Lanthanide-Based Optical Probes of Biological Systems

SUMMARY

The unique photophysical properties of lanthanides, such as europium, terbium, and ytterbium, make them versatile molecular probes of biological systems. In particular, their long-lived photoluminescence, narrow bandwidth emissions, and large Stokes shifts enable experiments that are infeasible with organic fluorophores and fluorescent proteins. The ability of these metal ions to undergo luminescence resonance energy transfer and photon upconversion further expands the capabilities of lanthanide probes. In this review, we describe recent advances in the design of lanthanide luminophores and their application in biological research. We also summarize the latest detection systems that have been developed to fully exploit the optical properties of lanthanide luminophores. We conclude with a discussion of remaining challenges and new frontiers in lanthanide technologies. The unprecedented levels of sensitivity and multiplexing afforded by rare-earth elements illustrate how chemistry can enable new approaches in biology.

Graphical Abstract

Lanthanide luminophores have unique photophysical properties that make them versatile reagents for optical imaging. In this review, Cho and Chen highlight advances that have unlocked the potential of these rare-earth metals as imaging tools and applications of lanthanide probes in cellular and organismal systems.

INTRODUCTION

Biology is becoming an increasingly molecular and quantitative science, driven by our curiosity about the machinery within cells and organisms. Technological advances have opened new windows into these complex processes, particularly methods that enable the observation of specific molecular structures, interactions, and reactions. A wide-ranging palette of synthetic and genetically encoded fluorophores have been designed (Rodriguez et al., 2017; Wang et al., 2019), and new modalities for photometry, flow cytometry, microscopy, and macroscale imaging have been developed (Feuchtinger et al., 2016; Power and Huisken, 2017; Sigal et al., 2018; Spitzer and Nolan, 2016; Wang and Yao, 2016). In combination, these advances have made possible the detection of transcripts, proteins, metabolites, and other molecules in diverse biological systems.

Despite this significant progress, major technical challenges continue to constrain our ability to visualize biology at the molecular level. For example, biological autofluorescence can preclude the detection of targeted transcripts or proteins at their physiological concentrations. While probes for these macromolecules can be used in conjunction with signal-amplifying technologies, these approaches typically require sample fixation and cannot be used to study living systems (Ali et al., 2014; Bi et al., 2017; Pichon et al., 2018). Commonly used fluorophores also are not optimal for macroscale imaging, as their excitation and emission wavelengths are readily absorbed or scattered by biological tissues.

Our ability to observe the molecular interactions and reactions that create living systems has been similarly constrained. Although fluorescence resonance energy transfer (FRET)-based sensors have been developed for a variety of intermolecular complexes, protein modifications, and metabolic products, their sensitivity and accuracy are limited by their broad excitation and emission bands and modest Stokes shifts (Algar et al., 2019; Bajar et al., 2016; Hochreiter et al., 2015). Achieving a systems-level understanding of biology also requires the simultaneous observation of multiple targets. However, multi-color detection is typically restricted to five or six spectral channels, even with state-of-the-art probes, lasers, and optical filters.

Our efforts to overcome these hurdles would benefit from new chemistries and detection modalities, and lanthanide-based technologies have shown particular promise. Lanthanide complexes and nanoparticles can have unique photophysical qualities, including millisecond-scale luminescence lifetimes, large Stokes shifts, ultra-sharp emissions, and the ability to undergo resonance energy transfer or photon upconversion. Lanthanide-based probes therefore can be adopted to achieve unparalleled detection sensitivity, multiplexing capacity, and imaging depth, allowing biological systems to be interrogated in novel ways. Yet lanthanide luminophores are not widely utilized for biological imaging. In comparison to conventional fluorophores and fluorescent proteins, lanthanide probes have been limited by their brightness, dependence on UV excitation, and low photon efflux. Recent progress in probe design, instrumentation, and imaging modalities have overcome many of these constraints, sparking renewed interest in these rare-earth metals.

This review focuses on the technological advances that have unlocked the potential of lanthanide luminophores as biological probes. We first describe the photophysics and chemotypes of lanthanide-based probes, comparing their attributes to those of other photoluminescent molecules. We then provide specific examples of how these luminescent metal ions have been applied in photometric assays, biological microscopy, and whole-body imaging. We conclude by discussing how lanthanide technologies could be further enhanced to create new opportunities for biological discovery.

CHEMISTRY OF LANTHANIDE PROBES

Lanthanide Photoluminescence

Lanthanide luminescence is fundamentally different from fluorescence. The latter is generated when a molecule transitions from one electronic state to the other, whereas lanthanide metal ions luminesce from a parity-forbidden transition within 4f atomic orbitals. This distinct photophysical process confers to lanthanides their characteristic spectral properties.

First, the 4f-4f transitions in luminescent lanthanide ions are “disallowed” by the Laporte selection rule, and their low probabilities result in extinction coefficients that are several orders of magnitude lower than those of commonly used fluorophores (<10 M⁻¹ cm⁻¹ vs 10,000 to 100,000 M⁻¹ cm⁻¹). Since efficient, direct excitation of these rare earth metals requires light intensities that are destructive to biological samples, ‘antenna’ chromophores are typically used to transfer energy to lanthanide ions (Figure 1A). Most of these chromophores absorb 300- to 400-nm light and have extinction coefficients that are greater than 10,000 M⁻¹ cm⁻¹ (Figure 1B, Table S1, and Table S2) (Armelao et al., 2010; Heffern et al., 2014; Mathis and Bazin, 2010), and a few that respond to visible light have been developed as well (Ma et al., 2018; Tian et al., 2014). The resulting photoluminescence is also associated with Stokes shifts that are several hundred nanometers, allowing clear separation of excitation and emission signals. Moreover, lanthanide probes are resistant to photobleaching since the metal ion serves as a triplet state quencher for the antenna chromophore (Xiao et al., 2011).

Figure 1. Photophysical properties and chemical structures of luminescent lanthanide complexes.

(A) Chemical structure of Eu³⁺-DTPA-cs124-CF₃ complex, highlighting the chelate and antenna moieties. (B) Commonly used chromophores in lanthanide chelates. (C) Theoretical decay curves of luminophores with 10-nanoseconds (red), 10-microsecond (green), and 1-millisecond (blue) lifetimes. (D) Emission spectra of trivalent lanthanide ions. Adapted with permission from (Bünzli, 2016).

Another consequence of the disallowed 4f-4f transitions is the slow decay of lanthanide excited states, which translates into exceptionally long emission lifetimes. While small organic fluorophores and fluorescent proteins emit light for only several nanoseconds, lanthanide luminescence can last for a few milliseconds (Figure 1C). For example, the luminescence lifetimes of commonly used Tb³⁺ and Eu³⁺ complexes range from 0.5 to 2.5 ms (Nishioka et al., 2006; Xiao and Selvin, 2001; Xu et al., 2011). Since the excited 4f states are shielded by the filled 5s² and 5p⁶ subshells, lanthanide luminescence is also highly insensitive to the chemical environment. Lanthanide emissions therefore have very narrow bandwidths (Figure 1D), and individual peaks that originate from different transitions are easily distinguishable. For example, the Tb³⁺ luminescence is characterized by line-like emissions at 490, 540, and 580 nm, and Eu³⁺ luminescence at 590, 610, and 720 nm. Several other lanthanides have ultrasharp emission peaks in the visible range, but their modest quantum yields (<10%) have limited their utility as imaging probes (Moore et al., 2009).

Synthetic Lanthanide Chelates

Chemists have elucidated many of the chemical prerequisites for making bright lanthanide complexes for biological applications. The antenna effect was discovered very early in the study of lanthanide luminescence, and the ability of EDTA and NTA (nitrilotriacetate) chelates to suppress aqueous quenching was described several decades ago (Horrocks Jr et al., 1977; Weissman, 1942). One of the first photoluminescent lanthanide chelates was Tb(DPA)₃³⁻ (DPA: dipicolinic acid) (Barela and Sherry, 1976) (Figure 2A), and subsequent efforts to develop new lanthanide-based optical probes have strived to maximize molar absorptivity, antenna-to-lanthanide transfer efficiency, quantum yield, and photo- and in vivo-stability. For example, the Raymond group has developed minimal chromophoric chelators that are optimized for Tb³⁺ and Eu³⁺: IAM (2-hydroxyisophthalamide) and 1,2-HOPO (1-hydroxypyridin-2-one) (Figure 1B) (Moore et al., 2009). These compounds are direct descendants of DPA; both IAM and 1,2-HOPO are bidentate ligands, and four units of IAM or 1,2-HOPO are combined to generate an octadentate complex. In particular, a Tb³⁺ cryptate that contains four IAM molecules, Tb³⁺-Lumi4, is one of the brightest and most long-lived Tb³⁺ complexes (Xu et al., 2011), with an extinction coefficient, quantum yield, and luminescence lifetime of 23,700 M⁻¹ cm-1, 50%, and 2.45 ms, respectively.

Figure 2. Lanthanide-binding chelates and peptides.

(A) Chemical structure of Tb(DPA)₃³⁻. (B) Lanthanide ligands with a polyaminocarboxylate backbone and a cs124 antenna. (C) Ligands with a terpyridine or terpyridine-like backbone. (D) Amino acid sequences of lanthanide ion-binding peptides.

Polyaminocarboxylates functionalized with a single antenna are among the most frequently used lanthanide chelates. The Selvin and Parker groups have synthesized and characterized several probes in this structural class, including DTPA, TTHA, DOTA, and N₂O ligands that are functionalized with carbostyril-based antennae (Ge and Selvin, 2008; Li and Selvin, 1995; Montgomery et al., 2009; Parker and Williams, 1996; Xiao and Selvin, 2001) (Figure 2B and Table S1). In particular, the cs124 chromophore has been coupled with both Tb³⁺ and Eu³⁺, and cs124-CF₃ preferentially transfers energy to Eu³⁺, with an extinction coefficient of 15,000 M⁻¹ cm⁻¹ at 360 nm. Lanthanide-polyaminocarboxylate complexes have varying biological stabilities in vivo, making certain classes more amenable to biological applications. For example, DOTA and N₂O chelators can bind trivalent lanthanide ions in the presence of millimolar nucleoside triphosphates (NTPs and dNTPs), but DTPA and TTHA cannot (Cho et al., 2018; Ge and Selvin, 2008; Hewitt et al., 2017).

Multivalent, unimolecular ligands can also be designed to be intrinsically chromophoric. Terpyridine (Mukkala et al., 1993) and terpyridine-like (Mukkala et al., 1996; Takalo et al., 1997) lanthanide chelates pioneered by the Takalo group are the earliest examples of this class (Figure 2C). In the case of 2,2’:6’,2”-terpyridine-based chelates, three nitrogens of the terpyridine moiety coordinate the central lanthanide ion, together with polycarboxylate arms attached to the 6 and 6” positions. The terpyridine doubles as a light harvesting center. Among the lanthanide complexes that adopt the terpyridine scaffold, Eu³⁺-ATBTA stands out for its brightness and stability (Nishioka et al., 2006). This lanthanide chelate has a 20,000 M⁻¹ cm⁻¹ extinction coefficient at 360 nm and a quantum yield of 9.1%, and it remains photoluminescent in phosphate-buffered saline (PBS) even though the pK_sp for EuPO₄ is ~25 (Firsching and Brune, 1991). The biochemical stability of Eu³⁺-ATBTA is further demonstrated by its perdurance for days in zebrafish embryos (Cho et al., 2018) and ability to withstand heating to 95 °C (Nishioka et al., 2006). Accordingly, structurally related Tb³⁺ and Eu³⁺ complexes have been employed as TaqMan probes for quantitative polymerase chain reaction, affording a linear detection range spanning six orders of magnitude when monitored by time-gated photometry (Gueimonde et al., 2004; Nurmi et al., 2002; Nurmi et al., 2000).

Lanthanide-Binding Protein Motifs

Trivalent lanthanide ions can also form stable complexes with polypeptides. It had long been known that lanthanides can replace Ca²⁺ in biological systems (Nieboer, 1975), and most peptidic lanthanide-binding motifs are derived from Ca²⁺-binding sites (Figure 2D). For example, the Szabo group showed that Tb³⁺ binds strongly to the 14-amino acid loop in EF hand motifs and that Tb³⁺ excitation is promoted by nearby aromatic residues, especially tryptophan (Hogue et al., 1992). The Imperiali group subsequently used a mutagenesis screen to optimize the EF hand-derived lanthanide-binding motif, developing peptidic ligand with a dissociation constant (K_d) of 220 nM (Franz et al., 2003). Further screening yielded an Tb³⁺-binding peptide with a 19 nM K_d (Martin et al., 2005), and antenna chromophores, such as cs124 and acridone, were introduced as unnatural amino acids (Reynolds et al., 2008). More recently, a naturally occurring lanthanide-binding protein lanmodulin was identified from M. extorquens. This gene product contains four EF hand-like motifs and binds most of the trivalent lanthanide ions with picomolar affinity (Cotruvo et al., 2018). However, it should be noted lanmodulin lacks a tryptophan sensitizer and therefore is not suitable for photometric assays or imaging.

Lanthanide-binding protein motifs have been extensively used in X-ray crystallography and NMR studies (Barthelmes et al., 2017; Barthelmes et al., 2015; Martin et al., 2007; Silvaggi et al., 2007; Welch et al., 2003; Wöhnert et al., 2003; Zhuang et al., 2008). Lanthanides bound to these peptidic chelators have electron densities and paramagnetic properties that facilitate the determination of macromolecular structures. However, the application of lanthanide-binding protein motifs in live cells and organisms has been limited. Their affinities for free Tb³⁺ and Eu³⁺ cannot prevent ion sequestration by nucleoside triphosphates (Cho et al., 2018; Ge and Selvin, 2008; Morrison and Cleland, 1983) and inorganic phosphate (Firsching and Brune, 1991). In addition, the ability of trivalent lanthanide ions to precipitate phosphoproteins (Güzel et al., 2012) could significantly impede the delivery of soluble lanthanide ions to genetically encoded chelators in live cells and organisms.

Nanoparticles

Nanoparticles and metal-organic frameworks represent a third class of photoluminescent lanthanide probes. While biological applications of metal-organic frameworks have been limited (Rocha et al., 2011; Xu et al., 2016; Zhao et al., 2018), lanthanide-containing nanoparticles are increasingly being used for in vivo imaging. The size of nanoparticles can restrict certain applications as their diameters range from a few nanometers to several tens of nanometers, but their unmatched molar absorptivity and photostability are advantageous (Wu et al., 2015). Typically, lanthanides ions are embedded in a fluoride-based nanoparticle such as NaYF₄, NaYbF₄, NaLuF₄, and CaF₂, and a variety of crystalline shapes have been developed (e.g., sphere, rod, cube, or polyhedron). The surface of a lanthanide-doped nanoparticle can be modified to enable specific biological applications, such as metabolite sensing and targeted delivery (Bagheri et al., 2016; Gee and Xu, 2018).

The photophysics and chemical preparation of lanthanide nanoparticles have been reviewed elsewhere extensively and will not be discussed further here (Chen et al., 2014b; Dong et al., 2015; Wen et al., 2018; Wu et al., 2015). However, one property that makes lanthanide nanoparticles particularly useful for whole-animal imaging is the ability of certain subtypes to undergo photon upconversion. In this process, two or more photons of low energy are absorbed and converted into a single emission photon of higher energy. In other words, the emission wavelength is shorter than that of excitation light (“anti-Stokes emission”). Upconverting nanoparticles have begun to garner the attention of biomedical scientists for two reasons. First, the probes can be both excited and detected by near-IR photons, enabling high-contrast, high-resolution deep-tissue imaging since light attenuation, autofluorescence, and scattering are minimized. Second, upconversion allows near-IR irradiation to trigger photochemical reactions that would otherwise require UV or visible light. For instance, it can be exploited to site-specifically release caged drugs or generate cytotoxic chemical species (Wu et al., 2015).

Lanthanide dopants in nanoparticles facilitate photon upconversion via several different mechanisms, of which the two most common are excited-state absorption and energy transfer upconversion (Figure 3C) (Chen et al., 2014b; Dong et al., 2015; Wang and Liu, 2009). Excited-state absorption is a process that makes use of a single dopant species. It refers to successive absorption of excitation photons by a lanthanide ion (typically Nd³⁺, Ho³⁺, Er³⁺, and Tm³⁺) that has a ladder-like energy level structure. Successive absorption is promoted by the long lifetime of the intermediate excited state of the lanthanide dopants, and emission occurs from a high-lying excited state. In comparison, energy transfer upconversion utilizes two or more lanthanide species. In this process, excitation energy harvested by a “sensitizer” ion (donor) is passed on to the “activator” ion (acceptor). The transfer can occur twice or more from the same sensitizing ion or from multiple sensitizers to a single activator. Yb³⁺ is the most preferred sensitizer since it has only one excited-state level and absorbs strongly in the 960- to 990-nm window, which can be efficiency targeted by several conventional IR lasers. Yb³⁺ is often coupled with Ho³⁺, Er³⁺, or Tm³⁺ activators, and these acceptor ions generate violet-to-red wavelength photons from IR excitation, depending on the degree of upconversion (Figure 3D).

Figure 3. Luminescence resonance energy transfer.

(A) Energy transitions involved in direct lanthanide and LRET emissions. EC, extinction coefficient; E_TRANS, antenna-to-lanthanide transfer efficiency; E_LRET, LRET efficiency; and QY, quantum yield. E_LRET is equal to 1 − τ_+LRET/τ_−LRET. (B) Concentration- and R₀-dependent effects of an LRET acceptor on lanthanide luminescence. (donor emission lifetime of 1 millisecond and distance of closest approach of 0.5 nm) (C) Energy transitions of excited-state absorption and energy transfer upconversion. Energy transfer steps are indicated by the dashed arrows. (D) Energy levels of Tm³⁺, Yb³⁺, and Er³⁺, and potential energy transfer upconversion pathways. Adapted with permission from (Wang and Liu, 2009).

It is important to note that photon upconversion fundamentally differs from two-photon absorption, a non-linear optical process exploited for two-photon confocal microscopy. The latter requires simultaneous interaction of two photons with a chromophore while the former relies on metastable states between the ground and excited states that facilitate sequential energy absorption. In comparison, upcoversion is a linear optical process that is orders of magnitude more efficient.

COMPARISON WITH OTHER PHOTOLUMINESCENT PROBES

Fluorescent Proteins

Fluorescent proteins have become the work horses of biological imaging due to their genetic encodability and broad color palette. Just as protein engineers have extensively mutagenized fluorescent proteins to optimize their greater brightness and photostability, chemists have developed synthetic lanthanide chelates with photophysical properties that are ideal for biological imaging. The latest lanthanide probes have overall quantum yields (E_TRANS × QY_Ln, Figure 3A) that are comparable to those of red fluorescent proteins, which typically range from 20 to 50% (Piatkevich and Verkhusha, 2011) (Table S2). For example, when Tb³⁺ or Eu³⁺ is complexed with a TTHA chelate functionalized with a cs124 or cs124-CF₃ chromophore, the resulting chelate has an overall quantum yield of approximately 40% (Chen and Selvin, 2000b; Xiao and Selvin, 2001).

In addition, the extinction coefficients of lanthanide probes are similar to those of blue and cyan fluorescent proteins (Day and Davidson, 2009), defined primarily by the antenna moiety (Figure 1B and Table S2). Most antenna chromophores are aromatic polycycles, and widely used Tb³⁺ and Eu³⁺ sensitizers include bipyridines, 1,10-phenanthroline, terpyridines, tetraazatriphenylene, coumarins, xanthones, and acridones (Armelao et al., 2010; Heffern et al., 2014; Mathis and Bazin, 2010). Appending a single chromophore to the core chelate scaffold can achieve molar absorptivities of 10,000–50,000 M⁻¹ cm⁻¹.

While fluorescent proteins can be appended to a protein of interest by genetic fusion, lanthanide complexes require a chemical handle for in vivo protein tagging. For example, trimethoprim-, O⁶-benzylguanine-, or O²-benzylcytosine-functionalized lanthanide complexes have been used to label dihydrofolate reductate-, SNAP-, or CLIP-fused proteins (Emami-Nemini et al., 2013; Faklaris et al., 2015; Levoye et al., 2015; Pou et al., 2012; Rajapakse et al., 2010). For in vitro labeling of macromolecules such as antibodies, streptavidin, and oligonucleotides, standard coupling methods can be employed. Azide (Pillai et al., 2012), isothiocyanate (Li and Selvin, 1997; Takalo et al., 1997), maleimide (Ge and Selvin, 2004), and cyanuric chloride (Mukkala et al., 1993; Mukkala et al., 1996; Nishioka et al., 2006) moieties have been utilized for the bioconjugation of lanthanide chelates.

Other Long-Lived Photoluminescent Reagents

The defining feature of lanthanide luminescence is its long lifetime, which is 10⁵ to 10⁶-fold greater than that of conventional fluorescence. Other luminophores with emission lifetimes that exceed tens of nanoseconds include organic compounds that emit delayed fluorescence or room-temperature phosphorescence, transition-metal complexes, and quantum dots.

Organic compounds can generate >10-ns photoluminescence via three different mechanisms: (1) n-to-π* excitation, (2) excimer formation, and (3) structural suppression of T₁-to-S₀ nonradiative decay pathways. Several organic molecules that exploit these photophysical processes have been reported (Maltman et al., 2010; Nau and Zhang, 1999; Patsenker et al., 2011; Rich et al., 2013; Smith et al., 2004; Wawrzinek et al., 2013), but in general they make poor biological probes. n-to-π* fluorophores have low extinction coefficients as the transition is parity-forbidden. Small molecules that form excimers have extended, highly conjugated ring systems, which decrease water solubility and increase nonspecific binding in biological systems. Room-temperature phosphorescence requires a binding cavity in a protein, cyclodextran, or polymeric matrix to minimize the nonradiative inactivation of T₁ states. For these reasons, their usage in biomedical imaging has been restricted.

Phosphorescence from transition-metal complexes (e.g., Re⁺, Ru²⁺, Os²⁺, Ir³⁺, Rh³⁺, Pt²⁺, Au⁺, and Cu⁺) have much in common with lanthanide-based luminescence: long Stokes shifts, μs- to ms-scale emission lifetimes, and sharp emissions (Baggaley et al., 2014; Lo et al., 2012; Yam and Wong, 2011). However, transition-metal phosphorescence stems from a totally disparate electronic process, namely ligand-to-metal or metal-to-ligand charge transfer. Like organic phosphors, these heavy-metal complexes are sensitive to oxygen and solvent polarity, making them versatile sensors but poor tags for visualizing proteins or oligonucleotides. In addition, they typically carry multiple polypyridine ligands (2,2’-bipyridine, terpyridine, or 1,10-phenanthroline), which results in low water solubility and high cytotoxicity.

Quantum dots can be engineered to have long emission lifetimes by changing their size (Chen et al., 2014a; Zhang et al., 2016) or doping these nanoparticles with transition metals (e.g., Mn²⁺, Cu²⁺, or Ni²) or trivalent lanthanide ion (e.g., Tb³⁺, Eu³⁺, Er³⁺, Tm³⁺, Ho³⁺, or Yb³⁺) (Chen et al., 2015; Wu and Yan, 2013; Zhang et al., 2017). The former strategy can extend the emission lifetime of a quantum dot to a few hundreds of nanoseconds at best, whereas the latter approach enables μs- to ms-scale delayed photoluminescence. More recently, silicon-based quantum dots with emission lifetimes in the double-digit μs range have been developed (Tu et al., 2016; Yang et al., 2019), but their large size (>100 nm) is prohibitive for many biological applications.

Luminescence Resonance Energy Transfer

Lanthanide luminophores can participate in an energy transfer process analogous to FRET, which is referred to as luminescence resonance energy transfer (LRET) (Figure 3A). Common synthetic fluorophores and fluorescent proteins have been successfully adopted as LRET acceptor molecules, including cyanine, Atto, and Alexa Fluor dyes (Geissler et al., 2013), eGFP (Rajapakse et al., 2010), TagRFP (Bhattacharya et al., 2018), mCherry (Kubota et al., 2014), and allophycocyanin (Mathis, 1993). The Hildebrandt group and others have demonstrated that even quantum dots can be LRET acceptors (Charbonniere et al., 2006; Cui and Parker, 2016; Geissler et al., 2010; Hildebrandt et al., 2005).

LRET has several unique advantages over FRET. The Förster radius of a donor-acceptor pair (R₀, the distance at which energy transfer is 50% efficient) is proportional to (1) the one-sixth power of the overlap integral of the donor emission and acceptor absorbance (J^1/6) and (2) the one-sixth power of the donor quantum yield (QY^1/6). Lanthanide-based donors have a comparative advantage over their fluorescent counterparts in both areas. Due to their line-like emission peaks, lanthanide probes achieve J values that are about an order of magnitude greater than those of fluorescent donors (George Abraham et al., 2015). The quantum yield of a Tb³⁺ complex can exceed 70% while those of fluorescent proteins that emit in a similar spectral window (e.g., eYFP, Venus, mKO2, and mOrange) are 60–70% (Day and Davidson, 2009). Similarly, the quantum yield of a Eu³⁺ complex can be as high as 43%, and those of the spectrally matched mCherry, mRuby, and mKate2 proteins are 20–40% (Day and Davidson, 2009). These attributes enable LRET donor-acceptor pairs to have larger R₀ distances than their FRET counterparts; LRET R₀ values range from 5 to 10 nm, whereas FRET R₀ values are typically between 1.5 and 7 nm (Bajar et al., 2016; Müller et al., 2013).

R₀ values are also influenced by the relative orientations of the donor and acceptor transition dipole moments (κ²), which can range from 0 to 4. κ² is usually assumed to be 2/3, which occurs when the resonance energy transfer pair achieves complete rotational diffusion during the donor lifetime. Since the molecular tumbling and donor decay timescales are both in the low-nanosecond regime (Gáspári and Perczel, 2010; Ortega et al., 2013), this assumption is often invalid for FRET. In contrast, lanthanide donors can remain in their excited states for a few milliseconds, providing sufficient time for both the donor and acceptor to rotationally diffuse and yielding a κ² value very close to 2/3 (Selvin, 1996). The minimal R₀ stochasticity associated with LRET allows donor-acceptor distances to be determined with greater accuracy than FRET-based methods.

Finally, lanthanide donors enable “diffusion-mediated” resonance energy transfer. This process depends not only on R₀ but also on the diffusion coefficients of the donor and acceptor molecules, the acceptor concentration, and the distance of closest approach for the donor-acceptor pair. Since lanthanide luminophores are long-lived, they can approach within one Förster radius of an LRET acceptor during their excited state (Thomas et al., 1978). The diffusion rates of small metal complexes and organic fluorophores are typically in the 10⁻⁵ to 10⁻⁶ cm² sec⁻¹ range (Furukawa et al., 2007; Nitsche et al., 2004), and near-unity LRET efficiency is therefore predicted to occur when acceptor concentrations are in the low micromolar range and R₀ is greater than 6 nm (Figure 3B). This has been experimentally confirmed with Tb(DPA)₃³⁻/rhodamine B (Thomas et al., 1978) and Eu³⁺/ATBTA/Atto 610 (Cho et al., 2018). When lanthanide donors are paired with fluorescent proteins for diffusion-mediated LRET, high acceptor concentrations are required since the β-barrel scaffold blocks close access to the genetically encoded fluorophore.

Multiplexing Capabilities

The long photoluminescence lifetimes, narrow emission bands, long Stokes shifts, and LRET capabilities of lanthanide chelates also make these probes ideal for multiplexing. Spectrally distinct lanthanide complexes can be used in combination, achieving a multiplexing capacity that far exceeds that of conventional fluorophores. Two different strategies have been pursued: one approach makes use of temporal multiplexing (in combination with spectral multiplexing), while the other employs mixtures of lanthanides at predetermined ratios, taking advantage of their ultrasharp emission peaks.

Multiplexing through temporal gating was first demonstrated by the Selvin group (Chen and Selvin, 2000a) (Figure 4A). They labeled the 5’ ends of DNA oligonucleotides with either Tb³⁺-DTPA-cs124 or fluorescein and generated duplexes containing 6, 8, or 10 base pairs. Due to their varying LRET efficiencies, the labeled duplexes had luminescence lifetimes of 62, 266, and 540 μs, respectively. In addition, a 10-base pair duplex with Eu³⁺-DTPA-cs124 and Cy5 labels exhibited an emission lifetime of 250 μs. Their work established a means of kinetically identifying specific double-stranded DNA structures. In principle, the number of resolvable targets can be quadratically increased by integrating both time and wavelength domains.

Figure 4. Multiplexing using lanthanide luminophores.

(A) Length-dependent LRET emission lifetimes for DNA duplexes functionalized with a lanthanide chelate and acceptor chromophore. (B) Emission lifetimes of microspheres containing a fixed amount of lanthanide chelate and variable acceptor loading. Adapted from (Lu et al., 2014a) under a Creative Commons license. (C) Lanthanide nanoparticles can be optically differentiated by adjusting the ratios of three trivalent ions (Sm³⁺, Eu³⁺, and Dy³⁺). A scatter plot of the Dy³⁺/Eu³⁺ and Sm³⁺/Eu³⁺ luminescence ratios for 1,926 beads with 24 distinct spectral codes is shown. Adapted with permission from (Gerver et al., 2012).

A similar concept was recently realized by the Jin group using two different lanthanide materials. They encapsulated in porous polystyrene beads a fixed amount of Eu³⁺- (thenoyltrifluoroacetonate)₃ and a varying amount of red-adsorbing coumarin LRET acceptor (Lu et al., 2014a), creating a dozen types of lifetime-encoded microspheres (Figure 4B; τ = 188 to 359 μs). The Jin group also developed ‘τ-dots’ with varying photon upconversion efficiencies and therefore unique lifetime codes (Lu et al., 2014b). These NaYF4:Yb³⁺ nanoparticles were doped with differing concentrations of Tm³⁺ activator ion to achieve emission lifetimes ranging from 26 to 662 μs, enabling the generation of images with multiple life-time encoded layers.

The Fordyce group has pushed the limits of lanthanide probe multiplexing (Gerver et al., 2012; Nguyen et al., 2017) (Figure 4C). Using Eu³⁺, Dy³⁺, and Sm³⁺ nanoparticles, they synthesized polymeric beads that contain predetermined ratios of Dy³⁺:Eu³⁺ and Sm³⁺:Eu³⁺. The resulting 24 spectrally encoded microspheres were successfully distinguished by linear unmixing using six reference spectra. The group has since increased the library size to 550 barcodes by adding Ce³⁺/Tb³⁺ and Tm³⁺ nanoparticles to the combinations. This technology, termed MRBLE (Microspheres with Ratiometric Barcode Lanthanide Encoding), has been used to screen a peptide library for calcineurin ligands (Nguyen et al., 2019), and the Cyert group used the resulting leads to search in silico for potential calcineurin substrates in the human proteome (Wigington et al., 2019). Of 691 candidate proteins, 45 novel calcineurin targets have been experimentally confirmed, including several nuclear pore complex and centrosomal proteins.

BIOLOGICAL APPLICATIONS OF LANTHANIDE PROBES

Photometric Assays

Lanthanide luminophores first emerged as an alternative to radioisotopic labels in immunoassays. Using a first-generation, temporally gated photometer (Figure 5A), the Soini group demonstrated that a 400-μs delay between the sample excitation and detection allowed the selective collection of photons emitted from a Eu³⁺ complex (Soini and Kojola, 1983). Background autofluorescence was completely suppressed, and the detection sensitivity of their system matched that of ¹²⁵I-based radiography. Ultrasensitive photometric assays remain the most common application of lanthanide probes in the biosciences.

Figure 5. Lanthanide luminescence-based bioanalytical techniques.

(A) Time-resolved luminescence detection using pulsatile excitation and temporally gated photon collection. (B) Schematic representation of DELFIA. (C) Detection of GPCR internalization by diffusion-enhanced LRET. (D) Common design principles for lanthanide-based chemosensors.

DELFIA (dissociation-enhanced lanthanide fluorescence immunoassay) developed by the Hemmilä group is another popular technique that takes advantage of lanthanide emission lifetimes (Dickson et al., 1995; Hemmilä, 1985, 1988; Hemmilä et al., 1984; Ogata et al., 1992) (Figure 5B). This method is a variant of ELISA (enzyme-linked immunosorbent assay), in which a secondary antibody is tagged with a Eu³⁺ complex rather than a signal-amplifying enzyme such as horseradish peroxidase. After washing, Eu³⁺ ions are released from the antibody-chelate conjugates by acidification, and luminescence “enhancers” such as 2-naphthoyltrifluoroacetone and trioctylphosphine oxide are then added for the time-gated detection of Eu³⁺ luminescence. DELFIA achieves attomolar detection sensitivity and its compatibility with several trivalent lanthanide ions allows multiplexing. The technique has been primarily used to quantify hormones and cytokines in human blood samples, with a fg/mL detection limit (Zhang et al., 2014b). The sensitivity of DELFIA can be further enhanced by orders of magnitude by employing Eu³⁺ nanoparticles that can each release thousands of metal ions at low pH (Zhou et al., 2014). This approach has been used to detect carcinoembryonic antigen protein, a colorectal cancer marker.

Time-resolved LRET assays are another exciting example that demonstrates the experimental capabilities enabled by lanthanide-based probes (Bhattacharya et al., 2018; Cha et al., 1999; Emami-Nemini et al., 2013; Faklaris et al., 2015; Gonzalez et al., 2008; Kubota et al., 2017; Mathis, 1995). For instance, LRET is an attractive alternative to FRET for probing protein conformations. In both cases, the protein of interest is typically labeled at two different sites, one with a donor and the other with an acceptor luminophore, and their distance from each other can be determined by the donor-acceptor energy transfer efficiency. FRET measurements are limited by (1) the overlap between the donor and acceptor spectra and (2) donor-only or acceptor-only species that can arise from incomplete bioconjugation. However, LRET emissions can be detected exclusively by combining spectral and temporal filtering (Chen and Selvin, 2000a; Cho et al., 2018; Selvin and Hearst, 1994; Selvin et al., 1994; Vereb et al., 1998; Xiao and Selvin, 2001). Furthermore, the luminescence lifetime of the LRET emission can be directly converted to LRET efficiency (E_LRET = 1 − τ_sensitized / τ_donor-only). For example, the Selvin and Bezanilla groups structurally characterized the Shaker potassium channel by modifying individual subunits with either Tb³⁺-DTPA-cs124 or fluorescein and monitoring the resulting LRET signals (Cha et al., 1999). The Shaker pore has four-fold rotational symmetry, and by fitting the LRET emission decay to a bi-exponential curve, they were able to measure inter-subunit distances and detect voltage-dependent changes in pore structure. A similar approach was used to investigate the actin-dependent conformational change of smooth muscle myosin protein (Xiao et al., 2003), demonstrating that the myosin lever arm swings only in the presence of both actin and ADP (Whittaker et al., 1995)

The ability of lanthanide chelates to undergo diffusion-mediated LRET has also been exploited to study protein function. For instance, the Durroux and Bachelerie groups developed an LRET-based photometric assay for quantifying ligand-induced internalization of G protein-coupled receptors (GPCRs), a process that regulates the de- and re-sensitization of these signaling proteins (Levoye et al., 2015) (Figure 5C). In this system, a SNAP tag is genetically fused to an extracellular region of the GPCR of interest and then covalently labeled with cellimpermeable Tb³⁺-Lumi4-O⁶-benzylguanine. GPCR ligands are then co-administered with fluorescein as a soluble LRET acceptor. Diffusion-mediated LRET between surface-localized GPCRs and the extracellular fluorescein gives rise to long-lived luminescence signals; however, once internalized, Tb³⁺ complex-labeled GPCRs cannot participate in this process. Thus, time-gated measurements of Tb³⁺-Lumi4 and fluorescein signals could be used to determine the degree of GPCR internalization in response to various ligands. Diffusion-mediated LRET also has been employed to determine the transmembrane depth of bacteriorhodopsin-bound retinal (Leder et al., 1989) and to quantify the lateral diffusion of membrane proteins (Lan et al., 2015).

Microscopy

Time-resolved luminescence microscopy employs the same detection principles as the photometric assays described above (Figure 5A). First-generation microscopes for time-resolved lanthanide imaging relied on xenon flashlamps for pulsed excitation, and temporal gating was achieved with mechanical chopper blades (Beverloo et al., 1990; Marriott et al., 1994; Seveus et al., 1992), a cost-effective approach that is still employed today (Jin and Piper, 2011; Zhang et al., 2014a). For example, the Jin group successful imaged Eu³⁺-labeled Giardia lamblia cysts and Cryptosporidium parvum oocysts in aqueous media with significant amounts of fluorescent background. A 5-μs delay between sample excitation and emission acquisition completely removed autofluorescence signals and allowed visualization of long-lived lanthanide luminescence within the microorganisms. Chopper-based microscopes also have been used to image lanthanide luminophores in Daphnia magna, Caenorhabditis elegans, and Danio rerio (Ma et al., 2017; Ma et al., 2018; Song et al., 2015; Zhu et al., 2016).

Other groups have built time-resolved microscopes with intensified charge-coupled device (ICCD) cameras which allow electronical time-gating and obviate the need of mechanical choppers (Beeby et al., 2000; Cho et al., 2018; Hanaoka et al., 2007; Rajapakse et al., 2010). For example, the Miller group used this system and a UV LED to visualize interactions between ZO-1 and Claudin-1, proteins that are involved in the formation of tight junctions between cells. The two proteins were fused to dihydrofolate reductase and GFP, respectively, and the former was labeled with trimethoprim-functionalized Tb³⁺-Lumi4. ZO-1/Claudin-1 binding was then confirmed by LRET between Tb³⁺-Lumi4 and GFP. In addition to xenon flashlamps and LEDs, pulsed lasers have been adopted as a light source for time-resolved lanthanide microscopy (Afsari et al., 2016; Beeby et al., 2000; Chen et al., 2018; Cho et al., 2018). For example, the Huang and Hildebrandt groups modified the surfaces of 6- and 12-nm quantum dots with either a Eu³⁺ or Tb³⁺ complex, generating LRET pairs with distinct emission lifetimes (Chen et al., 2018). The nanoparticles were used to label four different populations of HeLa cells, which were then mixed and imaged with a time-resolved microscope equipped with an ICCD camera and 349-nm pulsed laser. The differentially labeled HeLa cell populations could be readily discerned by their luminescence lifetimes, and in theory, the multiplexing capacity of this system could be increased by several fold employing quantum dots that emit at different wavelengths.

These advances in time-resolved microscopy have also fostered the development of lanthanide-based sensors for biological imaging. Chemosensors for cations, anions, pH, and metabolites have been developed steadily since the 1990s, but their applications were restricted to photometric assays until the 2010s. The two most common approaches for designing lanthanide chemosensors are to modulate lanthanide ion hydration or antenna activity (Aulsebrook et al., 2018; Shuvaev et al., 2017) (Figure 5D). The Yuan group has shown that these probes can be used to in conjunction with time-resolved luminescence microscopy to detect various biomolecules in cells and small organisms, such as Daphnia magna, Caenorhabditis elegans, and Danio rerio. For instance, lanthanide-based probes have been developed to image reactive oxygen species (Ma et al., 2018; Sun et al., 2015; Tang et al., 2019b), nitric oxide (Dai et al., 2017; Tian et al., 2015), hypochlorous acid (Liu et al., 2017; Ma et al., 2017), ascorbic acid (Song et al., 2015), Zn²⁺ (Ye et al., 2014), glutathione (Song et al., 2019), and carbon monoxide (Tang et al., 2019a). Since trivalent lanthanide ions have also been used as magnetic resonance imaging contrast agents (Heffern et al., 2014), certain lanthanide-based sensors could be used for both cell and whole-body imaging. The glutathione sensor, which was used for optical imaging in HeLa cells and zebrafish and for magnetic resonance imaging in mice, exemplifies this paradigm and demonstrates the versatility of lanthanide probes.

Despite their unique photophysical properties, lanthanide chelates are still not widely utilized for biological imaging, primarily due to their low photon efflux (Gahlaut and Miller, 2010; Jin et al., 2007a, b; Jin and Piper, 2011; Rajendran and Miller, 2015; Soini et al., 2003). It typically takes ~100,000-fold longer to collect the same number of photons from a lanthanide luminophore than from a fluorophore of the same brightness. Photon efflux is limited further by the pulsatile LED illumination typically employed for time-resolved microscopy, which excites <0.1% of the lanthanide probes per pulse (Cho et al., 2018). As a result, time-resolved lanthanide microscopy can require tens of seconds to acquire a single image with sufficient pixel intensities. Lanthanide imaging has also been constrained by the long-lived photoluminescence that is intrinsic to common microscope optics (Cho et al., 2018; Gahlaut and Miller, 2010; Rajendran et al., 2014) (Figure 6A). Although fluorescent emissions can be suppressed by a microsecond-scale delay, optics-derived background has lifetimes comparable to that of lanthanide luminescence and thus cannot be efficiently filtered. The undesired photoluminescence originates from UV-excitable materials in objective lenses, and it significantly degrades the sensitivity of lanthanide microcopy. Utilizing high-UV-transmission objectives or narrowband emission filters enhances signal-to-noise ratios by only a few fold (Cho et al., 2018; Rajendran et al., 2014). Together, these factors have prevented lanthanide imaging from recapitulating the detection sensitivities achieved in photometric assays.

Figure 6. Time-resolved luminescence microscopy.

(A) Micrographs of zebrafish embryos injected with Eu³⁺-DTBTA-dextran obtained by steady-state and time-resolved microscopy. The optics-derived photoluminescence (e.g., region within the dashed circle) overlaps spectrally and temporally with lanthanide emissions. (B) Optical paths of epi-illumination (left) and trans-reflected illumination (right) microscopy. (C) Comparison of conventional and LRET-enhanced time-resolved imaging. The integrated emission intensity (IE) for each condition is shown. (D) Suppression of optics-derived photoluminescence by trans-reflected illumination. (E) Comparison of steady-state immunofluorescence and trLRET microscopy, demonstrating the greater detection sensitivity afforded by lanthanide probes. Zebrafish embryos stained with an anti-MYH1E antibody labeled with either Alexa Fluor 594- or Eu³⁺-DTBTA are shown. Scale bars: 200 μm. Adapted with permission from (Cho et al., 2018)

We recently overcame these limitations with a new lanthanide imaging modality, trans-reflected illumination (Figure 6B) with luminescence resonance energy transfer (trLRET) (Cho et al., 2018). In trLRET, low lanthanide photon efflux is addressed by using an LRET acceptor that reduces luminescence lifetimes from the millisecond to microsecond regime. Lanthanide-based emission rates increase proportionally with LRET efficiency, permitting imaging cycle rates that are at least 40-fold higher than that of lanthanide donors alone (Figure 6C). Our imaging platform also utilizes a Q-switched laser (QSL) source rather than an LED, enabling 40% of the lanthanide probes to be excited in a single nanosecond-scale pulse. In combination, these two methods greatly enhance lanthanide photon flux.

To mitigate optics-derived background signals, trLRET employs trans-illumination rather than epi-illumination. This optical configuration allows a TiO₂-coated glass insert to be placed in the excitation light path. Since the TiO₂ coating attenuates UV light by 100,000-fold, this optical configuration minimizes the excitation of photoluminescent materials in the microscope objectives and can increase the signal-to-noise ratio by 75-fold (Figure 6D). trLRET microscopy has been used to image endogenous proteins in zebrafish embryos with at least an order of magnitude greater sensitivity than standard epifluorescence microscopy (Figure 6E) (Cho et al., 2018). trLRET can also enable the visualization of molecular interactions in vivo when the resonance energy transfer is predominantly proximity-dependent rather than diffusion-mediated (Cho et al., 2018).

Deep-Tissue Imaging

In addition to cells and embryos, deep tissues can be imaged with photoluminescent lanthanides. One approach utilizes lanthanide-based nanoparticles that are capable of photon upconversion, taking advantage of the paucity of biological molecules that can undergo this photophysical process. Since the upconverted signals have shorter wavelengths than the excitation light, they can be spectral filtered of biological autofluorescence with near-complete efficiency. The Li group showed that nanocrystals doped with Gd³⁺, Yb³⁺, and Tm³⁺ can be imaged in mice with a penetration depth of approximately 2 cm using 980-nm excitation and 800-nm detection (Liu et al., 2011). Similarly, the Prasad and Han groups showed that Tm³⁺-containing nanoparticles can be imaged in mice with negligible background signals (Figure 7A), and the lanthanide probes can even be detected through 3 cm of porcine tissue (Chen et al., 2012).

Figure 7. Photon upconversion and deep-tissue imaging using lanthanide luminophores.

(A) Whole-animal imaging of a mouse injected via the tail vein with a Tm³⁺-doped photon-upconverting nanoparticle (excitation: 980 nm; emission: 800 nm). Adapted with permission from (Chen et al., 2012). (B) Abdominal imaging of a mouse that has been orally administered Yb³⁺, Nd³⁺, and Tm³⁺-containing nanoparticles, which localize initially to the stomach. Micrographs obtained with or without a time delay are shown. Adapted from (Tan et al., 2018) under a Creative Commons license. (C) Imaging of a mouse after intracerebral injection of an Yb³⁺- and Nd³⁺-doped nanoparticle, demonstrating the removal of eye autofluorescence by temporal gating. Adapted from (Tan et al., 2018) under a Creative Commons license. (D) Temporally resolvable τ-dots can be used for multiplexed deep-tissue imaging. Adapted with permission from (Fan et al., 2018).

Lanthanide nanoparticle imaging can be further enhanced by time gating, as their luminescence lifetimes (tens to hundreds of microseconds) are orders of magnitude longer than that of sample autofluorescence. This mode of detection is compatible with both upconverting and downconverting probes. For example, the Jin group showed that a microsecond-scale delay facilitates the visualization of Yb³⁺- and Tm³⁺-doped nanoparticles that are subcutaneously injected into mice (Zheng et al., 2016). More recently, the Rodriguez and Chen groups synthesized lanthanide ion-doped NaGdF₄ and NaYF₄ nanoparticles with long emission lifetimes that can be excited with an 808-nm laser, thereby averting thermal damage to the biological specimen (Tan et al., 2018). The NaGdF₄ nanoparticle was orally administered to mice, and immediately after, its presence in the stomach was confirmed by time-resolved imaging (Figure 7B). Furthermore, the NaYF₄ nanoparticle was injected into the murine brain and its luminescence could also be detected with high contrast using temporal gating (Figure 7C). These examples illustrate how lanthanide photophysics enables greater contrast and penetration depth for deep-tissue imaging.

Lanthanide-based nanoparticles can also enable deep-tissue imaging with unprecedented levels of spatial resolution. This is achieved by targeting the near IR-IIb spectrum (1,500–1,700 nm) where light scattering is significantly diminished (Ding et al., 2018). Until very recently, there were only a handful of probes compatible with this optical window (Diao et al., 2015; Zhang et al., 2018), but several Er³⁺-doped nanoparticles with these spectral properties have been developed (Kamimura et al., 2017; Xue et al., 2018; Zhong et al., 2017). The 1,550-nm Er³⁺ emission affords a spatial resolution of a few tens to hundreds of micrometers, which is sufficient to visualize major blood vessels in live mice. Of note, the Piper and Zhang groups have generated Er³⁺-based nanoparticles with varying luminescence lifetimes (microsecond to millisecond) by controlling the energy relay structure (Fan et al., 2018). By conjugating these lifetime-encoded probes to antibodies against surface cancer markers, they were able to simultaneously image multiple antigens on millimeter-sized tumor xenografts in mice (Figure 7D). In principle, this multiplexing technology could be adopted for imaging targets in deeper tissues.

FUTURE FRONTIERS IN LANTHANIDE LUMINESCENCE

The unique photophysical properties of lanthanide-based probes can be exploited in multiple ways to advance our understanding of biological processes. The application of these metal-based luminophores was once limited to photometric assays, but they are now emerging as versatile tools for biological imaging. We anticipate that the exceptional sensitivity of time-resolved lanthanide luminescence microscopy will be increasingly exploited to visualize low-abundance biomolecules in live cells and animals. Conventional fluorescence microscopy can detect low nanomolar concentrations, which corresponds to a few thousand molecules in a typical mammalian cell. In comparison, the median copy number of mRNAs in mammalian cells is 17 (Schwanhausser et al., 2011). To overcome this limitation, current RNA imaging methods such as MS2 (Tutucci et al., 2018) or tDeg (Wu et al., 2019) tag recruit several fluorescent proteins to a target; however, this may affect the localization, stability, and function of the transcript. Lanthanide luminophores may afford a general solution for the live imaging of transcripts, as these molecular probes are small in size and their luminescence can be spectrally and temporally resolved from cellular and tissue autofluorescence. We further expect LRET probes will gain popularity as tools for visualizing molecular interactions in living systems. Compared to FRET, LRET enables: (1) longer R_o distances, (2) more accurate distance measurements, and (3) far greater signal-to-noise ratios. For example, LRET could significantly enhance the FRET-based mRNA imaging method developed by the Tyagi group, which utilized short oligonucleotides to juxtapose donor and acceptor molecules upon binding to the targeted transcript (Bratu et al., 2003).

As with any technology, lanthanide-based imaging has room for improvement. On the probe side, the development of less hydrophobic lanthanide chelates is necessary to avoid background from non-specific binding. Appending sulfonic acid groups could render these probes better suited for biological systems as in the case of Alexa Fluor dyes. On the other hand, cell-penetrating probes is also highly desirable. While several lanthanide chelates have been shown to cross the plasma membrane, they are typically sequestered to one or two organelles (Montgomery et al., 2009). A membrane-permeable probe with even distribution throughout the cell is needed. Furthermore, most lanthanide luminophores are excited by UV-A light, and chelates that can be excited by visible light (e.g., 405-nm lasers) could reduce phototoxicity and enable imaging modalities that utilize epi-illumination. While the Yuan group have demonstrated that the BPT antenna (2-(N,N-diethylanilin-4-yl)-4,6-bis(pyrazol-1-yl)-1,3,5-triazine) can be excited at 420 nm in an aqueous solution, it requires an auxiliary chelator (Ma et al., 2018; Tian et al., 2014). Combining these moieties into a unimolecular chelate could yield lanthanide probes that are suitable for wide range of biological applications.

On the instrumentation side, integrating time-gated detection and light-sheet illumination is a logical next step. Light-sheet illumination not only averts long-lived optics background like trans-reflected illumination but also enables optical sectioning of the specimen. Time-resolved light-sheet microscopy would in principle produce images free of biological autofluorescence, optics background, and out-of-plane signals. An imaging platform that can spectrally and temporally unmix lanthanide emissions in real time would also be of great use.

Finally, lanthanide-based optical imaging could be combined with other detection modalities. For example, Ce³⁺ and Pr³⁺ complexes of DTPA-(3,3’-diaminobenzidine)₂ have been used for multiplexed electron microscopy (Adams et al., 2016), and analogous probes that utilize brighter trivalent lanthanide ions and 3,3’-diaminobenzidine derivatives with greater molar absorptivity might enable correlative light-electron microscopy. Similarly, lanthanide isotope-labeled antibodies developed for mass spectrometry imaging (Angelo et al., 2014; Giesen et al., 2014) could be modified for optical imaging by adding an antenna group to the DTPA or DOTA chelate. Luminescent lanthanide probes can also double as magnetic resonance imaging contrast agents (Heffern et al., 2014; Song et al., 2019). Multimodal lanthanide imaging will enable us to visualize biological samples at wide-ranging spatial resolutions with the same probe set.

These and other future advances will help establish lanthanide probes as standard components of the biologist’s toolbox, enabling new experimental approaches to long-standing problems. Translating lanthanide chemistry into biological discovery will continue to require contributions from multiple fields, including chemical synthesis, photophysics, and microscopy. Chemical biologists will likely be central to these efforts, as they can bridge the development of new lanthanide technologies and their application to cellular and organismal systems.