|Home | About | Journals | Submit | Contact Us | Français|
An organism arises from the coordinate generation of different cell types and the stereotypical organization of these cells into tissues and organs. Even so, the dynamic behaviors, as well as the ultimate fates, of cells driving the morphogenesis of an organism, or even an individual organ, remain largely unknown. Continued innovations in optical imaging modalities, along with the discovery and evolution of improved genetically-encoded fluorescent protein reporters in combination with model organism, stem cell and tissue engineering paradigms are providing the means to investigate these unresolved questions. The emergence of fluorescent proteins whose spectral properties can be photomodulated is one of the most significant new developments in the field of cell biology where they are primarily used for studying protein dynamics in cells. Likewise, the use of photomodulatable fluorescent proteins holds great promise for use in developmental biology. Photomodulatable fluorescent proteins also represent attractive and emergent tools for studying cell dynamics in complex populations by facilitating the labeling and tracking of individual or defined groups of cells. Here, we review the currently available photomodulatable fluorescent proteins and their application in model organisms. We also discuss prospects for their use in mice, and by extension in embryonic stem cell and tissue engineering paradigms.
Starting off as a single cell, embryonic development relies on the generation of a tremendous diversity of cell types in a variety of locations, and requires dynamic cell behaviors that are precisely regulated. Cells become organized into different layers and tissues, often as a result of their division, migration or death. Therefore, knowledge of the dynamics and fate of cells during embryonic development and organogenesis through generation of comprehensive three-dimensional maps compiled at cellular resolution has long been a goal for developmental biology and tissue engineering. To this end, live imaging methods for following development and organogenesis at the level of individual cells are actively being developed.
A variety of approaches have been established for labeling and following single cells or populations of cells on model organisms. These include grafts of genetically-distinct, usually reporter expressing, cells or tissues.1,2 Vital dyes which are usually introduced by injection at the site of interest.3 Most recently electroporation of nucleic acids or proteins into cells or tissues has also found widespread application and been used to study cell dynamics in embryos, organ explant cultures or tissue engineering paradigms.4 In some animal systems such as the zebrafish an additional level of spatiotemporal control of cell labeling is possible because certain vital dyes are available in photoactivatable “caged” forms which can be “uncaged” and so rendered fluorescent when exposed to light of an appropriate wavelength. When injected into embryos, fluorescent cells can be followed and imaged over time during normal development and contrasted with the results of experimental perturbations, for example in mutants.5,6
In the mouse, pulse-chase labeling and tracking of populations of cells has also exploited transgenic technologies such as the binary transgenic genetic induced fate mapping (GIFM) strategies or the generation of mosaics in the form of chimeras.7–9 However, most of these techniques are usually either invasive to the embryo or only effective when cells or tissues of interest are easy accessible to manipulation, limiting their use to cells that are superficially located. Furthermore, with each of these methods it has been a challenge to label a precise spatially-defined cell population or even single cells. Importantly, many of these established techniques often only provide a static picture of morphogenetic events documenting the location of labeled cells and their final position.
The combination of advanced image acquisition and analysis techniques facilitates the investigation of large, dynamic cell populations within complex specimens such as developing embryos.10,11 These imaging approaches represent a unique opportunity to study embryonic morphogenesis and organogenesis in situ, from the level of cellular processes to the scale of an entire organism or organ (in toto imaging).10 Recent advances in the isolation and engineering of genetically-encoded fluorescent proteins (FP) have accelerated the development of bright, high-contrast live imaging reporters which can be used in studies of cell fate and dynamics. Often native (cytosolic) reporters are used, however, subcellularly-localized FPs provide a means to achieve greater cellular resolution.10,12,13 Moreover, when expressed under defined cis-regulatory elements both cytosolic and subcellularly-localized fluorescent proteins provide greater cell type specificity.14–17
The emergence of photomodulatable FPs (PMFPs) has provided the opportunity to non-invasively and selectively label and trace cells or proteins at regionally-defined spatio-temporal resolution. PMFPs comprise two main categories: photoactivatable (PA) and photoconvertible (PC) FPs. Of these reporter types there are also categories of those that are permanent or reversible in their photomodulation. Due to a conformational change in their fluorophore, PAFPs change from a non-fluorescent state to a fluorescent state upon irradiation with short wavelength light, whereas PCFPs convert from one fluorescent state to another with an accompanying change of color and leading to change in their emission and excitation spectra towards a longer wavelength.18,19 In this way cells of interest expressing these reporters can have fluorescence selectively activated or converted within a spatially-defined region of interest (Fig. 1). These physical properties make PMFPs attractive tools for the precise and non-invasive labeling of cells. With the ability to modulate their spectral characteristics in specific cohorts of cells or a large population of cells of interest, one has a greater control of labeling specificity. To date, several PMFPs and their variants have been reported from diverse marine species, with ongoing efforts focused on improving the physical properties of existing PMFPs, as well as identifying additional ones.
The goal of this review is to discuss PMFPs that are currently available (Table 1) with a view to their application for studying cell dynamics in mice, the preferred genetically-tractable mammalian model organism. The mouse is the best genetic model for the study of most human diseases and developmental disorders. A strength of the mouse system is the ability to manipulate the genome at base pair resolution so that genes of interest can be mutated and their function probed, and FP reporters can be expressed in specific cells or tissues. As with other mammals the mouse due to its in utero embryonic development represents an inherently more complex experimental system, and so lags behind other animal models in the methods and tools that are available for live imaging applications.20,21 However, these technical hurdles are slowly being overcome, and it can be expected that any utility demonstrated for mice will then be extended to mouse and human embryonic stem (ES) cells and tissue engineering applications.
The irreversible photoactivatable GFP PA-GFP, is a variant of GFP that is widely used in different species because it possesses similar characteristics to GFP (Table 1).22 It has been engineered from wild type (wt)-GFP and contains a single residue amino acid substitution (T203H) in the GFP protein.22 This amino acid substitution leads to a non-fluorescent neutral fluorophore, which upon exposure to short wavelength light, is irreversibly converted into an anionic form with a concomitant 100-fold increase in green fluorescence. The resulting high contrast of the activated as compared to the non-fluorescent form has made it useful for studying cell migratory behaviors in the different organisms.
Labeling of single cells or small groups of cells expressing PA-GFP has been used in a study investigating cell trajectories in the chick hindbrain.23 Another study in the chick used PA-GFP to specifically label and track neural crest cells to elucidate and discriminate their individual and population behaviors, as well as for mapping their trajectories. A plasmid driving widespread expression of PA-GFP was electroporated into the chick neural tube, the site from which neural crest cells emerge in the embryo. The reporter was selectively photoactivated in neural crest cells as they delaminated from the dorsal neural tube. Therefore in these experiments, the migratory streams of neural crest cells could be visualized from their origin in the neural tube to their final destination in the pharyngeal arches. Novel short and long-range cell-cell interactions were visualized between individual neural crest cells as they migrated in chains, and between neural crest cells and their surrounding cells.23 In the postnatal mouse, PA-GFP has been used to study protein dynamics in vivo in the developing neocortex.24 In Drosophila, a PA-GFP fusion to alpha-tubulin has been used to study cell behaviors in the early mesoderm, a cell layer that lies deep within the embryo. Live imaging of Drosophila embryos ubiquitously expressing the subcellularly-localized PA-GFP fusion reporter, and photoactivation of prospective mesodermal cells before they became internalized, enabled imaging of their migration over non-activated non-visible ectodermal cells thereby facilitating the specific tracking of mesoderm cell trajectories.25
The tetrameric kindling FP (KFP) has been engineered from the natural chromoprotein asulCP (asFP595) of the sea anemone Anemonia sulcata (Table 1). AsulCP changes from a non-fluorescent to a red fluorescent state upon exposure to green light. Since its fluorescent state is unstable, this change is reversible and the protein can convert back to a non-fluorescent state. However, its engineered variant KFP is capable of both reversible and irreversible photoactivation depending upon the intensity of the activating light.26,27
Dronpa, another reversible PAFP, was isolated from the family of the Pectiniidae (Table 1). It fluoresces green and can be converted to a non-fluorescent form upon exposure to short wavelength light.28 Dronpa has been successfully applied in live imaging experiments in embryos. In the zebrafish, Dronpa was used for labeling individual neurons to reconstruct the neuronal network.29 Dronpa has also been used to label cells in culture and explanted neurons in combination with an optical lock-in detection approach (OLID) modality to produce high contrast images by reducing the background-to-noise ratio of the fluorescent signal in both frogs and zebrafish.30 To date, use of neither KFP nor Dronpa has been reported in mice, but it is likely that studies incorporating them as live imaging reporters are currently underway.
Red fluorescent proteins (RFPs) have been much sought after for live imaging experiments as they are less toxic due to their longer wavelength emission. Three photoactivatable RFPs, PAmRFP1-1, PAmRFP1-2 and PAmRFP1-3, had been engineered from the monomeric RFP mRFP1 (Table 1). The brightest among these three variants is PAmRFP1-1. Its red fluorescence is initially weak but then increases about 70-fold upon photoactivation with short wavelength light (400 nm).31 However, compared to other PM-FPs its quantum yield is relatively low and so its overall brightness is dim which makes it unsuitable for many live imaging applications. However, PAmCherry1, one of the recently developed variants of mCherry which include PAmCherry1, PAmCherry2 and PAmCherry3 holds more promise for live imaging studies and for dual-color imaging due to its improved brightness and faster photoactivation.32
A recent study has directly addressed the intrinsic ability of many RFPs to be photoconverted. Five commercially available RFPs exhibiting bright fluorescence intensities were examined and their ability to photoconvert when excited with the appropriate wavelength illumination was investigated. Far-red proteins including the dimer Katushka, its monomeric version mKate and HcRedI were shown to convert from a red to a green fluorescent state using single and two-photon excitation (Table 1). Moreover, two orange FPs, mOrange1 and mOrange2, were shown to shift from the red fluorescence spectrum (emission peak at 562 nm) to a far-red fluorescent state (emission peak at 640 nm). In principle, mOrange1 and mOrange2 are ideal for live imaging experiments due to their brightness, high photoconversion rate and far-red emission. They are also favorable as they are monomeric and were shown to work when incorporated into fusion proteins, including a subcellularly-localized human histone H2B fusion which labels the nucleus to provide information on cell position, division or death, as well as in fusions to proteins that are directed to the plasma membrane which provide information on cell morphology. Furthermore, their relative pH insensitivity makes mOrange1 and mOrange2 attractive for the generation of fusion proteins localized to secretory vesicles, to probe the existence of differential polarities or distributions within populations of cells.33 To date, none of these proteins have been used in an organismal context, though given their characteristics it is likely only a matter of time until they are.
Several PC-FPs are currently available, these include PS-CFP (Photoswitchable Cyan Fluorescent Protein) and its successor PS-CFP2, as well as an expanding group of green-to-red PC-FPs. Of the latter group, EosFP, Kaede, KikGR, Dendra and Dendra2 are currently the most used in live imaging studies in various model systems (Table 1). Their high contrast when switching color, makes activated cells readily distinguishable from nonactivated cells, while the reduced cell toxicity of long wavelength illumination makes them attractive for live imaging studies. In the following section we will provide a brief overview of the available PC-FPs as well as a representative series of applications.
Both the monomer PS-CFP and its successor PS-CFP2 have been engineered from wt-GFP (Table 1). Upon exposure to short wavelength light these proteins convert irreversibly from a cyan to a green fluorescent state with an accompanying increase in the green-to-cyan fluorescent ratio.34,35 However, given the overlapping nature of the pre- and post-conversion spectra, significant signal bleed through from the CFP and GFP channels is observed (Fig. 2A–F), and distinct separation of photoconverted and non-photoconverted cells can only be achieved when imaged using spectral separation methods such as linear unmixing.36
EosFP, named after the Greek goddess of dawn, is a tetrameric green-to-red FP, originally isolated from the stony coral Lobophyllia hemprichii. Further genetic engineering resulted in a monomeric variant, mEosFP, that has been shown to work in a variety of protein fusion in cells.37 In the frog embryo this reporter has been successfully used to label different cell lineages and used to examine early embryo morphogenesis.38 Recently, a brighter version of EosFP, mEOS2, has been reported and will likely supercede EosFP in most applications.39
The engineered green-to-red monomeric PC-FP, Dendra and its successor Dendra2 exhibit high photostability in their red fluorescent states, and are also attractive for long-term reporter tracking in living cells.35,40 To date, Dendra has been used to study the dynamics of actin transport during axonogenesis in organotypic hippocampal slice cultures,41 while Dendra2 has been used to study protein dynamics in cell culture systems, in the nematode worm C. elegans and in plants, as well to investigate protein turnover in gap junction channels in cells grown in culture.42–45 The tetramer Kaede, Japanese for maple leaf, was isolated from the coral Trachyphyllia geoffroyi. The red and green fluorescent states are similar in their brightness and their stability. Photoconversion of Kaede results in a 2,000-fold increase of the red over the green fluorescent signal.46,47 Kaede has been used in a variety of systems. It was used to trace neurons in zebrafish embryos, and specifically for visualizing cell movements and the growth of neurites.48 In other species such as the frog, Kaede was used to investigate protein synthesis in retinal ganglion axons.49 In chick, it was evaluated along with other PMFPs to investigate and live image the migration of neural crest cells.50 Transgenic mice constitutively expressing Kaede have been generated and used to study cell movements from lymphoid organs to other tissues.51 Recently, ES cell lines ubiquitously expressing a tandem dimeric Kaede (tdKaede) have been reported. These cell lines may be prove useful for cell lineage tracing studies in mouse chimeras as well as in tissue engineering applications or transplantation medicine.52
Another PMFP is the tetramer KikGR (Kikume Green-Red), which was engineered from the protein KikG of the stony coral Favia flavus.53 Like Kaede, KikGR is one of the few photoconvertible proteins that has been successfully used in embryonic stem (ES) cells and transgenic mice.54 Comparative studies have suggested that KikGR is more advantageous for cell labeling and optical marking than Kaede, since photoconversion of the former is more efficient, meaning both, green and red, fluorescent states are brighter compared to those of Kaede (Fig. 2G–R), and switching from one state to another more rapid.54 Several mouse strains exhibiting widespread expression of KikGR have been established.54,55 In one study using these mice selective labeling of early blastomeres in preimplantation embryos demonstrated that the specification of the embryonic-abembryonic axis in the mouse is likely to be independent of early cell lineage.55 In another study selective labeling of an ROI of defined dimensions within the emergent mesoderm tissue demonstrated that this methodology can be used to investigate collective cell behaviors such as convergent-extension which produce a directional change in the dimensions of the ROI.54
To determine the reporters best suited for in vivo applications in embryos, the direct comparison of PA-GFP, PS-CFP2, Kaede and KikGR, four photomodulatable proteins, has been carried out in the chick,50 and also recently in the mouse.54 In the chick embryo electroporation of plasmid constructs produces robust levels of reporter expression. Live imaging of labeled neural crests revealed distinct advantages of each protein for specific applications. Due to their high photoefficiency, the amount of light required to achieve photoconversion, KikGR and Kaede were reported as more suitable for monitoring cell migratory behavior, whereas PS-CFP2 and PA-GFP were noted as being more photostable, and thus likely to perdure, making them attractive for long-term studies, such as cell fate mapping.50
In the mouse transgenesis or gene targeting techniques are used to create reporter expressing strains. Thus levels of fluorescent proteins are physiological and therefore not as high as those achieved by the transient transgenic plasmid electroporation methods used in the chick. PA-GFP, PS-CFP2, Kaede and KikGR have been comparatively evaluated for investigating cell dynamics and cell fate in mouse embryos and ES cells.54 Of the four, KikGR was found to be the most suitable. Despite its use in other model systems, which perhaps afford higher levels of transgene expression, PA-GFP was prone to autoactivation, whereas the fluorescent signal of Kaede was dim compared to that of KikGR. PS-CFP2 was bright but because it exhibited signal bleed through was not optimal for all imaging systems.
KikGR was bright and exhibited rapid and complete photoconversion (Fig. 2). Mouse transgenic strains with widespread expression of KikGR (namely the CAG-KikGR strain) are attractive tools to live image and study cell fate and cell dynamics in live mouse embryos (Fig. 3), organ explants or tumor models. CAG-KikGR transgenic mice are indistinguishable from their wild type littermates demonstrating that KikGR is developmentally neutral, such that expression of the reporter protein at readily detectable levels does not cause deleterious effects in embryonic development or disease progression. KikGR can therefore be expressed in mice at sufficient levels so that cells can be visualized before and after photoconversion of any region of interest, and for extended periods of time (Fig. 4). Importantly, these experiments demonstrated that photoconversion with short wavelength high-energy illumination, and subsequent 3D timelapse imaging for an extended period of time does not adversely affect development. These reporter strains of mice can be of use to investigate both individual and collective cell behaviors, as well as determining cell fate.54
PMFP variants constitute a rapidly increasing set of tools for selectively non-invasively labeling and following cell populations to determine cell dynamics and cell fate. The development of mouse strains exhibiting lineage-specific expression of PMFPs as well as those expressing subcellularly-localized PMFPs can be anticipated within the next few years (Fig. 5). These new generation tools should provide greater spatio-temporal control of selective cell labeling as well as improved resolution of imaging data for individual cells, as compared to strains exhibiting widespread expression of native, predominantly cytosolic, fluorescent proteins.10,20,56 Fusions of PMFPs to histones such as human histone H2B localize to active chromatin. The resulting nuclear-localized reporter allows image segmentation and in doing so single cells within a group can be visualized. In this way the position of cells can be determined and tracked over time. Since histone fusions are also reporters of cell division and death, they also provide information on cell proliferation and apoptosis within a population.
In an accompanying article in this issue of Organogenesis, Kulesa and colleagues report the generation and evaluation of a human histone H2B-PS-CFP2 reporter in the chick.59 They compare the H2B fusion with native PS-CFP2 and demonstrate that both reporters have similar photophysical properties, but that H2B-PSCFP2 is more effective for imaging single cells in dense tissue. We have generated a human histone H2B-KikGR fusion and evaluated it in COS-7 cells (Fig. 6). H2B-KikGR exhibits correct localization to the nucleus and can be efficiently converted. However, our attempts to generate ubiquitously H2B-KikGR expressing ES cells and transgenic mice have failed, likely as a result of reporter toxicity due to the tetrameric state of KikGR. Of note, attempts to generate H2B-mCherry mice have also failed.56 Monomeric proteins such as PAGFP, PS-CFP or Dendra and Dendra2 are likely better suited for incorporation into subcellularly-localized fusions. Therefore the recently reported engineered monomeric green-to red PMFPs, mKikGR and mEos2 may be of promise for use in mice. Indeed, several mEos2 fusions have recently been reported. They include an H2B-mEos2 fusion which labels the nucleus, an mEos2-connexin fusion labeling gap junctions, a vimentin-mEos2 fusion which labels intermediate filaments, as well as fusions labeling mitochondria and keratin.39 At present these mEos2 fusions have only been validated in cell culture systems.
Alternative types of fluorescent protein fusions desirable for in vivo imaging include those directed to the plasma membrane. This latter class of reporter provides information on cell morphology and membrane dynamics, including cellular protrusions such as filopodia and lamelopodia. Since it will be important for some applications to simultaneously compare the behaviors of different cell populations, dual labeling using different spectrally-distinct and subcelluarly-localized reporters will be necessary to provide high resolution information.21,33,56
We thank Ingo Bothe for discussions and comments on this review, and Phoebe Nowotschin for editorial assistance. Work in our laboratory is supported by the National Institutes of Health (RO1-HD052115 and RO1-DK084391) and NYSTEM. S.N. is supported by a postdoctoral fellowship from the American Heart Association.
Previously published online: www.landesbioscience.com/journals/organogenesis/article/10552