A central role for the transcriptional regulator VtlR in small RNA-mediated gene regulation in Agrobacterium tumefaciens

LysR-type transcriptional regulators (LTTRs) are the most common type of transcriptional regulators in prokaryotes and function by altering gene expression in response to environmental stimuli. In the class Alphaproteobacteria, a conserved LTTR named VtlR is critical to the establishment of host-microbe interactions. In the mammalian pathogen Brucella abortus, VtlR is required for full virulence in a mouse model of infection, and VtlR activates the expression of abcR2, which encodes a small regulatory RNA (sRNA). In the plant symbiont Sinorhizobium meliloti, the ortholog of VtlR, named LsrB, is involved in the symbiosis of the bacterium with alfalfa. Agrobacterium tumefaciens is a close relative of both B. abortus and S. meliloti, and this bacterium is the causative agent of crown gall disease in plants. In the present study, we demonstrate that VtlR is involved in the ability of A. tumefaciens to grow appropriately in artificial medium, and an A. tumefaciens vtlR deletion strain is defective in motility, biofilm formation, and tumorigenesis of potato discs. RNA-sequencing analyses revealed that more than 250 genes are dysregulated in the ∆vtlR strain, and importantly, VtlR directly controls the expression of three sRNAs in A. tumefaciens. Taken together, these data support a model in which VtlR indirectly regulates hundreds of genes via manipulation of sRNA pathways in A. tumefaciens, and moreover, while the VtlR/LsrB protein is present and structurally conserved in many members of the Alphaproteobacteria, the VtlR/LsrB regulatory circuitry has diverged in order to accommodate the unique environmental niche of each organism.


Results
Expression of the sRNA AbcR1 is dependent on VtlR, and deletion of vtlR in A. tumefaciens results in a significant lag in growth in vitro. The vtlR gene, designated as atu2186, is located on the circular chromosome of A. tumefaciens strain C58 (Fig. 1A). Directly upstream of vtlR is trxB (atu2185), a gene encoding a thioredoxin reductase; and downstream of vtlR are the genes encoding the AbcR sRNAs, abcR1 and abcR2. It is interesting to note that the genetic organization of trxB and vtlR is well conserved in the class Alphaproteobacteria.
In B. abortus, VtlR has been shown to positively regulate the tandemly encoded sRNA abcR2 on chromosome I; however the expression of abcR1 on chromosome II of B. abortus is not regulated by VtlR 22 . Dissimilar from B. abortus, abcR1 and abcR2 are encoded directly downstream of vtlR/lsrB on the same chromosome in A. tumefaciens str. C58 (Fig. 1A) and S. meliloti 1021. It was unknown, however, if either of these sRNAs were regulated by the homolog of VtlR/LsrB (Atu2186 and SMc01225) in A. tumefaciens or S. meliloti respectively. To test whether VtlR regulated abcR1 and/or abcR2 in A. tumefaciens, northern blot analyses were performed to measure expression of AbcR1 and AbcR2 in wild-type A. tumefaciens, the isogenic deletion strains of abcR1 (∆abcR1) and abcR2 (∆abcR2), a double deletion strain of abcR1 and abcR2 (∆abcR1/2), and an isogenic deletion strain of vtlR (∆vtlR) (Fig. 1B). Bands representing the AbcR1 and AbcR2 transcripts were clearly visible when the A. tumefaciens culture is grown to an O.D. of 4.0 (Fig. 1B). Importantly, the AbcR1 and AbcR2 transcripts are not present in the isogenic deletion strains, confirming that the strains are in fact deletions of the indicated genes. Furthermore, the northern blot analyses also demonstrated that the expression level of AbcR2 is unaffected in ∆vtlR, but AbcR1 production is abolished in ∆vtlR. These data indicate that VtlR positively influences the expression of AbcR1, but not AbcR2, in A. tumefaciens. A similar strategy was employed to measure the expression of AbcR1 and AbcR2 in S. meliloti 1021 and S. meliloti 1021:: ∆lsrB. Northern blot analysis revealed no change in the expression of the AbcRs in the absence of lsrB; indicating that lsrB does not regulate the expression of AbcR1 or AbcR2 in S. meliloti (Fig. S1).
In the plant symbiont S. meliloti, a deletion of lsrB resulted in a significant growth defect 18,20 ; however, a deletion of vtlR in B. abortus showed no significant differences when grown in nutrient rich or nutrient limiting media 22 . A. tumefaciens str. C58, ∆vtlR, ∆abcR1, ∆abcR2, and ∆abcR1/2 were grown in a nutrient-rich medium, and the number of colony-forming units (CFU) were measured every 12 h to examine growth over time in the deletion backgrounds to assess the necessity of vtlR for growth of A. tumefaciens (Fig. 1C). A. tumefaciens ∆vtlR displayed no difference in growth kinetics (i.e., doubling time during exponential growth) compared to the parental strain; however, similar to S. meliloti, A. tumefaciens ∆vtlR exhibited a lag in growth in liquid medium and small colony phenotype on agar medium when compared to the parental strain C58 (Fig. 1C,D). Isogenic deletions of abcR1 and abcR2, and a double deletion of abcR1 and abcR2, resulted in no differences in A. tumefaciens growth when compared to the parental strain (Fig. 1C). The lag in growth of the A. tumefaciens ∆vtlR was rescued by in-trans complementation with the plasmid pSRK-Km harboring an IPTG-inducible wild-type vtlR Scientific RepoRtS | (2020) 10:14968 | https://doi.org/10.1038/s41598-020-72117-0 www.nature.com/scientificreports/ gene ( Fig. 1D) 33 . As a control, A. tumefaciens ∆vtlR harboring an empty pSRK-Km plasmid showed no difference in growth compared to A. tumefaciens ∆vtlR (Fig. 1D). Altogether, these data reveal that VtlR positively regulates the sRNA AbcR1, and that VtlR is critical to the growth of A. tumefaciens in artificial medium.
VtlR is critical to the virulence, biofilm formation, and motility of A. tumefaciens str. C58. VtlR was previously shown to be necessary for the symbiosis and pathogenesis of S. meliloti and B. abortus 18,22 . Moreover, recent work has demonstrated a role for VtlR in the pathogenesis of A. tumefaciens for attachment of the bacteria to the roots of Arabidopsis plants, as well as for the efficient transformation of tobacco leaves 32 . Since attachment and transformation are crucial virulence factors for the pathogenesis of A. tumefaciens, we hypothesized that A. tumefaciens ∆vtlR would also exhibit reduced tumorigenesis compared to the wild-type strain C58. To test this hypothesis, A. tumefaciens ∆vtlR was assessed for its ability to form tumors in experimentally infected potatoes ( Fig. 2A). Compared to A. tumefaciens C58, the vtlR deletion strain caused, on average, the formation of fewer tumors 21 days post-infection ( Fig. 2A). However, deletion of abcR1, abcR2, or a double deletion of abcR1 and abcR2 did not have any statistical difference in tumor formation compared to C58 (data not shown).
To further examine the function of VtlR in A. tumefaciens, biofilm production and swimming motility by ∆vtlR was assessed (Fig. 2B,C). A. tumefaciens ∆abcR1 and ∆abcR2 displayed no differences in biofilm production nor swimming motility when compared to the parental strain C58; however, the ∆abcR1/2 strain displayed The vtlR gene (atu2186) is located on the A. tumefaciens circular chromosome, directly downstream of a thioredoxin reductase, trxB (atu2185). vtlR is upstream of the tandemly encoded AbcR sRNAs, abcR1 and abcR2. Directly downstream of abcR2 is an ArsR-type transcriptional regulator, arsR (atu2187). (B) Northern blot analyses for AbcR small RNAs. RNA was isolated from A. tumefaciens C58, ∆vtlR, ∆abcR1, ∆abcR2, and ∆abcR1/2 cultures grown to optical densities (OD 600 ) of 1.0, 3.0, or 4.0, separated on a denaturing polyacrylamide gels, transferred to nitrocellulose membranes, and probed with specific radiolabeled oligonucleotides. 5S ribosomal RNA was blotted for as a loading control. (C) In vitro growth kinetics of A. tumefaciens strains. The A. tumefaciens C58, abcR1, abcR2, abcR1/2, and vtlR strains were grown in LB broth, and at specified time points, samples from each culture were taken, serial diluted and plated on AT-agar plates to determine colony forming units (CFUs www.nature.com/scientificreports/ modest, but statistically insignificant, differences (data not shown). The A. tumefaciens ∆vtlR produced significantly less biofilm than the parental strain and displayed decreased swimming motility (Fig. 2B,C). Taken together, these data demonstrate that VtlR is important for the ability of A. tumefaciens to efficiently produce biofilms as well as sufficiently swim through agar, and this is consistent with previous work produced by Tang et al. 32 . However, it should be noted that it is possible that the growth defect exhibited by the vtlR deletion strain could be partly responsible for the observed decrease in motility, but the slower growth of ∆vtlR likely does not contribute to the decrease in biofilm formation, because the data are normalized to bacterial growth.
The small RNA AbcR1 primarily regulates ABC-type transport systems. To  www.nature.com/scientificreports/ deletion strain exhibits a growth defect compared to the other strains (Fig. 1C), RNA was isolated from cultures are the same OD 600 (OD 600 = 1.0) rather than a specific time point to ensure that the cells were in the same phase of growth. Additionally, the number of colony-forming units from each culture was also determined to confirm that the cultures contained similar numbers of viable bacteria. Although AbcR1 was dispensable for A. tumefaciens virulence, the regulon of AbcR1 is robust with almost 100 genes differentially expressed ≥ 3-fold in ∆abcR1 when compared to the wild-type strain C58 (Supplementary Table S1). In comparison, only 8 genes were differentially expressed in ∆abcR2, and interestingly, half of the AbcR2-regulated genes encode hypothetical proteins (Supplementary Table S2). The significant difference in terms of the number of transcripts showing altered levels between AbcR1 over AbcR2 is in line with previous studies showing that AbcR1 exhibits more regulatory activity than AbcR2 in Agrobacterium 25,29 . Overall, the majority of genes differentially expressed in the abcR1 deletion strain encode components of ABC-type transporters and membrane proteins (56%), followed by hypothetical proteins (22%), and genes involved in enzymatic processes (18%) (Fig. 3A). Many of these AbcR1 regulated genes have been confirmed by previously published proteomic analysis 25 . Overview of RNA-sequencing analyses of A. tumefaciens ∆abcR1 and ∆vtlR. A. Pie chart summarizing genes differentially expressed (≥ 3-fold) in A. tumefaciens ∆abcR1. Blue: transport systems and membrane proteins (56%); brown: hypothetical proteins (22%); yellow: metabolism, signaling, enzymatic processes (18%); red: phage-related (2%); orange: transcription, translation (1%); green: chemotaxis, motility (1%). (B) Pie chart summarizing genes differentially expressed (≥ 3-fold) in A. tumefaciens ∆vtlR. Blue: transport systems and membrane proteins (33%); brown: hypothetical proteins (33%); yellow: metabolism, signaling, enzymatic processes (19%); orange: transcription, translation (5%); grey: secretion systems (4%); pink: polysaccharide biosynthesis (3%); red: phage-related (2%); purple: conjugation (1% www.nature.com/scientificreports/ the A. tumefaciens VtlR transcriptional regulon is comprised of over 250 genes, including the AvhB type IV secretion system and a conjugal transfer system. RNA-seq analysis identified over 250 dysregulated genes in A. tumefaciens ∆vtlR when grown in LB broth (i.e., nutrient rich medium). Genes differentially expressed ≥ 3-fold in ∆vtlR are depicted in Supplementary Table S3, and qRT-PCR was utilized to validate differential expression of several of these targets in ∆vtlR (Table S4). An outline of the functional classification of the proteins encoded by the differentially expressed genes is shown in Fig. 3B. Of these dysregulated genes, 33% are predicted to encode membrane proteins and transport systems, 33% encode hypothetical proteins, and approximately 19% encode protein involved in metabolism, signaling, and enzymatic processes (Fig. 3B). A comparison of the VtlR and AbcR1 regulons sheds light on the dependent and independent regulatory functions each possesses (Fig. 3C). For example, the VtlR regulon was found to be significantly larger than the AbcR1 regulon, with 254 dysregulated genes in ∆vtlR compared to 95 dysregulated genes in ∆abcR1. Approximately half of the genes differentially expressed ≥ 3-fold in abcR1 were also differentially expressed in ∆vtlR. Altogether, these data reveal > 210 genes that could potentially be regulated by VtlR in an AbcR1-independent manner.
Agrobacterium tumefaciens VtlR directly activates the expression of abcR1 and atu1667, a small hypothetical protein. To further characterize the regulatory mechanism of A. tumefaciens VtlR, electrophoretic mobility shift assays (EMSAs) were employed to test for the ability of recombinantly purified VtlR (rVtlR) to interact directly with the promoter regions of putative regulatory targets. As expected, VtlR bound to the promoter region on abcR1 in a concentration-dependent manner (Fig. 4A). Moreover, the addition of unlabeled abcR1 promoter region DNA competitively inhibited binding between rVtlR and the radiolabeled abcR1 promoter, while excess unlabeled abcR2 promoter DNA did not affect the formation of the rVtlR-P abcR1 binding complex. Overall, these data demonstrate that VtlR binds directly to the abcR1 promoter region to activate abcR1 expression in A. tumefaciens.
In A. tumefaciens, the gene atu1667 encodes for a small hypothetical protein that is orthologous to BAB1_0914 and BAB2_0512 in B. abortus. bab1_0914 and bab2_0512 are directly transcriptionally activated by VtlR in B. Electrophoretic mobility shift assays (EMSAs) were carried out with recombinantly purified A. tumefaciens VtlR (rVtlR) and 32 P-radiolabeled abcR1 promoter region (P abcR1 ). Increasing concentrations of rVtlR were added to P abcR1 , and the binding reactions were incubated at room temperature for 30 min. To determine specificity of binding, specific (unlabeled P abcR1 ) and non-specific (unlabeled P abcR2 ) competitors were added to some binding reactions. (B) Northern blot analysis confirming VtlR activation of atu1667. RNA from A. tumefaciens C58 and ∆vtlR was isolated from cultures grown in LB broth to OD 600 0.6, separated on a denaturing polyacrylamide gel, transferred to a nitrocellulose membrane, and probed with radiolabeled oligonucleotides. 5S ribosomal RNA was used as a loading control. (C) EMSAs were performed with rVtlR and 32 P-radiolabeled atu1667 promoter region (P atu1667 ). To determine specificity of binding, specific (unlabeled P atu1667 ) and non-specific (unlabeled P abcR2 ) competitors were added to some binding reactions. (D) Nucleotide alignment of the promoter regions of abcR1 and atu1667. A consensus sequence is underlined in grey, with 100% identity of nucleotides highlighted in black.
Scientific RepoRtS | (2020) 10:14968 | https://doi.org/10.1038/s41598-020-72117-0 www.nature.com/scientificreports/ abortus, and thus, it was hypothesized that VtlR also directly activates the expression of atu1667 in A. tumefaciens 22 . To test this hypothesis, northern blot analysis was used to assess Atu1667 RNA levels in the A. tumefaciens vtlR deletion strain (Fig. 4B). Northern blot analysis and RNA-seq analysis (> 8-fold differentially expressed in ∆vtlR) showed that Atu1667 RNA levels were significantly decreased in A. tumefaciens ∆vtlR compared to the parental strain, suggesting that VtlR activates expression of atu1667. To assess potential binding between VtlR and the atu1667 promoter region, EMSAs were performed with a radiolabeled promoter region of atu1667 and rVtlR (Fig. 4C). These experiments determined that rVtlR binds directly and specifically to the promoter of atu1667, indicating that VtlR is a direct transcriptional activator of atu1667 in A. tumefaciens.
Identification of a VtlR-binding consensus sequence and discovery of a novel VtlR-regulated sRnA. A bioinformatic approach was employed to align the upstream regions of abcR1 and atu1667 and determine sequence similarities between the two promoters to identify a binding motif. Alignment of the two promoters led to the identification of a putative VtlR-binding sequence composed of 15 DNA base pairs (Fig. 4D). This putative VtlR consensus sequence was utilized to bioinformatically search for other potential VtlR binding sites in the A. tumefaciens genome. Putative VtlR sites were identified through the use of the online service Virtual Footprint 34 . Surprisingly, only one area displayed a match to the abcR1/atu1667 VtlR-binding consensus sequence using this approach, and the identified area corresponds to an intergenic region flanked by atu4669 and atu4670 (Fig. 5A,B). Subsequently, an EMSA was performed with rVtlR and DNA that encompasses the atu4669 and atu4670 intergenic region (Fig. 5C). rVtlR bound directly and specifically to this intergenic region DNA. Importantly, the promoter regions of atu4669 and atu4670 are not in this area of the chromosome; in contrast, this intergenic region is located between the 3′ ends of both atu4669 and atu4670 (Fig. 5B). Therefore, if VtlR is binding to a promoter region of a gene it regulates, then there may be a previously unannotated gene transcribed antisense to either atu4669 or atu4670. To determine if an unannotated gene in encoded in this region, northern blot analysis was carried out with RNA isolated from C58 and ∆vtlR. Northern blot analysis revealed the presence of a ~ 300 nucleotide transcript antisense to atu4670 (Fig. 5D). This small RNA was first identified as "L4" by Wilms et al. during a screen for previously unidentified sRNAs in the Agrobacterium tumefaciens genome 35 . The expression of this transcript was abolished in ∆vtlR. The evidence of direct binding of rVtlR to this DNA  34 . Nucleotide alignment of the promoter regions of abcR1, atu1667, and vrsA revealed a consensus sequence underlined in grey, with 100% identity of nucleotides highlighted in black. (B) The new VtlR-binding site is located on the linear chromosome in the intergenic region of two genes encoding for hypothetical proteins (atu4669 and atu4670). The VtlR-binding box is depicted by a light grey box, and the putative transcript is depicted by the black arrow antisense to atu4670. (C) EMSAs were carried out with rVtlR and 32 P-radiolabeled vrsA promoter region. To determine specificity of binding, specific (unlabeled P vrsA ) and non-specific (unlabeled P abcR2 ) competitors were added to some binding reactions. (D) Northern blot analysis was employed to determine if VtlR regulates the putative transcript. RNA from A. tumefaciens C58 and ∆vtlR was isolated from cultures grown in LB broth to OD 600 0.6, separated on a denaturing polyacrylamide gel, transferred to a nitrocellulose membrane, and probed with radiolabeled oligonucleotides. 5S ribosomal RNA was used as a loading control.  Table S5). Under the conditions tested, RNA was isolated from A. tumefaciens grown in LB broth to late exponential phase, only 7 of these sRNAs could be visualized via northern blot analysis and 2 were differentially expressed in A. tumefaciens ∆vtlR with restored expression in the complemented strain ∆vtlR::pSRK-Km-vtlR; one of which was the aforementioned VrsA ( Supplementary  Fig. S2). The second dysregulated sRNA in ∆vtlR is located within the intergenic region between atu0985 and atu0986 and is approximately 225 nucleotides in length; however, the proposed VtlR binding box is not found upstream of this new sRNA.
VrsA does not contribute to the growth kinetics, tumor or biofilm formation, or motility of A. tumefaciens and 17 genes are differentially expressed in ∆vrsA. Agrobacterium tumefaciens ∆abcR1 did not exhibit similar phenotypes observed in ∆vtlR and RNA-seq analysis showed that only a portion (~ 15%) of the genes dysregulated in ∆vtlR are potentially regulated in an AbcR1-dependent manner. With the discovery that VtlR also regulates the expression of vrsA, it is plausible that the > 210 AbcR1-independently regulated genes in ∆vtlR are regulated in a VrsA-dependent manner. To examine this hypothesis, an unmarked in-frame deletion of atu4670, the cis-encoded gene to vrsA, was constructed and utilized to perform phenotypic and RNA-seq analysis. Phenotypic analyses were performed to assess growth, tumor and biofilm formation, and motility of A. tumefaciens ∆vrsA similarly to those performed in Figs. 1 and 2. A. tumefaciens ∆vrsA growth in nutrient rich medium, tumor and biofilm formation, and motility were not significantly different when compared to the parental strain A. tumefaciens str. C58 (Supplementary Fig. S3).
RNA was isolated from cultures of A. tumefaciens str. C58 and A. tumefaciens ∆vrsA grown to late exponential phase in LB broth and analyzed via RNA-seq analysis similarly to the transcriptomics performed in Supplementary Tables S1-S3. The data revealed 17 genes differentially expressed ≥ 3-fold in ∆vrsA compared to the parental strain C58 (Supplementary Table S6). Genes dysregulated in ∆vrsA are organized in 3 loci on the A. tumefaciens genome. One locus including genes encoding a putative transport system, atu5126-atu5130, was downregulated in the ∆vrsA RNA-seq dataset. Two loci, one including genes encoding a putative transport system and the other including genes involved in denitrification, were upregulated in ∆vrsA RNA-seq dataset.
It was possible that differences observed in gene expression were actually due to the loss of atu4670 expression, rather than deletion of vrsA, because these genes are overlapping on the chromosome. To test this, we generated a strain harboring a mutated VtlR-binding box in the vrsA promoter that abolishes vrsA expression while not affecting the coding region of atu4670 ( Supplementary Fig. S4). Northern blot analyses revealed that the vrsA-SD strain produces no VrsA sRNA, and moreover, qRT-PCR analyses demonstrated that dysregulation of att, agl, and nor genes is similar between the ∆vrsA and vrsA-SD strains, indicating that disruption of atu4670 is not responsible for the observed gene expression differences identified in the RNA-seq experiments with the ∆vrsA strain.
Heterologous complementation of A. tumefaciens ∆vtlR with S. meliloti lsrB or B. abortus vtlR. VtlR/LsrB proteins are highly conserved in many members of the Rhizobiales, and thus, it was hypothesized that A. tumefaciens ∆vtlR could be functionally complemented with a heterologous allele. The A. tumefaciens VtlR and S. meliloti LsrB, proteins share over 88% identity in amino acid sequence, whereas A. tumefaciens VtlR and B. abortus VtlR share 68% amino acid sequence identity. In S. meliloti, deletion of lsrB results in a severe growth defect in rich medium 18 . In contrast, there was no difference in the growth kinetics of B. abortus when vtlR was deleted 22 . To test our hypothesis, vtlR from A. tumefaciens, vtlR from B. abortus, and lsrB from S. meliloti were individually cloned into the expression vector pSRK-Km to complement the A. tumefaciens vtlR deletion strain, and the strains were assessed for growth in vitro. When plated on AT agar, wild-type colony size was restored with complementation of all three vtlR genes (Figs. 1D, 6A). This revealed the ability of vtlR from A. tumefaciens, vtlR from B. abortus, and lsrB from S. meliloti to heterologously complement the small colony phenotype of A. tumefaciens vtlR.
To assess the ability of VtlR orthologs to complement the A. tumefaciens vtlR growth defect in liquid medium, growth curves over time were conducted in rich medium supplemented with 100 μM IPTG and 45 μg mL −1 kanamycin (Fig. 6B). As a control, the growth defect displayed by A. tumefaciens ∆vtlR was not complemented by the presence of an empty pSRK-Km plasmid. A. tumefaciens ∆vtlR harboring the lsrB allele from S. meliloti or vtlR from A. tumefaciens restored wild-type growth kinetics (Fig. 6B). However, vtlR from B. abortus only partially restored growth of the A. tumefaciens ∆vtlR, revealing a potential divergence in function of B. abortus VtlR from other Rhizobiales.

Discussion
In the present study, we have characterized a conserved LTTR named VtlR in the plant pathogen A. tumefaciens by defining its role in biological processes (i.e., tumorigenesis and biofilm formation) and genetic regulation (Fig. 1, 2, 3 (Fig. 6). This work also further characterizes the regulatory capabilities of the AbcR sRNAs in Agrobacterium pathogenesis ( Fig. 3; Supplementary Tables S1, S2), as well as a novel sRNA called VrsA (Supplementary Fig. S3; Supplementary Table S6). The requirement of vtlR in A. tumefaciens pathogenesis resembles what was previously reported in two other Rhizobiales, B. abortus and S. meliloti ( Fig. 2A) 18,22 . Moreover, our data support the observation that VtlR is important for host-bacterium interactions of A. tumefaciens with plants 32 . However, the dispensability of abcR1 and abcR2 in A. tumefaciens for efficient host-bacterium interactions and for proper growth of the bacteria differs from S. meliloti and B. abortus. In S. meliloti, deletion of abcR1 or abcR2 does not affect symbiosis, but a deletion of abcR1 causes an acute growth defect in nutrient-rich medium 27 . In B. abortus, AbcR1 and AbcR2 are functionally redundant, and despite displaying no differences in bacterial growth in nutrient-rich or nutrientlimiting conditions, a strain containing a deletion of both abcR1 and abcR2 in B. abortus is less able to colonize and survive in macrophages and experimentally infected mice 24 . Conversely, AbcR1 and AbcR2 in A. tumefaciens are not functionally redundant. In A. tumefaciens, AbcR1 contains two RNA-binding motifs, named M1 (for motif 1) and M2 (for motif 2), and these motifs are utilized by AbcR1 to interact with target mRNAs 25 . The reason AbcR2 lacks robust regulatory functionality may have to do with the absence of M1 from its nucleotide sequence 31 . Altogether, data from this study supports previous work, where RNA-seq analysis found a deletion of abcR1, not abcR2, to lead to significant gene dysregulation ( Fig. 3; Supplementary Tables S1, S2) 25 .
Although deletion of abcR1 did not result in any phenotypic differences with regards to growth in vitro, virulence, or biofilm formation, we sought to further analyze the AbcR1 regulon ( Fig. 3; Supplementary Table S1). RNA-seq analyses uncovered 95 differentially expressed genes (≥ 3-fold) in the abcR1 deletion strain compared to the parental strain C58. Previously, Overlöper and colleagues utilized proteomic and bioinformatics analyses to characterize 16 targets of AbcR1, the majority of which are components of ABC-type transport systems (e.g., chvE, malE, atu2422, atu4678, and atu1879) 25 . Indeed, our transcriptomic data largely resembles the proteomic data from that study (Supplementary Table S1). However, aside from the ABC-type transport systems, RNAseq revealed additional AbcR1 targets, such as genes encoding transcriptional regulators, chemoreceptors, and a variety of enzymes. Remarkably, several genes previously shown to be necessary for A. tumefaciens virulence,  Table S1) 36,37 . Yet, a deletion of abcR1 does not affect the ability of A. tumefaciens to form tumors in experimentally infected potatoes (Data not shown). A possible explanation for this may reside in the expression of these virulence-associated targets in ∆abcR1. While these target mRNAs exhibit decreased, but not completely abrogated, expression in ∆abcR1, it is possible that even the low level of expression of virulence-associated targets in ∆abcR1 is sufficient to sustain infection, thus resulting in a lack of attenuation. Further investigation is needed to fully define the direct regulation and sRNA-mRNA interactions between AbcR1 and the newly identified targets in A. tumefaciens.
In Agrobacterium, VtlR is the transcriptional activator of abcR1, and directly binds to the promoter region of abcR1 to exert its regulatory function (Figs. 1, 4). Following identification of gene dysregulation in both A. tumefaciens ∆abcR1 and ∆vtlR, we sought to compare the transcriptomic profiles of the two deletion strains (Fig. 3). Importantly, several genetic systems differentially expressed in ∆vtlR showed no difference in abcR1, suggesting these systems could be key for Agrobacterium pathogenesis. Of note, a type IV secretion system (avhB), a conjugation system (tra), genes necessary for polysaccharide biosynthesis (exo), as well as 10 transcriptional regulators are all differentially expressed in ∆vtlR and show no difference in expression in ∆abcR1. However, none of these genes has been shown to contribute to or is associated with the virulence of A. tumefaciens 38 . With regards to biofilm production, one gene, divK, may contribute to differences in biofilm production observed in ∆vtlR (Fig. 2B). A deletion of divK has been reported to result in decreased biofilm production in Agrobacterium 39 . In A. tumefaciens ∆vtlR, divK is over-expressed, suggesting the possibility of divK dysregulation could alter biofilm formation. While no other genes from the ∆vtlR RNA-seq analysis have been linked directly to biofilm formation, several genes have been speculated to be necessary for biofilm production (e.g., glcF and gguB) 40 .
The difference in size of the VtlR regulons in Brucella and Agrobacterium is striking. Indeed, the B. abortus VtlR regulon is comprised of 10 genes, while the A. tumefaciens VtlR regulon is comprised of > 200 genes (Supplementary Table S3) 22 . The complete regulon of LsrB in S. meliloti is currently unknown, although several genes involved in LPS synthesis, glutathione synthesis, and oxidative stress are regulated by LsrB 19,20 . Since S. meliloti LsrB was shown to directly bind to the promoter region of the lrp3-lpsCDE operon, it was hypothesized that the homologous lrp3-lpsCDE system in A. tumefaciens may be similarly regulated 20 . However, EMSAs showed no binding of A. tumefaciens rVtlR to the promoter region of lrp3-lpsCDE ( Supplementary Fig. S5). Furthermore, the VtlR/LsrB systems have diverged in their regulation of the abcR sRNAs between organisms, as S. meliloti LsrB does not regulate the abcR sRNAs under the conditions tested ( Supplementary Fig. S1). Regarding glutathione production, Tang et al. demonstrated that the genes gshA and gshB are regulated by OxyR 20 . OxyR, a LTTR responsible for aiding the cell in responding to oxidative stress, is transcriptionally activated by LsrB in S. meliloti 20 . Similar to S. meliloti, A. tumefaciens OxyR is necessary for protection against oxidative stress and important for host-microbe interactions 41,42 . In contrast to S. meliloti, oxyR does not appear to be regulated by VtlR in A. tumefaciens, as no difference in oxyR expression in A. tumefaciens ∆vtlR was found (Supplementary  Table S3). Overall, these findings suggest a divergence of function between S. meliloti LsrB and A. tumefaciens VtlR.
As mentioned in the introduction, the N-terminal domain LysR-type transcriptional regulators (LTTRs) most commonly include a helix-turn-helix domain, and regulated targets have a 'classic' LTTR-binding box sequence of TTA-N 7/8 -TAA, in an A/T rich region 1 . Thus, the proposed VtlR binding box confirmed in B. abortus and hypothesized in A. tumefaciens of GCAT-N 3 -TG-N 3 -T is divergent from typical LTTRs and may be unique to this regulator among the Rhizobiales.
Identification of a putative VtlR binding box, Fig. 4D, and genome-wide search of this sequence in A. tumefaciens identified a region upstream of a novel sRNA previously described as "L4" but has been renamed VrsA (VtlR-regulated sRNA) 35 . Our data shows that VrsA is directly activated by VtlR (Fig. 5). Phenotypic and transcriptional analyses described in this study revealed that a deletion of vrsA does not contribute to the growth, tumor or biofilm formation, or motility of A. tumefaciens (Supplementary Fig. S3). RNA-seq analysis revealed that 17 genes were differentially expressed in A. tumefaciens ∆vrsA grown in nutrient rich broth (Supplementary Table S6). The att operon (i.e., atu5126-atu5130) encoding a transport and attachment system was downregulated in the ∆vrsA strain, indicating that VrsA activates the expression of these genes 37 . This supports the VtlR regulatory model because att also had lowered expression in ∆vtlR. Thus, VtlR activates the expression of vrsA, which in turn activates the expression of att. The expression of two loci were increased in ∆vrsA, and these genes include a putative transport system, atu0591-atu0593, and the nor operon, which encodes a nitric oxide reductase important for denitrification. Overall, further mechanistic analyses are necessary to understand the regulatory role of the VrsA sRNA on these targets. Two other small RNA encoding genes were differentially expressed in ∆vtlR. atu1667 encodes a putative hypothetical protein that has not yet been characterized in Agrobacterium, but BAB2_0512, a homolog of Atu1667, encodes a small protein that is linked to the ability of B. abortus to utilize the sugar fucose 43 . It is unknown whether atu1667 also encodes a small protein or if it potentially functions as a small regulatory RNA. Northern blot analysis also revealed the presence of a small RNA in the intergenic region of atu0985 and atu0986 that is activated by VtlR (Supplementary Table S5; Supplementary Fig. S2) 32 . There are several similarities observed between our work and the study by Tang et al., but the studies also contain differences with regards to experimental approach and results observed. One major difference is the Agrobacterium strain utilized. The work presented here employed A. tumefaciens str. C58, which is a naturally occurring strain of A. tumefaciens, whereas Tang et al. utilized a strain named A. tumefaciens C58C1 32 . A. tumefaciens C58C1 is a modified A. tumefaciens strain in which the original virulence plasmid pTiC58 is replaced by the A. rhizogenes virulence plasmid, pRiA4b 44 . While the C58C1 strain is still able to form host-microbe interactions and cause disease in its plant host, this absence may affect the function of VtlR in A. tumefaciens.
While there are many similarities in gene expression when comparing the present RNA-seq dataset to the previously published ∆vtlR dataset, there are some significant differences (Supplementary Table S3) 32 . It should be noted that between the studies, A. tumefaciens strains were grown in different medium. In the presented study, RNA was isolated from A. tumefaciens grown in LB broth, whereas the previous study isolated RNA from bacteria grown in TY broth (Supplementary Table S3) 32 . The largest difference between the VtlR regulons from these two studies is the differential gene expression observed with regards to AbcR1. Our data and data previously presented by Overlöper and colleagues show differences in expression and regulation of chvE, atu2422, atu4678, and atu1879 in ∆abcR1 and ∆vtlR (Supplementary Tables S1, S3) 25 . None of these genes appear to be differentially regulated in the VtlR regulon described previously 32 . Altogether, these differences may be explained by the variation in experimental design, particularly in regards to the conditions under which the bacteria were cultured before RNA isolation.
In conclusion, VtlR is necessary for efficient in vitro growth and tumorigenesis of the plant pathogen A. tumefaciens. The A. tumefaciens VtlR protein binds directly to a conserved binding box in three promoter regions, but VtlR is linked to the regulation of over 250 genes. Taken together, we propose a model of regulation by VtlR in A. tumefaciens in which VtlR plays a central role in the activation of sRNAs that, in turn, control the expression of a wide variety of mRNA targets (Fig. 7). In this model, VtlR activates the expression of AbcR1, the newly described sRNA, VrsA, and the putative sRNA Atu1667, and these sRNAs are the major regulatory elements responsible for propagating the VtlR genetic circuit. Overall, this study demonstrates the functional importance of VtlR in A. tumefaciens, and provides insight into the evolutionary similarities and differences that exist in the VtlR/LsrB systems of members of the Rhizobiales.

Methods
Bacterial strains and growth conditions. Agrobacterium tumefaciens str. C58 and derivative strains were routinely grown on Luria-Bertani (LB) agar (Fisher Scientific Company LLC, Suwanee, GA) and cultures were routinely grown in LB broth. LB consists of 10 g NaCl, 10 g Bacto™-tryptone, and 5 g yeast extract per liter. , OD 600 = 3 and OD 600 = 4), the bacteria were cultured in LB broth for extended time periods (e.g., greater than 24 h), and tenfold dilutions of the culture were used to assess the optical density at 600 nm in the linear range of 0-1. For cloning, Escherichia coli strain DH5α was grown on tryptic soy agar (BD) or in Luria-Bertani (LB) broth. When appropriate, growth media were supplemented with kanamycin (45 μg mL −1 for E. coli and 300 μg mL −1 for A. tumefaciens), sucrose (5%), or IPTG (100 μM).
construction and complementation of A. tumefaciens deletion strains. Mutagenesis of A. tumefaciens vtlR, abcR1, abcR2, and vrsA. The vtlR gene (atu2186) was mutated utilizing an unmarked gene excision strategy previously described 22,26 . An approximately 1-kb fragment of the upstream region of atu2186 was amplified via PCR with A. tumefaciens genomic DNA, primers atu2186-Up-For and atu2186-Up-Rev, and Taq polymerase (Monserate Biotechnology Group). Similarly, an approximately 1-kb fragment of the downstream region of atu2186 was amplified with primers atu2186-Dn-For and atu2186-Dn-Rev. All oligonucleotides are listed in Supplementary Table S7. These fragments were then digested with the appropriate restriction enzymes, and subsequently phosphorylated with polynucleotide kinase (Monserate Biotechnology Group). Fragments were then combined in a single ligation with digested pNPTS138 and T4 DNA ligase (Monserate Biotechnology Group) 45 . The resulting plasmid was introduced into A. tumefaciens C58 by electroporation. Following electroporation of this plasmid into C58 primary integration of this non-replicating plasmid was confirmed using primer pairs USP003/atu2186-con-For and USP003/atu2186-con-Rev. Counter-selection on 5% sucrose resulted in excision of the integrated plasmid. Deletion of the atu2186 (vtlR) locus was confirmed using primer pair atu2186-con-For/atu2186-con-Rev. All plasmid constructs are listed in Supplementary Table S8. This method was utilized to construct the abcR1, abcR2, and abcR1/2 deletion constructs and strains.  Table S1. The fragment was digested with the appropriate restriction enzymes, and subsequently ligated into digested pSRK-Km. The resulting plasmid was then introduced into the A. tumefaciens vtlR strain by electroporation. The strain harboring the complementation plasmid was selected for on AT-agar plates supplemented with kanamycin (45 μg mL −1 ). All plasmid constructs are listed in Supplementary Table S2.
In-trans heterologous complementation of A. tumefaciens vtlR. Heterologous complementation was carried out as described above. For complementation with lsrB from Sinorhizobium meliloti, the lsrB gene (SMc01226) was amplified via PCR with S. meliloti 1021 genomic DNA, primers lsrB-comp-For and lsrB-comp-Rev and Taq polymerase. All oligonucleotides are listed in Supplementary Table S7. The fragment was digested with the appropriate restriction enzymes, and ligated into digested pSRK-Km. The resulting plasmid was then introduced into the A. tumefaciens vtlR deletion strain by electroporation. All plasmid constructs are listed in Supplementary  Table S8. For complementation with vtlR from Brucella abortus, the vtlR gene (bab1_1517) was amplified via PCR with B. abortus 2308 genomic DNA, primers vtlR-comp-For and vtlR-comp-Rev and Taq polymerase. All oligonucleotides are listed in Supplementary Table S7. The fragment was digested with the appropriate restriction enzymes, and ligated into digested pSRK-Km. The resulting plasmid was then introduced into the A. tumefaciens vtlR strain by electroporation. All plasmid constructs are listed in Supplementary Table S8.
Site-directed mutagenesis of VtlR binding motif in vrsA promoter. Phusion High-Fidelity DNA polymerase (New England BioLabs) was used to amplify a 1 kbp fragment from 554 nt upstream to 179 bp downstream of the vrsA coding sequence, using purified A. tumefaciens str. C58 genomic DNA primers USP204 and USP205. The 1 kbp amplicon was gel purified using the E.Z.N.A. gel extraction kit (Omega Bio-tek), A-tailed with Taq DNA polymerase (New England BioLabs), and ligated into vector pGEM-T Easy (Promega) to generate plasmid pJEH158. Plasmid pJEH158 was transformed into E. coli strain DH10B for propagation. Amplicon sequence was confirmed using Sanger sequencing (Genewiz).
For site-directed mutagenesis, 10 ng of plasmid pJEH158 served as template with 125 ng of each primer, USP206 and USP207. Phusion High-Fidelity DNA polymerase was used with the following cycling parameters. Initial denaturation: 95 °C, 30 s. Cyclic amplification: 95 °C, 30 s/55 °C, 1 min/68 °C, 4 min, for a total of 18 cycles in a volume of 50 μL. Following column purification of reaction products with E.Z.N.A. cycle pure kit (Omega Bio-tek) to 40 μL, 1 μL of DpnI and was added and the mixture incubated for 2 h at 37 °C. 5 μL of this reaction was then used to transform chemically competent E. coli strain DH10B. From a pool of transformants, multiple candidate mutagenized plasmids were purified and subjected to Sanger sequencing. Multiple mutant plasmids were identified and one, pJEH159, was retained for further use.
The entire 1-kb fragment originally cloned into pGEM-T Easy was sub-cloned from pJEH159 into the suicide plasmid pNPTS138 using engineered restriction sites for SpeI and SphI. The product of this sub-cloning, Scientific RepoRtS | (2020) 10:14968 | https://doi.org/10.1038/s41598-020-72117-0 www.nature.com/scientificreports/ pJEH060, was verified by Sanger sequencing and transformed into the conjugative mating strain of E. coli, S17-1 λpir. Allelic replacement of the 1-kb region encompassing vrsA and its mutated promoter was performed using the same steps as for generating the ∆vrsA mutant strain of A. tumefaciens C58, generating strain C58-JEH169. A 2,027 bp amplicon from this strain generated using Phusion High-Fidelity DNA polymerase and primers USP152 and USP153 was used for sequence confirmation. potato tumor assay. To test the virulence of the constructs, tumor formation on disks of red potato was measured. Organic, red potatoes were scrubbed to remove dirt and debris, sterilized in dilute bleach for 20 min, and finally sterilized with UV light for no less than 20 min. Potato disks were created by coring the potatoes and cutting the cores into 0.5-cm-wide slices. The slices were placed onto an agar plate with no added nutrients, and each plate had five technical replicates. The disks were inoculated with the indicated strains, which were grown overnight in ATGN. The overnight cultures were diluted to an optical density (600 nm) of 0.06 prior to inoculation, then ten microliters of the strain was placed on each potato disk. The plates were sealed with parafilm and left undisturbed at room temperature for 4 weeks. The tumor formation was counted at day 14 and 21.

Static biofilm assay.
To test the ability of a strain to form a biofilm, a static biofilm assay was performed.
Overnight cultures of each strain were grown in LB. The following morning each culture was subcultured to an optical density (600 nm) of 0.1, and once the cultures reach exponential growth, they were diluted to an optical density (600 nm) of 0.05. Three milliliters of each culture was placed in a nine well polystyrene plate. Previously, polyvinyl chloride coverslips had been placed in each well, and then the plates were UV sterilized for no less than 20 min. After inoculation, the plates were incubated at room temperature for 48 h. The coverslips were rinsed to remove excess or weakly attached organisms, then stained with 0.1% crystal violet. After staining, the excess crystal violet was rinsed off and the adherent crystal violet was re-solubilized in 33% acetic acid. Three hundred microliters of re-solubilized crystal violet and culture were loaded into a 96 well plate. The absorbance of the crystal violet solution was measured at 600 nm (A 600nm ), and the optical density of the culture at 600 nm was measured (OD 600nm ). For data presentation biofilm formation is normalized to growth using the formula A 600nm / OD 600nm and expressed relative to biofilm formation by the wild-type background. northern blot analysis. Agrobacterium tumefaciens RNA was isolated from cultures using the methodology previously described 22,26 . Ten micrograms of RNA were separated on a denaturing 10% polyacrylamide gel with 7 M urea and 1 × TBE (89 mM Tris base, 89 mM boric acid, 2 mM EDTA). To determine size, a low molecular weight DNA ladder (New England Biolabs) was labelled with [γ-32 P]ATP (PerkinElmer) and polynucleotide kinase (Monserate Biotechnology Group). Following electrophoresis, the ladder and RNA samples were transferred to a Amersham Hybond™-N + membrane (GE Heathcare) by electroblotting in 1 × TBE buffer. Samples were then UV cross-linked to the membrane, and membranes were then pre-hybridized in ULTRAhyb ® -Oligo Buffer (Ambion) for 1 h at 45 °C in a rotating hybridization oven. Oligonucleotide probes were end-labelled with [γ-32 P]ATP and polynucleotide kinase. All oligonucleotides are listed in Supplementary Table S7. Radiolabeled probes were incubated with pre-hybridized membranes at 45 °C in a rotating hybridization oven overnight. The following day, membranes were washed four times with 2 × SSC (300 mM sodium chloride and 30 mM sodium citrate), 1 × SSC, 0.5 × SSC, and 0.25 × SSC at 45 °C in a rotating hybridization oven for 30 min each. Each SSC washing solution contained 0.1% sodium dodecyl sulphate (SDS). Membranes were exposed to X-ray film and visualized by autoradiography.
Protein purification. Recombinant A. tumefaciens VtlR (rVtlR) was constructed utilizing the Strep-tag II system (IBA), and subsequently cloned and expressed in E. coli BL21 cells. The coding region of atu2186 was amplified via PCR using A. tumefaciens C58 genomic DNA as a template, primers rAtu2186-For and rAtu2186-Rev and Taq polymerase (Monserate Biotechnology Group). All oligonucleotides are listed in Supplementary  Table S7. The DNA was then digested with BsaI and ligated into pASK-IBA7, which encodes an amino-terminal Strep-tag II on the protein of interest. Following sequencing of the plasmid, the E. coli BL21 strain harboring rVtlR-pIBA7 was grown to an OD 600 nm of 0.7 before recombinant gene expression was induced by 200 μg mL −1 anhydrotetracycline (AHT). Following 3 h of constant shaking at 37 °C, cultures were collected by centrifugation (4,200×g for 10 min at 4 °C) and lysed with CelLytic B (Sigma) in the presence of the protease inhibitor phenylmethanesulfonylfluoride (PMSF). The supernatant from the lysed cells was cleared by centrifugation (14,000×g for 10 min at 4 °C) and passed through a Strep-Tactin Sepharose affinity column. The column was then washed two times with Buffer W (100 mM Tris-HCl, 300 mM NaCl, pH 8.0) and the rVtlR was eluted with 2.5 mM desthiobiotin in Buffer W. The degree of purity of the rVtlR was high as judged by visualization of a single band on SDS-PAGE.
Electrophoretic mobility shift assays (EMSAs). EMSAs with rVtlR were carried out as previously described 22 . All EMSAs were done in a final volume of 20 μL reaction mixture that included a binding buffer composed of 10 mM Tris-HCl (pH 7.4), 50 mM KCl, 1 mM dithiothreitol, 6% glycerol, 50 μg mL −1 bovine serum albumin and 50 μg mL −1 salmon sperm DNA. DNA fragments of the abcR1, abcR2, atu1667, and vrsA promoter regions were amplified by PCR using A. tumefaciens C58 genomic DNA at a template, gene-specific primers and Taq polymerase (Monserate Biotechnology Group). All oligonucleotides are listed in Supplementary Table S7. Fragments were then run on a 0.8% agarose gel, purified and end-labelled with [γ-32 P]ATP (PerkinElmer) and polynucleotide kinase (Monserate Biotechnology Group). Increasing amounts of rVtlR were added to DNA fragments in binding buffer, and subsequently incubated at room temperature for 30 min. In some gels, non-Scientific RepoRtS | (2020) 10:14968 | https://doi.org/10.1038/s41598-020-72117-0 www.nature.com/scientificreports/ radiolabeled specific DNA (i.e. promoters of abcR1 and vrsA) or non-radiolabeled non-specific DNA (i.e. promoter of abcR2) were added to reactions in 50 × molar concentrations. Binding reactions were run on 6% native polyacrylamide gels in 0.5 × TBE running buffer for 1 h. Gels were dried onto 3 mm Whatman paper using a vacuum gel drier system and visualized by autoradiography.

RNA-sequencing. RNA extraction and precipitation.
RNA extractions were carried out as previously described 22,24,26 . Agrobacterium strains were grown in triplicate to an OD 600 nm of 1.0 with constant shaking at 28 °C. An equal amount of 1:1 ethanol-acetone was added to cultures and stored at − 80 °C. For RNA isolation, the cell/ethanol-acetone mixtures were thawed and pelleted at 16,000×g for 3 min. RNA was isolated from cells by use of TRIzol reagents (Invitrogen) followed by ethanol precipitation. Following RNA isolation, genomic DNA was removed with DNase I (2 U; Thermo Fisher Scientific), where 30 μg of RNA was incubated with DNase I for 1 h at 37 °C. Samples were then cleaned up by phenol-chloroform extractions and subsequent ethanol precipitation. RNA samples were resuspended in nuclease-free H 2 O and purity of each sample was checked with a NanoDrop 1,000 spectrophotometer (Thermo Fisher Scientific). All samples had an A 260 /A 280 ratio of ~ 2.0 and a concentration yield of ~ 1 μg μL −1 . RNA samples (10 μg total) were then submitted to the Bioinformatics Institute at Virginia Tech for RNA-seq analysis.
Stranded RNA library construction for prokaryotic RNA-Seq. The RNA-sequencing libraries were prepared as described previously by our group 43 . Briefly, 1 μg of total RNA with RIN ≥ 8.0 was depleted of rRNA using Illumina's Ribo-Zero rRNA Removal Kit (Gram-Positive and Gram-Negative Bacteria) (P/N MRZB12424, Illumina, CA, USA). The depleted RNA is fragmented and converted to first strand cDNA using reverse transcriptase and random primers using Illumina's TruSeq Stranded mRNA HT Sample Prep Kit (Illumina, RS-122-2103). This is followed by second strand synthesis using polymerase I and RNAse H, and dNTPs that contain dUTP instead of dTTP. The cDNA fragments then go through end repair, addition of a single ' A' base, and then ligation of adapters and indexed individually. The products are then purified and the second strand digested with N-Glycosylase, thus resulting in stranded template. The template molecules with the adapters are enriched by 10 cycles of PCR to create the final cDNA library. The library generated is validated using Agilent 2100 Bioanalyzer and quantitated using Quant-iT dsDNA HS Kit (Invitrogen) and qPCR. 16 individually indexed cDNA libraries were pooled and sequenced on Illumina NextSeq to get a minimum of 25 million reads.
Illumina NextSeq sequencing. The Illumina NextSeq sequencing was carried out as described previously by our group 43  RNA-Seq data processing and analysis. The Agrobacterium tumefaciens (strain C58) gene and genome sequences, as well as corresponding annotations from NCBI (https ://www.ncbi.nlm.nih.gov/) were used as a reference. Raw reads were quality-controlled and filtered with FastqMcf 46 , resulting in an average of 1,936 Mbp (1,703-2,329 Mbp) nucleotides. The remaining reads were mapped to the gene reference using BWA with default parameters 47 . Quality control of the sequence data is depicted in Supplementary Fig. S6. Differential expression of genes was calculated using the edgeR package in R software (https ://www.r-proje ct.org/), with Benjamini-Hochberg adjusted P values of 0.05 considered to be significant 48 .