Abstract

   Maize rough dwarf disease (MRDD) threatens maize production globally.
   The P7-1 effector of the rice black-streaked dwarf virus (RBSDV)
   targets maize Rab GDP dissociation inhibitor alpha (ZmGDIα) to cause
   MRDD. However, P7-1 has difficulty recruiting a ZmGDIα variant with an
   alternative helitron-derived exon 10 (ZmGDIα-hel), resulting in
   recessive resistance. Here, we demonstrate that P7-1 can recruit
   another maize protein, gibberellin 2-oxidase 13 (ZmGA2ox7.3), which
   also exhibits tighter binding affinity for ZmGDIα than ZmGDIα-hel. The
   oligomerization of ZmGA2ox7.3 is vital for its function in converting
   bioactive gibberellins into inactive forms. Moreover, the enzymatic
   activity of ZmGA2ox7.3 oligomers increases when forming
   hetero-oligomers with P7-1/ZmGDIα, but decreases when ZmGDIα-hel
   replaces ZmGDIα. Viral infection significantly promotes ZmGA2ox7.3
   expression and oligomerization in ZmGDIα-containing susceptible maize,
   resulting in reduced bioactive GA[1]/GA[4] levels. This causes an
   auxin/cytokinin imbalance and ultimately manifests as MRDD syndrome.
   Conversely, in resistant maize, ZmGDIα-hel counters these virus-induced
   changes, thereby mitigating MRDD severity.

   Subject terms: Virulence, Gibberellins, Microbe
     __________________________________________________________________

   RBSDV infection increases ZmGA2ox7.3 expression and oligomerization in
   ZmGDIα-containing maize, leading to reduced bioactive gibberellins and
   MRDD syndrome. In resistant maize, ZmGDIα-hel attenuates these effects,
   thereby reducing MRDD severity.

Introduction

   Plant viruses, primarily transmitted by biotic vectors, pose a
   significant threat to global food security by inflicting substantial
   damage to crops^[42]1. To combat viral infections, plants have evolved
   multiple layers of defense responses, which are collectively referred
   to as active resistance^[43]2. In addition, plants have developed
   passive resistance strategies, enabling them to evade viral recognition
   or exploitation by modifying or eliminating susceptibility factors,
   also referred to as recessive resistance^[44]3. Most of the
   naturally-occurring recessive resistance genes are related to the
   translation initiation factors eIF4E and eIF4G or their homologs^[45]4.
   Other loss-of-susceptibility genes include ZmGDIα-hel, cPGK2,
   HvPDIL5-1, and among others^[46]5–[47]7.

   Maize rough dwarf disease (MRDD) is a devastating viral disease, often
   referred to as the “cancer” of maize. It was first reported in the late
   1940s in northern Italy and subsequently spread to countries such as
   Greece, France, and Israel^[48]8. MRDD is caused by various Fijivirus
   pathogens in the family Reoviridae, among which the maize rough dwarf
   virus was detected in Europe, Mal de Río Cuarto virus was reported in
   South America, and the major pathogen responsible in East Asia is the
   rice black-streaked dwarf virus (RBSDV)^[49]9,[50]10. RBSDV is
   transmitted to maize in a persistent manner by the small brown
   planthopper (Laodelphax striatellus)^[51]11. The typical symptoms of
   MRDD are characterized by severe growth abnormalities, including
   dwarfism, shortened internodes, abnormal ears/tassels, dark-green
   leaves, and white waxy enations on the backs of leaf veins^[52]12.
   RBSDV has a genome of 10 double-stranded RNAs (S1–S10), with sizes
   ranging from 1.8 to 4.5 kb^[53]13,[54]14. Each segment encodes one or,
   in the case of S5, S7, and S9, two proteins^[55]13. Notably, the P7-1
   effector, encoded by S7-1, forms tubular structures in RBSDV-infected
   plants and localizes to the plasmodesmata, which utilizes both the
   secretory pathway and actomyosin motility system for intracellular
   virus transport and movement^[56]15,[57]16.

   Phytohormones regulate plant growth, development, and stress
   responses^[58]17,[59]18. Viral infections can disrupt the balance of
   phytohormones, leading to disease syndromes^[60]17. In Arabidopsis, the
   replicase protein of the tomato mosaic virus (TMV) interacts with
   Aux/IAA, disrupting the auxin response pathway to manifest
   symptoms^[61]19. In rice dwarf virus (RDV) disease, the P2 protein
   interacts with ent-kaurene oxidase to reduce rice GA[1] levels,
   resulting in a dwarf phenotype^[62]20. The P2 protein encoded by rice
   stripe virus interacts with and degrades rice OsNPR1, thereby
   disrupting the OsNPR1-mediated interplay between salicylic acid and
   jasmonic acid, ultimately facilitating viral infection^[63]21. In
   Arabidopsis, the geminivirus AC2/AL2 protein can interact with
   adenosine kinase and lead to an increase the expression of primary
   cytokinin-responsive genes^[64]22. Sw-5b, a plant leucine-rich repeat
   immune receptors (NLR), directly utilizes the ABA central regulator
   PP2C-SnRK2 complex to activate an ABA-dependent defense against viral
   pathogens^[65]23.

   Gibberellins (GAs) are a class of tetracyclic diterpenoid hormones that
   regulate plant stress responses, seed germination/dormancy, stem
   elongation, and leaf expansion^[66]24. In plants, GA[4] (a type of 13-H
   GAs), GA[1] (13-OH GA[4]), GA[7] (another type of 13-H GAs), and GA[3]
   (13-OH GA[7]) are considered major bioactive types^[67]25. Among these,
   GA[4] exhibits higher bioactivity than GA[1], and gibberellin
   2-oxidases (GA2oxs) play a pivotal role in inactivating both GA[1] and
   GA[4]^[68]25–[69]27. GA2oxs are categorized into two groups,
   C[19]-GA2oxs (also divided into class I and class II) and C[20]-GA2oxs,
   based on their substrate specificity and enzymatic activity^[70]26.
   Overexpression of C[19]-GA2oxs leads to a reduction of endogenous
   bioactive GAs, causing GA-deficient phenotypes^[71]26,[72]28.

   In our previous work, we identified a helitron-induced variant
   ZmGDIα-hel that confers quantitative recessive resistance to MRDD by
   reducing disease severity index (DSI) by 24.2%–39.3%^[73]5. However,
   the detailed mechanism behind this resistance remains elusive.
   Specifically, we aim to understand how the interaction between the
   viral P7-1 effector and the host susceptibility factor (ZmGDIα)
   initiates MRDD and how the weaker interaction with ZmGDIα-hel
   alleviates MRDD? In this study, we identified a gibberellin 2-oxidase,
   ZmGA2ox7.3, as a key player in MRDD. During RBSDV infection, changes in
   the levels of ZmGA2ox7.3, ZmGDIα/ZmGDIα-hel, and P7-1, as well as their
   interactions, are crucial for modulating GA levels and, consequently,
   the balance of auxin/cytokinins. These factors ultimately determine the
   outcome of MRDD in maize.

Result

RBSDV-induced transcriptome reprogramming

   In our previous study, we generated a pair of near-isogenic lines
   (NILs): the susceptible NIL-S (with the susceptibility allele, ZmGDIα)
   and the resistant NIL-R (with the resistance allele, ZmGDIα-hel)^[74]5.
   Following RBSDV infection, viral accumulation was detected in NIL-S
   starting from 16 days post inoculation (dpi) and increased linearly
   thereafter to reach a high level^[75]5. Consistent with this
   observation, we found that the viral S10, encoding the coat protein,
   exhibited dramatic accumulation in NIL-S but not in NIL-R
   (Fig. [76]1a). Similarly, the viral S7-1 also showed significant
   accumulation (Fig. [77]1b). We performed RNA-seq analysis at 16 dpi
   between NIL-S and NIL-R. The differentially expressed genes (DEGs) were
   identified based on the criteria of fold-change > 2 and P < 0.05.
   Compared to NIL-R, NIL-S exhibited 24 downregulated and 21 upregulated
   DEGs (Fig. [78]1c, d). Gene Ontology (GO) term analysis revealed that
   these DEGs are mainly enriched in one molecular process: oxidoreductase
   activity, and in three biological processes: biology,
   oxidation-reduction, and single-organism (Supplementary Fig. [79]1a).
   KEGG analysis demonstrated significant DEG enrichment in three
   pathways: photosynthesis-antenna proteins, fatty acid elongation, and
   diterpenoid biosynthesis (Supplementary Fig. [80]1b).

Fig. 1. Transcriptome analysis of differentially expressed genes (DEGs)
induced by RBSDV between the two NILs.

   [81]Fig. 1
   [82]Open in a new tab

   a, b Relative expression of RBSDV S10 and S7-1 (n = 3) between the two
   NILs 16 days post RBSDV inoculation (dpi). A biological replicate in
   the NIL-R serves as a control sample for relative expression. c A
   volcano plot depicts the DEGs when comparing NIL-S to NIL-R at 16 dpi.
   Genes with significant down-regulation are represented by green dots,
   those with significant up-regulation by red dots, and those without
   significant change by blue dots. d A heatmap displays the 45 DEGs
   induced by RBSDV when comparing NIL-S to NIL-R. Red indicates
   up-regulation, while green denotes down-regulation. The red asterisk
   denotes ZmGA2ox7.3. e Enrichment of the ZmGA2ox7.3 transcript in NIL-S
   compared to NIL-R. A structure diagram of ZmGA2ox7.3 is provided at the
   bottom. f The relative expression of ZmGA2ox7.3 was verified by RT-qPCR
   (n = 3). A biological replicate in the non-infected NIL-S at 27 dpi
   serve as a control sample for relative expression. In (d, e), “rep1”
   and “rep2” indicate two biological replicates. In (a, b, f), data are
   presented as means ± s.e.m. Significance was determined by a two-tailed
   Student’s t test.

Screening of potential P7-1 targets required for RBSDV infection

   Four (ZmGA2ox7.3, ZmDAO1, ZmB1B2, and ZmPPI1) out of the 45 DEGs were
   captured in a previous Co-immunoprecipitation and mass spectrometry
   (Co-IP/MS) assay (Supplemental Table [83]1)^[84]5. Notably, the
   ZmGA2ox7.3 gene (Zm00001d039394), encoding a gibberellin 2-oxidase 13,
   was dramatically upregulated in NIL-S compared to NIL-R (Fig. [85]1d,
   e). In the reverse transcription-quantitative real-time PCR (RT-qPCR)
   assay, ZmGA2ox7.3 expression was significantly induced by RBSDV at 16
   dpi in NIL-S, while it only showed a slight increase in NIL-R. However,
   no induction of ZmGA2ox7.3 expression was observed in either NIL at 27
   dpi (Fig. [86]1f). Regarding the other three DEGs, ZmDAO1 (encoding a
   dioxygenase for auxin oxidation 1) exhibited downregulation in NIL-S in
   both RNA-seq and RT-qPCR analyses. ZmB1B2 (encoding photosystem II
   light harvesting complex) and ZmPPI1 (encoding peptidyl-prolyl
   cis-trans isomerase protein) showed no significant differential
   expression between the two NILs in the RT-qPCR, contrary to the RNA-seq
   data (Fig. [87]1d and Supplementary Fig. [88]1c, d). We further
   investigated potential interactions between these four proteins and
   ZmGDIα/ZmGDIα-hel using the bimolecular fluorescence complementation
   (BiFC) assay. ZmGA2ox7.3 showed a strong interaction with ZmGDIα but
   weaker with ZmGDIα-hel. In contrast, ZmDAO1 did not interact with
   either ZmGDIα or ZmGDIα-hel. ZmB1B2 exclusively interacted with ZmGDIα,
   and ZmPPI1 solely with ZmGDIα-hel (Supplementary Fig. [89]2a–e). All
   these findings were confirmed in the in vitro pull-down assays
   (Supplementary Fig. [90]2f and Fig. [91]4b).

Fig. 4. The interactions of ZmGA2ox7.3 with ZmGDIα, ZmGDIα-hel, and P7-1.

   [92]Fig. 4
   [93]Open in a new tab

   a Validation of the interactions of ZmGA2ox7.3 with ZmGDIα, ZmGDIα-hel,
   or P7-1 using an in vivo SLC assay. The color scale depicts the
   intensity of the luminescence signals. b Confirmation of the
   interactions of ZmGA2ox7.3 with ZmGDIα, ZmGDIα-hel or P7-1 using a
   Co-IP assay. c Confirmation of the interactions of ZmGA2ox7.3 with
   ZmGDIα, ZmGDIα-hel, or P7-1 using an in vitro GST-pull down assay. d
   Confirmation of the interactions of ZmGA2ox7.3 with ZmGDIα, ZmGDIα-hel,
   or P7-1 using an in vivo BiFC assay. Reconstituted YFP fluorescence
   signals were visualized using confocal microscopy and captured at 72 h
   post inoculation (hpi). Scale bar, 50 µm. The Co-IP, pull-down and BiFC
   assays were repeated at least four times.

   A previous report in rice stated that the overexpression of gibberellin
   2-oxidase genes resulted in stunted plants and shortened
   internodes^[94]27,[95]29, resembling typical MRDD symptoms. Combining
   expression patterns, protein-protein interaction profiles, and
   predicted functions, ZmGA2ox7.3 is likely another P7-1 target necessary
   for RBSDV infection.

Overexpression of ZmGA2ox7.3 causes dwarf plant

   We introduced an overexpression construct pUbi:ZmGA2ox7.3-GFP into the
   B73 recipient line, resulting in seven independent transgenic events.
   Compared to B73, the T[2] transgenic lines exhibited significantly
   enhanced expression of ZmGA2ox7.3, concomitant with plant dwarfism and
   internode shortening (Fig. [96]2a–e). The degree of internode
   shortening progressively intensified from the bottom to the top of the
   plant (Fig. [97]2f). When examining longitudinal sections of the
   uppermost internode, we observed that the average cell length in the
   T[2] transgenic lines was significantly shorter than in B73
   (Fig. [98]2g, h). Subcellular localization using transient expression
   assays in Nicotiana tabacum leaves and maize protoplasts indicated that
   ZmGA2ox7.3 is localized to the plasma membrane, cytoplasm, and nucleus
   (Supplementary Fig. [99]2g, h).

Fig. 2. Overexpression of ZmGA2ox7.3 causes dwarf plants.

   [100]Fig. 2
   [101]Open in a new tab

   a Schematic diagram of the overexpression construct
   pUbi:ZmGA2ox7.3-GFP. b, c Confirmation of seven transgenic events
   overexpressing ZmGA2ox7.3 (n = 3). The relative expression levels of
   ZmGA2ox7.3 in the transgenic lines (b). Immunoblot analysis of the
   exogenous ZmGA2ox7.3-GFP using an anti-GFP antibody, and of the actin
   protein using an anti-actin antibody (c). The B73 recipient line served
   as a negative control. A biological replicate in the B73 serve as a
   control sample for relative expression (b). d, e Plant heights of the
   transgenic lines overexpressing ZmGA2ox7.3. Statistical analysis of
   plant heights comparing B73 with the transgenic lines (d). The plant
   morphology of both B73 and the transgenic lines (e). Scale bar, 30 cm.
   f Measurement of internode lengths of B73 (n = 6) and ZmGA2ox7.3
   overexpression line OE-5 plants (n = 7), starting from the uppermost
   internode (the first stem node) and progressing downwards. g, h
   Characteristics of cells in the uppermost internode. Images of
   transections comparing B73 with OE-5 (g). Statistical analysis of cell
   lengths between B73 and OE-5 (h). Scale bar, 500 μm (white dashed
   line). In (b, f), data represent the mean ± s.e.m, individual data
   points are depicted. In (d), box plots are displayed. Error bars
   represent the maximum and minimum values; the central lines show the
   medians; and the box edges denote the 25th and 75th percentiles. In
   (h), data are shown as violin plots with individual data points.
   Statistical differences (P < 0.05) are highlighted using different
   lowercase letters, as determined by Kruskal-Wallis test (b, d).
   Significance levels are determined by a two-tailed Student’s t-test (f)
   and two-sided Mann–Whitney U test (f, h).

   To characterize ZmGA2ox7.3 further, we retrieved 24 GA2-oxidases in
   rice and maize from the NCBI public database
   ([102]https://www.ncbi.nlm.nih.gov/) and MaizeGDB database
   ([103]https://www.maizegdb.org/). Phylogenetic analysis revealed three
   distinct clades. ZmGA2ox7.3 shares its highest amino-acid identity with
   OsGA2ox7, followed by OsGA2ox3, in rice. These three proteins belong to
   the C[19]-GA2oxs class I family, which functions to catalyze the
   conversion of biologically active GAs (GA[1] and GA[4]) and their
   precursors (GA[9] and GA[20]) into inactive GA forms (GA[8], GA[34],
   GA[51], and GA[29]) (Supplementary Fig. [104]3a–c)^[105]25,[106]26.
   This was exemplified by rice OsGA2ox3, which demonstrated its
   capability to catalyze the conversion of bioactive GA[4] into inactive
   GA[34]^[107]27.To ascertain whether ZmGA2ox7.3 belongs to the
   C[19]-GA2oxs class I family, we tested it with three active
   gibberellins: GA[1], GA[3], and GA[4]. Of them, GA[1] and GA[4] serve
   as typical substrates for GA2oxs. GA[3] is the product converted from
   GA[5], which in turn is derived from GA[20], and both conversions are
   catalyzed by GA3oxs^[108]26. We sprayed GA[1], GA[3], and GA[4] on B73
   and its transgenic line overexpressing ZmGA2ox7.3 at their seedling
   stages. B73 showed increased plant height upon spraying with these
   active GAs. Notably, the dwarf phenotype of the transgenic line could
   be fully restored with exogenous GA[3], while it was partially restored
   with GA[4] and to a lesser extent with GA[1] (Fig. [109]3a, b). In B73,
   all three active GAs significantly stimulated the expression of
   ZmGA2ox7.3; however, in the transgenic line, they all significantly
   lowered ZmGA2ox7.3 expression (Fig. [110]3c). Additionally, GA[4] led
   to more accumulation of ZmGA2ox7.3 oligomers (Fig. [111]3d). These
   findings imply an inherent connection between ZmGA2ox7.3 and GA[4] or
   GA[1], consistent with the role of ZmGA2ox7.3 as a member of the
   C[19]-GA2oxs family.

Fig. 3. The phenotype of overexpression of ZmGA2ox7.3 to treatments with
exogenous GAs.

   [112]Fig. 3
   [113]Open in a new tab

   a, b Responses of B73 and OE-4 to treatments with GA[1], GA[3] or
   GA[4]. The plant morphology before and after GA[1], GA[3] or GA[4]
   treatment (a). Statistical analysis of plant heights between B73 and
   OE-4 (b). Mock, 0.1% Tween 20 served as a negative control. Scale bar,
   10 cm. c The relative expression levels of ZmGA2ox7.3 in B73 and OE-4
   before and after treatments with GA[1], GA[3] and GA[4], respectively
   (n = 3). A biological replicate in the B73 mock serve as a control
   sample for relative expression. d The exogenous ZmGA2ox7.3-GFP proteins
   of OE-4 treated with GA[4], GA[1] or GA[3] were subjected to 4–13%
   BN-PAGE and SDS-PAGE, followed by immunoblotting with an anti-GFP
   antibody, and of the actin protein immunoblotting with an anti-actin
   antibody, respectively. The B73 recipient line served as a negative
   control. In (b), data are shown as violin plots with individual data
   points. In (c), data represent the mean ± s.e.m, individual data points
   are depicted. Statistical differences (P < 0.05) are highlighted using
   different lowercase letters, as determined by Kruskal–Wallis test (b,
   c).

ZmGA2ox7.3 binds to ZmGDIα/P7-1 to form the hetero-oligomer

   In split luciferase complementation (SLC) assays, luminescence signals
   were observed upon co-expression of cLUC-ZmGA2ox7.3 with each of the
   three fusion genes (ZmGDIα-nLUC, ZmGDIα-hel-nLUC, and P7-1-nLUC) in
   Nicotiana benthamiana leaves (Fig. [114]4a). The co-immunoprecipitation
   (Co-IP) assay confirmed the associations of ZmGA2ox7.3 with each of the
   three proteins (Fig. [115]4b). The in vitro pull-down assay
   demonstrated that ZmGA2ox7.3 had a stronger interaction with ZmGDIα and
   P7-1, but only weaker with ZmGDIα-hel (Fig. [116]4c). The BiFC assay
   further confirmed that ZmGA2ox7.3 interacted more strongly with ZmGDIα
   and P7-1, and less so with ZmGDIα-hel (Fig. [117]4d).

   We next conducted multiple SLC and micro-scale thermophoresis (MST)
   assays to precisely measure the binding strengths of ZmGA2ox7.3 with
   two ZmGDIα isoforms and P7-1. ZmGA2ox7.3 exhibited stronger luciferase
   (LUC) activity when interacting with ZmGDIα than with ZmGDIα-hel.
   Despite the lower quantity of P7-1, the interaction between ZmGA2ox7.3
   and P7-1 also displayed strong LUC activity, albeit lower than that
   between ZmGA2ox7.3 and ZmGDIα (Supplementary Fig. [118]4a–f). MST
   assays definitely indicated that ZmGA2ox7.3 exhibits the highest
   affinity for P7-1, followed by ZmGDIα, and the lowest affinity for
   binding to ZmGDIα-hel (Supplementary Fig. [119]4g). To pinpoint the
   specific regions of ZmGDIα/ZmGDIα-hel that bind to ZmGA2ox7.3, we
   divided the ZmGDIα and ZmGDIα-hel cDNAs into multiple segments, as done
   in our previous study^[120]5. In SLC assays, ZmGA2ox7.3 demonstrated
   interactions with three segments of ZmGDIα, all of which included exon
   10 (ZmGDIα^1-333, ZmGDIα^292-445, and ZmGDIα^292-333). This indicates
   the involvement of exon 10 in the interaction between ZmGA2ox7.3 and
   ZmGDIα (Supplementary Fig. [121]5a). Similarly, interactions between
   ZmGA2ox7.3 and three ZmGDIα-hel counterparts (ZmGDIα-hel^1-318,
   ZmGDIα-hel^292-430, and ZmGDIα-hel^292-318) were also observed,
   implying that the helitron-derived exon 10 did not completely abolish
   its ability to interact with ZmGA2ox7.3 (Supplementary Fig. [122]5b).

   Our subsequent focus shifted to ascertain whether any other regions of
   ZmGDIα might also engaged in interactions with ZmGA2ox7.3.
   Intriguingly, both the N-terminus (ZmGDIα^1-291) and C-terminus
   (ZmGDIα^334-445) flanking exon 10 exhibited interactions with
   ZmGA2ox7.3 (Supplementary Fig. [123]5c). We then subdivided
   ZmGDIα^1-291 into two segments: ZmGDIα^1-145 and ZmGDIα^146-291. The
   LUC activity was observed when ZmGA2ox7.3 was co-expressed with
   ZmGDIα^146-291, but not with ZmGDIα^1-145 (Supplementary Fig. [124]5c).
   Consistent with our previous study^[125]5, the same four segments
   (ZmGDIα^334-390, ZmGDIα^334-378, ZmGDIα^379-445, and ZmGDIα^391-445)
   were co-expressed with ZmGA2ox7.3. Except for ZmGDIα^334-378, all the
   segments exhibited LUC activity, indicating that the C-terminal
   ZmGDIα^379-445 region can bind to ZmGA2ox7.3 (Supplementary
   Fig. [126]5c). Moreover, when the four fragments containing the
   ZmGDIα^334-378 segment (ZmGDIα^292-390, ZmGDIα^292-378,
   ZmGDIα-hel^292-375, and ZmGDIα-hel^292-363) were individually
   co-expressed with ZmGA2ox7.3, LUC activity was observed in all cases
   (Supplementary Fig. [127]5d). This indicates that ZmGDIα^334-378 does
   not hinder its flanking regions from interacting with ZmGA2ox7.3.

   To validate these interactions, we selected four segments,
   ZmGDIα^1-145, ZmGDIα^146-333, ZmGDIα-hel^146-318, and ZmGDIα^379-445,
   to test their interactions with ZmGA2ox7.3. In the pull-down assay,
   ZmGA2ox7.3 interacted strongly with ZmGDIα^146-333 and ZmGDIα^379-445,
   but not with ZmGDIα^1-145 or ZmGDIα-hel^146-318 (Supplementary
   Fig. [128]5e). Similarly, in the BiFC assay, ZmGA2ox7.3 interacted
   strongly with ZmGDIα^146-333 and ZmGDIα^379-445, weakly with
   ZmGDIα-hel^146-318, but not at all with ZmGDIα^1-145 (Supplementary
   Fig. [129]5f). Hence, segments ZmGDIα^146-333 and ZmGDIα^379-445 play
   pivotal roles in the interaction with ZmGA2ox7.3, while the
   helitron-derived exon 10 in ZmGDIα-hel diminishes this interaction.
   Taken together, we hypothesize that P7-1, ZmGDIα (or ZmGDIα-hel), and
   ZmGA2ox7.3 can interact with one another, with P7-1/ZmGDIα/ZmGA2ox7.3
   forming a tighter complex compared to P7-1/ZmGDIα-hel/ZmGA2ox7.3.

The ZmGDIα/P7-1 complex promotes the expression of ZmGA2ox7.3

   Given the dramatic increase in ZmGA2ox7.3 expression in NIL-S 16 days
   after viral infection (Fig. [130]1e), we sought to determine whether
   viral effectors, ZmGDIα, or both manipulate the expression of
   ZmGA2ox7.3, and what the difference is between the two ZmGDIα isoforms.
   Thus, we performed dual-luciferase transient expression assays using
   protoplasts of the two NILs. The reporter with the 3.0-kb ZmGA2ox7.3
   promoter was co-transformed with the effector constructs overexpressing
   viral cDNAs (Supplementary Fig. [131]6a). Out of the 13 effectors, P7-1
   exhibited significantly higher normalized LUC activity in the NIL-S
   protoplasts compared to the empty vector control, followed closely by
   P6. No significant differences were found for the remaining effectors
   (Supplementary Fig. [132]6b). In contrast, in NIL-R protoplasts, the
   LUC activities either significantly decreased (for P5-2, P6, P7-2, and
   P10) or remained unchanged (for P1-P5-1, P7-1, P8, P9-1 and P9-2)
   compared to empty vector controls (Supplementary Fig. [133]6c).
   Similarly, in susceptible B73 protoplasts, P7-1 showed significantly
   higher LUC activity compared to the empty vector control (Supplementary
   Fig. [134]6d).

   Next, we transiently expressed the 35S:GFP or 35S:S7-1-GFP constructs
   in protoplasts isolated from the two NILs. In NIL-S protoplasts, the
   expression of ZmGA2ox7.3 was induced by ~2-fold when transformed with
   P7-1-GFP compared to GFP. However, in NIL-R protoplasts, the expression
   of ZmGA2ox7.3 showed no difference between P7-1-GFP and GFP
   (Supplementary Fig. [135]6e). Furthermore, the expression of ZmGA2ox7.3
   was also significantly elevated in four transgenic lines overexpressing
   S7-1-GFP (Supplementary Fig. [136]6f–h). Collectively, the viral P7-1
   effector, when paired with ZmGDIα, can significantly enhance the gene
   expression of ZmGA2ox7.3. However, this effect was not observed with
   ZmGDIα-hel.

ZmGA2ox7.3 can form oligomers in the presence of GA[4]

   The rice OsGA2ox3 can oligomerize when exposed to GA[4]^[137]27. Being
   the close homolog in maize, ZmGA2ox7.3 might also undergo
   oligomerization in the presence of GA[4]. In the SLC assay, we observed
   that ZmGA2ox7.3 can interact with itself (Fig. [138]5a), a finding
   later confirmed by in vivo Co-IP and in vitro pull-down assays
   (Fig. [139]5b, c). Furthermore, both SLC and BiFC assays affirmed that
   exogenous GA[4] treatment can significantly enhance the ZmGA2ox7.3
   self-association (Fig. [140]5d, e). In rice, the oligomerization of
   OsGA2ox3 was reported to depend on three essential amino acids: C194,
   K308, and K313^[141]27. Sequence alignment revealed that ZmGA2ox7.3
   shares the first two amino acids at positions 198 and 312,
   respectively, and the third residue at the position 317 is Arg (R)
   (Supplementary Fig. [142]3b). We mutated C198, K312, and R317 to Ala
   (A) simultaneously to generate the ZmGA2ox7.3^3A variant. In the SLC
   assay, the ZmGA2ox7.3^3A variant failed to undergo self-association,
   even when treated with exogenous GA[4] (Fig. [143]5f). To assess
   molecular weights for potential oligomers of ZmGA2ox7.3 in gel
   filtration chromatography, we established a standard curve by plotting
   the peak volume against the corresponding molecular weight
   (Fig. [144]6a, b). We immunopurified both GST-ZmGA2ox7.3 and
   GST-ZmGA2ox7.3^3A (Fig. [145]6c). When exposed to 0.5 mM GA[4],
   multiple peaks were observed for ZmGA2ox7.3 in the gel filtration
   profile; however, these were significantly diminished in the absence of
   GA[4]. In contrast, ZmGA2ox7.3^3A exhibited no evident peaks,
   irrespective of the presence or absence of GA[4]. Based on the standard
   curve, these peaks corresponded to the ZmGA2ox7.3 tetramer (designated
   as ‘T’, ~240 kDa), dimer (designated as ‘D’, ~120 kDa), and monomer
   (designated as ‘M’, ~60 kDa), respectively (Fig. [146]6d). We also
   collected elution fractions of the three peaks of GA[4]-treated
   ZmGA2ox7.3 and displayed their compositions using blue native
   polyacrylamide gel electrophoresis (BN-PAGE) (Fig. [147]6e). Next, we
   compared the ‘T’ peaks (11.0–11.5 ml) between ZmGA2ox7.3 and its
   variant ZmGA2ox7.3^3A. In the presence of GA[4], a prominent band of
   ZmGA2ox7.3 tetramer emerged, while it was fainter without GA[4]. For
   ZmGA2ox7.3^3A, there was hardly any discernible bands. The intensities
   of these bands were further confirmed on an SDS-PAGE gel
   (Fig. [148]6f). These findings highlight the importance of GA[4] and
   the three key residues (K198, K312, and R317) in the GA-dependent
   oligomerization of ZmGA2ox7.3.

Fig. 5. The bioactive GA[4] promotes oligomerization of ZmGA2ox7.3 but not
its variant ZmGA2ox7.3^3A.

   [149]Fig. 5
   [150]Open in a new tab

   a The self-association of ZmGA2ox7.3 in an SLC assay. The color scale
   denotes the intensity of the luminescence signal. b Confirmation of the
   self-interaction of ZmGA2ox7.3 using a Co-IP assay. c Confirmation of
   the self-interaction of ZmGA2ox7.3 using an in vitro pull-down assay. d
   The impact of GA[4] on ZmGA2ox7.3 self-association in the SLC assay
   (n = 10). e The impact of GA[4] on ZmGA2ox7.3 self-interaction in the
   BiFC assay. f The self-interaction of the variant ZmGA2ox7.3^3A with or
   without GA[4] in SLC assays (n = 8). In (d, f), data are presented as
   box plots, with error bars indicating the maximum and minimum values.
   The medians are indicated by the central lines, and the box edges
   represent the 25th and 75th percentiles. In (e), data are depicted as
   violin plots with individual data points. Significance levels are
   determined using two-sided Wilcoxon matched-pairs signed rank test (d),
   two-sided Mann–Whitney U test (e) or a two-tailed Student’s t test (f).
   The Co-IP and pull-down assays were repeated at least three times.

Fig. 6. The formation of oligomer ZmGA2ox7.3 depends on bioactive GA[4].

   [151]Fig. 6
   [152]Open in a new tab

   a Chromatographic separation of a mixture of five standard proteins. b
   Plots of calibration curves for standard proteins. “R” indicates the
   correlation efficiency. c The quantities of purified GST-ZmGA2ox7.3 and
   GST-ZmGA2ox7.3^3A. d The gel filtration profiles of ZmGA2ox7.3 and
   ZmGA2ox7.3^3A with or without 0.5 mM GA[4]. The peaks correspond to the
   tetramer (T), dimer (D), and monomer (M) forms. e The molecular weights
   of the “T”, “D”, and “M” peaks of ZmGA2ox7.3 in the presence of 0.5 mM
   GA[4]. f The eluted “T” peak fractions from ZmGA2ox7.3 and
   ZmGA2ox7.3^3A with or without 0.5 mM GA[4]. g The plant morphology of
   Col-0, OE-ZmGA2ox7.3^3A, and OE-ZmGA2ox7.3 in Arabidopsis thaliana. The
   exogenous ZmGA2ox7.3-GFP^3A and ZmGA2ox7.3-GFP proteins were detected
   using immunoblotting with anti-GFP antibodies. Scar bar, 5 cm.
   Significance level was determined by a two-sided simple linear
   regression statistical test (b). The gel filtration chromatography
   assays were repeated three times.

Enzymatic activity of ZmGA2ox7.3 is mainly due to its oligomerization

   In rice, the formation of OsGA2ox3 multimers is known to determine its
   enzymatic activity^[153]27. Inspired by this, we posed two questions:
   (1) Does the formation of ZmGA2ox7.3 oligomers, similar to its rice
   counterpart, influence its enzymatic activity? (2) Do the
   ZmGA2ox7.3-interacting proteins, specifically P7-1 and the two ZmGDIα
   isoforms, affect its enzymatic activity as well?

   We firstly overexpressed ZmGA2ox7.3 and ZmGA2ox7.3^3A in Arabidopsis,
   with each construct yielding two independent transgenic events. As
   anticipated, overexpression of ZmGA2ox7.3 induced a dwarf phenotype,
   while overexpression of ZmGA2ox7.3^3A did not lead to dwarfing
   (Fig. [154]6g). Considering that ZmGA2ox7.3^3A was unable to form
   oligomers, this observation implies that the formation of the
   ZmGA2ox7.3 oligomer is a prerequisite for its enzymatic activity.

   To assess the enzymatic activity of ZmGA2ox7.3, we developed an assay
   by detecting the product GA[34] while using GA[4] as the substrate,
   modified from previous studies (Supplementary
   Fig. [155]3c)^[156]25,[157]26. We prepared five distinct concentrations
   of GA[4] and mixed each with 1 mg of ZmGA2ox7.3 (Fig. [158]7a). The
   enzymatic activity of ZmGA2ox7.3 was detectable at GA[4] concentrations
   as low as 5 × 10^−4 mM, and augmented with increasing GA[4]
   concentration. This activity plateaued at 0.5 mM GA[4] and did not
   increase further, even with GA[4] concentrations up to 5 mM
   (Fig. [159]7b). In parallel, we observed a proportional increase in the
   content of ZmGA2ox7.3 tetramers and dimers as the GA[4] concentration
   increased, culminating at 0.5 mM GA[4] on the gel filtration profile.
   Intriguingly, the curves for 0.5 mM and 5 mM GA[4] coincided perfectly,
   despite a tenfold disparity in GA[4] concentration (Fig. [160]7c). A
   strong relationship was observed between the content of ZmGA2ox7.3
   oligomers (tetramers and dimers) and the GA[34] product, with a
   correlation coefficient of 0.9819 (Fig. [161]7d). Moreover, we
   collected elution fractions from the ‘T’ and “D” peaks and subjected
   them to immunoblotting with an anti-GST antibody on BN-PAGE. The
   intensity of the ZmGA2ox7.3 oligomeric bands aligned well with its
   enzymatic activity across different GA[4] concentrations
   (Fig. [162]7e). These results indicate that the quantity of ZmGA2ox7.3
   oligomers directly reflect its enzymatic activity.

Fig. 7. Oligomerization of ZmGA2ox7.3 is closely associated with its
enzymatic activity.

   [163]Fig. 7
   [164]Open in a new tab

   a Detection of the purified GST-ZmGA2ox7.3. b The enzymatic activity of
   ZmGA2ox7.3 across a range of GA[4] concentrations, as measured by
   quantifying the product GA[34] (n = 3). c The gel filtration curves of
   ZmGA2ox7.3 under various concentrations of GA[4]. The peaks observed,
   from left to right, correspond to tetramer (T), dimer (D), and monomer
   (M). d The correlation between the product GA[34] and the peak area of
   ZmGA2ox7.3 oligomers (the red dashed box section in c) at various
   concentrations of GA[4]. “R” indicates the Pearson’s correlation
   efficiency. e Quantitation of ZmGA2ox7.3 tetramers and dimers
   (oligomers) under various concentrations of GA[4]. The eluted peak
   fractions (11.0–11.5 ml) and (12.0–12.5 ml) were subjected to 4–13%
   BN-PAGE, followed by immunoblotting with an anti-GST antibody. In (b),
   data (mean ± s.d.) are presented with individual data points, where
   different lowercase letters indicate a significant difference
   (P < 0.05) based on one-way ANOVA with Tukey’s test and two-sided
   Pearson R statistical test (d). The gel filtration chromatography assay
   was repeated three times.

The ZmGDIα/P7-1 complex affects the enzymatic activity of ZmGA2ox7.3

   To elucidate how the interaction of ZmGA2ox7.3 with P7-1, ZmGDIα or
   ZmGDIα-hel affects its enzymatic activity, we immunopurified all four
   proteins: MBP-P7-1, His-ZmGDIα, His-ZmGDIα-hel, and GST-ZmGA2ox7.3.
   Equal amounts of GST-ZmGA2ox7.3 were mixed with one or two of the other
   three proteins to prepare six distinct combinations (I:
   GST-ZmGA2ox7.3 + MBP + GST; II: GST-ZmGA2ox7.3 + MBP-P7-1 + GST; III:
   GST-ZmGA2ox7.3 + MBP+His-ZmGDIα; IV:
   GST-ZmGA2ox7.3 + MBP-P7-1+His-ZmGDIα; V:
   GST-ZmGA2ox7.3 + MBP+His-ZmGDIα-hel; VI:
   GST-ZmGA2ox7.3 + MBP-P7-1+His-ZmGDIα-hel) (Fig. [165]8a). When exposed
   to 0.5 mM GA[4], all six combinations displayed multiple oligomers in
   the gel filtration profile. In addition to the previously identified
   ‘T’, ‘D’, and ‘M’ peaks, we detected new ‘Mu’ peaks in combinations III
   (ZmGA2ox7.3/ZmGDIα) and IV (ZmGA2ox7.3/ZmGDIα/P7-1). However, these
   ‘Mu’ peaks nearly vanished when ZmGDIα was substituted by ZmGDIα-hel
   (Fig. [166]8b). We collected elution fractions from the two ‘Mu’ peaks
   and analyzed their molecular weights using BN-PAGE and compositions via
   mass spectrometry. The ‘Mu’ peak for combination III included a
   ZmGA2ox7.3 oligomer with ZmGDIα (~10.21 ml, ~410 kDa), while that for
   combination IV had a ZmGA2ox7.3 oligomer with ZmGDIα/P7-1 (~10.19 ml,
   ~440 kDa) (Fig. [167]8b–d). We also conducted immunoblotting assays to
   analyze the compositions of eluted fractions collected from the ‘Mu’
   (10.0 to 10.5 ml) and ‘T’ (11.0 to 11.5 ml) peaks across all six
   combinations. In the ‘Mu’ fractions, the content of ZmGA2ox7.3 was
   significantly higher in the combination IV than in other combinations
   (Fig. [168]8e). Conversely, in the ‘T’ fractions, the content of
   ZmGA2ox7.3 was significantly higher in the combination I than in other
   combinations (Fig. [169]8f). When considering both the ‘Mu’ and ‘T’
   fractions together, ZmGDIα did not significantly infleunce the
   oligomerization of ZmGA2ox7.3, whereas ZmGDIα-hel severely hindered
   this oligomerization. Notably, MBP-P7-1 could further enhance the
   formation of the ZmGA2ox7.3 oligomer when combined with ZmGDIα, but not
   with ZmGDIα-hel (Fig. [170]8e, f).

Fig. 8. The ZmGDIα/P7-1 complex enhances the enzymatic activity of
ZmGA2ox7.3.

   [171]Fig. 8
   [172]Open in a new tab

   a The compositions of six combinations. The mixtures were verified on a
   10% SDS-PAGE gel using an immunoblotting assay. b The gel filtration
   curves of oligomerized proteins from six combinations in the presence
   of GA[4]. The peaks correspond to multimer (Mu), tetramer (T), dimer
   (D), and monomer (M). The “Mu” peak was enlarged and placed below. c, d
   Analysis of compositions of the “Mu” peaks appeared in combinations III
   and IV. The eluted “Mu” peak fractions were stained with Coomassie
   brilliant blue (c). The various proteins were revealed by the LC-MS
   assay (d). e, f Quantification of ZmGA2ox7.3 oligomers for the “Mu” and
   “T” peaks from six combinations (n = 5). The eluted “Mu” (10.0–10.5 ml)
   (e) and “T” (11.0–11.5 ml) (f) fractions were subjected to 10%
   SDS-PAGE, followed by immunoblotting using anti-GST (for
   GST-ZmGA2ox7.3), anti-His (for His-ZmGDIα/His-ZmGDIα-hel), and anti-MBP
   (for MBP-P7-1) antibodies. g Measurement of enzymatic activities of
   ZmGA2ox7.3 for six combinations in the presence of 0.5 mM GA[4]. In (e,
   f), data (mean ± s.d.) are presented with individual data points, data
   are depicted as violin plots with individual data points (g). Different
   lowercase letters indicate a significant difference (P < 0.05) based on
   Kruskal–Wallis test (e) or one-way ANOVA with Tukey’s test (f, g). The
   gel filtration chromatography assay was repeated at least five times.

   With the same six combinations, we examined the potential impacts of
   P7-1 and the two ZmGDIα isoforms on the enzymatic activity of
   ZmGA2ox7.3. Compared to the mixture of 0.5 mM GA[4] and GST-ZmGA2ox7.3
   (combination I), the addition of MBP-P7-1 (combination II) led to a
   significant reduction in the GA[34] product. In contrast, adding
   His-ZmGDIα (combination III) did not significantly affect the GA[34]
   product. When both MBP-P7-1 and His-ZmGDIα were added simultaneously
   (combination IV), a significant increase in the GA[34] product was
   observed. However, the inclusion of His-ZmGDIα-hel led to a significant
   decrease in the GA[34] product, regardless of the absence (combination
   V) or presence (combination VI) of MBP-P7-1 (Fig. [173]8g). Taken
   together, these results suggest that the oligomerization and enzymatic
   activity of ZmGA2ox7.3 can be enhanced when mixed with ZmGDIα/P7-1, but
   are considerably diminished when ZmGDIα-hel replaces ZmGDIα.

Viral infection impairs the subcellular structures of susceptible maize

   To decipher the cytological changes caused by RBSDV infection that
   leads to MRDD symptoms, we observed the internode cells in the diseased
   NIL-S and the heathy NIL-R. The diseased NIL-S plants exhibited typical
   symptoms of stunted growth and shortened internodes compared to the
   healthy NIL-R plants (Supplementary Fig. [174]7a, b, g, h). In the
   upper internodes of the diseased NIL-S plants, devastating changes in
   subcellular structures were observed, including significant thickening
   of the cell walls and rupture of both the cell and organelle membranes
   (Supplementary Fig. [175]7c–f). In contrast, the cytoplasm in the
   healthy NIL-R plants was tightly organized around the cell membrane
   (Supplementary Fig. [176]7i–l). Notably, numerous virions were detected
   in the cytoplasm of the diseased NIL-S plants, while no viral particles
   were found in the healthy NIL-R plants (Supplementary Fig. [177]7e, f,
   k, l).

RBSDV-induced phytohormone change underlies MRDD symptoms

   To explore the mechanism by which RBSDV infection leads to MRDD
   symptoms, we used the two NILs to measure the levels of various GAs
   during viral infection. We infested maize seedlings at the budding
   stage with viruliferous planthoppers and harvested upper internodes
   (the shortest upper stem nodes) and upper leaves (the areas around
   white waxy enations) at 60 dpi. We then assessed the relative
   expression of the viral S7-1 and quantified the levels of eight GAs
   (GA[1], GA[4], GA[8], GA[9], GA[20], GA[34], and GA[51]). In the upper
   internodes, the S7-1 mRNA accumulated significantly more in NIL-S than
   in NIL-R after RBSDV infection (Fig. [178]9a). Comparatively, following
   viral infection, NIL-S exhibited a pronounced decrease in the levels of
   bioactive GA[4], its product GA[34] and precursor GA[9], relative to
   NIL-R (Fig. [179]9b). Additionally, RBSDV-infected NIL-S displayed
   substantially reduced levels of bioactive GA[1] and its product GA[8].
   Intriguingly, RBSDV infection led to a remarkable increase in GA[29]
   levels in both NIL-S and NIL-R (Fig. [180]9b). These findings provide
   insights into the severe internode shortening observed in
   RBSDV-infected NIL-S, as opposed to the unaffected NIL-R.

Fig. 9. RBSDV infection severely reduces the active gibberellins in the upper
internodes and upper leaves of the two NILs, and a working model for the
development of MRDD symptoms.

   [181]Fig. 9
   [182]Open in a new tab

   The relative expression of RBSDV S7-1 in the upper internodes (a) and
   upper leaves (c) of the two NILs at 60 dpi (n = 3). The contents of
   eight gibberellins in the upper internodes (b) and upper leaves (d) of
   the two NILs at 60 days after infestation with either RBSDV-carrying or
   RBSDV-free planthoppers (n = 3). In (a, c), data (mean ± s.e.m.) are
   presented with individual data points, as determined by a two-tailed
   Student’s t test. In (b, d), data represent the mean ± s.d. and
   different lowercase letters indicate a significant difference
   (P < 0.05) based on one-way ANOVA with Tukey’s test.

   In the upper leaves as well, the S7-1 mRNA accumulated significantly
   more in NIL-S than in NIL-R (Fig. [183]9c). In the infected NIL-S, the
   level of bioactive GA[4] nearly vanished, and there was also a
   significant decrease in the level of another bioactive GA[1].
   Additionally, other GAs, such as GA[9] and GA[51] were also reduced,
   and GA[29] and GA[34], were barely detectable (Fig. [184]9d). In NIL-R,
   there was no significant difference in the active GA[4] levels pre- and
   post-viral infection. However, the other active GA[1], along with four
   additional GAs (GA[1], GA[8], GA[9], and GA[20]), exhibited a
   significant reduction, though not as pronounced as in NIL-S. Both
   GA[29]and GA[34] were not detected (Fig. [185]9d).

   The white waxy enations on the abaxial side of the upper leaf veins
   arise from the abnormal proliferation of phloem cells in the vascular
   bundle^[186]5. Given the involvement of auxin, cytokinins, and GAs in
   cell expansion, we sought to ascertain whether auxin and cytokinins
   also contribute to the formation of white waxy enations. To do this,
   the two NILs were infested with viruliferous and non-viruliferous
   planthoppers, respectively. We collected samples at 16 and 60 dpi to
   assess auxin levels at both time points and cytokinin levels at 60 dpi,
   respectively. Following viral infection, RBSDV accumulated
   significantly in NIL-S but only slightly in NIL-R, as evidenced by the
   relative expression of viral S10 and S7-1 at 16 dpi (Supplementary
   Fig. [187]8a, b). Concurrently, the auxin content in the upper leaves
   significantly increased in NIL-S, but remained unchanged in NIL-R.
   Under non-infected conditions, no difference in auxin levels was
   observed between NIL-S and NIL-R (Supplementary Fig. [188]8c).

   By 60 dpi, both NILs displayed significantly elevated auxin levels in
   both the upper leaves and internodes compared to the uninfected
   control, with the auxin content in NIL-R surpassing that in NIL-S,
   particularly in the upper internodes (Supplementary Fig. [189]8d, e).
   Finally, we evaluated the cytokinin levels in the upper leaves. During
   the late stage of RBSDV infection, the concentrations of six cytokinins
   (iP, iPR, tZ, tZR, cZ and HZD) were significantly higher in NIL-R than
   in NIL-S, with two exceptions being cZR and DHZR (Supplementary
   Fig. [190]8f).

   The observed fluctuations in the three growth-promoting phytohormones
   underscore the profound disturbance of phytohormone balance in NIL-S
   following RBSDV infection. The coordinated interactions among GAs,
   auxin, and cytokinins are crucial for the manifestation of MRDD
   syndrome, which is mainly characterized by dwarfism.

A working model for the development of MRDD symptoms

   ZmGA2ox7.3 can form active oligomers to convert bioactive GA[1]/GA[4]
   into inactive GA[8]/GA[34]. During RBSDV infection, the expression of
   ZmGA2ox7.3 is substantially induced in NIL-S to produce abundant
   ZmGA2ox7.3 oligomers. These oligomers can act either independently or
   in conjunction with the P7-1/ZmGDIα complex to reduce GA[1]/GA[4]
   levels, disrupt the balance of auxin and cytokinins, and ultimately
   induce MRDD symptoms. Conversely, the lower levels of ZmGA2ox7.3
   oligomers in NIL-R are insufficient to effectively reduce GA[1]/GA[4]
   levels, leading to MRDD resistance (Fig. [191]10).

Fig. 10. A working model for the development of MRDD symptoms.

   [192]Fig. 10
   [193]Open in a new tab

   Left: RBSDV infection in NIL-S substantially stimulates the expression
   of ZmGA2ox7.3 to form abundant ZmGA2ox7.3 oligomers. These can act
   either alone or combined with P7-1/ZmGDIα to reduce active GA[1]/GA[4]
   levels, disrupt the balance of auxins and cytokinins, and ultimately
   induce MRDD symptoms. Right: RBSDV infection in NIL-R just slightly
   increases the expression levels of ZmGA2ox7.3, and the resultant low
   level of ZmGA2ox7.3 oligomers is insufficient to effectively reduce
   GA[1]/GA[4] levels, eventually leading to MRDD resistance.

Discussion

   Viral proteins can hijack host target proteins to disrupt the balance
   of plant hormones, which are crucial for facilitating viral systemic
   infection, ultimately leading to the appearance of viral
   symptoms^[194]24,[195]30,[196]31. However, little is known about the
   molecular mechanisms underlying the manifestation of MRDD symptoms. Our
   previous study demonstrated that the viral P7-1 effector binds more
   tightly to ZmGDIα than to ZmGDIα-hel, which underpins the differential
   resistance to RBSDV. In this study, we discovered that P7-1 can recruit
   another host target, ZmGA2ox7.3, which also exhibits a tighter binding
   affinity to ZmGDIα than to ZmGDIα-hel. During RBSDV infection, the
   allelic variation at the ZmGDIα locus, caused by helitron insertion,
   greatly influences both the quantities and interactions among
   ZmGA2ox7.3, ZmGDIα/ZmGDIα-hel, and the viral P7-1. All these factors
   play crucial roles in reducing active gibberellins and, consequently,
   disrupting the balance of auxin and cytokinins in maize, thereby
   determining the outcome of MRDD. Our discovery thus elucidates the
   intricate molecular mechanisms that underlie the development of MRDD
   symptoms.

   In susceptible maize, RBSDV infection could significantly enhance the
   gene expression of ZmGA2ox7.3. Given that no RBSDV protein serves as a
   transcription factor, the viral proteins likely recruit host
   transcription factors to promote ZmGA2ox7.3 expression. Previous
   researches indicated that GA-2 oxidase family genes are regulated by
   multiple transcription factors^[197]32,[198]33. Intriguingly, our
   previous Co-IP/MS assay revealed numerous transcription factors in
   immunoprecipitates associated with RBSDV infection, such as
   growth-regulating factor (GRF), AP2-EREBP, basic helix-loop-helix
   (bHLH), MADS-box, and GRAS (Supplemental Table [199]2). These findings
   suggest that, upon RBSDV invasion, the viral P7-1 effector may recruit
   certain maize transcription factors to promote the expression of
   ZmGA2ox7.3.

   OsGA2ox3 forms GA[4]-dependent multimers, displaying higher enzymatic
   activity than monomer^[200]27. Similarly, when exposed to GA[4],
   ZmGA2ox7.3 can form oligomers (tetramers and dimers) with enhanced
   catalytic activity, converting the bioactive GA[4] into its inactive
   GA[34]. In the ZmGA2ox7.3^3A variant, the three mutated residues are
   solely associated with the formation of oligomers, but not with
   enzymatic activity^[201]27. The failure to form ZmGA2ox7.3^3A oligomers
   leads to the absence of enzymatic activity, indicating that oligomer
   formation is a prerequisite for the enzymatic activity. Given that
   ZmGA2ox7.3 tetramers and dimers exhibited simultaneous increases in the
   presence of GA[4], we thus cannot distinguish between them in their
   enzymatic activities. Therefore, we generally conclude that ZmGA2ox7.3
   oligomers enhance enzymatic activity. In susceptible maize,
   RBSDV-triggered upregulation of ZmGA2ox7.3, combined with a massive
   accumulation of the viral P7-1 effector, leads to the formation of
   abundant ZmGA2ox7.3 oligomers. These oligomers, whether acting alone or
   in conjunction with ZmGDIα/P7-1, can efficiently convert active GAs
   into their inactive forms, thereby contributing to the manifestation of
   MRDD syndrome. Conversely, in resistant maize, the presence of
   ZmGDIα-hel dramatically inhibits the RBSDV-induced ZmGA2ox7.3
   expression and prevents ZmGA2ox7.3 oligomerization, resulting in
   minimal or no MRDD symptoms.

   Viral effectors such as TMV replicase and RDV-P2 disrupt plant hormone
   balance^[202]19,[203]34. In the current study, the RBSDV P7-1 effector
   not only stimulates ZmGA2ox7.3 expression but also amplifies its
   enzymatic activity when combined with ZmGDIα. Therefore, P7-1 plays a
   vital role in promoting ZmGA2ox7.3 to reduce GA[1] and GA[4] levels.
   Notably, in a previous Co-IP/MS assay, we identified seven auxin
   response factors (ARFs) (Supplemental Table [204]2). Given that ARFs
   act as transcriptional activators in the auxin signaling pathway, our
   findings hint at the possibility that the ZmGDIα/P7-1 complex might
   also interface with these ARFs, thereby potentially disrupting the
   auxin balance.

Methods

Plant materials

   A pair of near-isogenic lines (NILs), namely NIL-S and NIL-R, were
   derived from our previous study^[205]5. NIL-R carries the resistance
   gene ZmGDIα-hel, while NIL-S carries the susceptibility gene ZmGDIα.
   Transgenic plants (ZmGA2ox7.3-GFP-OE and P7-1-GFP-OE) were generated in
   the Maize Functional Genomic Project of China Agricultural University,
   using the susceptible maize inbred line B73 as the recipient.

Artificial inoculation of RBSDV

   Maize kernels were germinated in vermiculite soil in pots in a
   greenhouse. The seedlings were artificially inoculated by either
   viruliferous planthoppers carrying RBSDV or non-viruliferous
   planthoppers without RBSDV during the emergence stage for a duration of
   3 days^[206]35. After inoculation, the seedlings were transplanted into
   the field. The detailed methodology was described in our previous
   study^[207]5.

Cytological and histological examination

   Seedlings of the NILs were artificially infested with viruliferous
   planthoppers. At 60 days post-inoculation (dpi), the uppermost
   internodes of both NIL-S and NIL-R were sampled by cutting discs with a
   thickness of less than 1 mm. These samples were then immersed in a 2.5%
   glutaraldehyde stationary solution (Huayueyang Biotechnology, China)
   overnight at 4 °C. The samples then underwent a series of five
   processing steps, including fixation, post-fixation, dehydration,
   embedding, and sectioning, following the methodology outlined in the
   previous study^[208]36. Eventually, images of the samples were captured
   using Tecnai G2 spirit BioTwin (FEI, USA).

   The transgenic ZmGA2ox7.3-GFP-OE and the recipient B73 plants were
   grown until anthesis, at which point the uppermost internodes were
   sampled and immobilized using the same procedures as described above.
   The samples were washed three times with PBS buffer, followed by a 2-h
   treatment with 1% OsO4 in PBS buffer (136.8 mM NaCl, 2.7 mM KCl, 4 mM
   Na[2]HPO4, 1.8 mM KH[2]PO4). They were then washed an additional three
   times with PBS. The samples underwent a series of ethanol dehydration
   steps at concentrations of 30%, 50%, 70%, 80%, 90% and 100%, with two
   subsequent rounds using 100% ethanol. Afterwards, the samples were
   further dehydrated in an automatic critical point dryer (Leica EM CPD
   00, Germany). Ultimately, the samples were coated with a layer of gold
   and photographed using a S-3400N Hitachi scanning electron microscope
   (Hitachi Hi-Tech, Tokyo, Japan).

RNA extraction, RT-qPCR and RNA-seq

   Total RNA was extracted from maize leaves or internodes with a
   PLANTpure plant RNA Kit (Aidlab, China), following the manufacturer’s
   protocol. Reverse transcription was performed using TransScript® RT/RI
   reverse transcriptase with random primers (Transgen Biotech, China),
   according to the manufacturer’s protocol. The full-length cDNA sequence
   of ZmGA2ox7.3 was amplified from the two NILs. RT-qPCR assays were
   performed on a CFX Connect real-time RT-PCR system (BIO-RAD, USA) using
   a SYBR Green qRT-PCR kit (Takara, Japan) to detect the expression
   levels of specific genes. The primer pairs qZmGA2ox, qP7-1, qZmDAO,
   qZmPP1 and qZmB1B2 were used to quantify the transcript levels of
   ZmGA2ox7.3, RBSDV-S7-1, ZmDAO1, ZmPP1 and ZmB1B2 respectively
   (Supplementary Data [209]1). The expression level of ZmTubulin was used
   as an internal control to calculate the relative expression levels of
   other genes using the 2^−ΔΔCt method.

   RNA-seq was performed by Novogene Bioinformatics Technology (Beijing,
   China). The purity and integrity of the extracted RNA samples were
   evaluated using a Nanodrop spectrophotometer and an Agilent 2100
   Bioanalyzer, respectively. The libraries were constructed and sequenced
   on an Illumina Hiseq platform, generating 125 bp/150 bp paired-end
   reads. The raw data (raw reads) in FASTQ format were subjected to
   quality control, with the removal of adaptor sequences and low-quality
   reads to get clean reads. These clean reads were then aligned to the
   Maize genome (B73 RefGen_v4, AGPv4) using HISAT (2.0.4) with default
   parameters. FPKM (fragments per kilobase of transcript per million
   mapped reads) values of each gene were calculated using HTSeq (version
   0.6.1) software. Differentially expressed genes (DEGs) were analyzed
   using DESeq (1.12.0) software, with significant DEGs defined based on a
   P-value threshold (log2 fold change > 1, P < 0.05). The z-score scale
   represents the mean-subtracted, regularized log-transformed FPKM
   values. The abundance of reads was visualized using the Integrative
   Genomics Viewer (2.16.1). Gene ontology (GO) enrichment analysis was
   performed using the accession numbers of significant DEGs via GOseq
   (version 2.12). Pathway enrichment analysis of DEGs was conducted using
   KOBAS (version 2.0). The RNA-seq analysis included two independent
   biological replicates, with the whole uppermost leaves harvested from
   at least five individual plants for each sample.

Construct generation and plant transformation

   The primer sequences used in this study are listed in Supplementary
   Data [210]1. All constructs were generated using the homologous
   recombination cloning system (Yeasen Biotechnology, China). Sequencing
   was performed to confirm the absence of mutations in the PCR fragments
   that were inserted into vectors. For the generation of ZmGA2ox7.3-GFP,
   ZmGA2ox7.3-mCherry, and ZmGA2ox7.3-Myc constructs, the coding sequence
   (CDS) of ZmGA2ox7.3 was amplified and cloned into the Super1300-GFP,
   Super1300-Myc, and Super1300-mCherry vectors, respectively. Likewise,
   both the full-length and fragmentary coding sequences of ZmGDIα and
   ZmGDIα-hel were inserted into the Super1300-Myc vector.

   To generate SLC constructs, we amplified the coding sequences (CDSs) of
   ZmGA2ox7.3, ZmGDIα, and ZmGDIα-hel and subsequently cloned them into
   JW771-35S-NLuc and JW772-35S-CLuc vectors. A series of fragments from
   the ZmGDIα and ZmGDIα-hel genes were also amplified and inserted into
   JW772-35S-CLuc to fuse with the C-terminal fragment of the luciferase
   gene (cLUC) as described previously^[211]5.

   To generate the GST-ZmGA2ox7.3, His-ZmGA2ox7.3, His-ZmGDIα, and
   His-ZmGDIα-hel constructs, the CDSs were cloned into the pGEX6P-1,
   pCold-TF, and PHAT2 vectors, respectively. The
   ZmGA2ox7.3^C198A/K312A/R317A (GST-ZmGA2ox7.3^3A) construct was
   performed by ClonExpress MultiS One Step Kit (Vazyme, China).

   To generate transgenic maize and Arabidopsis thaliana plants, the CDSs
   of RBSDV-S7-1and ZmGA2ox7.3^3A were amplified and cloned into the
   Super1300-GFP vector, respectively. Using the pSuper:RBSDV-S7-1-GFP and
   pSuper:ZmGA2ox7.3-GFP constructs as templates, we amplified fragments
   of RBSDV-S7-1-GFP and ZmGA2ox7.3-GFP, which were then inserted into the
   pBCXUN vector to generate the pUbi:ZmGA2ox7.3-GFP and
   pUbi:RBSDV-P7-1-GFP constructs, respectively.

   The cDNA of ZmGA2ox7.3 was cloned into the P2YC or P2YN vector to
   generate cYFP-ZmGA2ox7.3/nYFP-ZmGA2ox7.3 constructs. The full-length
   sequences of RBSDV-S7-1, ZmGDIα/ZmGDIα-hel, ZmDAO1, ZmB1B2, ZmPP1, as
   well as truncated versions of ZmGDIα/ZmGDIα-hel, also were cloned into
   the P2YN or P2YC vectors. This resulted in the generation of the
   following constructs: nYFP-P7-1, nYFP-ZmDAO1, nYFP-ZmB1B2, nYFP-ZmPP1,
   nYFP-ZmGDIα/cYFP-ZmGDIα, nYFP-ZmGDIα-hel/cYFP-ZmGDIα-hel,
   nYFP-ZmGDIα^1-146, nYFP- ZmGDIα^146-333, nYFP-ZmGDIα-hel^146-319, and
   nYFP-ZmGDIα^379-445, respectively.

   To generate LUC constructs, a 3.0 kb promoter fragment of ZmGA2ox7.3
   was amplified and inserted into the pGreenII0800-LUC vector to generate
   the pZmGA2ox7.3:LUC construct. The 13 RBSDV sequences were cloned into
   the pGreenII62-SK vector to generate the 13 effector vectors. The
   P7-1-myc, P7-1-nLUC, MBP-P7-1, and pCold-TF-His-P7-1 constructs were
   obtained from our previous study^[212]5.

Subcellular localization

   To detect the subcellular localization of the ZmGA2ox7.3 protein, the
   pSuper: ZmGA2ox7.3-GFP, pSuper1300-GFP, and 35S:PIP2A-mCherry (a plasma
   membrane protein) constructs were transformed into Agrobacterium strain
   EHA105. After an overnight culture in LB media at 28 °C, equal volumes
   of Agrobacterium cultures, normalized to an OD600 value, were harvested
   for the GFP and mCherry constructs. These cultures were then
   resuspended in infiltration buffer (10 mM MgCl[2], 10 mM MES, 150 mM
   acetosyringone, pH 5.8) and incubated for 2 h at 28 °C. Subsequently,
   the suspensions were infiltrated into leaves of three-week-old N.
   benthamiana, and the fluorescence signals were visualized two days
   post-infiltration. Additionally, the pSuper:ZmGA2ox7.3-mCherry and
   pSuper:mCherry constructs were also transformed into maize protoplasts
   and incubated at 28 °C for 12 h. The GFP and mCherry fluorescence
   signals were visualized using confocal laser scanning microscopy
   (ZEISS880, Carl Zeiss) after 72 h of transformation.

Phylogenetic analysis

   The amino acid sequences of GA2ox from maize and rice were downloaded
   from the National Center for Biotechnology Information
   ([213]https://www.ncbi.nlm.nih.gov/) and the maize database
   ([214]https://www.maizegdb.org/), respectively. The phylogenetic tree
   was constructed using the neighbor-joining method with MEGA7.0 software
   ([215]https://www.megasoftware.net/).

Split luciferase complementation (SLC) assay

   The resulting constructs were transformed into Agrobacterium EHA105 and
   co-infiltrated into N. benthamiana leaves as previously described. 72 h
   after transformation, the initial infiltration sites were injected with
   1 mmol/L beetle luciferase (Promega, USA). The chemiluminescent signal
   was then captured using the Chemiluminescent Imaging System (Tanon,
   China). Additionally, to further assess the chemiluminescent signal,
   the infiltrated sites were injected with 0.5 mM GA[4].

   Total protein was extracted from the injected leaves using laemmli
   sample buffer (0.125 M Tris-HCl, 4% SDS, 20% glycerol, 10%
   2-mercaptoethanol, 0.004% bromophenol blue, pH 6.8). The resulting
   supernatants were subjected to immunoblotting with an α-LUC antibody
   (Abcam, UK) and a plant-specific actin antibody (ABclonal Technology,
   China). The fluorescence signals and immunoblot bands were quantified
   for their intensities using ImageJ Launcher software (National
   Institutes of Health), and the measurements were expressed as mean gray
   value.

Co-immunoprecipitation (Co-IP) assay

   The relevant constructs were co-expressed in N. benthamiana leaves
   through co-infiltration for 72 h. Total protein was then extracted from
   the infiltrated leaves using IP buffer (50 mM Tris-MES, 10 mM EDTA,
   0.5 M sucrose, 1 mM MgCl[2], 5 mM DTT, 1 mM PMSF, pH 8.0). The
   extracted supernatants were incubated with anti-GFP magnetic beads
   (MBL, China) for 2 h, after which the beads were washed three times
   with IP buffer. The immunoprecipitated complexes were heated in SDS
   loading buffer (Genstar, China) at 100 °C for 5 min and subsequently
   subjected to immunoblotting using an anti-Myc antibody (MBL, China).

Pull-down assays

   The relevant fusion constructs were transformed into Escherichia coli
   strain BL21 (DE3) (TransGen Biotech, China). The recombinant
   GST-ZmGA2ox7.3, His-ZmGDIα, His-ZmGDIα-hel, and His-P7-1 proteins were
   subsequently purified using affinity chromatography. GST-fused proteins
   were purified using glutathione sepharose (Yeasen, China), while
   His-fused proteins were purified using Ni Sepharose (GE Healthcare,
   USA), all according to the manufacturers’ instructions. GST or
   GST-ZmGA2ox7.3 was immobilized on glutathione agarose beads and
   incubated at 4 °C for 1 h. These immobilized proteins were then
   incubated with His-ZmGDIα, His-ZmGDIα-hel, and His-P7-1 for an
   additional 2 h. Following elution from the beads, the proteins were
   subjected to immunoblotting using an anti-His antibody (Easybio,
   China).

Bimolecular fluorescence complementation (BiFC) assay

   The relevant constructs were transformed into Agrobacterium EHA105 and
   were co-infiltrated into N. benthamiana leaves and allowed to express
   for 72 h. GFP fluorescence was visualized using confocal microscopy
   (ZEISS880, Carl Zeiss) with an excitation wavelength set at 488 nm. For
   the combinations involving nYFP-ZmGA2ox7.3 and cYFP-ZmGA2ox7.3, the
   infiltrated sites were treated with either 0.1% Tween 20 or 0.1% Tween
   20 with 0.5 mM GA[4]. Subsequently, the GFP fluorescence signals were
   assessed. The fluorescence intensities were quantified using ZEN
   software.

Microscale Thermophoresis (MST) Assay

   The purified proteins, GST-ZmGA2ox7.3, His-P7-1, His-ZmGDIα, and
   His-ZmGDIα-hel, were concentrated and dialyzed in PBS buffer (pH 7.4)
   using an Amicon Ultra-4 concentrator unit with a 10 kDa molecular
   weight cutoff (Millipore, USA). To investigate the strength of
   interaction between ZmGA2ox7.3 and P7-1, ZmGDIα or ZmGDIα-hel, the MST
   assays were subjected to performed according to the manufacturer’s
   instructions (Nano Temper Technologies, Cat# MO-L011). The binding
   reactions were measured by using a microscale thermophoresis instrument
   (Nano Temper Technologies) at 25 °C, 40% MST power and 20% LED power.
   The Nano Temper Analysis Software MO affinity Analysis was used to
   calculate the value of the dissociation constant (Kd).

Transient expression assays in maize protoplasts

   For dual-luciferase assays, the pZmGA2ox7.3:LUC construct served as the
   reporter gene, while the 62SK-RBSDVs and 35S:S7-1-GFP constructs,
   driven by the 35S promoter, acted as effectors. The REN gene, under the
   control of the 35S promoter, was included in the pGreenII0800-LUC
   vector as an internal control. Empty effectors pGreenII 62-SK and
   35:GFP were employed as controls. Appropriate combinations of reporter
   and effector constructs were co-transformed into maize protoplasts and
   incubated at 28 °C for 12 h. Luciferase signals were detected according
   to the instructions of the Dual-Luciferase Reporter Assay System
   (Promega, USA). The ratio of REN activity to LUC activity was taken as
   the relative LUC activity. Each experiment was performed with at least
   three biological replicates.

Gel filtration chromatography analysis

   Standard curves correlating elution volume to molecular weight were
   generated according to the manufacturer’s instructions (GE Healthcare,
   USA). The mixture, containing ~3 mg of each standard protein, was
   injected onto a Superdex^TM 200 Increase 10/300 GL column (GE
   Healthcare, USA). The standard proteins were added in the following
   order: thyroglobulin (669 kDa), ferritin (450 kDa), aldolase (158 kDa),
   conalbumin (75 kDa), and ovalbumin (42 kDa). Blue Dextran 2000 was used
   to define the column void volume (V[o]). The mobile phase consisted of
   0.01 M phosphate buffer saline at pH 7.4, and the flow rate was set to
   0.3 ml/min. A standard curve for Kav and Mr was constructed based on
   the formula K[av] = (V[e]-V[o])/(V[c]-V[o]) = a-b*log[10](Mr), where
   V[c] represents the column volume, V[e] is the elution volume, and Mr
   denotes the protein molecular weight, using data obtained from the
   standard proteins.

   The purified proteins, GST-ZmGA2ox7.3, GST-ZmGA2ox7.3^3A, MBP-P7-1,
   His-ZmGDIα, and His-ZmGDIα-hel, were concentrated and dialyzed in PBS
   buffer (pH 7.4) using an Amicon Ultra-4 concentrator unit with a 10 kDa
   molecular weight cutoff (Millipore, USA). These proteins were then
   incubated at 25 °C for 12 h in PBS buffer containing 10 mM GA[4], using
   ethanol as the control. After incubation, the proteins were centrifuged
   at high speed for 10 min and fractionated through size exclusion
   chromatography. The chromatography was executed using an AKTA Purifier
   System (Cytiva, USA) equipped with a Superdex 200 Increase 10/300 GL
   column. The system, equilibrated with PBS buffer, eluted samples with
   the same buffer at a flow rate of 0.3 ml/min at 4 °C. Eluted fractions
   (0.5 ml each) were continuously collected, and peak fractions were
   subjected to 4–13% Blue Native-PAGE and 10% SDS-PAGE, followed by
   immunoblotting with an anti-GST antibody.

Blue native polyacrylamide gel electrophoresis (BN-PAGE)

   To investigate the oligomerization of ZmGA2ox7.3, the peak fractions
   were subjected to 4–13% Blue Native PAGE (WSHTBio, China) according to
   the manufacturer’s instructions. The resolved samples were then
   transferred onto polyvinylidene difluoride (PVDF) membranes using a
   methanol-free transfer buffer (48 mM Tris-base, 39 mM glycine),
   followed by immunoblotting using the appropriate antibodies.

Enzyme activity assay

   For the enzymatic activity assay, we adopted specific procedures from a
   prior study^[216]27. The purified proteins GST-ZmGA2ox7.3, MBP-P7-1,
   His-ZmGDIα, and His-ZmGDIα-hel were each incubated in 1 ml of enzymatic
   activity buffer (100 mM Tris-HCl, 4 mM ascorbic acid, 4 mM
   2-oxoglutaric acid, 0.5 mM FeSO4, 4 mM DTT, 2 mg/ml BSA, and 1 mg/ml
   catalase, pH 7.9). The mixture was then incubated with 0.5 mM GA[4] at
   30 °C for 12 h. To terminate the reaction, 125 μl of acetic acid was
   added to each solution, and the mixture was kept at 30 °C for an
   additional hour. The level of the enzyme reaction product, GA[34], was
   subsequently quantified by Wuhan Greensword Creation Technology Co.
   Ltd. ([217]http://www.greenswordcreation.com). Each experiment was
   conducted with three biological replicates.

Mass Spectrometry (MS)

   The elution fractions were analyzed by mass spectrometry (MS) at the
   Biological Mass Spectrum Laboratory, College of Biological Sciences,
   China Agricultural University. The elution fractions from the two “Mu”
   peaks were collected and separated by BN-PAGE as described. The gel
   lanes were excised, sliced, subjected to in-gel digestion with trypsin,
   and redissolved in trifluoroacetic acid, following the methodology
   outlined in a previous study^[218]37. The extracted peptides were then
   analyzed using an LTQ Orbitrap Velos mass spectrometer (Thermo Fisher
   Scientific). Protein identification was performed using the Mascot
   search engine (Mascot Server 2.3, Matrix Science) by blasting with the
   UniProt protein database ([219]https://www.uniprot.org/). In the
   previous study, the false discovery rate (FDR) and significance
   threshold have been set^[220]37.

Transgenic tests in Arabidopsis thaliana

   The pSuper1300:ZmGA29X7.3-GFP and pSuper:ZmGA2ox7.3^3A-GFP constructs
   were transformed into Col-0 to obtain OE-ZmGA2ox7.3^3A (OE^3A−1 and
   OE^3A−2), and OE-ZmGA2ox7.3 (OE-1 and OE-2) transgenic plants,
   respectively. Stable T[2] generation of four above lines were used for
   experiments.

Measurement of plant hormones

   NIL-S and NIL-R were both artificially infested with planthoppers,
   either carrying RBSDV or not. Subsequently, the uppermost internodes
   and upper leaves were sampled at 60 dpi. The collected samples were
   used to measure the content of various endogenous phytohormones,
   including gibberellins (Gas), indole-3-acetic acid (IAA), and
   cytokinins (CKs). The quantification of these phytohormones was
   executed using UHPLC-MS/MS analysis (Thermo Scientific Ultimate 3000
   UHPLC coupled with TSQ Quantiva). This analysis was conducted by Wuhan
   Greensword Creation Technology Co. Ltd.
   ([221]http://www.greenswordcreation.com) according to established
   methods^[222]38. The experiments were conducted with three biological
   replicates.

Application of exogenous GAs

   Exogenous GA[3] and GA[4] treatments were performed following
   previously described methods^[223]39. Seeds from both the recipient B73
   and ZmGA2ox7.3-GFP-OE transgenic lines were germinated in pots filled
   with vermiculite soil and placed in a greenhouse. They were kept under
   conditions of 26 °C with a 16-h white light/8-h dark cycle for 6 days.
   Subsequently, the entire plants were uniformly sprayed with either
   GA[1](100 μg·ml^−1, 0.1% Tween-20), GA[3] (1 mg·ml^−1, 0.1% Tween-20),
   GA[4] (100 μg·ml^−1, 0.1% Tween-20) or 0.1% Tween-20 (as a control).
   Plants heights were measured 8 days post-treatment. Both the transgenic
   lines and the control plants were represented by at least three
   biological replicates.

Quantification and statistics analysis

   Statistical parameters are reported in the figures and figure legends.
   Statistical analyses were preformed using GraphPad Prism 8.3.0. Data
   were tested for normality using the Shapiro-Wilk normality test in
   GraphPad Prism 8.3.0. For normally distributed data, statistical
   significane between two groups was assessed using a two-tailed
   student’s t test or LSD’s t test. For comparisons among more than two
   groups, one-way ANOVA with Tukey’s test or LSD’s t-test was used, and
   different lowercase letters indicate a significant difference
   (P < 0.05). For data that were not normally distributed, the
   Mann-Whitney U test or Wilcoxon matched-pairs signed-rank test were
   used for two-group comparisons, while the Kruskal-Wallis test was
   employed for comparisons among more than two groups. Different
   lowercase letters indicate a significant difference (P < 0.05). The
   Pearson’s correlation coefficient (R) and P values were calculated
   using GraphPad Prism 8.3.0. All statistics analyses in this study were
   conducted using GraphPad Prism 8.3.0.

Reporting summary

   Further information on research design is available in the [224]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [225]Supplementary Information^ (9.4MB, pdf)
   [226]Peer Review File^ (6.4MB, pdf)
   [227]41467_2024_51726_MOESM3_ESM.pdf^ (71KB, pdf)

   Description of Additional Supplementary Files
   [228]Supplementary Data 1^ (16.1KB, xlsx)
   [229]Reporting Summary^ (4.1MB, pdf)

Source data

   [230]Source Data^ (38.1MB, zip)

Acknowledgements