Peptide Aptamers That Bind to Geminivirus Replication Proteins Confer a Resistance Phenotype to Tomato Yellow Leaf Curl Virus and Tomato Mottle Virus Infection in Tomato

Abstract

Geminiviruses constitute a large family of single-stranded DNA viruses that cause serious losses in important crops worldwide. They often exist in disease complexes and have high recombination and mutation rates, allowing them to adapt rapidly to new hosts and environments. Thus, an effective resistance strategy must be general in character and able to target multiple viruses. The geminivirus replication protein (Rep) is a good target for broad-based disease control because it is highly conserved and required for viral replication. In an earlier study, we identified a set of peptide aptamers that bind to Rep and reduce viral replication in cultured plant cells. In this study, we selected 16 of the peptide aptamers for further analysis in yeast two-hybrid assays. The results of these experiments showed that all 16 peptide aptamers interact with all or most of the Rep proteins from nine viruses representing the three major Geminiviridae genera and identified two peptide aptamers (A22 and A64) that interact strongly with different regions in the Rep N terminus. Transgenic tomato lines expressing A22 or A64 and inoculated with Tomato yellow leaf curl virus or Tomato mottle virus exhibited delayed viral DNA accumulation and often contained lower levels of viral DNA. Strikingly, the effect on symptoms was stronger, with many of the plants showing no symptoms or strongly attenuated symptoms. Together, these results established the efficacy of using Rep-binding peptide aptamers to develop crops that are resistant to diverse geminiviruses.

INTRODUCTION

Geminiviruses are a large family of plant viruses characterized by their single-stranded DNA (ssDNA) genomes and twin icosahedral particles (1, 2). They are classified into four genera (Begomovirus, Curtovirus, Topocuvirus, and Mastrevirus) depending on their insect vector, genome structure, and host range (2, 3). Geminiviruses have proliferated worldwide due in part to increasing insect vector populations, changes in crop cultivation, climate change, and increased movement of plant materials (4, 5). Geminivirus diseases cause significant losses in important staple crops such as maize, cassava, vegetable crops, including tomato and beans, and cash crops like cotton (6, 7). A variety of strategies have been used to combat geminivirus diseases, but the development of durable resistance has been confounded by the capacities of these viruses to evolve rapidly and adapt to new hosts and environments (8, 9).

Geminiviruses often occur in disease complexes with multiple viruses working in synergy to increase symptom severity in a host plant (10). Geminivirus genomes display high levels of genetic variability due to high mutation and recombination rates (11, 12). Begomoviruses with bipartite genomes can also undergo reassortment during mixed infections. Many geminiviruses associate with small satellite DNA molecules that carry genes encoding pathogenicity determinants that can also recombine and further contribute to genetic variability and rapid virus evolution in the field (13). All of these properties facilitate the emergence of new and more-virulent viruses (7, 14). Given the complexity and dynamic nature of geminivirus disease complexes, it is essential to develop novel strategies that target multiple viruses and result in stable, broad-based disease resistance.

Conventional breeding and genetic engineering approaches have been used to generate geminivirus-resistant plants. Conventional breeding has improved the productivity of some crops but has been limited by few sources of natural resistance, the multigenic nature of the resistance traits, and the time required for a breeding program (15, 16). Resistance created by conventional breeding can be broken by the appearance of a recombinant virus, association with a DNA satellite, or an unfavorable environment where many viruses are present (17). Several transgenic strategies based on viral sequences have been evaluated, including microRNAs (18), antisense RNAs, RNA interference (RNAi) constructs, and mutant viral proteins (8). Most of these approaches do not confer high levels of resistance or are limited to cognate and closely related viruses (9). Efforts are under way to produce generic geminivirus resistance through an in silico small interfering RNA (siRNA) search (19). Other genetic engineering strategies that confer some level of resistance to a specific geminivirus genus or species (8) include virus-inducible expression of toxic proteins to kill infected cells (20), custom Zn finger DNA-binding proteins that bind to viral replication origins and block replication (21, 22) and GroEL homologs that bind to virus particles and interfere with insect transmission (23).

All geminiviruses encode a replication protein (Rep) (also known as AL1, AC1, L1, C1, or C1:C2) that is essential for viral replication (24). Rep binds to the viral replication origin (25), catalyzes initiation and termination of rolling circle replication (RCR) (26, 27), and functions as the replicative DNA helicase (28–30). It also reprograms plant cell cycle controls to induce the synthesis of host replication machinery necessary for viral replication (31, 32). Rep is involved in a number of protein interactions that are necessary for viral replication, transcription, and infection (33). Its known partners include itself (34), other viral proteins (35–40), and a variety of host proteins involved in DNA replication, recombination, cell cycle regulation, and cell signaling (41).

The N terminus of Rep is highly conserved in the four geminivirus genera (42) and contains three well-characterized motifs that are found in many RCR initiators (43, 44). RCR motif I (FLTY) is involved in double-stranded DNA (dsDNA) binding specificity (45), RCR motif II (HLH) is a metal binding site (3), and RCR motif III [YXXK(D/E)] is the catalytic site for ssDNA cleavage (26, 27). The Rep N terminus also contains a fourth conserved motif designated the geminivirus Rep sequence (GRS) (42).

Peptide aptamers are recombinant proteins that bind to and inactivate a protein of interest (46–49). Peptide aptamers contain a short amino acid sequence inserted into a protein scaffold that constrains conformation, resulting in higher binding specificity and affinity than unconstrained peptides (50). Peptide aptamers can disrupt protein-protein interactions and protein-DNA interactions (47) and inhibit viral function by targeting viral proteins (49, 51–53). Expression of a peptide aptamer that interacts with the nucleocapsid proteins of different tospoviruses in transgenic Nicotiana benthamiana conferred strong resistance to four diverse tospovirus species (51), demonstrating that a single peptide aptamer can confer broad-based viral resistance if an essential viral protein is targeted at a conserved domain. Hence, peptide aptamers that bind to the conserved N terminus of geminivirus Rep proteins have the potential to confer broad-based resistance to geminivirus infection.

An earlier report identified a set of peptide aptamers that bind to the N terminus of the Rep protein from Tomato golden mosaic virus (TGMV) and interfere with viral replication in cultured tobacco cells (49). In this paper, we examined the abilities of selected peptide aptamers to bind to diverse Rep proteins and to confer resistance in tomato to the unrelated geminiviruses, Tomato yellow leaf curl virus (TYLCV) and Tomato mottle virus (ToMoV).

MATERIALS AND METHODS

Plasmid construction.

The cloning strategies are described in supplemental material. The yeast two-hybrid bait and prey plasmids used in this study are listed in Table S1 in the supplemental material, and the PCR primers used for cloning are listed in Table S2 in the supplemental material. Peptide aptamer prey plasmids from the pJM-1 library and the negative prey controls, pYesTrp-Jun and pNSB1172, have been described previously (49). The peptide aptamers were fused to the simian virus 40 (SV40) nuclear localization signal, the Escherichia coli B42 activation domain (AD), and the hemagglutinin (HA) epitope tag (54). The peptide aptamers were expressed from the yeast GAL1 promoter, which is repressed by glucose and activated by galactose. The bait plasmids were constructed by cloning Rep or Rep truncations into pEG202 or pNLexA to create N- or C-terminal LexA DNA binding domain (DBD) fusions, respectively (55) (Origene Technologies, Rockville, MD). The TYLCV and ToMoV plasmids used for agroinoculation are listed in Table S3. Partial tandem copies of the TYLCV, ToMoV A, and ToMoV B genomes were cloned independently into pMON721 (56). The Agrobacterium plasmids used for tomato transformation are listed in Table S4. The HA-tagged peptide aptamers (HA-peptide aptamers) flanked by the Cauliflower mosaic virus (CaMV) E35S promoter and the rbcS E9 terminator were cloned independently into pMON721. All of the clones were confirmed by DNA sequencing.

Yeast two-hybrid assays.

The Saccharomyces cerevisiae strains EGY48 (MAT his3 trp1 ura3-52 leu2::LexA6op-LEU2) and/or EGY191 (MAT his3 trp1 ura3-52 leu2::LexA2op-LEU2) were cotransformed with bait and prey plasmids and the lacZ reporter plasmid pSH18-34 using a lithium acetate-polyethylene glycol protocol (57). The bait, prey, and lacZ plasmids were selected in medium lacking histidine (−H), tryptophan (−W), or uracil (−U), respectively. The transformants were plated on synthetic dropout medium containing glucose and lacking histidine, tryptophan, and uracil (Glu−HWU). For yeast two-hybrid growth assays, 5-μl droplets of 1 × 10⁻² dilutions (optical density at 600 nm [OD₆₀₀] adjusted to 0.12 to 0.20) of fresh yeast colonies were plated onto synthetic dropout medium containing galactose but lacking histidine, tryptophan, uracil, and leucine (Gal-HWUL) and incubated at 30°C for 2 to 12 days with constant monitoring and photo documentation.

For β-galactosidase assays, freshly patched yeast cells were used to inoculate 1.2 ml of liquid medium (Gal−HWU) and grown overnight at 30°C shaking at 250 rpm. The cells were pelleted at 1,000 × g for 5 min, rinsed with Z buffer (0.1 M NaPO₄ [pH 7], 10 mM KCl, 1 mM MgSO₄), and resuspended in 300 μl of Z buffer. A 10-μl aliquot was diluted into 90 μl of water, and OD₆₀₀ was measured using a 96-well microplate reader. The remaining cells were subjected to five freeze-thaw cycles in liquid nitrogen. The cell lysate (150 μl) was added to 50 μl of Z buffer containing 40 mM β-mercaptoethanol. The β-galactosidase enzyme assays adapted from Clontech protocols (57) were performed as previously described (42). The different constructs were tested in a minimum of three experiments, each of which assayed three independent transformants per construct.

For immunoblot analysis of bait protein expression, soluble proteins were extracted from the yeast transformants as previously described (42). Total protein (50 μg) was resolved on a 12% polyacrylamide–SDS gel and analyzed by immunoblotting using an anti-LexA DBD polyclonal antibody (catalog no. ab50953; Abcam).

Tomato plant growth and transformation.

Tomato (Solanum lycopersicum cv. Micro Tom) seeds were germinated in soil placed in small pots (8.5-cm length by 6.5-cm width by 6-cm height) in a tray (24 small pots per tray). The trays were placed at 25°C under 16-h light/8-h dark photoperiod and 50% humidity. The trays were covered with a transparent dome for a week. The plants were watered every other day and fertilized once a week by adding Miracle-Gro (Scotts Miracle-Gro Company) to the water.

Agrobacterium tumefaciens ABI carrying a transformation plasmid for the HA-peptide aptamers (pNSB1735 for peptide aptamer A22 [peptide aptamers explained in Results and shown in Table 1], pNSB1785 for peptide aptamer A64, or pNSB1743 for TRX-GST [TRX stands for the thioredoxin scaffold, and GST stands for glutathione S-transferase]) was used to transform the peptide aptamers into tomato following a protocol adapted from Cortina and Culiáñez-Macià (58) and described in the supplemental material.

Table 1.

Rep-binding peptide aptamers

Aptamer	Sequence	Motif(s)a	Interference scoreb	Relative strengthc
A22	CRTRGCGCHLCRMLSQFTGG	4, 24	87	100
A40	LQYSWNLYSVASFKTRRVSS	1, 25, 27	62	35
A46	CYMEVEGRPRRWADSFFVAW	1	69	29
A59	AKDVERGAGGKIKACELCRL	24	53	53
A64	TELWWADFCKMHMEGGKGMC	4	80	82
A84	GGRQTEPSLTLLADLTLLLS	20	83	54
A99	RERGGDDYRRMMHPGAASGP	4, 20	80	19
A116	SCDEAFDAASVASELFCQPY	27	84	22
A127	TWGLVCTGTGWGLLDTVVRA	28	6	34
A132	GRVQLEILVSEAEEGVSPFL		57	38
A135	RDAEWQDVLGRARAVHLRGR	28	6	32
A147	GRGGCMLCDVDGSSAWLHTEGRLTGPITSQQCLSFQYLGNGEFIDG	24	87	18
A155	FCPECQMVAGAEDGDAIDLQ		70	32
A159	LWGGGTAWDFFVWGEDSAC	25	71	23
A160	GMSGRIPEPDDWVVLFITGC	20	84	28
A176	AWDSESLATWASVMPWPYPT	1, 25, 27, 28	41	24
TRX-GST (control)	ELPILGYWKIKGLVQPTRGP		6	6

^aThe aptamer motifs are amino acid sequences enriched in the 88 peptide aptamers in reference 49.

^bInterference score is the ratio (multiplied by 100) of the amount of TGMV dsDNA that accumulates in tobacco protoplasts cotransfected with a TGMV-A replicon and a peptide aptamer expression cassette versus the TRX-GST negative-control cassette reported in reference 49.

^cThe average of the β-galactosidase activity results for each aptamer normalized to A22 for the TGMV Rep_1–130 and Rep_1–352 baits in Fig. 1A.

HA-peptide aptamer protein expression in T1 transgenic lines.

Leaves from 3-week-old seedlings were excised, flash frozen in liquid nitrogen, and ground. The ground tissue (100 mg) was resuspended in 250 μl of homogenization buffer (40 mM Tris acetate, 100 mM potassium acetate, 1 mM EDTA, 1 mM dithiothreitol [DTT], 20% glycerol, and 1× protease inhibitor cocktail [catalog no. P9599; Sigma]) and centrifuged at 15,000 rpm for 20 min at 4°C. The supernatant was recovered, and the total protein concentration was determined using the Bio-Rad protein assay kit (Bradford assay kit). Total protein (150 μg) was resolved on a NuPAGE 4 to 12% Bis-Tris minigradient SDS gel (catalog no. IM-8042; Novex) and analyzed by immunoblotting using a rat monoclonal anti-HA antibody at 0.2 μg/ml (catalog no. 11 867 423 001; Roche).

Agrobacterium-mediated inoculation of tomato with TYLCV and ToMoV.

Agrobacterium tumefaciens ABI cells transformed with pNSB1736 (TYLCV), pNSB1906 (ToMoV A), pNSB1877 (ToMoV B), or pMON721 (empty vector) were streaked on LB plates containing 75 μg/ml spectinomycin and incubated at 30°C for 2 days. A single colony was used to inoculate LB medium containing 75 μg/ml spectinomycin and grown overnight at 30°C. The overnight culture was first diluted to an OD₆₀₀ of 1.0 and then diluted 1:100 for agroinoculation. Diluted agrobacteria transformed with the TYLCV agroinoculation plasmid were placed in a 5-ml syringe capped with a 27-gauge 0.5-in. needle. The diluted agrobacteria transformed with the ToMoV A or ToMoV B agroinoculation plasmids were mixed before they were placed in a syringe capped with a needle. Agrobacteria transformed with pMON721 were used as a negative control (mock inoculation).

Two-week-old tomato plants were gently pricked with a needle about 10 times in the stem immediately below the apex (59). A few small drops of the diluted Agrobacterium tumefaciens culture were released from the syringe through the needle onto the wound. Each tray contained 4 wild-type plants, 2 plants that were inoculated with virus and 2 plants that were mock inoculated, and 20 transgenic plants consisting of 10 plants per independent transgenic line, 9 of which were inoculated with virus and 1 that was mock inoculated. The inoculated plants were covered with a transparent dome and placed in the 25°C chamber under 16-h light/8-h dark photoperiod and 50% humidity. After 2 days, the domes were removed, and the plants were allowed to grow. The plants were scored for symptoms using a scale from 0 to 4 (0 for no symptoms and 4 for the most severe symptoms), and viral DNA loads were determined as described below.

Viral DNA detection in infected tomato plants.

Leaf samples were collected at 14, 21, and 28 days postinfection (dpi) for TYLCV-inoculated plants and at 7, 14, and 21 dpi for ToMoV-inoculated plants. For leaf disc printing, 60-mm leaf discs were excised from the second youngest leaf of the plant being tested and rubbed onto a nylon membrane with a plastic pestle (31, 60). Viral DNA in the leaf disc print was UV cross-linked twice to the membrane. A TYLCV-specific probe was prepared from a 1.6-kb ClaI fragment of pTYLC2. A 2.6-kb SacI fragment of pNSB1691 was used to generate a ToMoV A-specific probe. Probes were radiolabeled using [α-³²P]dATP. The membranes were hybridized overnight with ³²P-labeled DNA probes, and viral DNA was quantified by phosphorimager analysis (40).

RESULTS

Peptide aptamers A22 and A64 bind strongly to TGMV Rep.

In a previous paper (49), we described a yeast two-hybrid screen of the pJM-1 library that encodes E. coli thioredoxin (TrxA) with 2.9 × 10⁹ random 20-mer peptides in its active site (54). The screen identified 88 peptide aptamers that bind to the TGMV Rep protein and showed that 14 strongly interfere with TGMV replication in cultured tobacco cells (49). In this study, we selected 16 of the peptide aptamers for further analysis in yeast. The selected aptamers, which are referred to by the letter A followed by the peptide number (e.g., A22), encompass a range of replication interference activities and putative aptamer sequence motif groups (Table 1) (49). The selected peptide aptamers included seven that strongly interfere with TGMV DNA replication (interference score ≥ 80), seven that have moderate interference activity (interference score between 41 and 79), and two that do not impact viral replication in tobacco cells. The seven aptamer motif groups described previously are represented among the selected peptide aptamers. Several peptide aptamers contain multiple aptamer sequence motifs, while A132 and A155 do not contain a aptamer motif. TRX-GST is a negative control that does not bind to TGMV Rep or interfere with viral replication (49).

We first examined the relative binding activities of the selected peptide aptamers to TGMV Rep using semiquantitative β-galactosidase assays (61). The 16 peptide aptamer prey plasmids and the negative-control TRX-GST were fused to the E. coli B42 activation domain (AD) (49). They were cotransformed into yeast with TGMV Rep bait plasmids corresponding to full-length Rep_1–352 (Rep with amino acids 1 to 352) or Rep_1–130 (Rep with amino acids 1 to 130) fused to the LexA DNA binding domain (DBD) and the lacZ reporter plasmid pSH18-34 (62). Yeast transfected with Rep_1–130 expressed higher levels of the reporter in the presence of the aptamer preys than Rep_1–352, but the profiles of the different preys were similar with the two baits (Fig. 1A). The two baits were expressed at similar levels (Fig. 2C, compare lanes 1 and 3), so the difference in reporter level is likely due to reduced accessibility of the N-terminal region in Rep_1–352 than in Rep_1–130 bait fusion proteins.

Fig 1.

Characterization of peptide aptamer binding to TGMV Rep. (A) Each peptide aptamer was cotransformed into yeast with TGMV Rep_1–130 or TGMV Rep_1–352 and the lacZ reporter plasmid pSH18-34. The interaction was measured in semiquantitative β-galactosidase (β-gal) assays. The peptide aptamer lines (black bars) and the negative-control TRX-GST (white bar) are shown. (B) The regions in TGMV Rep that bind to peptide aptamers A22 and A64 were mapped using N-terminal (Rep_1–180, Rep_36–180, Rep_64–180, and Rep_98–180) and C-terminal (Rep_1–36, Rep_1–64, Rep_1–98, and Rep_1–130) truncations. (Top) A diagram of TGMV Rep_1–180 is shown with RCR motifs I, II, and III shown as gray boxes and GRS depicted as a black box. In the graphs, the interactions are represented as fold increase of β-galactosidase activity above the background level for A22 and A64. In the schematic representations of A22 and A64, the strongest interacting region is represented by the black line in the diagram for each peptide aptamer. The broken line indicates additional regions that might contribute to binding.

Fig 2.

Yeast two-hybrid growth assay. Selected peptide aptamers were tested for interaction with TGMV Rep_1–130. (A) Yeast cells containing the selected peptide aptamers and the GUS negative control or TGMV Rep_1–130 baits were analyzed for growth on Gal−HWUL medium. The negative prey controls AD:Jun and peptide aptamer TRX-GST are indicated by the white boxes. The day after plating is indicated below the pictures. (B) Key for the peptide aptamers on the plates shown in panel A (left) and day when growth was visible (right). The peptide aptamers are shown in the key without the initial A. A22 was the first to show growth on day 2. Most of the other peptide aptamers (A46, A59, A64, A84, A116, A127, A132, A147, A155, A159, A160, and A176) showed growth on day 3, while A40 grew on day 4 and A99 and A135 grew on day 5. The negative controls (AD:Jun and TRX-GST) began showing growth on day 6, indicating that the experiment could no longer be analyzed for interaction. (C) Total protein extracts from yeast transformed with LexA DBD fusions corresponding to TGMV Rep_1–352, GUS, TGMV Rep_1–130, BCTV Rep_1–358, ACMV Rep_1–358, ToMoV Rep_1–130, CLCuBV Rep_1–130, TYLCV Rep_1–130, CaLCuV Rep_1–349, EACMV Rep_1–359, and MSV RepA_1–272. The extracts (50 μg) were fractionated by SDS-PAGE and analyzed by immunoblotting with an anti-LexA DBD antibody.

The highest levels of β-galactosidase activity were seen with A22 and A64 peptide aptamers. A59 and A84 peptide aptamers also resulted in high β-galactosidase levels. The other peptide aptamers gave lower levels of reporter expression. However, the activities of all of the peptide aptamers were above that measured with the TRX-GST negative control. We cannot rule out the possibility that differences in reporter expression seen with the various peptide aptamer-AD fusions were due to differential accumulation, but this seems unlikely because the prey proteins are identical except for a 20-amino-acid sequence in the TRX-A active site.

The peptide aptamers bind to diverse geminivirus Rep proteins.

The N termini of geminivirus Rep proteins contain several conserved amino acid motifs that might be recognized by the peptide aptamers. To test this possibility, we screened the 16 selected peptide aptamers for binding to a diverse set of geminivirus Rep proteins. The Rep proteins in Table 2 represent the three major geminivirus genera, and eight are from viruses that cause losses in key crops around the world. East African cassava mosaic virus–Uganda 2 severe (EACMV-UG2) and African cassava mosaic virus–Cameroon (ACMV-CM) are part of the Cassava mosaic disease complex that is devastating to cassava in sub-Saharan Africa (63). Maize streak virus (MSV) is a major constraint on maize production in Africa (64). Cotton leaf curl Burewala virus (CLCuBV) is a member of the viral complex that severely impacts cotton production in the Indian subcontinent (65). ToMoV, Beet curly top virus–California (Logan) (BCTV-CA), and Cabbage leaf curl virus (CaLCuV) are associated with losses in vegetable crops in United States (66–69), while TYLCV is the major viral problem in tomatoes worldwide (70, 71). The selected Rep proteins are from highly diverse geminiviruses that include viruses of the Old World (OW) (Europe, Asia, Africa, and Middle East) and New World (NW) (North America, South America, and the Caribbean) lineages, viruses with monopartite and bipartite genomes, and viruses associated with β satellites or the putative DNAII/DNAIII satellites. The N termini of the selected Rep proteins have 82% identity and 92% similarity to 24% identity and 40% similarity to TGMV Rep_1–130.

Table 2.

Geminivirus Rep proteins tested for aptamer binding

Virusa	Genus	Geographyb	Genome	Satellite	% identity/% similarityc	GenBank accession no.
TGMV	Begomovirus	NW	Bipartite		100/100	NC_001507
ToMoV-FL	Begomovirus	NW	Bipartite		82/92	NC_001938
CaLCuV	Begomovirus	NW	Bipartite		49/66	NC_003866
CLCuBV	Begomovirus	OW	Bipartite	Beta	80/89	NC_012137
EACMV-UG2 severe	Begomovirus	OW	Bipartite	DNAII/DNAIII	72/82	NC_004674
ACMV-CM	Begomovirus	OW	Bipartite	DNAII/DNAIII	64/75	AF112352
TYLCV-DR	Begomovirus	OW	Monopartite		73/86	EF110890
BCTV-CA (Logan)	Curtovirus	NW	Monopartite		75/83	NC_001412
MSV-SA	Mastrevirus	OW	Monopartite		24/40	NC_001346

^aAbbreviations: TGMV, Tomato golden mosaic virus; ToMoV-FL, Tomato mottle virus–Florida; CaLCuV, Cabbage leaf curl virus; CLCuBV, Cotton leaf curl Burewala virus; EACMV-UG2 severe, East African cassava mosaic virus–Uganda 2 severe; ACMV-CM, African cassava mosaic virus–Cameroon; TYLCV-DR, Tomato yellow leaf curl virus–Dominican Republic; BCTV-CA, Beet curly top virus–California; MSV-SA, Maize streak virus–South Africa.

^bNW, New World (North America, South America, and the Caribbean); OW, Old World (Europe, Asia, Africa, and Middle East).

^cAmino acid sequence identity/similarity relative to TGMV Rep_1–130.

The 16 peptide aptamers and the negative-control TRX-GST were cotransformed into yeast with the nine Rep proteins either as N-terminal regions or full-length proteins fused to the LexA DBD and were analyzed in two-hybrid growth assays. The choice of Rep bait was based on the amount of background growth in bait-alone controls. Immunoblot analysis using an anti-LexA antibody showed that all of the baits were expressed, albeit at different levels (Fig. 2C and Table 3). We used β-glucuronidase (GUS) encoded as a DBD–β-glucuronidase fusion as a negative bait control. The interactions were screened in a low-stringency yeast strain (EGY48) except for assays containing the CaLCuV Rep_1–349 and EAMCV Rep_1–359 baits, which were screened in a high-stringency yeast strain (EGY191) because of bait autoactivation background at low stringency.

Table 3.

The peptide aptamers bind to diverse geminivirus Rep proteins

^aInteractions between the peptide aptamers and the nine Rep proteins are shown. The result of a given combination of peptide aptamer and Rep protein is listed in the table as “yes” if interaction was observed and as “no” if no interaction was seen. The combinations that were negative are shown shaded for emphasis.

^bThe relative expression of each bait protein in yeast was determined on immunoblots using an anti-LexA DBD antibody (Fig. 2C).

^cLow-stringency assays used reporters with six LexA DNA binding sites, while high-stringency assays used reporters with two LexA DNA binding sites.

For the growth assays, the optical densities of yeast cotransformed with a peptide aptamer and a Rep protein were measured, and equivalent amounts were plated on medium lacking leucine to select for protein-protein interaction. LEU2 reporter expression in yeast strains EGY48 and EGY191 is under the control of six copies or two copies, respectively, of the LexA operator sequence and the minimal LEU2 promoter. Activation of the LEU2 reporter was assayed by monitoring yeast growth over a week. The first day that growth was observed was noted, as shown for TGMV Rep_1–130 and the peptide aptamers in Fig. 2A and B. Once the negative controls began showing growth at about 6 days after plating, the experiment could no longer be analyzed for interaction, and the peptide aptamers that did not show growth before this day were considered negative for interaction. Table 3 summarizes the results of the yeast two-hybrid screens performed in triplicate for interactions between the peptide aptamers and the nine Rep proteins. The result of a given combination of peptide aptamer and Rep protein is listed in the table as “yes” if interaction was observed and as a “no” if no interaction was seen. Of the 160 peptide aptamer/Rep protein combinations analyzed, 146 combinations were positive for interaction. The control aptamer TRX-GST did not interact with any of the Rep baits.

Early growth on a plate most likely reflected strong interaction between the peptide aptamer and the geminivirus Rep protein. Yeast cotransformed with the A22 peptide aptamer showed the earliest growth (2 or 3 days after plating), and growth of yeast cotransformed with A64 was often the next to be observed (3 or 4 days after plating). This was generally true for all of the Rep baits, indicating that A22 and A64 peptide aptamers interact strongly with diverse Rep protein (Reps), analogous to our results for TGMV Rep in semiquantitative β-galactosidase assays (Fig. 1A).

A22 and A64 peptide aptamers bind to different regions in the Rep N terminus.

The peptide sequences of A22 and A64 both contain putative aptamer motif 4 and might interact similarly with Rep. To address this possibility, we mapped the A22 and A64 binding sites in the TGMV Rep N terminus using a series of truncated baits. The N-terminal truncations included Rep_1–180, Rep_36–180, Rep_64–, and Rep_98–180 fused to the LexA DBD at position 180 in the Rep protein. The C-terminal truncations included Rep_1–130, Rep_1–98, Rep_1–64, and Rep_1–36 fused to the LexA DBD at position 1 in the Rep protein. By positioning the LexA DBD at the opposite end of the truncations in the Rep protein, we sought to minimize any effects of the fusions on binding activity. The truncations separated the three RCR motifs and the GRS in the TGMV Rep N terminus (Fig. 1B).

The relative binding activities of the truncated Rep baits for A22 and A64 peptide aptamers were measured in semiquantitative β-galactosidase assays (Fig. 1B). A22 interacted strongly with TGMV Rep_1–180. The interaction was greatly reduced for Rep_36–180 and abolished for Rep_64–180 and Rep_98–180, showing that deletion of amino acids 1 to 35 greatly impairs binding. A22 interacted with Rep_1–130, Rep_1–98, and Rep_1–64 similarly but did not bind to Rep_1–36, indicating that amino acids 1 to 36 are not sufficient for interaction. Together, the N- and C-terminal truncation data indicated that A22 interacts with TGMV Rep between amino acids 1 and 64 with the primary interaction between amino acids 1 and 36 (Fig. 1B, A22, black and broken line). A64 interacted strongly with TGMV Rep_1–180, Rep_36–180, and Rep_64–180 but showed no interaction with Rep_98–180, demonstrating that deletion of amino acids 64 to 97 greatly impairs binding. A64 showed efficient binding with Rep_1–130 and Rep_1–98, reduced binding with Rep_1–64, and no binding to Rep_1–36, indicating that amino acids 1 to 64 are not sufficient for efficient interaction. Together, the N- and C-terminal truncation data showed that A64 interacts with TGMV Rep between amino acids 36 and 97 with the strongest interaction between amino acids 64 and 97 (Fig. 1B, A64, black and broken line).

We detected reporter activity with the truncated Rep baits in the presence of peptide aptamer A22 and/or A64, indicative of their expression in yeast. The one exception was Rep_1–36, and we cannot rule out the possibility that the lack of reporter activity with either A22 or A64 reflects poor expression of the bait. However, the data clearly showed that A22 and A64 peptide aptamers interact with different regions in the Rep N terminus. Together, the yeast two-hybrid studies indicated that A22 and A64 are strong candidates for binding to and interfering with Rep proteins from diverse geminiviruses.

Tomato plants expressing TRX-GST are susceptible to TYLCV infection.

To test whether the peptide aptamers can interfere with Rep function and confer resistance to geminivirus infection in plants, we selected tomato because of its high susceptibility to a large number of geminiviruses (72). For our studies, we chose the begomoviruses TYLCV and ToMoV, which have different lineages (Old World versus New World) and genome structures (monopartite versus bipartite). TYLCV is the major viral pathogen of tomato around the world (68, 73), while ToMoV is an important pathogen in the Americas. Because of the global impact of TYLCV, we first challenged aptamer-expressing plants with TYLCV and tested the best performing lines in ToMoV infection assays.

Our first step was to ask whether the TRX scaffold of the peptide aptamers alters TYLCV infection in plants expressing the negative-control TRX-GST. Tomato was transformed with a plant expression cassette corresponding to HA-tagged TRX-GST under the control of the CaMV E35S promoter with a duplicated enhancer region (74) and a pea rbcS 3′ end. Total protein extracts from young leaves from four independent T1 lines were subjected to anti-HA immunoblot analysis (Fig. 3A). A band of the predicted size was detected in extracts from the four TRX lines but not from wild-type tomato leaves, confirming transgene expression. Lines TRX-12 and TRX-25 showed the highest levels of transgene expression. The TRX-GST lines appeared normal, flowered, and produced fruit.

Fig 3.

Tomato lines expressing TRX-GST are susceptible to TYLCV infection. (A) Total protein extracts (150 μg) from 3-week-old transgenic tomato seedlings (T1 generation) were resolved by SDS-PAGE and visualized by anti-HA immunoblotting for the indicated lines. Wild-type (WT) seedlings were used as the negative control. (B) Scoring system for TYLCV symptoms (from 1 to 4). Scores: 1, 1 or 2 curled and wrinkled leaves; 2, ≥3 curled and wrinkled leaves; 3, ≥3 curled and wrinkled leaves with mild chlorosis at the leaf margins and overall stunting of the plant; 4, severe chlorosis and stunting. (C and D) Tomato plants agroinoculated with TYLCV were monitored for symptoms (C) and viral DNA levels (D) at 14, 21, and 28 dpi. (C) Symptoms were scored from 0 to 4 with 0 being no symptoms and 4 being the most severe symptoms. (D) Relative viral DNA levels were determined in leaf disc prints hybridized to a ³²P-labeled ToMoV probe. The hybridization signals were quantified and normalized to the wild-type control. Statistical significance (∗, P value < 0.01) was determined using a two-tailed Student's t test. The error bars show 2 standard errors (SE).

The TRX-GST lines were agroinoculated with a TYLCV replicon containing partial tandem copies of the viral genome. In the experiment, nine wild-type plants and nine plants for each of the four TRX-GST lines were inoculated with TYLCV, while nine wild-type plants and one plant for each transgenic line were mock inoculated with the empty vector control, pMON721. Positive- and negative-control plants were distributed throughout the trays containing the transgenic plants to control for any potential differences due to location in the growth chamber. Symptoms were scored at 14, 21, and 28 dpi using a symptom score scale from 0 to 4 (Fig. 3B). Wild-type plants inoculated with TYLCV showed curled and wrinkled leaves, mild chlorosis at leaf margins, and stunting typical of TYLCV infection at 14 dpi (Fig. 3C, WT). Symptoms progressively increased with most plants showing severe chlorosis and stunting at 21 dpi and 28 dpi. The four TRX-GST lines had slightly milder symptoms than wild-type plants at 14 dpi, but the differences were not statistically significant. At 21 and 28 dpi, the symptoms of the four TRX-GST lines were indistinguishable from those of wild-type plants and ultimately resulted in severe chlorosis and stunting (Fig. 3C). Similar results were obtained with four additional TRX-GST lines (not shown).

We also quantified viral DNA accumulation in TYLCV-inoculated and mock-inoculated plants at 14, 21, and 28 dpi using leaf disc prints hybridized with a ³²P-labeled TYLCV DNA probe (Fig. 3D). We chose this assay to quantify viral DNA accumulation over time (31, 60) because of the large numbers of plants and transgenic lines included in our analysis. No viral DNA was detected in mock-inoculated plants, which were used as background controls. For each time point, viral DNA was quantified using ImageQuant software and normalized to mean viral DNA accumulation in wild-type plants (set at 100 for each time point). This analysis confirmed the presence of viral DNA in TYLCV-inoculated wild-type plants at the three time points. Lower viral DNA accumulation was observed for the TRX-GST lines at 14 dpi, but the difference was only statistically significant (P < 0.01) for line TRX-17. Lower levels of viral DNA accumulation at 14 dpi were also seen for TRX-GST lines in other experiments, but the differences were not statistically significant (Fig. 4C and 5C). At 21 and 28 dpi, there were no differences in the levels of viral DNA accumulation between wild-type plants and the four TRX-GST lines. These results indicated that expression of TRX-GST may confer a low level of resistance to TYLCV at 14 dpi, but the effect is no longer apparent at 21 and 28 dpi, with the TRX-GST lines resembling wild-type plants at the later time points. The similarity in symptom severity and viral DNA levels established that TRX-GST expression in tomato does not confer TYLCV resistance.

Fig 4.

Two TYLCV-infected A22-expressing tomato lines have mild symptoms and lower viral DNA loads. (A) Total protein extracts (150 μg) from 3-week-old transgenic tomato seedlings (T1 generation) were resolved by SDS-PAGE and visualized by anti-HA immunoblotting for the indicated lines. Wild-type seedlings were the negative control, and TRX-25 seedlings were the positive control for transgene expression. (B and C) Tomato plants agroinoculated with TYLCV were monitored for symptoms (B) and viral DNA levels (C) at 14, 21, and 28 dpi. (B) Symptoms were scored from 0 to 4 with 0 being no symptoms and 4 being the most severe symptoms. Statistical significance was determined using a Wilcoxon rank sum test and indicated by asterisks as follows: ∗, P value < 0.01; ∗∗, P value < 0.001. (C) Relative viral DNA levels were determined in leaf disc prints hybridized to a ³²P-labeled TYLCV probe. The hybridization signals were quantified and normalized to the wild-type control. Statistical significance was determined using a two-tailed Student's t test and indicated by asterisks as follows: ∗, P value < 0.01; ∗∗, P value < 0.001. The error bars show 2 SE.

Fig 5.

TYLCV-infected A64-expressing tomato lines have milder symptoms and lower viral DNA loads. (A) Total protein extracts (150 μg) from 3-week-old transgenic tomato seedlings (T1 generation) were resolved by SDS-PAGE and visualized by anti-HA immunoblotting for the indicated lines. Wild-type seedlings were the negative control, and TRX-25 seedlings were the positive control for transgene expression. (B and C) Tomato plants agroinoculated with TYLCV were monitored for symptoms (B) and viral DNA levels (C) at 14, 21, and 28 dpi. (B) Symptoms were scored from 0 to 4 with 0 being no symptoms and 4 being the most severe symptoms. Statistical significance was determined using a Wilcoxon rank sum test and indicated by asterisks as follows: ∗, P value < 0.01; ∗∗, P value < 0.001. (C) Relative viral DNA levels were determined in leaf disc prints hybridized to a ³²P-labeled TYLCV probe. The hybridization signals were quantified and normalized to the wild-type control. Statistical significance was determined using a two-tailed Student's t test and indicated by asterisks as follows: ∗, P value < 0.01; ∗∗, P value < 0.001. The error bars show 2 SE.

Transgenic tomato lines expressing peptide aptamer A22 or A64 display resistance to TYLCV infection.

The yeast two-hybrid experiments indicated that the A22 and A64 peptide aptamers interact strongly with different regions in the TGMV Rep N terminus (Fig. 1) and that both bind to TYLCV and ToMoV Rep (Table 3). A22 bound to all nine of the Rep proteins tested for interactions, while A64 bound to eight out of nine Rep proteins tested. However, with the exception of A22, A64 displayed stronger Rep binding and/or replication interference activity than A46, A59, A84, and A147, each of which bound to all of the Reps. Hence, we transformed tomato with plant expression cassettes corresponding to HA-tagged versions of A22 or A64 in the same vector used for TRX-GST. We detected transgene expression in 11 and 25 independent lines corresponding to A22 and A64, respectively, by anti-HA immunoblotting. The results are shown for 7 A22 T1 lines in Fig. 4A and for 13 A64 T1 lines in Fig. 5A. The levels of transgene expression varied widely between lines, indicative of position effects. No phenotypic effects were observed in any of the transgenic lines irrespective of transgene expression level (not shown).

We challenged the T1 plants corresponding to 10 independent A22 lines and 14 A64 lines with TYLCV in four separate experiments. Each experiment was designed as described above for the TRX-GST lines with nine TYLCV-inoculated plants for each genotype (each A22 line, each A64 line, TRX-25, and wild type) and one mock-inoculated control for each transgenic line and nine for wild-type tomato. Lines that did not differ from the wild-type plants were tested in a single experiment, while lines that showed reduced viral symptoms and DNA accumulation were challenged in multiple experiments. Figures 4 and 5 show representative results of experiments with the A22 and A64 lines, respectively.

In Fig. 4B, all of the A22 lines showed reduced symptoms that were statistically different from those of the wild type at 14, 21, and 28 dpi. A22-5 and A22-82 plants scored between 0 and 1 at all three time points, indicating that they showed almost no visible symptoms (shown for A22-5 line at 21 dpi in Fig. 6). Leaf disc print analysis of viral DNA accumulation showed that A22-5 and A22-82 contained reduced levels of viral DNA that were statistically significant compared to those of the wild type at the three time points postinfection. A22-41, A22-56, and A22-95 also had lower viral DNA loads at 14 dpi but not at 21 and 28 dpi. Viral DNA levels in the other A22 lines were equivalent, and in some cases greater, than those of the wild type at 21 and 28 dpi. However, the differences were not statistically significant because of the greater variation observed between plants with high viral loads.

Fig 6.

Examples of A22, A64, and TRX-GST plants inoculated with TYLCV at 21 dpi and ToMoV at 14 dpi. Each panel shows a wild-type plant either mock inoculated (mock) or inoculated with TYLCV or ToMoV (wt). Transgenic (T1) plants A22-5, A64-21, or TRX-GST inoculated with TYLCV and ToMoV are shown.

In Fig. 5B, 10 of the 13 A64 lines also showed significantly milder symptoms than the wild type at 14, 21, and 28 dpi. The A64-21, A64-47, A64-64, A64-100, A64-107, A64-157, and A64-159 lines scored between 0 and 1 at all three time points, indicating that they had almost no visible symptoms (shown for A64-21 line at 21 dpi in Fig. 6). Lines A64-13 and A64-14 also showed reduced symptoms at early times, while line A64-62 showed no symptom reduction at any time. Leaf disc print analysis of viral DNA accumulation showed that lines A64-21, A64-47, and A64-157 contained statistically significant lower levels of viral DNA compared to the wild type at the three time points (Fig. 5C). A64-13 plants had lower viral DNA levels at 14 and 28 dpi, while A64-14, A64-100, A64-107, A64-129, and A64-159 plants had reduced levels at 14 dpi. Viral DNA accumulation did not differ from the wild type in A64-62, A64-64, A64-67, and A64-115 plants.

Although some transgenic lines expressing low levels of peptide aptamers developed more severe symptoms and contained higher levels of viral DNA (e.g., A22-33 and A22-95 lines in Fig. 4), there was no clear relationship between peptide aptamer expression level, reduced symptoms, and viral DNA levels (Fig. 4, compare the A22-82 and A22-41 lines; Fig. 5, compare the A64-157 and A64-67 lines). This result suggested that other factors influenced by the chromosomal position of the transgene, such as where the peptide aptamers are expressed in plants, may impact the resistance phenotype.

Transgenic tomato lines expressing peptide aptamer A22 or A64 also show resistance to ToMoV infection.

We asked whether the peptide aptamers also confer a resistance phenotype to ToMoV infection. For these studies, we challenged two A22 lines (A22-5 and A22-82) and two A64 lines (A64-21 and A64-47) that displayed reduced symptoms and viral DNA loads in the TYLCV infection assays. Our choice of lines was also dictated by the availability of T1 seed, which had been depleted for many of the lines by the TYLCV experiments. We used T1 plants because they are not segregating for the transgene and, thus, represent a uniform population. Under our assay conditions, ToMoV disease progression was quicker than observed for TYLCV. As a consequence, ToMoV symptoms and viral DNA loads were assayed 1 week earlier at 7, 14, and 21 dpi. ToMoV symptoms were scored using a numerical scale from 0 to 4 (Fig. 7A). Wild-type plants and plants expressing TRX-GST were tested in parallel in the ToMoV assays using the same tray design as described for the TYLCV experiments. The ToMoV studies were repeated twice with similar results.

Fig 7.

ToMoV-infected A22- and A64-expressing tomato lines have milder symptoms and lower viral DNA loads. (A) Scoring system for ToMoV symptoms. Scores: 1, 1 or 2 curled and wrinkled leaves; 2, ≥3 leaves curled upward and wrinkled; 3, ≥3 leaves prominently curled upward and mild chlorosis at leaf margins; 4, mottled pattern of chlorotic blotches on all leaves. (B and C) Tomato plants agroinoculated with ToMoV were monitored for symptoms (B) and viral DNA levels (C) at 7, 14, and 21 dpi. (B) Symptoms were scored from 0 to 4 with 0 being no symptoms and 4 being the most severe symptoms. Statistical significance was determined using a Wilcoxon rank sum test and indicated by asterisks as follows: ∗, P value < 0.01; ∗∗, P value < 0.001. (C) Relative viral DNA levels were determined in leaf disc prints hybridized to a ³²P-labeled ToMoV probe. The hybridization signals were quantified and normalized to the wild-type control. Statistical significance (∗∗, P value < 0.001) was determined using a two-tailed Student's t test. The bars show 2 SE.

In Fig. 7B, the A22 and A64 lines showed reduced ToMoV symptoms at 7, 14, and 21 dpi that were statistically different from those of the wild type. In contrast, there were no differences in the symptoms between the negative-control line TRX-12 and wild-type plants. At 7 dpi, the symptom scores of A22-5, A64-21, and A64-47 were between 0 and 1, with most plants displaying no visible symptoms (shown for A22-5 and A64-21 lines at 14 dpi in Fig. 6). Leaf disc print analysis showed that viral DNA accumulation was low in all four lines at 7 dpi (Fig. 7C). Symptoms were also attenuated at 14 and 21 dpi, with none of the aptamer-expressing plants achieving level 4 symptoms, although viral DNA levels approached wild-type levels at the later time points. These results demonstrated that expression of A22 or A64 in transgenic tomato can attenuate infection by diverse geminiviruses.

The peptide aptamers impact symptoms more strongly than relative viral DNA levels.

Both peptide aptamers appeared to have a stronger effect on symptoms than on viral DNA accumulation during TYLCV and ToMoV infection. To examine this observation further, we determined the overall distribution of low versus high symptoms and viral DNA loads (Fig. 8). For this analysis, we calculated the percentage of TYLCV-inoculated plants with symptom scores of 0, 1, 2, 3, or 4 at 14, 21, and 28 dpi (Fig. 8A). To perform a similar analysis for viral DNA loads, relative viral DNA levels were divided in 5 groups (0 to 5, 6 to 25, 26 to 50, 51 to 75, and 76 to 100+), and the percentage of plants in each group was determined at the different time points (Fig. 8B). We pooled the data for resistant peptide aptamer lines into two categories (A22 and A64) across two experiments and compared them to wild-type and TRX-GST plants from the same experiments. The analysis included 39 wild-type plants, 27 TRX-GST plants, 54 A22 plants from three transgenic events and 90 A64 plants from five transgenic events inoculated with TYLCV. A similar analysis was not performed for ToMoV because of the smaller size of the available data set.

Fig 8.

Relative distributions of TYLCV symptoms and viral DNA accumulation in transgenic lines. (A and B) The distributions for symptom scores (A) and relative viral DNA loads (B) at 14, 21, and 28 dpi are shown as the percentage of wild-type (WT), TRX-GST, A22, and A64 plants in each group. The groups and corresponding colors are shown in the key below the graphs. The transgenic lines (A22-5, A22-56, A22-82, A64-13, A64-21, A64-47, A64-107, and A64-157) shown were those showing the best resistance.

At 14 dpi, 69% of wild-type plants and 52% of TRX-25 plants showed severe symptoms with scores ≥ 3, while only 2% of A22 plants and 3% of A64 plants had high scores. At the other end of the symptom scale, no wild-type plants and only 4% of TRX-25 plants had symptom scores ≤ 1, while 63% of A22 plants and 83% of A64 plants had low symptom scores. At 21 and 28 dpi, nearly all of the wild-type and TRX-25 plants showed severe symptoms (scores ≥ 3) with no plants scoring ≤ 1. At 28 dpi, 72% of wild-type plants and 59% of TRX-25 plants had symptom scores of 4. In contrast, none of the A22 or A64 plants achieved level 4 symptoms during the experiments, and less than a third were scored as level 3 at the later time points. Instead, half of the A22 and A64 plants had symptom scores ≤ 1 at 28 dpi, with 28% of A22 plants and 27% of A64 plants remaining asymptomatic throughout the experiment. The symptom score distribution demonstrated that TYLCV symptoms are strongly delayed and attenuated in tomato plants expressing A22 or A64 compared to wild-type and TRX-25 plants.

We also observed a delay in viral DNA accumulation in the A22 and A64 plants. At 14 dpi, 90% of wild-type plants and 62% of TRX-GST plants were in the 51-to-75 and 76-to-100+ groups, and none were in the 0-to-5 group. In contrast, only 13% of A22 plants and 29% of A64 plants had high levels of viral DNA, and 24% of A22 plants and 28% of A64 plants were in the 0-to-5 group. Over the course of the experiment, the percentage in the top two groups increased to 61% for A22 plants and 47% for A64 plants but was still lower than wild-type (93%) and TRX-25 (92%) plants. At 28 dpi, 21% of A22 plants and 32% of A64 plants were in the 0-to-5 or 6-to-25 group, while no wild-type or TRX-25 plants were in these groups.

We then examined the relationship between viral DNA load and symptom score at 14, 21, and 28 dpi. The TYLCV-inoculated plants were grouped based on their symptom scores, and the average relative viral DNA level was determined for each symptom score at the three time points (Fig. 9). At 14 dpi, wild-type plants contained higher levels of viral DNA at all symptom scores than TRX-GST plants, but this difference was not apparent at 21 and 28 dpi. Wild-type and TRX-GST plants with low symptom scores contained less viral DNA than those with high scores at 14 dpi, but this was not seen at the later time points. Asymptomatic A22 and A64 plants (symptom score = 0) contained much lower levels of viral DNA at all three time points. In contrast, A22 plants with symptom scores of 1 to 3 contained similar levels of viral DNA, while the A64 plants showed increasing levels of viral DNA as the severity of symptoms increased at 14, 21, and 28 dpi.

Fig 9.

Relationship between viral DNA load and symptom score in transgenic lines infected with TYLCV. (A to C) Wild-type (WT), TRX-GST, A22, and A64 TYLCV-inoculated plants were grouped based on their symptom scores, and the average relative viral DNA level was determined for each symptom score at 14, 21, and 28 dpi. The number of plants included in each column is shown below each column. The symptom scores for the columns are shown in the key at the bottom of the figure.

Comparison of the distributions in Fig. 8A and B supported our perception that the aptamers have a stronger effect on symptoms than on viral DNA accumulation. It is striking that none of the A22 or A64 plants achieved level 4 symptoms, even though half had high levels of viral DNA comparable to the controls at 28 dpi. This difference may indicate that lower levels of viral DNA early in the infection process reduce symptom severity later. Alternatively, the differences in symptoms and viral DNA loads may be due to the disruption of an activity of Rep that is not related to its role in viral replication.

DISCUSSION

Geminiviruses cause diseases in food and cash crops, creating economic and socioeconomic problems worldwide (75). The abilities of geminiviruses to infect a wide variety of crops and to adapt to new hosts and geographic areas, along with their high mutation and recombination rates and association with nonspecific satellite molecules, provide a huge number of evolutionary possibilities for the emergence of new viral diseases. These properties coupled with the propagation of insect vectors and infected plant material negatively impact the breadth and durability of many resistance strategies. To overcome these challenges, we created a novel resistance strategy using peptide aptamers to target the N terminus of the geminivirus Rep protein. Rep is an essential player in geminivirus infection, and its sequence is highly conserved across geminivirus genera, making it an excellent candidate for developing broad-based disease resistance. Peptide aptamers can bind to proteins and interfere with their function in bacterial, yeast, animal, and plant systems (47). They can also interact with viral proteins and act as antiviral agents (49, 53, 76), and they have been used to confer broad-spectrum resistance to tospoviruses in N. benthamiana (51). In this report, we identified peptide aptamers that target the geminivirus Rep protein and interfere with viral function in planta.

We selected 16 peptide aptamers that were shown previously to interact with TGMV and CaLCuV Rep (49) for analysis in yeast and in tomato plants. When semiquantitative yeast two-hybrid assays were used to compare their relative strengths of binding to TGMV Rep, the highest binding activities were seen with A22, A59, A64, and A84 peptide aptamers. Although all four peptide aptamers contain aptamer sequence motifs 4, 20, and/or 24 (Table 1), they do not have any aptamer motifs in common, suggesting that they interact differently with Rep. This idea was supported by the mapping results for A22 and A64, the two peptide aptamers with the strongest binding activities. A22 requires the first 35 amino acids of Rep containing the RCR motif I for maximal binding, while A64 does not. Instead, A64 interacts primarily with Rep residues between residues 64 and 97 that include the GRS. Both A22 and A64 display weak binding to Rep residues 36 to 63, consistent with the peptide aptamers binding to an extended surface and/or two sites in the Rep N terminus. However, the data clearly indicated that the primary Rep contacts of A22 and A64 are distinct, raising the possibility that the two peptide aptamers disrupt Rep function via different mechanisms and potentially interfere with different activities of the multifunctional viral replication protein.

We used yeast two-hybrid growth assays to examine the abilities of the different peptide aptamers to bind to a diverse set of Rep proteins representing the three major Geminiviridae genera (Tables 2 and 3). We detected binding to all 9 Rep proteins for 5 peptide aptamers (A22, A147, A59, A84, and A46) and to 8 of 9 Reps for 10 aptamers (A64, A160, A159, A116, A40, A127, A135, A176, A132, and A155). These results were striking given the diversity of the 9 Rep proteins and strongly suggested that 15 of the 16 peptide aptamers interact with conserved sequence and/or structural motifs in the Rep N terminus. It is also noteworthy that with the exception of TGMV, all of the Rep proteins are from viruses associated with significant disease problems, indicating that the peptide aptamers can be used to target agronomically important viruses.

A22 and A64 peptide aptamers had several properties that suggested that they might confer broad-based geminivirus resistance in planta. Both bind strongly to TGMV Rep and interact with diverse Rep proteins. They bind to different regions in the Rep N terminus and may interfere with Rep activity differently. Importantly, both A22 and A64 interact with the Rep proteins of TYLCV and ToMoV, allowing us to test the peptide aptamer strategy in tomato, which is easily transformable and severely impacted by geminivirus disease in the field (72). We readily recovered transgenic tomato lines expressing A22, A64, or TRX-GST, which does not bind to Rep and served as a control for any effects due to the thioredoxin scaffold. The transgenic plants appeared phenotypically normal, indicating that the Rep-binding peptide aptamers are unlikely to interact with plant proteins and cause pleiotropic effects. In TYLCV infection assays, T1 plants expressing either A22 or A64 displayed attenuated symptoms and reduced viral DNA levels relative to nontransgenic tomato plants. A large fraction of the A22 and A64 lines showed a resistant phenotype, but the strength of the resistance varied widely. The best lines (A22-5, A22-82, A64-21, A64-47, and A64-157) showed little or no symptoms and had significantly reduced viral DNA loads, with about 25% of the plants showing no evidence of infection. Strikingly, 4 of the best performing lines against TYLCV also displayed reduced symptoms and viral DNA levels when inoculated with ToMoV. Although the effects against ToMoV were most apparent at 7 dpi, only a few plants achieved level 4 symptoms at 21 dpi, and 25% had symptom scores of 0 or 1. The weaker effect on ToMoV compared to TYLCV may reflect differences in the infection processes of the two viruses as evidenced by the different timing and nature of the symptoms. It is also possible that some of the A22 and A64 lines that were not tested in ToMoV assays would have exhibited stronger resistance phenotypes. However, every peptide aptamer line included some plants with symptoms or viral DNA, indicating that neither A22 nor A64 confer immunity to TYLCV or ToMoV infection.

Analysis of the TRX-GST plants suggested that the scaffold slows down viral infection early on. We often observed milder symptoms and less viral DNA in TRX-GST plants versus wild-type tomato at the earliest time points, but the differences were rarely statistically significant. Effects due to TRX-GST expression were not apparent at later time points. Overexpression of tobacco TRXh reduces RNA virus infection in tobacco, most likely by altering the cellular redox status leading to activation of the salicylic acid (SA) pathway (77). Expression of genes in the SA pathway is activated by geminivirus infection, and constitutive expression of the SA effector protein PR1 inhibits infection (31). The TRX scaffold containing the peptide aptamers is not catalytically functional due to the insertion of the peptide sequence in the active site, and as such, it is unlikely to upregulate the SA pathway. The 20-amino-acid sequence from GST in TRX-GST might impact host resistance, but this also seems unlikely given the small size of the peptide. Our results clearly show that any contribution of the scaffold is transient and that the Rep-binding peptide is the major contributor to the resistance phenotype.

Expression of A22 or A64 has a stronger effect on symptoms than on viral DNA accumulation (Fig. 8). We inoculated tomato plants at the two-leaf stage because young plants are more susceptible to geminivirus infection and develop stronger symptoms. Thus, one potential explanation of the stronger effect on symptoms is that a delay in viral DNA accumulation in very young plants would allow time for them to become more mature and less susceptible to viral effects manifested at the level of symptoms. This idea is consistent with the observation that TYLCV- or ToMoV-inoculated plants with no apparent symptoms or attenuated symptoms at 28 or 21 dpi, respectively, did not show stronger infection phenotypes when monitored for extended time periods of up to 2 months. In addition, there were no apparent differences with respect to flowering and fruit/seed production between mock-inoculated, wild-type plants and virus-inoculated A22 or A64 plants.

The differences in symptoms and viral DNA loads may also be due to the disruption of other Rep functions that are not related to its role in viral replication. Rep modulates many host functions through interactions with plant proteins. To date, two host factor binding sites have been located to or shown to overlap with the Rep N terminus. RBR, which is involved in plant cell cycle regulation and development (78), binds to Rep between residues 103 and 180 (79), but its primary binding site is between amino acids 144 and 154 in TGMV Rep (80). Thus, it is unlikely that Rep binding to either A22 or A64 impacts its interactions with RBR. In contrast, the SUMO-conjugating enzyme SCE1 binds between residues 85 and 114 in the TGMV Rep N terminus (81) such that A64 binding between residues 64 and 97 may interfere with Rep-SCE interactions. Interestingly, impaired Rep-SCE interactions reduce both viral DNA accumulation and symptoms (81), as was observed for A64 plants inoculated with TYLCV (Fig. 9). The apparent lack of correlation between symptoms and viral DNA levels for A22 plants is consistent with the peptide aptamer interfering with some other aspect of Rep function that is separate from replication. A recent study showed that ectopic expression of the Rep proteins of three geminiviruses in N. benthamiana causes leaf curling and yellowing characteristic of viral symptoms (82). One interesting possibility is that the extreme N terminus of Rep is involved in symptom production and A22 binding to this region interferes with a yet-to-be identified host interaction necessary for symptoms.

The experiments reported here represent the first step in the application of Rep-binding peptide aptamers to combat geminivirus disease. A number of parameters can be optimized to improve the effectiveness of the strategy. First, the T1 plants used in our studies were heterozygous for the peptide aptamer transgene. Homozygous plants may be more resistant if there is a dose-response effect of the transgene, and it will be important to challenge T2 homozygous plants in future experiments. Analysis of T2 plants will also confirm whether transgene expression is stable. Second, although some geminiviruses can invade mesophyll and epidermal cells (31, 32), many are limited to the phloem, and all must pass through vascular tissue during systemic movement and insect vector acquisition (7). Hence, expressing the peptide aptamers under the control of a phloem-specific promoter might enhance their effectiveness (83, 84). Third, A22 and A64 bind to different regions of the Rep N terminus and may act through different mechanisms. Therefore, their effects may be additive such that plants expressing both peptide aptamers are more resistant to infection as long as measures are taken to prevent gene silencing due to their common scaffold sequences.

We demonstrated the efficacy of using peptide aptamers that bind to viral Rep protein to confer broad-based resistance to geminivirus infection. Our data suggested that the peptide aptamers bind to conserved motifs in the Rep N terminus that are required for function (42, 85), reducing the likelihood of the evolution of Rep variants that do not interact with them. As a consequence, coupling the peptide aptamer strategy with existing resistance technologies with inherent specificity constraints provides a mechanism to extend the value of these technologies. For example, the success of creating resistant cultivars by conventional breeding has been limited because the resistance can be broken by the appearance of a recombinant virus or by an unfavorable environment where many viruses are present (17). Combining the peptide aptamers with conventional resistance would combat new geminivirus species or variants while maintaining conventional resistance against known viruses. RNA interference (RNAi), the primary technology used to generate transgenic resistance against plant viruses, is homology dependent and works only with closely related viruses (8). Pairing the peptide aptamer technology with successful RNAi resistance would alleviate this limitation while maintaining established resistance to closely related geminiviruses. The peptide aptamer approach can also be expanded to other geminivirus proteins and viral processes to enhance the resistance phenotype.