SEARCH

SEARCH BY CITATION

Keywords:

  • avian;
  • fate mapping;
  • developmental biology;
  • organogenesis;
  • microscopy

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

We describe the development of transgenic quail that express various fluorescent proteins in targeted manners and their use as a model system that integrates advanced imaging approaches with conventional and emerging molecular genetics technologies. We also review the progression and complications of past fate mapping techniques that led us to generate transgenic quail, which permit dynamic imaging of amniote embryogenesis with unprecedented subcellular resolution. genesis, 49:619-643, 2011. © 2011 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Embryogenesis is a complex and dynamic process that results from the integration of three fundamental cellular processes: growth, differentiation of cell types, and morphogenesis (DeHaan1967). Morphogenesis is the transformation of amorphous assemblies of cells into organized structures orchestrated through a complex medley of individual cell behaviors and collective cell movements. The study of morphogenesis embodies an integral part of developmental biology and has recently been rejuvenated by the creation of new imaging reagents, model organisms, imaging techniques, and growing molecular insight into its underlying mechanisms (Ando et al.,2002; Gurskaya et al.,2006; Kulesa et al.,2009,2010; Nowotschin and Hadjantonakis,2009a,b; Patterson,2008; Stark and Kulesa,2007; Wacker et al.,2007; Wiedenmann et al.,2004).

Embryogenesis occurs by a spectacular array of cell movements and rearrangements. To generate a three-dimensional (3D) embryo, cells move independently and collectively in every direction–anterior and posterior, medial and lateral, and dorsal and ventral. What's more, these cellular movements occur along the additional axis of time, presenting a strong argument that to accurately study embryogenesis, developmental biologists should employ the tools and approaches of dynamic analysis.

APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Fate Mapping

An important aspect of developmental biology is investigating where particular cells of interest originate in the early embryo, and how and when they arrive at their final locations. A classic approach to studying cell and tissue origins is to generate a fate map, which outlines the developmental history of specific cells or regions of the early embryo. Lineage analysis traces a progenitor cell through its divisions to form all of its daughter cells and tracks their contributions to the embryo to map the fate of that progenitor cell. It is a powerful tool for deciphering which and how many cells form the various structures of the developing embryo and how the structures evolve their shape. To produce a fate map, a single cell or a few cells are labeled in a specified location, and the label is transmitted to the progeny of the initial cell(s) independently of where the cells migrate, generating lineages of labeled progeny. The types of cells that derive from the original progenitor and their locations and lifespans can then be determined by assessing the positions, phenotypes, and behaviors of the labeled cells at successive time points following the initial labeling (Bronner-Fraser and Fraser,1989; Fox et al.,2008; Kulesa et al.,2009; Sato et al.,2010; Wacker et al.,2007).

A critical element of constructing a fate map is selecting a suitable lineage label or cell marking. Over the past 100 years, numerous different labeling methods have been attempted (Rodriguez-Gallardo et al.,2005). Researchers can label single cells using many different techniques and reagents, including but not limited to vital dyes, radioactive tracers, genetic markers, and retroviruses.

Vital Dye Labeling

Hubert Goodale was the first to study embryonic morphogenesis using vital dyes (Goodale,1911a,b). He placed spots of Nile blue sulfate around the equator of amphibian blastulas to label live cells. The spots of dye lengthened towards the edge of the blastopore, and Goodale concluded that the marginal zone and endodermal yolk cells must be migrating. Walter Vogt subsequently adapted vital dyes to follow the migration of cells in different regions of the amphibian embryo and track their cell fate at different stages of development (Vogt,1929). Vogt infused dye into agar coated on a microscope slide and then placed sections of dyed agar on amphibian embryos to label those regions of cells and track them. These studies provided a decisive foundation for mapping the derivatives of regions of the amphibian gastrula, but with limited resolution. Because the dyes were water-soluble, they diffused to neighboring cells and therefore confounded the results of the lineage tracing. The dyes are thus not sufficiently specific to label individual cells. Furthermore, the dye is lost over time, particularly in rapidly dividing tissues (Weston,1963).

Tissue Transplantation

In a landmark article in 1924, Spemann and Mangold conducted transplant experiments between Triton newt embryos that displayed different pigmentation to distinguish the contributions of donor and recipient tissue to various parts of the embryonic axis (Spemann and Mangold,1924). Hilde Mangold transplanted tissue from the dorsal blastopore lip of one gastrulating embryo onto the ventral side of a different embryo at the same stage. By using Triton species with different colorations, the tissues that originated from the donor graft could be distinguished from those deriving from the recipient host in the newt chimera. These experiments powerfully demonstrated the principle of induction and that an “organizer” region of the dorsal blastopore had the ability to instruct unspecified ectoderm to form neural tissue and ventral mesoderm to form somites (Spemann and Mangold,1924). This established an elemental principle of developmental biology that at permissive times during development, certain cells could induce other cells to assume a different fate as specified by their spatiotemporal position.

Radiographic Labeling Approach

The need for a generalizable method of labeling cells led to radiographic labeling. The autoradiographic mapping approach was first considered by Hughes et al. (1958) and tested in vitro by Trinkaus and Gross (1961) in cultured chicken melanocytes to determine the specificity of the labeling technique. Several groups then applied radioactive labeling to study cell migration in vivo in chick embryos (Orts-Llorca and Collado,1968; Rosenquist,1966,1970; Stalsberg and DeHaan,1969; Weston,1963). Lineage analysis by radioactive labeling involves removing grafts of tissue from a donor, immersing the cells or tissue in tritiated thymidine for its incorporation into DNA, and grafting the tissue back into the host. Tissue must be sectioned, and the approximate positions of nuclei in two-dimensions (2D) are detected by the beta emissions released. Centers of beta emission are captured on autoradiographic paper at specified time points after grafting (Weston,1963). No differences in cell behavior were detected between radioactively labeled cells versus unlabeled cells at the dosage used (Trinkaus and Gross,1961); however, subtle effects on cell survival or behavior could not be assayed.

Excess unlabeled thymidine was transferred with grafts to the recipient embryo to prevent unintended labeling of cells in the recipient embryo and ensure that only grafted cells were labeled. While this provided considerable specificity of cell labeling, label detection could not be resolved to single cells within dense tissue, emissions could only be detected from the surface of sections, and label was diluted with cell division (Hughes et al.,1958). Cell death could also result in uptake of labeled DNA by migratory macrophages, which would confound interpretations of labeled cell migration (Trinkaus and Gross,1961). Furthermore, developing the autoradiographs took days or weeks and required fixation and sectioning, resulting in snapshots of the general locations of some of the labeled nuclei at different time points. A more sensitive method for detecting label, and labels that could be detected in living tissue, were required for precise fate mapping and study of the dynamics of morphogenesis.

Chick-Quail Genetic Chimeras

In 1969, Nicole Le Douarin advanced the field of lineage analysis by establishing the use of chick-quail chimeras for fate mapping (Le Douarin,1969). Le Douarin recognized that the similarity between quail and chicken embryonic development would enable grafting between them without obstructing development. She capitalized upon the ability to distinguish quail and chicken cells from one another by staining DNA with Schiff's reagent. Quail interphase nuclei contain dense heterochromatin in the nucleoli, which stain brightly with Schiff's reagent, while nucleoli in the chick do not label distinctly (Le Douarin,1973,1996).

Thus, Le Douarin exploited an intrinsic nuclear marker that enabled her to identify grafted cells in a chimaeric embryo. Her approach involves transplanting cells or tissues between embryos of matched developmental stage in ovo, generally from the quail embryo into the corresponding region of the chick embryo after removing the identical chick tissue (Couly et al.,1992; Le Douarin,1973,1996; Le Douarin and Barq,1969). The eggs are reincubated for a desired period to enable the transplanted quail cells to assimilate into the host chicken tissue as normal development ensues. At later stages of development, the grafted cells are identified with microscopic analysis of stained tissue sections. The approach avoids the aforementioned problem of leaky dyes disseminating to neighboring cells and permits high cellular resolution for lineage analysis studies (Le Douarin,1986; Le Douarin et al.,1974).

The strengths of the chick-quail chimera technique are that the label is integral to the cell, and it does not spread to adjacent cells or dilute with cell proliferation (Balaban et al.,1988; Bortier and Vakaet,1992; Le Douarin,2008). Furthermore, the locations of grafted cells can be more precisely determined compared with the radiographic approach. The traditional drawbacks are that chick-quail chimeras are analyzed statically, the tissue contributions are attributed to groups of cells rather than individual cells, the requirement for precise staging between quail and chick complicates the method, the technique is not broadly applicable to all species, and the transplant scar can induce developmental problems.

Progenitor Cell Ablation

Not all model organisms are amenable to transplantation or chimeric studies for fate mapping. An alternative, reverse approach of removing small groups of cells early in development has been applied to map the derivatives of early progenitor cells (Underwood et al.,1980). In Drosophila melanogaster, for instance, the regions of the blastoderm that give rise to different appendages or body segments were determined by removing about 15 blastodermal cells at a time and observing the resulting anatomical defects in segmentation using electron microscopy. While this approach generated a useful overall lineage map of the blastoderm and demonstrated that progenitor cells of adjacent segments are contiguous on the blastoderm, borders cannot be precisely determined, and subtle defects are missed, so the contributions of the ablated cells are overlooked. Furthermore, the method clearly cannot be used to follow morphogenesis or determine how progenitor cells give rise to the structures they form.

Dye Labeling of Individual Cells

While tissue grafting techniques can map the embryonic contributions of groups of cells, the technique can disrupt development, and grafting is generally restricted to superficial sites on the embryo. Tracking the ancestry of tissues deep within an embryo requires a more generalizable means of labeling and visualizing individual progenitors.

Fifty years after Vogt's pioneering experiments using vital dyes to trace cell fate, breakthroughs were made to label living cells in a way that confined the label to the injected cells and circumvented the ambiguities resulting from dye leaking to adjacent cells. In 1978, Weisblat et al. fate mapped single cells (Weisblat et al.,1978). By injecting horseradish peroxidase into single leech blastomeres, they identified derivative tissues following a histochemical reaction with benzidine. They improved this approach by injecting dyes and dye-conjugated dextran, which can be directly visualized in the cytoplasm of cells (Gimlich and Braun,1985; Heasman et al.,1984; Weisblat et al.,1978,1980).

The development of vital dyes that are confined to labeled cells improved label resolution and permitted dynamic in vivo observation of individual cell movements. Dextran-dye conjugates were transmitted to daughter cells and remained traceable in the cytoplasm of cells for a week or more (Gimlich and Braun,1985). Conjugates of different colored dyes enabled multicolor tracing experiments, and fixable dextran-amines enabled immunostaining of dextran-labeled tissue sections (Gimlich and Braun,1985). Fluorescently labeled dextrans were injected into individual cells for fate mapping studies, for instance to trace the lineages of avian neural crest cells (Bronner-Fraser and Fraser,1989,1988,1991), Hensen's node in the chick (Selleck and Stern,1991), and zebrafish blastomeres (Kimmel and Law,1985), among others.

Carbocyanine Vital Dyes

The vital carbocyanine dyes DiA, DiI, and DiO simplifed fluorescent dye use as cell labels for lineage tracing (Schlessinger et al.,1977; Wu et al.,1977). These hydrophobic dyes fluoresce at distinct wavelengths and insert into lipid cell membranes (Axelrod,1979). Because the dyes integrate into the plasma membrane, they are distributed among the progeny of labeled cells during division, but the dyes usually do not flip into the membranes of neighboring cells (Honig and Hume,1989). Thus, they faithfully trace the lineages of the progenitor cells that incorporated the label. Vital carbocyanine dyes fluoresce with strong intensity, and the diverse colors that different dyes emit enable distinction of different populations of adjacent cells using multispectral techniques (Carette and Ferguson,1992; Clarke and Tickle,1999; Gan et al.,1999; Holland and Holland,2007; Honig and Hume,1989; Scherson et al.,1993; Serbedzija et al.,1989).

Carbocyanine dyes can also be used for time-lapse studies in living embryos (Ezin et al.,2009; Hatada and Stern,1994; Kulesa and Fraser,1998; Kulesa et al.,2000). The vital dyes can be focally applied to small numbers of desired cells of (i.e., 1–100 cells) or to larger areas of tissue in living embryos and then followed using fluorescence microscopy (Bildsoe et al.,2007; Wetts and Fraser,1988). Drawbacks with all of these direct cell-labeling methods are that labeling can be nonuniform, and the topically applied label is diluted upon cell division and is thus unsuitable for lineage analysis studies of extended duration (Bildsoe et al.,2007; Clarke and Tickle,1999; Dickinson et al.,2002; Frank and Sanes,1991; Honig and Hume,1989).

Viral Delivery of Genetic Markers

In the 1980s, Joshua Sanes and Constance Cepko and their coworkers engineered replication-defective retroviruses to label cells for lineage analysis (Price et al.,1987; Sanes et al.,1986). Upon retroviral infection of a cell, the marker gene [i.e., horse radish peroxidase, alkaline phosphatase, β-galactosidase, or fluorescent protein (FP)] becomes integrated into the cell genome, thereby stably labeling the cell and its descendants for tracking over time (Price et al.,1987; Sanes et al.,1986; Turner and Cepko,1987). The retrovirus is rendered replication-incompetent by elimination of its gag, pol, and env genes, which results in expression of the marker genes in a cell-autonomous manner in clonally derived cells (Cepko et al.,1998; Sanes1989). Because retroviral vectors require disassembly of the nuclear envelope to access the genome of an infected cell, they are only able to integrate into the genomes of actively dividing cells (Cepko et al.,1995). However, lentiviral vectors, which belong to the Retroviridae family, can stably deliver gene cargo into the DNA of both proliferating and quiescent cells (Naldini et al.,1996b).

Marking cells by viral infection has several advantages as the viruses themselves are relatively harmless to the infected cell, no infective particles are produced by infected cells so there is no transmission to neighboring cells from the infected cell, and the integrated retroviral genome is dependably inherited by all offspring of an infected progenitor cell (Cepko et al.,1998; Mikawa et al.,1992; Sanes1989; Turner and Cepko,1987; Turner et al.,1990). However, implementation of the technique can sometimes be complicated by the silencing of integrated proviruses, and infection of too many progenitors may confound interpretation of lineage relationships (Cepko et al.,1998; Poynter and Lansford,2008; Sanes1989).

When labeling cells by viral gene transfer, it can be difficult to identify all the cells derived from a single progenitor, at the exclusion of those derived from other labeled progenitors. Even injecting a very low titer of retrovirus typically labels several individual cells (Cepko et al.,1998; Sanes1989; Turner and Cepko,1987). If it is known that minimal migration and intercalation of cells occurs, then it can be presumed that groupings of cells marked with the reporter are clonally related, but if widespread intermixing of cells is suspected, then a more stringent evaluation of clonality is needed (Austin and Cepko,1990; Turner and Cepko,1987). In a laudable effort to verify lineage clonality, Cepko and coworkers devised a complex library of retroviral constructs in which a unique DNA oligomer tags each individual virus and can be identified using polymerase chain reaction (PCR) and DNA sequencing (Golden et al.,1995). Cells that are clonally related contain identical DNA tags. The drawback of this approach is that it is extremely arduous to characterize individual cells, and extracting and PCR-amplifying the contents of single cells are nontrivial.

Although the approach of recombinant DNA oligomer tagging would be too laborious to trace the lineages of thousands of cells during embryonic morphogenesis, the notion of marking individual cells and their descendants with a stable genetic label was an integral advancement for in vivo dynamic lineage mapping. The discovery of green FP (GFP) in jellyfish (Shimomura et al.,1962) and its cloning (Prasher et al.,1992) and expression in cells of other species (Chalfie et al.,1994; Gervaix et al.,1997; Okada et al.,1999; Tsien,1998) opened the door for fate mapping by dynamic imaging. Introducing FPs to cells by infection with lentiviral vectors stably marks the infected cells, and they reliably transmit the fluorescent marker to their progeny. Labeled cells proceed to divide, migrate, and differentiate as usual (Carleton et al.,2003; LaRue et al.,2003; Okada et al.,1999), and the stable genetic marking permits tracking of labeled progenitors and their descendants in living embryos in real time using time-lapse videography, confocal, and two-photon (2P) microscopies (Dickinson et al.,2002; Okada et al.,1999; Sato et al.,2010).

Lentiviruses are efficient gene delivery vehicles that can bestow enduring gene transduction to a range of cell types in vivo (Blomer et al.,1997; Leber et al.,1996; Lois et al.,2002; Naldini et al.,1996a,b; Watson et al.,2005). As described below, lentiviral vectors are particularly useful because they are able to stably express the delivered genes in a tissue-specific manner in somatic cells and in the germ line of transgenic animals (Kootstra and Verma,2003; Lois et al.,2002). Lentiviral vectors have been used to efficiently produce transgenic mice, rats, quail, chickens, monkeys, and other organisms with high expression of the fluorescent marker (Chapman et al.,2005; Lois et al.,2002; McGrew et al.,2004; Michalkiewicz et al.,2007; Palfi et al.,2002; Remy et al.,2010; Sasaki et al.,2009; Sato et al.,2010; Scott and Lois,2005; Sosa et al.,2010; Tarantal et al.,2001; Yang et al.,2008). Combining stable labeling of cells via lentiviral gene transduction with FP expression provides the means for decoding the complex developmental morphogenesis and lineage structures in vertebrates using dynamic multispectral imaging.

MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Direct observation of cell lineages generated the entire fate map from zygote to larva in the nematode Caenorhabditis elegans (C. elegans) (Sulston et al.,1983). This approach revealed that under certain conditions, cell lineages in C. elegans conform to the same stereotyped map from one embryo to the next. However, acquired fate maps in vertebrates have revealed less stereotyped development (Gimlich and Cooke,1983; Gimlich and Gerhart,1984; Harrison,1933; Jacobson,1984,1985; Lawrence and Green,1979; Nieuwkoop,1969; Zipursky and Rubin,1994), and to the contrary, have highlighted the essential role of cell-cell interactions in directing fate and orchestrating the extensive cell movements of vertebrate morphogenesis (Adams et al.,1999; Damsky et al.,1993; Etienne-Manneville and Hall,2002; Harden et al.,1995; Jacinto et al.,2001; McMahon et al.,2008,2010; Morriss-Kay and Tuckett,1985; Schock and Perrimon,2002; Supatto et al.,2005). Cohesive collections or sheets of cells coalesce, driven by a mosaic of cell behaviors that includes cell migration, division, growth, death, polarity, and cell shape changes that direct cells to move, to sense their environment, and to adhere to one another (DeHaan1967; DehHaan and Ebert,1964). Through individual cell activities that drive collective movements, these aggregations of cells become sculped into functional tissues and organs. Cell mechanics govern changes in cell shape and tissue morphogenesis by pushing, pulling, bending, and twisting individual cells and whole tissues into their requisite forms. Genetic studies have demonstrated that cell shape and motility derive from the interactions of subcellular structures such as the cytoskeleton and cell membrane adhering to the extracellular matrix and adjacent cells (Farhadifar et al.,2007; Friedl,2004a,b; Kafer,2007; Lecuit,2008; Mogilner and Keren,2009; Rosso et al.,2004).

While the molecular structures underlying these subcellular interactions have been studied to great depth, the dynamic cellular mechanisms that generate and sustain global tissue morphology are not understood. It is this problem of deciphering how morphogenesis occurs at the cellular level that multispectral dynamic imaging is suited to resolve.

MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

The derivation of embryonic structures from individual populations of progenitor cells can be visualized in live embryos by tracing the allocation of progenitors and their descendants to their final destinations. Realizing the power of fluorescent labeling for tracking the dynamics of embryonic morphogenesis requires a model system amenable to in vivo time-lapse imaging. Thus, one of the first considerations for dynamic fate mapping is choosing an appropriate model organism that exhibits the developmental program of interest.

Quail, an Amniote Ideal for Dynamic Studies

The Japanese quail, Coturnix coturnix japonica, is grouped with chickens in the Order Galliformes of Class Aves. The quail is a warm-blooded amniote vertebrate (Chojnowski et al.,2007), and like mammalian embryos, the quail embryo experiences considerable growth and tissue rearrangement during morphogenesis. There is general agreement that early embryogenesis in the chick (Hamburger and Hamilton,1951) and quail (Ainsworth et al.,2010; Padgett and Ivey,1959,1960; Ruffins et al.,2007; Zacchei,1961) proceed similarly (Bronner-Fraser,1996a). Coturnix quail present significant experimental advantages, including the availability of fluorescently labeled transgenic lines, the modest size of the breeding adults, which are easily bred in traditional animal facilities, its quick reproductive maturation (Huss et al.,2008), and a recently sequenced genome. The quail genome is ∼1.4 picograms, and a comparative map between quail and other avians with 15-fold coverage is being produced (Dr. Dave Burt, Roslin Institute, personal communication).

Because the quail embryo develops outside the mother, it is suited for extended time-lapse imaging to capture embryogenesis, either in ovo (in the egg) or ex ovo in culture (in vivo). Whereas imaging of mouse embryos in culture is generally restricted to a few hours, with the feasibility of imaging for up to 24 hours (Jones et al.,2002), an unperturbed embryo inside a windowed quail egg can be imaged continuously for several days. This makes the Japanese quail an ideal amniote model that enables the application of molecular genetic targeting techniques that have become commonplace in mouse studies, in conjunction with live dynamic imaging of fluorescently labeled cells to capture amniote embryogenesis with remarkable subcellular resolution.

METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Fluorescent Protein Labeling

Accurate tracking of cell lineages during morphogenesis requires unequivocal identification of the progenitors and their descendants through the duration of the study. Thus, the next steps in creating a dynamic fate map involve choosing an appropriate cell label and method of introducing the label. A clearly discernable marker that is well-tolerated by the specimen and robust to continuous imaging is a prerequisite for successful dynamic analysis. Over the past decade, there has been an explosion of development of new methods to label cells with fluorescent markers. Although dye labeling still retains some utility with its ease of application, the use of FPs has emerged as the most versatile and durable labeling technique. The labeled cells can express an FP in the cytoplasm, or the FPs can be fused to specific proteins or cell localization tags to define subcellular structures. The FPs fold and become visible within hours after expression, and they exist in a spectrum of colors (Shaner et al.,2004,2005,2007).

Depending on the desired application, fluorescent labels can be delivered by administering dye to cell membranes, injecting, electroporating, or transfecting DNA, RNA, or marker enzymes into the cytoplasm, or integrating reporters into the genome by viral infection or transgenesis (Bildsoe et al.,2007; Bronner-Fraser,1996a,b; Dickinson et al.,2002; Itasaki et al.,1999; Sato et al.,2010).

Transient Labeling

For short duration experiments, it is generally adequate to transiently express FPs through electroporation of DNA or RNA. FPs may be visible within tens of minutes when translated from RNA and in 3–6 hours when encoded by DNA, and expression can persist a few days [(Dickinson et al.,2002; Kulesa et al.,2010); RL, unpublished data]. The approach has some shortcomings in the form of uncertainty as to the amount of nucleic acid that has been introduced to cells, the dilution of transcripts and thus signal over time, and restrictions imposed by the delivery method. Electroporation can distribute nucleic acid unevenly across a tissue, with some cells receiving excessive amounts, and it is extremely difficult to determine the number of copies of DNA or RNA that are introduced. Over-expression of the exogenous DNA or RNA may alter cellular behaviors that are being imaged. For instance, the excess exogenous protein may overload a cell's proteasomal degradation pathway in an attempt to control its abundance, in the meantime disrupting degradation of proteins involved in directing normal cell behavior. Undegraded FP can further compromise interpretation of data as excess may pile up in the cytoplasm even when fused to a protein not normally found in the cytoplasm, falsely suggesting cytoplasmic localization.

Overloading some cells with excess plasmid copies also complicates imaging of the specimen because the signal in some cells will be saturated, while it may be faint in nearby cells (see Fig. 1). Setting imaging conditions to visualize the moderately expressing cells results in saturation of cells that express excess label (Fig. 1a–d, white arrowhead). Details of cell behavior, such as membrane dynamics, are also obscured by label saturation. In addition, not all cells are labeled (there is dark space in Fig. 1 where cells must exist), which facilitates visualization of labeled cells but provides no information about the surrounding unmarked cells. In Figure 1, the cell marked with a yellow outline may be interacting with another cell, but the reciprocal interactions cannot be seen because the partner cell is unlabeled. The figure also shows how electroporation frequently results in labeled debris that can complicate interpretation of what is being seen (small blue arrowhead in Fig. 1b). Furthermore, as mentioned above for all transient cell-labeling techniques, electroporated label dilutes with cell division.

thumbnail image

Figure 1. Uneven expression of electroporated plasmid in different cells. (ad) Four successive snapshots of migrating mesodermal cells demonstrate dynamic cellular movements. The cell marked with a large white arrowhead in all four images has received an excessive amount of plasmid for expression of the FP, while the cell outlined in dotted yellow is appropriately labeled for imaging, and its movements are easier to discern. (a) The cell outlined in yellow extends a membrane protrusion (small white arrows). (b) The small blue arrowhead is marking a spot of fluorescence that could be a concentrated bit of protein in a nearby cell or a bit of cell debris from a dying cell. An undesired side effect of electroporation is the death of some cells, leaving behind bits of labeled cell debris. (c) The saturated cell (arrowhead) could be two cells, but because the signal is so saturated, it obscures any borders of heightened fluorescence that might otherwise be visible between two less intensely labeled cells. (d) The cell outlined in yellow extends several lamellipodia (small, white arrows) as it crawls past the saturated cell(s). Scale bars = 20 μm.

Download figure to PowerPoint

Stable Genetic Labeling

Longer duration experiments may require stable integration of constructs via lentiviral infection or transgenesis for continuous FP expression to track cells without dilution of label. As described previously, these lentiviruses are engineered to be replication-incompetent so that only the progeny of the original infected cells are labeled, and no infectious particles that could infect neighboring cells are produced (Sanes1989). Lentiviral plasmids can be introduced into cells by electroporation and the lentiviruses by infection. Following retroviral integration into the genomes of infected cells, the genetically encoded cargo, such as FPs, are stably expressed (LaRue et al.,2003). If somatic cells are infected, the cells' progeny will all be labeled. When germ cells are infected, some of the offspring of the infected animal may be transgenic.

One consideration regarding the use of viral vectors such as retrovirus or lentivirus is the longer time interval–typically 8–24 hours–between infection and visual detection of fluorescence (Dickinson et al.,2002; LaRue et al.,2003). The provirus requires breakdown of the nuclear envelope before it can access and insert into the cell genome, so the rate of cell division accounts for the time delay in cargo expression (Cepko et al.,1995). However, viral infection provides the substantial advantages that once integrated, fluorescent label expression is maintained at physiologically appropriate levels, all daughter cells reliably inherit the FP gene, and the expression remains consistent over generations of cells rather than being diluted with cell division. Likewise, in transgenic animals, the FP is expressed at physiologically regulated levels for the duration of time that the promoter used is active during embryogenesis.

Modern vectors are typically decorated with the glycoprotein of the vesicular stomatitis virus (VSV-G), which nonselectively mediates entry into nearly all cell types, including those of chickens and quail, mice and rats, and humans (Zavada,1982). Importantly, the VSV-G also enables ultracentrifugation of viral supernatants to concentrate them so that tiny volumes are sufficient to infect a large number of cells (Burns et al.,1993). This makes embryo injection practical.

Transgenic Labeling

For many applications, it is sufficient to infect progenitor populations with lentiviruses and image their progeny to study morphogenesis. However, to harness the power of fluorescent labeling for studying embryogenesis without having to infect each specimen, we have developed transgenic quail that express various FPs in a tissue-specific manner. We presently generate transgenic quail by injecting HIV-based VSV-G pseudotyped lentiviruses into the subgerminal cavity of stage X blastodermal cells (Poynter and Lansford,2008; Poynter et al.,2009). These viruses integrate into all cells, and the transgenic offspring derived from infected primordial germ cells express an FP in a ubiquitous or tissue-specific manner. The success rate of hatching founders varies with the investigator's expertise, but of the founder birds that hatch, 25–40% are germline mosaic for the transgene. To date, transgenic quail that express FPs driven by vascular (Sato et al.,2010), neural [(Scott and Lois,2005); RL, unpublished data], and ubiquitous (RL, unpublished data) promoters have been generated. Transgenic quail lines provide unlimited samples with consistent labeling of cells at appropriate physiological expression levels. Individual lines can be crossed to obtain embryos with multiple subcellular structures labeled in spectrally distinct colors. In this way, transgenesis can tailor label expression to facilitate the study of specific cells and behaviors of interest in a convenient, dependable, and optically accessible model system.

Subcellular and Tissue-Specific Localization of FPs

Self-inactivating, replication-defective lentiviruses and retroviruses can be built to efficiently express their genetic cargo in desired tissues and cells using cell type-specific transcription units. Within these tissue-specific expression cassettes, FPs can be targeted to designated subcellular locations according to what cell behavior is to be studied. These tools allow for specific types of cells to be marked in defined subcellular locations. For instance, the histone 2B (H2B)-FP marker stays associated with the chromatin throughout the cell cycle, permitting cells to be tracked through cell divisions (see Fig. 2). In this way, any desired cellular structure can be labeled in one of many colors using tissue-specific or ubiquitous viral vectors by infecting tissue progenitors in ovo or by generating transgenic animals.

thumbnail image

Figure 2. Tracking fluorescently labeled nuclei through cell division. FP fused to H2B remains associated with chromatin through mitosis. The panels (af) show sequential snapshots every 8 min taken from a time-lapse of a Tg(tie1:H2B-eYFP) embryo. The eYFP (pseudocolored green) delineates the chromatin of the labeled dividing cell (marked with arrowheads) through all stages of mitosis, permitting tracing of cell lineages from progenitor to daughter cells through all cell divisions. Scale bars = 10 μm. Tg = transgenic, H2B = histone 2B, eYFP = enhanced Yellow FP.

Download figure to PowerPoint

Studies requiring cell type-specific gene expression have traditionally relied upon transcriptional control by tissue-specific promoters to drive FP expression, but new approaches to focus expression are emerging (Anliker et al.,2010; Cattaneo,2010; Watson et al.,2002). Targeted cell entry vectors have garnered deserved attention as they may circumvent undesired off-target expression and disruptive gene integration effects that can result from nonspecific cell infection and viral integration (Michou et al.,1999; Sugishita et al.,2004; Watson et al.,2002,2005). Thus, viruses that only enter desired target cells and leave all other cells untouched offer a new approach for targeting gene expression that minimizes the potential for perturbing development.

Multiplexed Labeling Strategies

FPs are available in many spectral variants, including CyanFP, GreenFP, YellowFP, RedFP, or CherryFP, to name a few (Campbell et al.,2002; Shaner et al.,2005; Wang et al.,2004). Different fluorescent colors can be localized to specific cellular organelles, allowing these structures to be imaged alone or in combination with other markers. Comprehensive analysis of cell interactions and other behaviors during morphogenesis often requires simultaneous visualization of several different tissues or subcellular structures. Fusing different FP markers to each of these different structures facilitates these studies. As noted above, nuclear-localized H2B-FP permits tracking and determining the fates of cells while directly assaying cell proliferation and cell deformation (Hadjantonakis and Papaioannou,2004; Kanda et al.,1998; LaRue et al.,2003). At the same time, if cell migration or behavior of particular cells is of interest, then it is useful to label the plasma membrane with a different FP spectral variant. Additional considerations of this approach are discussed below, but the basic premise allows for the dynamics of lamellipodial extensions and retractions, interactions between neighboring cells, and movements associated with division to be followed, as depicted in Figure 3 and the Supporting Information Movie 1. Using these multiple labeling techniques, one can now visualize in real time the lamellipodial extensions of individual cells migrating, for example, from the neural tube or along blood vessels (Bianco et al.,2007; Kulesa et al.,2010; Lawson and Weinstein,2002; Prasad and Montell,2007; Tekotte et al.,2007). The subcellular resolution permitted by multispectral fluorescent labeling of nuclei, membranes, or other cellular structures in an optically and genetically accessible amniote opens the door to unprecedented lineage analysis and dynamic developmental studies in living embryos.

thumbnail image

Figure 3. Visualizing cell behavior with multispectral labeling techniques. (ad) The panels show individual cells labeled with two spectrally distinct FPs to study cell behavior. Tg(tie1:H2B-eYFP) quail embryos were electroporated with a ubiquitously expressed plasmid DNA for membrane-localized CherryFP to label some but not all cells. The tie1+ nuclei are pseudocolored green and the membrane-CherryFP is red. Panels (a–d) are snapshots selected every 75 min from an in vivo time-lapse that demonstrate various cell behaviors that can be recognized and studied in living embryos. (a) The membrane label defines a membrane protrusion that a cell extends as it interacts with its environment (white arrowhead). (b) Two labeled neighboring cells make contact (indicated by the white arrowhead and outlined in dotted yellow and white at the region of contact). Note that the cells may also be interacting with other nearby cells, but those interactions are not visible unless both participating cells' membranes are labeled. (c) The cell indicated with a white arrowhead is about to divide, and fluorescent label is concentrated in a smaller volume, making the cell appear brighter. (d) The cell in (c) has divided, and a cell membrane projection is visible (white arrowhead). Scale bars = 10 μm. Tg = transgenic, H2B = histone 2B, eYFP = enhanced Yellow FP, FP = fluorescent protein.

Download figure to PowerPoint

It is often advantageous to express the genes for multiple different proteins from one expression vector to reduce the multiplicity of infections required to label all structures of interest. For instance, to label the nucleus green and the mitochondria red from a single infection, an internal ribosome entry site (Bienkowskaszewczyk and Ehrenfeld,1988; Jang et al.,1988; Shih et al.,1987) can be inserted between two genes, such as H2B-GFP and cytochrome C-RFP (Chen et al.,2003). One shortcoming of this approach is that the mRNA of the 3′ gene is often translated at lower levels relative to the 5′ gene (Jang et al.,1988,1989). An alternative option for coupled protein expression is to use an autoproteolytic peptide linker, such as the foot-and-mouth disease virus 2A sequence (Donnelly et al.,2001; Fang et al.,2005; Ryan and Drew,1994), which confers the stoichiometric coexpression of two or more genes linked by the 2A sequence. This multicolor approach can be used to define the entire cell and its various compartments (i.e., nuclear, cytoplasmic, and membrane) with one vector. When integrated into the cell genome by retroviral infection, this technology provides developmentally regulated, tissue-specific coexpression of FPs or functional reporters (DVB and RL, unpublished data). Using the 2A sequence approach, one can combine all the necessary tools into one vector to visualize the natural induction or suppression of promoter-enhancer units at specific developmental stages or positions and simultaneously track multiple subcellular structures in different colors.

IMAGING MODALITIES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Fluorescently labeled cells can be tracked using epifluorescent, confocal, or 2P microscopy to analyze the dynamic movements and interactions of cells as they undergo morphogenesis and organogenesis (Kulesa and Fraser,2000; Kulesa et al.,2000; LaRue et al.,2003; Petroll et al.,2004; Sato et al.,2010).

Widefield Epifluorescence Imaging

Using widefield epifluorescence imaging, fluorophores are excited at all depths of the embryo in the full field of view, which is illuminated, and photons emitted from all depths are simultaneously collected. Thus, epifluorescence imaging is fast, and the coincident excitation of the sample means low photodamage to tissue, permitting longer duration videos (Czirok et al.,2002). It can therefore be useful for studying rapid morphological changes at the surface of the embryo (Zamir et al.,2008). However, widefield epifluorescence microscopy obscures movements in the z-direction and cannot resolve depth. The image that is generated is a 2D projection of signal emitted from many layers of cells. Cells at the surface cannot be distinguished from those in deeper layers, making it very difficult to, for instance, tell which cells might be interacting, or even if apparent cell divisions are real or simply cells at different depths that had been traveling together subsequently moving apart from one another (see Fig. 4).

thumbnail image

Figure 4. Importance of imaging three spatial dimensions for cell tracking. Three spatial dimensions are necessary to interpret cell behavior when studying tissue more than one cell layer thick. Figures (ac) show confocal z-stacks of genetically labeled eYFP+ nuclei from a Tg (tie1:H2B-eYFP) quail embryo. Collecting data in each plane of focus preserves the depth of the sample. On the contrary, widefield epifluorescence imaging provides a 2D view, so cells at different layers cannot be distinguished. A similar effect can be achieved by projecting the z-stack onto a single 2D plane to flatten the image, or by viewing the volume from only one direction, as in (a) and (b). In (a), two cells at different depths appear as one single cell, and cells may travel together for considerable distances in such an alignment, obscuring the fact that they are actually distinct cells (cells of interest are marked by white arrowheads throughout the panels). (b) The cells may then begin to travel in separate directions, giving the appearance that a single cell has divided. When data is collected with widefield imaging, it can be very difficult to ascertain whether this is a true cell division. However, when depth is preserved as with confocal or 2P imaging, a simple rotation of the sample volume as in (c) demonstrates that these are indeed two individual cells at different depths, and not a cell division. Scale bars = 10 μm. Tg = transgenic, H2B = histone 2B, eYFP = enhanced Yellow FP.

Download figure to PowerPoint

Confocal and Two-Photon Imaging

Confocal and 2P microscopies are essential for distinguishing and tracking individual cells in tissues with multiple layers of cells. In confocal microscopy, light emitted from each point on the specimen is focused at a plane that is conjugate with the plane of focus of the objective lens that illuminates the specimen (Minsky,1961; Yuste and Konnerth,2005). A detector pinhole is used to restrict the width of the cone of light collected, and this in turn means that only photons emitted from within a restricted depth above and below the plane of focus are collected to form an image of that focal plane. Images are taken at successive depths to reconstruct a 3D image of the specimen that resolves structures in depth as well as in the plane of focus (Fine et al.,1988).

A disadvantage of confocal microscopy is that fluorophores at all depths of the specimen are activated by the incident laser light, even though the photons used to construct the image are restricted to the plane of focus. This unnecessary fluorophore activation can bleach the fluorophores and cause phototoxicity to the cells. In contrast, 2P microscopy generates resolution in the z dimension by only exciting fluorophores at the point of focus, and then all photons emitted in the direction of the objective are collected (Denk et al.,1990; Williams et al.,1994). To accomplish this, 2P microscopy uses femtosecond bursts of long-wavelength and high intensity light to excite fluorescent markers that typically absorb shorter wavelengths of light. 2P imaging is appealing because it causes less fluorophore bleaching and cellular phototoxicity than confocal microscopy since fluorophore excitation is restricted to the plane of focus. The longer-wavelength infrared light used for 2P microscopy also interacts less with biological samples. This further reduces tissue photodamage and permits imaging deeper within the specimen as fewer incident photons are scattered (Kulesa et al.,2009; McMahon et al.,2008,2010; Supatto et al.,2005).

4D Whole-Embryo Imaging

Time-lapse imaging can be performed using confocal or 2P microscopy to create a powerful approach called 4D or xyzt (xyz-in-time) imaging. Images are collected in overlapping slices through the desired depth of a specimen to cover an entire xyz volume of interest. This process is repeated at set time intervals to capture the dynamic behaviors of cells within the defined volume over time. The z-dimension or depth of study is increased by the adding additional slices in the z-stack. For the study of thick samples or regions beneath the surface of the specimen, 2P microscopy is preferred to improve signal at depth. The overall xy region of interest can be enlarged by collecting multiple overlapping z-stacks (tiles). The tiles are then stitched together using built-in microscope software or post-processing algorithms that align overlap between the individual fields of view (Canaria and Lansford,2010; Emmenlauer et al.,2009; Karen et al.,2003; Preibisch et al.,2009). Stitched time-lapse data sets can be analyzed using software such as ImageJ/Fiji, Bitplane Imaris, and Matlab (Canaria and Lansford,2010; McMahon et al.,2008,2010; Sato et al.,2010).

Using live embryo 4D imaging, embryonic morphogenesis can be studied with high spatiotemporal detail. Cells can be tracked in real time from their initial primordial sites to their final destinations in the developing amniote (see Fig. 5). Time-lapse imaging demands that the embryo must continue to develop and behave normally throughout the duration of imaging. To image a quail embryo successfully can be challenging: it must be cultured at physiologically appropriate temperature and CO2 levels, within the working distance of the objective, and with moderate laser power to delineate the cell behaviors of interest while minimizing phototoxicity in the tissues being imaged. To complicate the matter, collected images must have sufficient resolution to illustrate the cellular activities of interest, creating an inherent tradeoff between maximizing image quality and preserving specimen integrity, which must be optimized for each application. While many model organisms are less well suited for dynamic imaging, quail embryos are optically transparent and tolerate the conditions required for extended imaging of cells undergoing morphogenesis.

thumbnail image

Figure 5. Tracking cell movements over time. (a) DNA encoding GAP43 membrane-tethered CherryFP driven by the ubiquitous promoter phosphoglycerate kinase (PGK) was electroporated into the presomitic mesoderm of Tg(tie1:H2B-eYFP) quail embryos. The tie1+ nuclei are pseudocolored green and the membrane-CherryFP is red. In this snapshot from the z-stack time-lapse, a region of cells is identified for tracking over time. The cell marked with a white arrowhead in (a) is also indicated in (b). (b) A subset of cells in image (a) was tracked via the locations of their nuclei in each time-step of a 12 hour time-lapse. Deeper layers of cells were omitted for clarity. The squiggly white lines mark the paths traveled by each green nucleus over this time course up to the time point displayed in (b). Tracking of individual cells during morphogenesis reveals which progenitors contribute to each tissue and how they arrive at their final destinations. Viewing the track projections conjointly also provides valuable information about collective cell migration, tissue rotation, or other global morphogenetic patterns. Grid tick marks = 10 μm. Tg = transgenic, H2B = histone 2B, eYFP = enhanced Yellow FP, FP = fluorescent protein.

Download figure to PowerPoint

Transgenic quail embryos can be imaged, both in vivo (LaRue et al.,2003; Sato et al.,2010) and in ovo (RL, unpublished data), using microscope incubation chambers that maintain conditions permissible for development, as previously described for chicken embryos (Kulesa and Fraser,1999). Setting the incubation temperature around 36°C, just below physiological temperature, accommodates slight additional heating from the imaging lasers. Using confocal or 2P microscopy, it is feasible to collect one or multiple z-stacks every 5–10 min for up to 36 hours (Sato et al.,2010). Confocal microscopy permits imaging to a depth of about 100–180 μm before too much light is lost to scattering, while 2P can resolve ∼150–300 μm in the z-dimension (DVB and RL, unpublished observations). These conditions are sufficient to capture all the cell divisions and movements that generate the various developing embryonic tissues.

Optimizing Time-Lapse Imaging Settings

To maximize both resolution and field-of-view for 4D imaging, 10×–40× long-working distance objectives with high numerical aperture are preferred. The scan speed, number of slices in the z-stack, and number of tiles must be optimized to collect a complete, single or tiled z-stack within about 1–12 min (Canaria and Lansford,2010). The speed of cell behaviors being observed and the magnification and resolution required to discern them will dictate the optimal collection parameters [(Vermot et al.,2008); RL and DVB, unpublished data].

For instance, in Hamburger-Hamilton stage 6 quail embryos, one field of view imaged with a 20× objective is sufficient to image the ∼450 by 450 μm region of cell migration and involution at the primitive streak. Images can be collected approximately every 6 min to track cells migrating at 50–100 μm per hour. This contrasts with imaging organogenesis at later stages of development, where, for instance, three or four overlapping fields of view may be necessary to capture the formation of the mesonephros, but images can be collected every 10–12 min to track the locations of less rapidly moving cells. However, if the aim is to analyze the interactions between cells and changes in cell shape, then images must be taken at the minimum of twice the frequency of the changes, which generally necessitates imaging smaller regions and/or collecting images of 256 pixels2 (px2) to achieve the required time resolution (Vermot et al.,2008). Nevertheless, there is a tradeoff between time and spatial resolution. While collecting a 256 px2 image is four times faster than collecting a 512 px2 image, the x and y resolutions of the 256 px2 image are double (two-fold worse than) those obtained with a 512 px2 image. Increasing to 1,024 px2 gives a two-fold improvement in resolution over 512 px2, but again at the expense of time (by a factor of 4). Thus, the compromise between time and xy spatial resolution must be optimized for the aim of each study. Similar considerations apply to depth resolution. To resolve structures in the axial (z)-dimension, a good rule of thumb is that the thickness of each z-stack slice must be no more than half the axial thickness of the structure to be resolved. Even under these conditions, the structure may appear the be up to one and a half times its true thickness, so taking thinner overlapping slices improves resolution, but as always, this resolution is achieved at the expense of the time required to collect the stack of images and with potential phototoxicity to cells.

For the instances when the laser power and scan time can be diminished no further without compromising image quality and resolution, a few techniques exist to mitigate potential photodamage to cells. Biological tissue is less affected by lower-energy, longer-wavelength light, so using fluorophores that excite in the far-red range, or imaging with 2P microscopy, reduces cellular phototoxicity. 2P imaging also concomitantly reduces overall tissue exposure, as mentioned previously, because photon energy is confined to the focal plane. All of these considerations must be taken into account when planning experiments, as the best data sets are produced by optimizing the time, resolution, field of view, and photon intensity requirements.

MULTISPECTRAL IMAGING TECHNIQUES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Spatial and Spectral Separation

Capturing the complex activities of cells undergoing morphogenesis often requires concurrent imaging of multiple spectrally distinct labels. Both traditional confocal laser scanning microscopy and 2P laser scanning microscopy can effectively image several fluorescent labels in the same sample using spectral imaging approaches. Fluorophores with divergent spectra can be used, or computational methods can be employed to simultaneously resolve the signals from dyes or FPs with overlapping spectra (Dickinson et al.,2001; Lansford et al.,2001). Quantitative multispectral imaging enables precise extraction of the spectral signature of each label and distinguishes them from cellular autofluorescence. This technique even differentiates between fluorophores targeted to the same subcellular site (Teddy et al.,2005). A simpler method for simultaneously imaging several labels is to localize each spectrally distinct fluorophore to a separate subcellular location, thereby capitalizing on the benefits of multiple colors to elucidate cellular behaviors while enabling the use of traditional spectral filters to separate the FP signatures (LaRue et al.,2003).

It is also generally difficult to distinguish cells from one another in embryos in which just the cytoplasm or cell membrane are fluorescent because there is no clear visible separation between adjacent cells. This is particularly a problem when a single color labeling scheme is used (see Fig. 6). Nevertheless, experiments can be designed to avoid such issues by either labeling scattered cells in a tissue or labeling different cells different colors. First, RNA polymerase (RNAP) II tissue-specific promoters can be used to confine FP expression to desired cells or tissues. This reduces the number and uniformity of labeled cells. Alternatively, diluting the infection or electroporation volume results in sparser labeling of cells to create a “mosaic” or “salt and pepper” expression pattern. However, multispectral techniques offer a superior approach to distinguish closely opposed cells in dense tissue (Feng et al.,2000; Hadjantonakis et al.,2002; Hutter,2004; Kulesa et al.,2010; Livet et al.,2007). In this case, cells are labeled one of multiple colors, and concentrations are titered to ensure that adjacent cells are liable to express different colored labels and thus are plainly distinguishable from one another. This approach can be applied to differentially label nuclei, cytoplasm, or plasma membranes of adjacent cells in distinct colors to facilitate individual cell tracking. Additionally, new FPs are being engineered to help address the challenges of identifying individual cells in dense tissue and to enable selective marking of particular cells of interest during the course of imaging.

thumbnail image

Figure 6. Obscured cell boundaries with monochromatic membrane label. (ac) The panel shows three snapshots from a time-lapse of cells labeled with membrane-localized CherryFP described in the previous figures. When nearly all cells of a tissue are labeled with cytoplasmic or membrane label of the same color, as they are in this local collection of cells, it is difficult to distinguish between cells because there is no separation between them. It is even difficult to discern the boundaries of more brightly labeled cells against the backdrop of lesser label (Fig. 6a, approximate cell boundaries marked by a dotted white outline). The cell outlined in yellow is somewhat distinct in (a), but it becomes less identifiable in (b) as it moves closer to the pack of cells, and is indistinguishable from adjacent cells in (c), approximate cell boundaries marked in yellow. Scale bars = 15 μm. FP = fluorescent protein.

Download figure to PowerPoint

Photoactivatable and Photoconvertible FPs

An expanding collection of photoactivatable and photoconvertible FPs can be used independently or in combination with other FPs to visualize cell behaviors within the quail embryo (Chudakov et al.,2010; Nowotschin and Hadjantonakis,2009a; Patterson,2008). The photoactivatable green FP (paGFP) enables enhancement of label in single cells or groups of cells in densely populated tissue. paGFP is a variant of wild type GFP that was engineered to emit about a 100-fold increase in green fluorescence upon irreversible conversion to an anionic form induced by 405 nm laser light (Patterson and Lippincott-Schwartz,2002, 2004).

Other FPs have been developed that convert from one spectral signature to another upon a conformation change induced by radiation with a specific wavelength of light. One such example is the Kaede (“maple leaf”) protein, which is green until activated with 405 nm light, after which it fluoresces red (Ando et al.,2002). A distinct advantage of photoconvertible proteins such as Kaede is that prior to any photoactivation, labeled cells can be identified by the green label and followed with 488 nm radiation. The 488 nm photons do not convert the protein, so imaging can be performed before photoconverting to red emission with 405 nm laser excitation. The large (2,000-fold) increase in red-to-green signal intensity upon photoconversion permits visualization of labeled structures even when relatively few FP molecules are expressed.

The Dendra2 and mEos2 green-to-red photoconvertible proteins offer the additional advantage that they are inherently monomeric as opposed to tetrameric (as Kaede is), making them more suitable for fusion to cellular proteins (Adam et al.,2009; McKinney et al.,2009). Dendra2 can also be photoconverted by less phototoxic 488 nm light, in addition to 405 nm light (Adam et al.,2009; Gurskaya et al.,2006). These and other photoactivatable and photoconvertible proteins, such as KikGR, PS-CFP2, Dronpa, and others are promising probes for analyzing the behaviors of tagged proteins in time and space in living embryos using confocal or 2P microscopy (Ando et al.,2002,2004; Chudakov et al.,2003,2004,2010; Falk et al.,2009; Gurskaya et al.,2006; Habuchi et al.,2008; McKinney et al.,2009; Nowotschin and Hadjantonakis,2009b; Pantazis and González-Gaitán,2007; Patterson,2008; Post et al.,2005; Stark and Kulesa,2007; Tachibana et al.,2008; Wacker et al.,2007).

Individual cells or groups of cells can be photoactivated or photoconverted to distinguish them from surrounding cells for easier visualization and tracking. The labeling pattern that is desired will determine whether confocal or 2P imaging is most appropriate for activating or converting the FP. Confocal laser will activate FPs at all depths within the path of the laser. However, 2P imaging only provides sufficient excitation energy within a couple micron-diameter volume at the point of focus (Denk et al.,1990; Williams et al.,1994). Thus, using 2P laser illumination is advantageous for single cell labeling to prevent out-of-plane photoactivation, and it is also more effective to activate cells deep within a tissue (Kulesa et al.,2009; Pantazis and González-Gaitán,2007). Still, the volume marked for photoactivation must be entirely within the cell of interest, otherwise multiple cells may be photoactivated. Laser intensity must also be minimized to avoid phototoxicity or ablation of the cells. Generally 100 mW or less of laser power with ∼130–160 femptosecond pulse duration at 60 MHz is optimal for 2P photoactivation [(Pantazis and González-Gaitán,2007); DVB, unpublished data].

Optical targeting using confocal or 2P photoactivation or photoconversion of specific cells followed by time-lapse confocal or 2P image acquisition to track the cells offers an efficient, minimally invasive technique for tracing cell behaviors and movements in living embryos (Kulesa et al.,2009,2010). For instance, studies in the chick labeled neural crest cells with photoinducible vectors by electroporation into the hindbrain to characterize their migratory behaviors (Kulesa et al.,2008; Stark and Kulesa,2007). Dynamic time-lapse studies can also be semiautomated by photoactivating multiple regions within a single embryo and capturing multiple data sets at a time (Kulesa et al.,2009).

The advent of photosensitive regulatory proteins that permit activation or inactivation of specific behaviors has allowed more precise spatial and temporal study of the activities of these proteins (Chudakov et al.,2010; Wu et al.,2009). In an elegant example, investigators used a reversibly photoactivatable cytoskeletal protein Rac1 to induce cell membrane protrusions and retractions and direct cell migration, in tissue culture and in vivo in D. melanogaster (Wang et al.,2010; Wu et al.,2009). The ability to manipulate protein behavior in a cell has enabled them to study Rac1 inhibition of another cytoskeletal protein, RhoA, and by using biosensors that fluoresce upon activation of a target protein, they have precisely measured the coupling of cytoskeletal GTPase activation with membrane activity in time and space (Machacek et al.,2009; Nalbant et al.,2004; Wang et al.,2010). The creation of additional photoactivatable structural, regulatory, signaling, and transcriptional proteins will permit increasingly precise study of the behaviors of individual cells and proteins and how these cellular processes impact tissue morphogenesis and embryological development.

IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

An exceptional characteristic of avian embryos is the ability to combine classical embryological manipulations, such as tissue grafting experiments, with addition of exogenous factors, like biologically active molecules, for studying the effects of molecular perturbations or disease states. The accessibility of the avian embryo facilitates in vivo manipulations, such as ectopic expression of desired proteins, including integrin molecules, chemotactic or growth factors, or their naturally occurring inhibitors, or knockdown or ablation of proteins or genes. Critically, these manipulations can be performed quickly and dependably with spatiotemporal precision.

Studying Embryologic Origins of Disease

Many diseases originate from a failure of the embryo to develop properly (Bader et al.,2004; Desmet,2005; Fagman and Nilsson,2010; Francis-West et al.,2003; Geng and Oliver,2009; Groenman et al.,2005; Guerrini and Barba,2010; Jiang et al.,2006; Marchuk et al.,2003; Morello and Lee,2002; Nath et al.,2009; Oystreck et al.,2011). Fundamental insight into the rules of development and the mechanisms of disease onset can be gained from examining the interface of embryology and pathology. Independent lines of quail that recapitulate human disorders have been bred to study numerous conditions, such as atherosclerosis, diabetes, endocrine disruption, and Pompe's disease (Kikuchi et al.,1998; Minvielle et al.,2007; Ottinger et al.,2005; St Clair,1998). Transgenic quail models for human diseases can also be generated using gain of function and loss of function vectors, for instance with short hairpin RNA (shRNA). Such quail disease models will permit tracing individual cells from an initial embryonic insult to their adult disease manifestation as the disease process unfolds. These studies will help elucidate the mechanisms by which a disease state develops: understanding how specific cells fail to arrive at their proper destination, or where in the developmental process structures fail to form, or how disrupted signaling pathways have indirect downstream effects can all contribute to finding new interventions for disease.

Targeted Inactivation of Genes

The targeted gene inactivation approach is a powerful technique that has commonly been used in mice to determine the functional significance of genes (Colvin et al.,2011; Graziotto et al.,2011; Iotsova et al.,1997; Krege et al.,1998; Meyers et al.,1998; Sanford et al.,1997). Likewise, in quail embryos, the functions of cells and genes during development can be studied using conditional or constitutive lineage ablation or RNA interference (RNAi) (Hunter et al.,2005; Kim and Dymecki,2009; Metzger et al.,1995; Nagy,2000; Nov et al.,2007; Sosa et al.,2010; Wang et al.,2007). RNAi is a powerful technique whereby an antisense RNA probe hybridizes with a complementary cellular mRNA and either impedes protein translation or mediates the degradation of that mRNA, resulting in down-regulation of specific genes (Bartel,2004; Bernards,2006; Cullen,2004; Elbashir et al.,2001; Fire et al.,1998; Hammond et al.,2000; Lau et al.,2001; Lee et al.,1993,2002; Lykke-Andersen,2006; Meister and Tuschl,2004; Olsen and Ambros,1999; Wightman et al.,1993). These techniques can be applied to dynamically study embryogenesis and morphogenesis.

Conditional knock-down of gene expression in transgenic embryos can be accomplished with Cre-dependent or other inducible RNAi vectors or tissue-specific vectors (Gao and Zhang,2007; Hitz et al.,2007,2009; Kleinhammer et al.,2011; Matsuda and Cepko,2007; Tavernarakis et al.,2000). These vectors can use the microRNA (miR)-30 or miR-155-based shRNA expression system or analogous expression constructs (Chung et al.,2006; Matsuda and Cepko,2007; Paddison et al.,2004; Zeng et al.,2002). With this system, the microRNA of interest is designed into the human miR-30 or equivalent miR precursor sequence within a longer mRNA transcript, which is subsequently cleaved to extract the mature microRNA (Zeng et al.,2002). Compared with conventional RNAi vectors that express shRNAs from RNA polymerase III promoters such as the U6 promoter, these miR-based shRNA expression systems display several advantages. First, RNA polymerase II promoters can be used to express shRNAs (Chung et al.,2006; Du et al.,2006; Zeng et al.,2002). This has critical implications because tissue-specific RNA polymerase II promoters can drive microRNA expression and mRNA inactivation in a tissue-specific or inducible manner (Hitz et al.,2007; Stegmeier et al.,2005). Second, these constructs can be delivered to cells by simple plasmid transfection or viral infection (Stegmeier et al.,2005). Third, engineering regulatory elements, such as a transcriptional stop cassette, into expression vectors is straight-forward. Finally, multiple different shRNAs, tandem copies of the same shRNA, and shRNAs plus an FP can be coexpressed from the same plasmid or lentivector for multiplexed gene knock-down and for clear visual identification of labeled cells (Chung et al.,2006; Stegmeier et al.,2005). These shRNAs can be designed using algorithms tested for hybridization based on thermodynamic parameters (Matveeva et al.,2010; Shabalina et al.,2006; Walton et al.,2010).

Once infected with the shRNA expression vector, labeled embryos can then be analyzed using dynamic multispectral imaging techniques. Impacts on development from knocking down expression of one or more critical genes in live embryos can be measured in real time to assess the importance of environmental cues versus genetically encoded instructions for tissue morphogenesis. Resulting malformations of tissues or organs, delayed or stalled development, morphological deformities, and other effects can be characterized. These tools can be applied to study gross morphogenetic movements of tissues on a whole-embryo scale as well as intimate interactions between individual cells of a particular tissue.

IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Post-collection image processing poses equally formidable challenges to those presented by multispectral 4D image collection (Blanchard et al.,2009; England and Adams,2007; Megason and Fraser,2007). Dynamic time-lapse imaging methods generate 3D sequences of moving cells from which discernible lineages, migrations, and behaviors must be resolved. Comprehensive lineage analysis may involve analyzing thousands of cells moving convergently and divergently within the developing embryo, necessitating sufficient spatiotemporal resolution to follow every individual cell. Single-cell analysis is vital when inspecting the origin of cells in complex tissues as only clonal analysis can confirm, for instance, whether a single progenitor can engender multiple differentiated cell types.

Interdisciplinary collaborations between developmental biologists and applied mathematicians have successfully facilitated the derivation of algorithms that evaluate these complex data sets and properly identify and track the complicated movements of cells within living embryos (Berlemont and Olivo-Marin,2010; Dufour et al.,2005). Individual precursor cells and their subsequent progeny can be color-coded somewhat automatically with colocalization algorithms to simplify cell tracking and cell fate analysis of the image sets. Alternatively, labeling cell nuclei, cytoplasm, filamentous actin, or other structures in different colors facilitates segmentation of structures and analysis of cell movements and behavior. Label confined to the nucleus can be identified computationally due to the clearly defined boundaries, enabling automated tracking of cells by their nuclei, even in dense tissue (Bao et al.,2006; Kulesa et al.,2010; McMahon et al.,2008; Sato et al.,2010). Time-lapse data sets can also be quantitatively mined to characterize cell divisions and orientation, speed and movement trajectories, membrane dynamics and changes in shape, and cell spacing and collective motion to generate a comprehensive, quantitative portrayal of cell behaviors (England and Adams,2007; England et al.,2006; Sato et al.,2010). Integrating the tools of cell biology and systems biology is essential to interpret the mechanics of cell behaviors and more fully comprehend the driving forces of developmental biology.

As targeted labeling techniques, multispectral imaging approaches, and image processing algorithms continue to progress, multispectral imaging of living embryos is becoming an increasingly powerful tool for dynamic analysis of cell lineages, behaviors, and morphogenesis (Joyner and Zervas,2006; Kulesa et al.,2010; Teddy et al.,2005).

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

We contend that dynamic microscopic imaging of transgenic quail lines permits unparalleled comprehension of the behaviors of cells during amniote embryogenesis and organogenesis, providing valuable insight into how cell orientation, polarity, division, intercellular interactions, and mechanical stress drive morphogenesis. Considerable breakthroughs have been made in time-lapse microscopy and computational analysis–both of which portend innovative advancements in the emerging field of dynamic lineage analysis. As our imaging goals continue to increase in complexity, so will the demand for novel tools like photoactivatable recombinases, transcription factors, and receptor molecules to dynamically label and manipulate samples, more advanced imaging systems for rapid scanning of large fields of view at high resolution, and faster data analysis algorithms to process data in real time and enable live feedback from data to inform and optimize data collection during the course of imaging.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

In vitro Imaging

The pre-somitic mesoderm of Tg(tie1:H2B-eYFP) quail embryos was electroporated at Hamburger-Hamilton (HH) stage 9 with plasmid DNA encoding GAP43 membrane-tethered CherryFP driven by the ubiquitous promoter phosphoglycerate kinase (PGK). Avian embryos were cultured ex ovo as initially described by New (1955) and modified as described (Sato et al.,2010; Zamir et al.,2008). Embryos were extracted from the egg using filter paper rings and mounted ventral side up on a thin, semi-solid mixture of albumen and agar.

A custom-made chamber surrounding the microscope stage controlled the temperature at 37°C during the course of imaging. Confocal imaging was performed on a Zeiss 510 META laser scanning microscope using a Plan-Apochromat 20×/0.8 objective starting at approximately stage HH11.

Imaging parameters: Confocal: 488 nm (13%), 561 nm (9.9%), z-stack size: 512 × 512 × 54 px; 636.4 × 636.4 × 159.0 μm (3 μm z-interval) for 80 time points every 8 min (pixel dwell time = 2.56 μs).

Computational Analysis

Imaris software (Bitplane, Zurich, Switzerland) was used for image processing and cell tracking analysis of the time-lapse video data set. A 3 × 3 × 3 px median filter was applied for noise reduction in the green channel. Labeled nuclei were tracked as spots using an autoregressive motion algorithm, which models the motion of each object as an autoregressive AR1 process. The minimum travel distance for each spot from one time point to the next was 15 μm. Tracks generated automatically by Imaris software were also individually checked and corrected by hand. Levels on images were adjusted using Adobe Photoshop to match the full 8-bit output range of the collected intensity histogram to make images more clear when printed.

LITERATURE CITED

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. APPROACHES TO MARKING CELLS FOR LINEAGE ANALYSIS OF EMBRYOGENESIS
  5. MULTISPECTRAL DYNAMIC IMAGING TO DECIPHER TISSUE MORPHOGENESIS
  6. MODEL SYSTEM CONSIDERATIONS FOR DYNAMIC IMAGING
  7. METHODS TO MARK CELLS OF INTEREST FOR DYNAMIC TRACKING
  8. IMAGING MODALITIES
  9. MULTISPECTRAL IMAGING TECHNIQUES
  10. IN OVO MOLECULAR PERTURBATION FOR FUNCTIONAL STUDIES
  11. IMAGE PROCESSING AND QUANTITATIVE ANALYSIS OF MOVING CELLS
  12. CONCLUSION
  13. MATERIALS AND METHODS
  14. Acknowledgements
  15. LITERATURE CITED
  16. Supporting Information

Additional Supporting Information may be found in the online version of this article.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.