Identification of genes encoding ALMT and MATE transporters as candidate aluminum tolerance genes from a typical acid soil plant, Psychotria rubra (Rubiaceae)

To understand how tropical plants have adapted to acid soils, we analyzed the transcriptome of seedlings of Psychotria rubra, a typical species found on acid soils. Using RNA-seq, we identified 22,798 genes, including several encoding proteins of the Al3+-activated malate transporter (ALMT) and multidrug and toxic compound extrusion (MATE) families. Molecular phylogenetic analysis of ALMTs and MATEs revealed the grouping of those from P. rubra, which may be useful to select targets for elucidating the molecular basis of P. rubra adaptation to acid soils in the future. The transcriptome datasets obtained in this study would help us to further understand the physiological and ecological aspects of soil adaptation of Psychotria species.


INTRODUCTION
Understanding how plants adapt to various soils is essential in plant biology (Hiradate, Ma & Matsumoto, 2007) because plants are sessile and need to grow roots in settled soils. Adaptation to acid soils is an important issue because acid soils cover a considerable part of Earth's arable land and prevent agriculture of most plants (Von Uexküll & Mutert, 1995). In acid soils, aluminum is toxic to root tip growth, and various aspects, from molecular to physiological, of the mechanisms of aluminum toxicity have been explored in detail (reviewed in Ma, 2007).
Proteins of the Al 3+ -activated malate transporter (ALMT) and multidrug and toxic compound extrusion (MATE) families are likely involved in plant adaptation to acid soils (Delhaize, Gruber & Ryan, 2007;Delhaize, Ma & Ryan, 2012;Ma, 2007;Ma, Chen & Shen, 2014). ALMTs and MATEs release organic acids (malate and citrate, respectively), which bind Al 3+ and detoxify it. ALMTs and MATEs related to aluminum tolerance have been identified in model and agricultural plants (wheat: Sasaki et al., 2004;barley: Delhaize et al., 2004;Furukawa et al., 2007;maize: Maron et al., 2010;Arabidopsis: Hoekenga et al., 2006;Liu et al., 2009), but the composition of those families in non-model wild plants has hardly been explored. Psychotria (Rubiaceae) is a highly diversified genus comprising more than 1,600 species distributed in all tropical and some subtropical regions (Hamilton, 1989;Davis et al., 2001;Razafimandimbison et al., 2014). Because Psychotria species adapt to several types of soils (e.g., soils with high concentrations of nickel; Merlot et al., 2014), the genus is an ideal target to use to understand how adaptation of wild plants to different types of soils has evolved. In this study, we report ALMTs and MATEs of P. rubra, which grows on acid soils (Miyawaki, 1989).

Sampling, RNA extraction, RNA-seq library preparation, and sequencing
Seeds of P. rubra were collected on Mt. Nago-dake, in the north of Okinawa Island, and seedlings were grown in a greenhouse of the National Institute of Technology, Okinawa College ( Fig. 1). RNA was extracted from the seedlings (three-five cm height) using an RNeasy Plant Mini kit (Qiagen, Hilden, Germany). An RNA-seq library was prepared using a TruSeq Stranded mRNA Sample Prep Kit (Illumina, San Diego, CA, USA). The library was sequenced (100-bp paired-end reads) on an Illumina HiSeq 2500 platform. The above procedures on RNA-seq were outsourced to Hokkaido System Science Corporation, Japan. De novo assembly and annotation of transcriptome sequences FASTQ files were filtered, the reads with poor-quality bases (Q < 20) and those shorter than 20-bp were excluded, and adapter sequences were removed in cutadapt v.

Extraction of ALMTs and MATEs of P. rubra and molecular phylogenetic analysis
We searched for P. rubra ALMTs and MATEs by BLASTP (e-value <1e −5 ) using sequences of TaALMT1 (UniProt database ID: Q76LB1) and HvAACT1 (UniProt ID: A7M6U2; this was the first MATE identified in barley (Furukawa et al., 2007) as queries). We then performed BLASTP searches against the Swiss-Prot database (e-value <1e −5 ) with each ALMT and MATE of P. rubra and selected the top 10 hits for each. The amino acid sequences of ALMTs and MATEs from P. rubra and related sequences from Swiss-Prot and other studies (Dreyer et al., 2012;Liu et al., 2016) were aligned in MAFFT v. 7.407 software (Katoh & Standley, 2013). We excluded four P. rubra ALMT sequences (Prub_02169, Prub_02171, Prub_08405, Prub_08554) from the following analysis because of poor alignment. We selected only plant MATEs, including HvAACT1 and those from (Liu et al., 2016), for the following analysis. Neighbor-joining trees of ALMTs and MATEs were constructed in MEGA7 software (Kumar, Stecher & Tamura, 2016) with the following settings: Poisson model, Uniform rates, and Pairwise deletion. To evaluate the confidence of phylogenetic trees, bootstrap tests were performed with 1,000 replicates.   P . p . P h y p a t P p 1 s 5 3 1 3 6 V 6 . 1 P . p . P h y p a t P p 1 s 7 2 2 7 6 V 6 . 1 1 0 0 P . p . P h y p a t P p 1 s 6 8 1 0 8 V 6 . 1 P .p . P h y p a t P p 1 s 5 2 0 6 V 6 .1 P .p . P h y p a t P p 1 s 8 7 1 0 3 V 6 .1

RESULTS AND DISCUSSION
Our RNA-seq analysis of P. rubra yielded 57,110,261 paired-end reads, of which 53,994,410 remained after filtering. De novo assembly of the remaining reads resulted in 131,578 contigs (Table 1), in which we found 24,687 non-redundant amino acid sequences; 22,798 of them were expected to originate from P. rubra itself as indicated by BLASTP analysis of TAIR10 (the remaining 1,889 sequences were almost no-hit in Swiss-Prot database or included those from microorganisms, etc). Among these 22,798 sequences, 19,701 ones were hit against Swiss-Prot database, and gene ontology (GO) numbers were found in 1,0348 ones. From these sequences, we found 14 ALMTs and 12 MATEs (Table 2). A ra b id o p s is th a li a n a A tD T X 1 N ic o ti a n a ta b a c u m N tJ A T 1 9 9 A ra b id o p si s th a lia n a A tA L F 5 9 3 Ar ab ido ps is th ali an a At TT 12 Bra ssic a rapa BrT T12 Using 14 ALMTs from P. rubra as queries in a BLASTP search against the Swiss-Prot database, we found 14 homologs, all of plant origin (mainly from Arabidopsis). Using a similar approach, we found 13 homologs of MATEs (six from Arabidopsis and seven from non-plant organisms). Molecular phylogenetic analysis did not detect P. rubra orthologs of TaALMT1 (UniProt ID: Q76LB1) (Fig. 2). ScALMT1 (accession number: ABA62397) from rye (Secale cereale) is the only known clear ortholog of TaALMT1 (Delhaize, Gruber & Ryan, 2007). ALMT1 of Arabidopsis (UniProt ID: Q9SJE9), encoded by an aluminum tolerance gene (Hoekenga et al., 2006), is clearly distinct from TaALMT1 (Delhaize, Gruber & Ryan, 2007). Thus, ALMTs related to aluminum tolerance may have multiple origins. Molecular phylogenetic analysis of MATEs revealed no clear orthologs of HvAACT1 (UniProt ID: A7M6U2) in P. rubra (Fig. 3).
Expression and functional analyses of ALMTs and MATEs of P. rubra would be useful for understanding their roles in soil adaptation of Psychotria (e.g., with and without Al treatment). Another aluminum tolerance mechanism of plants (different from releasing organic acids), aluminum accumulation, has been reported in several species of the Rubiaceae (Jansen et al., 2003). Genes related to this function are also good targets for future studies to explain the molecular basis of acid soil adaptation of P. rubra.

CONCLUSIONS
We succeeded in identifying transcriptome sequences including ALMTs and MATEs from P. rubra in this study. Comparative transcriptome analysis of several Psychotria species would help us to clarify the physiological and ecological aspects of diversification of this genus (e.g., adaptation to metalliferous soils; Merlot et al., 2014). In particular, Psychotria manillensis, which is closely related to P. rubra, is reportedly adapted to non-acid soils (Miyawaki, 1989). Thus, comparative analysis of P. rubra and P. manillensis should help to explain how soil adaptation-related genes are involved in adaptive evolution of Psychotria species.