Precise, flexible and affordable gene stacking for crop improvement

ABSTRACT

The genetic engineering of plants offers a revolutionary advance for crop improvement, and the incorporation of transgenes into crop species can impart new traits that would otherwise be difficult to obtain through conventional breeding. Transgenes introduced into plants, however, can only be useful when bred out to field cultivars. As new traits are continually added to further improve transgenic cultivars, clustering new DNA near previously introduced transgenes keep from inflating the number of segregating units that breeders must assemble back into a breeding line. Here we discuss various options to introduce DNA site-specifically into an existing transgenic locus. As food security is becoming a pressing global issue, the old proverb resonates true to this day: “give a man a fish and you feed him for a day; teach a man to fish and you feed him for a lifetime.” Hence, we describe a recombinase-mediate gene stacking system designed with freedom to operate, providing an affordable option for crop improvement by less developed countries where food security is most at risk.

Introduction

The world population tripled from 2.5 billion in 1950 to 7.4 billion in 2015 and much of that is attributed to the less developed world that increased 3.6-fold from 1.7 billion in 1950 to 6.1 billion in 2015. With this trend, the projected population in less developed regions will reach up to 8.4 billion by 2050.¹ Yet the world per capita arable land has been decreasing from 0.45 hectares in 1960 to a projected 0.18 hectares by 2050, and even lower for the less developed regions at 0.14 hectares.² Not only will our world need to grow more food, but we must grow them with greater efficiency than in the past.

Although there are various solutions to increase food production, one solution that cannot be ignored is the bioengineering of crops through the tinkering of DNA. In 2015, global bioengineered (genetically modified) crops covered ∼180 million hectares, of which a third of them contain more than a single trait, or stacked traits. More than half (54%) of global biotech crops were grown by developing countries from Latin America, Asia and Africa.³ Despite the fact that to commercialize a biotech crop is not trivial, requiring a decade and a hundred million US dollars,⁴ the investment in plant molecular genetics research in public and private sectors should produce a steady stream of new trait genes for crop improvement. This will inevitably offer the opportunity to bioengineer better crops through transgenesis or gene editing.

In transgenesis, a transgenic trait not only needs to be incorporated into a crop genome, but also must be bred out to elite lines that grow in the field. Due to the technical difficulty in transforming elite lines (field cultivars), transformation is typically conducted with a laboratory line (or a field line that can be transformed). The transgenic trait is then introgressed into a variety of different elite lines through repeated backcrossing to a recipient genotype. This practice may not necessarily be disadvantageous, as once a trait is deregulated, i.e., approved by the regulatory authorities, that one deregulated event can then be bred out to the many commercial cultivars that grow in different parts of the world. The conversion from a laboratory line to an elite line requires that non-elite traits are crossed out of the final field cultivar. For crops that behave genetically as diploids, if the relevant genetic difference between a laboratory line and an elite line (or between one elite line of one location and another elite line of another location) resides in 8 different unlinked loci, then the probability of assembling 8 segregating elite loci into a homozygous individual from their cross hybridization would be (1/4)⁸ or 1 individual plant out of 65,536. Adding a first transgene, representing another independently segregating locus, would increase the probability to (1/4)⁹ or 1 individual plant out of 262,144. As more new traits are discovered through the years, and more transgenic loci are added to the genome at new locations, this would make the line conversion process more and more difficult. For example, adding another transgenic locus to the above example would yield the probability of assembling 10 independently segregating loci as (1/4)¹⁰ or 1 homozygous individual out of 1,048.576, and this illustration does not even factor in the possibility that some traits exhibit linkage drag, i.e., they do not segregate independently.

In practice, a breeder cannot possibly obtain such large pool of progenies from a single cross, but instead could only assemble several traits from each round of backcross. Nonetheless, the statistical probabilities illustrate the rapidly increasing difficulty with introgressing more and more transgenic traits that segregate independently, resulting in more and more backcrosses needed. To make matters worse, line conversion is not conducted just with a single elite line, but with numerous elite lines grown commercially in many parts of the world. This makes the breeding process a bottleneck step in the bioengineering of crop plants.

There are several solutions around the problem of increasing transgenic loci. The first is the “do-over” strategy. That is, multiple transgenes can be packaged into one single transformation event. This can be done through stacking multiple transgenes in vitro into a single DNA cassette, and the cassette can be introduced into the plant genome as a large contiguous DNA fragment. This means that previously introduced traits will be combined with new traits to integrate the entire package into a new genomic location. Developers will have to screen the hundreds to thousands of random integration events to find an insertion with appropriate expression of all transgenes introduced, with an integration structure that is relatively precise, and without causing unwanted changes in the plant due to its integration event. Moreover, although old traits had received prior deregulation in previously introduced cultivars, they will nonetheless have to go through the deregulation process again since the new integration event is no longer the same as in previously deregulated cultivars. This approach would have developers expending greater effort in selecting for a line with field efficacy of the numerous transgenic traits, as well as in deregulating and re-deregulating the numerous transgenes for commercial release.

Another solution is the “direct transformation into elite line” strategy. By advancing transformation efficiency of elite cultivars, or cultivars genetically closer to field cultivars, developers can exercise the option of introducing new transgenes directly into lines that will require fewer backcrosses to field cultivars. Success in transforming elite lines has been progressing steadily. However, with most crops, there are numerous different elite lines adapted to a wide range of habitats, and hence this approach would require numerous independent transformation events. The issue of regulatory approval would also magnify, as there would be a large number of independent integration events that must be deregulated, even though they all contain the same transgene.

A third solution would be the in vivo site specific integration of additional transgenes to the previously engineered transgenic locus, whether into an elite genome or into a laboratory variety that is easier to transform. As before, the appended transgenic locus from “site specific gene stacking” must be introgressed to the numerous locally adapted cultivars. However, this method maintains the original number of segregating loci to ease the downstream breeding effort. This approach could also ease regulatory approval of the new DNA as it resides in a previously deregulated genome location. Moreover, the previously placed transgenes would not be considered a new transgenic event that needs a fresh round of deregulation. In principle, site specific integration of additional DNA to a previously engineered transgenic locus could be accomplished using either plant endogenous homologous recombination or site specific recombination.

As the breakage of double stranded DNA enhances the normally low homologous recombination efficiency, sequence specific nucleases (SSN) have been developed to break double stranded DNA at specified sequences. These SSN include zinc-finger nuclease (ZFN),⁵ transcription activator-like effector nuclease (TALEN),⁶ homing endonuclease/meganuclease,⁷ clustered regulatory interspaced short palindromic repeats (CRISPR)/CRISPR associated protein (Cas)⁸ and the recently described argonaute protein from Natronobacterium gregoryi (NgAgo).⁹ These SSN appear highly efficient for generating mutations through a non-homologous end-joining reaction, with the exception of the NgAgo system, in which there are currently some issues concerning its reliability.¹⁰ In all likelihood, the next generation of bioengineered crops would comprise not only of those obtained by transgenesis, but also by SSN-directed gene editing.

For inserting a transgene, the use of SSN requires that a double-stranded break be repaired from a homologous template of the new DNA. This repair pathway is not as efficient as the non-homologous end-joining pathway, but success has been reported for ZFN and meganuclease-mediated integration of a new DNA fragment into a preexisting transgenic locus in major crop plants.^11,12 This demonstrates their potential use for repeated gene integration at the same genomic location.

Whether SSN is efficient for transgene insertion is not the most critical concern with commercial developers, as extra time and labor can compensate for operational inefficiency. The major concern is the issue of whether there is freedom to operate (FTO). Each of the SSN is patented and obtaining a license may be prohibitively expensive such that only major crop biotech companies can afford their use. For the commercial development of major crops, ZFN has been exclusively licensed since 2008 to Dow Agrosciences through Sangamo Biosciences,¹³ and recently Monsanto and Dow Agroscicences announced a non-exclusive global option and licensing agreement on using ZFN for crops.¹⁴ Through Vilnius University since June, 2015, Dupont has obtained exclusive licensing for all commercial use of CRISPR/Cas9,¹⁵ as well as announced a strategic alliance with Caribou Biosciences on CRISPR/Cas9.¹⁶ Likewise, Monsanto has received a license for use of CRISPR/Cas9 in agricultural applications from the MIT/Harvard Broad Institute.¹⁷ However, there is ongoing litigation on the CRISPR/Cas9 technology that could take years to sort out.¹⁸ The licensing of TALEN is less clear, although Calyxt/Cellectis Plant Sciences (Minnesota) has obtained rights to practice the invention.¹⁹ Whether NgAgo would work in crop plants is currently not known given its reproducibility problem in mammalian cells,¹⁰ as well as whether it will be deemed sufficiently novel and unobvious to be issued a patent. Thus, until patents expire, the accessibility of developing countries to use these SSN for the commercial development of bioengineered crop does not look promising. Yet it is in these developing countries where food security is most at risk.

In contrast to the use of SSN, the use of certain site-specific recombinases has FTO. Hou et al.²⁰ have described a recombinase-mediated gene stacking (RMGS) system through demonstration of 2 rounds of gene stacking in tobacco, as well as an important step of removing unneeded DNA from the gene stack (Fig. 1A). The Mycobacteriophage Bxb1 integrase was used to integrate new DNA into the plant genome at specific attachment sites previously placed in the genome, either attB or attP. In the absence of an excisionase, the integrase-catalyzed recombination between attB and attP is not reversible. The Coliphage P1 Cre recombinase that performs reversible site-specific recombination on lox recombination sites was used subsequently to remove unneeded DNA including selectable marker genes after site-specific integration. By alternating the use of attB- or attP-containing donor molecules, additional DNA can be precisely appended to the locus. Therefore, the Bxb1/Cre mediated gene stacking system provides the solution of not increasing the number of segregating loci when future crop plants are improved through the addition of more transgenes. As importantly, Cre mediated excision of unneeded DNA addresses biosafety issues, whether real or perceived, especially regarding selectable marker genes that encode antibiotic resistance.²¹

Figure 1.

Compatible in vivo and in vitro recombinase mediated gene stacking. (A) in vivo gene stacking: donor plasmid (1) integrates into genomic target (2) via Bxb1 mediated attB x attP reaction generating genome structure shown in (3). Followed by excision of unneeded DNA via Cre-mediated recombination between directly placed lox sites to generate genome structure as (4). Second donor construct (5) integrates into genome target via genome attB x plasmid attP recombination generating genome configuration in (6) and after another round of Cre-mediated recombination yields the structure shown in (7). Note that each donor construct contains 2 attB or attP which can recombine with genomic target to generate 2 different configurations (only desired configuration shown). Cre can also invert the DNA between oppositely oriented lox sites such as shown in (2), (4) and in (7). Directly placed RS2 sites can optionally excise transgenic DNA through CinH-mediated recombination, leaving only T-DNA borders (if they were used) and one RS2 site. (B) in vivo gene stacking: phiC31 integrase promotes in vitro attP x attB recombination between molecules shown as (8) and (9) to yield molecule shown as (10). Cre recombinase deletes the unneeded DNA between directly placed lox sites to produce molecule shown as (11). Third molecule shown as (12) is then added by phiC31-mediated recombination to produce molecule shown in (13), followed by Cre-mediated excision of unneeded DNA to produce molecule shown in (14). At any point, in vitro recombined molecules can be transferred to an Agrobacterium vector for plant transformation, such as shown by the cointegration between (14) and (15) to generate a plant transfer vector shown in (16). After gene transfer and another round of in vivo Cre-mediated recombination, the expected structure is shown in (17). Note that the single attP site in (17) can serve as a target site for further gene stacking in vivo by Bxb1-mediated integration. Alternatively, the in vitro generated molecule shown in (14) can be directly inserted in vivo into a genomic target site such as shown in (7) through Bxb1-mediated integration. After a subsequent round of Cre-mediated resolution, the expected structure is shown in (18). The compatibility between the in vivo and in vitro system gives flexibility in stacking single or multiple genes.

In the article by Hou et al.²⁰ an additional set of RS2 recombination sites was included to flank the transgenes, which could permit future deletion of the RS2-flanked transgenes by the Acetinetobacter-derived CinH recombinase. For example, this might be useful for the root-specific deletion of transgenes in certain crops, such as for creating transgene-free carrots derived from transgenic plants.

Recently, we also developed a recombinase-mediated in vitro gene stacking system²² where the Streptomyces phage phiC31 (ΦC31) integrase was used to co-integrate separate circular DNA molecules, followed by using Cre to delete unnecessary plasmid backbone (Fig. 1B). The in vitro derived gene stack can be incorporated into an Agrobacterium transformation vector by Bxb1 recombination in vitro, permitting the transfer of the transgene package to a single chromosome location, and in the process creates a new target site for the subsequent in vivo stacking of new DNA. Alternatively, the in vitro-derived gene stack can be integrated directly into the plant genome in vivo at a pre-existing chromosomal target. Being able to stack DNA in vitro as well as in vivo, and with compatibility between the 2 systems, brings new flexibility for using the RMGS system.

Most important of all, the RMGS system was designed with FTO.²³ The Dupont Cre-lox patent had expired²⁴ and the patent applications on the Bxb1-att, and the CinH-RS2 systems, previously filed by the USDA had been abandoned.²⁵ Although the particular method of DNA delivery described by Hou et al. was protoplast transformation, which has low applicability for major crop plants, recent work by Li et al.²⁶ have described a detailed protocol using biolistics to deliver the DNA into rice embryogenetic callus. Regenerated plants with precise site-specific integration events have been obtained and this biolistic-mediated site-specific integration method should be applicable to a wide variety of crop plants. As biolistics is a technology described since 1987 patent claims have long expired.²⁷ Like the in vivo gene stacking system, the in vitro gene stacking system also has FTO. Although patents have been granted for the in vivo use of the phiC31 recombination system,^28,29 the in vitro use of this system is not patented.

Recombinases, though proficient in site-specific recombination, cannot edit genes the same way as SSN. However, if obtaining a commercial license for SSN is not possible or affordable, there are other options. Though much less efficient than SSN, forward mutagenesis can generate most of the knockout or knock-down mutations, and these mutations can be screened by a gel electrophoresis method known as TILLING (targeting induced local lesions in genomes).³⁰ Transgenes can also be engineered to suppress or enhance native gene expression. Thus, despite recent company mergers that further limit the control of food production to even fewer multinational agbiotech corporations, developing countries and small biotech developers can still practice crop improvement through transgenesis and forward genetic approaches, and this open source RMGS system provides a tool for developing countries to solve their food security challenge.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Funding

Supported by Chinese Ministry of Agriculture Grant 2010ZX08010-001 and Guangdong Province, China Talent Funds 2010.