Dicamba Resistance: Enlarging and Preserving Biotechnology-Based Weed Management Strategies

See allHide authors and affiliations

Science  25 May 2007:
Vol. 316, Issue 5828, pp. 1185-1188
DOI: 10.1126/science.1141596


The advent of biotechnology-derived, herbicide-resistant crops has revolutionized farming practices in many countries. Facile, highly effective, environmentally sound, and profitable weed control methods have been rapidly adopted by crop producers who value the benefits associated with biotechnology-derived weed management traits. But a rapid rise in the populations of several troublesome weeds that are tolerant or resistant to herbicides currently used in conjunction with herbicide-resistant crops may signify that the useful lifetime of these economically important weed management traits will be cut short. We describe the development of soybean and other broadleaf plant species resistant to dicamba, a widely used, inexpensive, and environmentally safe herbicide. The dicamba resistance technology will augment current herbicide resistance technologies and extend their effective lifetime. Attributes of both nuclear- and chloroplast-encoded dicamba resistance genes that affect the potency and expected durability of the herbicide resistance trait are examined.

In the past decade, the availability of biotechnology-derived herbicide-resistant and insect-resistant traits has led to striking advancements in agricultural crop management systems throughout the world. These “input traits” have contributed to greater productivity per hectare, decreased production costs, greater flexibility and efficiencies in production regimes, reduced pesticide use, and improved farmer health (13). In 2006, more than 100 million hectares worldwide were planted with crops having biotechnology-derived traits (4). In the United States, for example, Roundup (glyphosate)–resistant crops were planted on almost 90% of the soybean acreage and 60% of the cotton acreage in 2005, along with about 18% of the corn crop (3). The recent emergence of weeds resistant to the herbicides used year after year for weed control in fields of herbicide-resistant crops has prompted serious concerns regarding the long-term availability of the facile and economically important weed control provided by current herbicide-resistant crop plants. Also at risk is the greatly expanded use of no-till or reduced-till planting procedures that are made possible by “burndown” of weeds before planting of herbicide-resistant crops. These integrated practices minimize soil loss due to water and wind erosion resulting from traditional methods of soil tillage (3).

Among the glyphosate-tolerant weed species currently posing the greatest danger to agricultural productivity are several broadleaf plants such as giant ragweed (Ambrosia trifida), horseweed (Conyza canadenis), waterhemp (Amaranthus rudis), Palmer amaranth (Amaranthus palmeri), and common ragweed (Ambrosia artemisifolia) (3, 5). To combat these pernicious weeds and to address the potential emergence of other herbicide-resistant broadleaf weeds, we have targeted the development of crop plants resistant to treatment with dicamba. Dicamba is a widely used, low-cost, environmentally friendly herbicide that does not persist in soils and shows little or no toxicity to wildlife and humans (610). Use of the dicamba resistance trait alone or in combination with other herbicide resistance traits will allow rotation of herbicides or use of mixtures of herbicides that will greatly suppress several present or future herbicide-resistant weeds. Here, we describe the use of a genetically engineered bacterial gene, DMO (dicamba monooxygenase), that encodes a Rieske nonheme monooxygenase capable of inactivating dicamba when expressed from either the nuclear genome or chloroplast genome of transgenic plants. The DMO enzyme acts to destroy the herbicidal activity of dicamba before the herbicide can build to toxic levels in dicamba-treated transgenic plants, as shown below.

As the first step in the complete mineralization of dicamba, the soil bacterium Pseudomonas maltophilia (strain DI-6) converts dicamba to 3,6-dichlorosalicylic acid (DCSA) (11, 12) (Fig. 1A), a compound that lacks appreciable herbicidal activity. The enzyme system responsible for this conversion in the bacterium is the three-component enzyme dicamba O-demethylase. This enzyme system serves as an electron transfer chain in which electrons from NADH (the reduced form of nicotinamide adenine dinucleotide) are shuttled through a reductase to a ferredoxin and finally to the terminal component DMO (1315). The ferredoxin component of dicamba O-demethylase closely resembles the ferredoxin found in plant chloroplasts. Thus, to potentially take advantage of a source of reduced ferredoxin in chloroplasts of transgenic plants to supply electrons for the DMO reaction (and to eliminate the need for the bacterial reductase and ferredoxin genes), we included a chloroplast transit peptide–coding region upstream of the DMO gene to allow targeting of DMO to the chloroplast. The DMO expression cassette (Fig. 1B) contained the strong peanut chlorotic streak virus gene promoter FLt36 (16) and a terminator region from the pea Rubisco (ribulose-1,5-bisphosphate carboxylase-oxygenase) small subunit gene. The goal then was to determine whether expression of DMO from this expression vector in transgenic broadleaf plants could provide protection against the normally lethal effects of dicamba.

Fig. 1.

Dicamba inactivation. (A) Conversion of dicamba to DCSA by DMO. (B) Genetically engineered version of the DMO gene for expression in higher plants, using the FLt36 promoter from peanut chlorotic streak virus, a translational enhancer from the tobacco etch virus (TEV), a chloroplast transit peptide–coding region from the pea Rubisco small subunit gene for chloroplast localization of DMO, and a terminator region from the pea Rubisco small subunit gene (rbcS3′).

Because of ease of transformation and regeneration, Arabidopsis thaliana, tomato, and tobacco were used as model systems to test whether expression of the DMO gene alone (i.e., without the ferredoxin and reductase components of dicamba O-demethylase) could impart herbicide resistance after application of dicamba. Agrobacterium-mediated gene transfer was used to introduce the DMO expression cassette into the nuclear genome of the respective plant species. In regard to tobacco, we used DNA, RNA, and protein blot analyses to test several independently derived T1-generation plants for the presence and expression of the DMO gene (fig. S2). RNA blots demonstrated highly variable levels of DMO mRNA in individual transformants that, in general, did not correlate closely with the amount of DMO enzyme produced. We noted (most easily in lanes 2 and 6, fig. S2) that although most of the precursor DMO molecule containing the chloroplast transit peptide was cleaved to the mature form, not all of the precursor was processed.

Most dicotyledonous plants, such as tobacco, are quite sensitive to treatment with dicamba, an auxin-type herbicide. Figure 2A illustrates this point by showing nontransgenic tobacco plants not treated (leftmost plant) and treated with increasing amounts of dicamba. Herbicide damage symptoms are pronounced after spraying dicamba even at the low level of 0.017 kg/ha. Symptoms are quite severe at 0.28 kg/ha and 0.56 kg/ha, the levels normally used for weed control in agricultural applications.

Fig. 2.

Effects of dicamba treatment on nontransgenic tobacco plants and plants transformed with a genetically engineered DMO gene. (A) Demonstration of the sensitivity of nontransgenic tobacco plants to treatment with increasing doses of dicamba (left to right: 0, 0.017, 0.034, 0.07, 0.14, 0.28, and 0.56 kg/ha). (B) Three independently derived T1-generation tobacco plants carrying the dicamba resistance gene (three plants at left) and a nontransgenic plant (right) treated with dicamba at a level of 5.6 kg/ha. (C) Top view of plants in (B).

Treatment of transgenic tobacco plants containing the DMO gene with 5.6 kg/ha (10 to 20 times the recommended application rate) caused few if any symptoms, whereas a nontransgenic plant suffered severe damage (Fig. 2B). Damage to the lower leaves of the transgenic plants could be duplicated by spraying plants with the surfactant-containing solvent solution used as the vehicle for dicamba application. Leaves produced after treatment of the transgenic plants with dicamba exhibited no visible signs of damage (Fig. 2C). Transgenic tomato plants carrying the genetically engineered DMO gene, likewise, showed no damage to newly emerged leaves (fig. S2B) after spraying with dicamba at concentrations as high as 5.6 kg/ha. Arabidopsis expressing the DMO gene also displayed strong resistance to treatment with dicamba at 1.12 kg/ha, the highest level tested (fig. S3). Over a range of dicamba concentrations tested, an unexpected finding was the observation that tobacco plants transformed with a DMO expression cassette lacking a transit peptide–coding region were resistant to treatments with dicamba at levels on average only slightly below that of plants containing DMO genes bearing transit peptide–coding regions (fig. S4; see below).

To determine whether DMO could function exclusively inside chloroplasts, we created the pDMO1 vector bearing the DMO gene coding region (fig. S5). This vector allows integration of the DMO gene into the chloroplast genome of tobacco by homologous recombination and the isolation of transformants through selection for antibiotic resistance. The DMO gene coding region was driven by the strong psbA chloroplast gene promoter, containing the complete psbA 5′-untranslated region sequence, to obtain high levels of DMO expression. Initial DNA blot analyses of antibiotic-resistant transgenic plants (fig. S6A) demonstrated the presence in chloroplast genomes of both the DMO transgene (5.6-kb band) and the native psbA gene region (3.3-kb band). Repeated regeneration and selection of transgenic plants on antibiotic-containing medium resulted in apparently homoplastidic chloroplasts bearing the DMO gene fragment but not the endogenous native gene region (fig. S6B). Only chloroplast transformants expressing the DMO enzyme were resistant to treatment with dicamba (fig. S7). T1, T2, and T3 generations of progeny from two independently derived chloroplast transformants were tested for resistance to treatment with dicamba at various doses. All exhibited high levels of resistance. Indeed, chloroplast genome transformants displayed no apparent damage (other than “solvent-only damage” to lower leaves) when sprayed with dicamba at a rate of 28 kg/ha (fig. S8). Only transitory damage was observed when plants were treated with extremely high dicamba applications of 112 and 224 kg/ha. At these extremely high levels, initial damage was caused primarily by surfactants and other components of the solvent in which dicamba was delivered. New apex tissues and leaves growing from the damaged plants displayed nearly normal to normal phenotypes, showed no decrease in growth rates, and retained the ability to produce usual numbers and quality of seeds.

The above results were consistent with the hypothesis that reduced ferredoxin in tobacco chloroplasts could be the donor to DMO of electrons needed for oxidation of dicamba to DCSA. As a direct test of this hypothesis, we examined the ability of purified spinach ferredoxin to support the conversion of dicamba to DCSA in the presence and absence of DMO purified from P. maltophilia (strain DI-6) or overproduced and purified from Escherichia coli (table S1). Results of these experiments demonstrated that reduced ferredoxin from spinach or Clostridium pasteurianum was fully capable of donating electrons to DMO in vitro, as measured either by dicamba degradation or by DCSA appearance.

The exceptionally high levels of resistance to dicamba displayed by tobacco plants carrying the DMO gene in the chloroplast genome, relative to plants bearing the DMO gene as a nuclear gene, suggested the possibility that chloroplast-encoded DMO was produced in greater abundance. Comparison of the amounts of oxygenase as percentage of total soluble protein, fraction of fresh weight, or fraction of dry weight (table S2) showed that chloroplast transformants produced about 20 times as much DMO as did nuclear transformants synthesizing DMO with a chloroplast transit peptide, and about 40 times as much DMO as did nuclear transformants synthesizing DMO without the peptide. The ability to achieve high levels of herbicide resistance and the ability to block gene dissemination through “pollen flow” are attractive features of incorporating the DMO gene into the chloroplast genomes of important crop plants as soon as the techniques for such approaches prove practical (17, 18).

Genetic studies of the inheritance of the DMO gene in chloroplast transformants revealed that inheritance was maternal, as expected, and was mostly Mendelian in the case of plants carrying DMO as a nuclear gene (table S3). Most plants examined by DNA blot analysis contained a single DMO gene insert. Moreover, T3 and T4 progeny maintained the original levels of expression in regard to herbicide resistance whether they contained single or multiple copies of the DMO gene.

The prime value of the dicamba resistance technology is related to its use in major field crops in which management of broadleaf weeds is essential to maximize production. Because soybean is one such crop, we transformed the soybean varieties Thorne (Ohio State University) and NE3001 (University of Nebraska) with the same DMO expression cassette (Fig. 1B) used to transform tobacco, tomato, and Arabidopsis. As a means to derive marker-free soybean transformants, a two–T-DNA binary plasmid was assembled. In this plasmid, the marker-gene T-DNA element carried a bar gene cassette under the control of the Agrobacterium tumefaciens nopaline synthase promoter (nos), and the second, separate, T-DNA element carried the DMO expression cassette. More than 50 transgenic soybean events were produced, and seeds from the T1, T2, and T3 generations were collected. Among the population of primary transformants generated, one marker-free event was identified that harbored only the DMO cassette. Most transgenic soybean events showed resistance to treatment with dicamba at 2.8 kg/ha and 5.6 kg/ha under greenhouse conditions (fig. S9) and complete resistance to dicamba at 2.8 kg/ha (the highest level tested in field trials) (Fig. 3). Initial field studies with five independent soybean events on University of Nebraska farms over the past 3 years revealed no compromise in agronomic performance—including yield, date to flowering, height, and lodging—in the transgenic plots treated with dicamba application (1.5 kg/ha) at preplant, V3 stage, or dual preplant spray treatment coupled with post-emergence treatments at the V3 stage of plant development when compared with non–herbicide-treated, weed-free plots of the parental soybean variety Thorne.

Fig. 3.

Effects of dicamba treatments on nontransgenic soybean plants and transgenic plants containing the genetically engineered DMO gene. Nontransgenic (1) and transgenic soybean plants (2) treated under field conditions with dicamba at a level of 2.8 kg/ha are pictured 8 days after spraying.

Dicamba resistance in all of the plants tested did not require cotransformation with either ferredoxin or reductase genes from P. maltophilia (strain DI-6). These results showed that the plants contained one or more molecules that could transfer the requisite electrons to DMO to allow conversion of dicamba to DCSA. The initial targeting of DMO to the chloroplasts by means of a transit peptide sequence was aimed at using reduced ferredoxin abundantly available in the chloroplasts. However, transformation of tobacco plants with a DMO gene construct lacking a chloroplast transit peptide–coding sequence unexpectedly resulted in plants that were highly resistant to treatment with dicamba. Results from our limited trials with a small number of T1-generation plants indicated that the level of resistance obtained with these transgenic plants was only slightly lower on average than that obtained with tobacco plants producing DMO containing a transit peptide.

These observations raise important questions in regard to the molecules in transgenic plants that can productively donate electrons to DMO. The fact that homoplastidic chloroplasts producing DMO internally from a DMO gene integrated into the chloroplast genome show resistance to extremely high levels of dicamba (fig. S8) and the fact that purified DMO can function in vitro with reduced spinach chloroplast ferredoxin (table S2) both suggest that chloroplast ferredoxin can productively interact with DMO to allow electron transfer. However, the source of electrons for DMO produced from nuclear genes lacking a chloroplast transit peptide–coding sequence remains unknown. Presuming that ferredoxins do not reside outside of the plant chloroplasts, one must consider the possibility that an unknown cytoplasmic protein can provide DMO with a steady supply of electrons. Alternatively, DMO itself might contain a gratuitous chloroplast transit peptide that allows sufficient DMO to enter the chloroplasts to provide protection from dicamba moving into the cell after dicamba treatment. Further studies, such as microscopic localizations in situ of DMO with and without a chloroplast transit peptide and/or isolation and identification of cytoplasmic proteins that can interact “indiscriminately” with DMO to supply electrons, will be needed to resolve the questions emanating from the present observations.

It is illuminating to consider that dicot plants like tobacco display distinct injury symptoms even at levels of dicamba treatment as low as 0.001 to 0.01 kg/ha (Fig. 2A). Many transgenic tobacco, tomato, Arabidopsis, and soybean plants containing a nuclear-encoded DMO gene were fully resistant to treatments with dicamba at or above 5.8 kg/ha. This demonstrates that the DMO gene, present even in a single copy and expressed at relatively moderate rates (table S2), is capable of decreasing the sensitivity of dicot plants to applications of dicamba by at least a factor of 5000.

Dicamba is an “auxin”-type herbicide that mimics the effects of excess quantities of the natural plant hormone indole-3-acetic acid (IAA) when applied to dicotyledonous plants. It has been used for more than 40 years to efficiently control most broadleaf weeds. Yet despite its widespread use, no new noxious and economically important dicamba-resistant weed species have appeared (5). One possible reason for such a situation may be that dicamba may act on some, if not all, of the IAA receptors that are essential in controlling normal growth and development of all plants. If so, the appearance of new dicamba-resistant weeds may not happen rapidly. This is especially true if the dicamba resistance gene is “stacked,” for example, with the widely used glyphosate resistance gene to allow farmers to alternate herbicide applications between dicamba and glyphosate or to use mixtures of the two herbicides together. In either case, appearance of weeds resistant to either dicamba or glyphosate will be greatly suppressed. Moreover, the ability to use either or both herbicides before planting or at a variety of points during crop development will allow producers excellent weed control with greater flexibility in their crop management practices. This may be particularly important in the control of existing glyphosate-resistant weeds, such as horseweed, in which application of dicamba before planting can control emerged or emerging glyphosate-resistant weeds. Thus, dicamba-resistant crops can be a valuable asset in strategies to control currently existing herbicide-resistant weeds and to suppress the appearance of additional herbicide-resistant weeds that ultimately could threaten the long-term use and value of current herbicides and herbicide-resistant crops. Likewise, dicamba-resistant crops should further encourage the use of conservation tillage practices that greatly decrease soil erosion and foster more sustainable and environmentally friendly farming.

Supporting Online Material

Materials and Methods

Figs. S1 to S9

Tables S1 to S3

References and Notes

View Abstract

Stay Connected to Science

Editor's Blog

Navigate This Article