Development and commercial use of Bollgard® cotton in the USA – early promises versus today's reality


*For correspondence (fax +1 636 737 6662; e-mail


Bollgard® cotton is the trademark given to a number of varieties of cotton bio-engineered to produce an insecticidal protein from Bacillus thuringiensis (Bt). When produced by the modified cotton plants, this protein controls certain lepidopterous cotton insect pests. Commercially available since 1996, these cotton varieties are purchased under a license agreement in which the growers pay a fee and agree to abide by the terms, which include a 1-year license to use the technology and agreement to participate in an insect resistance management program. Today, Bollgard® cotton is grown on more than one-third of all cotton acreage in the USA. This product has reduced cotton production costs and insecticide use by providing an effective alternative to chemical insecticides for the control of tobacco budworm, Heliothis virescens; cotton bollworm, Helicoverpa zea; and pink bollworm, Pectinophora gossypiella. The specificity and safety profile of the Bt protein produced in planta in cotton was maintained. It has retained its selectivity for lepidopterous insects and lacks the characteristics found in potential allergenic proteins. Fiber quality, the agronomic characteristics of the plant and seed composition remain unchanged. New cotton technology is being developed to provide improved insect control and a wider spectrum of activity. These future products could further reduce insecticide use in the production of cotton, while maintaining the high level of safety and reliability that has been demonstrated by five seasons of Bollgard® cotton use.


The promise of genetic engineering, a technology tool that emerged from molecular biology laboratories in the 1970s, fueled optimism, excitement and investment in the 1980s. It was seen as a tool to address the mysteries of medicine and hypothesized as a way to ‘engineer’ reduced reliance on fossil fuels. It was a way to address issues of food production such as making foods healthier and more nutritious, reducing the effects of environmental stress on crop yield, and decreasing ‘on farm’ reliance on pesticides. After this initial surge of possibilities, the difficulty of bringing novel, genetically engineered products to the market place took hold in the 1990s. Few concepts survived the harsh realities of practicality, profitability and utility, at least in the short term. The few agricultural concepts that evolved to commercial products in this decade include herbicide tolerance in soya, canola, corn and cotton; virus resistance in potatoes and squash; and resistance to specific insect pests in corn, potatoes and cotton.

Bollgard® cotton, bio-engineered to resist lepidopterous insect pests, has had a dramatic effect on cotton production. The primary product concept was straightforward. Bacillus thuringiensis (Bt), a microorganism found in soil, was known to produce a potent, specific and safe protein insecticide (Baum et al., 1999; Dulmage, 1981). This protein was well characterized, considered safe, and specific to the lepidopterous class of insects. By introducing the genetic information encoding the insecticidal protein of Bt into plants, these plants would produce their own insecticide. This simple concept promised to reduce insecticide applications, improve insect control, and provide opportunities for aggressive integrated pest management systems while maintaining the inherent safety of this insecticidal protein. The development of Bollgard® cotton required overcoming several significant technical hurdles unforeseen when the product concept was originally proposed in the early 1980s. Among these were concerns with the levels of in planta Bt gene expression (Fischhoff et al., 1987); a resistance management program (Fischhoff, 1992); suitability of this technology for cotton production systems (Jenkins et al., 1993; Wilson et al., 1992); and environmental risk–benefit studies, confirmation of safety on non-target insects, food and feed safety studies and persistence and fate studies of the Bt protein (Betz et al., 2000). Analysis was also required to confirm that there were no substantial differences in the properties of the cotton bio-engineered for insect resistance. The growth characteristics of the transformed cotton plants, the properties of their seeds (oil, protein, carbohydrate; Fuchs et al., 1993), and the germination of the seeds in the field were closely studied. Quality characteristics such as length, strength and micronaire or thickness and weavability of the fiber were studied to determine if the expression of Bt in planta, directly or indirectly, affected the quality of the inherent characteristics of the cotton varieties offered commercially (Etheridge and Hequet, 2000; Jenkins et al., 1997; Jones et al., 1996).

Having passed these technical challenges, Bollgard® cotton was successfully commercialized in 1996 and planted on 1.8 million US cotton acres (12% of US cotton acreage) in its first year of commercial availability. Adoption of Bollgard® cotton in the USA has grown each year to greater than one-third of the US acreage planted to cotton in 2000 (over 4 million acres). Most of the early promises of this technology have been realized. The use of Bollgard® cotton does reduce chemical insecticide use (Carpenter, 2001). Bollgard® has been responsible for improved cotton yields when compared to conventional cotton sprayed with insecticides. It is a safer, higher yielding, economic alternative to chemical insecticides for the control of tobacco budworm (Heliothis virescens), pink bollworm (Pectinophora gossypiella) and moderate infestations of cotton bollworm (Helicoverpa zea). An aggressive insect resistance management program, envisioned in 1991 (Fischhoff, 1992) and designed to delay the potential development of insect resistance to Bt, has been implemented. The characteristics of cotton have not been altered as a result of the addition of the Bt gene producing Bt protein. Fiber quality, plant agronomics and seed composition remain unchanged. The specificity and safety profile of the Bt protein produced in planta in cotton was maintained, and after 5 years of commercial use the safety and efficacy of Bollgard® cotton is appreciated in the market place.

The movement of the original product concept to commercialization was possible due to the extensive involvement of the cotton research community, which included universities, government agencies, growers and corporate participation. Improvements in this product are possible as new proteins, promoters and expression systems have become available. These new products in development with the cotton research community are intended to retain the safety and convenience of Bollgard® cotton while increasing the spectrum of insect activity and utility of bio-engineered cotton varieties available for cotton growers.

Cloning and expression of Bt genes in E. coli

The potential of genetic engineering to improve agriculture was recognized in the early 1980s. Both established agricultural companies and emerging technology companies started programs to identify ways of applying the new developments in genetic engineering to persistent agricultural problems. The vision of introducing a gene from Bt into plants to provide protection from insect injury was a common objective. The first report of cloning and expression of a Bt gene in Escherichia coli by Helen Whiteley's laboratory (Schnepf and Whiteley, 1981) indicated that, although the Bt protein appeared to be expressed in Bt under the same gene-expression control as sporulation, expression of a cry1Aa gene in E. coli was achieved. The protein retained insecticidal activity when expressed in E. coli, and it was subsequently reported that limited ‘truncation’ of the gene (Schnepf and Whiteley, 1985) resulted in a truncated protein that retained insecticidal activity. Numerous other Bt genes have been identified, and the rich diversity of Bt genes and their proteins is now appreciated (Aronson et al., 1986; Crickmore et al., 1998; Hofte and Whiteley, 1989).

Expression of native Bt insecticidal protein genes in plants

The expression of Bt genes for effective insect control in plants was not as simple as originally anticipated. Although Bt genes could be cloned, expressed in bacteria and used to construct chimeric genes that were inserted into plants, the expression levels of native Bt coding sequences in plants were disappointing (Fischhoff et al., 1987). Expression was much lower in plants than would have been predicted based on the limited experience with other bacterial genes such a neomycin phosphotransferase (Vaeck et al., 1987).

Early studies in the 1980s focused on Bt gene expression work in tobacco and tomato (Barton et al., 1987; Fischhoff et al., 1987; Vaeck et al., 1987). These species were found to be easily transformed, grew quickly, and could easily be grown in growth chambers or greenhouses in small pots. They were very susceptible to insect attack by several lepidopteran pests that were easy to rear (tobacco hornworm, Manduca sexta, and tobacco budworm) and very sensitive to Bt proteins (MacIntosh et al., 1990). Various chimeric genes comprising a plant promoter, a full-length Bt coding sequence and a 3′ untranslated region for polyadenylylation addition were reported to have no detectable expression in tobacco, and very low levels in tomato. Truncation of the Bt coding sequence by the elimination of the approximate 3′ half of the coding sequence improved expression to the point where protein and messenger RNA could be reliably detected. This work, along with other studies of this period, has been reviewed in several publications (Diehn et al., 1996; Estruch et al., 1997; Mazier et al., 1997).

Laboratory insect assays indicated that the insecticidal activities of the Bt genes expressed in plants were retained. In the first field tests of Bt plants in 1987, tomatoes transformed with a truncated version of the cry1Ab gene completely controlled tobacco hornworm, confirming prior laboratory tests (Perlak and Fischhoff, 1993). It was observed that there was a significant reduction in tomato damage due to tomato fruit worm (H. zea). Although the control of these two pests in a field situation was encouraging and confirmed by subsequent field tests (Delannay et al., 1989), Bt gene expression was judged to be too low to be consistently commercially successful. Expression of truncated Bt genes in cotton was reported to be very low, and a field test had indicated that there was no control of budworm in the field or when the plant samples from the field were tested in the laboratory (Jenkins et al., 1991).

Synthetic Bt genes and increased Bt gene expression in plants

The first published reports of in planta synthetic Bt gene expression were in cotton (Perlak et al., 1990). Levels of insect control protein (Bt protein) were increased dramatically from undetectable levels with the wild-type gene, to 0.05–0.1% of total soluble protein in plants with the synthetic gene. In a subsequent publication (Perlak et al., 1991) the essential modifications to Bt synthetic genes for increased plant expression were described in greater detail. The evidence indicated that potential polyadenylylation signal sequences were present in a relatively small segment of the sequence (30 bp) that seemed to be associated with lower levels of expression. Alteration of this sequence increased expression significantly (10-fold). Others confirmed this work (Van der Salm et al., 1994). A re-synthesized Bt gene was designed and constructed that lacked potential polyadenylylation signal sequences, areas of A–T richness and ATTTA sequences. Plant codon usage was also considered in the design of this synthetic gene. The plants showed significant increases in Bt protein levels and mRNA production, and it was concluded that several factors and regions of the gene were involved in the low levels of insect control protein observed with the native Bt genes in plants. Recent research has confirmed that premature polyadenylylation contributes to the poor expression of Bt genes in plants (De Rocher et al., 1998; Diehn et al., 1998).

The most dramatic result from the cotton plants described in an early publication (Perlak et al., 1990) was the increase in insect control. Greenhouse tests that simulated heavy field insect pressure showed remarkable protection (70–87%) under conditions in which most (88–100%) of the bolls and squares (cotton flowers) on control plants were destroyed by cotton bollworm. Field testing and screening would be necessary to identify a transformed cotton event suitable for commercialization, but these preliminary results were very encouraging.

Development of Bollgard® cotton

Selection of a transformed cotton line for commercialization

The report on insect-resistant cotton (Perlak et al., 1990) signaled the beginning of a detailed and extensive assay program to select a line that would be suitable for back-crossing and eventual commercialization. The initial efficacy data were collected from a limited number of events or individual plant lines, descended from a single plant after transformation and regeneration. A much larger program, with hundreds of transformed cotton plants, was initiated prior to commercialization.

Cotton was reported to have been regenerated (Trolinder and Goodin, 1987) and transformed (Firoozabady et al., 1987; Umbeck et al., 1987) by Agrobacterium-mediated transformation, although only at very low frequencies and essentially only in Coker cotton variety backgrounds. The process was very long (12–18 months from bacterial infection to plants) and very labor intensive (estimated at 60–80 man hours per plant). Because of the Coker background requirement for transformation, a back-crossing program with elite cotton germplasm was required to produce commercial varieties, further lengthening the time to commercialization. A large program of cotton transformation was established to select a cotton line that had good levels of Bt gene expression and desirable agronomic characteristics. The large number of initial transformed events permitted extensive testing of promoter–structural gene combinations. Bt gene expression work in the early 1990s (Wong et al., 1992) reported that high levels of expression of the Bt gene were possible in tobacco. However, repeated attempts to extend these results to cotton were unsuccessful. A large number of synthetic, full-length Bt coding sequence vectors, as well as truncated versions of the Bt coding sequence, were tested in greenhouse assays and in the field.

Pre-commercial field-testing program

The field-testing program was highly regulated by the US Department of Agriculture (USDA) and the Environmental Protection Agency (EPA). It involved key researchers from the USDA and universities in cotton-producing states. The cotton-growing regions in the USA represent a wide diversity of climates and insect pests. Localized testing for product suitability is an important aspect of the commercialization process.

The earliest field tests were done with plants expressing truncated Bt genes, and were primarily efficacy trials (Jenkins et al., 1997). These results indicated excellent control of tobacco budworm, and good control of cotton bollworm could be attained (Benedict et al., 1996; Rummel et al., 1994; Wilson et al., 1992). Key agronomic characteristics such as yield, lint quality, morphology, and other factors were assessed in field tests. The extent of testing expanded when the Bt gene was crossed back into elite cotton varieties. Further testing and experience after commercialization (Jenkins et al., 1997) indicated that protection of the cotton plants from damage by tobacco budworm was very good, but that some applications of insecticide were occasionally required under conditions of high insect infestation by cotton bollworm. These results were consistent with previous published reports (Mahaffey et al., 1994). It was also noted that some cotton lines exhibited a genotype-by-event interaction. Eventually, a single event, Monsanto 531, was chosen as the event from which all current Bollgard® cotton varieties have descended. This cotton line was transformed with a vector containing a full-length synthetic cry1Ac-like Bt coding sequence driven by an enhanced 35S promoter (Kay et al., 1987). This line exhibited consistent target-insect control in the greenhouse and the field, and very good agronomic characteristics in a wide assortment of cotton varieties that were back-crossed with the original transformed cotton line.

Commercial Bollgard® cotton varieties were initially available in the USA in 1996, and almost 1.8 million acres were planted that year. Bollgard® cotton acreage has expanded to more than one-third of the total US cotton acreage in 2000 (over 4 million acres). Because of the wide acceptance of Bollgard® cotton during the past 5 years, much has been learned about the efficacy, benefits for the grower, benefits for the environment, and the impact of Bollgard® cotton on agronomic practices.

As part of the commercial licensing agreement, farmers are required to participate in the resistance management plan. This program requires that growers have a percentage of their crop that does not contain the cry1Ac-like gene. This portion of the crop functions as a refuge, providing an arena for Cry1Ac-susceptible insects to develop. The farmers have two refuge choices: one that represents 5% of their crop area where no insecticides are sprayed for the target pests; or where 20% or more of the crop area is planted to non-Bollgard® cotton that can be sprayed with any insect management regime except microbial Bt sprays. Growers receive a detailed guide to the concept of using refuges in relation to the resistance management program prior to planting Bollgard® cottonseed, and compliance is mandatory.

Efficacy of Bollgard® cotton in controlling lepidopterous insects

Bollgard® cotton was envisioned as an alternative to chemical insecticides for controlling the key lepidopterous pests of cotton. Early studies in the greenhouse and in small field plots demonstrated that Bollgard® provided insect control equivalent to, or better than, control with insecticides used on conventional cotton. The performance of Bollgard® cotton in providing a commercial level of insect control in the field has been outstanding. Control of tobacco budworm and pink bollworm in fields of Bollgard® cotton has been very good across the entire cotton-growing area in a diverse number of cotton varieties. Control of cotton bollworm with Bollgard® has been less complete, requiring occasional supplemental insect control with chemical sprays. The protein produced in Bollgard® cotton, Cry1Ac, has very good activity against these three major cotton pests when compared to other Bt proteins (MacIntosh et al., 1990). The cotton line selected for commercialization, Monsanto 531, one of hundreds of transformed cotton plants screened, has provided consistent floral protection and good gene expression in a wide variety of commercial genetic backgrounds. The difficulty in providing complete protection against cotton bollworm was predicted from the earliest greenhouse studies (Perlak et al., 1990) and subsequent field work (Jenkins et al., 1997), especially under extremely high insect-pressure situations with cotton bollworm. Expression of Bt protein in the pollen of Bollgard® cotton is low, and pollen is a food source for developing bollworms (Greenplate et al., 1998). Bollworms surviving on Bollgard® cotton have been observed feeding in fresh blooms and on small bolls under bloom tags (Brickle et al., 2001).

Early studies (Perlak et al., 1990) indicated some control of beet armyworm (Spodoptera exigua) with Bt cotton. However, some of these plants produced Cry1Ab protein, a protein more effective on beet armyworm and less effective on cotton bollworm when compared to Cry1Ac. The current Bollgard® cotton product has little effective activity against beet armyworm and fall armyworm (Spodoptera frugiperda), two insects with little sensitivity to the Cry1Ac protein (MacIntosh et al., 1990).

Safety of Bollgard® cotton

There were several basic safety assumptions when the product concept for Bollgard® cotton was proposed. One was that the specificity and safety profile of the Bt protein produced in planta would be maintained. A second assumption was that the expression of the cry1Ac gene would not affect the agronomic characteristics of the plant, the characteristics of the seeds, or the quality of the fiber. To confirm these assumptions, numerous studies have been completed.

The Cry1Ac protein produced by Bacillus thuringiensis is known to be extremely selective against lepidopteran insects and harmless to humans, fish, wildlife and agriculturally beneficial insects (McClintock et al., 1995). The mode of action of this protein requires binding to mid-gut epithelium receptors in the insect (Hofmann et al., 1988a; Van Rie et al., 1990a; Van Rie et al., 1990b). Equivalent receptors have not been identified on intestinal cells of mammals (Hofmann et al., 1988b; Sacchi et al., 1986), explaining the distinct absence of toxicity of these proteins to non-target organisms. Studies confirmed the spectrum of insecticidal activity of the plant-produced protein to target and non-target insect species (Sims and Martin, 1997). In addition, high-dose feeding studies of the Cry1Ac protein, as well as extensive field observations, confirmed no adverse effects on non-target insects including adult and larval honeybees, ladybird beetles, collembola, parasitic wasps and green lacewings (Armstrong et al., 2000; Sims and Martin, 1997).

Proteins, as a class, are not generally toxic to humans and animals (Betz et al., 2000). The low mammalian toxicity of Bt microbial insecticides and the absence of toxicity to Bt insecticidal proteins have been demonstrated in extensive safety studies conducted over the past 40 years, a number of which have been published (Fisher and Rosner, 1959; Meeusen and Atallah, 1990; Siegel and Shadduck, 1990). The EPA has concluded: "The large volume of submitted toxicology data allows the conclusion that the tested subspecies are not toxic or pathogenic to mammals including humans" (EPA, 1998a; EPA, 1998b).

Confirmation of human and animal safety of the Cry1Ac protein produced in Bollgard® cotton plants included studies characterizing the biological and physicochemical equivalence of the plant-produced protein to that produced by Bt, as well as assessments of the potential mammalian toxicity and allergenic potential of the protein. The Cry1Ac protein produced in cotton has the expected molecular weight, insecticidal activity and immunoreactivity as that produced by Bt (Fuchs et al., 1993). Neither protein is glycosylated, as expected due the lack of targeting sequences for glycosylation. Further, bioinformatic analyses indicate that the Cry1Ac protein is not structurally similar to any known protein toxins or other proteins associated with adverse mammalian or human health effects. The potential toxicity of proteins can be determined in short-term laboratory studies with surrogate species (Sjoblad et al., 1992). No acute toxic effects were observed in mice based on body weights, food consumption and gross necropsy following gavage administration of high levels of the Cry1Ac protein.

Important considerations contributing to the allergenicity of proteins ingested orally include the assessment of factors that contribute to exposure, such as stability to digestion and prevalence in the food (Metcalfe et al., 1996). Important protein allergens tend to be stable to peptic digestion and the acidic conditions of the stomach if they are to reach the intestinal mucosa where an immune response can be initiated (Astwood et al., 1996). The Cry1Ac protein was confirmed to degrade rapidly via digestion in model systems, with a half-life of approximately 15 sec. Another significant factor contributing to the allergenicity of certain food proteins is their high concentrations in foods (Taylor and Lehrer, 1996). Most allergens are present as major protein components in the specific food. In contrast to this generality for common allergenic proteins, Cry1Ac protein is produced at very low levels in cotton tissues, ranging from 0.0005 to 0.0020% FW. It is also important to establish that the protein does not represent a previously described allergen. Bacillus thuringiensis and its formulations used as microbial pesticides have not been described as sensitizing allergens, including introduction through oral exposure (McClintock et al., 1995). The amino acid sequence of the Cry1Ac protein produced in cotton was compared to protein sequences associated with allergenicity found in publicly available genetic databases (GenBank, EMBL, PIR and SwissProt) and those documented in the current literature. Cry1Ac does not share a structurally significant sequence similar to sequences within the allergen database, and does not share potential immunologically relevant amino acid sequences greater than seven contiguous identical amino acids.

The environmental safety of the Cry1Ac protein as produced by Bt is also extensively known (EPA, 1998a; EPA, 1998b). In addition to the lack of effects on non-target organisms, the proteins in general, and Cry1Ac specifically, are known to degrade rapidly in the environment (Palm et al., 1993; Palm et al., 1994; Palm et al., 1996) and thus do not accumulate in the soil or leach into groundwater. Laboratory and field soil-degradation studies confirmed these environmental characteristics for the Cry1Ac protein produced in Bollgard® cotton (Betz et al., 2000). These studies addressed the issues considered principal for the environmental safety assessment by the EPA Scientific Advisory Panel, which included the amount of Cry protein that enters the soil over time, and the length of time that the protein persists in a biologically active form. Further, the bioengineering of cotton does not generate a heightened concern for the movement of introduced genes from Bollgard® into wild cotton relatives. Only two wild relatives of cotton occur in the USA: Gossypium thurbei in Arizona and Gossypium tomentosum in Hawaii. Gossypium thurbei is not sexually compatible with commercial varieties of Gossypium hirsutum. Gossypium tomentosum is sexually compatible with commercial varieties of G. hirsutum; however, cotton is not grown commercially in Hawaii (Betz et al., 2000).

The second assumption made in the product concept phase for insect-protected cotton was that the genetic enhancement would not change the cotton plants' fundamental characteristics. This basic principle has been termed ‘substantial equivalence’ and was established in the early 1990s by the Food and Agriculture Organization of the United Nations, the World Health Organization and the Organization for Economic Cooperation and Development. Agronomic equivalence was demonstrated in extensive observations of plants in greenhouse and field testing over several years, and included analysis of more than 40 components of plant morphology, maturity, development, yield, fiber characteristics, and disease and insect susceptibility. In addition, more than 2500 separate analyses were performed on 67 components of the cottonseed and oil derived from Bollgard® cotton plants, including nutrients, amino acids, fatty acids, vitamins and minerals, as well as natural anti-nutritional components such as gossypol. These components were confirmed to be substantially equivalent to comparable conventional cotton varieties (Berberich et al., 1996). The conclusions regarding the lack of toxic and allergenic properties of the Cry1Ac protein produced in cotton and the nutritional equivalence of the cottonseed were additionally confirmed in a number of animal feeding studies conducted with rats, quail, catfish, goats and cows.

Cotton fiber from commercial Bollgard® varieties has also been extensively studied and demonstrated to be unaffected by the presence of the inserted genetic material. Fiber quality data from commercially grown, ginned and classed cotton bales were collected in 1999 and 2000 from across the cotton belt. Average staple length and micronaire from thousands of bales of cotton produced from conventional varieties, bales of Bollgard® and bales produced from Bollgard® with Roundup Ready® cotton varieties were compared by high-volume instrumentation. The quality of the fiber produced from Bollgard® cotton varieties was not different from the quality of fiber from conventional varieties (Figures 1 and 2). Several other reports confirm the results of these studies (Etheridge and Hequet, 2000; Kerby et al., 2000a; Kerby et al., 2000b; Moser et al., 2000; Speed and Ferreira, 1998).

Figure 1.

Figure 1.

Micronaire ratings.

(a) Micronaire ratings of 19 031 cotton bales. Data collected in 1999 from AR, AZ, LA, MS, NC, TN, TX, VA. Dependent variable = technology, type III F = 662.22, P > 0.0001.

(b) Micronaire ratings of 76 992 cotton bales. Data collected in 2000 from AR, AZ, CA, MS, NC, TN. Dependent variable = technology, type III F = 536.81, P > 0.0001.

Figure 1.

Figure 1.

Micronaire ratings.

(a) Micronaire ratings of 19 031 cotton bales. Data collected in 1999 from AR, AZ, LA, MS, NC, TN, TX, VA. Dependent variable = technology, type III F = 662.22, P > 0.0001.

(b) Micronaire ratings of 76 992 cotton bales. Data collected in 2000 from AR, AZ, CA, MS, NC, TN. Dependent variable = technology, type III F = 536.81, P > 0.0001.

Figure 2.

Figure 2.

Staple ratings.

(a) Staple ratings of 19 031 cotton bales. Data collected in 1999 from AR, AZ, LA, MS, NC, TN, TX, VA. Dependent variable = technology, type III F = 862.40, P >0.0001.

(b) Staple ratings of 76 814 cotton bales. Data collected in 2000 from AR, AZ, CA, MS, NC, TN. Dependent variable = technology, type III F = 1845.57, P >0.0001.

Figure 2.

Figure 2.

Staple ratings.

(a) Staple ratings of 19 031 cotton bales. Data collected in 1999 from AR, AZ, LA, MS, NC, TN, TX, VA. Dependent variable = technology, type III F = 862.40, P >0.0001.

(b) Staple ratings of 76 814 cotton bales. Data collected in 2000 from AR, AZ, CA, MS, NC, TN. Dependent variable = technology, type III F = 1845.57, P >0.0001.

The environmental and mammalian safety of the Cry1Ac protein was confirmed in planta, and the equivalence of the transformed cotton plant and its products to conventional cotton was demonstrated.

Benefits of Bollgard® cotton

Safety benefits

The budworm/bollworm complex is a major cotton pest, causing serious damage to floral structures of the cotton plant and requiring extensive treatment with chemical insecticides in conventional cotton. The effectiveness of Bollgard® cotton has resulted in a dramatic reduction in the use of chemical insecticides. Estimates of the reduction indicate that 2.2 fewer sprays are used on an annual basis for the control of bollworm and budworm across the entire cotton belt (Benedict and Altman, 2000). There are reduced trips across the field, reduced worker exposure, reduced chemical load on the environment, and reduced exposure of communities surrounding cotton-producing areas. It is estimated that almost 2 million pounds of insecticide active ingredient were not sprayed in 1998 due to the use of Bollgard® cotton (Benedict and Altman, 2000). Other studies (Gianessi and Carpenter, 1999; Gianessi and Carpenter, 2001) report very similar findings (2.7 million pounds per year in 2000). These economic benefits are reviewed elsewhere (Edge et al., 2001; Falck et al., 2000; Falck-Zepeda et al., 1999). The reduction of insecticide use is more dramatic in some areas than others. In Arizona, where Bollgard® cotton has had rapid and significant adoption in the last 5 years, there has been a steady and dramatic decrease in overall insecticide use in cotton (Agnew and Baker, 2001).

Economic benefits

Bollgard® cotton varieties are purchased under a license agreement in which the growers pay a fee to Monsanto averaging between US$20 and $32 per acre, depending on seeding rate, agronomic practices in a given region, and typical annual lepidopteran insect infestation levels. The growers agree to abide by terms that include a 1-year license to use the technology, and required participation in an insect resistance management program. Despite the initial expense for this technology, Bollgard® cotton has consistently provided value to growers, demonstrated by the adoption of Bollgard® on more than one-third of all cotton acreage in the USA. The value of Bollgard® cotton for a grower is variable, and depends on three criteria: how well the grower manages his crop for the environmental conditions of a particular year; the amount of insect pressure; and the price of cotton.

Bollgard® cotton is a tool that provides growers with an economical approach to manage tobacco budworm, pink bollworm and cotton bollworm, but it does not address all insect pest management issues. As cotton yield potential increases, due, in part, to varietal management and satisfactory environmental factors, the value of the protection provided by Bollgard® increases. Secondly, the greater the target pest infestation level, the more money is saved on insect control costs compared to conventional cotton managed with conventional insecticide programs. Insect management programs for conventional cotton often require additional monitoring and insecticide applications under high pest-infestation levels, whereas these applications are not necessary in Bollgard® fields. Finally, the price of cotton directly influences the value that Bollgard® cotton delivers to the grower. As Bollgard® cotton protects yield, the higher the price of cotton per pound, the higher the value. In addition to economics, Bollgard® can provide benefits such as time savings, management simplicity and increased safety through reduced insecticide applications. These factors can significantly affect the overall value that an individual grower places on Bollgard® cotton.

Economic comparisons of Bollgard® to conventional cotton using chemical insect control programs have been conducted in many areas of the US cotton belt since its introduction in 1996 (Bryant et al., 1999; Cooke et al., 2000; Karner et al., 2000; Mullins and Mills, 1999; Reed et al., 2000; ReJesus et al., 1997; Seward et al., 2000; Stark, 1997; Weir et al., 1998). Comparisons conducted by independent third parties show that, on average, growers benefit from Bollgard® cotton due to increased yield, decreased insect control costs, or both. Monsanto has conducted economic comparisons since 1995 on 485 large field sites, resulting in a database covering a variety of growing conditions and target pest populations. Most studies have indicated that Bollgard® cotton yields tend to exceed yields of conventional cotton fields, even under low pest infestation levels. This is due to the control provided by Bollgard® cotton of target pest populations when populations are below treatment thresholds or simply cannot be detected by routine scouting (Moser et al., 2000). The prevention of injury, either under low infestation levels or when thresholds are only marginally met, has been termed ‘sub-threshold’ protection. In independent trials, the average economic advantage of Bollgard® over conventional cotton from 1995 to 1999 was US$49.80, with an average yield advantage of 10% over non-Bollgard® comparisons. This is consistent with the average advantage (US$44.70 and 7% increase per acre) calculated for Bollgard® from Monsanto-sponsored trials. The Bollgard® advantage is related to pest population levels in a given season. However, it is apparent that even in very light insect years, such as 1997 and 1999, when insect control costs were higher in Bollgard® cotton (technology fee included) than conventional for some trials, there was an overall economic advantage from higher yields in Bollgard® cotton. This consistent yield advantage may be explained either by agronomic advantages of the Bollgard® varieties and/or better insect control (including sub-threshold protection) with the Bollgard® varieties (Ihrig and Mullins, 2001).

Development of Bollgard® II

From the earliest development of Bollgard® cotton, it was anticipated that the stacking of a second Bt gene would provide increased activity and an expanded spectrum of insect control. A number of candidate proteins were considered. The Cry2A proteins were leading candidates because they lack immunological cross-reactivity with the Cry1Ac protein, allowing the Cry2A protein to be followed independently when ‘stacked’ or bio-engineered into Bollgard® cotton. The Cry2A proteins appear to have distinctly different characteristics compared to the Cry1Ac protein found in Bollgard® (English and Slatin, 1992; English et al., 1994).

Dose–response studies on the susceptibility of the different insect species to various toxins were performed by diet incorporation (Stone et al., 1989). Laboratory-reared tobacco budworm, cotton bollworm, beet armyworm and fall armyworm were obtained from Ecogen, Inc., Langhorne, PA, USA, and pink bollworm was obtained from the insect-rearing facility at the Western Cotton Research Laboratory, Phoenix, AZ, USA. Recombinant strains of Bt were used to express the primary toxins: Cry1Ac, Cry2Ab2 and Cry1Fa, and the Cry1Ac/Cry1Fa hybrid toxin. The toxins were then isolated and purified from sporulated, lysed cultures utilizing standard procedures (Donovan et al., 1988; Donovan, 1991; Malvar et al., 1994). The crystalline preparations of the proteins were then treated with high-pH buffer to solubilize the proteins, after which they were run on SDS–PAGE gels (4–20% acrylamide) and quantified against a bovine serum albumin standard (Dankocsik et al., 1990). A series of six to eight concentrations prepared by serial dilution was used in each instance. Neonates were infested onto the diet. Mortality measurements were recorded 7 days after infestation. Concentration–mortality regressions were estimated assuming the probit model (SAS Institute, 1995). Results were expressed as LC50 in µg ml−1 diet.

Results from these studies are presented in Table 1. The spectrum of activity of the Cry2Ab2 protein indicated that it would be complementary to the activity of Cry1Ac protein, especially against both armyworm cotton pests.

Table 1.  Toxicity data (LC50) for primary toxins and the chimeric protein against cotton pests in the USA
 Cry1AcCry2Ab2Cry1FaCry1Ac/Cry1Fa hybrid
  1. LC50, lethal concentration in µg ml−1, at which 50% of the larvae are dead or not moulted.

Tobacco budworm>>0.020.44>>0.610.04
Cotton bollworm>>2.1116.75>>1002.26
Pink bollworm>>0.010.04>>2.240.01
Beet armyworm>>10043.81>>4.731.93
Fall armyworm>>10076.31>>3.813.99

In order to expedite the introduction of cotton expressing two Bt genes, particle-gun transformation was utilized (McCabe and Martinell, 1993) to transform a commercial cotton cultivar with the cry2Ab2 gene. The Delta and Pineland variety DP50 containing the Bollgard® gene (cry1Ac) was transformed with a synthetic version of the cry2Ab2 gene driven by the enhanced 35S promoter (Kay et al., 1987) and the NOS polyadenylylation 3′ end. The resulting R1 lines were screened using several criteria, including plant phenotype, bioactivity, and expression of the Cry2Ab2 protein. Bioassays were performed using both fresh squares and leaf disks against cotton bollworm and fall armyworm. Lines were selected based on improved efficacy against cotton bollworm and fall armyworm (bioactivity correlating to expression) and phenotype. Enzyme-linked immunosorbent assay (ELISA) was used to determine the expression levels of both Cry2Ab2 and Cry1Ac proteins.

It was determined experimentally that the expression of the cry1Ac gene was unaffected by the introduction of the cry2Ab2 gene. The expression pattern of the cry1Ac gene in the different plant parts, and at different life stages of plant growth, was the same in Bollgard® cotton as in Bollgard® cotton plants transformed with cry2Ab2. The in planta expression level of Cry2Ab2 is higher than that seen for Cry1Ac in Bollgard® cotton. Because the Cry2Ab2 protein has been a challenging protein to work with due to its poor solubility and tendency to aggregate, an alternative method was employed that had been developed for quantifying Cry protein levels. This technique for evaluating the expression level of Cry2Ab2 protein involves the use of a quantitative bioassay developed by Greenplate (1999). This method has the advantage of using a very susceptible insect, tobacco budworm, to estimate the amount of Cry protein, based on developmental inhibition. It is a more direct measure of the activity of the Cry proteins, and is not subject to the difficulties encountered in protein extraction and solubility. Utilizing a Cry1Ac standard curve, the amount of total lepidopteran activity could be estimated and expressed in Cry1Ac equivalents (Greenplate et al., 2000). Expression levels in samples collected from 2 years’ field trails have been estimated using the quantitative bioassay. These field trials were performed in four locations and were sampled over several time points throughout the season. Data derived from field samples have shown that the cry2Ab2 gene present in the transformed cotton events increased the overall efficacy against the key lepidopteran insect pests of cotton about fourfold. Field trials using Bollgard® II cotton events in 2000 showed that the increased lepidopteran activity observed in laboratory assays translated to greater control of lepidopteran pests in the field (Voth and Greenplate, 2001). Similar varieties of cotton bio-engineered with Bollgard®, Bollgard® II or conventional cotton were compared in small-plot replicated trials. Each plot consisted of four rows 50 feet long, replicated four times. The plots were not sprayed for lepidopteran insects, so the relative efficacy of each treatment could be determined.

In the field it has been observed that control of tobacco budworm and pink bollworm is essentially complete for Bollgard® and is the same for Bollgard® II. Of particular interest is the efficacy of Bollgard® II on those species that are not completely controlled with Bollgard®. Figure 3 shows the relative efficacy on the four primary species where increased control would have value to cotton growers. For cotton bollworm, Bollgard® usually provides good control, although over sprays are sometimes needed. For fall armyworm, beet armyworm and soybean looper Bollgard® provides limited control, whereas Bollgard® II provided excellent control. These results correlate well with estimates of LC50 determined utilizing diet-incorporated bioassays on laboratory colonies of cotton bollworm, beet armyworm and fall armyworm (Sivasupramaniam et al., 2000; Table 1). The increased spectrum of lepidopteran pests controlled by Bollgard® II cotton is attributed to the wider spectrum of activity of Cry2Ab2, as well as to the higher expression of this gene in the Bollgard® II line (Figure 3). Results from 3 years’ field trials show that Bollgard® II provides excellent control of the economically important lepidopteran species in the USA. The improved control of Bollgard® II has resulted in increased boll retention and higher yields (data not shown).

Figure 3.

Relative insect efficacy of Bollgard® and Bollgard® II cotton compared to unsprayed conventional treatments.

The data indicate that the combination of the cry2Ab and cry1Ac genes in Bollgard® II cotton gives superior control of lepidopteran pests. This observed improvement in the control of these pests should result in a further decrease in the use of conventional pesticides in cotton production (Ridge et al., 2000). The higher levels of expression and related improvements in bioactivity provide excellent control of all bollworm species attacking cotton, and have positive implications for resistance management issues, especially with reference to cotton bollworm. Information relating to the environmental, food and feed safety of Bollgard® II has been submitted to the appropriate US regulatory agencies and is currently in the process of review.

Future improvements to Bollgard® cotton

As a corollary to the improvements rendered by the development of Bollgard® II cotton, protein-engineering efforts have focused on enhancing the activity of the cry1Ac gene. Further improvements in insect efficacy were achieved by the use of Cry1Ac/Cry1Fa chimeric proteins (Malvar et al., 2000). Efficacy studies have been conducted on a Cry1Ac/Cry1Fa chimeric protein in cotton (Sivasupramaniam et al., 2000). By replacing the domain III and a small part of domain II of the activated cry1Ac gene with the corresponding segment of cry1Fa, an increase in the spectrum of insects that is controlled has been observed (Table 1). Bioassays with both cotton bollworm and fall armyworm fed on fresh cotton leaves and squares expressing this chimeric protein have shown significantly higher mortality of fall armyworm, and at least equivalent activity against cotton bollworm. Greenhouse studies indicate that these plants have normal, acceptable phenotypes. Several events have been earmarked as a result of these laboratory investigations. Field trials with these new cotton events will be conducted in the USA in 2001. Successful demonstration of efficacy on fall armyworm and beet armyworm in these trials may lead to the development of a cotton product that provides excellent control of all lepidopteran pests of cotton: budworms/bollworms, armyworms, cutworms, loopers and perforators.


Bollgard® cotton has been available in the market place for 6 years. This product has been widely planted on more than one-third of the total annual cotton acreage in the USA. Bollgard® cotton provides control of budworm and bollworm that is superior to chemical insecticides. It allows the grower the option of reducing his insect control costs and increasing his cotton yields. Some of the benefits of Bollgard® cotton are shared by the grower with both the community surrounding cotton-growing regions, and the consumers. The reduction of chemical insecticide use results in less insecticide exposure for the grower, his family, his neighbors and the surrounding community. Reduction of production costs through the use of Bollgard® cotton allows growers to remain competitive in the global market place, and provides stability and sustainability in cotton production that benefits both the cotton industry and the consumer. Monsanto, as the technology provider, and seed companies also share in the profitability of cotton production using Bollgard® cotton. This return on investment has spurred investment and competition in cotton research which will result in products for improved cotton production. This eventually benefits the grower as well. Bollgard® cotton is a glimpse of what is possible in providing safe, improved alternatives for agricultural production through the use of new technology.


The authors would like to thank Roy Fuchs, Karen Marshall and Sandra Hurston for critical review of the manuscript.