RNA interference suppression of AGAMOUS and SEEDSTICK alters floral organ identity and impairs floral organ determinacy, ovule differentiation, and seed-hair development in Populus

Abstract

The role of the floral homeotic gene AGAMOUS (AG) and its close homologues in development of anemophilous, unisexual catkins has not previously been studied. We transformed two RNA interference (RNAi) constructs, PTG and its matrix-attachment-region flanked version MPG, into the early-flowering female poplar clone 6K10 (Populus alba) to suppress the expression of its two duplicate AG orthologues. By early 2018, six out of 22 flowering PTG events and 11 out of 12 flowering MPG events showed modified floral phenotypes in a field trial in Oregon, USA. Flowers in catkins from modified events had 'carpel-inside-carpel' phenotypes. Complete disruption of seed production was observed in seven events, and sterile anther-like organs in 10 events. Events with strong co-suppression of both the two AG and two SEEDSTICK (STK) paralogues lacked both seeds and associated seed hairs. Alterations in all of the modified floral phenotypes were stable over 4 yr of study. Trees from floral-modified events did not differ significantly (P < 0.05) from nonmodified transgenic or nontransgenic controls in biomass growth or leaf morphology. AG and STK genes show strong conservation of gene function during poplar catkin development and are promising targets for genetic containment of exotic or genetically engineered trees.

Keywords: AGAMOUS; SEEDSTICK; Populus; RNAi; containment; flowering; matrix attachment region; sterility.

Figure 1

The PtAG2‐derived RNAi hairpin target and construct design. (a) Transcript of the PtAG2 gene. Exons are indicated by boxes. The portion of the complementary DNA sequence used to make the RNA interference (RNAi) hairpin is indicated by a dashed blue line. The location of the MADS box is indicated by a red line. (b)  Diagram of the RNAi construct PTG. (c) Diagram of the RNAi construct MPG. LB, left border; tNOS, terminator of the nopaline synthase gene; nptII, the neomycin phosphotransferase gene; pNOS, nopaline synthase promoter; tOCS, terminator of the octopine synthase gene; PtAG‐intron‐PtAG, inverted repeat of a segment of the PtAG2 gene; p35S, cauliflower mosaic virus 35S promoter; RB, right border; MAR, matrix attachment region derived from the tobacco RB7 gene. Arrows indicate the direction of transcription.

Figure 2

Layered carpels and anther‐like organs observed in altered events. (a) A nontransformed (NT) control Populus alba catkin. (b) An individual flower from the NT control catkin. (c) Hand‐sectioned NT control flower with ovules. (d) Hand‐sectioned NT control flower with immature seeds and cottony seed hair. (e–h) Catkin and flowers from altered event PTG 199 (e–g) or PTG 6‐2 (h). (i–l) Catkin and flowers from altered event MPG 165‐1 (i–k) or MPG 211 (l). st, stigma; ca, carpel; dis, cup‐shaped disk; ov, ovule; sed, seed; ant, anther‐like organ. Bars, 1 mm. Images a–c, e–g and i–k taken on 20 February 2015; images d, h and l taken on 21 March 2016.

Figure 3

Morphological variation was commonly observed within events. (a–d) Populus alba catkins collected from the same branch from event PTG 6‐2 (a, c) and individual flowers isolated from these catkins (b, d). (e–h) Catkins collected from the same branch from event MPG 193 (e, g) and individual flowers isolated from these catkins (f, h). Bars, 1 mm. Arrows in images b, d, g and h indicate anther‐like organs. All images taken on 21 March 2016.

Figure 4

Both constructs produced events that lacked seeds and seed hair. (a) Populus alba seed production and germination by event. Normal events and nontransformed (NT) control are highlighted in gray. Seedless events are indicated by the orange box. (b) Seed pods from NT control released seeds (indicated by red arrows) and cottony seed hair upon maturation and drying. (c) No seed hair was produced or released from dry seed pods from altered event PTG 6‐2. (d) No seed hair was detected in hand‐dissected seed pods (indicated by red circle) from altered event MPG 211. Images (b–d) taken on 22 April 2016.

Figure 5

Relative transcript level of PaAGs and PaSTKs in developing floral buds. (a) Relative PaAGs expression by event, category (combinations of construct and floral morphology), and construct. (b) Relative PaSTKs expression by event, category, and construct. (c) Combined PaAGs and PaSTKs expression by event. Error bars represent ± SE of means. ***, P < 0.001; **, P < 0.01; *, P < 0.05, compared with nontransformed (NT) control. Different letters denote significant differences (P < 0.05) in total expression of both orthologues. Normal events and NT control are highlighted in gray. Seedless events are indicated by orange boxes. PTG/A, altered PTG events; PTG/N, normal PTG events; MPG/A, altered MPG events; MPG/N, normal MPG events; NT/N, nontransgenic control.

Figure 6

Normal Populus alba tree architecture and growth observed in RNA interference events. (a) Tree architecture of nontransformed (NT) control, imaged on 12 March 2014. (b) Tree architecture of the altered event MPG 119, imaged on 14 March 2014. (c) Bar plots of spatially adjusted growth parameters, trunk diameter at breast height (DBH), tree height, and trunk volume index of trees from 2015. PTG/A, altered PTG events; PTG/N, normal PTG events; MPG/A, altered MPG events; MPG/N, normal MPG events; NT/N, nontransgenic control. Bars and error bars represent marginal means and ± SE from the fitted models, respectively.

Figure 7

Normal Populus alba leaf morphology and growth observed in RNA interference events. (a) Leaf scan of nontransformed (NT) control. (b) Leaf scan of the altered event MPG 119. Images (a) and (b) taken on 21 July 2015. Bars, 5 cm. (c–g) Bar plots of leaf parameters, leaf area (c), leaf density (d), total leaf Chl indicated by SPAD meter reading (e), petiole length (f), and petiole width (g) of leaves collected in summer 2015. PTG/A, altered PTG events; PTG/N, normal PTG events; MPG/A, altered MPG events; MPG/N, normal MPG events; NT/N, nontransgenic control. Bars and error bars represent marginal means and ± SE from the fitted models, respectively.