Evolution of development studies are a logical next step in furthering our understanding of cambial function and evolution, through synthesis of anatomical, developmental genetic, and genomic approaches and data within a phylogenetic context (Cronk, 2001). Evolution of development approaches can address questions central to secondary vascular growth, such as: What are the evolutionary origins of secondary vascular growth in angiosperms and gymnosperms, and was there a single or multiple origins? What are the ancestral genes and mechanisms regulating secondary vascular growth? What is the genetic basis for observed phylogenetic variation in secondary vascular growth, including cambial variants? As discussed in Section three, a crucial missing component of current research is comparative studies of regulatory genes and mechanisms across taxa.
1. Evolution of development studies require synthesis of approaches and data types
Evolution of development studies of secondary vascular growth will require multiple steps. First, taxonomic relationships among plants must be determined with reasonable certainty and precision. Taxonomic relationships are increasingly well resolved at various taxonomic levels for large numbers of plants through the construction of DNA sequence-based phylogenies (Angiosperm Phylogeny Group, 2003; Palmer et al., 2004). Second, developmental and anatomical variation in secondary vascular growth should be identified at different taxonomic levels, ranging from generalized traits at broad taxonomic levels to more unusual or subtle traits at lower taxonomic levels. Examples of such variation have already been discussed in section III, and there is an extensive wood anatomical literature and data that can be referenced (e.g. Carlquist, 2001; http://insidewood.Lib.Ncsu.Edu/search). Third, major regulatory genes and mechanisms must be identified using model plants for which extensive genomic tools and the ability to assess gene function (e.g. through transformation) are available. As discussed above, this work is incomplete but is accelerating with the availability of the Populus genome (Tuskan et al., 2006) and functional genomic tools. Fourth, the development of new woody plant models at key taxonomic positions must be undertaken to enable comparative studies of gene function. Lastly, variation for both phenotypes and underlying genetic regulatory genes and mechanisms must be surveyed in species at appropriate taxonomic positions, which show variation for traits of interest (Soltis & Soltis, 2003).
2. New woody model species should have desirable attributes and taxonomic positions
Currently there are significant genomic resources for a handful of woody species, most of which were selected because of their economic importance. For gymnosperms, there are large numbers of expressed sequence tags (ESTs) and genetic maps for several members of the Pinales, including species of Pinus, Abies and Picea (Dean, 2006). Ongoing development of existing Pinales models is accelerating, although transformation is limiting for many species. Similarly, there are a number of angiosperm species with significant EST sequences and other resources, although all are clustered within the Rosids. Notable are the forest trees of Populus (Tuskan et al., 2006) and Eucalyptus (http://eucalyptusdb.bi.up.ac.za/), the secondarily woody perennial papaya (Carica papaya) (Ming et al., 2008), and the woody vines of Vitus (The French–Italian Public Consortium For Grapevine Genome Characterization, 2007), for which full genome sequences are available, and members of the Fagaceae, for which extensive sequencing is underway (Fig. 3). However, some of these current models lack desirable attributes for model species (see below paragraph), and are not fully representative of taxonomic variation desired for robust comparative studies of gene function.
There are several desirable attributes of new woody plant models. Importantly, new models must be amenable to detailed analysis of gene function. In general for woody perennials, this requires the ability to transform with transgenes, which can be used to knock down or change the expression of genes of interest. Interestingly, novel approaches have been developed for transformation of cambial tissue with Agrobacterium after bark removal (Van Beveren et al., 2006), which could be applicable even to currently recalcitrant species. In addition, it is highly desirable to be able to perform controlled crosses and produce pedigrees for genetic mapping, which supports efforts ranging from whole genome sequencing to analysis of quantitative trait loci. Important variables influencing construction of pedigrees for woody perennials include the time to sexual maturity and degree of inbreeding depression. It is encouraging that many other important techniques, including sequence-based evaluation of gene expression and various ‘omics’ technologies, are relatively easy to apply to new organisms. There are, however, significant practical hurdles in establishing new models, including an often limited number of researchers working on the model who must establish and curate databases, annotations, bioinformatic tools, and germplasm while making research progress (Abzhanov et al., 2008).
For addressing questions concerning the ancestral origins of vascular cambia in angiosperms and gymnosperms, Ginkgo biloba could be a valuable new model gymnosperm. Ginkgo is a basal gymnosperm that has changed surprisingly little over hundreds of millions of years, and culture methods have been reported that could support transformation (Dupre et al., 2000). Ginkgo could be used as a highly complementary reference to models being developed for more derived Pinales.
There are a number of potential model woody species that could be developed to assess broader taxonomic variation within angiosperms (Fig. 3). These models would provide opportunities for comparative functional developmental and genomic studies that could address the degree of variation in core mechanisms regulating secondary growth, as well as evidence for homologous origins of such mechanisms. For example, the magnoliid Liriodendron tulipifera (tulip tree, order Magnoliales) is an attractive candidate, being a basal angiosperm and large forest tree for which a somatic embryogenesis-based transformation system is available (Dai et al., 2004). There are also existing pedigrees from forest tree improvement programs, and L. tulipifera is interfertile with the Asian Liriodendron chinense. Platanus species (sycamores, order Proteales) are basal basal eudicots, can be transformed, and have existing pedigrees. Liquidambar styraciflua (sweet gum, order Saxifragales), within the core eudicots (but outside the rosids), also has the benefit of transformation systems (Dai et al., 2004) and pedigrees. Within the core eudicots, existing rosid woody models (including fully sequenced Populus and Eucalyptus) would be complemented by comparisons with woody members of the asterids such as Fraxinus spp. or Paulownia spp. (see paragraph below) for which transformation (Giri et al., 2004) and pedigrees are available. In addition, all of these species are of environmental and/or economic importance. Even more challenging will be comparative studies between angiosperms and gymnosperms. While initial comparisons have demonstrated the ability to recognize general homologies between genes from A. thaliana and pines (Kirst et al., 2003), determining orthologous and functional relationships will be challenging.
Selection of new angiosperm woody models should also maximize the information gained from relationships to existing models, including comparisons with nonwoody species (Fig. 3). Selection of woody species within families with highly developed herbaceous annual models could allow for powerful comparative studies of herbaceous annuals and woody perennials. For example, the order Fabales includes both the sequenced Medicago (Cannon et al., 2006) and several notable tree species of the Acacia family. Furthermore, selection of new models should include consideration of variation for secondary vascular growth in closely related plants, including cambial variants. For example, the order Lamiales occupies a key taxonomic position within the Asterids, and also contains important variation ranging from the forest trees of Oleaceae (e.g. Fraxinus spp. (ashes)) and Paulowniaceae (e.g. Paulownia spp.), to the previously mentioned lianas within the family Bignoniaceae with amazing diversity in secondary vascular growth (Pace et al., 2009).
3. Genomic approaches can identify genes regulating development in model species, and survey phylogenetic variation
A comprehensive strategy for evolution of development studies of secondary growth will begin with detailed characterization of the regulation of secondary growth using functional genetic and genomic tools in taxonomically diverse model species. Currently, this strategy is best illustrated by Populus, where combinations of developmental genetic studies using transgenesis and genomics are revealing major regulatory genes and mechanisms. A desirable next step is to move from single-gene views of development resulting from transgenesis-based developmental studies and genomics studies that are largely descriptive, to modeling of biological networks (e.g. transcriptional networks) underlying key secondary growth traits (Du & Groover, 2010). Network-level models of secondary growth would provide new levels of resolution of regulatory mechanisms, provide predictive capabilities to inform new research, and directly support detailed comparative studies of gene expression and regulation. Specifically, network-level models of secondary growth regulation could identify both putative basal regulatory genes and regulatory modules that are shared among diverse taxa, as well as genes or modules whose expression or function may be variable and responsible for observed phenotypic variation.
As sufficient knowledge of regulatory genes and networks is developed in model species, surveys of additional species based on comparative gene expression during secondary growth can be used for comparative analysis outside of fully developed models. Genomic and sequencing technologies are increasingly extensible to new species, a feature that is highly supportive of comparative surveys that can include species that do not enjoy the full range of tools available for model species. For example, large-scale sequencing of ESTs can be accomplished now at reasonable cost using next-generation sequencing (Mardis, 2008; Schuster, 2008), and requires only the ability to isolate high-quality RNA from appropriate tissues. Such sequencing efforts can provide a wealth of information, including evolutionary histories, such as gene duplication events that often underlie important new evolutionary novelties through acquisition of new protein function by a duplicated gene, subfunctionalization of the original functions between paralogs, or acquisition of new expression patterns by duplicated genes (Ganfornina & Sánchez, 1999). Importantly, high-throughput sequencing not only assays sequence variation, but can also provide information about gene expression levels through quantification of the frequency at which a given gene’s transcript appears in a sequence run (Mardis, 2008). Changes in expression levels can be reflective of evolutionarily significant mutations in cis regulatory elements or activity of trans-acting factors (e.g. transcription factors). For sequencing-based surveys, a highly informative tissue for assay would be cambial and developing xylem tissue, which can typically be harvested in relatively large amounts from actively growing stems, and would allow simultaneous assay of both cambial meristem regulatory genes and genes involved in cell differentiation and wood formation.
4. Computational approaches and databases are central to evolution of development studies of secondary growth
Development of computational methods and bioinformatics tools, database creation and curation, gene and genome annotation, and curation of biological stocks will all present major challenges for establishing the field of evolution of development for secondary growth. Luckily, many of these needs are shared by other communities (Abzhanov et al., 2008), and major efforts have been undertaken to address at least some of these needs by creation of generalized resources. For example, the Generalized Model Organism Database (GMOD) tools provide ‘off the shelf’ database and informatic tools which can be relatively easily extended to new species (http://gmod.org/wiki/Overview). Other efforts (e.g. TAVERNA; http://taverna.sourceforge.net/) are underway to allow users without extensive informatics resources to create data manipulation and analysis pipelines through selection, modification, and joining of modular scripts. Examples of sophisticated database and analysis tools for comparative genomics across taxa include PHYTOZOME (http://www.phytozome.net/). In short, while the challenge is significant, it seems likely that database and informatic tools are increasingly accessible and could be effectively leveraged to allow even a small community of researchers to undertake ambitious evolution of development studies for secondary growth. While beyond the scope of this review, projects using genomic approaches (e.g. association mapping; Neale & Ingvarsson, 2008) to understand genetic variation responsible for variation in wood traits are underway in several woody species including Pinus and Populus spp. These studies will be synergistic with the comparative studies described above.