Genetically encoded fluorescent sensors for metals in biology

Abstract

Metal ions intersect a wide range of biological processes. Some metal ions are essential and hence absolutely required for the growth and health of an organism, others are toxic and there is great interest in understanding mechanisms of toxicity. Genetically encoded fluorescent sensors are powerful tools that enable the visualization, quantification, and tracking of dynamics of metal ions in biological systems. Here, we review recent advances in the development of genetically encoded fluorescent sensors for metal ions. We broadly focus on 5 classes of sensors: single fluorescent protein, FRET-based, chemigenetic, DNAzymes, and RNA-based. We highlight recent developments in the past few years and where these developments stand concerning the rest of the field.

Graphical Abstract

Introduction

This review focuses on genetically encoded fluorescent sensors for studying metals in biology. A close examination of the term “metal” reveals that this term encompasses many elements. Metals are frequently defined either by their physical properties (electrical conductivity, malleability) or propensity to form cations and engage in metallic bonding. Figure 1 shows the periodic table, emphasizing groups that are defined as metals and highlighting subsets of metals that frequently intersect biology. These metals typically fall into three general classes: metals that are considered essential for biological organisms, metals that are toxic to biological organisms¹, and metals that can serve to modulate biology along with metals used in medicine², even if they are not considered essential. For example, cisplatin is a platinum compound that is widely used as a chemotherapy agent due to its ability to inhibit DNA replication.³ There is also a fourth category of metals which are considered to be environmental contaminants. Here, fluorescent sensors are expressed in a biological organism (such as bacteria or yeast) to detect metals in environmental samples. Heavy metal pollutants and electronic waste can give rise to many metals contaminating the environment.^1,4

Figure 1. Classification of metals from the periodic table.

A) Periodic table classifying metals in the following categories: alkali metals (light green), alkaline earth metals (green), transition metals (dark green), lanthanide metals (turquoise), actinide metals (aqua), nonmetals (light blue), metalloid (blue), post transitional metals (dark blue) and halogen (teal). B) List of different categories of metals: common essential metals, metals widely recognized as toxic, metals that can modulate biology and/or metals used in medicine, and metals that are frequently detected in polluted environments. While most metals are toxic at high doses, the ones listed as toxic have a very high degree of toxicity and are often considered a threat to public health. Created with BioRender.com.

The largest collection of genetically encoded sensors exists for the group II alkaline earth metal calcium. Calcium has long been studied as a second messenger in signal transduction as calcium dynamics influence a host of cellular processes.⁵ An overview of the current state of genetically encoded calcium indicators, often referred to as GECIs, has been reviewed recently.⁶ In the last few years there have been notable developments in sensors for magnesium⁷, and the group I alkali metals sodium⁸ and potassium⁶. These metals are present at much higher concentrations than calcium; free metal concentrations are ~0.5–1 mM for magnesium, ~10 mM for sodium, ~100 mM for potassium compared to ~100 nM for calcium.⁹ These ions play important roles in neutralizing the charge of nucleic acids and ATP (particularly for Mg²⁺), maintaining membrane potential, and osmolarity of cells. There is evidence that dysregulation of previously mentioned metals can lead to pathological conditions, which motivates the development of fluorescent sensors to quantify these metals in healthy and diseased cells.¹⁰

There has also been significant interest in fluorescent sensors for essential transition metals such as manganese, iron, copper, and zinc. While manganese, iron, and copper have multiple biologically accessible oxidation states, most interest is focused on Mn²⁺, Fe²⁺, and Cu¹⁺, given the reducing environment of the cytoplasm. Mn²⁺ and Fe²⁺ have historically been challenging targets due to their tendency to form weaker complexes, according to the Irving-Williams series.¹¹ For Mn²⁺, Fe²⁺, and Cu¹⁺, there has been more progress in the development of small molecule fluorescent sensors than in the realm of genetically encoded sensors¹², particularly in the area of activity-based sensors that leverage metal-catalyzed chemistry.¹³ However, the recent report of a genetically encoded FRET-based sensor for Mn²⁺ based on a reengineering of a lanthanide-binding protein indicates that it is possible to overcome expectations of the Irving-Williams series.¹⁴ The concentration of labile transition metal ions is many orders of magnitude lower than other ions such as calcium, magnesium, sodium, and potassium so there is an important concern regarding buffering and perturbation of cellular pools of these metal ions. The largest class of genetically encoded transition metal sensors is for Zn²⁺, as described below and reviewed in.¹⁵

In this review, we define a sensor as genetically encoded if at least one component of the sensor can be encoded by DNA and incorporated into biological organisms or cells via transformation, transfection, viral transduction, or genome editing. These sensors can be comprised of DNA, RNA, peptides, or proteins. Figure 2 provides an overview of the different kinds of genetically encoded sensor platforms, which are further described below.

Figure 2. Summary of genetically encoded metal ion senor platforms.

A) Schematic of a representative cpFP sensor platform where metal binding to the sensing domain changes the pKa of the chromophore and induces an increase in fluorescence intensity. B) Schematic of a representative FRET sensor platform where metal binding to the sensing domain induces a conformational change in the sensor which alters the efficiency of energy transfer between CFP and YFP, Typically, metal binding causes an increase in FRET and a decrease in donor emission. C) Representative chemigenetic sensor platform where metal binding changes the intensity of the HaloTag dye. D) DNAzyme sensor platform used for live cell imaging, where metal-catalyzed cleavage of mRNA encoding a fluorescent protein leads to a decrease in fluorescence intensity of one FP. E) Generalized cartoon of an RNA-based metal ion sensors, where metal binding promotes folding of a fluorogenic RNA aptamer, enabling the fluorophore DFHBI-1T to bind to the Spinach or Broccoli aptamer. Note, for cpFP and FRET sensor platforms, the metal binding stoichiometry varies for different sensors. For example, some sensors bind 1 metal ion, some bind 2, some bind 4. However, the principle of metal detection remains the same. Created with BioRender.com.

Main text

Single fluorescent protein sensors

A popular genetically encoded sensor platform utilizes one fluorescent reporter protein and a protein or peptide sensing domain that serves to bind an analyte, for example, a metal ion. The vast majority of sensors within this platform use an intrinsic fluorescent protein (FP) as the reporter protein, but there are examples of sensors based on CreiLOV^16,17 and near infrared (NIR) fluorescent proteins^18–20 as well. CreiLOV and NIR fluorescent proteins bind a cofactor, flavins or biliverdin, respectively. The FP is usually circularly permuted (cpFP), meaning the protein is engineered with new N- and C-termini closer to the chromophore.²¹ The protein or sensing domain is introduced close to the chromophore so that when a metal ion binds to the sensing domain, it changes the microenvironment of the chromophore and influences the intensity of the fluorescence.²¹ Usually, an increase in fluorescence is observed after a binding event (Fig 2a). This sensor design was popularized for Ca²⁺ with the development and optimization of GCaMP, GECO, and jRCaMP.^22–24 Most cpFP sensors are based on green fluorescent protein (GFP) or yellow fluorescent protein (YFP). It has been harder to generate robust cp scaffolds for red FPs because of problems with low brightness, photo switching and chromophore maturation.²⁵ This platform design has been used to generate fluorescent sensors for the following metal ions: Zn^{2+ 15}, Ca^{2+ 6}, Cu^{1+ 26}, Cu^{2+ 27} and K^{+ 28}. Some notable advances in the past few years include the following. There have been significant advances in increasing the kinetics of the Ca²⁺ indicator jGCamP8 for neural activity reporting²⁹. Engineering efforts to improve the sensitivity of the single FP Zn²⁺ sensor platform led to GZnP3, which reveaed that Zn²⁺ can be released from endolysosomal vesicles in primary hippocampal neurons via TRPML1. ³⁰ There has been a flurry of activity leading to the development of multiple single FP sensors for K⁺.^28,31 Of these, it is notable that GINKO2 was used to detect K⁺ in bacteria, plants, and mice.³²

The main advantages of single FP sensors are: high dynamic range and the opportunity to multiplex with sensors of different colors to monitor multiple analytes simultaneously. The main disadvantage is the need to ensure that the sensing capabilities are not influenced by changes in intracellular pH, as the sensing mechanism involves modulation of the pKa of the chromophore.³³ An additional limitation of single FP sensors is that the readout typically involves change in fluorescence intensity (increase or decrease) and thus, such intensiometric sensors are sensitive to the sensor concentration, movement artifacts, and sample thickness. Some intensity-based sensors show a shift in the excitation spectrum upon metal binding and hence can be used as a ratiometric sensor by collecting the fluorescence intensity at two different excitation wavelengths.

FRET-based sensors

The largest sensor platform is based on Föster Resonance Energy Transfer (FRET) between a donor FP and an acceptor FP (Fig 2b). A protein or sensing domain is introduced strategically so that a binding event can either increase or decrease FRET efficiency. Prominent and widely used examples of FRET-based sensors are YC Cameleons, D-family cameleons, TN-XXL and iGECI for Ca²⁺⁶, and ZapCY, ZapCV and eCALWY for Zn²⁺.¹⁵ eZinCh2 is another prominent Zn²⁺ sensor platform, although it has been shown to form oligomers in an oxidizing environment such as the ER, so use of this platform should be restricted to the cytosol.³⁴ The majority of these sensors use variants of cyan fluorescent protein (ECFP, cerulean or turquoise) and yellow fluorescent protein (citrine, Venus and cpVenus). FRET sensors have also been developed for the following metal ions: Mg^{2+ 35}, K^{+ 36},Cu^{+ 37}, Co^{2+ 38},Ni^{2+ 39}, Fe^{2+ 40}, Ag^{+ 41}, Au⁺⁴², Mo^{+ 43},Mn^{2+ 14,40},La^{3+ 44}, As^{3+ 45}, Pb^{2+ 46}and Hg^{2+ 47}. The primary application of these FRET-based sensors is in cultured cells, as CFP and YFP are not particularly useful for in vivo imaging due to high background, poor tissue penetration, and scattering of the high energy excitation and emission. These limitations can be overcome by using fluorophores in the near infrared (IR) or by using bioluminescent proteins. Both approaches have been adopted for Ca²⁺ sensing. A recently engineered a near infrared genetically encoded Ca²⁺ indicator (iGECI), based on miRFP670 and miRFP720, was developed for in vivo Ca²⁺ imaging⁴⁸, and there are multiple sensor designs based on the bioluminescent NanoLuc luciferase protein.^49,50 These tools offer insight into future engineering directions if more biosensors are to be translated to in vivo imaging platforms.

The main advantage of a FRET sensor is that the ratiometric readout can be used for quantification of the target metal ion⁵¹, and contrary to single FP sensors, the sensor can have a bright fluorescence signal even in the absence of the target metal. The disadvantages of FRET-based sensors are that their dynamic range tends to be lower than single FP sensors leading to lower sensitivity, and they take up more spectral bandwidth so it can be challenging to multiplex with other sensors.

Chemigenetic sensors

A hybrid platform that involves a genetically encoded element coupled with a small molecule probe is rapidly gaining popularity (Fig 2c). While there is quite a bit of diversity in the design of different sensors, the genetically encoded element is typically a protein that reacts with a specific small molecule ligand; common examples are HaloTag⁵², SNAP-Tag⁵³ and FAST⁵⁴ which react with chloroalkane, benzylguanine, and hydroxybenzylidene rhodamine analogs, respectively. The metal binding component can either be derived from proteins (as in the case of HaloCaMP⁵²) or synthetic chelators (as in the case of HaloGFP-Ca1–4 and HaloGFP-Na1–14)⁵⁵. To be a sensor, metal binding needs to change the intensity of the fluorescence read-out. One example of how this can be accomplished is HaloCaMP, which consists of a cpHaloTag fused to two Ca²⁺ sensing domains (Calmodulin and a peptide from myosin light chain kinase); this is the part of the sensor that is genetically encoded. Then, a fluorogenic cell permeable JaneliaFluor(JF)-HaloTag ligand is added to the cells. The key to the fluorescent sensing mechanism is that the JF dye exists in an equilibrium between a lipophilic, non-fluorescent form and a polar, fluorescent zwitterion.⁵⁶ What pushes the equilibrium towards fluorescence, for HaloCaMP, is the binding of Ca²⁺ ions. A chemigenetic sensor based on split-FAST (Fluorescence-Activating and absorption-Shifting Tag) has also been developed.⁵⁴ One of the earliest examples of this class used SNAP-tag reacted with a fluorescent zinc sensor (ZP1) conjugated to benzylguanine⁵³, and a similar approach has been used to create chemigenetic sensors for Ca^{2+ 52,55}, K^{+ 57}, and Mg²⁺.⁵⁸ Finally, the newest sensors in this platform use a cpGFP inserted into HaloTag and a synthetic metal chelator (for either Ca²⁺ or Na⁺) conjugated to a HaloTag-reactive chloroalkane. The chelator is positioned close to the chromophore of cpGFP and metal binding shifts the excitation spectrum, giving rise to an excitation ratiometric sensor.⁵²

The primary advantage of chemigenetic sensors is that they can leverage the strengths of small molecule fluorophores (brightness, photostability, chemical properties) with proteins (genetic encodability). Because of the spectral diversity of small molecule fluorophores and narrow excitation and emission bandwidths compared to FPs, chemigenetic sensors are likely to be particularly useful for multiplexing.

Genetically encoded sensors based on nucleic acids

The final class of genetically encoded sensors is based on nucleic acids: DNAzymes and RNA-based sensors. DNAzymes are DNA molecules that can perform catalytic reactions with specific metal ions as cofactors. This specificity can be exploited for a sensing functionality of specific metal ions in environmental samples, point-of-case diagnostics, cellular imaging, and in vivo imaging.⁵⁹ DNAzymes for environmental detection and point-of-care diagnostics typically exploit metal-catalyzed, DNA-mediated cleavage of fluorescently tagged molecular beacons. Because DNAzyme sensors are activity-based, specificity for the target metal is achieved by selecting DNA sequences that react more quickly and at a lower concentration than a competing metal. Some sensors show very strong selectivity (> 10⁵-fold for a competing metal) whereas others are more modest (~ 4-fold for a competing metal).⁵⁹ For cellular and in vivo imaging, DNAzyme activity is used to regulate the expression of FPs via mRNA cleavage, rather than a binding event regulating the fluorescence intensity of an FP. An example of this sensor platform is the expression of Clover2 (GFP variant) and mRuby2 (RFP variant) for detection of Mg²⁺. In the presence of excess Mg²⁺, a Mg-driven DNAzyme catalyzes a decrease in the expression of Clover2, while mRuby2 expression is unaffected, creating a ratiometic sensor⁶⁰ particularly powerful application of this technology was the detection of tightly bound Zn²⁺ in blood samples.⁶¹ One limitation of the DNAzyme-based sensors for cellular imaging is the slow response time and irreversibility of the response. DNAzyme sensors have been developed for a wide range of metals including: Cu⁺¹, Ca²⁺, Ag⁺, Hg²⁺, Zn²⁺, Cd²⁺, Cr³⁺, Pb²⁺, Na⁺, Th⁺, Li⁺, and Mg²⁺.⁶⁰

The discovery of metalloregulatory riboswitches⁶² coupled with the recent development of RNA-based sensors⁶³ opens the possibility of designing RNA-based sensors for metal ions. Riboswitches are naturally evolved RNA elements that bind small molecule metabolites (or metal ions). They are highly prevalent in microbes and are broadly used to regulate gene expression. The creation of RNA-based sensors typically involves linking a riboswitch to a fluorogenic RNA aptamer such as Broccoli. There is currently one example of an RNA-based sensor for a metal ion that involves engineering an Ag⁺ binding site into the Broccoli fluorogenic RNA aptamer.⁶⁴ This sensor was applied to image Ag⁺ in bacterial cells. The czcD metal-binding riboswitch was also converted to a metal sensor by fusion with the fluorogenic RNA aptamer Spinach2; this sensor detects Fe²⁺, Mn²⁺, Co²⁺, Ni²⁺, and Zn²⁺, but does not distinguish between these metals.^{55, 65}

State of the field

With the wide array of genetically encoded sensors^6,15,25,51, it can be challenging for end-users to identify the best tool for a chosen application. Table 1 summarizes the most widely used sensor platforms for common metal ion targets. These sensors are typically designed to detect the labile or accessible pool of metals. In organisms and in cells, metal ions exist in multiple states: tightly bound to proteins and other biomolecules or hydrated, loosely bound and accessible. While the intent of most researchers is to measure the labile or accessible pool, it is important to evaluate whether sensor expression perturbs this labile pool by measuring the metal concentration as a function of sensor concentration^66,67 and/or using multiple sensors with different metal ion affinities.⁶⁸ Genetically encoded sensors for metal ions have been applied in bacteria, yeast, plants, and mammalian cells. But, except for genetically encoded calcium indicators, they have rarely been widely applied to flies, worms or mice. Given that genetically encoded Ca²⁺ indicators (GECIs) have the longest history and greatest diversity of sensor platforms, the evolution of GECIs provides a valuable roadmap for developers of fluorescent sensors for other metal ions. In particular, while CFP-YFP FRET sensors dominated the landscape for many years, for specialized applications like in vivo imaging, the field has moved toward single FP sensors, near-IR FRET pairs, bioluminescent tools, and chemigenetic sensors that can be generated using long wavelength fluorophores.

Table 1: List of most widely used sensors for common metal targets.

Many of these sensors have undergone multiple rounds of optimization leading to robust, high performance sensor platforms. More details on these sensors can be found in the following reviews: ^6,15,25,51

Target metal ion	Single FP platform	FRET-based platform	Chemigenetic platform
Ca2+	GCamP	YC cameleons	HaloCamP
	GECO	D-family cameleons	HaloGFP-Ca1–4
	jRCamP	TN-cameleons	splitFAST-Ca2+
		iGECI
Zn2+	GZnP1–3	ZAPCY family
	ZnGreen	ZapCV family
	ZnRed	eCALWY family
	ZIBG1–2	eZinCh2
K+	KRaIONI	GEPII family
	GINKO family	KIRIN family
Mg2+		MARIO	HaloTag-MGH
Na+			HaloGFP-Na1
Heme-Fe		CISDY
		HS1
La3+		LaMP1

Conclusions

A wide variety of genetically encoded sensors for metals in biology exist. While calcium and zinc indicators have dominated the field, there are exciting recent advances in sensors for essential group I alkali metals such as K⁺ and Na⁺. One of the clear growth areas is the expansion of the chemigenetic sensor platform. There is now a diverse array of designs that leverage highly fluorogenic JF⁵⁶ and MaP⁶⁹ dyes, natural binding domains or synthetic chelators, and genetically encodable protein domains. Another area of active development is extending these sensor platforms to a wider array of metal ions, particularly metals that are often found in environmental samples.

Highlights.

• Genetically encoded sensors are powerful tools for visualizing metal ions.

• Both protein and nucleic acid sensor platforms have been developed.

• Ca²⁺ and Zn²⁺ have the greatest variety of sensor platforms.