Plant Biotechnology Journal (2014) 12, pp. 685–693

doi: 10.1111/pbi.12172

Transgenic alteration of ethylene biosynthesis increases grain yield in maize under field drought-stress conditions Jeffrey E. Habben*, Xiaoming Bao†, Nicholas J. Bate‡, Jason L. DeBruin, Dennis Dolan, Darren Hasegawa, Timothy G. Helentjaris§, Renee H. Lafitte, Nina Lovan, Hua Mo, Kellie Reimann and Jeffrey R. Schussler DuPont Pioneer, Johnston, IA, USA

Received 27 September 2013; revised 11 December 2013; accepted 22 December 2013. *Correspondence (Tel 1(515)535-4130; fax 1(515)535-4788; email [email protected]) † Present address: DBN Biotechnology Center, P.O. Box 5109, No 2 Yuanmingyuan West Rd., Haidian Dist., Beijing 100193, China. ‡ Present address: Syngenta Biotechnology, Inc., 3054 East Cornwallis Road, P.O. Box 12257, Research Triangle Park, NC 27709-2257, USA. § Present address: 5525 N Desert Wood Place, Tucson, AZ 85745, USA.

Keywords: ethylene, maize, grain yield, drought tolerance, ACC

Summary A transgenic gene-silencing approach was used to modulate the levels of ethylene biosynthesis in maize (Zea mays L.) and determine its effect on grain yield under drought stress in a comprehensive set of field trials. Commercially relevant transgenic events were created with down-regulated ACC synthases (ACSs), enzymes that catalyse the rate-limiting step in ethylene biosynthesis. These events had ethylene emission levels reduced approximately 50% compared with nontransgenic nulls. Multiple, independent transgenic hybrids and controls were tested in field trials at managed drought-stress and rain-fed locations throughout the US. Analysis of yield data indicated that transgenic events had significantly increased grain yield over the null comparators, with the best event having a 0.58 Mg/ha (9.3 bushel/acre) increase after a flowering period drought stress. A (genotype 9 transgene) 9 environment interaction existed among the events, highlighting the need to better understand the context in which the downregulation of ACSs functions in maize. Analysis of secondary traits showed that there was a consistent decrease in the anthesis-silking interval and a concomitant increase in kernel number/ ear in transgene-positive events versus nulls. Selected events were also field tested under a lownitrogen treatment, and the best event was found to have a significant 0.44 Mg/ha (7.1 bushel/ acre) yield increase. This set of extensive field evaluations demonstrated that down-regulating the ethylene biosynthetic pathway can improve the grain yield of maize under abiotic stress conditions.

synthase, phytohormone.

Introduction Water is the most limiting resource for crop production (Boyer, 1982); consequently, there is considerable interest in understanding drought tolerance and creating drought-tolerant germplasm. Drought is a complex trait in that the timing and degree of water limitation produce a multitude of stress-induced responses. In maize, a drought stress that occurs during the vegetative phase of growth and development can reduce plant height, decrease leaf elongation and trigger leaf wilting. If the stress continues unabated, leaf senescence ensues followed by leaf death. Drought stress occurring during the reproductive phase can cause pollen sterility, tassel necrosis, delayed silk exsertion and kernel abortion. All of these reactions can contribute to a reduction in grain yield and likely evolved as conservative, adaptive mechanisms by the plant. However, these mechanisms are less than ideal for obtaining maximum grain yield in modern production agriculture. The basic premise is that maize is too conservative in its overall response to drought for the objective of maintaining high yield in agricultural environments and that modulating this conservatism could lead to an increase in grain yield. Phytohormones are potent molecules that control diverse plant phenotypes. One of these, ethylene, has been demonstrated to regulate many different aspects of growth and development, particularly under abiotic stresses. Ethylene is a gas that is synthesized in almost all plant tissues in the presence of oxygen (Lin et al., 2010). Methionine is the starting point in the ethylene

biosynthetic pathway, and it is converted into S-adenosylmethionine (SAM) by methionine adenosyltransferases (Yang and Hoffman, 1984). ACC synthase (ACS) catalyses the conversion of SAM into 1-aminocyclopropane-1-carboxylic acid (ACC), the first committed step of ethylene biosynthesis. ACC is then converted by ACC oxidase into ethylene with ACC oxidase activity largely considered as constitutive in plants (Yang and Hoffman, 1984). Ethylene becomes the effector molecule that triggers subsequent reactions. Signalling is initiated via the interaction between the ethylene ligand and its receptors localized in the endoplasmic reticulum (Lin et al., 2010). This binding shuts down receptor signalling, releasing the pathway from inhibition and setting forth a cascade of downstream cellular actions (Alonso and Ecker, 2001; Bleecker and Kende, 2000; Klee, 2002). The ethylene pathway has been linked to many diverse physiological processes in vegetative organs of maize. One of the best characterized is the role of ethylene in aerenchyma formation in roots under water-logging conditions. Seedling roots exposed to hypoxia exhibit an increased rate of ethylene evolution, greater activities of ACS and ACC oxidase relative to controls and consequently greater programmed cell death that manifests itself as aerenchyma (He et al., 1996). A drought study with maize seedlings demonstrated there was an inverse relationship between endogenous ethylene levels and root
n et al., 2009) and that elongation was elongation (Alarco inhibited when roots were exposed to ACC. In another example,

ª 2014 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd

685

686 Jeffrey E. Habben et al. Young et al. (2004) generated knockouts of ACSs in maize and documented that mutants had a reduction in ethylene emission as well as an inhibition in drought-induced senescence in older leaves. In maize, as well as other cereals, ethylene has also been linked to numerous facets of growth and development in reproductive organs. Similar to the response shown in flooded roots (He et al., 1996), elevated ethylene levels were coupled to the triggering of programmed cell death in endosperms of developing kernels (Young et al., 1997). Under water-limiting conditions, kernel abortion in maize typically occurs at the ear tip. In a study by Feng et al. (2011), ethylene production in kernels that ultimately aborted at the ear tip declined more slowly and was maintained at a higher level than that of kernels on the rest of the ear. An in vitro study with maize kernels demonstrated that adding ACC to cultured kernels caused them to abort and those that remained viable had reduced mass (Hanft et al., 1990). Application of ACC to developing kernels of field-grown plants reduced the mass of apical kernels that led the authors to conclude that ethylene was involved in kernel abortion (Cheng and Lur, 1996). Cox and Andrade (1988) determined that application of ethephon (a chemical that increases ethylene evolution) to field-grown maize hybrids caused a reduction in kernel number/ear. In small grain cereals, a related set of studies has shown similar ethylene functionality. In wheat (Triticum aestivum L.), a heat stress was applied during reproductive development to heat-tolerant and heat-susceptible cultivars (Hays et al., 2007). This stress caused a 6-, 7- and 12-fold change in ethylene emission in developing kernels, embryos and the flag leaf, respectively, in the heatsusceptible cultivar, but no change in ethylene levels in the heattolerant one. Yang et al. (2006) exposed developing rice (Oryza sativa L.) kernels to an ethylene inhibitor and measured an increase in cell division rate, maximum cell number, grain-filling rate and grain mass of inferior spikelets. In rice endosperms, cell division rate and starch concentration were negatively correlated with ethylene levels (Panda et al., 2009). From these studies, as well as others (Beltrano et al., 1994, 1999; Mohapatra et al., 2009; Yang et al., 2007; Zhang et al., 2009), it is apparent that ethylene physiology plays an important role in the decreased yield of cereals grown under abiotic stress conditions (Wilkinson et al., 2012). We are interested in ethylene as it relates to improving drought tolerance in maize because of the numerous pharmacological and morphometric studies that have associated this hormone with grain yield stability. Additionally, we are interested in this phytohormone because its biosynthesis, catabolism and signalling have been well characterized at the molecular level, and consequently, the pathway is readily amenable to transgenic modification in plants. As the conversion of SAM to ACC is the rate-limiting step in production of ethylene (Woeste et al., 1999), it follows that ACSs would be a focal point for control of ethylene biosynthesis. The objective of this research was to use a transgenic approach to reduce the synthesis of ethylene in maize via expression of an ACS RNA interference construct and to determine its subsequent effect on plant performance in abiotic stress environments.

construct was transformed into maize, and numerous events with single-gene insertions were created, and from this set, nine independent events were selected for further testing. Subsequent analysis of the maize genome indicated the presence of a putative unannotated ACS gene (GI: 414875885) that we named ZM-ACS3. The overall per cent homology between ZM-ACS3 and ZM-ACS6 is 62.9%. Within the hairpin sequence, there was a continuous 44 bp region that aligns with ZM-ACS6 as well as ZM-ACS3 (Figure S2), making it plausible that the hairpin construct would affect not only ZM-ACS6, but also ZM-ACS3 transcripts. Therefore, we analysed transcript abundance via quantitative real-time PCR (qRT-PCR) of both genes and found that their transcript levels were significantly decreased relative to wild type (WT) (Figure S3). To determine the impact of ACS silencing on ethylene biosynthesis, we analysed the ethylene emission of field-grown plants from three selected events (DPE29, DP-E21 and DP-E12) plus WT. Overall, there was a 53% decrease in ethylene emission from leaves of transgene-positive plants relative to WT, demonstrating the effectiveness of the ACS hairpin construct in decreasing ethylene biosynthesis in leaves. In a separate experiment, ethylene levels were determined in ear spikelets, and a similar reduction in ethylene emission in transgenic events was measured relative to WT (data not shown). When ethylene emission levels were measured from leaves of plants grown either in a low or high environmental stress condition (see Experimental procedures), the transgenic events’ average ethylene emission was decreased 57% in the low-stress environment class and reduced 49% in the high-stress environment class, relative to WT. DP-E12 showed the greatest decrease in ethylene emission (55%) in the high-stress environment class (Figure 1), while DP-E29 and DP-E21 were more similar in their response (decreased 49% and 44%, respectively).

Grain yield of ACS6 RNAi events under drought conditions The effect of silencing ACSs in transgenic maize hybrids was evaluated in an extensive set of field trials conducted over a 2-year period. Our overarching objective was to create transgenic

Results Ethylene emission in ACS6 RNAi events To silence ACSs in maize, a maize ubiquitin promoter was fused to a ZM-ACS6 hairpin that consisted of a fragment of the ZM-ACS6 sequence in an inverted repeat structure with the maize ADH1 Intron1 as the loop sequence of the hairpin (Figure S1). This

Figure 1 Per cent ethylene emission from leaves of wild-type (WT) and three ACS6 hairpin events (DP-E12, DP-E21, DP-E29) in a high-stress environment (six locations) and low-stress environment (five locations) across nine hybrids in 2011. The ethylene level was standardized to 100. Error bars represent the SEM (N = 144 for high-stress environment, N = 63 for low-stress environment).

ª 2014 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd, Plant Biotechnology Journal, 12, 685–693

Altering ethylene biosynthesis increases maize yield 687 events that increase grain yield under drought-stress conditions and, at a minimum, had no yield penalty under well-watered conditions. Initially, the nine events were evaluated in nine unique elite hybrids in 2010. In all cases, the nontransformed WT hybrid was used as the null comparator. In Table 1, yield data from a managed drought-stress location as well as from a high-yield location are presented. The drought-stress location was near Woodland, CA, where rainfall during the growing season was 500 mm (20 in)]. The average yield of WT hybrids was 11.34 Mg/ha (180.5 bu/ac) (Table 1), which is typical for this region of the US. None of the events altered yield compared with WT, indicating that the transgenic construct did not impose a yield penalty at this high-yield site. When these entries were broadly tested across the US Corn Belt, and a combined analysis performed on the yield data (Table 2), more variable performance was observed relative to the managed drought-stress location in CA (Table 1). Only event DP-E29 showed a statistically positive impact over all locations and all hybrids, and two events (DP-E12 and DP-E17) caused a significant grain yield reduction compared with WT. The modest yield effects of the transgenic construct measured in this analysis indicate that a considerable (genotype 9 transgene) 9 environment interaction was present. To determine the reproducibility of these yield results, field trials were conducted again in 2011 but with only three of the events: DP-E12, DP-E21 and DP-E29. Yield trials in 2011 were conducted with a different set of nine elite hybrids across 12 locations. After harvest, each location was analysed and placed into either a low-stress or high-stress environment class using the proprietary EnClass environmental classification system (Loffler et al., 2005). Those locations where mean yields were reduced 30%–70% due to documented plant water deficits were included in the high-stress environment class, while the remaining locations were placed in the low-stress environment class. The mean yield of the WT entries in the lowstress environment class was 12.77 Mg/ha (203.3 bu/ac) (Table 3). The mean yield in the high-stress environment class was 7.31 Mg/ha (116.3 bu/ac), a 43% reduction in yield driven by drought stress. In the low-stress environment class, no significant yield penalty was observed with either DP-E21 or DP-E29, while DP-E12 caused a yield reduction of 0.24 Mg/ha

DP-E12

6.85 (108.9)

0.17 (2.8)

27

DP-E15

6.95 (110.6)

0.28 (4.4)*

26

DP-E17

7.12 (113.3)

0.44 (7.1)*

24

DP-E18

6.84 (108.9)

0.17 (2.7)

26

DP-E21

7.07 (112.5)

0.39 (6.3)*

27

DP-E26

7.02 (111.8)

0.35 (5.6)*

18

DP-E28

7.04 (112.0)

0.36 (5.8)*

27

DP-E29

7.26 (115.5)

0.58 (9.3)*

27

DP-E43

7.14 (113.7)

0.47 (7.5)*

18

Yield prediction

Predicted difference

Wild type

6.67 (106.2)

–

54

Entry

Mg/ha (Bu/ac)

Mg/ha (Bu/ac)

DP-E12

11.40 (181.5)

0.06 (1.0)

14

DP-E12

8.47 (134.8)

0.11 ( 1.8)*

475

DP-E15

11.44 ( 182.1)

0.10 (1.6)

14

DP-E15

8.54 (136.0)

0.04 ( 0.6)

471

DP-E17

11.34 (180.5)

0.00 (0.0)

12

DP-E17

8.51 (135.4)

0.07 ( 1.1)*

413

DP-E18

11.40 (181.5)

0.06 (1.0)

11

DP-E18

8.54 (135.9)

0.04 ( 0.7)

459

DP-E21

11.50 (183.0)

0.16 (2.5)

12

DP-E21

8.63 (137.4)

0.05 (0.8)

470

DP-E26

11.32 (180.2)

0.02 ( 0.3)

11

DP-E26

8.51 (135.5)

0.07 ( 1.1)

317

DP-E28

11.44 (182.2)

0.10 (1.7)

10

DP-E28

8.55 (136.1)

0.03 ( 0.4)

456

DP-E29

11.37 (181.0)

0.03 (0.5)

14

DP-E29

8.65 (137.7)

0.07 (1.2)*

466

DP-E43

11.40 (181.5)

0.06 (0.9)

8

DP-E43

8.59 (136.7)

0.01 (0.1)

308

Wild type

11.34 (180.5)

–

27

Wild type

8.58 (136.6)

–

962

Table 2 Grain yield of ACS6 RNAi events and wild type at all locations

N

High-yield location

N, number of hybrid 9 replication data points included in the statistical

N, number of hybrid 9 replication 9 location data points included in the

analysis.

statistical analysis.

Data are from nine individual transgenic maize events and wild type pooled

Data are from nine individual transgenic maize events and wild type pooled

across nine elite hybrids at a drought flowering-stress and high-yield location in

across nine elite hybrids and across 22 locations in 2010. Predicted difference

2010. Predicted difference for each event is compared with the wild type. All

for each event is compared with the wild type. All analyses were implemented

analyses were implemented using ASReml with output of the model presented

using ASReml with output of the model presented as best linear unbiased

as best linear unbiased predictions (see Experimental procedures).

predictions (see Experimental procedures).

*Predicted difference significant at P < 0.05.

*Predicted difference significant at P < 0.05.

ª 2014 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd, Plant Biotechnology Journal, 12, 685–693

688 Jeffrey E. Habben et al. (3.8 bu/ac) (Table 3). In the high-stress environment, DP-E29 and DP-E21 significantly increased yield 0.16 Mg/ha (2.5 bu/ac) and 0.17 Mg/ha (2.7 bu/ac), respectively, while DP-E12 had no significant effect. Further dissection of the DP-E29 yield data showed that the nine hybrids in the high-stress environment class ranged in predicted yield difference from a low of 0.02 Mg/ha (0.4 bu/ac) to a high of 0.32 Mg/ha (5.0 bu/ac). Thus, over the 2-year period, DP-E29 and DP-E21 repeatedly demonstrated a positive yield response compared with WT under drought stress with no yield penalty under normal conditions (Tables 1 and 3); in contrast, DP-E12 exhibited a neutral yield response under drought stress and either a neutral or negative yield response under normal conditions (Tables 1 and 3).

Anthesis-silking interval and kernel set of ACS6 RNAi events under drought conditions Data on multiple secondary traits were collected to better understand the differential yield response of the events. The anthesis-silking interval (ASI) in maize under typical well-watered conditions exhibits minimal asynchrony between the time pollen sheds from tassels to the time silks exsert from ear husks. However, drought stress can often delay silk exsertion (increase in ASI), which leads to a decrease in pollination of ovaries (Campos et al., 2004). When pooled across low-stress and highstress environment classes, ASI was reduced (i.e. more synchronous shedding/silking) in DP-E29 and DP-E21 relative to the WT, but was unaffected in DP-E12, indicating enhanced silk growth in the positive yielding events (Table 4). In addition to ASI, kernel set also showed a significant change. Overall, the kernels/ ear yield component was strongly impacted in the high-stress environment class, where a WT mean of 310 kernels/ear was measured, in contrast to a WT average of 564 kernels/ear in the low-stress environment class (Table 4). DP-E29 and DP-E21 increased kernels/ear by an average of 15.1 and 13.4, respectively, compared with the WT in the high-stress environment class, while DP-E12 showed no statistically significant increase. Table 3 Grain yield of ACS6 RNAi events and wild type in two environments

Entry

Yield prediction

Predicted difference

Mg/ha (Bu/ac)

Mg/ha (Bu/ac )

N

Entry

ASI prediction

Predicted difference

GDU °C

GDU °C

N

Pooled across low- and high-stress environments DP-E29

66.2

3.6*

264

DP-E21

67.6

2.3*

257

DP-E12

68.3

1.5

263

Wild type

69.9

–

258

Entry

Kernels/ear

Predicted difference

prediction

Kernels/ear

N

Low-stress environment class DP-E29

579.1

15.0*

217

DP-E21

578.4

14.4*

213

DP-E12

576.9

12.9*

215

Wild type

564.0

–

209

High-stress environment class DP-E29

325.1

15.1*

217

DP-E21

323.5

13.4*

213

DP-E12

318.1

8.1

215

Wild type

310.0

–

209

GDU, growing degree units. N, number of hybrid 9 replication 9 location data points included in the statistical analysis. Data are from three individual transgenic maize events and wild type pooled across nine elite hybrids in either pooled or two environment classes in 2011. Predicted difference for each event is compared with the wild type. All analyses were implemented using ASReml with output of the model presented as best linear unbiased predictions (see Experimental procedures). *Predicted difference significant at P < 0.05.

All three events demonstrated an increase in kernels/ear in the low-stress environment class compared with WT. Taking the secondary trait data as a whole, events with increased grain yield exhibited a reduction in ASI and a concomitant increase in kernels/ear.

Grain yield of ACS6 RNAi events under low-N conditions

Low-stress environment class (six locations) DP-E29

12.65 (201.4)

0.13 ( 2.0)

314

DP-E21

12.68 (201.8)

0.10 ( 1.6)

275

DP-E12

12.54 (199.6)

0.24 ( 3.8)*

305

Wild type

12.77 (203.3)

–

303

High-stress environment class (six locations) DP-E29

7.46 (118.8)

0.16 (2.5)*

314

DP-E21

7.48 (119.0)

0.17 (2.7)*

275

DP-E12

7.34 (116.9)

0.04 (0.6)

Wild type

7.31 (116.3)

–

305 303

N, number of hybrid 9 replication data points included in the statistical analysis. Data are from three individual transgenic maize events and wild type pooled across nine elite hybrids in two environment classes in 2011. Predicted difference for each event is compared with the wild type. All analyses were implemented using ASReml with output of the model presented as best linear unbiased predictions (see Experimental procedures). *Predicted difference significant at P < 0.05.

Table 4 Anthesis-silking interval (ASI) (top) and kernels ear (bottom) of ACS6 RNAi events and wild type in two environments

As reproductive growth was enhanced in the transgenic ACS6 RNAi events under drought-stress conditions, testing was also conducted to determine whether events could enhance grain yield under an alternative abiotic stress, low nitrogen. DP-E29 and DP-E21 were tested in 2011 in nine hybrids at three locations where only 78 kg/ha (70 lb/ac) of N was applied to plots. In-season precipitation was normal for each of the three locations. The average yield of the hybrids across the two events at these locations was 8.7 Mg/ha (139 bu/ac) compared with an average yield of around 12.5 Mg/ha (200 bu/ac) at these locations when normal N amounts (224 kg/ha) (200 lb/ac) were applied to trials. Multiple hybrids showed a statistically significant increase in yield relative to WT under low-N conditions with the best event 9 hybrid combination (Hybrid 1, DP-E29) showing an increase of 0.44 Mg/ha (7.1 bu/ac) (Table 5). This result indicates that expression of the ACS6 hairpin vector can enhance maize hybrid yield not only under drought-stress conditions, but also under limited-N conditions.

ª 2014 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd, Plant Biotechnology Journal, 12, 685–693

Altering ethylene biosynthesis increases maize yield 689

Discussion Our hypothesis is that maize is too conservative in its response to drought stress in modern production agriculture, in that it overreacts to limiting water conditions by restricting female reproductive growth, in favour of plant survival. This response is possibly a vestige of the monoecious nature of maize, but is counter-productive to its use in commercial grain production. It has been proposed that this evolutionarily derived strategy is something that breeders continually confront and endeavour to counteract (Skirycz and Inz
e, 2010; Tardieu, 2012). Here, we used a transgenic approach to offset this inherent survival response and showed that by manipulating ethylene biosynthesis we can stabilize female maize reproductive growth and significantly increase grain yield under abiotic stresses. Grain yield is the culmination of a cascade of molecular, biochemical and physiological processes that occur throughout the growth and development of the crop; thus, deconvoluting the source of a yield increase is challenging. As a starting point, we measured and analysed numerous secondary traits. Under drought stress, the most consistent secondary trait changes were improvements in ASI and kernels/ear (Table 4), which is advantageous as drought typically delays silk emergence from husks resulting in unpollinated ovaries and reduced kernel set. By manipulating the ethylene pathway, we have improved silking dynamics such that more ovules are pollinated, resulting in an improved kernel set, up to 15 kernels/ear (DP-E29, Table 4). As kernel row number was not changed in transgenic events, essentially one extra ring of harvestable kernels was added per ear, and at a typical Corn Belt plant population (80 000 plants/

ha) (32 000 plants/ac) each extra ring of kernels can theoretically increase grain yield approximately 0.3 Mg/ha (5 bu/ac). Thus, small increases in kernel set can potentially translate into significant overall grain yield increases. From a molecular perspective, we show that constitutive over expression of the ACS6 hairpin consistently decreased ZM-ACS6 and ZM-ACS3 transcript levels as well as reduced ethylene emission levels in transgenic events relative to WT. Several studies have focused on the role of ethylene in influencing maize reproductive growth (Cheng and Lur, 1996; Cox and Andrade, 1988; Hanft et al., 1990). All of these reports are consistent with the hypothesis that decreasing the biosynthesis of ethylene can lead to enhanced maize productivity. However, it is too simplistic to directly relate ethylene suppression to a positive yield response. As an example, in the low-stress environment class, DP-E29 is yield neutral, while DP-E12 is yield negative (Table 3). In the highstress environment class, DP-E12 is yield neutral, while that of DP-E29 is yield positive. Examination of transgenic event ethylene levels shows that in the low-stress environment, DP-E29 has lower ethylene levels than DP-E12, but in the high-stress environment this difference is reversed (Figure 1). This outcome is not unexpected given that aspects of plant growth and development are regulated not just by ethylene biosynthesis, but also by ethylene perception and subsequent downstream signalling (Pierik et al., 2007), as well as by the interaction of ethylene with other hormones (Wang and Irving, 2011). In production agriculture, transgenic events would need to perform under diverse abiotic stress conditions. To this end, the ACS6 hairpin events were tested in a low-N treatment that was approximately threefold less than that of typical application levels

Table 5 Grain yield of ACS6 RNAi events and wild type under low-N conditions Event yield prediction

Wild-type yield prediction

Predicted difference

Mg/ha (Bu/ac)

Mg/ha (Bu/ac)

Mg/ha (Bu/ac)

Hybrid 1

9.15 (145.7)

8.71 (138.6)

0.44 (7.1)*

12

Hybrid 2

9.61 (152.9)

9.38 (149.3)

0.23 (3.6)

12

Hybrid 3

8.64 (137.5)

8.36 (133.1)

0.27 (4.4)*

12

Hybrid 4

8.12 (129.2)

7.94 (126.4)

0.18 (2.8)

11

Hybrid 5

9.06 (144.3)

8.86 (141.0)

0.21 (3.3)

11

Hybrid 6

8.57 (136.4)

8.26 (131.4)

0.31 (4.9)*

12

Hybrid 7

8.23 (131.0)

8.18 (130.2)

0.05 (0.8)

11

Hybrid 8

8.86 (141.1)

8.65 (137.7)

0.21 (3.3)

12

Hybrid 9

8.55 (136.1)

8.17 (130.0)

0.38 (6.1)*

8

Hybrid 1

9.08 (144.6)

8.71 (138.6)

0.37 (6.0)*

12

Hybrid 2

9.54 (151.9)

9.38 (149.3)

0.16 (2.5)

12

Hybrid 3

8.70 (138.5)

8.36 (133.1)

0.34 (5.4)*

12

Hybrid 4

8.00 (127.4)

7.94 (126.4)

0.06 (0.9)

7

Hybrid 5

9.10 (144.8)

8.86 (141.0)

0.24 (3.9)

10

Hybrid 6

8.66 (137.9)

8.26 (131.4)

0.41 (6.5)*

12

Hybrid 7

8.29 (132.0)

8.18 (130.2)

0.11 (1.8)

11

Hybrid 8

8.93 (142.2)

8.65 (137.7)

0.28 (4.5)*

12

Hybrid 9

8.19 (130.4)

8.17 (130.0)

0.02 (0.3)

6

Entry

N

DP-E29

DP-E21

N, number of hybrid 9 replication 9 location data points included in the statistical analysis. Data are from two individual transgenic maize events (DP-E29 and DP-E21) and wild type in nine elite hybrids under low-N application (78 kg/ha) (70 lb/ac) conditions across three locations in 2011. Predicted difference for each event is compared with the wild type. All analyses were implemented using ASReml with output of the model presented as best linear unbiased predictions (see Experimental procedures). *Predicted difference significant at P < 0.05.

ª 2014 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd, Plant Biotechnology Journal, 12, 685–693

690 Jeffrey E. Habben et al. in the Corn Belt, with the result being a significant increase in grain yield in transgenic events relative to the comparator (Table 5). There was no drought stress in these low-N locations, as confirmed by our EnClass environmental characterization system (see Experimental procedures), and thus, the yield increase was not due to the improved drought tolerance of transgenic events. Similar to what we observed under water-limiting conditions, the increase in grain yield was driven by an increase in kernels/ear (Table S1) and may also be impacted by a reduction in ASI (Table S2). Previous studies from CIMMYT demonstrated that nontransgenic maize selected for enhanced drought tolerance had improved performance under low-N conditions (B€ anziger et al., 1999, 2002) and that ASI was one of the secondary traits that was improved under this situation (B€ anziger and Lafitte, 1997; Zaidi et al., 2004). While informative, these studies were conducted with tropical germplasm, which has not been as intensively selected for grain yield as temperate commercial hybrids. Here, we demonstrate that elite ACS6 RNAi transgenic events that yield better under drought stress can also yield better in low-N environments. Importantly, the cross-validation of improved yield in transgenic events propagated under low N verifies the ethylene pathway as a valuable target for mitigation of abiotic stresses in maize. From a commercial perspective, transgenes have been successfully integrated and used in row crops for herbicide tolerance and insect resistance (James, 2011; Shi et al., 2013). However, transgenic events that directly enhance yield under drought stress have been less forthcoming. This, in part, could be a result of the use of drought screens that are too focused on plant survival traits, in lieu of drought field screens targeted to plant productivity (Belimov et al., 2009). Nevertheless, the literature is replete with publications declaring drought-tolerant transgene efficacy (Lawlor, 2013). One major problem is that the definition of drought tolerance is highly context dependent. To the plant breeder, validation of improved drought tolerance is typically increased field yield stability. The breeder definition of drought tolerance is our definition, and there are only a small number of publications that report improved transgenic crop yields in field trials conducted under drought stress (Castiglioni et al., 2008; Nelson et al., 2007; Qin et al., 2011; Wang et al., 2005). Because of the dearth of such publications, critics have questioned whether or not transgenes can improve a quantitative trait like grain yield under field drought-stress conditions (e.g. Lawlor, 2013). To our knowledge, we describe the most extensive field drought study conducted on a transgenic crop reported in the literature. We show that in multiple genetic backgrounds and in multiple environments, expression of an ACS6 RNAi construct can significantly increase maize grain yield. Moreover, these data were collected in elite, commercially relevant maize hybrids, and not in experimental, nonelite germplasm. This hairpin construct provides biological efficacy, measured as reduced ethylene emission, alteration of key secondary traits and enhanced grain yield. Given this success, we continue to probe the ethylene pathway as a means to develop commercially viable, droughttolerant germplasm.

Experimental procedures Vector construction, plant transformation and transgene expression analysis To construct the ZM-ACS6 double-stranded RNAi plasmid for expression in maize, a 487-bp fragment was amplified from

genomic DNA (GenBank AY359571). The ZM-ACS6 (TR) fragments were PCR-amplified with BamHI and PstI sites [ZM-ACS6 (TR3)] as well as AgeI and SacI sites [ZM-ACS6 (TR4)]; the fragments were cloned into Invitrogen’s (Carlsbad, CA) pCR4Blunt cloning vector and sequence verified. The two ZM-ACS6 fragments were then ligated with a 537-bp PstI-XmaI (AgeI compatible) ZM-ADH1 Intron1 fragment and a BamHI-SacI Gateway (Invitrogen) compatible backbone vector containing the ZM-UBIQUITIN promoter (Christensen et al., 1992), the ZM-UBIQUITIN 5′ UTR and the ZM-UBIQUITIN INTRON1. The Gateway entry vector was recombined into an intermediate destination vector containing the MOPAT selectable marker using Invitrogen’s LR Clonase II Plus enzyme mix. The ZM-UBIQUITIN PROMOTER: ZM-UBIQUITIN 5′ UTR: ZM-UBIQUITIN INTRON1: ZM-ACS6 hairpin + ZM-UBIQUITIN PROMOTER: ZM-UBIQUITIN 5′ UTR: ZM-UBIQUITIN INTRON1: MOPAT: PINII TERMINATOR intermediate plasmid was then cointegrated into a pSB1 vector in Agrobacterium tumefaciens strain LBA4404. The Agrobacterium containing the binary vector was then used for maize transformation (Unger et al., 2001), which was conducted with a proprietary inbred. Multiple glufosinate-resistant T0 events were grown in a greenhouse and T1 seed collected for subsequent propagation. Genomic DNA was extracted from leaves of transgenic maize plants and checked by PCR for insert intactness. Only events with single-gene insertions were advanced. To magnify expression differences, we stressed seedlings by flooding them and determined levels of ZM-ACS6 and ZM-ACS3 transcripts. F1 transgene-positive and WT seed were germinated, grown for 8 days under normal conditions and then exposed to a flooding treatment for 30 h. Total RNA was isolated from leaf tissue following a protocol modified from Verwoerd et al. (1989), and 2 lg was used as template to synthesize cDNA using Qiagen’s (Valencia, CA) QuantiTect Reverse Transcription Kit. qRT-PCR experiments were performed in 50 lL reactions using 5 lL of 1 : 5 (v/v) dilution of first-strand cDNA, 25 lL TaqMan Universal Master Mix (Applied Biosystems, Carlsbad, CA), 300 nM forward and reverse primers, as well as 100 nM TaqMan probe. In the reaction plate, all samples were analysed in triplicate with the ABI 7500 System (Applied Biosystems) using the following step-cycle program: 2 min at 50 °C, 10 min at 95 °C, 40 cycles of 15 s at 95 °C and 1 min at 60 °C. Relative quantitation values were determined using the difference in Ct from the target gene and ZM-UBQ5 (internal control). Primers employed were ZM-ACS6 Forward Primer (PHN143189)—TCCGCTTCAGCTGGGCTAA; ZM-ACS6 Reverse Primer (PHN143191)—ACTGGTGGTGAGCTGTGAAGCT; ZM-ACS6 FAM Probe (PHN143473)—CGACCGGAAGGCCGAGCG; ZM-ACS3 Forward Primer (PHN144755)—ACGAGCTGC TCACGTTCGT; ZM-ACS3 Reverse Primer (PHN144756)— CCAGGATAGTAAGGAGTAGGGATCAG; ZM-ACS3 FAM NFQMGB Probe (PHN144757)—AACCCGGGAGACGC; ZM-UBQ5 Forward Primer (endogenous control) (PHN133881)—CCACTTC GACCGCCACTACT; ZM-UBQ5 Reverse Primer (PHN133882)— CGCCTGCTGGTTGTAGACGTA; ZM-UBQ5 VIC NFQ-MGB Probe (PHN136465)—CGGTAAGTGTGGCCTC.

Hybrid yield testing To evaluate the yield potential of the transgenic events, field trials were conducted with elite commercial hybrids over a range of environments in the US in 2010 and 2011. Using standard backcrossing techniques, the insertion was backcrossed from a donor inbred line into two commercial inbred lines, one in each of the two major complimentary heterotic groups. Using herbicide markers, positive individuals were advanced to subsequent

ª 2014 Society for Experimental Biology, Association of Applied Biologists and John Wiley & Sons Ltd, Plant Biotechnology Journal, 12, 685–693

Altering ethylene biosynthesis increases maize yield 691 backcrossing generations. Both genotypic and phenotypic selections were used to identify positive plants and lines with minimal donor contribution. Homozygous-positive BC4F2 and BC4F4 plants were used for 2010 and 2011 hybrid seed production, respectively. Nine proprietary inbred lines were used as testers, six with one conversion and three with a second conversion. In the same nursery where the topcross transgenic F1 seed was produced, the WT version of the same hybrid was created by crossing the same two elite inbreds not containing the transgenic construct. All subsequent yield comparisons were made between the F1 transgene-positive hybrid and the F1 WT. In some cases, other transgenic deregulated traits including herbicide tolerance and/or insect protection were already included in these hybrids. When this occurred, the same deregulated genes were present in both the experimental and WT entries of a hybrid. Multiple individual events were backcrossed to determine effect of insertion site on efficacy. Experimental and wild-type hybrid pairs representing nine independent events were grown in field environments at research centres and/or grower fields in Woodland, CA; Garden City, KS; Manhattan, KS; La Salle, CO; Firth, NE; Keosauqua, IA; West Bend, IA; Johnston, IA; Opolis, MO; Seymour, IL; Farmington, IL; Streator, IL; Sciota, IL; Fort Branch, IN; and Edmonson, TX. Yield studies were set up as a randomized complete block design in a split-plot arrangement with hybrids as main plots and transgene status (positive or WT) as the split plot. One to three replicates were established at each location, and the plant population density used was typical for growers in that particular region. In addition, in 2011, a limited-N yield trial was also grown at Sciota, IL; Johnston, IA and York, NE. In this trial, four replications were established at each location where only 78 kg/ha (70 lb/ac) of N was applied. Plots in all studies were four rows wide, with rows spaced at 76 cm (30 in) and plot length ranged from 4.3 to 5.4 m (14.1–17.8 ft). Grain yields were recorded using a research combining harvesting ears from the two centre rows of each plot. This ensured that harvested plants were competing against themselves and not against hybrids of differential plant stature. Harvest weight and grain moisture from each plot were used to calculate yield/area at a constant moisture. The 2010 and 2011 yield experiments were conducted across a range of environments, including dryland, limited irrigation and fully irrigated grower fields (target population of environments, or TPEs) as well as managed stressed environments (MSEs) where subsurface drip irrigation was used to impose precise intensities of plant water deficits at flowering time. In the MSEs, drip irrigation was typically withheld 2–5 weeks prior to anthesis to impose a water-limiting treatment during the critical flowering period. Plant available water was depleted to the level where significant leaf rolling was observed daily for at least 1–2 weeks prior to flowering, and continuing through flowering. Plant height in the flowering stress treatment was generally reduced 10%–25%, compared with full irrigation treatments, and some leaf senescence due to prolonged drought stress was observed in plants in the flowering stress treatment at the time pollen shed and silking occurred. At approximately 2 weeks after flowering, irrigation was provided to rehydrate the plants, terminating the stress treatment. Irrigation was continued for the remainder of the growth cycle to provide adequate water for plants to fill the kernels established during the flowering stress treatment. To determine the efficacy of transgenic events over a variety of stress and nonstress environments, at both MSE and TPE locations, soil moisture down to 1.5 m was monitored via

capacitance probes (AquaSpy, San Diego, CA) to document soil water depletion within the root zone during the development of the crop. These data were combined with meteorological data recorded on-site (temperature, precipitation, light irradiation) and inputted into the EnClass environmental characterization system to document the timing of drought stress at each location (2011). This model outputs a daily water supply–demand ratio throughout the entire growth cycle of the crop. This ratio predicts the daily plant water status of maize based on demand for water use (evapotranspiration in mm/day) vs. remaining soil moisture in the effective root zone of the crop (mm/M). In 2011, all locations were placed into either a high-stress or low-stress environmental class. Locations were binned into the high-stress environmental class when a low water supply– demand ratio was observed during flowering and/or grain fill, indicating water as a significant limiting factor for yield. Final yield at these high-stress locations was