• Open Access

RNAi-mediated Ultra-low gossypol cottonseed trait: performance of transgenic lines under field conditions


Correspondence (Tel 979-862 4795; fax 979-862 3414; email rathore@tamu.edu)


Cottonseed remains a low-value by-product of lint production mainly due to the presence of toxic gossypol that makes it unfit for monogastrics. Ultra-low gossypol cottonseed (ULGCS) lines were developed using RNAi knockdown of δ-cadinene synthase gene(s) in Gossypium hirsutum. The purpose of the current study was to assess the stability and specificity of the ULGCS trait and evaluate the agronomic performance of the transgenic lines. Trials conducted over a period of 3 years show that the ULGCS trait was stable under field conditions and the foliage/floral organs of transgenic lines contained wild-type levels of gossypol and related terpenoids. Although it was a relatively small-scale study, we did not observe any negative effects on either the yield or quality of the fibre and seed in the transgenic lines compared with the nontransgenic parental plants. Compositional analysis was performed on the seeds obtained from plants grown in the field during 2009. As expected, the major difference between the ULGCS and wild-type cottonseeds was in terms of their gossypol levels. With the exception of oil content, the composition of ULGCS was similar to that of nontransgenic cottonseeds. Interestingly, the ULGCS had significantly higher (4%–8%) oil content compared with the seeds from the nontransgenic parent. Field trial results confirmed the stability and specificity of the ULGCS trait suggesting that this RNAi-based product has the potential to be commercially viable. Thus, it may be possible to enhance and expand the nutritional utility of the annual cottonseed output to fulfil the ever-increasing needs of humanity.


Cotton, one of the world's most important cash crops, is grown in more than 80 countries across the world. While cotton is cultivated for its fibre, the fact remains that the plant produces approximately 1.6 times more seeds than lint, on a weight basis. Cottonseed is rich in oil (approximately 21%) and protein (approximately 23%) that are of relatively high quality. The amount of protein present in the global cottonseed output can potentially provide the basic protein requirements of half a billion people, annually. However, due to the presence of toxic gossypol, a cardio- and hepatotoxic terpenoid, cottonseed is unfit for consumption by humans or monogastric animals (Risco and Chase, 1997). Gossypol and other related terpenoids are present in the lysigenous pigmented glands, located on the surface, throughout the cotton plant (Adams et al., 1960) and play an important role in defending the plant against pests and some diseases (Bell and Stipanovic, 1978; Bottger et al., 1964; Hedin et al., 1992). Gossypol is the major terpenoid present in the seed glands, and most commercial cottonseeds contain 0.52%–1.01% gossypol (Calhoun et al., 2004). Seed-gossypol concentration can vary depending on the variety and the environmental conditions during the growing season. Currently, the main usage for cottonseed is as a cattle feed, before or after oil extraction. Ruminant animals are able to tolerate a limited amount of gossypol in their diets. However, these animals are highly inefficient in converting feed into meat (feed conversion ratio for beef cattle: 5.8) (Klopfenstein et al., 1991). In contrast, poultry and a number of aquaculture species (e.g. shrimp, catfish, tilapia and salmon) are 3–5 times more efficient in feed conversion (Rathore et al., 2007), but are much more vulnerable to gossypol toxicity. Therefore, elimination of gossypol from the cottonseed will make this abundant and nutritionally rich resource safe for use as feed for various monogastric animals and possibly even as human food. While it is possible to remove glands/gossypol using physical/chemical means, the processing steps are cost prohibitive. Therefore, the discovery of a glandless mutant in the 1950s (McMichael, 1959, 1960) and subsequent introgression of this trait into commercial cultivars by breeders generated a great deal of excitement and provided hope for the utilization of the seeds as feed and food (Lusas and Jividen, 1987). A number of studies, conducted in various parts of the world, showed that glandless cottonseed could serve a useful source of nutrition not just for monogastric animals, but also for humans (Lusas and Jividen, 1987; Rathore et al., 2007). Unfortunately, due to the lack of the glands and therefore the protective terpenoids in the vegetative and floral parts of the plant, the glandless varieties suffered higher pest damage and had lower yields under field conditions (Jenkins et al., 1966; Vaissayre and Hau, 1985). Thus, although the glandless cottonseed proved fit as a source of feed for monogastric animals and even as food, it was not accepted widely by growers. The unfulfilled goal of utilizing cottonseed to meet the nutritional needs of the ever-increasing human population, however, can be achieved if the elimination of gossypol is restricted to the seed so that the rest of the plant still retains its natural, terpenoid-based defensive capability against pests and pathogens.

By interfering with the expression of the δ-cadinene synthase gene through RNAi during seed development, we have disrupted gossypol biosynthesis to generate ultra-low gossypol cottonseed (ULGCS; Sunilkumar et al., 2006; Rathore et al., 2007). Studies conducted on several ULGCS lines, under greenhouse conditions, showed that the RNAi-mediated trait was stable and heritable. Furthermore, the levels of gossypol and related terpenoids in the foliage and floral organs were not diminished, and thus their potential function in plant defence remained intact. In a follow-up study, Rathore et al. (2012) showed that ULGCS lines were capable of launching a terpenoid-based defence response during the early stages of seedling development. However, all the results described in these two reports were obtained from plants that were grown under controlled, greenhouse/growth chamber conditions. There have been instances where a transgene-mediated trait expressed in the greenhouse was unstable under field conditions. For example, transgenic tobacco, expressing a mutant form of csrl-1 gene that had shown a high level of resistance to sulfonylurea herbicide in the greenhouse, became susceptible to this herbicide under field conditions (Brandle et al., 1995). Similarly, Sharp et al. (2002) observed that transgenic wheat plants expressing viral coat proteins against wheat mosaic virus were resistant in greenhouse studies, but lost this trait in the field. In a separate study conducted on wheat, researchers reported that the effect of transgene expression varied from year to year, based on the climatic conditions of a particular growing season (Bahieldin et al., 2005). Given these uncertainties in transgene behaviour, it was considered essential to investigate the performance of ULGCS lines in the field, with its inherent biotic and abiotic pressures. Here, we present the results of field trials conducted over 3 years where the stability and specificity of the trait was studied and the performance of selected ULGCS lines was compared with the wild-type parent (Coker 312). We provide data not only on the yield of fibre and seed, but also on the quality and composition of these two important products of the cotton crop.


RNAi lines maintain the ultra-low gossypol levels in the seed

Gossypol content in the seed kernels from transgenic and nontransgenic plants grown over three seasons in the field was analysed by high pressure liquid chromatography (HPLC). The results presented in Figure 1 show clearly that the transgenic lines maintained the ULGCS trait under field conditions. The seed-gossypol content in the two original RNAi lines (66-49B and 66-81) continues to be below the FDA-approved limit of 450 ppm in food products. The new line 66-250, tested for the first time in the field in 2011, had seed-gossypol value of 450 ppm. Six plants were also grown from each of these lines in the greenhouse at the same time for comparison during years 2009 and 2010. Again, the seed-gossypol content of the transgenic plants was lower than 450 ppm for the greenhouse-grown plants (Figure S1).

Figure 1.

Levels of gossypol in pooled, kernel samples obtained from 12–15 mature seeds, each from nontransgenic Coker 312 plants and the transgenic lines (66-49B, 66-81 and 66-250) grown under field conditions. Note that the RNAi line 66-250 is relatively new and was grown only in the 2011 season. Data represent mean ± SE, ***P < 0.001; n = 12 individual plants (2009), n = 3 rows, with each row representing average of 10 plants (2010), and n = 4 rows, with each row representing average of 10 plants (2011). Mean kernel-gossypol level for each type of plant is shown above each bar. Note that the predominant terpenoid in the cottonseed glands is gossypol. A representative seed kernel, sliced in half to expose the glands, is also shown for each type of plant. Note that the glands are still present in the ULGCS; however, these lack the dark brown coloration of the wild-type seeds.

Nonseed tissues maintain normal, wild-type levels of gossypol and related terpenoids

Terpenoid analyses were performed each year on six different tissues (leaves, bracts, terminal part of the axillary branch, floral buds, petals and 2-day-old bolls) that were collected about 8–10 weeks after planting (Table 1). Unlike the seeds where we observe over 95% reduction in the levels of gossypol in the transgenic lines, comparatively minor differences were observed for various terpenoids in the nonseed tissues between the controls and RNAi lines. Although some of the terpenoid levels in 66-49B are lower compared with the controls, the results for the other two lines suggest that this is not due to their inherent transgenic status. Thus, the vegetative and floral tissues of the transgenic lines should retain their terpenoid-based defence capabilities. The field was monitored closely for insect pests and diseases on a regular basis. Some of the usual cotton pests, such as thrips, aphids, flea hoppers and stink bugs as well as foliar pathogens, including Alternaria macrospora, Alternaria alternata, Cercospora gossypina, etc. that cause leaf spot disease, were observed in the field. However, we did not notice any differences between the ULGCS lines and their nontransgenic counterparts in terms of the susceptibility to pests or pathogens. The same six types of tissues were also collected for terpenoid analysis from the greenhouse-grown plants, on the same day as in the field, in years 2009 and 2010. Overall, no major differences in the terpenoid levels in nonseed tissues were found between the transgenic lines and the nontransgenic controls (Table S1).

Table 1. Levels (μg/mg dry weight) of gossypol and related terpenoids [Mean (±SE), n = 3] in different tissues obtained from cotton plants grown in the field
Coker 31266-49B66-81Coker 31266-49B66-81Coker 31266-49B66-8166-250
  1. G, gossypol; HGQ, hemigossypolone; H1-H4, heliocides.

  2. Note that the predominant terpenoid in the flower petal glands is gossypol.

G0.08 (±0.008)0.06 (±0.003)0.11 (±0.011)0.11 (±0.01)0.10 (±0.03)0.12 (±0.02)0.08 (±0.00)0.06 (±0.00)0.16 (±0.02)0.11 (±0.00)
HGQ0.15 (±0.02)0.10 (±0.012)0.10 (±0.014)0.17 (±0.04)0.12 (±0.04)0.11 (±0.03)0.25 (±0.04)0.24 (±0.02)0.32 (±0.08)0.38 (±0.02)
H1–H41.25 (±0.05)0.89 (±0.06)1.05 (±0.10)1.78 (±0.32)1.59 (±0.46)1.44 (±0.32)1.51 (±0.25)1.07 (±0.59)1.75 (±0.47)1.82 (±0.09)
Floral buds
G5.65 (±0.42)5.04 (±0.39)6.41 (±0.91)3.15 (±0.28)3.14 (±0.43)3.47 (±0.19)3.09 (±0.10)3.17 (±0.05)4.01 (±0.08)3.93 (±0.07)
HGQ1.11 (±0.14)0.96 (±0.11)0.88 (±0.23)0.45 (±0.09)0.36 (±0.06)0.40 (±0.07)0.51 (±0.06)0.61 (±0.09)0.66 (±0.02)0.87 (±0.04)
H1–H42.39 (±0.09)1.99 (±0.22)1.90 (±0.40)1.47 (±0.32)1.46 (±0.19)0.90 (±0.03)1.89 (±0.13)2.07 (±0.19)2.30 (±0.13)2.40 (±0.12)
Terminal part of axillary branch
G1.04 (±0.30)0.99 (±0.26)0.82 (±0.03)1.02 (±0.03)0.98 (±0.24)1.15 (±0.12)0.92 (±0.14)1.15 (±0.08)1.32 (±0.31)1.17 (±0.19)
HGQ1.62 (±0.16)1.22 (±0.19)1.15 (±0.17)1.24 (±0.06)0.98 (±0.05)1.12 (±0.09)2.22 (±0.21)1.89 (±0.19)1.96 (±0.27)2.27 (±0.12)
H1–H42.84 (±0.27)1.77 (±0.23)2.43 (±0.40)1.64 (±0.14)1.74 (±0.25)1.94 (±0.17)3.19 (±0.43)2.70 (±0.11)2.91 (±0.21)3.67 (±0.16)
G0.34 (±0.08)0.19 (±0.02)0.48 (±0.13)0.80 (±0.01)0.43 (±0.04)0.97 (±0.04)0.65 (±0.09)0.62 (±0.04)1.12 (±0.12)0.63 (±0.01)
HGQ1.66 (±0.49)0.78 (±0.06)1.37 (±0.37)2.25 (±0.23)1.25 (±0.09)2.09 (±0.08)4.04 (±0.40)3.73 (±0.30)4.38 (±0.18)3.57 (±0.15)
H1–H42.68 (±0.36)1.77 (±0.23)2.46 (±0.76)1.81 (±0.38)0.91 (±0.07)1.14 (±0.09)2.94 (±0.25)2.77 (±0.59)3.51 (±0.47)2.57 (±0.11)
2-day bolls
G1.69 (±0.09)1.37 (±0.08)1.88 (±0.07)2.02 (±0.24)1.22 (±0.05)1.47 (±0.06)1.06 (±0.06)0.86 (±0.02)1.04 (±0.06)0.89 (±0.01)
HGQ7.29 (±0.05)6.43 (±0.23)6.58 (±0.41)4.29 (±0.28)3.69 (±0.24)3.05 (±0.05)4.95 (±0.41)4.57 (±0.26)3.98 (±0.25)4.32 (±0.03)
H1–H411.03 (±0.14)9.05 (±0.54)7.64 (±0.39)11.27 (±3.58)9.96 (±1.88)5.81 (±1.31)8.66 (±0.73)8.74 (±0.54)5.51 (±0.60)7.62 (±0.38)
G5.70 (±0.15)6.28 (±0.22)6.25 (±0.25)4.77 (±0.26)5.92 (±0.13)5.97 (±0.05)3.60 (±0.19)3.78 (±0.30)4.15 (±0.18)4.22 (±0.14)

ULGCS lines do not suffer seed/fibre-yield penalty

Seed cotton was hand harvested from individual plants, air dried and then ginned on a roller gin to separate the fibre from the seed. The fibre yields for the transgenic lines were higher than those of nontransgenic controls (Coker 312) in the years 2009 and 2011 (Figure 2a). For reasons unknown to us, we observed much higher incidence of tight lock (adherence of seed cotton in compact wedges in the locules of opened bolls) in the nontransgenic controls in the field during these 2 years. Whatever the cause, it may account for the lower yield of the fibre in the controls during those 2 years. The occurrence of tight lock was rare in any of the previous greenhouse studies, including the one in 2009, in either transgenic or nontransgenic plants. Overall, the fibre yield was lower in 2010 in the field for both transgenic lines and controls, with line 66-49B being the lowest. In 2009, the seed yield for nontransgenic controls was lower compared with the transformants, while the seed production was similar for each type of plants in the following two seasons (Figure 2b). As was the case for fibre, the seed yields were lower in 2010 for all types of plants. Similar to the observations under field conditions, the seed and fibre yields for the transgenic lines were comparable to or better than the nontransgenic controls in both 2009 and 2010 in the greenhouse (Figure S2).

Figure 2.

Average yields for fibre (a) and seed (b) for nontransgenic Coker 312 plants and the transgenic lines (66-49B, 66-81 and 66-250) grown under field conditions. Note that the RNAi line 66-250 is relatively new and was grown only in the 2011 season. Data represent mean ± SE, *P < 0.05; ***P < 0.001, n = 12 individual plants (2009), n = 3 rows, with each row representing average of 10 plants (2010), and n = 4 rows, with each row representing average of 10 plants (2011).

With the exception of gossypol, ULGCS composition is similar to the wild-type cottonseeds

Compositional analysis results for acid-delinted, cottonseed samples obtained from the 2009 field trial are presented in Tables S2–S5. Along with two transgenic lines and the nontransgenic parental cultivar (Coker 312) from the field trial, this analysis was also performed on seeds of a glandless cotton cultivar (Acala GLS) and a commercial cultivar (DP50) for comparison. We have also provided a range of values obtained from International Life Sciences Institute database for comparison with our seed composition results (International Life Sciences Institute (ILSI), 2010). Results of this analysis show that the levels of proximate components (protein, carbohydrate, moisture and ash) in the seeds of ULGCS lines were comparable to that of the nontransgenic control (Coker 312). As described later, glandless cottonseed is known to have higher oil content. Thus, it was interesting that the seed oil content of the two transgenic lines was slightly higher compared with Coker 312 (Table S2). Fatty acid composition results, presented in Table S3, do not show any major differences between transgenic lines and the respective parental control. Amino acid composition results, presented in Table S4, show that their levels in the seeds of ULGCS lines are comparable to those in the seeds from the controls. No major difference was observed in the seed mineral composition (Table S5) between transgenic lines and the nontransgenic parent, except for calcium. ULGCS line, 66-49B, has substantially lower calcium content compared with the parental line. This lower level, however, is similar to the commercial line DP50 and is within the range of values reported for U.S. cotton in the ILSI database. The levels of vitamin E were also lower in 66-49B, but within the range provided in the ILSI database. The lower values for these two components, however, are not inherent properties of their transgenic status, because the ULGCS line 66-81 showed levels similar to the wild type. Total and free gossypol contents were also measured. Indeed, the seeds from the two ULGCS lines have substantially lower gossypol content compared with Coker 312 seeds. As expected, the seeds of glandless Acala GLS show very low levels of gossypol.

ULGCS have equivalent protein content and slightly higher oil levels compared with the wild-type cottonseeds

As protein and oil are the two most valuable components of the cottonseed, we examined their levels in the seeds obtained from all 3 years. We observed year-to-year differences in the seed protein levels, with the highest protein values occurring in the year 2010 (Figure 3). Despite these yearly fluctuations, the overall seed protein contents for the RNAi lines were similar to those of nontransgenic controls. It is interesting that seed oil levels also show the year-to-year differences across all entries, but the trend is reversed in relation to the protein values (Figure 3). Thus, there is an inverse correlation between protein and oil content in the cottonseed. Also, we observed another interesting and useful characteristic in the ULGCS. The seed oil values were consistently higher in the transgenic lines for all 3 years. Even the new ULGCS line (66-250) that was tested in the year 2011 for the first time showed this trend. Importantly, the year-to-year fluctuation in seed oil content does not affect the relative gain in oil levels in the ULGCS lines.

Figure 3.

Protein (a) and Oil (b) levels in samples obtained from seeds of nontransgenic Coker 312 plants and the transgenic lines (66-49B, 66-81 and 66-250) grown under field conditions. Note that the protein values were measured in seed kernels, while oil content was determined in whole seeds. The RNAi line 66-250 is relatively new and was grown only in the 2011 season. Data represent mean ± SE, *P < 0.05; **P < 0.01, n = 3.

Fibre quality of the ULGCS lines is maintained

Cotton's ultimate value is as a fibre crop, so it is critical that we examine the quality of the fibre for any inadvertent effects of metabolic engineering of the seed. We used High Volume Instrumentation ([USTER® HVI (Uster Technologies AG, 8610 Uster, Switzerland)] to measure various fibre quality parameters (micronaire, MIC; upper half mean length, UHM; uniformity index, UI; strength, STR; and elongation, ELO). The results from this analysis are presented in Figure 4 and Figure S3. Of the five most important quality parameters for cotton fibre, only micronaire values showed some differences between RNAi lines and nontransgenic controls grown in the field. Specifically, micronaire values were higher for the transgenic lines versus the Coker 312 control in the years 2009 and 2011 but not in 2010. The higher incidence of tight lock in the control plants in 2009 and 2011 may account for the differences in micronaire values. For plants grown in the greenhouse, with the exception of micronaire in 2010, the fibre quality parameters of transgenic lines again were similar to those of nontransgenic Coker 312 (Figure S3). In conclusion, there was no diminution in the quality of the fibre in ULGCS lines.

Figure 4.

Fibre quality parameters (MIC: micronaire; STR: strength; UHM: upper half mean length; UI: uniformity index; ELO: elongation) as measured by High Volume Instrumentation (USTER® HVI) for samples derived from nontransgenic Coker 312 plants and the transgenic lines (66-49B, 66-81 and 66-250) grown under field conditions. Note that the RNAi line 66-250 is relatively new and was grown only in the 2011 season. Data represent mean ± SE, *P < 0.05; **P < 0.01; ***P < 0.001, n = 12 individual plants (2009), n = 3 rows, with each row representing average of 10 plants (2010), and n = 4 rows, with each row representing average of 10 plants (2011).


In our previous studies, we had examined nine different ULGCS lines for five generations under greenhouse conditions and found the trait to be heritable and stable (Rathore et al., 2012; Sunilkumar et al., 2006). As mentioned earlier, there are reports showing that transgene efficacy that was observed under greenhouse conditions either weakened or showed variability under field conditions (Bahieldin et al., 2005; Brandle et al., 1995; Dietz-Pfeilstetter, 2010; Sharp et al., 2002; Stam et al., 1997). Prior to our use of RNAi to obtain the ULGCS trait, we had attempted to employ antisense technology to achieve the same objective. As reported in our recent publication (Rathore et al., 2012), antisense-mediated reduction of seed gossypol was unstable, even under greenhouse conditions. As RNAi technology is just beginning to be used to obtain valuable traits, there are only a handful reports on the field performance of transgenic plants engineered through RNAi-mediated gene silencing (Fornalé et al., 2012; Huang et al., 2009; Niu et al., 2008). Therefore, it was critically important to evaluate the stability of the RNAi-mediated ULGCS trait along with the agronomic performance of the plants under field conditions. Here, we present a comprehensive report on the first such study conducted to evaluate an RNAi-mediated trait in the field in cotton. Field trials were conducted on selected ULGCS lines over 3 years (2009–2011) that included examination of trait stability, tissue specificity, yield, seed composition and fibre quality. These trials established that the ULGCS lines maintained their transgenic trait under field conditions, and their agronomical performance was not diminished.

The specificity of the ULGCS trait was already shown under greenhouse conditions (Sunilkumar et al., 2006). As the greater susceptibility of glandless cotton to insects under field conditions was the main factor for its commercial failure, it was all the more important to establish that the nonseed parts of the ULGCS lines retain normal levels of defensive terpenoids in the field to defend the plant against a variety of biotic pressures. The results presented in this paper show that the terpenoid levels in six different vegetative and floral tissues in the ULGCS lines were comparable to those in the parental cultivar. Thus, unlike glandless cotton, the ULGCS plants should be able to defend themselves against pests and pathogens to the same extent as their wild-type parent under field conditions. Indeed in our previous study, we have shown that the ULGCS seedlings were able to launch a terpenoid-based defence response when challenged with a pathogen under laboratory conditions (Rathore et al., 2012).

Fibre and seed yields of the ULGCS lines were comparable or slightly better than the nontransgenic controls in the field. Compared with the years 2009 and 2011, the seed cotton yields for the year 2010 were lower for the transgenic lines and the controls (Figure 2). It is possible that various climatic factors played a role in reducing the yields during the 2010 season. Compared with 2009 and 2011, there was higher precipitation in 2010 (Table S6). As a result, there was a noticeably higher incidence of diseases and insect pests during the 2010 season. Also, Cabangbang and Manguiat (1989) reported that cotton cultivars grown in high rainfall and low solar radiation had significantly lower yields compared with the same cultivars when grown in low rainfall and higher solar radiation environments as long as the crops were adequately irrigated. The number of cloudy days in the growing season (April-August) for the years 2009, 2010 and 2011 were 30, 48 and 19, respectively (http://www.srh.noaa.gov). Thus, lower solar radiation, due to more cloudy days, coupled with higher incidence of diseases and insect pests may have been responsible for the lower yields in the year 2010. Nevertheless, the RNAi lines, in general, provided yields similar to or better than the nontransgenic controls.

Potential utility of the ULGCS will be as a source of protein and oil. Therefore, we examined the levels of these two important components in the cottonseed obtained from all 3 years of field plantings. Our results show no significant difference in the seed protein content between the transgenic and nontransgenic plants for 3 years. However, the total seed oil content of the ULGCS lines was higher than that of the control plants, and the differences were statistically significant in most cases (Figure 3). In oilseed crops, oil and protein constituents are synthesized at different rates and times during seed development (Blumenthal et al., 2008). Gallup (1927, 1928) showed that the accumulation of oil and gossypol in the cottonseed occurs at about the same time. Thus, a reduction in gossypol production is expected to result in energy/resource surplus. It is possible that the cottonseed may divert this surplus to produce more oil. Support for this notion comes from the fact that glandless cottonseed has higher oil content (Ramos and Kohel, 1987). Thus, the consistently higher oil levels in the ULGCS may well be due to the substantial reduction in seed-gossypol content. Extensive experimentation will be required to identify a mechanism that might underlie this relationship between the oil and gossypol accumulation in cottonseeds.

As removal of gossypol from the oil of glanded cottonseed is an expensive process, the use of ULGCS should reduce the cost of refining cottonseed oil before it is fit for human consumption. The oil derived from cottonseed can also be a good source of lecithin, which is an excellent emulsifying agent in food products. However, the gossypol in the glanded seed binds to lecithin in oil and makes it harder and more expensive to extract. Cherry et al. (1981) have shown that the glandless cottonseed can be an excellent source of food-grade lecithin because of the absence of gossypol in the seed. Thus, ULGCS with its ultra-low levels of gossypol offers a potentially more economic source of lecithin.

There was year-to-year fluctuation in the level of protein in cottonseed across all entries. Similar yearly variation was observed for the seed oil content, but in the opposite direction (Figure 3). Such an inverse correlation between the oil content and protein levels has been reported previously. There are several reports showing that the protein and oil contents of cottonseed are negatively correlated (Horn et al., 2011; Pandey and Thejappa, 1975; Stansbury et al., 1956). In oil seed crops, nitrogen fertilization results in higher protein levels in the seed while reducing the oil content (Blumenthal et al., 2008). Such an effect of nitrogen fertilization has been shown for sunflower (Geleta et al., 1997; Mathers and Stewart, 1982; Steer et al., 1984), rapeseed (Holmes and Ainsley, 1979; Krogman and Hobbs, 1975), soybean (Hassan et al., 1985; Jadhav et al., 1994) and cotton (Sawan et al., 2006). The amount of nitrogen applied to our field plot was 35, 42 and 12 lbs/acre in the years 2009, 2010 and 2011, respectively. Pettigrew and Dowd (2011) observed that frequent irrigation of cotton fields resulted in increased oil content and decreased protein content in the seed. Due to scant rainfall, our field was irrigated frequently in 2009, and even more in 2011. This along with other climatic factors and the amount of nitrogen applied may account for the year-to-year differences in oil and protein levels in cottonseeds in our field trials. The relatively higher oil content in the ULGCS compared with the nontransgenic seeds was maintained despite the year-to-year fluctuations in the overall oil levels. This observation provides additional support for the notion that elimination of gossypol from cottonseed improves its oil content.

The results from composition analysis conducted on seeds from the 2009 trial suggest that silencing of the δ-cadinene synthase gene in the seed through RNAi largely affects the levels of gossypol in the seed. The fatty acid and amino acid profiles of the two most important nutritional components of the seed are similar to those of wild-type seeds. This is also true for the levels of minerals and vitamins (as mentioned earlier, the lower values of vitamin E and calcium in one of the RNAi lines do not represent the inherent property of these transgenics). Thus, the overall quality of the ULGCS is not affected in any other major way, and therefore, the nutritional value of these seeds should be substantially higher than those of the nontransgenic seeds because of the virtual elimination of gossypol. ULGCS therefore has the potential to become a suitable source of nutrition for monogastrics. Gossypol binds to the epsilon-amino group of lysine in a protein and thus reduces the bioavailability of this important amino acid. Therefore, even ruminant animals that can tolerate a limited amount of gossypol in the diet should benefit from ULGCS by the greater availability of lysine. It should be noted that the levels of cyclopropenoid fatty acids (CPFA; sterculic acid, malvalic acid and dihydrosterculic acid) in the ULGCS are similar to those in the seeds from parental plants. The anti-nutritive properties of these fatty acids are a result of their ability to inhibit desaturation of stearic acid to oleic acid (Johnson et al., 1967; Reiser and Raju, 1964). During the processing of cottonseed, the bulk of CPFAs are retained in the oil portion, leaving the meal nearly free of these compounds. Further refining, especially the deodorization process, virtually eliminates CPFAs from the oil (Eaves et al., 1968). Thus, consumption of either the meal or refined oil should pose little risk to either animals or humans. Also, it is possible to eliminate these compounds by silencing the gene(s) that encode enzyme(s) involved in their production (Liu et al., 2004; Yu et al., 2011).

Cotton will always be grown for its fibre and not for its seed irrespective of the improvements made in the quality of this by-product. For this reason, if genetic manipulation of the seed quality affected the quality or yield of the fibre, the product will not be commercially viable. Technically, the fibre cells are of maternal origin, and we have utilized a promoter that is exclusively active in the developing embryo (Sunilkumar et al., 2002). Therefore, it is unlikely that genetic manipulation of the seed kernel will have a direct effect on fibre yield/quality. The results presented in this paper show clearly that neither the yield nor the quality of the fibre is adversely affected in the ULGCS lines. In summary, this study demonstrated the robustness of the RNAi-mediated, ULGCS trait under field conditions over several generations. Routine, visual monitoring of plants did not reveal any differences between the wild-type and transgenic plants in their susceptibility to pests or pathogens. Importantly, ULGCS lines were not compromised in terms of yield, seed composition or fibre quality. Thus, unlike the glandless trait developed some decades ago, the ULGCS trait has the potential to be commercially viable. Additional, large-scale, multi-location field studies will be conducted to better understand the agronomics including yield and transgene performance in the elite germplasm. Seed composition analyses will be performed on seed samples from various trials to ensure substantial equivalence and maintenance of the ULGCS trait. Cottonseed obtained from these field trials will be used for various animal feeding studies to ascertain its safety and nutritional efficacy.

Materials and methods

Field planting

Field trials were conducted over 3 years (2009, 2010 and 2011) from April to October at Texas A&M Field Laboratory, Somerville, TX. Two different RNAi lines, 66-49B and 66-81, were used along with the nontransgenic parent (Coker 312). Prior to their use in the field trials, the two transgenic lines had been tested for five generations under greenhouse conditions, and the ULGCS trait was found to be stable (Rathore et al., 2012). In the third year, a recently generated RNAi line, 66-250, was also evaluated in the field. In 2009, 30 plants from each transgenic line and controls were grown in a single block design without replications. The data presented for 2009 represent seed cotton collected from 12 plants in the middle of the row for each line, with each plant being considered a replicate. In the subsequent years, the plants were grown in a randomized complete block design with 10 plants each per plot, with three replications in 2010 and with four replications in 2011. The trials were conducted in the same field every year. Harvesting was carried out by hand, multiple times every season, and the seed cotton was then left to dry in paper bags for a few weeks before ginning on a roller gin. Weight measurements were carried out for fibre and fuzzy seeds from each individual plant. In years 2009 and 2010, we also grew six plants of each type in the greenhouse for comparison.

Determination of Gossypol and Related Terpenoids

Each year, six different tissues (leaves, bracts, terminal ends of axillary branches, floral buds, petals and 2-day-old bolls) were harvested 8–10 weeks after sowing, freeze-dried and ground to a fine powder. Terpenoid analyses were performed on these samples and cottonseed kernels using HPLC following the methods described by Stipanovic et al. (1988) and Benson et al. (2001). Briefly, the finely ground green tissue/petal sample (approximately 100 mg) was extracted by ultrasonication (10 min) in 5 mL of solvent containing acetonitrile/water/phosphoric acid (80 : 20 : 0.1) in a 15-mL polypropylene tube. Following centrifugation at 2800 × g for 5 min, a 50-μL fraction of the extract was analysed on an Agilent Technologies (Palo Alto, CA) 1200 liquid chromatograph, equipped with a diode array detector for compound spectral identification (Stipanovic et al., 1988; Sunilkumar et al., 2006).

A slightly different procedure was used for extracting terpenoids from seeds. Twelve to 15 seeds from each plant were dehulled, and the kernels were ground to a fine powder, weighed and then mixed with 50 mL of solvent containing ethanol/ether/water/glacial acetic acid (59 :17 : 24 : 0.2). The suspension was agitated on a shaker at room temperature for 1 h to facilitate extraction of terpenoids. The final sample volume was adjusted to 50 mL to account for evaporation and centrifuged for 15 min at 2800 × g. A 50-μl fraction of the extract was analysed using the HPLC as described above.

Seed compositional analysis

Compositional analyses were conducted by Covance Laboratories, Inc. (Madison, WI) on the seed samples obtained from the 2009 harvest. Acid-delinted cottonseed samples of the two transgenic lines (66-49B and 66-81) and the nontransgenic Coker 312 controls were analysed for proximates (ash, moisture, fat, carbohydrates and protein), acid detergent fibre, neutral detergent fibre, minerals (calcium, copper, iron, magnesium, manganese, phosphorus, potassium, sodium and zinc), amino acids, fatty acids, cyclopropenoid fatty acids, phytic acid, free gossypol, total gossypol and vitamin E. Seeds from a glandless cotton cultivar (Acala GLS) and one commercial, glanded cultivar (‘Deltapine 50’, DP50) were also included in this analysis for comparison.

Seed oil and protein content

AOAC Method 990.03 (1989) was used to estimate the total protein content in the cottonseed kernels. Briefly, the ground kernel sample was encapsulated within a tin receptacle and dropped into the combustion unit and heated to 950 °C. The gaseous products of combustion were then collected, cooled and allowed to mix thoroughly. A 10-mL aliquot of this gas mixture was taken and passed through columns containing copper, CO2 absorbent and desiccant to reduce nitrogen oxides to N2 and to remove all CO2 and H2O, respectively. The amount of nitrogen within the sample was measured by a thermal conductivity cell and reported as per cent nitrogen of the sample. Per cent protein was calculated by multiplying the reported nitrogen by 6.25, the protein conversion factor.

Whole cottonseed oil content was measured using an Oxford 4000 continuous wave Nuclear Magnetic Resonance (NMR) analyser. The NMR method is based on the principle of activating hydrogen atoms in the sample using electromagnetic radiation in a magnetic field. The NMR reading is a measure of the number of activated hydrogen atoms (Kohel, 1980, 1998). Cottonseed samples were prepared by acid-delinting whole fuzzy seeds, drying under forced air at 38 °C for 24 h and equilibrating to room temperature in a glass desiccator prior to taking measurements. The sample size was approximately 10 g, and the NMR integration period was approximately 30 s. Commercially available 27% and 51% cottonseed oil samples were used as controls. The average of two readings for each sample was used as the final measure of oil content.

Fibre quality

Seed cotton was ginned using a roller gin, and fibre properties for each sample were determined using a USTER® HVI line by Cotton Incorporated (Cary, NC). The fibre quality parameters examined were micronaire (MIC, a measure of fibre fineness and maturity), upper half mean length (UHM, average length of the longer one-half of the fibres), uniformity index (UI, the ratio of the mean length to UHM), strength (STR, force required to break a bundle of fibres one tex unit in size) and elongation (ELO, degree of extension of the fibres before a break occurs when measuring strength).

Statistical analysis

The data that showed certain clear-cut trends were subjected to statistical analysis using one-way ANOVA in Microsoft Excel 2007 spreadsheets. Data are presented as mean ± SE, and < 0.05 was considered as statistically significant.


We are thankful to Jim Balthrop (Office of Texas State Chemist) for his help with seed protein analysis and Ms. Shanna Sherwood for her help with ginning. This research was supported by funds from Cotton Incorporated, Texas Cotton Biotechnology Initiative (TxCOT) and Texas AgriLife Research. Seed composition analysis was performed by Covance Laboratories, Inc., Madison, WI, USA. The authors declare no conflict of interest.