence in the circular fusion and linearizes the formation (opened linear form). In the linear form, the CP FLuc restores the optical intensities, which indexes activity of caspase-3.

Unique molecular designs in genetically encoded nano-imaging agents have been developed. A unique bioluminescent probe named “molecular-tension probe” was developed on the basis of molecular tension of a luciferase that is artificially appended by protein–protein binding [75] (Figure 4J). Unlike other probe design, this probe makes use of the full length of a luciferase, which is sandwiched between a pair of proteins of interest. In the probe design, the linker length between the ingredients is minimized to exert an efficient molecular tension to the sandwiched luciferase. This simply designed molecular probe is surprisingly sensitive to estrogens.

All of the genetically encoded probes are roughly divided into two categories: GTP and NTP. Recently, a probe system that combines the characteristics of the two groups has been developed [65] (Figure 4K). The probe system consists of two ligand-sensing machineries: (i) a reporter-gene system carrying a glucocorticoid response element (GRE) promoter; and (ii) a single-chain probe sensing glucocorticoids. The cDNA encoding the single-chain probe is situated downstream of the GRE promoter, indicating that this system carries two on–off switches for glucocorticoids. Initially, glucocorticoids activate the GRE promoter, which starts expressing the single-chain probe. The same glucocorticoids are repeatedly recognized by the expressed single-chain probe, which emits bioluminescence. The probe system comprises both the characteristics of genetic and nongenetic probes, and thus is characteristically classified into an interface region between the two categories.