Origins of Ca²⁺ imaging with fluorescent indicators

Abstract

In 1980, Roger Tsien published a paper, in this journal (Biochemistry, 1980, 19 (11), 2396), describing “New calcium indicators and buffers with high selectivity against magnesium and protons: design, synthesis, and properties of prototype structures.” These new buffers included 1,2-bis(o-aminophenoxy)ethane-N,N,N’,N’-tetraacetic acid, or BAPTA, which remains in wide usage today. And so, the world was set alight with new ways in which to visualize Ca²⁺. The ability to watch fluctuations in intracellular Ca²⁺ revolutionized the life sciences, although the fluorescent indicators used today, particularly neurobiology, no longer rely exclusively BAPTA but on genetically-encoded fluorescent Ca²⁺ indicators. In this Perspective, we reflect on the origins of Ca²⁺ imaging with a special focus on the contributions made by Roger Tsien, from the early concept of selective Ca²⁺ binding described in Biochemistry, to optical Ca²⁺ indicators based on chemically-synthesized fluorophores, through to genetically encoded fluorescent Ca²⁺ indicators.

Graphical Abstract

Introduction

Ca²⁺ imaging in neurons is a mainstay of modern neurobiology. Using genetically encoded Ca²⁺ indicators (GECIs), combined with state-of-the-art optics and computational approaches, recording the activity of ensembles of neurons across large areas of the brains of model organisms is routine. From worms,¹ to flies,² to mice³ and marmosets,⁴ fluorescent Ca²⁺ indicators⁵ have been deployed to paint a picture – or more fittingly, record a movie – of brain activity with cellular resolution. Together with advances in optogenetic approaches^6–8 – controlling neuronal activity with light – neurobiology has made substantial strides towards both measuring and manipulating brain activity with light.⁹

This Perspective aims to examine some of the origins of Ca²⁺ imaging. There have been many important contributions and contributors to this field. Inspired by a report published in 1980 by Roger Tsien in this journal,¹⁰ we trace the involvement and influence of Roger Tsien on the field of Ca²⁺ imaging, starting with the development of BAPTA during his studies at Cambridge,¹¹ moving through synthetic Ca²⁺ indicators pioneered in his labs, and proceeding to the first variants of genetically encoded Ca²⁺ indicators. It is perhaps a fitting testament and tribute^11–14 to the prowess of the late Prof. Tsien that these innovations and their descendants are still in wide usage today.

BAPTA

Roger’s initial paper describing BAPTA was a triumph of molecular design (Figure 1). The main Ca²⁺ chelator/buffer of the day, EGTA, solved the problem of achieving Ca²⁺/Mg²⁺ selectivity (Mg²⁺ exists in a ~10,000:1 excess over cellular Ca²⁺),¹⁵ as Roger eloquently described in the original paper. The unique tetra carboxylate motif of EGTA could bind Ca²⁺ preferentially over Mg²⁺ because the larger ionic radius of Ca²⁺ allowed the carboxylates to form a perfectly-sized chelate. Mg²⁺, on the other hand, with its smaller ionic radius, would force the anionic carboxylates into much closer proximity, inducing an energetic penalty for Mg²⁺ binding on account of charge repulsion.¹⁰ However, the problem with EGTA was that binding of Ca²⁺ didn’t induce a detectable optical change in the molecule. Furthermore, the aliphatic amines critical for binding Ca²⁺ were extremely basic: with pKa values of 9.0 and 8.6,¹⁰ at physiological pH, the amines are primarily protonated. Not only did EGTA bind Ca²⁺ quite slowly, the binding kinetics were pH sensitive. Roger’s simple, ingenious solution was to incorporate the aliphatic amine scaffold of EGTA into an aromatic framework, converting the alkyl amines into anilines (Figure 1). And thus, BAPTA was born.

Figure 1.

Structures of EGTA and BAPTA. Annulation of EGTA results in BAPTA, which retains the overall shape of EGTA (conferring Ca²⁺ selectivity over Mg²⁺), but with less basic aniline nitrogens, which lowers the pH sensitivity and improves binding kinetics of BAPTA. Binding of Ca²⁺ results in a shift from an “aniline-like” structure to a “benzene-like” structure.

BAPTA preserved the unique size and shape of EGTA which conferred its notable selectivity for Ca²⁺ over Mg²⁺: BAPTA binds Ca²⁺ with a K_d of approximately 100 nM and binds Mg²⁺ with a K_d of approximately 20 mM.¹⁰ The Ca²⁺ K_d of BAPTA (~100 nM) is well-matched to the Ca²⁺ K_d of EGTA (~150 nM, pH 7.2, 100 mM KCl).¹⁶ This structural change also endowed BAPTA with a few critical improvements over EGTA. First, because the pKa of aniline nitrogens is several orders of magnitude lower than aliphatic amines (the two highest BAPTA pKa values are 6.4 and 5.5),¹⁰ BAPTA binding of Ca²⁺ was no longer pH dependent in the physiological range, and the binding kinetics of Ca²⁺ were improved compared to EGTA, allowing for reversible monitoring of both Ca²⁺ binding and release.¹⁰

However, and perhaps, more importantly, the aromatic rings of BAPTA now provided a visual cue that Ca²⁺ had bound to the BAPTA chelator. Upon complexation with Ca²⁺, the lone pairs of electrons on the BAPTA nitrogens twist out of conjugation with the aromatic ring (Figure 1), shifting the absorbance band of BAPTA from “aniline-like” with a maximum of 254 nm in the absence of Ca²⁺ to one that was more “benzene-like”, with a maximum of 203 nm, when Ca²⁺ bound. The world now had an optical indicator that showed good selectivity for Ca²⁺ over Mg²⁺ (>10⁵ selectivity) and H⁺.

BAPTA had a few design limitations that precluded its use in living cells: 1) the absorbance of BAPTA was obscured by the more abundant protein and nucleic acids present in cells, which absorb in the same region of the spectrum, and 2) BAPTA is not fluorescent at visible wavelengths, making it unusable in cells. In addition to the wavelength problem, the other obstacle, at least to the application of BAPTA-based indicators to living cells, was that the structural motif essential for Ca²⁺ binding – the tetra carboxylic acids – rendered BAPTA-type chelators unable to cross cellular membranes. Roger would set out to solve both of these problems,¹¹ first in another sole-author paper¹⁷ prior to starting his independent career at UC Berkeley, and second, in the unveiling of fura-2 in 1985.¹⁶

AM-esters

Roger realized that delivery of Ca²⁺ indicators to cells via microinjection would limit their wide-spread utility. So, he devised a method to chemically mask the tetra-carboxylate binding motif as an acetoxy-methyl ester (AM ester, Figure 2). The lipophilic tetra-ester crosses cell membranes, and ubiquitous cellular carboxy esterases remove the acetyl group, the resulting hemi-acetal decomposes to release an equivalent of methylene glycol (hydrated form of formaldehyde), and the exposed carboxylates can now bind Ca²⁺. This method, inspired by pro-drug strategies used for antibiotics¹⁸ remains the standard for delivery of synthetic Ca²⁺ indicators and other anionic small molecules to living cells. Recent innovations include bulky, hydrolytically stable AM esters for delivery to defined cells.¹⁹

Figure 2.

Delivery of small molecules to cells via acetoxymethyl (AM) esters.

Ratio-based Ca²⁺ imaging

The other challenge still facing those who wanted to visualize Ca²⁺ in living systems was the wavelength problem associated with BAPTA. Although BAPTA provided a visual cue that Ca²⁺ binding had taken place, the wavelengths were far too short (<254 nm) to be useful in cells. To address this, Roger built a chromophore around the BAPTA binding motif. A prototype of this strategy was presented alongside BAPTA in the 1980 Biochemistry paper. The quinoline-containing Ca²⁺ chelator (quin-2),¹⁰ along with the recently developed AM esters, enabled bulk measurement of intracellular Ca²⁺ dynamics in intact lymphocytes.²⁰ The low brightness of quin-2 prompted the exploration of additional motifs for visualizing intracellular Ca²⁺. In 1985, Roger reported the improved Ca²⁺ indicator fura-2, in which the BAPTA core was extended to include a benzofuran fluorophore (Figure 3).¹⁶ Binding of Ca²⁺ induced the same conformational rotation of the aniline and shifted the absorbance maximum of fura-2 by about 30 nm (Figure 3b). The emission wavelength was relatively unchanged as a result of Ca²⁺ binding, which allowed practitioners to switch excitation wavelengths to alternatively excite the Ca²⁺-bound form (335 nm) or the Ca²⁺-free form (362 nm) of fura-2 while monitoring same emission band, likely because Ca²⁺ dissociates from the amine in the excited state.²¹

Figure 3.

Fura-2 structure and Ca²⁺ binding. a) Structures of the Ca²⁺-free and Ca²⁺-bound fura-2. b) Binding of Ca²⁺results in a blue-shift of fura-2 absorbance, with little change in the emission wavelength.

By taking the ratio of the emission intensity in cells excited at the wavelengths corresponding to the Ca²⁺-bound and -free forms of fura-2, one could estimate an actual Ca²⁺ concentration in living cells. The two excitation wavelengths allowed for correction of experimentally confounding factors like uneven loading of the dye in different cellular compartments or tissues, fluctuations in excitation source intensity, drifts in detector sensitivity or photo-bleaching. The bathochromic wavelength shift in fura-2 compared to BAPTA and quin-2 was far enough to push fura-2 into a range that could be used in living cells. Despite the requirement for UV excitation, which can be damaging to cells and tissue, fura-2 was received with open arms. It was widely adopted and remains the best and most widely-used ratio-based Ca²⁺ indicator, some 35 years after its initial report. The original paper has >21,000 citations according to Web of Science, and, in 2006 was the sixth most cited article in the history of the Journal of Biological Chemistry.²² Another compound in the same article, indo-1, nicely complements fura-2, affording ratio-based imaging of Ca²⁺ via emission, and making it useful for flow cytometry²³ and confocal microscopy.²⁴

Intensity-based Ca²⁺ imaging

Fura-2 combined several key intellectual developments to provide ratio-based optical estimations of intracellular Ca²⁺ concentration: the BAPTA chelator to achieve selectivity for Ca²⁺ over Mg²⁺ and H⁺, AM esters to deliver optical indicators into living cells, and a benzofuran fusion with BAPTA to provide wavelengths that could be used in living cells. However, high energy UV excitation was prohibitive in some contexts. To expand into the visible wavelengths, Roger decoupled the BAPTA Ca²⁺-sensing unit from the fluorescent reporter. Whereas in fura-2, BAPTA was incorporated into the benzofuran fluorophore, this next generation of Ca²⁺ indicators divorced the Ca²⁺ sensing motif (BAPTA) from the fluorescent reporter (which now had the freedom to be any fluorophore). The first two examples of this new strategy made use of fluorescein and rhodamine. To achieve Ca²⁺ sensing with these xanthene dyes, Roger connected BAPTA to the pendant aryl ring of the fluorescein or rhodamine, generating the Ca²⁺-sensitive fluorophores christened fluo-1, −2, or −3 and rhod-1 or −2 (Figure 4).

Figure 4.

Structures of synthetic Ca²⁺ indicators. Quin-2, Fura-2, and Fluor-3 are all first generation indicators developed by Tsien and co-workers. Fura-2, Fura Red, and BTC are ratio-based indicators. Fluo-3, OGB-1, CaTM-3, Ca Ruby Nano, X-Rhod, CaSiR-2, and Compounds 1–5 are all intensity-based indicators.

These new indicators sense Ca²⁺ via a different mechanism from fura-2, a mechanism that involves photoinduced electron transfer, or PeT (Figure 5a). In the absence of Ca²⁺, the electron-rich BAPTA short circuits fluorescence by donating, or transferring, an electron into the photoinduced excited state of the fluorescein or rhodamine. When BAPTA binds Ca²⁺, this lowers the highest occupied molecular orbital (HOMO) of BAPTA, because the molecular orbitals of the xanthene fluorophore and the BAPTA chelating group are largely electronically decoupled because of the orthogonality of the fused xanthene ring system and the aryl ring at the 9’ position²⁵ (although the original intent of molecules like fluo-2 was to maintain some electronic coupling and induce a Ca²⁺-dependent shift in absorbance, excitation, or emission wavelength)²⁶

Figure 5.

Fluo-3 structure and mechanism. a) Structure and binding of fluo-3 to Ca²⁺. In the absence of Ca²⁺, photoinduced electron transfer (PeT) from the electron-rich BAPTA group quenches fluorescence. Binding of Ca²⁺ relieves PeT, and the dye becomes more fluorescent. b) Representative plot of fluo-3 absorbance (grey) and emission (green) vs. wavelength with increasing Ca²⁺ concentration. Although emission intensity increases upon addition of Ca²⁺, the absorbance spectra of fluo-3 remains relatively unchanged.

This lowers the free energy of PeT and makes the PeT process less efficient. As a result, binding Ca²⁺ increases fluorescence and the dye brightens (Figure 5b). Unlike fura-2, the Ca²⁺-induced change in the optical properties of fluoand rhod-type dyes is an increase in the fluorescence quantum yield at a single wavelength, precluding estimates of actual Ca²⁺ concentration. However, fluo-2 and its progeny had the advantage of matching commonly available excitation and emission courses in microscopes world-wide, and the ability to track changes in Ca²⁺ levels, in real time, using commercially available microscopes was quickly adopted.

Second generation intensity-based Ca²⁺ indicators

As the Tsien lab increasingly turned its attention to the development of GFP, others continued the development of new variants of Ca²⁺ indicators (many of these efforts are reviewed more extensively elsewhere);^27–28 fluo-4, which used the recently developed 2’,7’-fluorofluorescein,²⁹ is still in use to this day, as is Oregon Green BAPTA (OGB) which, while also using the fluorofluorescein developed at Molecular Probes, made use instead of an amide linkage to connect BAPTA to the fluorophore (Figure 4). This insulating attachment (rather than direct incorporation as in the fluo-X series), along with the presence of the 3-carboxylate on the aryl ring, gives OGB a higher fluorescence quantum yield than its corresponding fluo-X cousins.²⁷ Since these initial reports, there have been a profusion of intensity-based Ca²⁺ indicators (up to fluo-8, at the present), which largely use BAPTA, although with some modification to the substituent para to the BAPTA nitrogen, which allows tuning of BAPTA’s affinity for Ca²⁺ across a wide swatch of useful concentration ranges, from nanomolar to low micromolar. Excellent reviews compiling wavelength, binding affinities, and usage guidelines for the wide array of chemically-synthesized Ca²⁺ indicators are available.^27–28

Improving intensity- and ratio-based indicators

Ca²⁺ imaging at even longer wavelengths can be achieved with rhodamine fluorophores, such as X-Rhod-1 or Ca Ruby,³⁰ which contain a julolidine-derived³¹ rhodamine 101³² core (Figure 4). Development in xanthene dye chemistry has allowed access to orange, red, and far-red emitting fluorophores that possess the compact, three-ring structure of fluorescein and rhodamine. The groups at the fore-front of exploring these new chemistries applied the lessons of Roger Tsien’s lab to generating new Ca²⁺ indicators that push Ca²⁺ sensing into new regions of the visible spectrum with indicators like Ca Tokyo Magenta (CaTM, Figure 4)^33–34 or Ca silicon rhodamine (CaSIR2, Figure 4).³⁵ Advances in rhodamine synthesis also allowed BAPTA to be positioned outside of the canonical 9’ position on the xanthene dyes, resulting in Ca²⁺ indicators with improved brightness and response to Ca²⁺ (Compounds 1–5, Figure 4).³⁶

On the other hand, improvements and derivatives of fura-2 have been fewer. Fura-red (Figure 4) attempts to extend ratio-based imaging into the visible spectrum, but fura-red³⁷ is quite dim and Ca²⁺ binding further lowers the fluorescence quantum yield.³⁸ Benzothiazolium-coumarin dyes like BTC (Figure 4) show red-shifted ratio-based responses to Ca²⁺, but with binding affinities in the >7 μM range.³⁹ Other improvements include clever ways to anchor fura-2 (and derivatives) to macromolecules⁴⁰ or self-labeling enzymes⁴¹ to improve cellular retention and localization.⁴² But robust methods for ratio-based imaging with long wavelengths using a fura-type scaffold remain an outstanding challenge. In particular, the move to replace mercury arc lamps with light-emitting diodes (LEDs) means that the ability to perform ratio-based Ca²⁺ imaging in living cells is becoming the provenance of specialized microscopes: LEDs with excitation wavelengths at the 340/380 nm wavelengths for fura-2 are not common and require bespoke instrumentation.⁴³

Alternative sensing modalities

Chemically-synthesized Ca²⁺ indicators afford the opportunity to extend to modalities beyond visible light. For example, magnetic resonance imaging (MRI) agents like DOPTA-Gd that change contrast in response to Ca²⁺ binding make use of BAPTA to achieve Ca²⁺-sensitive MR contrast (Figure 4).^44–46 Synthetic efforts continue to push Ca²⁺ imaging into the near infrared region of the electromagnetic spectrum.^47–48 More recently, chemical-genetic hybrids have been employed, combining the Ca²⁺-binding properties of genetically-encoded approaches^49–50 (see below) with modern fluorophores.^51–52

Genetically encoded Ca²⁺ indicators (GECIs)

Delivery of chemically-synthesized Ca²⁺-sensitive fluorophores, especially into tissues within intact organisms, remains a challenge. Although AM ester allow passage of fluorescent dyes across cellular membranes, their inclusion substantially decreases water solubility of the masked Ca²⁺ indicators. The advent of green fluorescent protein (GFP), and the optimized⁵³ and multi-color variants⁵⁴ developed in the Tsien lab, fundamentally altered this landscape. Today, the use of genetically-encoded Ca²⁺ indicators (GECIs) based off of GFP derivatives is ubiquitous across the field of neurobiology, especially for applications that involve in vivo imaging of neuronal activity.

Two-component indicators, or cameleons

The first example of a genetically encoded Ca²⁺-sensitive indicator (GECI) shows not only the rapid advances that had been made since the first demonstration of heterologous expression of GFP in 1994,⁵⁵ but also the advances in creative nomenclature compared to synthetic indicators.

Cameleon was the first FP-based Ca²⁺ indicator to be expressed in cells and used to monitor Ca²⁺ dynamics (Figure 6a).⁵⁶ Cameleon used two variants of GFP, cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP). CFP was fused to the N-terminus of calmodulin (CaM), a naturally-ocurring Ca²⁺-binding protein, which was in turn connected to the N-terminus of the CaM-binding peptide, M13, fused to YFP. Because of the close proximity of the two FP fluorophores and the overlap of the emission spectra of CFP with the excitation spectra of YFP, excitation of CFP could induce emission from YFP via radiationless dipole-dipole interactions known as Förster resonance energy transfer, or FRET.⁵⁸ Increases in Ca²⁺ levels results in the formation of a complex between CaM and Ca²⁺, which can in turn bind to M13. Changes in the conformation of the intervening linker between CFP and YFP increases the FRET ratio. Practically speaking, the ratio of emission from YFP when exciting CFP (donor) to the emission from YFP (acceptor) when exciting it directly would reveal the extent of Ca²⁺ binding to CaM and the concentration of intracellular Ca²⁺. Cameleon was so named because it changes color upon binding Ca²⁺ and extends a long tongue (M13) into and out of the mouth of calmodulin, which is abbreviated CaM.⁵⁶

Figure 6.

Genetically encoded Ca²⁺ indicators of a) “cameleon” or b) “camgaroo” lineages. Cameleon-style indicators use two fluorescent proteins to alter FRET efficiency upon Ca²⁺ binding. Camgaroo-style indicators use a single fluorescent protein (cpGFP, or GCaMP, is shown in this figure).

Similar to fura-2, cameleon and related FRET indicators can provide an estimate of intracellular Ca²⁺ concentrations by use of a ratio-based correction. Descendants of cameleons include indicators that use troponin C as a template for Ca²⁺ binding,⁵⁹ that use computationally-designed CaM-M13 pairs that are orthogonal to the native versions,⁶⁰ or use circularly permuted YFP.⁶¹ However, it was the intensity-based indicators that eventually became more widely used in neurobiology for functional imaging in animal brains.

Single-component indicators, or camgaroo and GCaMP

Intensity-based genetically encoded Ca²⁺ indicators use a single FP to monitor changes in Ca²⁺ concentration. Here, too, Roger Tsien and co-workers showed the early way forward.⁴⁹ Circular permutants⁶² of GFP, in which the N and C termini are migrated to different loops and regions of the FP and the original termini joined via a linker, place the FP chromophore closer to the solvent, breaking the protective “shell” formed by the β-barrel of GFP. Appending CaM/M13 at the ends of the new termini allowed conformational changes wrought by Ca²⁺ interactions with CaM and M13 to be effectively translated to the protein region surrounding the FP fluorophore, creating Ca²⁺-dependent fluorescence. This type of configuration carried out with enhanced yellow fluorescent protein (EYFP) was christened camgaroo. A cheeky footnote in the original manuscript states that camgaroo is an “apt nickname” because “it is yellowish, carries a smaller companion (calmodulin = M13) inserted in its ‘pouch,’ can bounce high in signal, and may spawn improved progeny.”⁴⁹

Camgaroo did indeed spawn more progeny. Both G-CaMP⁵⁰ and pericam⁶³ are circularly permuted GFPs or YFPs that use the same CaM/M13 strategy. Subsequent generations of G-CaMP (eventually shortened to GCaMP) have been used widely across model organisms, including fish, flies, worms, and mice.^{3, 5} Multiple rounds of systematic mutagenesis led to generations of GCaMPs^{5, 64} – all of which kept the cpGFP core and CaM/M13 binding partners – improving Ca²⁺ binding kinetics, increasing brightness, and enhancing the optical response to Ca²⁺ binding. The sixth generation, GCaMP6,³ has become nearly ubiquitous in neurobiological imaging, owing, in part, to the ease of use, and the availability of indicators with tailored properties: fast variants (GCaMP6f), with fast Ca²⁺ binding and release kinetics, slower variants (GCaMP6s) with larger increases in fluorescence, but slower Ca²⁺ response and release kinectics.³ Subsequent generations of GCaMP continue to improve brightness and reduce toxicity: jGCaMP7⁶⁵ was the first generation to include “j,” presumably to indicate the provenance of these discoveries, Janelia Research Campus. Like synthetic Ca²⁺ indicators, GECIs have expanded into other wavelengths,^66–69 although the GFP-based indicators remain the most widely used.

Conclusions/Future Directions

Ca²⁺ imaging has a rich history and a bright future. Our goal in this Perspective was two-fold: to trace some of the origins of Ca²⁺ imaging to the seminal paper published in Biochemistry¹⁰ and to highlight the contributions that one scientist in particular, Roger Tsien, made towards illuminating our understanding of neurobiology. Science, of course, is a team- and community-based endeavor, and our goal here is not to overlook the contributions made by others, but to highlight one scientist’s involvement in so many facets of this journey.

In the future, inspired by the contributions over the last 40 years, we look forward to new innovations, with improved ratio-based imaging at long wavelengths, advanced microscopies for imaging deeper in the brain,^70–72 and less invasive methods for monitoring Ca²⁺ that do not rely on light alone.^71–72