Salicylic acid and RNA interference mediate antiviral immunity of plant stem cells

Significance

Plant viruses, like those infecting animals, threaten the health of their hosts, can spread rapidly and globally, and challenge agricultural productivity in many species. Understanding antiviral defense and keeping plants virus-free is therefore of the utmost importance. Virus exclusion from stem cells is not only relevant for the infected individual and the potential to recover from acute infection but is believed to also block infection of the host germline, ultimately preventing vertical transmission of disease from parent to offspring and spread of viral infections via seeds. In this paper, we describe a stem cell–specific antiviral pathway in plants, which is of significant biological and economic relevance as it maintains stem cells virus-free.

Abstract

Stem cells are essential for the development and organ regeneration of multicellular organisms, so their infection by pathogenic viruses must be prevented. Accordingly, mammalian stem cells are highly resistant to viral infection due to dedicated antiviral pathways including RNA interference (RNAi). In plants, a small group of stem cells harbored within the shoot apical meristem generate all postembryonic above-ground tissues, including the germline cells. Many viruses do not proliferate in these cells, yet the molecular bases of this exclusion remain only partially understood. Here, we show that a plant-encoded RNA-dependent RNA polymerase, after activation by the plant hormone salicylic acid, amplifies antiviral RNAi in infected tissues. This provides stem cells with RNA-based virus sequence information, which prevents virus proliferation. Furthermore, we find RNAi to be necessary for stem cell exclusion of several unrelated RNA viruses, despite their ability to efficiently suppress RNAi in the rest of the plant. This work elucidates a molecular pathway of great biological and economic relevance and lays the foundations for our future understanding of the unique systems underlying stem cell immunity.

Diseases caused by plant viruses are a constant threat to food and economic security worldwide, a reason for the extensive scientific investigation of plant–virus interactions. It remains poorly understood how viruses are excluded from stem cells in the shoot apical meristem (SAM) (1), even though this was first observed almost a century ago (2) and is common to many viral infections that efficiently spread throughout the rest of the plant. This particular antiviral capability of stem cells has been used to generate virus-free plants by tissue culture of meristems (3). After transition to flowering, SAM stem cells also generate floral organs containing the germline, so the absence of virus in these cells is thought to play a key role in restricting vertical transmission of infection to the host progeny (1). Although the meristematic transcription factor WUSCHEL is involved in RNA virus exclusion from stem cells in Arabidopsis thaliana (4) and RNAi and its suppression by viruses have also been implicated (1, 5, 6), the molecular mechanisms and dynamics of virus exclusion remain to be resolved.

To understand the events maintaining a virus-free niche in SAM stem cells, we challenged A. thaliana mutants lacking components of the RNAi pathway with Turnip mosaic virus expressing a fluorescent protein located at viral replication complexes (TuMV-6K2:Scarlet). Loss of RNA-dependent RNA polymerase 1 (RDR1) caused TuMV to invade stem cells (SI Appendix, Fig. S1). To document the dynamics of infection in wild type (WT) and rdr1, we performed time-course experiments to assess virus propagation in the stem cell layers expressing a nuclear reporter expressed through the pCLV3 promoter (Fig. 1A). This allowed a semiquantitative approach and revealed temporary entry of TuMV in the top L1-L2 stem cell layers at 13 to 15 days postinoculation (dpi), followed by subsequent exclusion (Fig. 1B). By contrast, rdr1 mutants showed consistent virus infection of stem cells through time (Fig. 1B). This occurred even earlier in a double mutant with rdr6 (SI Appendix, Fig. S2), while in a dcl2/dcl3/dcl4 (dcl234) mutant unable to generate small interfering (si)RNA, we observed the highest levels of viral fluorescence in stem cells (Fig. 1C). These results, confirmed by in situ hybridization (Fig. 1D), portray a dynamic and layered RNAi antiviral network specifically protecting stem cells from infection. Moreover, RNAi did not exclude TuMV at the earliest time points, in accordance with observations with cucumber mosaic virus (4). TuMV infection always caused loss of apical dominance, but while WT plants ultimately generated fertile flowers, rdr1 mutants did not (SI Appendix, Fig. S3), leading to sterility (Fig. 1E). RDR1 contributes to antiviral RNAi by increasing production of 21- to 22-nt-long virus-derived siRNA (vsiRNA) (7), presumably by generating double-stranded RNA (dsRNA) substrate for dicer enzymes. RDR1 significantly contributes to siRNA production from the whole TuMV genome (Fig. 1F and SI Appendix, Fig. S4 A, C, and D) but, surprisingly, it does not affect overall TuMV accumulation (SI Appendix, Fig. S4B). Finally, complementing rdr1 with WT or RNA polymerization–deficient alleles of RDR1 provides evidence that dsRNA synthesis by this protein determines vsiRNA amplification (Fig. 1G), exclusion from stem cells (Fig. 1H) and fertility (SI Appendix, Fig. S3).

Fig. 1.

Arabidopsis RDR1 protects meristematic stem cells from TuMV infection through dsRNA synthesis and small RNA amplification. (A) Determination of virus entry into the stem cell area by quantification of fluorescence in plants expressing H2B:Clover (green) in SAM stem cell nuclei and infected with TuMV-6K2:Scarlet (magenta). (B) Fluorescence values as in (A) from the top 35 μm of wild-type (WT) and rdr1 SAMs between 10 and 17 dpi. Color legend at the bottom. N: number of meristems analyzed. (C) As in (B), fluorescence values in WT and dcl234 triple mutants. (D) In situ hybridizations in vertical sections of WT, rdr1, and dcl234 meristems to detect TuMV RNA (purple) in infected plants at 15 and 22 dpi. (E) Seed production by mock and TuMV-infected WT and rdr1 plants. Each data point represents progeny of one plant. n.s.: P > 0.05; ***P < 0.001. (F) Distribution of vsiRNA along the TuMV-6K2:Scarlet genome, assessed by sRNA sequencing on duplicates of mock- and TuMV-infected apices (meristem and small flower buds). Red bars indicate siRNA revealed by the @6K2 northern blot probe. (G) Northern blot detection of TuMV-derived sRNA in rdr1 expressing WT (RDR1-WT:Clover) or catalytically inactive (RDR1-D801A:Clover) alleles of RDR1. RNA was extracted from systemically infected leaves; snRNA U6 is used as loading control. (H) Laser confocal microscopy of meristems from the lines in (G), 18 dpi. DAPI fluorescence in grayscale, Scarlet in orange-to-yellow, scale bar 20 μm.

RNAi in plants has both local and remote, mobile silencing capabilities (8), the latter being well documented for gene and transgene silencing but postulated indirectly for antiviral activity (9, 10). To assess whether RDR1 can act locally in stem cells, we generated rdr1 lines expressing pCLV3:RDR1. These were able to restore TuMV exclusion (Fig. 2A). Interestingly, the exclusion zone was expanded to the whole CLV3 promoter expression domain, establishing that RDR1 can prevent TuMV proliferation very efficiently and locally in stem cells. Yet, transcriptional reporters for the RDR1 promoter showed that both in non- and TuMV-infected plants, it drove expression in the lower meristem dome and the tissues below, but never in the core domain of stem cell virus exclusion (L1+L2 layers) (Fig. 2B). Along with reported expression in vasculature (11), this suggests that RDR1 is not produced in stem cells but prevents TuMV proliferation there through remote activity. Next, we asked whether RDR1 excludes TuMV from stem cells through sensu stricto antiviral RNAi or by regulation of gene expression through previously reported (12) and here confirmed host gene–derived virus-activated siRNA (vasiRNA) (SI Appendix, Fig. S5). To this end, we generated transgenic rdr1 lines producing RDR1-independent antiviral siRNA (siScar) through a hairpin (Fig. 2C and SI Appendix, Fig. S6). Production of siScar in stem cells of rdr1 restored TuMV exclusion in a sequence-specific manner (Fig. 2D). Equally, production of siScar in subjacent non–stem cell tissues through the RDR1 promoter yielded the same result (Fig. 2E), allowing us to conclude that RDR1 excludes TuMV from stem cells by remotely providing viral RNA sequence information to the RNAi machinery, without the need for host gene–derived siRNA. Interestingly, we did observe a significant TuMV-induced increase in RDR1-dependent vasiRNA derived from two of the methyltransferase genes suppressed by WUS (4), suggesting a further silencing mechanism (SI Appendix, Fig. S7A). However, no significant difference between mock and infected was observed in the accumulation of transcripts of WUS or any of the downstream methyltransferase genes previously reported (SI Appendix, Fig. S7B) (4), suggesting that either the WUS pathway is no longer active at our observation time point or that any changes in stem cell transcript abundance are diluted within the larger tissue harvested and therefore indiscernible.

Fig. 2.

RDR1 immunizes the SAM stem cells at a distance by providing TuMV-specific small RNA. (A) Laser confocal microscopy of meristems from rdr1 lines expressing RDR1 through the stem cell–specific pCLV3 promoter, infected with TuMV-6K2:Scarlet. (B) As in (A), of two independent transgenic lines expressing H2B:Clover (green) through the pRDR1 promoter after mock or TuMV-6K2:Scarlet inoculation. Insets and white arrowheads: L1+L2 core virus exclusion zone. (C) Schematic representation of the siScar experiments: rdr1 mutants generating Scarlet-specific siRNA (siScar) through a hairpin transgene are infected with TuMV containing the siRNA target sequence (TuMV-6K2:Scarlet) or not (TuMV-6K2:Clover). (D) Meristems of an rdr1 mutant line expressing siScar in stem cells through the pCLV3 promoter, infected with TuMV-6K2:Scarlet or TuMV-6K2:Clover. (E) As in (D), but lines expressing siScar through the pRDR1 promoter. (F) Meristems of lines expressing P19 through the pRDR1 promoter, after infection with TuMV-6K2:Scarlet. (G) As in (F), but of lines expressing P19 under the pUBI promoter. (A, D–G): DAPI fluorescence in grayscale, Scarlet or Clover in orange-to-yellow, scale bar 20 μm.

As siRNAs are the mobile signal in RNAi (10, 13), we tested whether blocking 21- to 22-nt-long siRNA in cells expressing RDR1 with the viral RNAi suppressor protein P19 (13) would stop the mobile signal and suppress the stem cell antiviral pathway. Surprisingly, this was not the case (Fig. 2F), in contrast to suppression of the pathway by P19 overexpression in all tissues—including stem cells (Fig. 2G and SI Appendix, Fig. S8). This suggests that the mobile RDR1-dependent antiviral signal is either not 21- to 22-nt siRNA or a high load of 21- to 22-nt vsiRNA that to be blocked requires a larger amount of P19 than that produced through the pRDR1 promoter. Both possibilities may explain why TuMV, which encodes a strong siRNA-sequestering RNAi suppressor (HC-Pro) (14, 15), cannot block this RNAi-based stem cell defense mechanism, at least in early stages of infection before large amounts of HC-Pro are produced. Our results do not exclude movement of RDR1 protein and/or mRNA, yet if this is the case, they are accompanied by an RNA molecule conferring virus sequence-specificity.

RDR1 expression is increased by salicylic acid (SA) in several plant species (16–21). SA is a key hormone in the activation of plant defenses against pathogens (22), including viruses (23), so we asked whether SA plays a role in TuMV exclusion from SAM stem cells. Indeed, TuMV completely invades stem cells of NahG plants (Fig. 3A) expressing a bacterial enzyme degrading SA (24). TuMV infection greatly increases SA accumulation in WT plants but not in NahG plants (Fig. 3B), and SA induction is required for RDR1 upregulation upon infection (Fig. 3C). Increasing the steady-state amount of SA in plants lacking the SA-degrading DMR6 gene (25) also leads to RDR1 upregulation (Fig. 3D and SI Appendix, Fig. S9 B and C). Interestingly, knock-out of the main SA isochorismate biosynthesis pathway through the sid2 mutation did not lead to stem cell invasion (SI Appendix, Fig. S9A), a result in line with previous evidence for a second SA biosynthetic pathway (26, 27). Accordingly, sid2 mutants show a partial activation of SA-responsive gene PR1 upon TuMV infection and transcriptional activation of RDR1 comparable to WT (SI Appendix, Fig. S9 D and E). The incomplete transcriptional activation of PR1 in sid2 suggested that stem cell exclusion may be uncoupled from the canonical NPR1-dependent transcriptional response to SA. This is indeed the case as npr1 mutants (28) were able to prevent TuMV invasion of stem cells (SI Appendix, Fig. S9F), coherently with previous reports of NPR1-independent antiviral mechanisms triggered by SA (29, 30). Crucially, the TuMV-dependent SA response does not change in rdr1 mutants (Fig. 3D), confirming that RDR1 activation depends on SA and not vice versa. As artificial overexpression of RDR1 in NahG plants does not restore TuMV exclusion from stem cells (nor reduces the increased TuMV accumulation observed in NahG) (SI Appendix, Fig. S10 A–D), transcriptional upregulation of RDR1 alone is not sufficient for SA-dependent virus exclusion. Therefore, either SA positively influences the RDR1 pathway by additional means such as RDR1 protein activity/stability, as suggested by a previous report on MtRDR1 (31), and/or ubiquitous RDR1 overexpression does not recapitulate SA-dependent induction. Our results do not exclude the possibility that SA acts through other molecular antiviral pathways. Nevertheless, for SA activation to result in stem cell exclusion, RDR1 must be present, since rdr1 mutants show SA activation yet virus meristem invasion (Figs. 1B and 3D). Next, we asked whether SA activation is linked to stem cell exclusion of other virus species. We found that Turnip crinkle virus (TCV, family Tombusviridae) and Turnip yellow mosaic virus (TYMV, family Tymoviridae), species taxonomically distant from each other and TuMV (family Potyviridae), both elicit an SA response in WT Arabidopsis, albeit to different extents (Fig. 3E). TCV, the stronger inducer of SA, also up-regulates RDR1 expression (SI Appendix, Fig. S10 E and F). In situ hybridizations revealed that both TYMV and TCV were excluded from SAM stem cells (Fig. 3 F and G). Conversely, Tobacco rattle virus (TRV, family Virgaviridae), which infects meristems in Nicotiana benthamiana (6), did not elicit an SA response (Fig. 3H) and was not excluded from stem cells in A. thaliana (Fig. 3I). These results on four unrelated virus species therefore suggest that SA activation is correlated to the maintenance of virus-free SAM stem cells.

Fig. 3.

Increased salicylic acid (SA) production upon infection determines TuMV stem cell exclusion, increases RDR1 expression, and correlates with the exclusion of other virus species from stem cells. (A) Laser confocal microscopy of meristems from WT and SA-suppressing NahG plants infected with TuMV-6K2:Scarlet. (B) SA accumulation in WT and NahG plants before infection (T0) and after mock or TuMV-6K2:Scarlet inoculation. Each dot is a biological replicate: pool of tissues from five to six plants. (C) qRT-PCR on RNA from samples in (B) to assess RDR1 mRNA accumulation. Each bar is a biological replicate; each dot is a technical replicate. (D) As in (B) on WT, NahG, rdr1, and dmr6 plants. (E) As in (B), but on WT plants infected with TYMV or TCV. Mock values are the same as in (D). (F) In situ hybridization to detect TYMV RNA (purple) in meristems of mock- or TYMV-inoculated WT plants, 15 dpi. (G) As in (F), to detect TCV RNA in meristems of mock- or TCV-inoculated WT plants, 15 dpi. (H) As in (B), but on WT plants infected with TRV-Scarlet. (I) As in (A), on WT plants after mock or TRV-Scarlet infection. (A), (I): DAPI fluorescence in grayscale, Scarlet in orange-to-yellow, scale bar 20 μm. (B, D, E, and H): n.s.: P > 0.05; *P < 0.05; **P < 0.01, ****P < 0.0001.

Next, we investigated whether RNAi and SA are necessary for stem cell exclusion of TCV and TYMV. TYMV and TCV can completely invade stem cells of dcl234 mutants (Fig. 4 A and B), indicating that small RNAs are required for exclusion. Furthermore, the expansion of the TCV exclusion zone over time is also dependent on small RNAs (Fig. 4C). Interestingly, neither rdr1 nor NahG plants showed stem cell invasion, indicating that the SA/RDR1 pathway is not necessary for exclusion of these two viruses. TCV strongly induced SA/RDR1 production (Fig. 3E and SI Appendix, Fig. S10), suggesting that this pathway may be involved in—but not strictly necessary for—TCV exclusion from stem cells. These results suggest that for TCV and TYMV, either primary DCL products are sufficient and RDR1-dependent amplification of vsiRNA production is not required or other RDR enzymes are involved and necessary here. Taken together, our observations establish that RNAi is essential in maintaining a virus-free SAM stem cell niche. This is remarkable since, like TuMV, also TCV and TYMV encode for potent suppressors of RNAi (32, 33). Accordingly, dcl234 mutants showed a modest increase in viral RNA accumulation, if any (Fig. 4D), indicating that host RNAi has little effect on TCV and TYMV replication/propagation in Arabidopsis plants at large. Strikingly however, in addition to ensuring stem cell exclusion, RNAi is required for TYMV- and TCV-infected plants to produce seeds (Fig. 4E). Whether virus stem cell exclusion and fertility are connected remains to be determined, but artificial exclusion of TuMV through RDR1 or siScar expression in stem cells alone (Fig. 2 A and D) does not rescue seed production in rdr1 (SI Appendix, Fig. S11), suggesting that virus exclusion per se is not sufficient to ensure fertility.

Fig. 4.

Small RNAs determine exclusion from stem cells of viral species unrelated to TuMV and are required for fertility of infected plants. (A) In situ hybridization to detect TYMV RNA in meristems of TYMV-inoculated WT, dcl234, rdr1, and NahG plants, 15 dpi. Viral RNA results in blue-purple color. (B) As in (A) to detect TCV RNA in meristems of the same genotypes after TCV inoculation, 15 dpi. (C) As in (B), showing whole floral apices from WT and dcl234 infected with TCV, 15 and 22 dpi. (D) Northern blot analysis of viral RNA accumulation in TYMV- and TCV-infected WT and dcl234 plants, each in duplicate, from systemically infected leaves at 9 dpi. Methylene blue staining is used as loading control. (E) Seed production by mock-, TYMV-, and TCV-infected WT and dcl234 plants. Each data point is progeny of one plant. n.s.: P > 0.05; ****P < 0.0001.

In summary, our study describes a broad-range antiviral RNAi pathway, which in the case of TuMV is non-cell autonomous and activated by salicylic acid, that maintains the vital plant SAM stem cells free of pathogenic viruses. Crucially, unlike RNAi in the rest of the plant, this pathway can successfully evade viral suppression, pointing to vital aspects of small RNA biology that remain to be elucidated. This work provides a robust molecular framework for a plant stem cell–specific defensive system of great biological and economic relevance.

Materials and Methods

Molecular Cloning.

All the binary plasmids used in this study were generated through Golden Gate assembly. Refer to Data, Materials, and Software Availability for the complete plasmid sequences. All transgenes for Arabidopsis transformation were assembled using the GreenGate system (34), by BsaI digestion (BsaI-HF-v2, New England Biolabs #R3733S) and T4 Ligase ligation (Thermo Scientific #EL0014) of entry vectors into binary plasmids. The entry vectors were made by amplifying the sequences of interest by PCR using Q5 HiFi DNA polymerase (NEB #M0491L) with primers containing the BsaI cut site in the appropriate orientation and the standard “sticky ends” corresponding to GreenGate A-F units (34), then ligating them into the pGGA-F GreenGate vectors or into the Golden Gate-ready pMiniT™2.0 (NEB #E1203S). Sequences containing BsaI cut sites (such as AtRDR1) were divided into several entry vectors, where final assembly would introduce silent mutations, preventing further digestion of the assembled products. Transgenes were assembled into the GreenGate pGreen-based pGGZ003 destination vector (pCLV3::H2B:Clover; pUBI::H2B:Clover and pUBI::P19) or pGGSun, a version of pSUN (35) we adapted for GreenGate cloning (pCLV3::RDR1-WT:Clo; pCLV3::siScar; pRDR1::RDR1-WT:Clo; pRDR1::RDR1-D801A:Clo; pRDR1::siScar; pRDR1:P19; pUBI::RDR1-WT:Clo; pUBI::siScar) (see SI Appendix, Table S1 for entry vectors used). The Golden Gate assembly-ready pSun, pGGSun, was obtained by amplifying i) the pSun backbone adding BsaI sites and ii) a ccdB selection cassette to insert between the BsaI sites. The two PCR products were then assembled with Gibson Assembly Mastermix (NEB #E2611). All RDR1 constructs contain the genomic sequence of RDR1 (AT1G14790). The catalytically inactive RDR1-D801A allele was generated by mutating aspartic acid 801 in the RDR1 protein to alanine. A corresponding mutation of the last aspartic acid in the conserved DxDxD triplet was shown to abrogate RNA polymerization capability in A. thaliana RDR2 and RDR6 (36, 37). The constructs using the pRDR1 promoter do not only include the sequence upstream of RDR1 but also the sequence downstream of the RDR1 gene, inserted downstream of the sequence of interest (RDR1/H2B:Clover/siScar/P19). The same is valid for constructs with the pCLV3 promoter.

TuMV and TRV2 virus clones were generated by cloning segments of the viral genomes into pMini entry vectors, as described above, then seamlessly assembling them into pGGSun downstream of a 35S promoter and followed by a NosT terminator. In the case of TRV2-Scarlet (sequence of the PPK20 isolate), an HDV ribozyme was placed after the viral sequence to ensure cleavage for correct 3′ ending of the RNA. The mScarlet sequence, preceded by the PEBV CP subgenomic promoter, was inserted after the TRV CP-coding sequence. In the case of TuMV-6K2:Scarlet and TuMV-6K2:Clover, a sequence coding for the viral 6K2 protein (38) fused to Scarlet or Clover, respectively, and flanked by amino acid sequences cut by the viral proteases was inserted into the polycistronic TuMV sequence (UK1 isolate) between the P1- and HC-Pro-coding sequences. The TRV1 plasmid (p1586 - pCB-TRV1) was generated by cloning the cDNA from TRV1 isolate PPK20 into binary vector pDIVA (39), between the 35S promoter and the HDV ribozyme, by blunt ligation into the PCR-amplified backbone.

Plant Material.

For ease of interpretation, in this manuscript, wild type (WT) is used to refer to A. thaliana Col-0 ecotype plants, which is also the genetic background of all mutants. Arabidopsis mutant lines rdr1-1 (40), rdr6-15, rdr1-1/rdr6-15 (14), dcl2-1/dcl3-1/dcl4-2 (41), NahG (24), sid2-1 (42), npr1-1 (28), and dmr6-2 (25) were previously described (See SI Appendix, Table S2 for stock and genotyping information). Genotyping was performed by standard PCR of leaf DNA extracts. Transgenic Arabidopsis lines were generated by transforming A. tumefaciens GV3101 with the plasmid of interest and using the resulting cultures to perform floral dip. The transformants were selected in the appropriate manner (antibiotic resistance or seed coat fluorescence) and propagated to the third generation after transformation, when seed stocks homozygous for the transgene were selected and further used for infection experiments. The lines used for time-course experiments (pCLV3::H2B:Clover in Col-0, rdr1, rdr6, rdr1/rdr6, and dcl2/dcl3/dcl4 backgrounds) were obtained by crossing a Col-0/pCLV3::H2B:Clover line with rdr1-1/rdr6-15 or dcl2-1/dcl3-1/dcl4-2 and selecting the various mutant combinations by genotyping. All other transgenics were obtained by directly transforming the genotypes in question. In all infection experiments, plants were grown on soil at 12 h/12 h day/night cycles until infection, when they were moved to 16-h-/8-h-long day conditions to induce flowering. Plants were infected 3.5/4 wk after germination (TuMV, TCV, TYMV) or 2 wk after germination (TRV).

Virus Infection, Tissue Sampling, and Meristem Preparation.

Inoculum of TuMV-6K2:Scarlet and TuMV-6K2:Clover was obtained by inoculating N. benthamiana plants with A. tumefaciens cultures containing the respective plasmids as previously described (43) followed by harvesting and freezing the systemically infected leaves. Inoculum of TCV and TYMV was obtained by harvesting and freezing Arabidopsis leaves systemically infected after rub inoculation. Inoculum of TRV-Scarlet was obtained by harvesting and freezing Arabidopsis leaves systemically infected after inoculation of A. tumefaciens cultures containing TRV1 and TRV2-Scarlet plasmids as previously described (43). During infection experiments, inoculum was prepared by grinding frozen plant tissue in liquid nitrogen with a mortar and pestle and then resuspending the powder in 50 mM sodium phosphate buffer, pH 7.2, 0.2% sodium sulfite. After incubating on a wheel at 4 °C for 10 min, the homogenate was centrifuged at 1,000 g for 2 min, and the supernatant was kept on ice and used as inoculum. Plants were sprinkled with Celite 545 (Merck), and a cotton swab was dipped in the inoculum and used to gently rub the leaves, five to six leaves per plant. For molecular analysis, tissues were harvested at 8 to 9 dpi (systemic leaves) or 15 to 16 dpi (inflorescence apices), frozen, and stored at −70 °C. Each sample is a pool of tissues from four to five plants. For meristem preparations, the main inflorescence of each plant was removed and dissected under a light microscope until only the smallest flower buds and shoot apical meristem remained along with 1 to 2 mm of stem. Unless indicated otherwise in figures or figure legends, meristems were generally sampled at 15 to 18 dpi, depending on the experiment, with the exception of the pRDR1::H2B:Clover experiments at 12 to 13 dpi. Precise time points are indicated in the additional microscopy data (see Data, Materials, and Software Availability). If meristems were to be observed by confocal microscopy, the dissected meristems were incubated 40 min in fixing solution (44) (1× MTSB, 2% paraformaldehyde, 0.1% Triton-X) at 37 °C and then stored in MTSB at 4 °C for a maximum of 10 d. Then, 3 to 4 d before observation, the meristems were incubated in ClearSee (10% w/v xylitol, 15% w/v sodium deoxycholate, 25% w/v urea) at 4 °C, with the addition of 10 mg/L DAPI the day before observation. If the meristems were to be used for in situ hybridizations, they were incubated after dissection over night at 4 °C in fixing solution (4% formaldehyde, 50% ethanol, 5% glacial acetic acid, 1× PBS) and dehydrated by changing the buffer to 50%, then 70% ethanol in 1× PBS.

Confocal Microscopy and Image Analysis.

Meristems were mounted on glass slides in ClearSee and imaged with a Zeiss LSM880 laser confocal microscope. The following laser wavelengths were used: 405 nm for DAPI, 488 nm for Clover, and 561 nm for Scarlet. Further image processing was carried out with FIJI/ImageJ. For single meristem image assembly, images were cropped, rotated if necessary, split into single channels, LUTs were assigned (grayscale for DAPI, green for Clover, OrangeHot for Scarlet or Clover in Fig. 2 D and E), and brightness/contrast were adjusted. For Scarlet fluorescence signal, brightness was regulated until the high-signal zones were in yellow color, with the same settings for all genotypes. The single channel images were then merged, and a 20-μm scale bar was added (SI Appendix, Fig. S1). For time course quantification experiments (Fig. 1 B and C and SI Appendix, Fig. S2), 7 to 12 meristems were imaged per genotype/time-point without changing laser intensities within an experiment. Images were then analyzed with FIJI (45) using a macro developed for this task (see Data, Materials, and Software Availability on Zenodo): With meristems oriented vertically, an equally wide vertical section of each was selected for Clover and Scarlet fluorescence quantification, one measurement every 149 nm. The data were then imported into Microsoft Excel spreadsheets. Since differences in sample depth and degree of clearing caused differences in absolute fluorescence between meristems, the Clover fluorescence values in each meristem were converted to values on a 0-to-100 scale. The corresponding Scarlet values were normalized for each data point. These normalized values were then used to calculate the plotted average and SD.

In Situ Hybridization.

Meristems, after being prepared as described above, were stained in 1% w/v eosin in 70% ethanol and then infiltrated with xylene substitute and paraffin in a Diapath Donatello I tissue processor, after which they were cast into paraffin blocks using a Sakura Tissue Tek TEC5 (approximately 20 meristems/block). The blocks were then cut into 2-μm-thick sections that were transferred onto glass microscopy slides, which were screened for ones containing central sections of meristems. DIG-labeled RNA probes were generated with the DIG RNA Labeling Kit T7/SP6 (Roche #11175025910); see SI Appendix, Table S3 for the primers used to generate the DNA templates. In situ hybridization was then performed as previously described (46, 47), with minor variations, all solutions being prepared with DEPC-treated water. Slides were twice incubated 10 min in Histo-Clear II (National Diagnostics #HS-202), 5 min in 100% ethanol twice and rehydrated through serial passages in 90%, 70%, 50%, and 30% ethanol, and then in Tris-EDTA pH 7.5. Sections were then treated with Proteinase K (Roche #3115836001), washed in 1× PBS, incubated 10 min in 4% paraformaldehyde, dehydrated through serial ethanol washes, and air-dried. After probe denaturation for 3 min at 80 °C, hybridization with 50 to 100 ng DIG-labeled probes per slide was carried out O/N at 50 °C in 150 μL hybridization solution: 50% formamide, 10 mM Tris base, 300 mM NaCl, 5 mM EDTA, 10 mM Na₂HPO₄, 1× Denhardt’s solution (Sigma Aldrich #D2532-5ML), 10% dextran sulfate, and 0.5 μg/μL tRNA (Roche #10109517001). Slides were briefly washed in 2× SSC, then incubated in 0.2× SSC for 2 h at 55 °C and treated with RNase A (Thermo Scientific #EN0531) at 37 °C for 30 min, and then 1 h in 0.2× SSC at 55 °C. Slides were washed 10 min in washing buffer and incubated 1 h in blocking buffer (both Roche #11585762001). Anti-DIG antibody was added (Roche #11093274910 - 1:1500 dilution) and incubated for 1 h 45 min at room temperature, washed for 1 h, incubated in TNM5 (100 mM Tris pH 9.5, 100 mM NaCl, 5 mM MgCl₂) three times for 2 min, then O/N in TNM5 with 10% w/v polyvinyl alcohol, and 10 μL/mL NBT/BCIP (Roche #11697471001). Slides were mounted with Aqua-Poly/Mount (Polysciences #18606-20) and scanned with a Pannoramic 250 slide scanner at 40× magnification.

Salicylic Acid Quantification.

For SA quantification, four to five replicates of each genotype/virus were collected, each replicate being a pool of systemically infected tissue from five plants. Tissues were frozen, pulverized, and stored at −70 °C. Aliquots of tissue were weighed, ground with glass beads, and 1 mL of 80% acetonitrile (Sigma-Aldrich #34881) and 50 μL internal standard (5-fluorosalicylic acid, 1 mg/L) were added per sample. The resulting solution was then vortexed and placed in a shaker at room temperature for 1 h at 1,400 rpm shaking speed. Samples were centrifuged 5 min at 13,000 rpm, and supernatant was transferred to new tubes. After drying most of the liquid with a vacuum pump, quantification was performed by HPLC as previously described (48), with the difference that a Nucleodur 100-5 NH2 125- × 4-mm column (Macherey-Nagel, #760730.40) and an eluent consisting of 8.5% acetonitrile and 25mM formic acid pH 4 were used (49). Data were analyzed and plotted with Microsoft Excel; P values were calculated through standard pairwise Student t tests, two-tailed, assuming unequal variance.

Northern Blotting and qRT-PCR.

RNA extraction was performed with TRI Reagent (Zymo Research #R2050-1-200). Briefly, flash-frozen plant tissues were pulverized with glass beads, 1 mL TRI Reagent and after clearing 300 μL chloroform were added. After shaking and centrifugation, one volume isopropanol was added to the aqueous phase and incubated at least 1 h on ice. After centrifugation, the pellet was washed with 80% ethanol, dried, resuspended in RNase-free water, and the RNA concentration measured. RNA was stored at −20 °C. Small RNA northern blotting was performed as previously described (50) on 10 to 50 μg RNA, using a standard BioRad PAGE system for electrophoresis and EDC chemical cross-linking (Sigma Aldrich #E7750) onto Hybond NX nylon (GE Healthcare #RPN203T). Membranes were probed with α-³²P-CTP-labeled (Agilent #300385) PCR products (@6K2, @Scarlet) or γ-³²P-ATP-labeled (Thermo Scientific #EK0031) DNA oligonucleotides (@U6), hybridizing overnight at 42 °C in 1 mM EDTA, 7% SDS, and 500 mM sodium phosphate, pH 7.2. After three washes of 15 min at 45 °C in 2% SDS, 2× SSC, membranes were exposed to phosphor screen and signals revealed by an Amersham Typhoon. For high molecular weight northern blotting to detect viral RNA, 5 μg RNA was initially denatured by incubating with 15% v/v deionized glyoxal at 50°C for 1 h. Samples were then run in a 1% agarose gel in 20 mM sodium phosphate pH 7.2; capillary transfer to nylon membrane was performed overnight followed by UV cross-linking. After staining with methylene blue, membranes were probed with γ-³²P-ATP-labeled DNA oligonucleotides (@TCV, @TYMV) as described above. For qRT-PCR quantification, 5 μg RNA was treated with TURBO™ DNase (Invitrogen #AM2238), and 500 ng of this was used for cDNA synthesis with oligo-dT primer using RevertAid H Minus First Strand cDNA Synthesis (Thermo Scientific #K1632). qPCR on cDNA was performed with the FastStart Essential DNA Green Master kit (Roche #06402712001) using a Roche LightCycler 96 and corresponding proprietary software. Expression levels of RDR1 and PR1 mRNA were normalized to housekeeping gene AtSAND (AT2G28390), while levels of TuMV gRNA were normalized to AtGAPDH (AT1G13440). Data were analyzed and plotted with Microsoft Excel. P values were calculated through standard pairwise Student t tests, two-tailed, assuming unequal variance, pooling technical and biological replicates for each genotype/condition tested. See SI Appendix, Table S3 for primer sequences.

RNA Sequencing and Analysis.

All RNA libraries and sequencing were performed by the Next Generation Sequencing Facility (Vienna BioCenter Core Facilities). sRNA libraries were generated with QIAgen miRNA library kit (QIAgen #331502). Prior to analyzing the sequencing data, adapters were removed from sRNA library data by using cutadapt v1.18, selecting read length from 18 to 26 nt. Processed reads were aligned to the Arabidopsis genome (TAIR10) and TuMV-Scarlet sequence using bowtie2 v2.3.5 (51), i) allowing unique mapping to the TuMV-6K2:Scarlet sequence to assess the proportion of viral sRNAs and ii) allowing 1,000 times multimapping for gene-derived sRNA enrichment analysis. The aligned reads from multimapping were sorted by size (21 nt, 22 nt, and 24 nt) for further analysis. The small RNA metaplots were generated by using Deeptools v.3.3.1 (52) with “bamCoverage” adding “CPM” parameter. The annotation of small RNAs to genes was done by using featureCounts (53) with Araport 11 annotation. DESeq2 (54) was used to analyze small RNA enrichment on genes with a cutoff of P.adj. < 0.05, log2 fold change > |1| and > 10 reads in both replicates. Visualization of the data was done by using the packages tidyverse (55) and ggplot2 (56). For mRNA seq, Smart-seq3 sequencing libraries were generated and sequencing reads were processed with nf-core/rnaseq (57). Due to the redundancy of the TAIR annotations “transposable element” and “transposable element gene,” we used a custom annotation file containing TAIR10 features plus “transposable elements” without “transposable element genes” and added sequences of transgenes (see ref. 58 for details). The following specifications deviated from the default of the rnaseq pipeline: --additional_fasta (containing the sequence for Clover and TuMV_6K2_Scarlet), --clip_r1 19, --three_prime_clip_r1 2. Differential gene expression analysis was performed with DESeq2 (54). Visualization of the data was achieved using R and Bioconductor (59) including the packages tidyverse and ggplot2.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Additional and source data have been deposited on Zenodo at the following DOI: 10.5281/zenodo.8316269 (60). This includes panels with complete confocal and in situ microscopy experiments, raw data from time course fluorescence quantifications, blotting, qRT-PCRs, SA measurements, complete sequences of virus and transgene plasmids, and sRNA and mRNA analyzed data. Small RNA sequencing data have been deposited on GEO under accession number GSE221157 (61) and mRNA sequencing under accession number GSE242499 (62).