Imaging calcium signals in vivo: a powerful tool in physiology and pharmacology

Abstract

The design and engineering of organic fluorescent Ca²⁺ indicators approximately 30 years ago opened the door for imaging cellular Ca²⁺ signals with a high degree of temporal and spatial resolution. Over this time, Ca²⁺ imaging has revolutionized our approaches for tissue-level spatiotemporal analysis of functional organization and has matured into a powerful tool for in situ imaging of cellular activity in the living animal. In vivo Ca²⁺ imaging with temporal resolution at the millisecond range and spatial resolution at micrometer range has been achieved through novel designs of Ca²⁺ sensors, development of modern microscopes and powerful imaging techniques such as two-photon microscopy. Imaging Ca²⁺ signals in ensembles of cells within tissue in 3D allows for analysis of integrated cellular function, which, in the case of the brain, enables recording activity patterns in local circuits. The recent development of miniaturized compact, fibre-optic-based, mechanically flexible microendoscopes capable of two-photon microscopy opens the door for imaging activity in awake, behaving animals. This development is poised to open a new chapter in physiological experiments and for pharmacological approaches in the development of novel therapies.

LINKED ARTICLES

This article is part of a themed section on Imaging. To view the other articles in this section visit http://dx.doi.org/10.1111/bph.2011.163.issue-8BJP has previously published an Imaging in Pharmacology themed section, edited by A Davenport and C Daly. To view this section visit http://dx.doi.org/10.1111/bph.2010.159.issue-4

Introduction

Ca²⁺ is an ubiquitous intracellular messenger that regulates multiple cellular functions such as secretion, contraction, cellular excitability and gene expression in all organ systems. The earliest recognition that calcium ions are essential for regulating biological processes was by Sydney Ringer, who serendipitously found that ‘lime salt’ is necessary to maintain the contractions of an isolated frog heart (Ringer, 1883). Integrated function of organ systems in general is achieved by concerted signalling by entire cellular ensembles in organs in synchronous and/or asynchronous modes, and Ca²⁺ ions play a central role in such signalling. Therefore, understanding Ca²⁺ signals, and their temporal and spatial characteristics in cells and tissues is crucial to both gain knowledge of the physiological regulation of the organ system, and for developing pharmacological approaches. The pioneering chemical synthesis of organic Ca²⁺ indicators based on the Ca²⁺ chelating properties of EGTA in the 1970s by Roger Tsien and colleagues, made direct monitoring of cellular Ca²⁺ signals possible (Tsien, 1980; 1981; Grynkiewicz et al., 1985). Early studies focused on measuring Ca²⁺ signals in acutely isolated living cells and cells in tissue culture. It is however necessary to understand the signal flow between cells in intact systems, and characterize the spatially discrete and temporally complex signalling in order to describe systems level physiological regulation. Since many physiological signalling scenarios cannot be accurately reproduced in vitro, it is essential to record and analyse integrated signals between cells in living tissue in vivo. Advances in optical and computing technologies yielded powerful microscope systems that enabled the realization of this goal. By combining appropriate Ca²⁺ indicators with appropriate optical imaging techniques, cellular Ca²⁺ signals and sometimes fluxes have been monitored with a high degree of spatial and temporal resolution. Such imaging approaches have been developed for a number of tissues and organ systems in various animal models, and in the last decade, this effort has led to breathtaking insights into signalling in intact systems, and have been reviewed extensively (Garaschuk et al., 2007; Paredes et al., 2008; Rochefort et al., 2008). In this review I present recent technical developments in the fields of Ca²⁺ indicators and in vivo Ca²⁺ imaging. These technical advances have made Ca²⁺ imaging a powerful approach for functional and pharmacological studies in vivo.

In vivo calcium imaging – an overview

Imaging Ca²⁺ signals in cells in culture and tissue slices became routine in most laboratories over the last two decades. During this time, methods were developed to investigate the cellular physiology and pharmacology of Ca²⁺ signals in many different cellular and tissue systems in different animal models (see for review Bootman and Berridge, 1995). Most initial studies were restricted to imaging cells in culture. Very early on, however, it was recognized that cellular Ca²⁺ signals have complex spatio-temporal forms and spread as waves within cells, and in between cells. For that reason, acutely isolated tissue and organs were loaded with Ca²⁺ indicator dyes and imaged (Duffy and MacVicar, 1995; Porter and McCarthy, 1995). An alternative preparation for brain tissue studies is the use of organotypic cultures of brain slice preparations (Dani et al., 1992). Here, the brain slices are maintained in culture in an air–water interphase where the slices thinned out but retained more or less the cytoarchitecture of the original brain circuits (Gahwiler, 1981). These cultures were labelled and imaged to study physiological signalling modes (Dani et al., 1992). By this time, the recognition that Ca²⁺ signals are organized in specialized signalling microdomains in cells and coordinated in specific temporal sequences led to experiments using more intact systems.

Ideally, it is necessary to image organ systems in the living animal to gain understanding of signalling patterns within tissue. In experiments with liver cells in culture and perfused intact livers, it was found that agonist-evoked Ca²⁺ signals were qualitatively and quantitatively different in the intact preparation compared with those in isolated hepatocytes (Gaspers and Thomas, 2005). The brain is unique in its signal organization, where sensations and behaviours are encoded in patterns of activity in neuronal networks. Through a sequence of network, activity information is relayed to other associative brain regions where signal integration and activity patterns are achieved. This chain of events perhaps occurs in a series of important processing steps in local microcircuits. Activity in local circuits is not accurately recorded in isolated brain slice or culture preparations. It is necessary to image intact brain circuits in the living animal in order to understand the cellular signalling in physiology and information processing. In addition, a number of studies have shown that dendritic excitability is modulated by various factors including ionic composition, excitatory and inhibitory inputs, and activation and inactivation of voltage-dependent conductances. For that reason, dendritic signals recorded in brain slice preparations are different from in situ recordings in intact animals (Svoboda et al., 1999). These observations provided the impetus for the development of methods to discretely deliver Ca²⁺ indicators to cells in situ either using cell injection or bulk loading for imaging in the intact living animal.

Thus, it is only more recently that live animals were imaged to investigate physiological signalling in intact organ systems in vivo. The earliest experiments were carried out in invertebrates, worms and insects, plants, and small vertebrates, for example, zebrafish (Friedrich and Korsching, 1997; Dal Santo et al., 1999; Plieth, 2001). These initial studies used either expression of the luminescent Ca²⁺ sensor protein apoaequorin in transgenic organisms (Creton et al., 1997) or bulk loading of cell permeant versions of organic fluorescent Ca²⁺ indicators such as Fura-2 and Fluo-3 (Stosiek et al., 2003; Ohki et al., 2006; Takano et al., 2007). Luminescence signals were recorded using photomultiplier tubes, and fluorescence-based organic dyes were imaged using charge coupled device (CCD) camera systems with or without image intensification for signal enhancement. The development of more sensitive CCD sensors led to studies on sensory neural systems in non-mammalian brains, such as olfactory systems in the honeybee (Galizia et al., 1999), Drosophila (Karunanithi et al., 1997) and turtle (Wachowiak et al., 2002) as well as the auditory system in the cricket (Sobel and Tank, 1994). The choice of these readily accessible neural systems in lower animals was due to the very high light-scattering property of neural tissue, which restricted depth penetration in conventional imaging systems available at that time. In most tissues, particularly the brain, light scattering prevents the use of wide-field and confocal microscopy for imaging cells of interest located deep within tissue. The implementation of two-photon laser scanning microscopy finally allowed for deep tissue imaging in the brain (Hoogland et al., 2009; Grewe et al., 2010; Lutcke et al. 2010). A number of novel fibre-optic-based two-photon microendoscopes have been recently developed (Helmchen et al., 2001; Gobel et al., 2004), which makes the technique capable for both recording physiological activity in vivo and also for investigating pathophysiology of disease states. Indeed most in vivo imaging to date has been carried out in neural systems (see Helmchen and Kleinfeld, 2008; Homma et al., 2009 for reviews). The development of multiphoton laser scanning microscopy, together with the design of fluorescent protein-based Ca²⁺ sensors that could be targeted to specific cell types in transgenic animal models, has made in vivo Ca²⁺ imaging into a powerful tool generally available to most physiologists and pharmacologists (see Rochefort et al., 2008 for review). Two-photon confocal microscopy implemented on fixed stage microscopes are still too costly for many laboratories. Fibre-optic-based microendoscopes capable of two-photon microscopy have become more reasonably priced and are accessible by many investigators.

Calcium indicators

The ubiquitous nature of Ca²⁺ as an intracellular messenger encouraged scientists to search for Ca²⁺ indicators that would enable optical measurements of Ca²⁺ concentration in the cytoplasm of cells. The earliest measurements of intracellular Ca²⁺ dynamics occurred following the extraction and characterization of the Ca²⁺-sensitive bioluminescent protein aequorin from the jellyfish, Aequoria victoria (Shimomura et al., 1962). Aequorin extracted from jellyfish tissue was microinjected into cells to monitor rapid changes in intracellular Ca²⁺ by measuring luminescence changes (Hastings et al., 1969; Baker et al., 1971; Shimomura et al., 1993). About this time metalochromatic dyes like murexide and arzenazo III were also used to measure intracellular ionized Ca²⁺ concentration (Jobsis and O'Connor, 1966; Dipolo et al., 1976; Miledi et al., 1977). Although these dyes were initially promising, the Ca²⁺-dependent spectral changes were highly sensitive to pH fluctuations that were known to occur in cells (Russell and Martonosi, 1978). Aequorin, on the other hand, was not readily available and required microinjection into individual cells, and for that reason its use remained in the domain of few physiologists. Furthermore, aequorin is inactivated during light emission and hence its use was confined to experiments of short duration. With the design and development of organic fluorescent Ca²⁺ indicators for fluorimetric microscopy, a new field of research in cellular Ca²⁺ signalling began in earnest. Currently a large number of organic fluorescent probes, and fluorescent protein-based indicators are available for routine use. The most commonly used indicators are listed in Table 1.

Table 1.

Most commonly used fluorimetric calcium indicators

Indicator	Species	Application/tissue studied	Labelling technique	Imaging technique	Spatial resolution	Response amplitude	References
Organic synthetic fluorescent dyes
Fura-2	Blowfly	Visual system	Cell microinjection	Epifluorescence/CCD camera	Cellular/ subcellular	10–20%	Borst and Egelhaaf, 1992
Fura-2	Cricket	Auditory system	Cell microinjection	Epifluorescence/CCD camera	Cellular	6–20%	Sobel and Tank, 1994
Fluo-3	C. elegans	Apoptosis	Bulk microinjection	Confocal microscopy	Cellular	15%	Jain et al., 1993
Fluo-3	Drosophila	Motor nerve terminals	Bulk loading	Confocal microscopy	Cellular	100–200%	Karunanithi et al., 1997
Fura-2, Fluo-4, Indo-1	Mouse	Neocortex	Bulk microinjection	Two-photon microscopy	Cellular	20–50%	Stosiek et al., 2003
Calcium green-1	Honeybee	Olfactory system	Bulk microinjection	Epifluorescence/CCD camera	Glomerular	2–5%	Galizia et al., 1999
Calcium green-1	Zebrafish	Olfactory bulb	Bulk microinjection	Epifluorescence/CCD camera	Glomerular	5–10%	Friedrich and Korsching, 1997
Calcium green-1	Turtle	Olfactory system	Bulk loading	Epifluorescence/CCD camera	Cellular	5–20%	Wachowiak et al., 2002
Oregon green BAPTA	Mouse	Neocortex	Cell microinjection	Two-photon microscopy	Cellular/ subcellular	40–200%	Helmchen et al., 1999; Svoboda et al., 1999
Oregon green BAPTA	Cat	Visual cortex	Bulk microinjection	Two-photon microscopy	Cellular	40–200%	Ohki et al., 2006
Oregon green BAPTA	Ferrett	Visual cortex	Bulk microinjection	Two-photon microscopy	Cellular	10–30%	Shummers et al., 2008
Rhod-2	Mouse	Cortical astrocytes	Bulk microinjection	Two-photon microscopy	Cellular	10–100%	Takano et al., 2007
Aequorin-based luminescence calcium indicators
Aequorin	Tobacco	Whole plant	Transgenic	Luminescence detection	Bulk tissue	NA	Knight et al., 1991
Aequorin	Arabidopsis	Whole plant	Transgenic	Luminescence detection	Bulk tissue	∼10-fold	Knight et al., 1996; Liu et al., 2006
Aequorin	Zebrafish	Development	mRNA injection/transgenic	Luminescence detection	Tissue/cellular	5- to 10-fold	Creton et al., 1997; Cheung et al., 2006
Aequorin	Drosophila	Brain/mushroom bodies	Transgenic	Luminescence/photon counting	Bulk tissue	>100%	Martin et al., 2007
Fluorescent protein-based calcium indicators
DsRed/inverse pericam	C. elegans	Pharyngeal muscles	Transgenic	Epifluorescence/CCD camera	Bulk tissue	20–30%	Shimozono et al., 2004
YC 2.1, YC 3.1	C. elegans	Brain/sensory neurons	Transgenic	Epifluorescence/CCD camera	Bulk tissue	50–60%	Kerr et al., 2000
G-CaMP	Drosophila	Olfactory system	Transgenic	Two-photon microscopy	Bulk cellular	100%	Wang et al., 2003
YC 3.1	Drosophila	Flight muscle	Transgenic	Confocal microscopy	Bulk cellular	12%	Gordon and Dickinson, 2006
G-CaMP2, synapcam, YC 2.3, TN-L15	Drosophila	Motor neurons	Transgenic	Epifluorescence/CCD camera	Cellular	30–700%	Guerrero et al., 2005; Mank et al., 2006; Reiff et al., 2005
YC 2.1	Zebrafish	Spinal cord/neurons	Transgenic	Confocal microscopy	Cellular	15%	Higashijima et al., 2003
YC 2.12	Zebrafish	Development	Transgenic	Epifluorescence/CCD camera	Bulk cellular	NA	Tsuruwaka et al., 2007
Camgaroo, inverse pericam	Mouse	Olfactory bulb/neurons	Transgenic	Epifluorescence/CCD camera	Bulk tissue	1–3%	Hasan et al., 2004
G-CaMP2	Mouse	Cerebellum/neurons	Transgenic	Epifluorescence/CCD camera, two-photon microscopy	Bulk tissue	50%	Diez-Garcia et al., 2007
G-CaMP2	Mouse	Heart	Transgenic	Two-photon microscopy	Bulk tissue	60–70%	Tallini et al., 2006
YC 3.12	Mouse	Brain/neurons	Transgenic	Two-photon microscopy	Cellular	10–30%	Hasan et al., 2004
Cer TN-L15	Mouse	Cortex/neurons	Transgenic	Two-photon microscopy	Cellular	5–10%	Heim et al., 2007
Cer TN-L15	Mouse	Brain/astrocytes	Transgenic	Two-photon microscopy	Cellular	10–20%	Atkin et al., 2009
YC 3.60, Cer TN -L15, G-CaMP3	C. elegans, mouse	Cortex/neurons	Transgenic	Two-photon microscopy	Cellular	30–500%	Tian et al., 2009

C. elegans, Caenorhabditis elegans; CCD, charge coupled device; NA, not available.

Organic Ca²⁺ indicators

In the late 1970s Roger Tsien and colleagues synthesized the first organic fluorescent Ca²⁺ indicator, quin-2, which showed a remarkable increase in fluorescence emission upon binding Ca²⁺. The K_d of Ca²⁺ binding was approximately 100 nM, which made quin-2 suitable as a cytoplasmic Ca²⁺ indicator. While it was put to extensive use for measurement of cytoplasmic Ca²⁺ concentration in living cells, quin-2 had a number of shortcomings (Tsien et al., 1982; Grynkiewicz et al., 1985). Tsien and his colleagues improved on this initial design and synthesized a series of similar compounds that altered their fluorescence properties upon Ca²⁺ binding with varying affinities (Tsien, 1980; 1981; Grynkiewicz et al., 1985). This development sparked an explosion of research on cellular Ca²⁺ signalling and led to the development of a family of stable, brightly fluorescent Ca²⁺ indicators with a range of affinities for Ca²⁺ in physiological concentrations (see Paredes et al., 2008). Ca²⁺-sensitive dyes can be classified into two categories: single wavelength indicators and ratiometric indicators. The former exhibit significant Ca²⁺-dependent change in fluorescence intensity without shifting their excitation or emission wavelengths. Ratiometric indictors, on the other hand, shift their peak wavelength of either excitation or emission upon binding Ca²⁺. For that reason this class of dyes permit very accurate quantification of Ca²⁺ concentration that can be corrected for uneven dye loading, dye leaking photobleaching and changes in cell volume. Fura-2, Fluo-3, Fluo-4, calcium green-1 and Oregon green BAPTA were all of this family of organic Ca²⁺ indicators (Table 1). Methods were developed to fill cells with these organic molecules in high-enough concentrations suitable for fluorescence microscopy (Tsien, 1981) (see later). A number of laboratories developed technologies to investigate all aspects of the cell biology of Ca²⁺ signalling in different cell and organ systems.

A large number of fluorescent organic Ca²⁺ indicators became available with varying affinities towards Ca²⁺ and binding rate constants (Paredes et al., 2008). The high-affinity dyes with dissociation constants in submicromolar concentrations (0.1–0.3 µM) were useful for measurements of cytoplasmic Ca²⁺ concentrations (Kao et al., 1989). The low-affinity indicators with dissociation constants in hundreds of micromolar Ca²⁺ were useful for measurement of Ca²⁺ within cellular compartments where Ca²⁺ concentrations ranged in the low millimolar range (Park et al., 2002). Since cytoplasmic Ca²⁺ signals are generally transient in nature, they are measured under non-equilibrium conditions. Because of that reason, it is necessary to be aware of the speed with which an indicator binds Ca²⁺. Most commonly used organic fluorescent Ca²⁺ indicators have equilibrium binding rate constants between 5 and 20 µs. To date, there are more than 25 different organic fluorescent dyes that are commercially available from different sources. These dyes vary widely in their Ca²⁺-binding affinity and binding time constants (see Paredes et al., 2008 for review; see also Invitrogen).

Fluorescent protein-based Ca²⁺ indicators (FPIs)

The discovery of fluorescent proteins occurring in nature resulted in the widespread use of green fluorescent protein (GFP) in cell biological research (Tsien, 1998). FPIs are Ca²⁺-sensitive fluorescent or luminescent proteins and their chimeric constructs that can be expressed in specific subsets of cells in the tissue under investigation, by the use of cell and tissue specific promoters (Miyawaki et al., 1997; Hasan et al., 2004; Rogers et al., 2005; Heim et al., 2007). Once the codons were humanized, these proteins were genetically targeted and expressed in specific cells and cellular compartments non-invasively using molecular biology and transgenic techniques. Mutant forms of these fluorescent proteins were designed with widely differing spectral emission properties (Heim et al., 1994; Cubitt et al., 1995), and this opened the possibility for developing fluorescent proteins with Ca²⁺-induced Foster resonance energy transfer (FRET) properties (Miyawaki et al., 1997; 1999; Heim et al., 2007). Chimeric constructs of fluorescent proteins and Ca²⁺-binding proteins such as calmodulin and troponin-C were engineered yielding cell targetable Ca²⁺ indicators. Some of these new generation fluorescent protein Ca²⁺ indicators (FPIs) were designed with two different GFP mutants [e.g. cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP)] based on changes in FRET efficiency upon Ca²⁺ binding. FPIs are a powerful tool for in vivo Ca²⁺ imaging, since they support detection of signals within cells and cellular compartments from discrete cell types in intact tissue. This has been the holy grail for the field, particularly in the nervous system, where using this technique, local network activity can be monitored in a physiologically behaving animal with appropriate techniques.

During the 1990s, two classes of genetically encodable Ca²⁺ indicators became available, those based on fluorescent proteins, for example, cameleons (Miyawaki et al., 1997; 1999; Heim et al., 2007; see also Tian et al., 2009), and those based on luminescent proteins, aequorin and obelin (Illarionov et al., 1995; Martin et al., 2007; Tian et al., 2009). The very first of the genetically encoded Ca²⁺ sensors were different subclasses of circularly permuted GFP, where mutating different amino acids near the chromophore portion of GFP yielded indicators that reversibly changed fluorescence intensity when the ambient Ca²⁺ concentration changed. These were the pericams that increased fluorescence intensity with increasing Ca²⁺ concentration and inverse pericams that decreased fluorescence intensity with increasing Ca²⁺ concentration and ratiometric pericams (Nagai et al., 2001). Pericams have been successfully used in many physiological imaging applications from Caenorhabditis elegans muscle contraction to mouse neurons (Robert et al., 2001; Hasan et al., 2004; Shimozono et al., 2004). The second class of fluorescent protein Ca²⁺ indicators were camgaroos, and they were designed with the Ca²⁺-binding protein calmodulin inserted between positions 144 and 146 of YFP. Ca²⁺-dependent conformational change in the calmodulin moiety induces ionization of the chromophore resulting in a seven- to eightfold increase in YFP fluorescence (Baird et al., 1999). Camgaroos, like pericams, have proven to be useful for physiological recording in vivo in Drosophila and mice, where activity in olfactory bulb neurons was recorded (Hasan et al., 2004). The third iteration of circularly permuted GFP was G-CaMP, and this new design resulted in a large increase in GFP fluorescence upon Ca²⁺ binding (Nakai et al., 2001). G-CaMP produced much greater fluorescence changes in response to increases in Ca²⁺ concentration compared with the FRET-based designs, and were quickly adopted to monitor Ca²⁺ signals in vivo in a number of species (Wang et al., 2003; Pologruto et al., 2004). Although G-CaMP's slow kinetics limited the temporal resolution of signals, the extraordinary brightness and therefore the increased signal-to-noise were attractive. A number of improvements were made by point mutation strategies resulting in G-CaMP2 and, more recently, G-CaMP3 with improved kinetics (Chaigneau et al., 2007; He et al., 2008; Hoogland et al., 2009). G-CaMP3 is the most recent development and has at least a threefold increased brightness and dynamic range making it potentially a very valuable tool for in vivo imaging (Tian et al., 2009). G-CaMP3 was used for high-speed imaging of network activity in the motor cortex, and detected Ca²⁺ transient amplitudes linearly dependent on action potential number (Tian et al., 2009).

Using mutated GFP variants, cameleons, a new class of chimeric constructs were designed, which produce Ca²⁺-dependent FRET between the two fluorescent proteins in the chimera of CFP and YFP/Venus (Miyawaki et al., 1999) or cerulean and citrine (Griesbeck et al., 2001). Calmodulin or troponin-C served as the Ca²⁺-sensitive domains in the construct which underwent conformational change leading to increased FRET efficiency. A number of different indicators of this class were synthesized including YC 2.1, YC 3.1, YC 3.60 (Miyawaki et al., 1997; Hasan et al., 2004; Nagai et al., 2004; Reiff et al., 2005) and troponin-C-based indicators such as CerTN-L15, and TN-XXL (Reiff et al., 2005; Heim et al., 2007; Mank et al., 2008). Many of these FRET-based cameleons have been successfully used to record Ca²⁺ signals in different animal models in vivo, including C. elegans pharyngeal muscle (Kerr et al., 2000), Drosophila flight muscle (Gordon and Dickinson, 2006) and in rodents (Hendel et al., 2008; Atkin et al., 2009). Table 1 lists the most commonly used fluorescent protein Ca²⁺ indicators for in vivo Ca²⁺ imaging.

The most recent development in the class of genetically encodable FPIs is the design of a chimeric construct of aequorin, the luminescent Ca²⁺ sensor, and GFP. Aequorin is a bioluminescent photoprotein that emits light upon binding to Ca²⁺ (Shimomura et al., 1990). This property of aequorin has been successfully used for in vivo Ca²⁺ imaging in the zebrafish (see Shimomura et al., 1993; Creton et al., 1997). The low quantum yield of apoaequorin luminescence makes it difficult to detect Ca²⁺ events with short integration times (<1 s) or with single cell resolution. In order to overcome this problem, a chimeric construct was recently designed which mimics the native protein arrangement in the jellyfish (Shimomura et al., 1993), where the Ca²⁺-sensitive aequorin luminescence produces bioluminescence resonance energy transfer (BRET) to GFP, emitting light in the visible range (509 nm) (Baubet et al., 2000; Rogers et al., 2005). Using this principle, aequorin-GFP chimera were engineered, and were shown to be fast responding Ca²⁺ sensors that produce bright light emission in the visible range, over a large dynamic range of Ca²⁺ concentrations (0.1 µM–1.0 mM), and are insensitive to pH in the physiological range (Rogers et al., 2005; Curie et al., 2007; Martin et al., 2007). Compared with aequorin alone, the light emission of the chimeric construct with the optimized flexible linker region is more than 60-fold higher (Baubet et al., 2000). This Ca²⁺ sensor was named GA (GFP-aequorin), and is a powerful tool for functional mapping of neuronal circuits in vivo (Martin, 2008). GA offers a very high contrast, high signal-to-noise emission by virtue of the fact that light is emitted in the absence of illumination, and thus autofluorescence is non-existent. This lack of autofluorescence is an asset when imaging small structures deep within tissue. Another major advantage over fluorescence is that there is no phototoxicity and or bleaching of the fluorophore. While these are significant advantages, the major drawback of the GA luminescence imaging is the poor spatial resolution compared with fluorescence-based imaging. Temporal resolution however is most often adequate even with photon accumulation by integration over time (Pozzan and Rudolf, 2009; Drobac et al., 2010). A recent review compares the advantages and disadvantages of fluorescence imaging over bioluminescence imaging using GA (Martin, 2008).

The foregoing discussion illustrates the vigorous research activity in the development of tools for imaging Ca²⁺in vivo with good temporal and spatial resolution. While a number of Ca²⁺ sensors have been invented which have contributed to our understanding of cellular signalling in the context of intact animals, none have proved to be the all purpose, ideal indicator. Close attention should be paid to the choice of probe and the imaging technique that will suit the experimental requirements, in order to improve data quality and reduce toxicity. Development of probes that would allow imaging deep within tissue in vivo will prove to be essential.

Indicator loading techniques

Synthetic Ca²⁺ indicators

Imaging Ca²⁺ signals in vivo requires filling the cells with appropriate amounts of a fluorescent or luminescent Ca²⁺ indicator. Organic synthetic Ca²⁺ indicators are maximally fluorescent and have their optimal Ca²⁺ binding affinities in their ionized salt forms. These ionized species of dyes however are hydrophilic and therefore are not membrane permeant, and do not enter cells. To enable loading cells with the indicator noninvasively, Tsien and colleagues developed the esterified forms of the dyes (Tsien, 1981). Acetoxymethyl (AM) ester derivatives of the dyes, by virtue of their uncharged state at physiological pH were readily permeable across the cell membrane. Once in the cytoplasm, endogenous esterases in the cell de-esterified the dye into the charged species, making them both highly fluorescent and with high affinity to Ca²⁺ ions. De-esterified dye molecules are charged and are impermeant and therefore are trapped with the cell, thus achieving significantly higher concentrations than in the incubate. This method of loading synthetic fluorescent indicators into cells became routine for most applications. One major problem encountered in this type of cell loading is compartmentalization of dye into cellular organelles (Del Nido et al., 1998). This could be controlled to some extent by carefully optimizing incubation temperature and time for each cell and tissue type.

Microinjection of Ca²⁺-sensitive dyes was initially the method of choice, and single cells were injected with the indicator in cells in culture, in isolated tissues, as well as in intact animals (Sobel and Tank, 1994;Helmchen et al., 1999; Svoboda et al., 1999). While the luminescent Ca²⁺ indicator aequorin was the first compound microinjected into cells (Hastings et al., 1969; Baker et al., 1971), for in vivo imaging the synthetic fluorescent indicators were the first to be used (Sobel and Tank, 1994; Svoboda et al., 1999). Single cell microinjection is an excellent choice to image cells in vivo, since the cells of interest can be studied with high contrast, as the neighbouring cells are not filled. The major drawback is that only one cell is recorded at a time, and recording Ca²⁺ signals within cell ensembles is most often required to understand physiological signalling within tissue. This drawback can be overcome by a variation of this technique using bulk loading of cells with cell permeant dyes or by controlled permeabilization of cells at the vicinity of dye microinjection (see Figure 1B). Electroporation has been successfully used to fill a collection of cells (10–100) with chemical dyes such as calcium green-1 and Oregon green BAPTA-1 (OGB-1) (Nagayama et al., 2007). In the brain, this technique provides a powerful method for loading a local neural circuit for imaging in vivo. Embryonic brain cells were labelled in this manner in utero (Mank et al., 2008; Holtmaat et al., 2009).

Figure 1.

Different modes of transcranial imaging cortical cells using two-photon microscopy of a living mouse. A, Drawing of a fibre-optic endoscopic design for two-photon microscopy. The fibre is implanted and imaging is done in an awake, behaving mouse to image cortical neurons or glial cells. B, Imaging cortical cells through an acutely or chronically implanted cranial window. A glass coverslip is glued over a hole drilled through the skull using dental glue. In both cases, cells under the viewing port are labelled either by local injection of acetoxymethyl ester form of an organic dye or virus packaged fluorescent sensor, or transgenically expressed fluorescent Ca²⁺ sensor. C, Photograph of a mouse with an implanted cranial window. D, Photograph of transcranial two-photon imaging of an anaesthetized mouse with cranial window. These transgenic mice express the YC 3.60 cameleon in astrocytes under the control of the S-100b promoter. E, A field of astrocytes in the somatosensory cortex of the transgenic mouse imaged through the cranial window as in panel D. The image is an overlay of CFP and YFP channels. Images were acquired at 3 Hz. Scale = 50 µm. F, Spontaneous Ca²⁺ transients occurring in astrocytes numbered in panel E. Traces represent YFP/CFP ratios of intensities of pixels in regions of interest drawn around each cell, plotted against time. Data are from the author's laboratory.

Using cell-permeant forms of synthetic chemical Ca²⁺ indicators, ensembles of cells within tissue can be effectively labelled in living animals for imaging. This bulk loading approach has been used successfully in diverse tissues in many species of animals (see diagram in Figure 1B). These include C. elegans (Jain et al., 1993; Dal Santo et al., 1999; Kerr et al., 2000), Drosophila motor neurons (Karunanithi et al., 1997) zebrafish nervous system (Gahtan et al., 2002; Takahashi et al., 2002), and mammalian brain (Ohki et al., 2006; Takano et al., 2007; Shummers et al., 2008). In these experiments, cell-permeant AM ester forms of chemical dyes were microinjected into tissue in vivo. This results in labelling of a group of cells at the site of injection, which are then imaged using wide-field, confocal or two-photon microscopy techniques (Stosiek et al., 2003; Takano et al., 2007). Both the electroporation of cell impermeant indicators and bulk loading of cell permeant AM ester forms of Ca²⁺ indicators achieve universal labelling of all cell types in the vicinity of dye delivery. In the brain, this is a problem since both neurons and glial cells are simultaneously labelled. To overcome this problem a counterstaining approach has been used where specific cell types are stained in situ with a dye with different emission properties than the indicator. In bulk-loaded brain tissue, where both neurons and glial cells are labelled with indicator, sulfarhodamine 101 was used in conjunction to identify astrocytes. Astrocytes stained with sulfarhodamine 101 appear red, while OGB-1-AM, which fluoresces green monitors Ca²⁺ in both neurons and astrocytes (Nimmerjahn et al., 2004; Shummers et al., 2008).

Loading dye into specific neurons innervating a local area has also been achieved by two different means. In one, cell-permeant AM forms of chemical dyes were microinjected into axon tracks of known neural circuits which resulted in loading of the axons. Over time (4–24 h), the dye is transported in both anterograde and retrograde directions such that neuronal cell bodies and the nerve terminal regions were labelled, and these regions could be imaged without contamination of signal from non-neuronal cells (O'Donovan et al., 1993; Kreitzer et al., 2000). In another variation of this method, cell-impermeant, charged forms of dyes were microinjected into nerve fibre pathways such that some of the axons were damaged and the dye entered the cytoplasm which were then transported anterogradely to the cell bodies and imaged. Neural pathways in zebrafish spinal cord were injected with OGB-1 which resulted in specific labelling of neuronal cell bodies in the fish brain (Gahtan et al., 2002). Similarly, Mauthner cell circuit in the zebrafish was labelled with calcium green dextran and imaged in the living fish (Takahashi et al., 2002). These two specialized techniques for cell-specific dye loading are uniquely relevant only in the nervous system to image Ca²⁺ signals in collections of neurons with long processes, and may not be useful in other tissues.

Loading FPIs into cells

An alternative, newer technology for tissue and cell-specific indicator labelling is the development of genetically encodable, fluorescent, protein-based indicators (for reviews, see Knopfel et al., 2006; Kotlikoff, 2007; Martin, 2008). The advantage of the genetically encoded Ca²⁺ indicators compared with synthetic dyes is the possibility of targeting the indicator specifically to the cells of interest, using plasmid designs containing specific promoters, and targeting sequences. Furthermore, this allowed for long-term imaging of Ca²⁺ signals discretely, not only in selected cell populations, but also in specific subcellular compartments in intact living animals. In addition, this allows for repeated imaging of the same tissue in the same animal. Both transfection techniques and transgenic technologies were readily available for long-term stable expression of indicators in the tissue of choice.

The most commonly used method to target FPIs to cells in animals is using transgenic methodologies (Heim et al., 2007; Hendel et al., 2008). Plasmids are engineered with FPI coding sequences with appropriate promoters to specifically target cells of interest in the model organism. Plasmids were then delivered to recipient cells using pronuclear injection, homologous recombination or retroviral transfer to early-stage embryos. Transgenic targeting of fluorescent proteins to specific cell types has been achieved to every organ in small invertebrates like Drosophila and large mammals, including primates. It is critical to choose particular promoters to avoid leaky expression to other cell types than the intended target. In addition, the level of expression of the fluorescent protein depends on how strong the promoter expression is. Ubiquitous expression of FPIs in all cells has also been carried out using promoters found in most cell types in the body such as β-actin (Nagai et al., 2004). Specific expression of cameleons in neurons was achieved in transgenic mice using the promoters for the thy1 protein (Heim et al., 2007). One caveat in using transgenic technology for cell-specific expression is that many promoters have strong context-sensitive expression patterns depending upon the site of transgene insertion. Therefore, expression patterns in different founder lines can differ widely. However, since large numbers of founders can be achieved, transgenic technology remains the method of choice for expression of fluorescent proteins in specific genetic lines.

DNA constructs containing appropriate promoters engineered with FPI coding sequences can also be packaged into plasmids and were transfected into cells using cell permeabilizing techniques such as liposome-mediated transfer or electroporation. Electroporation was used in utero to deliver plasmid constructs that direct expression of fluorescent proteins to embryonic brain (Petreanu et al., 2009). Fluorescent proteins and FPIs have been directed to specific cells and cell groups both in brain and other tissues using viral vectors. In one experiment, YC 3.60 cameleon was packaged into an AAV (adenoassociated virus) vector to direct the FPI expression in neurons, which resulted in robust expression in various brain regions in mice. In vivo imaging of neural activity was successfully carried out in freely moving mice using a fibre-optic-based single-photon optical system (Wallace et al., 2008; Lutcke et al., 2010). In another study, pseudoherpes simplex virus vector was used to package another FPI, troponeon, and was delivered to cardiac muscle cells in a dog (Prorok et al., 2009). Cell-specific expression using viral vectors are possible by using specific targeting strategies such as using Cre recombinase. A Cre-recombinase-dependent AAV was used to target a fluorescent protein in Cre-expressing transgenic mice to achieve specific labelling of cells in mouse brain (Petreanu et al., 2009).

Synthetic organic dyes versus FPIs

One major advantage of using synthetic chemical indicators over FPIs is the wide range of indicators with widely differing Ca²⁺ affinities that are readily available commercially. It is also possible for the experimenter to introduce these dyes into cells with ease and use in experiments. Cell-loading conditions and protocols have been optimized over a number of years. Furthermore, in some instances, the synthetic organic Ca²⁺ indicators are superior to FPIs, in terms of the sensitivity, light throughput, signal-to-noise ratio, and dynamic range are considered. One particular disadvantage of organic synthetic dyes is they cannot be specifically targeted to particular organelles or compartments in the cell. Although a number of strategies have been attempted to solve this problem, the results have been less than optimal (see Paredes et al., 2008; Pozzan and Rudolf, 2009). In addition, they are at times compartmentalized into cellular organelles rendering them inaccessible in the cytoplasm (Del Nido et al., 1998). To date, however, when signals deep within tissue need to be imaged with adequate signal-to-noise, at rates high enough to capture physiologically relevant events, synthetic dyes such as OGB-1 has been the indicators of choice (Grewe et al., 2010).

Genetically encodable FPIs, however, are easily targeted to specific cell types in different organs in the body (Heim et al., 2007), as well as to specific organelles such as endoplasmic reticulum or mitochondria (Pozzan and Rudolf, 2009). A major advantage of genetically encoded FPIs compared with synthetic dyes is the possibility of performing experiments over the long-term in vivo both at the single cell level and at subcellular levels (Kotlikoff, 2007). Furthermore, expressing FPIs within cells through genetic manipulations allows for repeated imaging of cells and circuits in the same animal over long periods of time. Thus both issues of development and functional plasticity can be studied. A number of different classes of chimeric designs of GFP and its mutated congeners have been engineered as useful Ca²⁺ indicators. Some of the new dyes such as G-CaMP3, TN-XXL and GA might turn out to be most useful tools for in vivo imaging. While much effort has yielded a number of fluorescent protein-based indicators, they still need further improvement. Currently available FPIs have narrow dynamic ranges compared with synthetic dyes, and in addition have slow response times. Though binding and unbinding time constants are not available for many of the FPIs, functional data suggest that under in vivo conditions, the kinetic properties of modern FPIs are not dramatically different from synthetic dyes like OGB-1. The decay time constant of Ca²⁺ transients in neuronal cell bodies measured in vivo using CerTN-L15 was 1.68 s (Heim et al., 2007), which is only approximately twofold higher than the one measured with OGB-1 at 0.82 s (Kerr et al., 2005). Indeed using d3cpv, a calmodulin and M13-based Ca²⁺ sensor expressed in mouse brain Ca²⁺ responses to single action potentials were reliably recorded as long as the firing rate did not exceed 1 Hz (Wallace et al., 2008). The most recently developed G-CaMP3 appears to be well suited for imaging deep within brain tissue. This dye was recently utilized for imaging in motor cortex of mouse brain, and in a head-fixed fruit fly with good signal-to-noise ratios (Tian et al., 2009; Seelig et al., 2010). It would be important to make further improvements to increase the brightness, dynamic range and reduce response time constants of responses in FPIs such that physiological signals can be monitored in real time.

Imaging techniques

The earliest recordings of in vivo Ca²⁺ signals were made in preparations that were easily accessible for imaging. These were in invertebrates and small, transparent lower vertebrates and included C. elegans (Jain et al., 1993), auditory system of the cricket (Sobel and Tank, 1994), olfactory system of the blowfly (Borst and Egelhaaf, 1992), honeybee (Galizia et al., 1999) and zebrafish (Friedrich and Korsching, 1997). These early experiments on intact animals utilized organic synthetic Ca²⁺ indicator dyes and epifluorescence microscopy using CCD cameras. The use of zebrafish embryos was adopted early since they are small, transparent vertebrates and accessible to confocal microscopy when loaded with synthetic organic dyes (Ritter et al., 2001). In combination with injection of indicator dyes and epifluorescence or confocal microscopies, various superficial structures, particularly sensory systems were investigated. The choice of these structures in invertebrate or small vertebrate animal models was mainly due to the extremely limited depth that could be imaged by conventional imaging techniques. Light scattering prevents recording fluorescence changes in cells using epifluorescence and confocal microscopy more than about 20 µm below the surface.

The advent of multiphoton laser scanning microscopy added a powerful tool for imaging cellular and subcellular structures even in highly scattering tissue. The concept of two-photon confocal microscopy was demonstrated by Denk et al. (1990), when they showed that fluorophores in living tissue can be excited with a stream of femtosecond infrared laser pulses. This principle was implemented in confocal microscopy and was quickly adopted by biologists for imaging isolated tissue and intact animals. This technology is undergoing continuous major technical developments. One important advantage of two-photon laser scanning microscopy is that it allows for imaging cells several hundred microns deep within tissue, enabling high-resolution functional imaging in living animals, including mammals. Innovative scanning technologies have been developed for ultrafast time-resolved imaging of large regions of tissue comprised of many cells in 3D space (Salome et al., 2006; Duemani Reddy et al., 2008; Grewe and Helmchen, 2009).

In vivo Ca²⁺ imaging using two-photon confocal microscopy is almost always implemented on a fixed-stage, upright microscope platform, which requires the animal to be anaesthetized and immobilized with the imaged region exposed to the objective. Although motion due to respiration and pulse cannot be avoided, fixing the animal minimizes motion artefacts. Over the last decade a large number of functional Ca²⁺ imaging studies have been carried out using this approach in a number of organ systems and animal models (Svoboda et al., 1999; Takano et al., 2007; Mostany and Portera-Cailliau, 2008; Shummers et al., 2008; see for review Helmchen and Kleinfeld, 2008). Most of this work focused on imaging activity in the brain, and a number of advances were realized. Imaging neuronal populations allows for simultaneous recordings of spontaneous and evoked responses in a large number of cells such that function of entire circuits can be inferred. By staining large populations of cells using strategies of injecting a bolus of cell permeant form of organic dyes (Garaschuk et al., 2006), injection of dextran conjugated dye into axon bundles (Hirase et al., 2004), or using viral infection using FPI packaged into viral vectors (Mank et al., 2008), large populations of neurons can be imaged using these methods for simultaneous recordings of spontaneous activity (Adelsberger et al., 2005) as well as evoked responses (Tabor et al., 2004). Various brain regions such as the olfactory bulb (Wachowiak et al., 2004), cerebellum (Diez-Garcia et al., 2007), visual cortex (Ohki et al., 2006) and somatosensory cortex (Stosiek et al., 2003) have been imaged. In the brain, the bulk loading of cells followed by two-photon microscopy allowed for single cell resolution and revealed the precisely ordered pinwheel centres in the visual cortex of cats (Ohki et al., 2006). Recent experiments in mice using particularly organic Ca²⁺ indicators such as OGB-1, and ultrafast imaging techniques neural network activity was recorded in collections of cells at rates exceeding 0.5 kHz (Grewe et al., 2010; Lutcke et al., 2010). Ca²⁺ transients in response to single action potentials were recorded with signal-to-noise ratios in the order of 2–5 enabling determination of spike times with millisecond precision.

Because the zebrafish embryos are transparent, they were most suitable for developmental studies. Two-photon imaging of larvae revealed that retinal ganglion cell axonal arbors in the optic tectum develops rapidly and this development depends upon activity-dependent competition between neighbouring axons (Hua et al., 2005). A separate two-photon imaging study of tectal neurons showed that soon after dendritic outgrowth and synaptogenesis, most visual properties were already fully established. This development did not require complex visual experience (Niell and Smith, 2005). This type or in vivo two-photon Ca²⁺ imaging in the visual cortex and olfactory bulb of zebrafish embryos were most useful in establishing the functional topology of neuronal activity patterns, and also allowed for analysis of the role of activity in circuit development (Wachowiak et al., 2004; Yaksi et al., 2007).

One major problem with bolus loading of large populations of cells is that all the different types of cells such as neurons and glial cells in the region are labelled. Different strategies are available to solve this problem at least partially. As mentioned earlier, sulfarhodamine 101 was successfully used to identify astrocytes during imaging experiments (Nimmerjahn et al., 2004; Winship et al., 2007). Another method is labelling by microinjection of axonal bundles with dextran conjugated indicators and using axonal transport over several hours to both retrogradely and anterogradely label neurons prior to imaging (Hirase et al., 2004). This approach has been successfully used in the olfactory bulb (Oka et al., 2006) and in the cerebellar olivary nucleus (Kreitzer et al., 2000). This approach of retrograde labelling the olfactory system in the Drosophila revealed that odorant receptor-specific glomeruli showed different ligand spectra and much higher sensitivity in vivo compared with similar experiments in vitro (Oka et al., 2006). The problem of specific labelling is also effectively solved by the use of genetically encodable FPIs to target cells by the use of cell-specific promoters (Knopfel et al., 2006; Mank et al., 2008; Atkin et al., 2009). Figure 1E,F shows an example of two-photon imaging of astrocytes in the somatosensory cortex specifically expressing YC 3.60 sensor in a transgenic mouse (see also Atkin et al., 2009). This approach was successfully used to study parallel fibre pathway activation in the cerebellum in mice expressing G-CaMP2 in neurons in vivo (Diez-Garcia et al., 2007). These authors succeeded in measuring Ca²⁺ transients in beams of parallel fibres through the intact skull. In another study, the cameleon YC 3.1 was expressed in Drosophila flight muscles to record Ca²⁺ transients in the myoplasm while simultaneously recording the power output of the muscle (Gordon and Dickinson, 2006). Similarly G-CaMP2 was expressed in the cardiac muscles in transgenic mice to record Ca²⁺ transients associated with the heartbeat in developing hearts within embryos (Tallini et al., 2006). The authors were able to describe the organization of atrioventricular conduction and the development of the atrioventricular node. In the brain, the non-invasive nature of in vivo Ca²⁺ imaging research alters the research focus to recording from central nervous system microcircuits, opening the door for systems neuroscience. Furthermore, technical innovations yielded methods for routine transcranial imaging (Helmchen and Kleinfeld, 2008; Mostany and Portera-Cailliau, 2008), including repeated imaging of the same neural circuits in a given individual animal over a period of weeks and months (Holtmaat et al., 2006; Mostany and Portera-Cailliau, 2008). Cortical circuits were imaged in a head-fixed position through a transparent cranial window, implanted under anaesthesia. These methods proved to be a boon to the burgeoning research in the neurosciences over the last decade allowing investigation of communication within neural microcircuits in the brain.

More recently, efforts have been underway to develop techniques to image tissue with cellular level resolution in awake, behaving animals. In one method, transcranial imaging of cellular Ca²⁺ signals was carried out in an anaesthetized, head-fixed mouse to study signalling in the cortical layers (Dombeck et al., 2007). Ca²⁺ signals were recorded in layer 2/3 cortical pyramidal neurons and astrocytes loaded with calcium green-1 in response to sensory stimuli. Technological innovations are sorely needed to improve imaging in unanaesthetized, freely moving animals. Several technical hurdles need to be overcome to realize this goal. First, the microscopes must be sufficiently small to be borne by the animal which is freely mobile. Second, the device should retain the functionality of the conventional microscope. Thirdly, the instrument must be robust enough to minimize voluntary and involuntary tissue motion artefacts. Realizing all three of these goals has been difficult, but work towards them took a decisive step forward with the invention of a fibre-optic-based miniature two-photon microendoscope which could be mounted on a rat or a mouse (see Figure 1A for diagram) (Helmchen et al., 2001). This device incorporated fibre optics to bring in both the excitation beam and to collect the emitted light from cortical pyramidal neurons in awake animals (Helmchen et al., 2001). Although this form of imaging suffers from restricted field of view and low spatial resolution, it offers excellent temporal resolution, and can report from large cellular ensembles (Grewe and Helmchen, 2009; Grewe et al., 2010). In the brain, microendoscopic, two-photon microscopy affords monitoring synchronous activity within entire neural circuits, and can be used to study any region of the brain (Adelsberger et al., 2005). The microendoscopes use a novel approach involving the use of micro-electro-mechanical scanners coupled with gradient-index (GRIN) lenses on optical fibres has been developed (Gobel et al., 2004; Engelbrecht et al., 2008; Murayama and Larkum, 2009). These engineering developments made miniaturizing the scanning device possible while allowing for a fibre-optic tether. Since the GRIN lens refracts light through a refractive index gradient, its optical surfaces are flat and allows for efficiently focusing and collecting light, while keeping the overall form factor small. While much work is focused on recording Ca²⁺ signals in neural circuits in the brain, microendoscopy in non-neural tissue has remained in the sphere of diagnostic endoscopy, and is routine in the diagnosis of gastrointestinal and airway pathologies (see Tsesmeli and Coumaros, 2009 for review). Thus far, fibre-optic-based microendoscopy systems have not achieved comparable optical image quality to that of stage-mounted, fixed microscopes in recording cellular Ca²⁺ signals in tissue. Technical innovation to improve fibre-optic, microendoscopic, two-photon microscopy is necessary since imaging awake, freely behaving animals is ultimately required to address many physiological questions in in vivo studies.

Recently, in a technical tour de force, scientists have developed methods to train mice to cooperate in imaging experiments using two-photon microscopy (Sato and Svoboda, 2010). Svoboda and colleagues have carried out transcranial imaging in head-fixed, awake behaving mice and succeeded in recording Ca²⁺ signals from neural circuits in the barrel cortex associated with sensory-motor activity (Komiyama et al., 2010; O'Connor et al., 2010). In one study, the functional properties of barrel cortex neurons projecting to the primary motor cortex was imaged. The authors found that neurons projecting to the primary motor cortex have large receptive fields showing that intermingled neurons in the primary sensory areas send specific stimulus features to different parts of the brain (Sato and Svoboda, 2010). In another series of experiments, mice were trained to respond with a specific behaviour, that is, lick for a water reward, in response to particular odours and not produce the behaviour (not lick) for another odour. Cortical neurons were imaged during this behaviour through a cranial window, and the authors were able to identify two non-overlapping tongue motor cortex areas (Komiyama et al., 2010). Many neurons in these regions showed modulation coinciding with or preceding the licking action, consistent with their involvement in motor control. Furthermore, they found that nearby neurons showed profound coincident activity and that this temporal correlation in activity increased as task learning progressed, suggesting a learning related structural plasticity (Komiyama et al., 2010). The same group of researchers, in another set of experiments, trained mice to respond to a whisking cue with a specific behaviour (lick water) and not lick if the whisking cue was out of reach of the whiskers (O'Connor et al., 2010). Mice had to learn to lick if the pole was in the go position (within reach of the whiskers) and not lick when the pole was in the no-go position (out of reach of whiskers). Cortical circuit activity was recorded as sensory neuronal Ca²⁺ signals to map the local circuit that regulates sensory-motor function. Mice learnt this complex task in 1–2 weeks and performed localizations using a sophisticated strategy that is expected to yield large differences in barrel cortex activity patterns for the different trial types. Mice maximized whisker contact with the go stimulus and minimized contact with the no-go stimulus. It is likely that this motor strategy produces very different patterns of activity across the barrel cortex for go and no-go trials, and this might help the mouse choose between go and no-go responses (O'Connor et al., 2010).

In a parallel development, innovations in the use of chimeric aequorin-GFP (GA) constructs have been advanced by a group of researchers for in vivo Ca²⁺ imaging in awake freely moving animals (Curie et al., 2007); (see for review Martin et al., 2007). The luminescent biosensor is genetically encoded into the tissue of interest, and the animals are imaged using high sensitivity electron-multiplying CCD cameras (Rogers et al., 2008; Roncali et al., 2008). Awake, behaving animals are imaged over long periods of time albeit with relatively poor spatial resolution. While this method affords reasonable temporal resolution, cellular resolution in space is rarely possible, which is a major disadvantage. The signal-to-noise, however, is excellent due to the total absence of autofluorescence, since no optical excitation is necessary. Ca²⁺-induced aequorin luminescence is amplified by resonance energy transfer to GFP (Shimomura et al., 1993; Curie et al., 2007). This approach will have to await the invention of super-sensitive image capture devices to realize the spatial and temporal resolution necessary to make it a useful tool for in vivo Ca²⁺ imaging.

Ca²⁺ imaging in disease states

Ca²⁺ imaging has been used to study various disease states for some time using animal models in in vitro studies, mostly in cell culture ad brain slice preparations. Recently, however, novel imaging methods with cellular resolution have been used more and more for investigation of disease models in vivo with the goal of a detailed characterization of the pathophysiology. Since many disease states are not faithfully reproduced in in vitro models, this development holds much promise. Furthermore, the problems in depth penetration during imaging prevents access to deeper regions of the brain and other tissues. However, the availability of genetic tools significantly adds to development of in vivo imaging techniques to investigate disease states. A wide variety of transgenic mice are now available, which are models of neurological disease, and mice with specific cell types or cellular compartments labelled with indicators have become valuable tools.

Two-photon Ca²⁺ imaging has been used in the study of glial cells signalling in the brain (Hirase et al., 2004; Atkin et al., 2009). Recent in vivo studies showed that astrocytes show rapid Ca²⁺ transients upon sensory stimulation, and these signals are perfectly tuned to the neuronal signals in the vicinity (Wang et al., 2006; Shummers et al., 2008). Astrocytes also have a role in regulation of local blood flow by regulating blood vessel diameter (Takano et al., 2006). Astrocytes are involved in many normal and pathological brain functions and may represent a promising therapeutic target. Astrocytes have been implicated in epilepsy, stroke and Alzheimer's disease (Tian et al., 2005; Takano et al., 2007). A recent in vivo study of an Alzheimer's disease mouse model revealed a number of new observations including rapid development local neuronal toxicity, appearance of dysmorphic neurites and microglial infiltration (Meyer-Luehmann et al., 2008). In a mouse model of Alzheimer's disease, astrocytic Ca²⁺ homeostasis and Ca²⁺ signals were recorded in vivo using OGB-1 as the indicator (Kuchibhotla et al., 2009). Synchronously coordinated, rapidly spreading Ca²⁺ waves were recorded in these animals (Kuchibhotla et al., 2009). The Ca²⁺ waves originated near Alzheimer's plaques and radially spread over hundreds of micrometers. A recent review summarizes much of the imaging data on Alzheimer models available to date (Dong et al., 2010).

Progressive axonal and neuromuscular junction alterations in mice with different gene mutations were identified by fibre-optic-based confocal microendoscopy in vivo (Wong et al., 2009). This group were able to visualize YFP expression in peripheral nerves and follow motor neuronal degeneration in Wld^s (Wollerian degeneration spontaneous mutant) mice. In addition, the modification of motor neuron degeneration in Wld^s mice by mutating N-ethyl-n-nitrosourea was characterized. Furthermore, in the same study, motor neuron degeneration in Sod1 mutation was also followed by confocal imaging (Wong et al., 2009). Two-photon confocal microscopy was applied to visualize changes in microvasculature in the brain more than a decade ago (Kleinfeld et al., 1998). Using this approach the development of stroke and its pathophysiology have been studied using two-photon microscopy. These studies have revealed the rapid changes to dendritic spine structure during ischaemia (Zhang and Murphy, 2007). The extent of the functional recovery upon restoration of normal blood flow after ischaemia was also characterized. Furthermore, this approach allowed for rapidly alternating between ischaemia and reperfusion while monitoring functional alterations in the brain. This study revealed that the apical dendritic structure in layer V cortical neurons was dramatically degraded by ischaemia (Murphy et al., 2008).

Epilepsy is another common neurological disorder that has been investigated using in vivo Ca²⁺ imaging. Epilepsy is characterized by recurrent seizure activity caused by bursts of synchronous and rhythmic neural circuit hyperactivity. Intrinsic optical signals were monitored in the very first in vivo studies of epileptic seizures (Schwartz and Bonhoeffer, 2001).Two-photon imaging was used in brain slice preparations in models of epileptic seizures to achieve single cell resolution (Badea et al., 2001). In vivo Ca²⁺ imaging was first used for investigation of astrocytic activity during seizure events (Tian et al., 2005). Two-photon Ca²⁺ imaging in vivo showed that astrocytic Ca²⁺ signals were strongly correlated with the epileptogenic neuronal bursting activity in this animal model (Tian et al., 2005). In a recent study, an in vivo seizure model was developed in an albino Xenopus laevis tadpole where pentylenetetrazol-induced seizure activity was directly monitored (Hewapathirane et al., 2008). In vivo Ca²⁺ imaging is being increasingly used to investigate a number of neurological diseases in animal models. It is quite likely that similar experimental approaches would be valuable in other organ systems as well. Two-photon microendoscopy designs could become potentially valuable tools to investigate the pathophysiology and disease progression in other metabolic and endocrine disorders.

While in vivo Ca²⁺ imaging has become a centrepiece of research in physiological and cell biological processes, it is only rarely used as a tool in pharmacological research. High-throughput Ca²⁺ imaging is increasingly utilized in multi-well sample configurations in drug discovery research, particularly for screening of unknown compounds (Miret et al., 2005). In vivo Ca²⁺ imaging could potentially play an important role in pharmacological characterization of potential drugs. It might be possible to record drug effects on cellular Ca²⁺ signals in vivo in transgenic animals expressing fluorescent Ca²⁺ indicators in specific tissues. For example, animals expressing genetically encodable Ca²⁺ indicators in neurons or astrocytes can be imaged to monitor actions of a given drug on activity in specific networks in accessible regions of the cortex. Similarly, in vivo imaging of different splanchnic organs with a multiphoton microscope or a fibre-optic-based microendoscope may be possible to monitor drug effect on the liver, pancreas and kidney. The newly developed luminescence-based indicators such as GA would also be useful to image internal organs in real time albeit in somewhat lower resolution than with fluorescence imaging. However, very little published work exists on this line of work in the literature until now.

Technical considerations

As discussed earlier, there are a number of options available for filling cells with Ca²⁺ indicators for in vivo Ca²⁺ imaging. Of these, filling single cells by microinjection provides the most discrete, high signal-to-noise, high contrast fluorescence signals (Sobel and Tank, 1994; Helmchen et al., 1999). However, since most in vivo imaging applications target cell–cell communication and signalling within cellular ensembles, it is often necessary to fill populations of cells in local areas within tissue. In the brain, most investigations target local microcircuits that contain in the order of 1000–10 000 cells which are synaptically connected. This is achieved by various means, and the most often used is multi-cell bulk loading with either cell permeant synthetic organic dyes (Stosiek et al., 2003; Shummers et al., 2008), passive diffusion techniques (Zimprich et al., 1998; Takahashi et al., 2002) or local electroporation (Nagayama et al., 2007). Genetically encodable, fluorescent, protein-based indicators are expressed in cells using local application of plasmids followed by electroporation (Tabata and Nakajima, 2001; Mank et al., 2008; Holtmaat et al., 2009), or using plasmids packaged in viral vectors (Dittgen et al., 2004; Mank et al., 2008; Holtmaat et al., 2009). In all of these methods, cell populations are filled with indicators. The most selective labelling is generally achieved through transgenic technologies using any one of the genetically encodable Ca²⁺ sensors engineered to be expressed under the control of cell-specific promoters (see Mank et al., 2008).

Imaging in vivo with cellular resolution requires data sampling from many hundreds of cells with good temporal resolution, and this is not trivial. In normal scanning mode, the laser is scanned within one focal plane in a raster-like fashion. Using this approach, data acquisition is slow, since the scan pattern needs to be repeated several times to image large fields. A number of alternative strategies have been developed to achieve more flexibility in imaging large areas in three dimensions. Entirely new scanning regimes need to be developed to increase speed of data collection and to acquire images in 3D space, since local neural circuits in the brain are distributed in 3D volume of brain tissue (Grewe and Helmchen, 2009; Grewe et al., 2010). In order to image physiological signals in such circuits with good temporal resolution, novel high-speed xy scan techniques were developed (Gobel and Helmchen, 2007; Grewe et al., 2010; Lutcke et al., 2010). A recent review elegantly addresses many of the imaging techniques and the technical innovations that were developed to realize them (Gobel and Helmchen, 2007; Tian et al., 2009; Grewe et al., 2010). Innovations using resonant scanners such as acousto-optical devices (AODs) for random access scanning in 2D, and the use of multiple AODs, provided inertia-free, fast scanning in 3D (Hammond and Glick, 2000; Nguyen et al., 2001; Gobel and Helmchen, 2007; Grewe et al., 2010). With fibre optic illumination to predestined locations by positioning fibre tips appropriately, 3D measurements of in the order of 1000 cells with 100-Hz temporal resolution was achieved (Rozsa et al., 2007; Lutcke et al., 2010). These technological innovations have greatly extended the utility of two-photon laser scanning microscopy for in vivo imaging.

Steady technical innovations over the last decade have provided the tools necessary for imaging Ca²⁺ signals in the living animal with near real-time temporal resolution in 3D space deep within tissue. In conjunction with multi-cell bolus loading and two-photon laser scanning microscopy, large neuronal populations in the brain have been imaged with cellular resolution (Winship et al., 2007; Shummers et al., 2008; Grewe et al., 2010; Lutcke et al., 2010). The capability to image large scale structures revealed the extraordinary variability of odorant receptor maps in the olfactory bulb in mice and led to the finding that the receptor-defined olfactory glomeruli in vivo showed very high sensitivities compared with pericams in in vitro experiments (Oka et al., 2006). The availability of methods for in vivo Ca²⁺ imaging provided the means for directly imaging astrocytes as they monitor neuronal activity in a functioning neural network in the brain, providing the first direct experimental evidence for neuronal-glial cell signalling (Winship et al., 2007; Shummers et al., 2008). Imaging specific cell types in tissues was achieved with the different genetically encodable indicators such as G-CaMP2, G-CaMP3 and TN-XXL which can be targeted to defined populations of cells using specific promoters in transgenic animals (Hendel et al., 2008; Lutcke et al., 2010). This allows for discretely recording activity within defined cellular networks in the brain such as recording parallel fibre-Purkinje cell synapses in the living animal. Ca²⁺ signals in astrocytic networks in response to neural activity for sensory stimulation (forepaw stimulation) was imaged in transgenic mice expressing the YC 3.60 cameleon in astrocytes (Atkin et al., 2009). While these genetically encoded probes initially did not perform as well in mice compared with C. elegans or zebrafish, the new constructs are superior and have been shown to detect Ca²⁺ signals in response to single action potentials and responses to very low network firing rates (Tian et al., 2009; Lutcke et al. 2010). In the brain, detection of Ca²⁺ signals from dendrites and dendritic spines is extremely difficult because of very low indicator content within them. Image contrast is also poor, since in addition to cells, the many fine processes in the surrounding neuropil are also stained. Single cell microinjection was the only technique that provided sufficient contrast in isolated tissue slice preparations, which is not easily done in living animals. This problem may now be partially solved with the use of novel indicators and ultrafast 3D volume imaging in two-photon laser scanning microscopy (Gobel and Helmchen, 2007; Tian et al., 2009; Grewe et al., 2010; Lutcke et al., 2010). In this respect organic indicators are superior for high-speed imaging of Ca²⁺ transients in response to action potential trains. OGB-1 has been successfully used in a number of recent studies to record activity in neuronal microcircuits at rates better than 200 Hz (Grewe et al., 2010). In vivo optical imaging of brain function has benefited most from the developments in two-photon laser scanning microscopy, and similar advances in imaging Ca²⁺ signals in vivo are eagerly anticipated in other organ systems.

Future outlook

Most of the effort in imaging cellular Ca²⁺ signals in vivo has been focused on the nervous system. Now that the instrumentation and methods have been developed for brain imaging, they can be applied to other organ systems in the near future. In vivo Ca²⁺ imaging potentially could be used to monitor altered signalling in disease states including cancer, and such detection systems could provide a platform for pharmacological testing and drug development. In the brain, in vivo Ca²⁺ imaging is proving to be an invaluable tool to investigate disease states that were not accessible to in vitro experimentation, because isolated brain slices from diseased and aged brain are of poor quality. In the nervous system, combining Ca²⁺ imaging with electrophysiology and behavioural approaches is poised to yield important data for understanding the neural code for complex behaviours including learning and memory. Further advances in cellular labelling and imaging techniques, especially ultrafast 3D imaging tools, are imminent and expected to enable imaging of large populations of cells in the brain. Improvements in indicators, including genetically encoded indicators, will facilitate functional characterization of microcircuits in the brain and animal models of neurological disorders. Although two-photon laser scanning microscopy provides better depth penetration compared with wide-field microscopy, in vivo two-photon imaging is still restricted to very superficial layers of a mouse brain. Improved technologies would be necessary for imaging in deeper layers of the brain and indeed other tissues. Improvements in indicators and image sensors would be required to achieve this goal. Bioluminescence-based probes such as GA could become useful for cellular level imaging deep within tissue if appropriate light transmission technologies and super-sensitive sensors with large pixel formats become available. Optimization of fluorescence signal detection and improved signal analysis tools are expected and should provide methods for single spike detection in 3D neural networks. Improved brightness and signal-to-noise in indicators should help to accelerate development of methods to image tissue in awake, freely moving animals. In this regard, the optimization of microendoscopic design and miniaturization should spur in vivo imaging in all organ systems, particularly the brain. The availability of microendoscopes should make in vivo imaging generally possible for all organ systems since trans-integumental imaging using endoscopes are already in routine diagnostic use. In vivo imaging using microendoscopic, two-photon microscopy would be expected to provide experimental models for testing effectiveness of drugs in specific disease states such as epilepsy, cancer, animal models of diseases of the liver and pancreas. Finally, in vivo imaging of animals trained for imaging without anaesthesia while engaged in complex behaviours is expected to yield important data on physiological communication within cellular networks in different organ systems.

Glossary

Abbreviations

BRET

bioluminescence resonance energy transfer

CFP

cyan fluorescent protein

FPI

fluorescent protein indicators

FRET

foster resonance energy transfer

GFP

green fluorescent protein

YFP

yellow fluorescent protein

Conflict of interest

I can categorically certify that there are no conflicts to declare.

Supporting Information

Teaching Materials; Fig 1 as PowerPoint slide.