The VviERF6L clade was expanded to 18 members in the PN40024 reference genome
Novel VviERF6Ls genes were discovered by manually searching for conserved amino acid (AA) motifs in the newly annotated PN40024 genome. Along with the 12 previously identified VviERF6L genes , five additional genes were first identified (Table 1) from this manual curation. These additional genes were identified in unannotated sections of chromosome sequences by searching for specific AA motifs across the individual chromosomes using tools in ORCAE  where the reference grape genome sequence, PN40024, is stored. Structural models were confirmed in ORCAE using both mRNA  and expressed sequence tag (EST) data to confirm 5’ and 3’ ends of annotated sequences .
An additional VviERF6L was identified with an in-silico detection strategy. The manually curated VviERF6Ls were confirmed and substantiated as members of this clade from protein motifs identified in MEME [22, 23]. MEME revealed VviERF6L proteins consist of nine highly conserved AA motifs (Fig. 1 and Additional Files 1 and 2). The nine motifs are referred to in order of E-value with the lowest value motif corresponding to Motif 1 (Additional File 2). To identify additional novel VviERF6Ls that may have been overlooked with the manual annotation, an in-silico detection strategy was devised using the first (Motif 5) and last (Motif 4) spatial AA motifs to query the Vitis proteome. Genome coordinates that contained either the first, last, or both motifs were extracted corresponding to the potential proteins containing the motif(s) of interest. When only the first or the last motif was detected, the putative protein sequence was extended to 280 AA to obtain the potential full-length protein. This strategy confirmed the five novel VviERF6L genes from the manual curation and identified a sixth, increasing the members of the VviERF6L clade from 12 in the V1 annotation to 18 in the V2 annotation of PN40024 (Table 1).
The nine VviERF6L protein motifs were detected as being significantly present (p < 1.73x10-179) and conserved (E < 8.8x10-14) among the 18 VviERF6Ls (Additional File 2). These motifs had an average length of ~280 AA with the longest and shortest motifs (Motif 1 and Motif 9) having lengths of 50 and 7 AA, respectively (Additional File 2), resultant of the MEME settings used. Specific VviERF6L protein motif sequence and location per VviERF6L can be found in Additional File 3. Four protein motifs were identifiable and had previously been characterized as regulatory domains of other ERF Group IX TFs (Fig. 1) . These protein motifs included the AP2/ERF domain (DNA-binding; Motifs 1 and 2), the CMIX-2 (N-terminal acidic transactivation; Motif 5) the CMIX-5 (MAP kinase phosphorylation site; Motif 4), and the CMIX-6 (MAP kinase phosphorylation site; Motif 9) domains (Fig. 1). Motifs 1, 2, and 7 were present in all eighteen VviERF6Ls. Motifs 3, 4, 5, 6, and 9 were present in 77.8% of VviERF6L proteins, and Motif 8 was present in nine of the 18 VviERF6Ls (calculated from Additional File 3).
The VviERF6L AP2/ERF domain is homologous to that of Arabidopsis (At)ERF1 and 096. To identify VviERF6L sequence conservation with proteins in Arabidopsis thaliana, the nine motifs were queried in InterPro and the AP2/ERF domain was modeled in SWISS-model [24, 25]. The AP2/ERF domain (Motifs 1 and 2) of VviERF6L1, 2, 3, 4, 5, 6, 7, and 13 had the closest identity with that of AtERF1 with an average identity of 75.5% (Additional File 4). VviERF6L8, 9, 10, 11, 12, 15, 16, 17, and 18 had an average identity of 70.8% with the AtERF096 AP2/ERF domain, identified as the closest ortholog (Additional File 4).
VviERF6L12 and 14 appear to be truncated proteins. VviERF6L14 lacks the first 4 N-terminal motifs (Fig. 1), with no matching publicly available RNA-Seq or EST reads and insufficient sequence information in this region of the PN40024 genome in ORCAE. Besides VviERF6L15, VviERF6L14 is the only other non-mono-exonal VviERF6L. The true start codon of VviERF6L14 may exist in what is currently the un-sequenced region that is presently annotated as an intron and can be viewed in ORCAE . Despite potential mis-annotation of VviERF6L14 gene coordinates, promoter (see later) and protein motif analysis (Fig. 1, and Additional Files 1 and 2) validate this gene as a VviERF6L. VviERF6L12 appears to be a functional truncated protein (Fig. 1), supported by mRNA and EST read mapping across the length of the transcript in ORCAE. VviERF6L12 lacks the first 4 N-terminal motifs, which correspond to potential regulatory domains including the CMIX-2 domain (Fig. 1). VviERF6L12 is also missing Motif 4 corresponding to a CMIX-5 domain. VviERF6L3, 6, 9, 11, 13, 16, and 18 do not share consensus Motif 8 (Fig. 1). These proteins have higher amino acid variability in this region. VviERF6L10, 15, and 17 are also missing Motif 4 (Fig. 1).
The 18 VviERF6L proteins are a conserved clade. A multiple sequence comparison by log-expectation (MUSCLE) multiple sequence alignment (MSA) was performed to better understand the diversity within the VviERF6L clade. Percent identity was extracted from a MUSCLE alignment (Fig. 2). The 18 VviERF6Ls share high sequence conservation (average of 73.8%), with VviERF6L12, one of the truncated VviERF6Ls, diverging the most with an average percent identity of 50.9% (calculated from Fig. 2).
The VviERF6L clade is expanded in Vitis vinifera relative to other plant species. The number of ERF6L paralogs were identified in the species that had the genes with the highest orthology to VviERF6L1 from the Pan-taxonomic Compara Gene Tree in Gramene update 2018 containing 44 genomes  including the V1 annotation of PN40024 . The number of potential ERF6L orthologs was quantified in carrot (D. carota), soybean (G. max), tomato (S. lycopersicum), and potato (S. tuberosum) (Additional File 5). Vitis vinifera had 4.5-fold more ERF6L paralogs than tomato and potato, 9-fold more than soybean, and 17 more potential ERF6L genes than carrot (Additional File 5).
The VviERF6L clade is expanded across Vitis genotypes
Additional VviERF6Ls were identified in the translated Cabernet Sauvignon (CS) genome  indicating the VviERF6L clade members vary with grape genotypes. The nine PN40024 VviERF6L protein motifs were utilized to detect VviERF6Ls in the proteome sequence of CS using TOMTOM . Translated genes that contained at least three of the nine PN40024 VviERF6L protein motifs were extracted and analyzed with MEME as potential VviERF6Ls. These genes were used to identify CS specific VviERF6L protein motifs. TOMTOM used the CS cultivar specific VviERF6L protein motifs to identify additional potential CS VviERF6Ls that were missed using the nine PN40024 motifs (Additional Files 2 and 6-9). Thirteen highly conserved (E < 1.3x10-2) CS protein motifs (Additional File 2) were identified. The CS protein motifs were very similar to those of PN40024 (Additional Files 1, 7, and 9-10). Homology between PN40024 and CS VviERF6L protein motifs was quantified with protein BLAST  (Additional File 10). CS Motifs 1-6 shared 100% identity with corresponding PN40024 motifs, and CS Motif 7 shared ~71% with PN40024 Motif 9 (Additional File 10). In total, 26 CS VviERF6Ls were identified (Additional Files 8-9, and 11). Interestingly, unique VviERF6L sequences were identified in CS like VvCabSauv08_H0036F_008.ver1.0.g139880, which appears to be a novel VviERF6L not conserved in PN40024 (Additional Files 12-13). Lengths of CS VviERF6L proteins are in Additional File 11. CS VviERF6Ls (~300 AA residues) were on average 20 AA residues longer than PN40024 (~280 AA residues) (Table 1 and Additional File 11). CS lacked paralogs similar to PN40024 VviERF6L3, 8, 11, and 14 and had distinct VviERF6Ls (like VvCabSauv08_P0367F.ver1.0.g601540 and VvCabSauv08_H0036F_008.ver1.0.g139880), without a clearly distinguishable PN40024 equivalent. VvCabSauv08_H0036F_008.ver1.0.g139910 (596 AA residues) contained duplicated Motif 1-4, 5-9, and 11 (Additional File 6 and 9). VvCabSauv08_H0036F_008.ver1.0.g139950 (839 AA residues) consisted of duplicate Motif 1, 2, 7, and 12 and triplicate Motif 3, 5, 6, and 11. These two genes were about two and three times the length of the average CS VviERF6L (~300 AA residues (calculated from Additional File 10)), respectively. VvCabSauv08_H0036F_008.ver1.0.g139990 was a severely truncated VviERF6L (106 AA residues), completely lacking any conserved N-terminal motif (Additional Files 6 and 9). VvCabSauv08_P0070F.ver1.0.g450750 was of comparable length (243 AA residues) to the average CS VviERF6L, but this gene had more variable sequence, containing only four of the thirteen conserved motifs (Additional Files 6 and 9).
Chardonnay (CH)  and Carménère (CA)  also have expanded VviERF6L clades with 15 and 14 VviERF6Ls respectively (Additional Files 12-13). VviERF6Ls from PN40024 and CS were queried in protein BLAST in genome sequences of CH and CA to identify VviERF6Ls in these genotypes. The genomes of CH and CA were not released at this time; only BLAST was publicly available. Additional novel VviERF6Ls may exist in these genotypes, which may be identified using the motif detection strategy described for PN40024 and CS when the genomes become fully available.
The PN40024, CS, CH, and CA VviERF6Ls were more similar across Vitis vinifera genotypes than to other VviERF proteins (Additional File 12). To distinguish relationships between the highly homologous members of the VviERF6L clade in the AP2/ERF subfamily IX , a maximum likelihood phylogenetic tree was generated from Vitis vinifera PN40024, CS, CH, and CA VviERF6L paralogs and PN40024 VviERF proteins (gene names and protein sequences available in Additional File 13). The tree was created using the Jones-Taylor-Thornton model with the Bootstrap method test in MEGA X . Sequences were extracted from the PN40024 V2 assembly V3 structural annotation , CS genome , CH BLAST-tool , and the CA BLAST tool . All predicted VviERF6L proteins grouped together from the four genotypes examined (Additional File 12). Vitvi05g01525, corresponding to a putative disease related PRF protein , clustered with CH and CA VviERF6Ls in the multi-gentoype VviERF6L clade. This gene is inadequately sequenced on ORCAE, having 4,512 N’s in the coding sequence, and it is unclear if this gene is correctly positioned or annotated. The VviERF6L clade is distinct from other members of the AP2/ERF family (Additional File 12), and the VviERF6L protein sequences are highly conserved.
VviERF6L promoter regions are distinct with several conserved motifs
The PN40224 VviERF6Ls have variable promoter regions with several conserved and repeated cis-acting elements. To gain insight into the transcriptional regulation of the highly conserved VviERF6L genes in the PN40024 genome, -3000 bp upstream from the transcription start site (TSS) for the 18 PN40024 VviERF6Ls was analyzed with PLACE , and a multiple sequence alignment was performed to compare the putative promoter regions (Fig. 3). These regions showed greater diversity than VviERF6L protein sequences, averaging 48.7% relative to 81.05% identity (calculated from DNA sequences). PN40024 VviERF6L promoter region motif coordinates are in Additional File 14. A total of 200 known motifs were identified in the VviERF6L -3,000 bp putative promoter regions (calculated from Additional File 14). Of these cis-elements, 42 were present in all VviERF6L upstream regions (Additional File 15). The CAATBOX1 was the most over-represented motif across VviERF6L putative promoters, repeated 885 times, followed by DOFCOREZM (864 repetitions) and CACTFTPPCA1 (845 repetitions) (Additional File 15). These three motifs had an average of ~46 repeats per VviERF6L promoter. Numerous other cis-elements were repeated hundreds of times including ARR1AT and MYCCONSENSUSAT motifs.
Although the VviERF6Ls shared several highly repeated cis-regulatory elements, there were numerous differences across the VviERF6L promoter regions. The VviERF6L12 promoter region contained the highest number of ROOTMOTIFTAPOX1 (82 repeats), with the closest VviERF6L, VviERF6L14 having 41 repeats and the other VviERF6Ls having even fewer. The VviERF6L12 promoter region also had the most duplication of TATABOX3, ACGTATERD1 (analogous with VviERF6L1), SEF1MOTIF, SEF4MOTIFGM7S, WBOXATNPR1, and LECPLEACS2 (Additional Files 15 and 16). Three unique motifs were detected in the promoter region of VviERF6L12 that were not present in any other VviERF6L promoter: ABREZMRAB28, PALBOXLPC and UP1ATMASD (Additional File 14). Although the VviERF6L protein sequences are highly conserved, there is considerable variation in the VviERF6L promoter regions, indicating these genes are under unique transcriptional regulation.
VviERF6L genes are expressed in many organs and tissues
Examining the grapevine gene atlas , VviERF6Ls were expressed in numerous grapevine organs, across developmental stages, and in tissues including berries, stamen, buds, tendrils, flowers, pollen, seeds, leaves, and roots (Additional File 17). VviERF6L1, 5, and 12 were the most commonly expressed VviERF6Ls across various tissues. VviERF6L6, 10, and 11 were less broadly expressed across tissues. VviERF6L6 was only expressed in the rachis, carpel, petal, leaves, roots, and buds while VviERF6L10 and 11 were expressed in these tissues as well as berry flesh. VviERF6L8 was expressed in the same number of tissues as VviERF6L2 and 3 and comparable to VviERF6L4 and 7. Breaking down the berry into pulp, seed, and skin VviERF6Ls generally had significantly higher expression in the skin at pre-veraison and seed at maturity when the berries are red, soft, and ready to harvest (RSH) (Additional File 18) [GSE49569] . VviERF6L12 was the only VviERF6L to increase in signal intensity in the pulp as berries developed, and this gene maintained the highest expression level in all berry tissues at all developmental stages.
VviERF6L meta-data analysis parameters
VviERF6L expression was extensively examined across existing transcriptomic data in the literature. To better differentiate and understand the potential functional roles of VviERF6L genes in Vitis, VviEFRF6L gene expression was examined in a meta-data analysis of VviERF6L gene expression performed using 75 publicly available microarray and 24 RNA-Seq data series downloaded from NCBI Gene Expression Omnibus (GEO)  and Sequence Read Archive  data bases. The following data are examples of results found, but many other datasets demonstrate similar patterns. The example data series selected are simplified for visualization purposes. The data series (Additional File 19) investigated for VviERF6L gene expression met the following criteria: the experiment contained at least three individual biological or experimental replicates, VviERF6L gene expression was detectable, and at least one VviERF6L was identified as a differentially expressed gene (DEG) in the author’s original differential expression analysis (DEA). Results are discussed based on the author’s original DEA and statistical analysis unless otherwise specified.
Four and twelve probe sets were utilized on the grape Affymetrix and NimbleGen microarrays, respectively, with possible cross-hybridization occurring amongst the 18 PN40024 VviERF6Ls . Numerous occurrences of probe cross-hybridization for NimbleGen microarrays of the VviERF6L genes were previously determined  (Additional File 20), making it important to consider these results in terms of the VviERF6L clade response as opposed to individual gene responses. With the short-read length of the RNA-Seq data sets analyzed here and the high homology of the VviERF6Ls, it is unclear how distinguishable VviERF6Ls are individually in the RNA-Seq analysis.
Data series are referred to as in original publications. SRP117281, PRJNA516950, GSE67191, GSE62744, and GSE62745 were chosen as representative RNA-Seq data series of abiotic stress, berry development, and biotic stress to re-analyze with the V3 structural annotation of PN40024. The data series selected for re-analysis with the V3 structural annotation of PN40024 were used for weighted gene co-expression network analysis (WGCNA) to identify genes that share the same expression pattern as the VviERF6Ls under various stress conditions and developmental stages. All other data series demonstrated in the figures were graphed based on the original author’s transcript abundance quantification and statistics.
VviERF6L genes are involved in multiple abiotic stress responses
VviERF6L genes respond to water deficit and salinity
VviERF6Ls were differentially expressed in response to numerous abiotic stresses including water deficit and salinity. VviERF6Ls were significantly differentially expressed in CS leaves exposed to rapid dehydration for one hour (Fig. 4) [GSE78920] . The VviERF6Ls shared the same general expression pattern in response to rapid dehydration with a significant increase in transcript abundance early in the experiment followed by decreased transcript abundance and plateauing thereafter. VviERF6L12 and 9 demonstrated the highest and lowest levels of expression respectively. VviERF6L1 was among the most responsive VviERF6Ls, increasing quickly to the severe stress within one hour of treatment and decreasing at all time points after that to eventually be at the same level of expression as control plants after 24 hours of treatment. VviERF6L1 was chosen as a representative gene of the VviERF6L clade for subsequent RT-qPCR and overexpression experiments. RT-qPCR was performed for VviERF6L1 on CS leaves treated with 10 µM Protone (s-ABA). VviERF6L1 transcript abundance in CS leaves did not respond to ABA treatment (Additional File 21), indicating the water deficit response may follow an ABA independent pathway.
CS shoot tips exposed to a severe 16-day salt and water deficit treatment also show significantly increased VviERF6L transcript abundance (Fig. 5) . Probe sets 1618661_s_at (VviERF6L12) and 1619390_at (VviERF6L11) were highly induced by extreme water deficit and salt stress, but 1613698_at (cross-hybridizes to VviERF6L2 and VviERF4) and 1613799_at (VviERF6L3) were not induced. The accumulation of VviERF6L transcripts on day 16, the point at which both abiotic stresses were most severe indicate some VviERF6Ls may play a role in extreme salt and water deficit responses in grapevine.
VviERF6L differential expression in response to water deficit is supported by recent more comprehensive RNA-Seq data in which four Vitis species (Vitis vinifera cv. CS, Vitis champinii cv. Ramsey (RA), Vitis riparia cv. Riparia Gloire (RI), and Vitis vinifera x Vitis girdiana cv. SC2 (SC)) were treated with well-watered and moderate water deficit (WD) conditions in the form of a natural dry-down for two weeks . The grapevines demonstrate significantly differential VviERF6L expression patterns within each species (Fig. 6). For example, VviERF6L1 is not expressed in SC, but it is expressed in the three other species. Within each Species x Organ x Treatment group, the 18 VviERF6Ls follow similar expression patterns to each other (Fig. 6). VviERF6Ls were differentially expressed in leaves and roots in response to the WD (Fig. 6). VviERF6Ls were significantly more highly expressed in roots than in leaves. Consistently, the VviERF6Ls have higher expression in roots than leaves under both Control and WD conditions (13 average TPM (standard error of the mean (SEM) = ± 1.5) vs 4.6 average TPM (SEM = ± 0.51) for roots and leaves, respectively). As a general trend VviERF6L transcripts were decreased in response to WD (e.g. VviERF6L1). The majority of VviERF6Ls have relatively low expression levels apart from VviERF6L12 that demonstrated a significantly higher level of expression (Z-score for two population proportions p < 0.00001). VviERF6L8 was consistently the lowest expressed VviERF6L across organs and treatments. Interestingly, leaves from RA (a drought tolerant rootstock originating from Texas, USA) had a significantly higher accumulation of VviERF6L12 transcripts in week 2 WD relative RA Control leaves as well as compared to 2 WD treatment leaves of the other three species. The other species, which are more drought sensitive, did not exhibit an increase or as high of an increase in VviERF6L12 transcripts in the leaves in response to two weeks of WD.
Amongst the VviERF6L clade, DEA showed VviERF6L1 and 18 were the most common DEGs in DEA contrasts of interest. DEA was performed for each genotype for WD vs. Control and for each WD treated species to the others for weeks one and two for roots and leaves individually. A list of DEA contrasts of interest for this data set are located in Additional File 22. The frequency at which each VviERF6L was a DEG in the DEA contrasts of interest was quantified (Additional File 23). VviERF6L1 was identified as the most responsive VviERF6L, being a DEG in 14 DEA contrasts of interest followed by VviERF6L18 (a DEG in 10 DEA contrasts of interest); however, both genes were expressed at relatively low levels (Fig. 6). The other VviERF6Ls varied in DEG frequency in the DEA contrasts of interest (Additional File 23). VviERF6L7 and 8 were not DEGs in any contrast of interest. The range of frequencies each VviERF6L was a DEG in the DEA contrasts of interest was consistent with the results of the promoter analysis, indicating that while these genes are highly conserved, they are under distinct regulation.
VviERF6L genes respond to chilling and cold
VviERF6Ls were differentially expressed in leaves in response to chilling, cold, and freezing in the meta-data analysis. Many of the VviERF6Ls responded with analogous expression patterns. Recent RNA-Seq data in which five Vitis vinifera cultivars (Cabernet Franc, Chardonnay, Riesling, Sangiovese, and Tocai Friulano) were treated with chilling (ACC), freezing (FRZ) or a chilling acclimation followed by the freeze treatment (A+F) demonstrate significant VviERF6L differential expression (Fig. 7) . As with WD response, various V. vinifera cultivars demonstrated unique VviERF6L expression in leaves in response to cold treatments. For example, comparing freezing vs. control DEA across cultivars, Chardonnay had the highest significant increase in VviERF6L1 transcript abundance while Tocai Friulano and Sangiovese did not demonstrate ERF6L1 differential expression (Fig. 7 and DEA from original publication ). All cultivars had a decrease in transcript abundance of VviERF6L1 in the chilling vs. control treatment comparison (Fig. 7). The transcript abundance of VviERF6L11 and VviERF6L12 was increased in all genotypes in response to the freezing treatment (Fig. 7). The results in Figure 7 are supported by microarray data in which VviERF6Ls were differentially expressed in CS shoot tips exposed to a chilling treatment for 0, 4, and 8 hours (Additional File 24) . To confirm these results, RT-qPCR was performed on VviERF6L1 for RA, RI, CS, and SC whole canopy and single leaves treated with 4°C chilling for 0-2 hours. In contrast to the previous freezing and chilling treatments, these chilling experiments did not result in a significant difference of VviERF6L1 transcript abundance relative to control; there was however, a significant increase in CBF1 transcript abundance used as a positive control in chilled samples (Additional File 25). It is possible VviERF6L1 was not the most responsive VviERF6L in these species under these conditions. Together these examples from the meta-data analysis reveal VviERF6Ls are differentially expressed with complex responses to temperature reduction in cultivar-, temperature- and time-dependent manners and may play a role in cold response in grapevine.
VviERF6L genes respond to light intensity
VviERF6Ls were significantly differentially expressed in response to increased light exposure. CS berries exposed to varying light intensity through leaf removal or leaf movement at veraison demonstrated reduced VviERF6L12 transcript abundance in de-seeded berries (pulp and skin only) at late veraison and harvest. The majority of the other VviERF6Ls (all with lower transcript abundance than VviERF6L12) had enhanced transcript accumulation at harvest (particularly with leaf removal) relative to control conditions in which no leaves were (re)moved  (Fig. 8) [GSE121146]. In this experiment, leaves were cut off vines for the leaf removal treatment and physically bound in place for the leaf moving treatment. These actions may have elicited a wounding response. However, as the berries remained intact on the vine and only the leaves were removed, it is likely that the increased VviERF6L transcript abundance in the berries is associated with increased light exposure and not a wounding response. The accumulation of VviERF6L transcripts with enhanced light exposure at harvest in combination with the abundance of VviERF6L promoter motifs associated with light response indicates VviERF6Ls may play a role in grapevine response to light intensity. Supportive of these data and VviERF6L light response, VviERF6Ls were also differentially expressed in berries grown under a double cropping system with summer and winter harvests [GSE103226] . In the summer, CS berries grown in this system had the highest level of VviERF6L expression at the end of veraison (EL36), while there was a distinct depression in VviERF6L transcript level for winter berries (Additional File 26). These data sets support the hypothesis that VviERF6Ls have a cultivar-specific response to abiotic stress and may play a role in response to light intensity.
VviERF6L genes are involved in various biotic stress responses
VviERF6L genes respond to Neofusicoccum parvum
VviERF6Ls are differentially expressed in response to Neofusicoccum parvum. CS plants inoculated with N. parvum had significantly enhanced VviERF6L transcript accumulation in woody stems two weeks after treatment (Fig. 9) [GSE97900] . Interestingly, VviERF6Ls also responded to the wounding aspect of this treatment, which consisted of taking a power drill to the woody stem. The wounding response remained significant for the majority of VviERF6Ls up to two weeks after the treatment (Fig. 9). VviERF6L12 and 8 consistently demonstrated the highest and lowest expression levels, respectively (Fig. 9).
In general, VviERF6L expression levels significantly increased in response to E. necator inoculation in leaves of V. vinifera cv. Carignane and six partially resistant Asian accessions (DVIT3351.27 (DVIT3351), Hussiene, Karadzhandal, Khalchii, O34-16, Sochal, and Vavilov) [GSE67191] . The cultivars showed similar expression patterns with differences in expression levels of the VviERF6Ls (Additional File 27). VviERF6L12 and 16 generally had the highest expression in response to the powdery mildew inoculation (Additional File 27). VviERF6L8 generally had low, but detectable expression with the exception of DVIT3351 in which VviERF6L8 had higher expression levels, similar to those of VviERF6L3-7, and Vavilov that did not demonstrate any VviERF6L8 expression at either time point or treatment. It is possible VviERF6Ls play diverging roles in response to various pathogens.
Whole Zinfandel berries had high levels of VviERF6L12 counts across berry development in both control and red blotch-associated virus treatment in two separate vineyards (Fig. 10) [GSE85812] . In control berries, the expression of the VviERF6L clade was highest in the pre-veraison (PRV) stage with subsequent decline in transcript abundance as the berry maturity stage increased. VviERF6L8 expression was only detectable in one vineyard at pre-veraison; in all other cases it does not appear to be expressed (Fig. 10). VviERF6L expression at pre-veraison significantly decreased in response to red blotch-associated virus in at least one vineyard (Fig. 10) . VviERF6Ls showed variable expression across the vineyards in response to red blotch-associate virus particularly at pre-veraison. VviERF6L5, 7, and 10 had increased counts in infected samples at veraison (Fig. 10). At post-veraison, VviERF6L1, 5, and 11 had higher expression in red blotch-associated virus samples in both vineyards, and at harvest, VviERF6L3 had lower counts in infected berry tissue (Fig. 10). While various VviERF6Ls were significantly differentially expressed in response to these pathogens (Figs. 9-10 and Additional File 27), the pattern and degree of expression across genotypes was not consistent.
VviERF6L genes are involved in berry development
Unlike abiotic and biotic stress responses, VviERF6L gene expression patterns and levels show general conservation across cultivars with VviERF6L12 consistently having the highest transcript abundance in red and white berries . One study examining red and white berry development over four developmental stages (pea sized (Pea), touching (Touch), softening (Soft), and harvest (Harv)) showed differential expression of VviERF6Ls across berry ripening, but at a low expression level (Fig. 11). VviERF6Ls had the highest number of VviERF6L transcripts at the pea-sized and touching stages of berry development. VviERF6L transcript abundance decreased as berries softened and was even lower at harvest (Fig. 11). VviERF6L2, 12, and 13 were among the highest expressed VviERF6Ls in white berries with the addition of VviERF6L5, 15, and 16 for red cultivars in the early stages of berry development. In the later stages of berry development, VviERF6L12 clearly had the highest transcript abundance (Fig. 11). From pea-sized to touching berries, VviERF6L8 was expressed in white berries (except Passerina) to a comparable level with other VviERF6Ls, including VviERF6L7 and 9. At all other developmental stages, VviERF6L8 was negligibly expressed. In red berries, VviERF6L8 was only expressed in Barbera in pea-sized berries. VviERF6L expression across berry development is also consistent across vineyards and years [GSE97578]  [GSE41633]  (Additional Files 28-29). VviERF6L signal intensity peaked significantly at pre-veraison and generally declined as berries approached full ripening (FR) (Additional File 28). There were subtle changes in signal intensity level over the years and vineyards (Additional File 29), but generally expression levels were similar and the VviERF6L expression pattern remained conserved, indicating these genes may not be strongly influenced by environment during berry development.
VviLISCL3 and VviCML45 were the most connected genes to the VviERF6Ls
Two genes share a similar expression pattern as the 18 VviERF6Ls across various cultivars, organs, and treatments. The meta-data analysis was completed with a gene co-expression analysis to identify genes sharing expression patterns for all VviERF6Ls as a clade between the five data series that were re-analyzed with the V3 annotation of PN40024 (SRP117281, PRJNA516950, GSE67191, GSE62744, and GSE62745). The top 100 genes most connected to each VviERF6L were extracted from the TOM (Topological Overlap Matrix) for each WGCNA. Common co-expressed genes were identified by comparing these sets of genes. In total, two genes were identified in all five data series that were the most connected to the VviERF6L clade across the various conditions and variables of each data series (Additional File 30). The two co-expressed genes were a Scarecrow-like transcription factor, VviLISCL3 (Vitvi06g00489), and a calmodulin-like protein, VviCML45 (Vitvi14g01975). There were several other genes that shared expression patterns with the VviERF6L clade in four out of the five data series including VviERF1 and VviWRKY33 (Additional File 31). The full list of genes co-expressed with the VviERF6L clade in four of the five data series is in Additional File 31. Six of the 16 genes sharing the VviERF6L expression pattern were unannotated.
After analyzing the VviERF6L clade as a whole, the co-expression analysis was repeated for each VviERF6L individually. Surprisingly, no gene was connected to any VviERF6L in all five data series in this individual analysis, not even the other VviERF6Ls. Because no genes were co-expressed with any VviERF6L individually, genes co-expressed in four out of the five RNA-Seq data series were considered. VviWRKY33 was the only gene to be co-expressed in four out of the five data series in this individual VviERF6L gene co-expression analysis but only for VviERF6L11 and 16. The low number of genes co-expressed with VviERF6Ls in all five RNA-Seq series (16 for the clade co-expression analysis (Additional File 31) and 0 for the individual VviERF6L co-expression analysis) may be a result of the diverse RNA-Seq series utilized that examined various organs, genotypes, developmental stages, and stresses.
Summary of the meta-analysis
Generally, VviERF6Ls were lowly expressed in all data sets, but these genes demonstrate striking fold changes in expression levels and significant differential expression under numerous conditions. VviERF6Ls are broadly expressed across grapevine organs, tissues and in response to various abiotic and biotic stresses as well as throughout berry development (Fig. 12). VviERF6Ls appear to increase in expression in response to severe water deficit and salinity. However, over more long-term moderate water deficit, VviERF6Ls are generally decreased. VviERF6Ls have distinct differential expression in response to cold and light in different cultivars. VviERF6Ls are differentially expressed in response to various pathogens, but the level of expression varies depending on the pathogen and cultivar. VviERF6Ls are also differentially expressed across berry development with the highest expression levels as berries transition into veraison. VviERF6L expression patterns are highly conserved across cultivars, vineyards, and years. The broad range of VviERF6L expression across tissues and expression patterns are conserved throughout numerous data series. The transcriptional response for each member of the VviERF6L clade was dependent on a number of factors (organ, time, treatment, duration, severity, genotype, etc.). The transcriptomic response for each experimental design, while generally conserved, was unique for each VviERF6L (i.e. some members of the clade increased in transcript abundance, some decreased, and others did not respond under a specific condition). The individuality emerging in the VviERF6L clade as well as divergent observations for different severities of similar treatments makes it difficult to generalize common responses. However, a diagram (Fig. 12) was constructed to summarize the conditions that elicited VviERF6Ls responses as well as conditions requiring additional data to further elucidate the role VviERF6Ls play in grapevine. As transcriptomic technology evolves, the VviERF6Ls will be able to be better differentiated and understanding of the clade will be improved.
Two results are clear, first, in all the data series discussed, VviERF6L12, one of the truncated VviERF6Ls, repeatedly demonstrated significantly higher expression than any of the other VviERF6Ls. VviERF6L12 had 2–228 times more transcript abundance than the average of all other VviERF6Ls. Across the treatments and conditions of the selected data sets, the transcript abundance of VviERF6L12 averaged 25.6-times more RMA-normalized signal intensity, counts, FPKM, or TPM than the average expression of the other VviERF6Ls (Additional File 32). Second, VviERF6L8 was frequently the lowest expressed VviERF6L with no detectable expression in certain cultivars. The vast range of VviERF6L expression levels made it necessary to log2 transform the data in the meta-data analysis, so each VviERF6L expression was visible. Although the VviERF6L clade is highly conserved, each VviERF6L is under unique transcriptional regulation.
Overexpressing VviERF6L1 in Vitis had minor impact on the transcriptome and phenotype
In previous microarray studies, VviERF6L1 appeared to be the most responsive VviERF6L, with transcript abundance increasing in CS in response to severe leaf dehydration  and with changing sugar levels in a study of the late stages of berry ripening . Further investigation of ERF6L1 function was investigated with VviERF6L1 overexpression and knockdown lines. Attempts to establish Vvierf6l1 knock-down lines failed; plants were unable to be re-established after transformation with a T-DNA insertion. An empty vector control (G1) and VviERF6L1 overexpression lines (L12-1, L12-2, L12-3, L12-11, and L12-23) were created in a Seyval Blanc background under the control of a bi-directional duplex 35S promoter fused to EGFP/NPTII in pECBC . Overexpression was confirmed with semi-quantitative PCR and RT-qPCR to verify stable over-expression (Additional File 33). Extensive phenotyping revealed VviERF6L1 overexpression lines did not exhibit a morphological phenotype under control conditions, or in response to salinity, water deficit or pathogen spread treatments (Additional File 34). Potential downstream targets of VviERF6L1 were determined with differential expression analysis on RNA-Seq data from leaves of the empty vector control (G1) and VviERF6L1 overexpression lines (L12-3, L12-11, and L12-23). VviERF6L1 was the only VviERF6L gene with significantly higher expression in the overexpression lines compared with G1 (Additional File 35). In total, only fourteen genes were significantly differentially expressed in all three overexpression lines relative to G1 (Additional Files 36 and 37). Up regulated genes included: VviERF6L1 (Vitvi16g00350), CYP722A1 (Vitvi04g01352), CRK8 (Vitvi11g01160), LAC14 (Vitvi18g01479). Down regulated genes included: three PRB1 (pathogenesis-related protein 1) genes (Vitvi03g00757, Vitvi03g01649, Vitvi03g01651), unannotated genes (Vitvi03g01650, Vitvi07g01985, Vitvi11g01692, Vitvi18g02319), MET1 (Vitvi12g02119), WAKL1 (Vitvi18g00024), and LIMYB (Vitvi01g01444).