To obtain candidates of grape MATEs, BLASTP search (e ≤ 0.1) was performed against the CRIBI v2.1 genome annotation based on the V. vinifera 12X.0 genome assembly (Adam-Blondon, 2014) in Phytozome v13¹ (Goodstein et al., 2012) using the amino acid sequences of Arabidopsis MATEs (Li et al., 2002) as queries. Against obtained non-redundant candidates of grape MATEs, the presence of conserved MATE domain was confirmed by Pfam² (El-Gebali et al., 2019) and by the Conserved Domain Database of NCBI (CDD,³ Shennan Lu et al., 2020). Among the candidates, the proteins with predicted MATE domain by both Pfam and CDD were identified as grape MATEs. Molecular weight and isoelectric point were estimated by ProtParam⁴ (Gasteiger et al., 2005). The number of transmembrane domains was predicted by the Simple Modular Architecture Research Tool (SMART,⁵ Letunic et al., 2021).

MATEs in other grape genome annotations, that is, the CRIBI v1, the Genoscope, and the Vcost.v3 were identified as follows. BLASTP was performed against the CRIBI v1 genome annotation based on the V. vinifera 12X.0 genome assembly in EnsemblPlants.⁶ Whole protein sequences of the Genoscope genome annotation based on the V. vinifera 12X.0 genome assembly were downloaded from URGI⁷ or those of the Vcost.v3 genome annotation based on the V. vinifera 12X.2 genome assembly (Canaguier et al., 2017) were downloaded from URGI⁸ and in house BLASTP was performed.

Amino acid sequences of MATEs of grape and other plant species (Li et al., 2002; Takanashi et al., 2014; Biala-Leonhard et al., 2021) were aligned using the ClustalW program.⁹ The phylogenetic tree was created by MEGA X (Kumar et al., 2018) using the neighbor-joining model.

Gene structure (exon–intron structure) was schematized using coding sequences (CDS) and genomic sequence of the grape MATEs by the Gene Structure Display Server (GSDS,¹⁰ Hu et al., 2015). The Multiple Expectation Maximization for Motif Elicitation (MEME,¹¹ Bailey et al., 2009) was used to identify the conserved motifs in the amino acid sequences of grape MATEs. The parameters were set as follows: site distribution on zero or one occurrence per sequences, the maximum number of motifs was 12, and motif width is between 6 and 50.

The gene expression data in various organs, tissues, and developmental stages of grape by microarray analysis were downloaded from Grape eFP Browser (Fasoli et al., 2012).¹² The gene expression data were converted to base 10 logarithm and then heatmap was created by using HemI 1.0 (Deng et al., 2014).¹³

To identify MATEs in the grape genome, BLAST searches were performed against the grape genome assembly 12X.0 (Adam-Blondon, 2014) and genome annotation CRIBI v2.1 (Vitulo et al., 2014) using Arabidopsis MATEs as queries. Consequently, 78 genes were hit in the grape genome. The presence of MATE domains in the candidates was confirmed using domain search tools Pfam and CDD. As a result, two of the 78 hits did not have a motif predicted by both Pfam and CDD and were therefore excluded from further analyses. Fifty-eight candidates predicted to possess the MATE domain by both Pfam and CDD were identified as grape MATEs and analyzed in further study. The 18 candidates predicted by either Pfam or CDD to possess the MATE domain are listed in Supplementary Table S2.

The whole grape genome was sequenced and annotated in 2007 (Jaillon et al., 2007; Velasco et al., 2007). Since then, the sequence information was improved from 8X to 12X in 2009 (Adam-Blondon, 2014), and annotation was revised to CRIBI v1 (Jérôme et al., 2012), Genoscope, CRIBI v2 (Vitulo et al., 2014) and Vcost v3 (Canaguier et al., 2017). Different annotations were assigned different gene IDs, and they were not unified. This has led to confusion in grape research. Therefore, the corresponding gene IDs in the different annotations were linked in this study.

A BLAST search of the 58 MATE genes identified in CRIBI v2 to the genes in the three annotations (CRIBI v1, Genoscope, and Vcost v3) was performed. Most genes showed 1:1 correspondence; however, some genes corresponded to 2. These genes are discussed below.

Two genes (VIT_217s0000g02990 and VIT_217s0000g03000) in CRIBI v2 corresponded to one gene (GSVIVT01008369001) in Genoscope (Table 1), although these genes had a 1:1 correspondence in CRIBI v1 and Vcost v3. These suggest that the annotation of Genoscope may be miss annotation.

VIT_211s0052g01560 in CRIBI v2 corresponded to two genes (VIT_11s0052g01570 and VIT_11s0052g01560) in CRIBI v1, two genes (GSVIVT01029131001 and GSVIVT01029132001) in Genoscope, and two genes (Vitvi11g01693 and Vitvi11g01692) in Vcost v3. In the CRIBI v2 annotation, for VIT_211s0052g01560, eight splicing variants (VIT_211s0052g01560.1–8) were suggested, and VIT_211s0052g01560.1, 3, 4, and VIT_211s0052g01560.2, 5, 6, 7, and 8 were derived from different genome positions (Supplementary Figure S1). VIT_211s0052g01560.1, 3, and 4 correspond to VIT_11s0052g0157 in CRIBI v1, GSVIVT01029131001 in Genoscope, and Vitvi11g01693 in Vcost v3. In contrast, VIT_211s0052g01560.2, 5, 6, 7, and 8 corresponded to VIT_11s0052g0156 in CRIBI v1, GSVIVT01029132001 in Genoscope, and Vitvi11g01692 in Vcost v3, respectively (Supplementary Table S1). From these results, we concluded that the CRIBI v2 annotation VIT_211s0052g01560 contains two MATE genes, and 59 genes encoding MATEs are present in the grape genome.

For the 59 identified grape MATEs, information was collected based on the 12X.0 genome assembly and CRIBI v2.1 annotation. The number of amino acid residues, molecular weight, isoelectric point, and the number of transmembrane domains of MATE proteins and splicing variants are summarized in Table 1. The predicted subcellular localizations are listed in Supplementary Table S1. The number of amino acids was 262–689, molecular weights were 28,382–75,281, isoelectric points were 5–10, and the number of transmembrane domains was 3–12.

A phylogenetic tree of the amino acid sequences of 59 grape MATEs, 56 Arabidopsis MATEs, and the analyzed MATEs of various plants was generated (Figure 1). The tree revealed four groups of grape MATEs: 14 grape MATEs in Group 1, 28 in Group 2, 7 in Group 3, and 10 in Group 4. The 59 grape MATEs were named according to the gene ID in each group: VvMATE1-14 in Group 1, VvMATE15-42 in Group 2, VvMATE43-49 in Group 3, and VvMATE50-59 in Group 4 (Supplementary Table S1).

Exon-intron structures and protein motifs were shown with the phylogenetic tree generated by amino acid sequence homology (Figure 2A). The gene structures of grape MATEs were schematized in Figure 2C. The exon–intron structures were conserved in each group. The MATEs in group 1 had 5–8 exons and tended to have a long intron between the first and second exons. The MATEs in Group 2 have 7–9 exons, with long introns between the first and second exons and between the second and third exons. The MATEs in Group 3 had a larger number of exons than the other groups, that is, 10–13 exons. In contrast to Group 3, the MATEs in Group 4 have smaller exon numbers than the other groups, that is, 1–3 exons.

Protein motifs of grape MATEs were analyzed and schematized, as shown in Figure 2B. Twelve motifs were identified in grape MATEs. Generally, the motif patterns in Groups 1, 2, and 4 were similar. The motif patterns of Group 3 were different from those of the other groups, suggesting that the evolution and function of Group 3 were different from those of the other groups.

As for plant species, such as rice (Wang et al., 2016a) and flax (Ali et al., 2021), the grape MATEs were classified into four groups. Previous reports (Wang et al., 2016a) suggested that group members share characteristics and transport similar substrates.

Group 1 included AtDTX19, AtDTX1, and AtDTX6 in Arabidopsis, NtJAT1 in tobacco, and LaMATE2 in white lupin (Figure 1). AtDTX19 transports tetramethylammonium and protects the roots from toxic compounds in the soil (Diener et al., 2001). AtDTX1 transports alkaloids, antibiotics, and toxic compounds (Li et al., 2002). AtDTX6 transports paraquat and relates to paraquat tolerance (Xia et al., 2021). NtJAT1 transports alkaloids, such as nicotine (Morita et al., 2009). LaMATE2 is responsible for exporting isoflavonoids, such as genistein, from the roots to recruit Rhizobia (Biala-Leonhard et al., 2021). These results suggested that grape MATEs in Group 1 are involved in the transport of toxic compounds, alkaloids, and flavonoids. VvMATE1 and VvMATE2 have high homology with LaMATE2, suggesting that VvMATE1 and VvMATE2 have the potential to transport flavonoids.

Group 2 included AtDTX41, AtDTX35, AtDTX33, and AtDTX25 of Arabidopsis and NtJAT2, NtMATE1, and NtMATE2 of tobacco (Figure 1). AtDTX41 is responsible for accumulating proanthocyanidin precursors in vacuoles and for anthocyanin transport (Marinova et al., 2007). AtDTX35 is involved in flavonoid transport (Thompson et al., 2010), and also reported to be a chloride channel (Zhang et al., 2017). AtDTX33 also reported to be a chloride channel (Zhang et al., 2017). AtDTX25 transports ascorbate to vacuole and relates to Fe transport to seeds (Hoang et al., 2021). NtJAT2, NtMATE1, and NtMATE2 are involved in the transport of alkaloids, such as nicotine (Morita et al., 2009; Shoji et al., 2009; Shitan et al., 2014). Based on these reports, the grape MATEs in Group 2 are considered to transport flavonoids and alkaloids. The possibility that the grape MATEs in Group 2 were involved in the transport of polyphenolic compounds will be discussed later.

Group 3 included AtDTX43 and AtDTX43 in Arabidopsis and ZmMATE1 in maize (Figure 1). AtDTX43 contributes to iron and zinc homeostasis by transporting citric acid (Pineau et al., 2012). ZmMATE1 and AtDTX42 are responsible for the aluminum detoxification by transporting citrate (Maron et al., 2010; Wang et al., 2016b). These results suggested that the grape MATEs in Group 3 were responsible for organic acid transport.

Group 4 included AtDTX50 and AtDTX51 in Arabidopsis (Figure 1). AtDTX50 is responsible for the efflux of abscisic acid (ABA; Zhang et al., 2014). AtDTX51 has been reported to contribute to auxin and salicylic acid (SA) signaling (Sun et al., 2011; Li et al., 2014). These results suggested that some grape MATEs in Group 4 function in the transport or signaling of phytohormones, including ABA, auxin, and SA, and play an important role in the grapes’ growth and defense response.

To understand the functions of MATEs in the growth and berry development of grapes, the gene expression in various organs, tissues, and growth stages of berries was profiled. Microarray data were obtained from the Grape eFP Browser, and a heat map was created (Figure 3). Although the genes of most grape MATEs were uniformly expressed in various organs, tissues, and growth stages of berries, some genes were expressed in specific organs, tissues, or growth stages of berries.

VvMATE31, VvMATE42, and VvMATE46 showed extremely high expression in all organs, tissues, and growth stages of berries, suggesting that they function as the major MATEs in grapes. VvMATE22 was highly expressed in seed and berry tissues at early developmental stages, and its expression decreased with development, suggesting that VvMATE22 function is important in young grape berries. VvMATE36 was highly expressed in pollen, and VvMATE50 expression was higher in the rachis than in other organs, although their expression was not specific.