Twenty-four cultivar RH89-039-016 (RH89) potato plants (The Potato Genome Sequencing Consortium, 2011; tissue culture plants originally obtained from Wageningen University) were grown in 6-in. greenhouse pots, and one-half were inoculated with Streptomyces caniscabiei sp. nov. NE06-02D (“Strep”) at 2 × 10⁶ CFU/pot as previously described (Weisberg et al., 2021) for the RNA sequencing experiment. At 4 weeks after planting (wap), six inoculated and six non-inoculated plants were sprayed until runoff with 200 mg L⁻¹ of Weedar 64 2,4-D (Nufarm, Alsip, IL, USA) supplemented with 0.05% Tween 80 (Sigma-Aldrich, St. Louis, MO, USA). The remaining 12 plants (six inoculated and six non-inoculated) were sprayed with a mock treatment of water supplemented with 0.05% Tween 80. At 12 wap (8 weeks after 2,4-D treatments), tubers from all treatment groups were harvested for the collection of tuber periderm tissue for RNA extraction and scoring of disease severity as previously described (Clarke et al., 2020).

For each of the four treatment conditions (“mock,” non-inoculated and mock-treated; “2,4-D,” non-inoculated and 2,4-D-treated; “Strep,” inoculated and mock-treated; “2,4-D + Strep,” inoculated and 2,4-D-treated), three biological replicates of tuber periderm tissue and three biological replicates of leaf tissue from a single experiment were sampled (except as noted in Supplementary Table S1 ). Each biological replicate consisted of samples pooled from two plants, with six total plants used for each treatment group. Samples were stored at −80°C. Leaf samples were taken 2 days post-2,4-D treatment, and tuber samples were taken at harvest (8 weeks post-treatment). For each sample, 100 mg of tissue was ground in liquid nitrogen using a mortar and pestle, then vortexed in 1 mL of TRI Reagent (Sigma-Aldrich). RNA extraction was conducted via RNEasy Mini Kit with on-column RNase-Free DNase Set (Qiagen, Hilden, DE) with modifications. After incubation for 30 min at room temperature (RT), samples were vortexed in 600 µL of Buffer RLT supplemented with 6 µL of β-mercaptoethanol, then incubated at RT for 5 min. Samples were shaken for 1 min with 200 µL of chloroform, then incubated at RT for 10 min, followed by centrifugation at 14,000 rpm for 30 min at 4°C; this process was repeated using 1 mL of the resulting aqueous phase. Six hundred μL of aqueous phase was gently mixed with 300 µL of 100% ethanol and transferred to RNeasy Mini spin columns, with the remaining steps carried out per manufacturer’s instructions.

The RNA sequencing pipeline was conducted by Azenta Life Sciences (South Plainfield, NJ, USA). RNA samples were quantified via Qubit 2.0 Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA), and RNA integrity was verified via 4200 TapeStation (Agilent Technologies, Palo Alto, CA, USA). RNA sequencing libraries were prepared with the NEBNext Ultra II RNA Library Prep Kit for Illumina using the manufacturer’s instructions (New England Biolabs, Ipswich, MA, USA). Sequencing libraries were pooled and clustered on two lanes of a flow cell and loaded onto the Illumina HiSeq 4000 or equivalent (Illumina, San Diego, CA, USA). Samples were sequenced using a 2 × 150-bp paired-end (PE) configuration. Image analysis and base calling were conducted using HiSeq Control Software. Raw sequence data generated from Illumina HiSeq were converted into FASTQ files and de-multiplexed using Illumina’s bcl2fastq 2.17 software. One mismatch was allowed for index sequence identification.

From the resulting paired-end FASTQ files, Illumina adapter and leading 15-bp sequences were trimmed from reads via Trim Galore (https://www.bioinformatics.babraham.ac.uk/projects/trim_galore/) with default parameters and verified via FastQC reports for analysis of FASTQ file quality. For alignment of trimmed reads, genome indexing and mapping to the reference genome was performed via HISAT2 (https://daehwankimlab.github.io/hisat2/) with default options and utilized the DM 1-3–516 R44 (NCBI accession number PRJNA63145) annotated genome v6.1 (https://spuddb.uga.edu/dm_v6_1_download.shtml). Output SAM files were converted to BAM format, and mapping quality was analyzed via Samtools (http://www.htslib.org/). The High Confidence Gene Model Set annotations file for DM v6.1, in GFF3 format, was converted to GTF format via gffread (http://ccb.jhu.edu/software/stringtie/gff.shtml). For the alignment of reads to gene models, featureCounts (http://subread.sourceforge.net/) was employed using the representative, high-confidence gene model GTF and with the following options: -p (paired-end FASTQ files), –countReadpairs (count read pairs), and –transcript (assign reads at transcript-level). For pseudoalignment of trimmed reads, Salmon (https://combine-lab.github.io/salmon/) was employed to index, and assign read pairs to, the representative, high-confidence cDNA sequences for DM v.6.1.

DEG modeling and estimate dispersion analysis and principal component analysis (PCA) of assigned read pairs were performed via the “DESeq2” package (Love et al., 2014) in R Studio with R version 4.1.2. Pairwise comparison groups for DEG analysis were assigned as follows: “response to Strep,” “Strep” vs. “mock”; “response to -Strep_2,4-D,” “2,4-D + Strep” vs. “2,4-D”; “response to 2,4-D_Strep,” “2,4-D + Strep” vs. “Strep”; and “response to 2,4-D,” “2,4-D” vs. “mock” ( Figure 1 ). Gene ontology (GO) term assignments for annotated genes for DM v6.1 were downloaded from http://spuddb.uga.edu. Additional analyses of differentially expressed genes (DEGs) were performed in R Studio with the following packages: pheatmap (DEG FPKM heat map), VennDiagram (DEG Venn diagrams), and clusterProfiler (GO term enrichment). In GO plots, “obsolete” or select, highly redundant GO terms were removed for clarity.

RH89 potato plants were grown as previously described, and at 4 wap, foliage from three plants was sprayed until runoff with each of the following treatments: 200 mg L⁻¹ of 2,4-D, 400 mg L⁻¹ of 2,4-D, or water (“mock”). Each treatment was supplemented with 0.05% Tween 80. At approximately 11 weeks, tubers for each sample (n = 9) were harvested for isolation of approximately 20 g of tuber skins/sample. Lipids were extracted and saponified with 0.5 N of methanolic sodium hydroxide, then methylated with 14% boron trifluoride methanol. Fatty acid methyl esters were extracted via heptane and analyzed for fatty acid content (% by sample mass) via gas chromatography-flame ionization detection (GC-FID). Sample extraction and GC-FID were performed by Eurofins Food Chemistry Testing Madison, Inc. (Madison, WI, USA).

RH89 potatoes were grown in 6-in. greenhouse pots and inoculated with Streptomyces as described above but utilizing strains of either S. acidiscabies ATCC49003 or S. stelliscabiei NY02-3A, separately. Streptomyces stelliscabiei is closely related to S. caniscabiei, which was used in the transcriptomic experiment, and disease from all three species is dependent on the thaxtomin A phytotoxin. 2,4-D treatments were performed as described above but included an additional 400 mg L⁻¹ treatment group. Tubers were harvested 12 weeks after inoculation (8 weeks post-2,4-D treatment). At least two tubers were pooled from separate individual plants for the collection of each of the four biological replicate tuber periderm and tuber medulla samples from each combination of pathogen inoculum and 2,4-D treatment. Tissue samples were ground in water using Agdia (Elkhart, IN) mesh sample extraction bags. Collected samples were then assayed for sugar content using the Megazymes (Sydney, Australia) Sucrose/D-glucose/D-fructose assay kit following the manufacturer’s recommended protocols, except that samples were syringe-filtered through 0.8 µm filters instead of through filter paper.

We first sought to develop a greenhouse assay in which the efficacy of 2,4-D leaf sprays for suppressing common scab disease severity could be observed. As noted in previous field trials, low-dose 2,4-D treatments can induce minor alterations to plant foliage, including very minor leaf curling or early flowering (data not shown), and common scab infections of tubers do not present visible phenotypes in leaf tissue. Upon tuber harvesting, tuber samples are surveyed for common scab disease severity, with severe disease characterized by raised, pitted lesions and low-severity disease associated with superficial lesions. Leaf spray treatment with 200 mg L⁻¹ of 2,4-D of 4-week-old potato plants growing in pots inoculated with pathogenic Streptomyces substantially suppressed common scab severity on tubers harvested 8 weeks following the spray treatment ( Figure 2A ), with tubers from 2,4-D-treated plants exhibiting significantly less severe common scab symptoms ( Figure 2B ). For downstream transcriptomics analyses, to avoid sampling tubers for which Streptomyces infection may not have been well established in soil, tubers with reduced disease symptoms, and not asymptomatic tubers, were selected for 2,4-D-treated, Streptomyces-inoculated plants. Accordingly, leaf samples for 2,4-D-treated, Streptomyces-inoculated plants were selected on the basis of the presence of low-severity disease symptoms observed in their corresponding tubers.

To begin to survey the transcriptome in response to 2,4-D, Streptomyces (“Strep”), and combinations therein, tuber and leaf samples, each corresponding to the four treatment groups ( Figure 1 ), were processed for RNA extraction, total RNA library preparation, and double-stranded sequencing via the Illumina platform. Following quality control trimming of Illumina adapter sequences, the resulting FASTQ files contained an average of 69,419,353 individual reads per sample ( Supplementary Table S1 ). For alignment, HISAT2 (Kim et al., 2019) was utilized to map the reads to the doubled monoploid reference genome DM v6.1, with a mean mapping rate, including both primary and secondary mappings, of 82.46% ( Supplementary Table S1 ). Prior to determining sets of DEGs, we separately evaluated alignment and pseudoalignment methods to assign read pairs to the reference genome. For alignment, featureCounts (Liao et al., 2014) was utilized to align read pairs to the representative gene models for DM v6.1, resulting in a mean of 25,428,076 read pairs per sample, representing an alignment rate of 72.67%, aligned to features ( Supplementary Table S1 ). Comparatively, pseudoalignment via Salmon (Patro et al., 2017) resulted in a pseudoalignment rate of 73.55% ( Supplementary Table S1 ). Due to the marginally higher alignment performance of Salmon, we utilized this data set for identifying DEGs. Separately, pseudoaligned read pairs from tuber (n = 12) and leaf (n = 10) samples were analyzed and normalized for differential expression via DESeq2 (Love et al., 2014) in four pairwise comparisons: “response to Strep,” “response to Strep (2,4-D),” “response to 2,4-D,” and “response to 2,4-D (Strep)” ( Figure 1A ). For tuber samples, PCA conveyed that PC1 and PC2 explained 56% and 30% of total variance, respectively, and that Mock and 2,4-D samples were most closely related; Strep and 2,4-D + Strep samples formed additional, separate clusters, with concordance among biological replicates within each treatment group ( Figure 1B ). Leaf sample PCA resulted in PC1 and PC2 explaining 57% and 19% of total variance, respectively, and illustrated clustering of 2,4-D + Strep and 2,4-D samples separate from the Mock and Strep samples ( Figure 1B ). Generally, large DEG counts were observed for all treatment group comparisons except for tuber “response 2,4-D” and leaf “response to Strep (2,4-D)” ( Supplementary Figure S1 ).

In the tuber sample “response to Strep” comparison, we identified 800 DEGs, consisting of 772 upregulated genes and 28 downregulated genes ( Figure 3A ; Supplementary Table S2 ). Furthermore, response to Strep was also assessed via a comparison of the two groups of 2,4-D treated samples [“response to Strep (2,4-D),” i.e., response to Streptomyces infection in the context of 2,4-D treatment], yielding 1,580 DEGs, consisting of 1,135 and 445 upregulated and downregulated genes, respectively ( Figure 3A ; Supplementary Table S3 ). Overlap analysis of the DEGs from these two comparisons illustrated a specific response to Streptomyces, including 775 upregulated genes and 430 downregulated genes, in the context of foliar 2,4-D treatment, indicative of a possible enhancement to transcriptional sensitivity in tubers by foliar treatment with 2,4-D. To analyze the transcriptomic response to Streptomyces infection in tubers, we incorporated GO assignments for the DEGs identified in “response to Strep” and “response to Strep (2,4-D),” then performed GO enrichment analysis. Among upregulated DEGs in “response to Strep” were significantly enriched GO terms associated with pathogen defense, such as “response to wounding,” “cell wall modification,” “defense response to fungus,” “response to oxidative stress,” and the cellular component term “extracellular region,” with underlying DEGs corresponding to products such as pathogenesis-related (PR) proteins, chitinases, peroxidases, and jasmonate (JA) and ethylene (ET) pathway proteins ( Figure 3B ; Supplementary Tables S2 , S4 ). Furthermore, several enriched GO terms were associated with fatty acid biosynthesis, with underlying genes corresponding to lipases and fatty acid desaturases. Downregulated DEGs contributed to five enriched GO terms, largely corresponding to underlying dehydration response element-binding (DREB) proteins. Analysis of upregulated DEGs in “response to Strep (2,4-D)” identified enriched GO terms largely overlapping with those from “response to Strep,” with additional terms including “defense response to bacterium,” “response to jasmonic acid,” “phospholipase,” and “acylglycerol lipase” ( Figure 3B ; Supplementary Table S4 ). Furthermore, downregulated DEGs in “response to Strep (2,4-D)” conferred significant enrichment for 13 GO terms, such as “lipid transport,” “extracellular region,” and “cell wall biogenesis.”

Analysis of the systemic response to Streptomyces in leaf samples yielded a smaller transcriptional response, as leaf tissue samples were collected early in the tuber development process, prior to the establishment of common scab symptoms. The foliar “response to Strep” comparison yielded 148 upregulated genes and 17 downregulated genes, and 11 upregulated DEGs and 15 downregulated DEGs were identified in “response to Strep (2,4-D)” ( Supplementary Figure S2A ; Supplementary Tables S5 , S6 ). GO enrichment analysis indicated significant enrichments only from upregulated DEGs in “response to Strep,” with terms corresponding to the apoplast, extracellular matrix, and cell wall, and multiple molecular function terms associated with underlying DEGs encoding germin proteins ( Supplementary Figure S2B ; Supplementary Tables S5–S7 ).

To assess direct transcriptional responses to 2,4-D in leaf tissue, we sampled leaf tissue for RNA-seq 2 days after 2,4-D treatment, corresponding to 8 weeks prior to tuber harvest and preceding the development of common scab symptoms in tuber tissue. Leaf tissue was sampled within the context of Streptomyces pathogen in the soil (“response to 2,4-D (Strep)”) and without Streptomyces present in the soil (“response to 2,4-D”). In the leaf response to foliar 2,4-D alone, we identified 840 DEGs, consisting of 649 upregulated genes and 191 downregulated genes ( Figure 4A ; Supplementary Table S8 ). Of the 231 upregulated DEGs in “response to 2,4-D (Strep),” 156 overlapped with “response to 2,4-D,” as did 58 of 198 downregulated DEGs. Via GO enrichment analysis, pathways activated by 2,4-D corresponded to functions such as growth and differentiation (e.g., “cytokinin-activated signaling pathway,” “G1/S transition of mitotic cell cycle,” “meristem determinacy,” “microtubule binding”), biotic stress response (e.g., “defense response to fungus,” “innate immunity activating, cell surface,” “response to herbivore”), and carbohydrate metabolism (“carbohydrate metabolic process”) ( Figure 4B ; Supplementary Tables S8–S10 ). Accordingly, underlying genes were associated with a variety of protein classes, such as expansins, auxin efflux carriers and response proteins, cyclins, PR proteins, chitinases, peroxidases, and thioredoxins. GO enrichments derived from downregulated genes were limited to seven total terms and included “carbohydrate transport,” associated with underlying SWEET protein genes, as well as “solute:sodium symporter activity” and “positive regulation of programmed cell death” ( Supplementary Figure S3 ; Supplementary Tables S8–S10 ).

Because foliar 2,4-D treatment enhances resistance to common scab in tubers, we next sought to explore the transcriptional response to foliar 2,4-D treatment, alone and in the context of Streptomyces infection, in tuber tissue. 2,4-D was applied early in the tuber development process, and transcriptional response to foliar 2,4-D alone (“response to 2,4-D”) was limited, with only six upregulated DEGs identified ( Figure 5A ; Supplementary Table S11 ). Among these DEGs were two genes corresponding to Nodulin MtN3 family proteins, members of the SWEET family of sucrose transporters, and GO enrichment analysis conveyed strong (363–892-fold) enrichments for five corresponding GO terms, such as “sucrose transport,” from these two genes alone ( Figure 5B ; Supplementary Tables S11 , S13 ). However, a stronger transcriptional response was observed in the “response to 2,4-D (Strep)” comparison, including 342 upregulated and 583 downregulated DEGs ( Figure 5A ; Supplementary Table S12 ). Eight significantly enriched GO terms were identified from upregulated genes, including “response to hydrogen peroxide,” associated with underlying heat shock protein and WRKY transcription factor genes, as well as “acylglycerol lipase” and “phospholipase” terms corresponding to patatin and phospholipase genes ( Figure 5B ; Supplementary Tables S12 , S13 ). Downregulated genes contributed to 21 enriched GO terms, including several terms associated with growth, such as “synctium formation” and “plant-type secondary cell wall biogenesis.”

Among the enriched GO terms identified in the DEGs from tuber responses to 2,4-D were the molecular function terms “acylglycerol lipase” and “phospholipase” ( Figure 5B ; Supplementary Table S13 ), and enrichment for these terms was also observed in 2,4-D-treated tubers in response to Streptomyces infection (“response to Strep (2,4-D)”; Figure 3B ; Supplementary Table S4 ). Accordingly, the DEG sets from these comparison groups were found to include a number of genes with putative lipase, fatty acid desaturase, fatty acid hydroxylase, lipid transfer, and other functions of interest ( Supplementary Tables S3 , S12 ). To probe a potential role for tuber fatty acid metabolic profiles in the systemic response to foliar 2,4-D, we analyzed fatty acid contents in tuber tissue harvested from plants treated with 200 mg L⁻¹ of 2,4-D, 400 mg L⁻¹ of 2,4-D, or mock treatment ( Supplementary Table S14 ). Esterified fatty acids from tuber skin samples were quantified via the GC-FID panel; however, most classes of fatty acids were below the detection limit, potentially indicating insufficient sensitivity of lipidomic profiling, and no significant differences between treatment groups were identified.

Upregulated DEGs in the tuber response to 2,4-D contributed to GO term enrichments associated with sucrose transport ( Figure 5B ; Supplementary Table S13 ), and among upregulated tuber DEGs in “response to Strep (2,4-D),” “response to 2,4-D,” and “response to 2,4-D (Strep)” were genes corresponding to SWEET proteins and sucrose synthases ( Supplementary Tables S3 , S11 , S12 ). Therefore, we analyzed tubers for sucrose, glucose, and fructose content separately in both periderm and medulla tissue following spray treatment with 200 mg L⁻¹ of 2,4-D, 400 mg L⁻¹ of 2,4-D, or mock treatment. Tuber samples were collected at harvest, 12 weeks after inoculation of the potting mix with a mock inoculum, S. stelliscabiei strain NY02-3A, or S. acidiscabies strain ATCC49003 ( Supplementary Figure S4 ). While 2,4-D treatments did lead to increased abundance of sucrose in both periderm and medulla tissue, sucrose was detected only at low levels in the periderm (the site of infection), and there was no significant dosage effect for any inoculation condition–tissue type combination based on the Kruskal–Wallis non-parametric rank-sum test. Treatment with 2,4-D significantly increased glucose abundance in the medulla and fructose abundance in the medulla and periderm tissue of mock-inoculated plants. However, these increases in glucose and fructose were not observed in pathogen-inoculated tubers.