# MECHANISMS OF PLANT-ALUMINUM INTERACTIONS IN ACIDIC SOILS

EDITED BY : Jian Li Yang, Wei Fan and Peter Ryan PUBLISHED IN : Frontiers in Plant Science

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88945-909-4 DOI 10.3389/978-2-88945-909-4

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# MECHANISMS OF PLANT-ALUMINUM INTERACTIONS IN ACIDIC SOILS

Topic Editors: Jian Li Yang, Zhejiang University, China Wei Fan, Yunnan Agricultural University, China Peter Ryan, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia

Citation: Yang, J. L., Fan, W., Ryan, P., eds. (2019). Mechanisms of Plant-Aluminum Interactions in Acidic Soils. Lausanne: Frontiers Media. doi: 10.3389/978-2-88945-909-4

# Table of Contents


Hans-Peter Maurer, Walter J. Horst and Emmanuel Delhaize

*23 Emerging Pleiotropic Mechanisms Underlying Aluminum Resistance and Phosphorus Acquisition on Acidic Soils* Jurandir V. Magalhaes, Miguel A. Piñeros, Laiane S. Maciel and Leon V. Kochian

*35 Aluminum–Nitrogen Interactions in the Soil–Plant System* Xue Q. Zhao and Ren F. Shen


Gongle Wang, Mingyue Li, Xingsheng Li and Yang-Dong Guo

*102 Soybean NADP-Malic Enzyme Functions in Malate and Citrate Metabolism and Contributes to Their Efflux Under Al Stress* Ying Zhou, Zhenming Yang, Yuezi Xu, Haoran Sun, Zhitao Sun, Bao Lin,

Wenjing Sun and Jiangfeng You

# Mining Beneficial Genes for Aluminum Tolerance Within a Core Collection of Rice Landraces Through Genome-Wide Association Mapping With High Density SNPs From Specific-Locus Amplified Fragment Sequencing

#### Minghui Zhao1†, Jiayu Song1†, Aiting Wu<sup>1</sup> , Tao Hu<sup>1</sup> and Jinquan Li 2,3 \*

*<sup>1</sup> Rice Research Institute, Shenyang Agriculture University, Shenyang, China, <sup>2</sup> State Key Laboratory for Conservation and Utilization of Subtropical Agro-Bioresources, South China Agricultural University, Guangzhou, China, <sup>3</sup> Department of Plant Breeding and Genetics, Max Planck Institute for Plant Breeding Research, Cologne, Germany*

Trivalent Aluminum (Al3+) in acidic soils is harmful to root growth and significantly reduce crop yields. Therefore, mining beneficial genes for Al tolerance is valuable for rice production. The objective of this research is to identify some beneficial genes for Al tolerance from rice landraces with high density SNP set from SLAF-seq (Specific-Locus Amplified Fragment sequencing). A total of 67,511 SNPs were obtained from SLAF-seq and used for genome-wide association study (GWAS) for Al tolerance with the 150 accessions of rice landraces in the Ting's rice core collection. The results showed that rice landraces in the Ting's rice core collection possessed a wide-range of variation for Al tolerance, measured by relative root elongation (RRE). With the mixed linear models, GWAS identified a total of 25 associations between SNPs and Al tolerant trait with *p* < 0.001 and false discovery rate (FDR) <10%. The explained percentage by quantitative trait locus (QTL) to phenotypic variation was from 7.27 to 13.31%. Five of twenty five QTLs identified in this study were co-localized with the previously cloned genes or previously identified QTLs related to Al tolerance or root growth/development. These results indicated that landraces are important sources for Al tolerance in rice and the mapping results could provide important information to breed Al tolerant rice cultivars through marker-assisted selection.

Keywords: Aluminum tolerance, genome-wide association mapping, relative root elongation, rice landraces, Al tolerant QTL, SLAF-seq

#### Edited by:

*Peter Ryan, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia*

#### Reviewed by:

*Harsh Raman, New South Wales Department of Primary Industries, Australia Wricha Tyagi, Central Agricultural University, India Saul Justin Newman, Australian National University, Australia*

> \*Correspondence: *Jinquan Li lijinquan@scau.edu.cn*

*†These authors have contributed equally to this work*

#### Specialty section:

*This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science*

Received: *09 March 2018* Accepted: *27 November 2018* Published: *18 December 2018*

#### Citation:

*Zhao M, Song J, Wu A, Hu T and Li J (2018) Mining Beneficial Genes for Aluminum Tolerance Within a Core Collection of Rice Landraces Through Genome-Wide Association Mapping With High Density SNPs From Specific-Locus Amplified Fragment Sequencing. Front. Plant Sci. 9:1838. doi: 10.3389/fpls.2018.01838*

**4**

## INTRODUCTION

Rice (Oryza sativa L.) is an important crop in the world. There is about 13% of global rice field on acidic soils (Vonuexkull and Mutert, 1995). Trivalent aluminum (Al3+) in acidic soils is harmful to root growth and reduces significantly rice yield (Liu et al., 2012). It is a major toxin for plants on acid soils (Delhaize et al., 2012). Therefore, exploring the genetic mechanism of Al tolerance in rice is of importance to understand why Al3<sup>+</sup> is toxic to the plants and to breed Al tolerant varieties for rice production.

Serval researches on the genetic mechanism of Al tolerance in rice have been reported (e.g., Famoso et al., 2010, 2011; Cai et al., 2011). Previous researchers have identified a number of quantitative trait loci (QTLs) for Al tolerance in rice (Nguyen et al., 2002; Ma and Furukawa, 2003; Mao et al., 2004; Xue et al., 2007; Famoso et al., 2011), and found a few genes linked to Al tolerance (e.g., Yokosho et al., 2011; Chen et al., 2012; Huang et al., 2012; Xia et al., 2013; Li et al., 2014). In the context of QTL mapping for Al tolerance, most of these previous researches were conducted with bi-parent segregation populations and linkage mapping. Genome-wide association study (GWAS) make it possible to exploit natural genetic diversity and mine beneficial genes in the genome (Zhu et al., 2008). It is important to apply GWAS with modern genotyping technology for QTL mapping for Al tolerance.

In recent years, many QTLs for multiple traits have been identified using GWAS. For example, Huang et al. (2010) conducted GWAS for 14 agronomic traits with high density SNP set and 517 indica landraces of rice. Using GWAS with a diverse rice set of 383 accessions, Famoso et al. (2011) found 48 QTLs for Al tolerance, four of which co-localized with previously identified candidate genes for Al tolerance and two of which colocalized with previously identified Al-tolerant QTLs. Using 274 SSR markers and the same populations as this study, Zhang et al. (2016) performed an association study and found a total of 23 QTLs for Al tolerance. However, to our knowledge, no GWAS for rice Al tolerance has been performed using high-density SNPs with a core collection of rice landraces.

Recently, the whole genome sequencing technology is being increasingly used to accurately and rapidly detect numerous variants across the entire genome at the molecular level. The recently developed next-generation sequencingbased genotyping approach, i.e., specific-locus amplified fragment sequencing (SLAF-seq) method is a simplified genome sequencing technology that has shown to be highly accurate and cost-effective (Sun et al., 2013). SLAF-seq has been applied in genetic map construction, QTL mapping, and molecular breeding. However, to our knowledge, no high-density SNPs obtained from SLAF-seq technology has been applied for GWAS in a core collection of rice landraces.

Abundant germplasm resources for Al tolerance are available in the Asian cultivated rice, especially in rice landraces. As early as in 1920–1964, a total of 7,128 accessions of rice landraces had been collected by Prof. Ying Ting, which was named as Ting's rice collection (Li et al., 2011). They were from all over China as well as from some main rice cultivation countries. Based on 48 phenotypic data, Li et al. (2011) has constructed a rice core collection consisting of 150 accessions. The analysis of population structure indicated that there existed two subgroups mainly corresponding to indica and japonica subspecies and the LD decays to the threshold, i.e., the 95% quantile of r <sup>2</sup> between unlinked loci pairs, at 1.03 cM in the entire collection, which was about 200–500 kb in physical distance (Zhang et al., 2011; Li and Zhang, 2012). The large variation within the core collection provides an important gene pool of genetic diversity and beneficial genes for rice breeding. Therefore, it is worth to perform GWAS with such a core collection for Al tolerance in rice.

The objectives of the study were to (1) perform GWAS for rice Al tolerance to reveal the genetic basis for this complex trait; (2) identify novel functional candidate genes underlying the mapped regions; and (3) to mine the beneficial genes within the Ting's core collection of rice landraces with the newly developed high-density SNP set from SLAF-seq approach.

## MATERIALS AND METHODS

#### Plant Material

The Ting's rice core collection, i.e., a total of 150 accessions of rice landraces were used to screen their Al tolerance (**Table S1**). The core collection was constructed from 2,262 accessions of 7,128 based on a strategy of stepwise clustering and preferred sampling on adjusted Euclidean distances and weighted pair-group average method using integrated qualitative and quantitative traits (Li et al., 2011). It represents the diversity in the Ting's rice collection. Furthermore, Nipponbare and Xiangnuo 1 (Yang et al., 2007) were chosen as tolerant control and Nante (Fu et al., 2010), Xiangzhongxian 2 (Xu et al., 2004), and IR64 (Khatiwada et al., 1996) for Al sensitive control. These varieties were used to identify an appropriate concentration for Al toxicity.

#### Phenotyping for Al Tolerance

The Al tolerance for the 150 accessions of rice landraces were examined according to our previous research (Zhang et al., 2016). To choose an optimal Al3<sup>+</sup> concentration to screen Al tolerance, the seedlings for two Al tolerant and three Al sensitive rice varieties were exposed to 0.5 mM CaCl<sup>2</sup> (pH = 4.0) containing 50, 100, 150, 200, 250, 300, 350, 400, 450, and 500µM AlCl<sup>3</sup> (no other nutrient solution was applied), respectively. The Al3<sup>+</sup> concentration under which the largest difference of relative root elongation length (RRE) between the sensitive and tolerant varieties was chosen as an optimal Al3<sup>+</sup> concentration for screening of Al tolerance in the following experiment. In this case, the largest difference in RRE was observed at 100µM between the two tolerant and three sensitive varieties. Therefore, the 100µM AlCl<sup>3</sup> was used for screening of Al tolerance.

The 150 accessions of rice landraces cultivated at the farm of South China Agricultural University, Guangzhou (23◦ 16N, 113◦ 8E), during late season (July-November) in 2008 and 2009. The seeds were harvested each year. Uniform seeds in each year were surface sterilized in 1% H2O<sup>2</sup> for 30 min and rinsed with deionized water. Then the seeds were put into deionized water at 30◦C for 2 days in darkness for germination. The uniform seedlings were transferred to a net floating on a 0.5 mM CaCl<sup>2</sup> (pH = 4.0) solution in a 1.5 L plastic container. A randomized complete block design (RCBD) with three replicates was applied. Seedlings were grown at 28◦C for 48 h before being used for Al toxicity treatment. Then, the seedlings were exposed to 0.5 mM CaCl<sup>2</sup> (pH = 4.0) containing AlCl<sup>3</sup> for 24 h, and the root elongation length was measured for each sample. Then RRE was used to evaluate the degrees of Al tolerance of all landraces. RRE was calculated as follows: (root elongation length with Al treatment)/(root elongation length without Al treatment). Root length of 10 seedlings in each treatment was measured before and after treatments. The RRE for each genotype across the three replicates for 2008 and 2009 were calculated, respectively. The mean of RRE for 2 years was also calculated for each genotype. These phenotypic data were used for GWAS.

#### Genotyping of SNP Markers

The SLAF sequencing were conducted based on the standard protocol from Beijing Biomarker Technologies Corporation (http://www.biomarker.com.cn) and the introduction by Sun et al. (2013) and Song et al. (2018). To simplify, the first step was to perform a SLAF pre-design experiment with 8 accessions of landraces and different enzymes combinations. This step was used to evaluate the appropriate enzymes and sizes of restriction fragments. The SLAFs obtained by this step should be evenly distributed ascross the genome. The second step was to construct the SLAF library in accordance to the pre-design scheme. In this step, genomic DNA was digested by enzymes designed for individuals. Double barcodes were added to two rounds of PCR reactions to discriminate each individual and to facilitate the pooling of samples. In the third step, the purified DNA tags with indices and adaptors (SLAFs) of 300–400 bp were used and diluted for pair-end sequencing on an Illumina High-seq 2500 sequencing platform according to the Illumina sample preparation guide (Illumina, Inc.; San Diego, CA, US) at Beijing Biomarker Technologies Corporation. All polymorphic SLAF loci were genotyped according to the SNP loci at the reference genome. The SNPs with missing data > 20% across all genotypes as well as the SNPs with a minor allele frequency (MAF) (<5%) were excluded for the following statistical analysis. After filtration, 150 accessions of rice landraces with a total of 40,708 polymorphic SNPs were used for GWAS. The data of the SLAF sequencing have been uploaded to the BioSample database (BioSample accession SAMN10448484).

#### Statistical Analyses

The statistical model used for GWAS analysis was the PK mixed:Mip = µ + a<sup>p</sup> + Pz u=1 Diuυ<sup>u</sup> + g<sup>i</sup> <sup>∗</sup> + eip, where Mip was the phenotypic value of the ith entry carrying allele p, a<sup>p</sup> the effect of allele p, eip the residual, υ<sup>u</sup> the effect of the uth column of the population structure matrix D, and g ∗ <sup>i</sup> was the residual genetic effect of the ith entry (Yu et al., 2006; Stich et al., 2008).

Principal coordinate analysis (PCoA) was performed based on all SNPs after filtration. The first and second principal component was used as a D matrix of the above-mentioned association approach.

The kinship coefficient Kij between inbreds i and j were calculated on the basis of all SNP markers according to:KTij = Sij−1 <sup>1</sup>+<sup>T</sup> <sup>+</sup>1, where <sup>S</sup>ij was the proportion of marker loci with shared variants between inbreds i and j and T the average probability that a variant from one parent of inbred i and a variant from one parent of inbred j are alike in state, given that they are not identical by descent (Bernardo, 1993). For the series of Tvalues 0, 0.025, . . . , 0.975 **K** matrix between all inbreds was calculated. Negative kinship values between inbreds were set to 0. The optimum T-value was calculated according to Stich et al. (2008).

The R package EMMA Kang et al. (2008) was used to perform GWAS. The significance threshold of 0.001 and a false discovery rate (FDR) <10% were applied to test for significant associations between the traits and the SNP markers. The Bonferroni correction was used to adjust false positive rate in the multiple tests (Pocock et al., 1987). The FDR was calculated according to Benjamini and Hochberg (1995). For genome-wide studies with high density SNPs, one must consider the nonindependence of SNPs because of linkage disequilibrium (LD) when interpreting statistical significance (Li et al., 2012). To achieve this, we followed Duggal et al. (2008) to randomly select 1 SNP per LD block in addition to all the SNPs outside of blocks. The p-values for these SNPs were used for calculation of the FDR. The significantly associated SNPs with Al tolerance within the LD decay distance (i.e., 500 kb from our previous study) was grouped as one QTL. The percentage of genotypic variation explained by the significant SNPs was calculated by R 2 LR = 1 − exp(− 2 n (log L<sup>M</sup> − log L0)), where exp is an exponential function, logL<sup>M</sup> is the maximum log-likelihood of the model of interest, logL<sup>0</sup> is the maximum log-likelihood of the intercept-only model, n is the number of observations (Sun et al., 2010).

#### Searching Candidate Genes

To validate our mapping results and find a robust set of candidate genes, we searched the flanking regions ± 500 kb (the maximum LD decay distance in the core collection) of the significant associated SNP loci with Al tolerance to find previously mapped QTLs from the Rice QTL Map database (http://qtaro.abr.affrc. go.jp/qtab/table). Similarly, we searched the flanking regions of the significant associated loci (±500 kb) with Al tolerance to find previously cloned/identified candidate genes related to Al tolerance from the QTARO database (http://qtaro.abr.affrc.go.jp/ ogro/table). Because our measurement for Al tolerance was the relative root length with/without Al treatment, i.e., RRE, we think that the genes related to root development are corresponding to Al tolerance. Therefore, we mainly searched the candidate genes related to root development and Al tolerance within the searching regions. Aluminum tolerance genes identified by reverse genetics were found in the OryGenesDB (http://orygenesdb.cirad.fr/cgibin/searching.pl).

## RESULTS

The landraces in the Ting's core collection have RRE values ranged from 0.22 to 0.95, indicating a large variation for Al tolerance. The phenotypic distribution of RRE showed a normal distribution, indicating that aluminum tolerance is a quantitively inherited trait. The broad-sense heritability was 88.73% for Al tolerance.

With SLAF sequencing approach, a total of 116,643 highquality SLAFs were detected, with 24,889 polymorphic SLAF tags and a polymorphism rate of 21.34%. Each SLAF tag had an average coverage depth of 5.2×. The inner region of the polymorphic SLAF tags were further sequenced and a total of 67,511 SNPs were detected. After filtering the SNPs with missing data ≧20% across all genotypes and MAF ≦0.05, a total of 40,708 polymorphic SNPs were used for GWAS.

PCoA indicated that there were two clusters for the entire population (**Figure 1**), which was corresponding to their classification as indica and japonica types. Most of the kinship coefficients for any pair of landraces were zero (**Figures S1**, **S2**), indicating that these landraces are unrelated, which is due to that they were collected from a world-wide area and were from a core collection. There were also a few pairs of landraces showing high kinship coefficients.

A mixed linear model, i.e., PK model (Yu et al., 2006; Stich et al., 2008), which accounts for population structure and kinship, was used for GWAS for Al tolerance. To balance the false positive and negative rate, p < 0.001 and FDR <10% were used as the significant threshold to indicate whether a SNP was significantly associated with Al tolerance. The QQ plot indicated that the PK

to the first and second principal components, respectively. The number in the brackets indicated the percentage of genotypic variance explained by the

model effectively control the false positive (**Figure 2**). A total of 25 SNP regions were shown significantly associated with Al tolerance (**Table 1**, **Figure 3**, **Figures S3**, **S4**), but none of them reached the Bonferroni threshold (with a raw p < 2.46 × 10−<sup>8</sup> ). They were located on chromosomes 1-4, 6-7, 9 and 11. The QTLs explained individually from 7.27 to 13.31% of the phenotypic variance. The fixed effect was ranged from 0.092 to 0.256. There were different number of QTLs on each chromosome ranged from 1 to 7. The number of significantly associated SNPs for each QTL ranged from 1 to 17. Among them, qALT3.3 and qALT7.2 were detected in both years data as well as the mean of both years. Most QTLs were detected with the mean of both years data. The beneficial alleles for each significant QTL and their genotype

background were further examined (**Tables S2**, **S3**). To validate the mapping results, two databases, i.e., QTARO database and OryGenesDB, were used to screen the previously cloned genes and mapped QTLs around the flanking regions ±500 kb of the significant associated SNP loci with Al tolerance. A total of three QTLs mapped in this study were mapped to the same regions as the previously mapped QTLs for Al tolerance (**Table 1**). The closest distance for the previously mapped QTLs to the QTLs in this study ranged from 60.15 to 446.02 kb.

A total of three QTLs in this study were co-localized with the previously clone/identified genes (**Table 1**), including the wellknown Al tolerance gene STAR1. The candidate genes functions include Al tolerance, root growth, root development (e.g., root length, elongation, crown root). The closest distance for the candidate genes to QTLs in this study ranged from 3.85 to 449.75 kb.

Furthermore, one Al tolerance genes identified by mutation analysis from previous research, i.e., Os02g49790.1, was colocalized with the QTLs in this study (**Table 1**). The distance for the gene to the QTLs in this study is 264.06 kb. The gene has the

principal components.


**8**

function as aluminum-activated malate transporter or aluminum resistance protein.

In total, five of 25 QTLs identified in this study were co-localized with the previously cloned genes or previously identified QTLs related to Al tolerance and root traits.

#### DISCUSSIONS

Asian cultivated rice (Oryza sativa) was domesticated from its wild relative O. rufipogon (Kovach et al., 2007; Sang and Ge, 2007). Because of domestication and artificial selection of rice, genetic diversity has been remarkably reduced in many cases, and favorable alleles or genes might have been lost in the modern cultivars. Rice landraces are the intermediate form between modern cultivars and their ancestral species. Because of the less impact by artificial selection, landraces contain abundant genetic diversity and useful beneficial genes for modern cultivars. Moreover, transfer of beneficial genes from the intermediate forms to modern cultivars is considerably easier than from the ancestral wild species. Therefore, the identification and utilization of valuable genetic resources in landraces can be highly valuable for the genetic improvement of modern rice cultivars, for example, breeding varieties for Al tolerance.

The Ting's core collection of rice landraces is one of the earliest rice collection in China. Our previous studies indicated that two subgroups were presented in Ting's core collection, corresponding to indica and japonica subspecies (Zhang et al., 2011). Association studies were performed with 274 SSR markers for important agronomic trait and Al tolerance (Zhang et al., 2014, 2016), which confirmed that the core collection is a good population to map natural variations existing in the rice landraces. Compared to this study, a total of three QTLs, i.e., qALT1.1, qALT3.1, and qALT3.2, were identified in both our results and the previous research of Zhang et al. (2016). The closest distance for the QTL of Zhang et al. (2016) to the QTLs in this study ranged from 60.15 to 623.69 kb. In addition, there were three QTLs in the research of Zhang et al. (2016) having a distance between 1.7 and 2 Mb (about 8 cM) to the QTLs in this study. However, the mapping results were limited by the numbers of markers used in the previous studies. Because of its high-throughput and cost-effective nature, SLAF-seq is an ideal method for genotyping by sequencing and hence has been applied in this study. This method allowed us to obtain a total of 67,511 high-quality SNPs, which provide good foundation for our GWAS in this study.

In this study, we detected a total of 25 QTLs for Al tolerance (**Table 1**). The significant associations were distributed 8 of 12 chromosomes in rice, which was in accordance with the research of Famoso et al. (2011). The QTLs explained individually from 7.27 to 13.31% of the phenotypic variance in this study, which was smaller than those in research of Nguyen et al. (2003) and Famoso et al. (2011), while it was larger than that in research of Xue et al. (2007). This might be explained by different mapping populations used in the aforementioned studies and the total phenotypic variation were different. A total of 5 from 25 QTLs in this study were co-localized with the previously cloned genes or previously identified QTLs related to Al tolerance and root traits (**Table 1**), which was similar with the research of Famoso et al. (2011).

Huang C. F. et al. (2009) and Huang X. H. et al. (2009) found two genes, i.e., STAR1 and STAR2, responsible for Al tolerance in rice. The QTL qALT6.2 was co-localized with STAR1 with a minimum distance of 449.75 kb. Moreover, qALT6.2 was also colocalized with other two candidate genes OsHMA2 and OsPTR9 with a minimum distance of 61 and 237.92 kb, respectively, which have the function of Zn and Cd translocation, and lateral root formation.

Furthermore, the qALT1.1 in this study located at 938.62 kp away from the Al tolerance gene OsCDT3, which was identified by knockdown method (Xia et al., 2013). This QTLs was at 623.69 kb away from a previous mapped Al tolerance QTL (Zhang et al., 2016). The qALT1.6 in this study was co-localized with a candidate gene, i.e., OsFRDL4 with a distance of 433.58 kb (**Table 1**). OsFRDL4 is an Al tolerance gene identified by mutant method (Yokosho et al., 2011). Moreover, this QTL was colocalized with QTLs for Al tolerance identified by some previous researches (Wu et al., 1999, 2000; Mao et al., 2004) (**Table 1**) with a minimum distance of 108.94 kb.

The QTL qALT2.2 was co-localized with an aluminumactivated malate transporter gene (Os02g49790.1) with a minimum distance of 264.06 kb (**Table 1**). A previous mapped QTL associated with arsenic accumulation was located 228.11 kb away (Zhang et al., 2008), which implied that the mechanism of tolerance to metal ion (for example, Fe, As, Zn, Cd, etc) might have a similar metabolism way. The explanation could be supported by the observation that several candidate genes colocalized with the QTLs in this study have the functions on Fe and Cadmium uptake, Zn and Cd translocation, etc. The QTL qALT3.1 was co-localized with an Al tolerant QTL (Zhang et al., 2016).

It is interesting that the regions between 18,562,071 and 21,188,740 on chromosome 3 showed several peaks (the number of significant SNPs ranged from 6 to 14), corresponding to qALT3.2, qALT3.3, qALT3.4, and qALT3.5, significantly associated with Al tolerance in this study. However, only six candidate genes were identified within this region from the QTARO database and only one candidate gene was related to Cadmium and Iron uptake. No candidate genes were identified to associate with Al or other metal ion tolerance. As the mapping results in this study were highly significant, this region as well as other QTLs (qALT1.2, qALT1.5, qALT2.3, qALT3.6, qALT3.7, qALT4.1, qALT6.1, and qALT9.1) in this study where no candidate genes/previous mapped QTLs were found, could be new loci for Al tolerance and required further research.

## AUTHOR CONTRIBUTIONS

JL and MZ designed the study. MZ, JS, AW, and TH performed SLAF-seq experiment and data analyses. JL performed GWAS and statistical analyses. MZ, JL, and JS performed searching candidate genes/QTLs. JL wrote the paper. All authors read and approved the final manuscript.

## FUNDING

This work was supported by Fund of the National Natural Science Foundation of China grant 30700494, 31430062, National Natural Science Foundation of Liaoning province (201602670) and the National Key Research and Development Program of China (2016YFD0200707).

## ACKNOWLEDGMENTS

We are grateful to Dr. Peng Zhang from South China Agricultural University (SCAU) for providing aluminum tolerance measurement data, to Dr. Xiaoling Li, Dr. Xuelin Fu, and Prof. Xiangdong Liu for their assistance in the experiment, to Academician Prof. Yonggen Lu from SCAU and Academician Prof. Wenfu Chen from Shenyang agricultural university (SYAU) for their useful instructions.

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018. 01838/full#supplementary-material

Figure S1 | Kinship distribution for the landraces in the Ting's core collection.

Figure S2 | Kinship distribution for the landraces in the Ting's core collection, where the kinship coefficient equal to zero were excluded.

Figure S3 | Manhattan plot for genome-wide association study (GWAS) for aluminum tolerance measured by relative root elongation length (RRE) in 2008. The green line indicated the significant threshold at *p* < 0.001, and the red line indicated the Bonferroni significant threshold at *p* < 0.001 (with a raw *p* < 2.46 × 10<sup>8</sup> ).

Figure S4 | Manhattan plot for genome-wide association study (GWAS) for aluminum tolerance measured by relative root elongation length (RRE) in 2009. The green line indicated the significant threshold at *p* < 0.001, and the red line indicated the Bonferroni significant threshold at *p* < 0.001 (with a raw *p* < 2.46 × 10−<sup>8</sup> ).

Table S1 | Variety names, origin, and classification of *indica*-*japonica* for the 150 accessions of rice landraces in the Ting's core collection, where TI, IC, TJ, JC represent typical *indica*, *indica*-clined, typical *japonica*, *japonica*-clined rice, respectively.

Table S2 | SNP data for significant QTLs for each genotype with their relative root elongation length (RRE) values.

Table S3 | Allelic effect for the significant QTLs.

#### REFERENCES


construct core collection of rice (Oryza sativa). Rice Sci. 18, 46–55. doi: 10.1016/S1672-6308(11)60007-3


Zhu, C. S., Gore, M., Buckler, E. S., and Yu, J. M. (2008). Status and prospects of association mapping in plants. Plant Genome 1, 5–20. doi: 10.3835/plantgenome2008. 02.0089

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhao, Song, Wu, Hu and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Assessing How the Aluminum-Resistance Traits in Wheat and Rye Transfer to Hexaploid and Octoploid Triticale

Peter R. Ryan<sup>1</sup> \*, Dengfeng Dong1,2, Felix Teuber <sup>1</sup> , Neele Wendler 1,3, Karl H. Mühling<sup>3</sup> , Jie Liu1,4, Muyun Xu<sup>1</sup> , Naike Salvador Moreno1,5†, Jiangfeng You<sup>6</sup> , Hans-Peter Maurer <sup>7</sup> , Walter J. Horst <sup>8</sup> and Emmanuel Delhaize<sup>1</sup>

<sup>1</sup> CSIRO Agriculture and Food, Canberra, ACT, Australia, <sup>2</sup> College of Agriculture, Guangxi University, Nanning, China, 3 Institute of Plant Nutrition and Soil Science, Kiel University, Kiel, Germany, <sup>4</sup> College of Agronomy, Sichuan Agricultural University, Chengdu, China, <sup>5</sup> Department of Genetics, Faculty of Biology, Universidad Complutense, Madrid, Spain, <sup>6</sup> Laboratory of Soil and Plant Molecular Genetics, College of Plant Science, Jilin University, Changchun, China, <sup>7</sup> State Plant Breeding Institute, Universitaet Hohenheim, Stuttgart, Germany, <sup>8</sup> Institute for Plant Nutrition, Leibniz University Hanover, Hanover, Germany

#### Edited by:

Meixue Zhou, University of Tasmania, Australia

#### Reviewed by:

Atanas Ivanov Atanassov, Joint Genomic Center, Bulgaria Gaofeng Zhou, Murdoch University, Australia

> \*Correspondence: Peter R. Ryan peter.ryan@csiro.au

#### †Present Address:

Naike Salvador Moreno, Department of Cancer Biology and Comprehensive Cancer Center, Wake Forest University School of Medicine, Winston-Salem, NC, United States

#### Specialty section:

This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science

Received: 30 April 2018 Accepted: 24 August 2018 Published: 15 October 2018

#### Citation:

Ryan PR, Dong D, Teuber F, Wendler N, Mühling KH, Liu J, Xu M, Salvador Moreno N, You J, Maurer H-P, Horst WJ and Delhaize E (2018) Assessing How the Aluminum-Resistance Traits in Wheat and Rye Transfer to Hexaploid and Octoploid Triticale. Front. Plant Sci. 9:1334. doi: 10.3389/fpls.2018.01334 The mechanisms of aluminum (Al) resistance in wheat and rye involve the release of citrate and malate anions from the root apices. Many of the genes controlling these processes have been identified and their responses to Al treatment described in detail. This study investigated how the major Al resistance traits of wheat and rye are transferred to triticale (x Tritosecale Wittmack) which is a hybrid between wheat and rye. We generated octoploid and hexaploid triticale lines and compared them with the parental lines for their relative resistance to Al, organic anion efflux and expression of some of the genes encoding the transporters involved. We report that the strong Al resistance of rye was incompletely transferred to octoploid and hexaploid triticale. The wheat and rye parents contributed to the Al-resistance of octoploid triticale but the phenotypes were not additive. The Al resistance genes of hexaploid wheat, TaALMT1, and TaMATE1B, were more successfully expressed in octoploid triticale than the Al resistance genes in rye tested, ScALMT1 and ScFRDL2. This study demonstrates that an important stress-tolerance trait derived from hexaploid wheat was expressed in octoploid triticale. Since most commercial triticale lines are largely hexaploid types it would be beneficial to develop techniques to generate genetically-stable octoploid triticale material. This would enable other useful traits that are present in hexaploid but not tetraploid wheat, to be transferred to triticale.

Keywords: roots, acid soil, malate, citrate, Secale cereale, Triticum aestivum

## INTRODUCTION

Many important crop species are stable allopolyploids resulting from hybridisations between two separate but related species. Triticale (× Triticosecale Wittmack) is an allopolyploid because it is a hybrid between rye (Secale cereale L.) and wheat (Triticum aestivum L.). Triticale is a valuable grain crop that combines useful traits from wheat and rye. Wheat has greater yield and superior grain quality while rye is a forage crop with outstanding resistance to many biotic and abiotic stresses

**13**

including pathogens, low nutrient availability, soil pH and low temperatures. Wheat is typically used as the female parent and rye as the male parent because crosses are more stable if female plants have the larger ploidy of the two parents. The grain from this hybridisation are often sterile so the zygote from this cross is treated with colchicine to induce polyploidy and improve fertility (Mergoum and Gómez-Macpherson, 2004). When rye (diploid with genome RR) is hybridized with a hexaploid or bread wheat (hexaploid with genome AABBDD) the result is an "octoploid" triticale (AABBDDRR). When rye is hybridized with a tetraploid wheat (AABB) the result is a "hexaploid" triticale (AABBRR). Therefore triticale is amphidiploid meaning that it is diploid for the two parental genomes. Commercial triticale lines are mostly second generation hexaploid types because they often show better stability and performance than the octoploid types (Mergoum and Gómez-Macpherson, 2004).

Aluminum (Al) toxicity is a major limitation to crop production on acid soils because the concentration of soluble trivalent cations (Al3+) increases when soil pH falls below ∼4.5. Many species show a significant genotypic variation in resistance to Al stress and this is also the case for rye and bread wheat but not for durum wheat which is very sensitive of Al. Rye is among the most Al-resistant cereal species along with rice (Oryza sativa L.). Aniol and Gustafson (1984) investigated the Al resistance of triticale, wheat and rye and concluded that Al resistance of the wheat parent was an important determinant of the Al resistance of triticale. They also found that the strong resistance of rye was partially suppressed in the hybrid. When that report was published little information was available on the mechanisms of Al resistance in any plant species. It was later revealed that the major mechanisms for Al resistance in wheat and rye involve the release or efflux of malate and citrate anions from the root apices (Li et al., 2000; Ma et al., 2000; Delhaize et al., 2007; Stass et al., 2008; Ryan et al., 2011). Stass et al. (2008) compared contrasting genotypes of wheat and rye with the triticale hybrids and concluded that the Al resistance of triticale was mostly determined by citrate efflux, a trait that was largely controlled by the wheat genome.

Differences in malate efflux account for most of the genotypic variation in Al resistance in bread wheat but citrate efflux is important when malate efflux is absent (Ryan et al., 2009). Malate efflux is facilitated by an anion channel encoded by the aluminum-activated malate transporter, TaALMT1, gene on chromosome 4DL (Sasaki et al., 2004; Raman et al., 2005). Alresistant genotypes show a greater constitutive expression of TaALMT1 in the root apices than sensitive genotypes which is not affected by Al treatment. However, the TaALMT1 protein requires Al3<sup>+</sup> cations to trigger the malate release which means Al rapidly activates malate release (Sasaki et al., 2004). This rapid activation has been described as a Type I response which is consistent with the channel proteins being constitutively expressed and activated by Al (Ma et al., 2001). Citrate release from bread wheat is controlled by TaMATE1B, a transporter from the multidrug and toxic compound exudation (MATE) family (Ryan et al., 2009; Tovkach et al., 2013). TaMATE1B is encoded by a gene on chromosome 4BL. The greater citrate efflux is caused by a transposable element-like insertion near the transcription start site of TaMATE1B which results in a greater level of constitutive expression (Tovkach et al., 2013).

Members of these two gene families also control the release of malate and citrate from rye. Fontecha et al. (2007) identified a rye homolog of the wheat TaALMT1 gene on chromosome 7RS and showed that its expression was induced by Al to a greater degree in the resistant cultivar Ailés than the sensitive cultivar Riodeva. Quantitative trait loci (QTL) for Al resistance were subsequently linked to this same region in two separate rye populations (Benito et al., 2005; Matos et al., 2005; Collins et al., 2008; Silva-Navas et al., 2012). In one of these populations, generated from the M39A-1-6 (resistant) and M77A-1 (sensitive) haplotypes, a cluster of ScALMT genes and one ScMATE gene was located on the 7RS locus (Collins et al., 2008). The resistant parent had five copies of the ScALMT gene and expression of two of these (ScALMT1-M39.1 and ScALMT1-M39.2) was induced by Al in the root apices. By contrast, the sensitive parent had two copies of the ScALMT gene but only one (ScALMT1-M77.1) was induced by Al (Collins et al., 2008). Collins et al. (2008) was able to segregate the MATE gene from the resistance locus indicating that it was not contributing to the variation in Al resistance of that population. Those authors concluded that the ScALMT genes on 7RS controlled the Al-dependent efflux of malate from rye.

The first MATE gene in rye associated with citrate efflux from roots was the ferric reductase-like 2 gene (ScFRDL2) (Yokosho et al., 2010). The expression of ScFRDL2 in the roots was induced 15-fold by 50µM Al and closely coincided with the Al-dependent changes in citrate efflux. Another MATE gene identified in the same study, ScFRDL1, was considered unlikely to be involved in Al resistance because it was induced by iron deficiency and not by Al treatment (Yokosho et al., 2010). Silva-Navas et al. (2012) later examined the population generated from Ailés and Riodeva and mapped a MATE gene which they named aluminumactivated citrate transporter 1 (ScAACT1) in the Al-resistance QTL on chromosome 7RS. The authors proposed that ScAACT1, ScFRDL1, and ScMATE are all the same gene but this conclusion remains uncertain. For example, unlike ScAACT1, expression of ScFRDL1 was not induced by Al treatment according to Yokosho et al. (2010) and the Al resistance QTL excluded the MATE gene in the population described by Collins et al. (2008). Silva-Navas et al. (2012) argued that these inconsistencies could be explained partly by differences in the parental lines and partly by differences in the length of treatments and Al concentrations used. Whereas Yokosho et al. (2010) used 50µM Al treatments over 12 h, Silva-Navas et al. (2012) used 300µM Al treatment over 24 h. The relatedness of these MATE genes requires further clarification.

The aim of the present study was to examine how well the Al-resistance traits in the wheat and rye parental lines were transferred to the allopolyploid triticale. Two sets of diverse lines were used for this purpose. One set included octoploid triticale lines generated from an Al-resistant rye and hexaploid wheat. The second set included hexaploid triticale lines generated by crossing a durum line with rye. Measurements were made of relative Al resistance, anion efflux and expression of selected Al-resistance genes in the parental material and triticale lines.

## Genetic Material

Two sets of germplasm were used in the experiments (**Table 1**). The first set included two wheat cultivars, Carazhino and Egret, an Al-resistant rye line, **L185**, and two second generation octoploid triticale lines generated from these wheat and rye parents. The triticale lines are depicted as Carazinhox**L185** and Egretx**L185**. Carazinho is a highly Al-resistant wheat cultivar from Brazil that shows the Al-activated malate efflux controlled by TaALMT1 and the constitutive release of citrate controlled by TaMATE1B. Egret is Al-sensitive and shows little or no organic anion efflux with or without Al treatment. The second set of germplasm included a tetraploid (durum) wheat named 5020-30, an Al-resistant rye, **390**, and a closely-related but Al-sensitive rye, **389**, and the two primary hexaploid triticale lines derived from crossing these parents designated as 5020-30x**390** and 5020-30x**389**. The triticale lines were generated at the University of Hohenheim, Germany.

### Aluminum Resistance

Seeds were germinated for 2 days on moist filter paper and then planted over 20 L of aerated nutrient solution on laboratory benches. To estimate relative root length (RRL) the length of the longest root was measured before and after 4 days growth in the same nutrient solution with different Al concentrations. Therefore RRL was calculated as (net root growth in Al treatment net root growth in control solution) × 100.

#### Measurement of Citrate and Malate Efflux

The measurement of organic anion efflux from intact seedlings followed the procedures described previously (Delhaize et al., 1993; Ryan et al., 1995; Wang et al., 2007). Briefly, seeds were surface sterilized with bleach and thoroughly rinsed in sterile water. In preliminary experiments the seedlings were grown in aerated 20 L tubs with nutrient solution (pH 4.4) or in sterile conical flasks with 20 mL of 0.2 mM CaCl<sup>2</sup> (pH 4.3) on a rotary shaker and exudates collected from excised roots. The large volume of the tubs maintained the root relatively free of microbial contamination and so both growth methods gave similar exudate results. Only results from the tubs are presented here. The excised root segments (eight to twelve per replicate) were washed in small vials with 1 mL of control solution (0.2 mM CaCl2, pH 4.3) for 1 h on a platform shaker (60 rpm). The solutions were rinsed and replaced by 1 mL of treatment solution (control solution with or without 40µM AlCl3) and returned to the shaker for 2 h. After 2 h collection the malate and citrate concentrations in each solution were estimated enzymatically as described by Ryan et al. (1995). Malate assays used 0.1 mL of each sample and citrate assay used the remaining 0.9 mL. For the citrate assays the solutions were dried on a rotary vacuum drier and resuspended in 80 µL of assay solution as described by Ryan et al. (2009). All chemicals were obtained from Sigma-Aldrich Pty. Ltd. (Castle Hill, Australia). The concentrations were corrected to obtain the original malate and citrate contents in each sample and efflux was standardized for the number of apices and time of collection. In other experiments half of the seedlings were pretreated with 30µM AlCl<sup>3</sup> for at least 24 h prior to measurements as described in the figure legends.

#### Measurements of Gene Expression

RNA was extracted from the root apices with the RNeasy PlantMini Kit (Qiagen) after grinding tissues in liquid nitrogen. cDNA was synthesized with the SuperScript III First-Strand Synthesis System (Invitrogen) as recommended using 1 µg RNA of each extraction. Gene expression was determined by qRT-PCR using the SYBR Green Supermix (Bio-Rad) kit on a Bio-Rad CFX96 Real Time System. Data were analyzed with the Bio-Rad CFX Manager software. The primers selected for measuring the expression of the Al-resistance genes in wheat and rye were specific for those genes and did not hybridize with sequences in the other species. Primers for TaALMT1 expression in bread wheat were (5′ -3′ ) CGTGAAAGCAGCGGA AAGCC (fwd) and CCCTCGACTCACGGTACTAACA (rev). Primers for TaMATE1B expression in bread wheat were AGG GTGGTAGCAGTGACTTC (fwd) and GCGGCAATCACCTTC TTGTG (rev). Annealing temperatures during cycling were 67◦C for TaMATE1B and 61.5◦C for TaALMT1. The primers for measuring ScALMT1 expression in rye were GCAAACAAT ACCGTGGTTGTG (fwd) and ATCCCTCGAGTTAAGGCACC (rev). These primers could amplify products from the expressed copies of ScALMT1 in the resistant and sensitive haplotypes of rye (ScALMT1-M39.1, ScALMT1-M39.2, ScALMT1-M77.1) described by Collins et al. (2008). We measured expression of ScFRDL2 which is one of the candidate Al-resistance genes in rye because its Al-induced expression in the root apices by 50µM Al is closely correlated with the release of citrate from roots (Yokosho et al., 2010). Primers used for measuring ScFRDL2 expression were GGCTGCATTCCAGATTTGCTTG (fwd) and AGAAGCCCCAAGATCAATCCG (rev). Annealing temperatures were 68◦C for ScFRDL2 and ScALMT1.

The reference genes are important in the expression analyses because of the genetic differences between wheat and rye. Therefore the two reference genes chosen have previously been shown to be relatively stable across members of the triticaeae (Paolacci et al., 2009; Giménez et al., 2011). These gene are glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Ta30768, Genbank EF592180) with primers GTTGAGGGTTTG ATGACCAC (fwd) TCAGACTCCTCCTTGATAGC (rev) and the cell division control protein (AAA-superfamily of ATPases) (CDC; Ta54227) with primers GCCTGGTAGTCGCAGGAGAT (fwd) and ATGTCTGGCCTGTTGGTAGC (rev). In preliminary tests reliable amplicons were generated from wheat, rye and triticale with both sets of reference primers. Relative expression levels of the Al resistance genes were generally similar with both references genes and the results using CDC are presented.

#### Statistical Analysis

Al resistance was estimated by calculating relative root length (RRL) since this accounts for inherent differences in growth between different species (see above). Since RRL is a ratio of means (net root growth in different treatments) each of which has an error, then the result requires a new accumulated error. The formula for calculating this accumulated error and the procedure

#### TABLE 1 | Summary of germplasm used in this study.


† Likely mechanism from previous work but not previously investigated in these rye and triticale lines.

††These include known Al-resistance genes and candidate resistance genes. Note that some of the genes listed for citrate efflux might represent the same gene.

used for determining whether two RRL values are statistically different from one another is described previously by Zhou et al. (2013). The assumptions for this test are that the data are normally distributed and the variances are not different.

Other statistical analysis used the statistical software in SigmaPlotTM ver 14.0. Anion efflux results were analyzed with a one way ANOVA. In cases where the data failed an initial normality test the data were first transformed with the natural log function (ln). Analyses were then determined by applying the Student-Newman-Keuls method for multiple pairwise comparisons. Analysis of gene expression was similar and used three biological replicates except as stated. Note that rye was not included in the analysis of the expression of wheat genes, and conversely, wheat lines were not included in the analysis of the expression of rye genes.

#### RESULTS

Two sets of germplasm were compared for Al resistance, organic anion efflux and expression of selected Al-resistance genes. The first set was comprised of two bread wheat cultivars (Carazinho and Egret), a rye cultivar (**L185**) and the two octoploid triticale lines generated from crossing the rye to each of the wheat cultivars (Carazinhox**L185** and Egretx**L185**). Egret is an Alsensitive cultivar that shows little or no malate or citrate release. Carazinho is an Al-resistant wheat that has the Al-resistant alleles for TaALMT1 and TaMATE1B. Carazinho has greater expression of these two genes than Egret and displays an Al-activated efflux of malate and a constitutive release of citrate from the root apices (Ryan et al., 2009). The second set of germplasm included a tetraploid (durum) wheat line (5020-30), two closely-related lines of rye with contrasting resistance to Al (**390** resistant and **389** sensitive) and the two hexaploid triticale lines generated by crossing the durum wheat with each rye line (5020-30x**390** and 5020-30x**389**). The mechanisms of Al resistance in the **390** have not previously been investigated in detail.

#### Octoploid Triticale

Al resistance of the wheat, rye and octoploid triticale lines was compared by estimating relative root length after 4 d growth in a range of Al concentrations (**Figure 1**). Egret wheat was sensitive of all Al treatments while the Carazinho wheat and triticale lines were more resistant. At the highest Al treatment L185 rye was most resistant with 62% RRL while the wheat and triticale lines were similar at 25%. These results indicate that rye could contribute to the Al resistance of triticale because EgretxL185 triticale was significantly more resistant than Egret wheat. However, the Al resistance of rye and wheat was not additive in triticale because the resistance of CarazinhoxL185 was no greater than either the wheat or rye parent. It would be instructive to confirm the Al resistance measured in hydroponics reflects the measurements in field trials with acidic soil.

Malate and citrate release are known mechanisms for Al resistance in rye and wheat and fluxes of these organic anions were measured from each genotype. The results showed some variation between replicated experiments, especially in the triticale lines and rye material so the experiments were repeated several times. In the first series of experiments seedlings were grown in control nutrient solution and then malate and citrate efflux were measured in the presence or absence of Al. This means that the root tips were only exposed to Al for 2 h as exudates were collected. In the absence of Al, malate efflux from all genotypes was less than 0.05 nmol apex−<sup>1</sup> h −1 (**Figure 2**). When 40µM Al was included in the treatment solution, malate efflux increased significantly in all genotypes except for Egret wheat. The largest malate release was from Carazinho and Carazinhox**L185**. This

result is consistent with the Al-activation of malate efflux from wheat reported previously (Delhaize et al., 1993; Ryan et al., 1995) and indicates that the malate efflux trait from Carazinho wheat was fully expressed in the Carazinhox**L185** triticale. Citrate efflux from Carazinho and Carazinhox**L185** was large in the presence and absence of Al and indicates that citrate efflux was constitutive in these genotypes (**Figure 2**). Citrate efflux from Egret and Egretx**L185** was smaller regardless of Al, while efflux from rye was very variable. These results support previous observations in Carazinho and Egret and indicate that the large constitutive efflux of citrate from Carazinho was also transferred to the triticale.

Organic anion efflux in some plant species is induced by Al treatment over many hours or longer (Pellet et al., 1995; Li et al., 2000; Ma et al., 2001; Magalhaes et al., 2007; Delhaize et al., 2012). Anion release was therefore measured after pretreating the seedlings in Al. In these experiments, half the seedlings were pretreated in 30µM AlCl<sup>3</sup> for at least 24 h prior to the measurements and the other seedlings were only

exposed to Al during the 2 h collection period. The results in **Figure 3A** show that the pretreatment in Al did not affect malate efflux from any genotype. Efflux from Carazinho wheat and Carazinhox**L185** triticale remained greater than from Egret, **L185** rye and Egretx**L185** triticale. Citrate efflux from Carazinho and **L185**xCarazinho was large and unaffected by pretreatment in Al (**Figure 3B**). Citrate efflux from **L185** rye increased significantly after Al pretreatment while efflux from **L185**xEgret showed a small but significant increase following pretreatment. These results demonstrated the following: (i) the Al-activated efflux of malate and the constitutive efflux of citrate was fully transferred from Carazinho wheat to the Carazinhox**L185** triticale; (ii) Al pretreatment enhanced the efflux of citrate but not of malate from **L185** rye; (iii) the citrate efflux phenotype in rye was not fully transferred to triticale.

We next measured the expression of the ALMT genes, TaALMT1 and ScALMT1, that control malate release from wheat and rye roots. Note that the primers used for ScALMT1 recognize several copies of the ScALMT1 genes located in the Alt4 locus in rye as reported by Collins et al. (2008). Expression of the wheat gene TaALMT1 was greater in Carazinho and Carazinhox**L185** than the other genotypes and unaffected by pretreatment with Al (**Figure 4A**). TaALMT1 expression was low in Egret and

Egretx**L185** as expected. ScALMT1 expression was significantly greater in **L185** rye than the two triticale lines (**Figure 4B**). Pretreatment tended to induce expression but the difference in this experiment was not significant. These results indicate that expression of the rye ScALMT1 gene was suppressed in octoploid triticale.

The TaMATE1B and ScFRDL2 genes encode transporters that likely facilitate citrate efflux from wheat and rye respectively. TaMATE1B expression levels were high in Carazinho and Carazinhox**L185** and unaffected by pretreatment with Al (**Figure 5A**). Little or no expression was detected in Egret and Egretx**L185**. These data indicate that TaMATE1B was expressed similarly in wheat and triticale. ScFRDL2 expression was detected in **L185**, Egretx**L185** but it was suppressed in Carazinhox**L185** (**Figure 5B**). These data indicate that the rye ScFRDL2 gene was expressed in triticale but the level of expression varied with the different wheat parents.

#### Hexaploid Triticale

The second set of experiments examined the rye lines **390** and **389**, a tetraploid (durum) wheat line (5020-30) and the two hexaploid triticale lines generated from crossing each rye line to the durum wheat (5020-30x**390** and 5020-30x**389**).

The **390** rye is resistant to Al and **389** is closely related but more sensitive to Al. The Al resistance of these lines was compared by estimating relative root length after 4 d growth in 0, 15, and 60µM Al (**Figure 6**). Rye **390** showed no inhibition of root growth at 60µM Al which is consistent with it being the most resistant genotype. RRL for most other genotypes was 20% or less for all treatments. The exception was 5020-30x**390** triticale where RRL was ∼50% in 15µM Al (**Figure 6B**). These data indicate that the Al resistance of the **390** rye was incompletely transferred to hexaploid triticale.

Malate and citrate efflux from these genotypes was measured with or without a pretreatment in 30µM Al (**Figure 7**). Malate efflux from the Al-resistant rye **390** was induced by Al pretreatment and was five-fold greater than the other genotypes (**Figure 7A**). Citrate efflux from the resistant **390** rye showed a large induction by Al pretreatment while efflux from the 5020-30x**390** triticale showed a smaller induction reaching only ∼30% of the rye (**Figure 7B**). Efflux from the other genotypes remained small. These results show that malate and citrate efflux likely contribute to the Al resistance of the **390** rye. They also indicate that the malate efflux detected in the **390** rye was not transferred to the hexaploid triticale 5020-30x**390** while citrate efflux was only partially transferred to triticale.

Expression of the rye genes contributing to malate and citrate, ScALMT1 and ScFRDL2 respectively, were then measured with and without pretreatment in Al (**Figure 8A**). Without a pretreatment, ScALMT1 expression was low in all lines. After a pretreatment in Al, ScALMT1 expression in **390** rye increased 10-fold but was not induced in any of the other lines. These responses are consistent with the measured efflux of malate. The expression of ScFRDL2 was significantly increased by Al pretreatment in **390** rye and 5020-30x**390** triticale but remained lower in the other lines (**Figure 8B**). These results indicate that the ScFRDL2 gene was induced by Al in the Al-resistant **390** rye and the 5020-30x**390** triticale but that expression of ScALMT1 was suppressed in the 5020-30x**390** triticale.

#### DISCUSSION

Al-resistance in hexaploid wheat and rye relies on the efflux of malate and citrate anions from the root apices. These phenotypes are controlled in part by the TaALMT1 and TaMATE1B genes in hexaploid wheat and by the ScALMT and ScFRDL2 genes in rye. This study investigated the transfer of these resistance mechanisms from wheat and rye lines to triticale. The first

set of germplasm examined included Al-sensitive (Egret) and resistant (Carazinho) hexaploid wheat cultivars, an Al-resistant rye (**L185**) and the two octoploid triticale lines generated by crossing each wheat with the rye. All of these lines except for the Egret wheat showed strong resistance to Al toxicity (**Figure 1**). The following conclusions can be made from these first set of lines: (1) The wheat and rye parents both contributed to the Al-resistance of octoploid triticale. Support for this conclusion comes from the finding that Egretx**L185** triticale was significantly more Al-resistant than Egret (**Figure 1**) which indicates that **L185** contributed to the phenotype. Further, the malate and citrate efflux in Carazinhox**L185** resembled the responses in Carazinho wheat but not that of rye (**Figures 2**, **3**) indicating that wheat contributed to those traits. These responses generally reflected the relative expression of the genes involved. (2) Al resistance of the parental lines was not additive in triticale. This supported by the finding that neither triticale line was more resistant than the rye or Al-resistant wheat parents. (3) Function of the Al resistance genes in hexaploid wheat were more completely transferred to triticale than the rye genes. This is shown by the expression levels of the two wheat genes TaALMT1 and TaMATE1B which were similar in Carazinho and Carazinhox**L185** whereas expression of the rye genes, ScALMT and ScFRDL2, in triticale was inconsistent (**Figures 4**, **5**). In a

previous study, Stass et al. (2008) concluded that the Al resistance of triticale was determined by citrate efflux which was largely controlled by the wheat parent. The present results indicate that malate efflux from hexaploid wheat can also contribute to the resistance of triticale. (4) The expression of the rye ScFRDL2 gene in triticale depended on the genotype of the wheat parent. This is supported by the observations that relative expression of ScFRDL2 was significantly greater in Egretx**L185** than Carazinhox**L185** (**Figure 5**). It is interesting to speculate whether this finding is related to the different expression levels of the wheat gene TaMATE1B in Carazinho and Egret.

The second set of germplasm included a durum wheat (5020- 30), a pair of closely-related rye lines (**389** and **390**) that differed in Al resistance and the two hexaploid triticale lines generated by crossing each rye with the durum. The main conclusions drawn from those results include the following: (1) Malate and citrate efflux contribute to the Al resistance of **390** rye and these fluxes were correlated with increases in ScALMT1 and ScFRDL2 expression in **390** rye. (2) The Al resistance of 390 rye was not fully transferred to the 5020-30 x 390 triticale (**Figure 6**). This was consistent with the reduced efflux of organic anions from 5020-30 x 390 compared to the resistant rye (**Figure 8**). The Al resistance of these lines appeared to be most closely correlated with citrate efflux and expression of ScFRDL2.

The important observation from both sets of lines was that the Al resistance of rye was incompletely transferred to triticale—whether to a hexaploid or a octoploid triticale. Similar observations have been made previously for other rye genes in triticale (Neves et al., 1995; Kalinka and Achrem, 2018). By contrast, the Al resistance traits from hexaploid wheat did transfer more successfully to octoploid triticale.

The incomplete transfer of the Al-resistance traits of rye to triticale may be explained by the modifications that commonly occur to the genome of de novo allopolyploids mentioned above. When related species such as wheat and rye hybridize to form a stable allopolyploid many genes become duplicated and genetic changes that occur can affect gene expression. Furthermore, homeolog copies of all genes in allopolyploids are not expressed equally. Sequences can be lost and mutations generated due to chromosomal rearrangements or transposon activity (Ma and Gustafson, 2008), and gene transcription can be affected by epigenetic modifications and microRNAs (Cheng and Murata, 2002; Kashkush et al., 2002, 2003; Kraitshstein et al., 2010; Li et al., 2014, 2015; Kalinka and Achrem, 2018). Cytosine residues in DNA are prone to methylation when they occur as CpG, CpHpG, and CpHpH sites (where H represents any nucleotide except guanine) and methylation of gene promoter regions can interfere with transcription. In de novo allopolyploids, such as the primary triticale lines used here, DNA methylation appears to be a more important factor decreasing gene expression than genetic instability with some estimates suggesting 1 to 12% of genes are silenced this way (Kashkush et al., 2002; He et al., 2003; Mochida et al., 2004; Bottley et al., 2006). The homeologous genes from one parent genome in de novo allopolyploids can be silenced more than the other parent and this can even vary between different organs (Bottley et al., 2006; Zhao et al., 2011).

We propose that the rye traits are incompletely transferred to triticale because its genome is naive to the polyploid environment and therefore more prone to epigenetic modification. Hexaploid wheat, by contrast, has emerged from two major hybridization events. The first hybridization occurred about 0.5 million years ago between a diploid species (likely Triticum uratu, AA genome) and another unknown parent with the BB genome which generated an ancestral tetraploid wheat. The second event occurred only 10,000 years ago between a tetraploid species such as Triticum turgidum (BBAA) and the diploid grass Aegilops tauschii (DD) and generated hexaploid wheat. Since wheat has been subject to epigenetic silencing pressure for a long period, its genome is likely to be more resistant to further silencing processes than the rye genome in a wheat-rye hybrid. This idea is consistent with the outcome of previous studies that compared the expression of genes in tetraploid and diploid lines with their expression in de novo hexaploid wheat lines. For example, microarray analysis and RNA-seq techniques demonstrated that more genes from the diploid parent had reduced expression levels in the hexaploid line than the tetraploid parent. This indicates an "expression bias" toward the tetraploid genome parent compared to the diploid genome (Akhunova et al., 2010; Li et al., 2014). Future work will test this hypothesis by investigating how methylation states of specific Al-resistance genes in the wheat and rye parents change in the primary triticale lines.

Significant variation was detected between certain repeated experiments and in some instances anion fluxes did not correlate well with gene expression. For example, the relative expression of ScALMT1 in **L185** and Egretx**L185** did not reflect the measured fluxes of malate. This variation could be related, in part, to the variable delay in gene induction by Al and the possible involvement of other Al resistance genes not targeted in this study. More than one ALMT and MATE gene could contribute to anion efflux (see Introduction). The variation may also be related to the instability of the primary triticale lines which can continue in subsequent generations (Ma and Gustafson, 2008; Kalinka and Achrem, 2018). The grain used in these experiments were bulked on two occasions so some variation was not unexpected. However each experiment was performed several times and the results presented here reflect the same general trends. Future studies could, nevertheless, quantify the stability of the triticale lines by

#### REFERENCES


determining the chromosome number of individual plants within each line and in different generations. The expression levels of target genes could also be measured and Al-resistance assessed in field trials on acidic and limed soils over several sites and seasons. These experiments would provide further insight into the genetic stability of the primary triticale material.

Triticale was developed to combine the favorable attributes of rye and wheat and to generate diversity. The triticale material used in this study were primary triticale lines generated by crossing rye pollen to female wheat plants. The reverse cross (rye as the female parent) is possible but less successful. Nevertheless, since the mitochondrial and plastid genomes are maternally inherited some traits are under cytoplasmic control (Thiede, 1998; Battich et al., 2015). Additional diversity might be generated if improved technologies enabled the reverse crosses to occur more efficiently so that the origins of the mitochondrial and plastid genomes in triticale would be rye instead of wheat.

Most triticale grown around the world are hexaploid types because they tend to show better vigor and stability (Mergoum et al., 2009). The present study found that the strong Alresistance of rye was not fully expressed in either the hexaploid or octoploid triticale whereas the Al-resistance traits derived from Carazinho hexaploid wheat did transfer to octoploid triticale more successfully. If this pattern is indicative of other phenotypes then hexaploid triticale is potentially missing other valuable traits that may occur in some hexaploid wheat but not tetraploid wheat. Consideration should be given to developing cytological techniques that improve the genetic stability of octoploid triticale so that beneficial traits of hexaploid wheat (e.g., flour quality, nutrient content) can be captured in the hybrid. Such an approach could further improve the value of triticale production.

#### AUTHOR CONTRIBUTIONS

WH and PR conceived the project and coordinated with the breeder to access the germplasm. Experimental word was performed by DD, NW, JL, MX, NSM, JY, ED and PR. Supervision of experimental work was provided by PR, ED and KM. The manuscript was drafted by PR with help from all authors.

#### ACKNOWLEDGMENTS

The authors are grateful to the critical contribution from H-PM for generating and supplying the germplasm. We acknowledge FT for providing technical assistance and Anthony Condon for critically reading the manuscript.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Ryan, Dong, Teuber, Wendler, Mühling, Liu, Xu, Salvador Moreno, You, Maurer, Horst and Delhaize. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Emerging Pleiotropic Mechanisms Underlying Aluminum Resistance and Phosphorus Acquisition on Acidic Soils

Jurandir V. Magalhaes1,2 \*, Miguel A. Piñeros<sup>3</sup> \*, Laiane S. Maciel1,2 and Leon V. Kochian<sup>4</sup>

<sup>1</sup> Embrapa Maize and Sorghum, Sete Lagoas, Brazil, <sup>2</sup> Departamento de Biologia Geral, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil, <sup>3</sup> Robert W. Holley Center for Agriculture and Health, USDA-ARS, Cornell University, Ithaca, NY, United States, <sup>4</sup> Global Institute for Food Security, University of Saskatchewan, Saskatoon, SK, Canada

#### Edited by:

Peter Ryan, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia

#### Reviewed by:

Takayuki Sasaki, Okayama University, Japan Manny Delhaize, Plant Industry (CSIRO), Australia

#### \*Correspondence:

Jurandir V. Magalhaes jurandir.magalhaes@embrapa.br Miguel A. Piñeros map25@cornell.edu

#### Specialty section:

This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science

Received: 01 March 2018 Accepted: 06 September 2018 Published: 26 September 2018

#### Citation:

Magalhaes JV, Piñeros MA, Maciel LS and Kochian LV (2018) Emerging Pleiotropic Mechanisms Underlying Aluminum Resistance and Phosphorus Acquisition on Acidic Soils. Front. Plant Sci. 9:1420. doi: 10.3389/fpls.2018.01420 Aluminum (Al) toxicity on acidic soils significantly damages plant roots and inhibits root growth. Hence, crops intoxicated by Al become more sensitive to drought stress and mineral nutrient deficiencies, particularly phosphorus (P) deficiency, which is highly unavailable on tropical soils. Advances in our understanding of the physiological and genetic mechanisms that govern plant Al resistance have led to the identification of Al resistance genes, both in model systems and in crop species. It has long been known that Al resistance has a beneficial effect on crop adaptation to acidic soils. This positive effect happens because the root systems of Al resistant plants show better development in the presence of soil ionic Al3<sup>+</sup> and are, consequently, more efficient in absorbing sub-soil water and mineral nutrients. This effect of Al resistance on crop production, by itself, warrants intensified efforts to develop and implement, on a breeding scale, modern selection strategies to profit from the knowledge of the molecular determinants of plant Al resistance. Recent studies now suggest that Al resistance can exert pleiotropic effects on P acquisition, potentially expanding the role of Al resistance on crop adaptation to acidic soils. This appears to occur via both organic acid (OA)- and non-OA transporters governing a joint, iron-dependent interplay between Al resistance and enhanced P uptake, via changes in root system architecture. Current research suggests this interplay to be part of a P stress response, suggesting that this mechanism could have evolved in crop species to improve adaptation to acidic soils. Should this pleiotropism prove functional in crop species grown on acidic soils, molecular breeding based on Al resistance genes may have a much broader impact on crop performance than previously anticipated. To explore this possibility, here we review the components of this putative effect of Al resistance genes on P stress responses and P nutrition to provide the foundation necessary to discuss the recent evidence suggesting pleiotropy as a genetic linkage between Al resistance and P efficiency. We conclude by exploring what may be needed to enhance the utilization of Al resistance genes to improve crop production on acidic soils.

Keywords: abiotic stress resistance, transporters, plant breeding, pleiotropy, aluminum tolerance, phosphorus acquisition, phosphorus efficiency

## INTRODUCTION

fpls-09-01420 September 24, 2018 Time: 15:52 # 2

Acidic soils are globally widespread, extending to more than half of the world arable lands (von Uexküll and Mutert, 1995). These low-pH soils, which are commonly found in tropical and subtropical regions, include areas where food production needs to be increased to cope with a continuously growing population (Godfray et al., 2010). For example, there have been a number of studies in the literature addressing the extent of Al-toxic acidic soils in Africa, with approximately 25% of the soils being highly acidic (FAO and ITPS, 2015; Tully et al., 2015). Two of the major constraints for crop production on acidic soils, including those in Sub-Saharan Africa (Doumbia et al., 1993, 1998), are aluminum (Al) toxicity and low phosphorus (P) availability.

Aluminum and iron (Fe) oxides, which are enriched in the clay fraction of acidic soils upon intensive weathering of primary minerals (Shaw, 2001), drive both types of abiotic stresses, causing a general coincidental occurrence of Al toxicity and low P availability on tropical soils (Sanches and Salinas, 1981). Phosphorus forms strong, covalent bounds with these oxides, becoming highly unavailable for uptake by the plants (Marschner, 1995; Lynch, 2011), due to restricted P diffusive fluxes from the soil toward the root surface. In addition, P diffusion on highly weathered soils is highly dependent on the soil water content (Novais and Smith, 1999), which varies during the crop season, making P supply to the plant and, consequently, P uptake, highly discontinuous. Under low pH, Al present in aluminosilicates and oxides is released as the rhizotoxic Al3<sup>+</sup> ion into the soil solution, damaging the root system and inhibiting root growth (Delhaize and Ryan, 1995).

Aluminum resistance has long been associated with overall crop adaptation to acidic soils by indirectly enhancing mineral nutrient uptake and drought resistance (Foy et al., 1993). Accordingly, undamaged, "Al resistant" root systems are more effective in absorbing sub-soil water, and nutrients, particularly those that are highly unavailable on acidic soils, such as P. It is important to note that Al toxicity typically extends to sub-soil layers, where liming is highly ineffective in increasing soil pH, enhancing the deleterious effects of drought stress in reducing crop yields.

The widespread nature of Al toxicity and its global impact has spurred extensive research on the physiological, genetic, and molecular mechanisms that enable crops to withstand Al toxicity on acidic soils. Clearly, impressive progress has been made in the last two decades on the molecular underpinnings of crop Al resistance (reviewed by Kochian et al., 2015). These discoveries led to the isolation of a number of the previously anonymous molecular determinants of Al resistance in loci that had been identified previously via genetic mapping in crops such as wheat, barley, rye, sorghum, and maize, as well as in model systems such as in Arabidopsis thaliana.

It is reasonable to expect that the identification of the molecular drivers of plant Al resistance can be instrumental in the development of novel strategies for improving crop performance on acidic soils in a more efficient way. Marker-assisted backcross to improve Al resistance based on single major loci has been a feasible approach long before major Al resistance genes were cloned. Beyond that, these genes now offer opportunities for large scale germplasm screening approaches based on functional markers, which can streamline the utilization of large germplasm banks in favor of plant breeding (Tanksley and McCouch, 1997; Hufnagel et al., 2018). Most importantly, it is possible that the value of Al resistance for crop production in the context of the multiple stress scenario on acidic soil regions (Bahia Filho et al., 1997) has been somewhat underappreciated. Some possible reasons for that are the lack of systematic efforts to map Al saturation both in the surface and below ground soils and a rather incomplete quantification of the grain yield effect of known Al resistance genes in soil, which is to some extent understandable due to the highly complex chemical nature of acidic soils.

There is now an interesting body of emerging evidence suggesting that Al resistance genes may have an additional, pleiotropic effect on acidic soils, which involves enhancement of P acquisition. In conjunction with the known effect of Al resistance in enhancing water and mineral uptake, by promoting better root growth on acidic soils, this would further justify deliberate efforts to design novel, gene-based molecular breeding strategies aimed at developing cultivars adapted to acidic soil regions. These strategies can help in realizing the great potential there is in expanding the world's agricultural frontier, by exploring the vast areas under acidic soils in the tropics and subtropics, which show in general a favorable topography for agriculture (Sanches and Salinas, 1981).

Here, our objective is not to review the current available information on plant Al resistance or P efficiency, which is defined here as improved performance in soils with low P availability. For that, readers are directed to many available comprehensive reviews (Delhaize and Ryan, 1995; Kochian, 1995; Ma et al., 2001; Kochian et al., 2004; Delhaize et al., 2007, 2012; López-Arredondo et al., 2014; Eekhout et al., 2017). Our goal here is to explore the emerging connections between Al resistance genes and P deficiency responses that help maintain favorable P nutrition, which happens possibly via alterations in root system architecture. We recognize these studies are just emerging and are still found largely in the realm of model species, in this case, Arabidopsis. This makes some of the croprelated implications drawn in this paper somewhat speculative in nature. However, due to the efficacy and breeding potential of common mechanisms underlying two important abiotic stress factors on acidic soils, taking advantage of the convergence of Al resistance and P efficiency via pleiotropic genes could have a significant impact in enhancing global food security. In the next section, we will briefly review the components comprising mechanisms that might jointly control Al resistance and P nutrition. We will then explore the emerging, underlying basis for such pleiotropy and will close with a brief discussion of the future directions to further explore Al resistance genes as tools to improve P acquisition and crop performance on acidic soils.

#### OVERVIEW OF POSSIBLE PLEIOTROPIC MECHANISMS CONTROLLING BOTH Al RESISTANCE AND ROOT TRAITS THAT MAY LEAD TO ENHANCED PHOSPHORUS ACQUISITION UNDER LOW P CONDITIONS

#### Physiological Basis

fpls-09-01420 September 24, 2018 Time: 15:52 # 3

The ability of a plant to tolerate low P availability in the soil may be achieved both by internal mechanisms, acting to optimize the way plants internally utilize phosphorus, and by mechanisms to improve phosphorus acquisition from the soil. Mendes et al. (2014) genetically assessed the contribution of those mechanisms in maize grown on a tropical soil with low P availability and found that 80% of the QTLs mapped for P acquisition efficiency co-localized with those for P use efficiency (i.e., the ratio between grain yield and the amount of P supplied to the crop), indicating that the efficiency in acquiring P is the main determinant of P use efficiency in tropical maize. Since P acquisition efficiency achieved via changes in root morphology is the physiological basis of possible pleiotropy between Al resistance and better P nutrition, here we will briefly discuss this mechanism. For a broader view of mechanisms possibly contributing to enhanced crop performance under low P, which may involve modulation of P transporters, root system architecture modifications in response to low P, exudation of organic acids (OAs) and phosphatases, and mycorrhizal associations, in addition to internal mechanisms of P efficiency, readers are directed to recent reviews in this area (e.g., López-Arredondo et al., 2014).

Since P is in general highly unavailable on acidic soils, results such as those reported by Mendes et al. (2014) are expected, as enhanced capacity to acquire P is the logical first limiting step for P efficiency. However, other mechanisms have also been shown to exert beneficial effects on crop performance under low P in the field (López-Arredondo et al., 2014). The work by Gamuyao et al. (2012) provided a molecular foundation for the importance of root system architecture on the efficiency with which plants acquire P on soils with low P availability. The rice serine/threonine receptor-like kinase, OsPSTOL1, which is a member of the LRK10L-2 subfamily, was shown to enhance early root growth and grain yield on a P-deficient soil via increased P uptake, regulating crown root development (Gamuyao et al., 2012). Subsequently, a low but positive correlation between root surface area assessed in younger plants and grain yield under low P was instrumental in the identification of sorghum homologs of OsPSTOL1, designated SbPSTOL1 genes, that also act to enhance root growth, thereby leading to enhanced P acquisition and grain yield in a sorghum association panel (Hufnagel et al., 2014). Mechanistically, plant P deficiency leads to inhibition of primary root growth due to a shift from an indeterminate to a determinate developmental program, which is caused by reduced cell elongation followed by the loss of meristematic cells in the root apical meristem (RAM) (Sánchez-Calderón et al., 2005). Hence, this release of apical dominance leads to enhanced proliferation of lateral roots, and increased lateral root branching increasing P uptake as observed in maize (Zhu and Lynch, 2004; Postma et al., 2014).

From the physicochemical standpoint, the supply of a nutrient like P from the soil solution toward the root surface via a diffusive flow can be modeled by the Fick<sup>0</sup> s law (Nobel, 1991), which depends on the P concentration gradient generated by the interplay between root P absorption and P in the soil solution. This concentration gradient can thus be thought as the "force" driving diffusion fluxes; as the root system grows into new soil regions still rich in P, the distance through which diffusion occurs is reduced, thus enhancing the diffusive flow (Novais and Smith, 1999), which is also maintained by the uptake process. Finally, we point out that changes in the three-dimensional configuration of the root system, such as proliferation of shallow roots, can also enhance P uptake [for more details on such mechanisms, please see Li et al. (2016) and Lynch (2011)].

### Molecular Basis Malate and Citrate Transporters

Organic acid transport and homeostasis is emerging as a central hub in a network of acidic soil stress responses. The first OA transporters involved in Al resistance were the wheat TaALMT1 and Arabidopsis AtALMT1, both shown to encode plasma membrane anion channel proteins that mediate root tip malate efflux (Sasaki et al., 2004; Hoekenga et al., 2006; Piñeros et al., 2008; Zhang et al., 2008). Although being the founding members of a novel class of plant anion transporters, it is now well established that, as a family, ALMT functions extend well beyond Al resistance, and participate in a variety of other physiological processes, including guard cell regulation, fruit quality, anion homeostasis, seed development, and plant–microbe interactions (Sharma et al., 2016). However, electrophysiological analysis of TaALMT1 and AtALMT1 (i.e., those transporters associated with Al-dependent responses) in heterologous systems has shown a distinct functional feature of these two transporters in that although they have transport activity in the absence of extracellular Al3+, this activity is enhanced by extracellular Al3<sup>+</sup> (Hoekenga et al., 2006; Piñeros et al., 2008). This so-called "Al activation" is analogous to processes occurring in ligand-gated channels, with the agonistic binding of Al3<sup>+</sup> to the ALMT protein triggering a conformational change that favors its open state, consequently increasing its transport activity and facilitating anion (i.e., malate) flux. Although the molecular determinants involved in the binding of Al3<sup>+</sup> to the ALMT protein remain unknown, a combination of functional analysis of structurally modified TaALMT1 and AtALMT proteins and phylogenetic studies on ALMTs indicate that several different domains in these two proteins are likely to act together in the Al-mediated enhancement of transport activity (Sasaki et al., 2004; Furuichi et al., 2010; Ligaba et al., 2013). Overall, the Al-dependent enhancement of the transport activity of an anion channel mediating the selective efflux of malate represents an elegant regulatory component of root malate exudation associated with Al exclusion processes.

More recently, a second novel transport substrate and new regulatory mechanisms have been described for the TaALMT1 transporter (Ramesh et al., 2015, 2018). It has generally been assumed that malate efflux is the primary transport function associated with TaALMT1. Recently, it was shown that TaALMT1 also has a high permeability to the non-protein amino acid, gamma-aminobutyric acid (GABA), a zwitterion molecule associated with signaling cascades in plants. GABA is not only transported by TaALMT1 but also modulates the activity of the transporter protein. Similarly, the apoplastic pH and anion composition also appear to regulate TaALMT1 transport activity, such that increased anion concentrations and/or more alkaline apoplastic conditions stimulate transport activity (Ramesh et al., 2015). These functional characteristics provide additional regulatory layers to Al3+-mediated regulation of TaALMT1 activity. Consequently, in alkaline environments, enhancement of TaALMT1 activity resulting in both malate and GABA efflux has been suggested by Ramesh et al. (2015) to promote extracellular acidification via H<sup>+</sup> efflux coupled to the efflux of the malate anion, thereby potentially ameliorating and providing tolerance to high pH soils. Verification of such a tolerance mechanism operating in response to alkaline environments, and validation of the tantalizing functional plasticity of TaALMT1 in tolerance to abiotic stresses, awaits further investigation. It should be noted that the initial studies on this topic have not found increased tolerance or malate efflux in plants grown on alkaline soils and hydroponic media simulating alkaline field conditions (Silva et al., 2018).

The second type of Al resistance OA transporters belong to a subgroup of plasma membrane-localized MATE transporters identified from the map-based cloning of the major Al resistance loci in sorghum (SbMATE) (Magalhaes et al., 2004, 2007) and barley (HvAACT1) (Furukawa et al., 2007; Wang et al., 2007). Functional characterization of SbMATE, HvAACT1, and subsequently identified homologs in Arabidopsis (AtMATE1) (Liu et al., 2009), maize (ZmMATE1) (Maron et al., 2009), wheat (Ryan et al., 2009; Tovkach et al., 2013), rice bean (VuMATE1/2) (Yang et al., 2011; Liu et al., 2018), and rice (OsFRD2/4) (Yokosho et al., 2011, 2016) indicates that this subgroup of MATE transporters mediate citrate transport, and therefore as with ALMTs, these transporters underlie Al-exclusion via root tip OA root release. However, it is worthwhile to comment about the common assumption that ALMTs and MATEs are functionally very similar, as this is not the case. The functional analysis of several of the MATE transporters involved in Al resistance has established that, when expressed in heterologous systems, this subgroup of MATE transporters mediates constitutive pHdependent citrate transport that is not activated by Al3<sup>+</sup> in Xenopus oocytes (Magalhaes et al., 2007; Maron et al., 2009; Yang et al., 2011; Melo et al., 2013; Doshi et al., 2017; Liu et al., 2018), although some exceptions have been also reported both in X. oocytes (Furukawa et al., 2007; Yokosho et al., 2011) and tobacco suspension cells (Yokosho et al., 2016). Electrophysiological analysis indicates that, in the absence of exogenous intracellular citrate, these MATE transporters mediate an electrogenic transport that appears to be due to a large cation influx (H+, Na+, and/or K+). Differences in the OA transport mechanism between ALMTs and MATEs raises interesting questions. Because of the large inwardly directed voltage gradient or membrane potential across the root cell plasma membrane, the efflux of the malate and citrate anions is a thermodynamically passive process. This is consistent with the ALMT transporters functioning as anion channels mediating the passive movement of the malate anion out of the root cell.

On the other hand, the MATE transporters use a thermodynamically active (H+-driven) antiport mechanism associated with the passive efflux of citrate2<sup>−</sup> anions down its outwardly directed electrochemical gradient. One interesting and quite speculative explanation for this is that an alternative substrate, rather than the free citrate2<sup>−</sup> anion, is the substrate being transported out of the root cells. In the recent publication by Doshi et al. (2017), electrophysiological, radiolabeled, and fluorescence-based transport assays in two heterologous expression systems (oocytes and yeast) demonstrated that SbMATE has a fairly broad substrate recognition, mediating proton and/or sodium-driven efflux of the <sup>14</sup>C-citrate anion, as well as efflux of the organic monovalent cation, ethidium, but not its divalent analog, propidium.

Consistent with those findings, MATE proteins were found to transport a wide range of organic substrates (Omote et al., 2006), both anionic and cationic (Tanihara et al., 2007), and including ethidium in the case of the first characterized MATE family protein, the bacterial MATE, NorM (Morita et al., 2000). Nevertheless, it was somewhat surprising to the field of MATE researchers when it was discovered that the plant MATEs involved in Al resistance mediate the efflux of the anion, citrate. Thus, the findings in the recent Doshi et al. publication showing that at least SbMATE has a more broad transport substrate recognition allows us to very speculatively propose that SbMATE (and its orthologs) mediate the efflux of a complexed rather than free anionic form of citrate. This alternative could help explain the antiporter nature of these MATE transporters, as Al–citrate complexes, for instance, could actively be removed from the symplasm in a process energized by passive H<sup>+</sup> influx. Under this scenario, this group of MATE transporters would still mediate an Al resistance response by actively removing and detoxifying Al from the symplasm of root cells (i.e., mediating resistance), rather than mediating a process where Al is prevented from entering the root cell.

#### Malate and Citrate Transporters as Part of a Common Stress-Responsive Hub

Transcription factors including the Cys2His2-type zinc finger transcription factors OsART1 in rice (Yamaji et al., 2009) and AtSTOP1 and 2 (Sawaki et al., 2009), AtWRKY46 (Ding et al., 2013) in Arabidopsis, and the rice ASR (abscisic acidic, stress, and ripening) 1 and 5 (Arenhart et al., 2013, 2016; Lima et al., 2011), are involved with the regulation of membrane transporter genes. OsART1, an AtSTOP1 ortholog, modulates the expression of a number of membrane transporters involved in rice Al resistance, OsNrat1, OsMGT1, and OsFRDL4 (Xia et al., 2010; Yokosho et al., 2011; Chen et al., 2013). Similarly, AtSTOP1 modulates the expression of membrane transporters associated with Al resistance including AtALMT1, AtMATE1,

and AtALS3 (Liu et al., 2009; Sawaki et al., 2009), in response to both Al and H<sup>+</sup> rhizotoxicity. Recently, as discussed in the next sections, changes in AtSTOP1 regulation of AtALMT1 have been shown to constitute a major component of P sensing pathways (Balzergue et al., 2017; Mora-Macías et al., 2017). Likewise, expression of AtALMT1 is also regulated by other signaling pathways involving reactive oxygen species (ROS) and phytohormones (Daspute et al., 2017). Biotic stresses, such as that caused by infection of shoots by pathogenic Pseudomonas syringae, also triggered upregulation of AtALMT1 expression and increased root malate exudation, which attracts the beneficial rhizobacterium, Bacillus subtilis, into the root microbiome and stimulates Arabidopsis immune responses (Rudrappa et al., 2008). Overall, these more recent observations indicate that the regulatory role of AtSTOP1 on AtALMT1 expression and associated physiological stress responses extend well beyond the original signaling roles associated with Al and H<sup>+</sup> stress.

#### Al Resistance Transporters That Do Not Transport Organic Acids: Aluminum-Sensitive 3 (ALS3)

Screening for Arabidopsis mutants with altered responses to Al toxicity led to the identification of mutants with increased sensitivity to Al, within which the recessive Al sensitive mutant, als3, showed 80% root growth inhibition by Al compared to 24–38% inhibition in the wild type (Larsen et al., 1996). This Al sensitive response was unrelated to enhanced Al uptake by als3 plants (Larsen et al., 1997). Subsequently, map-based cloning identified ALS3 as an ABC transporter-like protein that is localized to leaf hydathodes and the phloem, in addition to the root cortex (Larsen et al., 2005). Based on its likely plasma membrane localization, it was suggested that ALS3 functions in an Al-specific manner to move Al away from sensitive tissues, thus providing Al resistance. ABC transporters contain both a nucleotide (ATP)-binding domain and a transmembrane (TM) domain (Rea, 2007). Larsen and colleagues noted that both ALS3 and the homologous putative bacterial metal resistance protein, ybbM, do not possess the ATP binding domain, which is normally needed for ABC transporters to function.

The ABC transporter, sensitive to Al rhizotoxicity (AtSTAR1), which possesses only the ATP-binding domain and not the TM domain, was implicated in Al resistance in Arabidopsis (Huang et al., 2010). AtSTAR1 is a homolog of rice OsSTAR1. Huang et al. (2009) showed that OsSTAR1 (which contains the nucleotidebinding domain) forms an ABC complex with OsSTAR2 (which contains the TM domain), which results in an active ABC transporter involved in Al resistance possibly by mediating UDP glucose efflux into the rice root cell wall. The actual mechanism whereby this activated form of glucose may provide Al tolerance still remains to be elucidated. However, Huang and collaborators hypothesize that UDP glucose may be transported by membranelocalized STAR1–STAR2 from the cytosol into vesicles, from which either UDP-glucose or derived glycoside would be released into the apoplast via exocytosis across the plasma membrane, and used to mask the sites for Al binding in the cell wall, thus providing Al resistance. In Arabidopsis, Huang et al. (2010) presented findings suggesting that AtSTAR1 may form a complex with ALS3, with ALS3 providing the TM domain enabling the formation of a functional AtSTAR1/ALS3 complex, which may mediate Al efflux from the outer cell layers of the root tip. These findings indicate that Arabidopsis Al resistance is complex, and also include AtALMT1 (Hoekenga et al., 2006) and AtMATE (Liu et al., 2009) providing root Al exclusion via root malate and citrate efflux. In addition to ALS3, a number of other putative Al transporters have been identified that could mediate Al resistance. These include OsNrat1, a rice root plasma membrane uptake transporter that ultimately results in Al storage in the root vacuole (Xia et al., 2010), AtNIP1, a root tip plasma membrane aquaporin protein that mediates root Al uptake (as an Al–malate complex) and sequestration (Wang et al., 2017), and another Arabidopsis ABC transporter, ALS1 (Larsen et al., 2007; Nezames et al., 2012).

Research based on suppressor screens have focused on the identification of molecular factors in the form of mutations that could complement the Al-sensitive phenotype of als3 (Gabrielson et al., 2006). These studies implicated DNA damage as a biochemical target of Al (Rounds and Larsen, 2008; Nezames et al., 2012; Sjogren et al., 2015; Sjogren and Larsen, 2017), which is viewed as a possible venue to enhance crop Al resistance (Eekhout et al., 2017). One component is the cell cycle checkpoint factor, ALUMINUM TOLERANT2 (ALT2), which may recruit members of the machinery involved with the detection and repair of DNA damage elicited by Al toxicity (Nezames et al., 2012). Accordingly, it was proposed that ALT2, and also ataxia telangiectasia-mutated and Rad3-related (ATR), impair the cell cycle and drive quiescent center differentiation in response to DNA damage caused by Al, leading to root growth arrest elicited by Al. It will be very interesting to assess the effect of the molecular factors involved with the biochemical targets of Al toxicity, such as DNA damage, in enhancing crop performance on acidic soils. Genetic manipulation of the underlying factors for Al toxicity is thought to hold potential for increasing global food security on acidic soils (Rounds and Larsen, 2008). Within the realm of natural variation for Al resistance in crop plants, the allelic effects of such factors may prove to be milder compared to that of major Al resistance genes encoding plasma membrane transporters. Nevertheless, exploiting such distinct biochemical pathways in concert, in the context of plant breeding, may offer potential for identifying transgressive segregants that could enhance even further crop perform on acidic soils.

## POSSIBLE PLEIOTROPIC EFFECTS UNDERLYING Al RESISTANCE AND P ACQUISITION EFFICIENCY

## SbMATE and TaALMT1 Increase Grain Yield on Al-Toxic and P-Deficient Soils

Overexpression of the wheat Al resistance gene, TaALMT1, in transgenic barley under the control of the ubiquitin promoter has been shown to enhance both P uptake and grain production on an acidic, high P-fixing soil (Delhaize et al., 2009). This effect was attributed in large part to the role of TaALMT1 in maintaining root growth under soil acidity, which likely results

from Al resistance. However, the observed greater P uptake per unit length in TaALMT1-expressing barley lines might also have resulted to some extent from P mobilization from the soil clays by the malate released into the rhizosphere, thus favoring P uptake (Delhaize et al., 2009). When the soil was limed, which substantially reduced Al saturation, grain yield of the transgenic and non-transgenic lines were similar, suggesting that enhanced P uptake under soil acidity was indeed largely achieved as an indirect effect of TaALMT1 enhancing Al resistance. It should be noted that clay acidic soils generally have a strong buffering capacity and, although liming can be used to reduce Al3<sup>+</sup> in the topsoil, neutralization of subsoil Al3<sup>+</sup> is often difficult to achieve. In the absence of liming, Al resistance can have an important indirect effect on crop performance via both enhanced root proliferation in the topsoil, where P is primarily located on acidic soils (Lynch and Brown, 2001), and improved water acquisition by better root development in the subsoil. With liming, Al tolerance may most strongly benefit crop yields by enhanced water acquisition from deeper, acidic soils layers.

Allelic variation at the sorghum chromosome 3 Al resistance locus, AltSB (Magalhaes et al., 2004), where the citrate transporter, SbMATE, resides (Magalhaes et al., 2007), explains a large portion of the sorghum Al resistance phenotype. Recently, a sorghum recombinant inbred line (RIL) population was assessed for Al resistance both in lab-based hydroponics (relative root growth) and in the field (grain yield) under +/−Al exposure, in a phenotyping site located at the Embrapa Maize and Sorghum station in Brazil (Carvalho et al., 2016). In that study, sorghum hybrids were also constructed that were either homozygous for the Al-sensitive or -resistant SbMATE allele, or heterozygous for SbMATE. These hybrids were isogenic, so that AltSB alleles from different donors could be compared within a homogeneous genetic background, thus isolating the effect of SbMATE from genetic background effects.

The resulting isogenic hybrids were assessed for grain yield in the field on control (absence of Al toxicity in the soil) or in an Al toxic soil with 56% Al saturation in the top soil (0–20 cm) and ∼70% Al saturation in the sub-soil (20–40 cm). A major QTL underlying both Al resistance assessed in hydroponics and grain yield under Al toxicity in the field was co-located with SbMATE on sorghum chromosome 3, and explained a large portion of the genetic variance in the Al toxic but not in the non Al-toxic soil. The allele associated with increased Al resistance was donated by the Al tolerant parent, SC283, and the Al resistance allele did not decrease grain yield in the absence of Al toxicity, indicating that no yield penalty arises from Al-induced citrate release elicited by SbMATE. This genetic approach allowed the authors to estimate a consistent effect of a single Al resistance allele of SbMATE as a grain yield increase of ∼0.6 ton ha−<sup>1</sup> , both in the RILs and in hybrid combinations. The rather additive gene action of SbMATE in grain yield production indicates that, when in homozygosity, SbMATE increases grain yield by more than 1.0–ha−<sup>1</sup> , or more than 50% over the population mean. The Al saturation level in the Al toxic site, 56%, is well above the 20% critical level beyond which sorghum yields are reduced (Gourley, 1987). Therefore, most of the yield advantage of SbMATE is likely caused by its effect on Al resistance itself. However, the typical acidic soil in question also has high P fixation capacity and P diffusion is known to be highly depend on the soil water content (Novais and Smith, 1999). Therefore, as Al stress and low P availability in general co-exist on acidic soils, a smaller portion of the yield advantage caused by SbMATE may have originated from citratebased enhanced P mobilization (Drouillon and Merckx, 2003) from the soil clays into the root surface, which is expected to favor P uptake.

A more compelling evidence for a pleiotropic effect of SbMATE on P acquisition comes from a genome-wide association mapping study conducted in West Africa (Leiser et al., 2014), which included gene-specific markers developed for SbMATE (Caniato et al., 2014). This study revealed that SbMATE SNPs were highly associated with grain yield and the associations were found especially under low P conditions for sorghum cultivated in soils at 29 different sites in West Africa, explaining up to 16% of the genotypic variance (Leiser et al., 2014). The average Al saturation was only 10% in the 16 field trials that were analyzed for Al saturation in the Leiser et al. (2014) study, and only one site had Al saturation reasonable above (27.5%) the critical level of Al saturation determined for sorghum (20%, Gourley, 1987). This suggests a direct pleiotropic effect of Al-activated citrated release promoted by SbMATE in enhancing P uptake and sorghum yields under low P availability in West Africa. It should be noted, however, that Al toxicity varies according to the chemical and mineral nature of the soils, which ultimately controls free Al3<sup>+</sup> activity in the soil solution. Therefore, in sandy soils, such as those commonly found in West Africa, we cannot rule out that higher Al3<sup>+</sup> activity in some of the sites may have led SbMATE activity to improve sorghum grain yield via Al resistance.

#### Evidence for a Pleiotropic Role of the STOP1/ALMT1 Module and ALS3 on P Acquisition via Changes in Root Morphology in Response to P Deficiency

Recent research findings exposed a possible direct link between AtALMT1 function and both Al resistance and changes in root growth triggered by response to low P (Balzergue et al., 2017; Mora-Macías et al., 2017). Previously, an antagonistic connection was established between phosphate and Fe availability, leading to adjustments in root growth (Müller et al., 2015). It was found that the LPR1 (ferroxidase)/PDR2 (P5-type ATPase) module enhances cell-specific Fe and callose deposition in the meristem and elongation zones under low P conditions. Under low Pi, accumulated ROS, possibly resulting from Fe toxicity triggered by Fe3<sup>+</sup> accumulation in the apoplast via LPR1-dependent Fe oxidation, may lead to callose deposition. In turn, according to the proposed model, callose deposition in the RAM under low P impairs cell-to-cell movement of the SHORT-ROOT (SHR) transcription factor, which is important for stem cell maintenance, hence providing a checkpoint for primary root growth control in response to low P.

A mutation screen in Arabidopsis indicated that both ALMT1 and its transcriptional regulator, STOP1, repress primary root growth under −P conditions (Mora-Macías et al., 2017).

Furthermore, P deficiency was also shown to upregulate ALMT1 expression in Arabidopsis, and experiments where exogenous malate was applied to the RAM restored the short root phenotype in almt1 and stop1 mutants in a concentration-dependent manner. Fe accumulation in the RAM was found to be required to activate the inhibition of primary root growth under −P conditions (Müller et al., 2015). Hence, the primary root growth inhibition by malate was suggested to occur via malate chelating and solubilizing Fe in the rhizosphere, which would promote Fe accumulation in the RAM apoplast (Mora-Macías et al., 2017). Accordingly, the resulting RAM exhaustion process leading to inhibition of the primary root growth under low P (Sánchez-Calderón et al., 2005) happens in the presence of Fe in the growth medium. Callose deposition, which is stimulated by ROS, may be involved in the root elongation inhibition following the model proposed by Müller et al. (2015). Hence, impaired cell-to-cell movement of the SHR transcription factor, which is important for stem cell maintenance, was suggested to lead to meristem exhaustion, inhibiting primary root growth (Müller et al., 2015; Mora-Macías et al., 2017). Because the enhanced proliferation of lateral roots coincides with the inhibition of the primary root (release of root apical dominance) under low P conditions (Sánchez-Calderón et al., 2005), ALMT1 may ultimately increase P uptake on acidic soils via increases in total root surface area, thereby favoring P diffusion toward the root surface.

A strikingly similar mechanism for an Al resistance gene leading to changes in root growth as a response to P deficiency has been proposed for ALS3 (Larsen et al., 1996, 2005) and AtSTAR1 (Huang et al., 2010; Belal et al., 2015; Dong et al., 2017). Together, STAR1 and STAR2 (a rice homolog of als3) form an ABC transporter implicated in Al resistance likely via the transport of UDP glucose into the root apoplast, which is believed to modify the cell wall leading to Al resistance (Huang et al., 2009) as previously discussed in Section "Al Resistance Transporters That do not Transport Organic Acids: Aluminum-Sensitive 3 (ALS3)." The commonality between the putative pleiotropic pathways mediated by ALMT1 and ALS3/AtSTAR1 is striking, particularly taking into consideration that those genes underlie distinctly different Al resistance mechanisms. Both pathways involve cross-talk between low P responses and Fe homeostasis, with involvement of LOW PHOSPHATE ROOT (LPR) oxidases; mutations in LPR leads to reduced Fe3<sup>+</sup> accumulation in roots and thereby root growth insensitivity to low Pi (Müller et al., 2015; Dong et al., 2017; Mora-Macías et al., 2017). However, the ALS3 pathway involves UDP glucose, which reverses Fe3<sup>+</sup> overaccumulation and rescues the short root phenotype in als3 subjected to −P conditions (Dong et al., 2017). However, unlike the T-DNA mutants for AtALMT1 and STOP1, als3 shows enhanced inhibition of primary root growth under P deficiency (Dong et al., 2017), suggesting possible antagonism between Al resistance conferred by ALS3 and P acquisition.

These studies offer a radically different stance on root OA release enhancing resistance to low P solely via increased P availability in the rhizosphere, as root developmental changes caused by ALMT1/STOP1 and ALS3 appear to be a low P-specific response that is focused on root development. A common physiological basis centered on Fe homeostasis underlying the effect of distinctly different Al resistance pathways encoded by ALMT1/STOP1 and ALS3 on root remodeling under low P seems likely. Should those responses prove to persist for crops cultivated on acidic soils, it will be tempting to speculate that the close soil chemistry associations between Al toxicity and low P availability, which is centered on the presence of Fe and Al oxides, may have resulted in coselective pressure for pleiotropic mechanisms enabling plants both to tolerate Al3<sup>+</sup> and to acquire P more efficiently. Nevertheless, there is a strong need for strategies to validate whether the direction of this hypothetical pleiotropic effect is consistent with a positive net benefit on acidic soil performance.

### Are Wall-Associated Kinases Associated With a Joint Effect on Al Resistance and P Acquisition?

Wall-associated kinases (WAKs), which are receptor-like kinase proteins (Kohorn and Kohorn, 2012) that span the plasma membrane and extend out into the cell wall (He et al., 1999), have been shown to play roles in cell expansion, development, morphogenesis, and defense responses to environmental stimuli (Sivaguru et al., 2003; Brutus et al., 2010; Kohorn and Kohorn, 2012; Gramegna et al., 2016; Mangeon et al., 2016). Sivaguru et al. (2003) reported that AtWAK1 expression was rapidly induced by Al and disappeared after 9 h of Al exposure and that transgenic plants overexpressing AtWAK1 showed enhanced Al resistance. Recently, a T-DNA knockout of the glycine-rich protein, AtGRP3, which interacts with AtWAK1 (Park et al., 2001), has also been shown to enhance Al resistance in Arabidopsis, similar to AtWAK1 (Mangeon et al., 2016). However, AtGRP3 expression was not modulated by Al and grp3 had a long root phenotype in the absence of Al exposure. Therefore, it remains to be verified whether the lower root growth inhibition in grp3 exposed to Al compared to the wt is in fact due to a mechanism enhancing Al resistance or is influenced to some extent by a leaky grp mutation, based on the role for AtGRP3 in repressing root growth.

Wall-associated kinases form a subfamily within the receptor kinase (RLKs)/Pelle superfamily, which includes other subfamilies such as WAK-like kinase (WAKL) and Leaf rust 10 disease-resistance locus receptor-like protein kinase (LRK10), that share similar protein architectures with the WAK proteins (Shiu and Bleecker, 2003; Hou et al., 2005; Lim et al., 2014). The general WAK protein architecture features an extracellular moiety containing a cysteine-rich (Cys-rich) galacturonanbinding domain (Gub\_Wak), epidermal growth factor (EGF) repeats, and a TM domain, in addition to a cytoplasmic serine/threonine kinase domain (Anderson et al., 2001; Decreux and Messiaen, 2005; Decreux et al., 2006).

Using association mapping, Hufnagel et al. (2014) showed that sorghum homologs of the rice serine/threonine receptor kinase, OsPSTOL1 (Gamuyao et al., 2012), are involved in increases in root surface area leading to enhanced P acquisition

FIGURE 1 | Structure and possible function of WAK, WAKL, and LRK10 members of the RLKs/Pelle superfamily. Protein domains are represented with the following colors: purple (Gub\_Wak), red (Wak association), and blue (Kinase). Amino acidic sequences were obtained from the National Center for Biotechnology Information (NCBI; http://www.ncbi.nlm.nih.gov) and Phytozome (www.phytozome.org) databases. The SbWAK protein sequence is available at Phytozome under accession Sobic.004G008100.1. The amino acidic alignment was performed with ClustalW (www.ebi.ac.uk/Tools/msa/clustalw2), and protein domain prediction was carried out using the Pfam (http://pfam.xfam.org/) and Smart (http://smart.embl-heidelberg.de/) tools. Inferences on functions for AtWak1(Anderson et al., 2001; Sivaguru et al., 2003; Brutus et al., 2010; Gramegna et al., 2016), OsWAK11(Hu et al., 2014), ZmWAK (Zuo et al., 2015), HvWAK1 (Kaur et al., 2013), TaWAK (Yang et al., 2014), OsDEE1S (Wang et al., 2012), OsWak1(Li et al., 2009), AtWAKL4 (Hou et al., 2005), Sb03g006765 (Hufnagel et al., 2014), and ZmWak-RLK1 (Hurni et al., 2015) are shown.

and grain yield under low P availability in the soil. In sorghum, these SbPSTOL1 proteins are predicted to have a signal peptide consistent with the targeting to a secretory pathway, as well as a TM domain and cell wall association domains. For example, the Sb03g006765 protein associated with P efficiency and increased root surface area is predicted to have a Cys-rich GUB\_Wak domain and a wall-associated receptor kinase domain (WAK\_association) located C-terminal to the GUB\_Wak domain. Similarities between SbPSTOL1 and WAK proteins such as AtWAK1, which appears to be involved in Al resistance (Sivaguru et al., 2003), arise primarily from the presence of the GUB\_Wak and TM domains, similar intron–exon organization, and a genomic localization in tight physical clusters (Hufnagel et al., 2014). Recent studies have suggested that amino acids in the Gub\_Wak domain bind covalently to native pectins and oligogalacturonides in the cell wall (Verica and He, 2002; Decreux and Messiaen, 2005; Decreux et al., 2006; Kohorn and Kohorn, 2012; Kohorn et al., 2016). This leads us to speculate that SbPSTOL1 proteins may function as WAKs, functioning as receptors for the activation of signaling cascades in response to extracellular stimuli (in this case, P deficiency). However, in place of the EGF repeats, which is a hallmark of WAK proteins (Kanneganti and Gupta, 2008), WAKL and LRK10 members, including Sb03g006765, possess a WAK\_association domain.

The GUB\_Wak domain is present in certain plant proteins suggested to be involved in responses to abiotic and biotic stresses that belong to three subfamilies in the RLK superfamily, the WAKL, WAK, and LRK10 subfamilies (with Sb03g006765 within the LRK10 subfamily). These proteins are depicted in **Figure 1**. Sequence alignment of the GUB\_Wak amino acidic sequences in these proteins does not show a high degree of conservation. However, this domain has conserved clusters of hydrophobicity that are essential for the association of these proteins via the extracellular residues, including a Cys-rich region and a conserved YPF motif. Therefore, it remains to be seen whether the SbPSTOL1 proteins functionally work as WAKs such as AtWAK1. If so, given the predicted role for SbPSTOL1 in enhancing root growth and P uptake in sorghum, this class of proteins could jointly control Al resistance and P uptake.

### CONCLUSION

We are at a stage in research on crop plant adaptation to acidic soils where a number of different Al resistance genes have been identified. These genes have been discovered using a variety of both forward and reverse genetic strategies, ranging from candidate genes validated primarily via ectopic overexpression in transgenic plants or identified via mutant screens to map-based cloning of Al resistance genes underlying loci previously known to play a role in the genetic variation of Al resistance. In most cases, very little work has been done to translate the findings from the basic research used to identify and characterize the genes to practical applications to generate crop varieties in breeding programs. The research that connects with genetic variation present within crop species to identify Al resistance genes is certainly the most amenable to providing molecular tools for the breeding of crops with improved production on acidic soils. In the cases where genetic determinants of Al resistance have been found by other

approaches, efforts to assess whether those determinants are also active in crop plants in field conditions are sorely needed if the ultimate goal is indeed to generate crops more adapted to cultivation on acidic soils. While the effect of Al resistance on crop performance on acidic soils is known, pleiotropic effects of such genes on P uptake efficiency needs to be explored in crop species grown in the field. In both cases, detailed quantification is needed to gage the true potential of Al resistance genes in coping with agriculture in stress-prone areas. Particularly in a scenario where global climate change is resulting in greater drought stress, the potential of those genes to ensure food security worldwide may be far greater than initially believed.

#### REFERENCES


### AUTHOR CONTRIBUTIONS

JM and MP delineated and wrote this review. LM and LK wrote this review. LK also edited the manuscript.

## FUNDING

JM and LM acknowledge support from the Fundação de Amparo a Pesquisa do Estado de Minas Gerais (FAPEMIG) and the National Council for Scientific and Technological Development (CNPq).




gene HvMATE controlling aluminium tolerance in barley (Hordeum vulgare L.). Theor. Appl. Genet. 115, 265–276. doi: 10.1007/s00122-007-0562-9


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer MD and handling Editor declared their shared affiliation.

Copyright © 2018 Magalhaes, Piñeros, Maciel and Kochian. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Aluminum–Nitrogen Interactions in the Soil–Plant System

Xue Q. Zhao<sup>1</sup> and Ren F. Shen1,2 \*

<sup>1</sup> State Key Laboratory of Soil and Sustainable Agriculture, Institute of Soil Science, Chinese Academy of Sciences, Nanjing, China, <sup>2</sup> University of Chinese Academy of Sciences, Beijing, China

Aluminum (Al) is the most abundant metal in the Earth's crust and is not an essential element for plant growth. In contrast, nitrogen (N) is the most important mineral element for plant growth, but this non-metal is often present at low levels in soils, and plants are often N deficient. Aluminum toxicity is dominant in acid soils, and so plants growing in acid soils have to overcome both Al toxicity and N limitation. Because of low N-use efficiency, large amounts of N fertilizers are applied to crop fields to achieve high yields, leading to soil acidification and potential Al toxicity. Aluminum lowers plant N uptake and N-use efficiency because Al inhibits root growth. Although numerous studies have investigated the interactions between Al and N, a complete review of these studies was lacking. This review describes: (1) the link between plant Al tolerance and ammonium/nitrate (NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup>) preference; (2) the effects of NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup> and pH on Al toxicity; (3) the effects of Al on soil N transformations; and (4) the effects of Al on NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup> uptake and assimilation by plants. Acid soils are characterized chemically by a relatively high ratio of NH<sup>4</sup> <sup>+</sup> to NO<sup>3</sup> <sup>−</sup> and high concentrations of toxic Al. Aluminumtolerant plants generally prefer NH<sup>4</sup> <sup>+</sup> as an N source, while Al-sensitive plants prefer NO<sup>3</sup> <sup>−</sup>. Compared with NO<sup>3</sup> <sup>−</sup>, NH<sup>4</sup> <sup>+</sup> increases the solubilization of toxic Al into soil solutions, but NH<sup>4</sup> <sup>+</sup> generally alleviates Al phytotoxicity under solution culture because the protons from NH<sup>4</sup> <sup>+</sup> compete with Al3<sup>+</sup> for adsorption sites on the root surface. Plant NO<sup>3</sup> <sup>−</sup> uptake and nitrate reductase activity are both inhibited by Al, while plant NH<sup>4</sup> + uptake is inhibited to a smaller degree than NO<sup>3</sup> <sup>−</sup>. Together, the results of numerous studies indicate that there is a synergistic interaction between plant Al tolerance and NH<sup>4</sup> <sup>+</sup> nutrition. This has important implications for the adaptation of plants to acid soils that are dominated chemically by toxic Al as well as NH<sup>4</sup> <sup>+</sup>. Finally, we discuss how this knowledge can be used to increase plant Al tolerance and N-use efficiency in acid soils.

Keywords: aluminum, nitrogen, ammonium, nitrate, interaction, plant, acid soil

## INTRODUCTION

Acid soils cover approximately 30% of the ice-free land and up to 70% of potentially arable soils worldwide (von Uexküll and Mutert, 1995). Acid soils occur mainly in humid tropical and temperate areas (von Uexküll and Mutert, 1995), where water and heat are generally abundant for plant growth, implying that acid soils have huge productive potential. However, plant productivity

#### Edited by:

Peter Ryan, Commonwealth Scientific and Industrial Research Organisation (CSIRO), Australia

#### Reviewed by:

Brent N. Kaiser, University of Sydney, Australia Francesco Di Gioia, University of Florida, United States

> \*Correspondence: Ren F. Shen rfshen@issas.ac.cn

#### Specialty section:

This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science

Received: 06 February 2018 Accepted: 25 May 2018 Published: 18 June 2018

#### Citation:

Zhao XQ and Shen RF (2018) Aluminum–Nitrogen Interactions in the Soil–Plant System. Front. Plant Sci. 9:807. doi: 10.3389/fpls.2018.00807

**35**

in acid soils is limited primarily by aluminum (Al) toxicity accompanied by deficiencies of some nutrients (Zhao et al., 2014). The improvement of crop productivity in acid soils depends on the dual enhancement of plant Al tolerance and nutrient-use efficiency.

Nitrogen (N) is the most abundant mineral nutrient required by plants. Soil N availability greatly affects the growth and development of crops worldwide (Gutiérrez, 2012). Nitrogen deficiency is a widespread problem for plants grown in terrestrial ecosystems (Vitousek and Howarth, 1991), and it is also a major factor limiting plant growth in acid soils (Fageria and Baligar, 2001). Large amounts of N fertilizers are used in agriculture to grow crops that feed an increasing global population every year. Erisman et al. (2008) estimated that N fertilizer has supported around 4 billion people born since 1908, accounting for approximately 27% of the world's population over the past century. At the same time, excess N fertilization is causing environmental problems such as water eutrophication, greenhouse gas emissions, nitrate (NO<sup>3</sup> <sup>−</sup>) loss, acid rain, and soil acidification due to low N-use efficiency (Ju et al., 2009). High yields and high nutrient-use efficiency are essential for contemporary agriculture. Therefore, there is an urgent need to increase plant N-use efficiency by understanding the responses to N (Kant et al., 2011).

Aluminum is the most abundant metal in the Earth's crust. It is not an essential element for plants, and excess Al is toxic to most plants. The primary symptom of Al phytotoxicity is the inhibition of root elongation, which can occur after exposure to Al3<sup>+</sup> at concentrations as low as µM levels within 1 h (Matsumoto, 2000; Kochian et al., 2005; Ma, 2007). This inhibition can be caused by reductions in cell elongation and cell division, which are attributed to Al interference with the cell wall, plasma membrane, the cytoskeleton, oxidative stress, signal transduction pathways, cytoplasm calcium homeostasis, magnesium uptake, and auxin polar transport (Ma, 2007). Plants have two strategies to detoxify Al (Ma, 2007). One is to exclude Al from the root tips (exclusion mechanism) and the other is to tolerate Al that enters the plant body (internal tolerance mechanism). Roots are the main organ for plants to take up nutrients from the growth medium, so Al toxicity inevitably affects the ability of plants to acquire nutrients from acid soils. On one hand, the inhibitory effects of Al on root growth can reduce the amounts of nutrients taken up by plants because of the small root volume. On the other hand, Al may directly affect the transport and metabolism of nutrients within plants. Interactions between Al and many nutrients often occur within soils and plants (Zhao et al., 2014). Most reports have focused on the effects of various externally added nutrients on Al phytotoxicity (Zhao et al., 2014), but the effects of Al on the uptake of these nutrients by plants and their corresponding mechanisms have received relatively little attention.

Aluminum is beneficial and even potentially essential for some plant species (Bojórquez-Quintal et al., 2017), because of the Al-induced stimulation of nutrient uptake (Watanabe and Osaki, 2002). Aluminum supply was shown to stimulate N uptake by several plant species adapted to acid soils (Osaki et al., 1997), and Al treatments increased shoot N contents in wheat and rye (Dinev and Stancheva, 1993). In contrast, Al reduced root N uptake and its upward translocation to shoots in sorghum and corn (Gomes et al., 1985; Pintro et al., 1996). Aluminum promoted the growth of plants supplied with ammonium (NH<sup>4</sup> <sup>+</sup>) but inhibited that of plants supplied with NO<sup>3</sup> <sup>−</sup> (Zhao et al., 2014). Nitrogen is a metabolic element involved in the synthesis of amino acids and proteins within plants. Knowledge about Al–N interactions may supply new information to explain instances where Al benefits plant growth.

Several reviews have focused on the interactions between Al and phosphorus (Chen et al., 2012), calcium (Rengel and Zhang, 2003; Meriño-Gergichevich et al., 2010), magnesium (Bose et al., 2011; Chen and Ma, 2013), boron, and silicon (Hodson and Evans, 1995; Horst et al., 2010). Aluminum is a metal and a toxic element to many plants, while N is a non-metal and is an essential element for all plants. More than 100 papers have reported on Al–N interactions so far, highlighting the importance of this topic. Despite the large amount of literature on Al–N interactions, there has been no systematic review of this topic so far. Here, we provide a detailed description and analysis of studies on the interactions between Al and N, including the link between plant Al tolerance and NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup> preference, the effects of NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup> and pH on Al toxicity, the effects of Al on soil N transformations, and the effects of Al on NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup> uptake and assimilation. We also propose a strategy for improving plant Al tolerance and N-use efficiency in acid soils.

#### LINK BETWEEN PLANT Al TOLERANCE AND INORGANIC N PREFERENCE

Acid soils are characterized by poor nitrification and high levels of soluble Al, while neutral to calcareous soils show high nitrification and lower levels of Al toxicity (Zhao et al., 2014; Che et al., 2015). The two main inorganic N sources available for plant growth are NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup>. Therefore, on the basis of the environment driving evolution, plants originating from acid soils are Al tolerant and prefer NH<sup>4</sup> <sup>+</sup> to NO<sup>3</sup> <sup>−</sup>, while those originating from neutral to calcareous soils are Al sensitive and prefer NO<sup>3</sup> <sup>−</sup> to NH<sup>4</sup> <sup>+</sup> (Gigon and Rorison, 1972; Foy and Fleming, 1978; Rorison, 1985; Falkengren-Grerup, 1995; Marschner, 1995; Maathuis, 2009; Zhao et al., 2013b) (**Table 1**). For instance, the growth of lowbush blueberry, which is adapted to strongly acid soils, was shown to be greatly promoted by NH<sup>4</sup> + but strongly inhibited by NO<sup>3</sup> <sup>−</sup> (Townsend, 1966; Townsend and Blatt, 1966). Wheat and barley are Al-sensitive and prefer NO<sup>3</sup> − (Malhi et al., 1988; Cramer and Lewis, 1993; Famoso et al., 2010), while tea and rice are Al-tolerant and prefer NH<sup>4</sup> <sup>+</sup> (Ruan et al., 2007; Famoso et al., 2010; Zhao et al., 2013b). The activity of NO<sup>3</sup> <sup>−</sup> reductase could not be detected in some calcifuge species, suggesting that they have a restricted ability to utilize NO<sup>3</sup> − (Havill et al., 1974). Rice (Oryza sativa) has two subspecies, indica and japonica. Indica rice cultivars generally prefer NO<sup>3</sup> <sup>−</sup>, while japonica cultivars prefer NH<sup>4</sup> <sup>+</sup> (Zhao et al., 2013b; Hu et al., 2015). Correspondingly, indica rice cultivars are generally Al sensitive, while japonica cultivars are Al tolerant (Zhao et al., 2013b). Among different rice cultivars, Al tolerance is closely related to NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> preference (Zhao et al., 2013b).

#### TABLE 1 | Aluminum tolerance and NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup> preference of plant species.


The above analyses collectively suggest that Al-tolerant plant species and genotypes utilize NH<sup>4</sup> <sup>+</sup> more efficiently than NO<sup>3</sup> − (**Table 1**). This knowledge is helpful for the selection of crop genotypes with both high Al tolerance and N-use efficiency via breeding or genetic modification. The selection of such genotypes should reduce the amount of N fertilizer required and improve plant growth in acid soils. However, the molecular mechanism underlying the link between plant Al tolerance and inorganic N preference is unclear. The two characteristics of grain protein content and acidity tolerance were found to be positively correlated among different wheat lines (Mesdag et al., 1970). In addition, a quantitative trait locus genetic analysis revealed that loci associated with Al tolerance and NH<sup>4</sup> <sup>+</sup> utilization were located in similar regions of rice genome (Ogawa et al., 2014). An important goal for future research is to uncover the mechanism of the link between plant Al tolerance and inorganic N preference at the molecular and genetic levels.

#### EFFECTS OF NH<sup>4</sup> <sup>+</sup>, NO<sup>3</sup> <sup>−</sup>, AND pH ON Al TOLERANCE

In recent decades, various anthropogenic activities have greatly accelerated soil acidification in Chinese crop fields (Guo et al., 2010; Liang et al., 2013). Among these activities is the excess use of NH<sup>4</sup> <sup>+</sup> fertilizer (Barak et al., 1997; Fang et al., 2014). Atmospheric NH<sup>4</sup> <sup>+</sup> deposition is also an important factor resulting in soil acidification (van Breemen et al., 1982). Nitrification is the mechanism by which NH<sup>4</sup> <sup>+</sup> acidifies soils. During the nitrification of NH<sup>4</sup> <sup>+</sup> to NO<sup>3</sup> <sup>−</sup>, H<sup>+</sup> are released into soils, which increase the concentration of soluble Al (van Breemen et al., 1982; Mulder et al., 1989; Mulder and Stein, 1994; Che et al., 2015) (**Figure 1**). Thus, NH<sup>4</sup> <sup>+</sup> facilitates the occurrence of Al toxicity much more than NO<sup>3</sup> <sup>−</sup> does. However, increased soluble Al content in soils caused by low pH does not always increase Al phytotoxicity, because lower pH can result in the desorption of Al from plant roots into the rhizosphere solution (**Figure 1**).

Early studies showed that changes in root zone pH due to ion uptake imbalances were related to Al tolerance in triticale, wheat, and rye under certain solution and soil conditions (Mugwira and Patel, 1977). The plant growth medium can be acidified due to NH<sup>4</sup> <sup>+</sup> uptake by plant roots and the nitrification of NH<sup>4</sup> <sup>+</sup> to NO<sup>3</sup> <sup>−</sup>. Alternatively, the growth medium can be alkalinized due to the uptake of NO<sup>3</sup> <sup>−</sup> by plant roots. Because Al toxicity occurs in acid soils, one could speculate that the preferential utilization of NO<sup>3</sup> <sup>−</sup> relative to NH<sup>4</sup> <sup>+</sup> can enhance plant Al tolerance through increasing the pH of the growth medium via NO<sup>3</sup> <sup>−</sup> uptake. The Al tolerance of some wheat varieties was attributable to their abilities to preferentially utilize NO<sup>3</sup> <sup>−</sup> relative to NH<sup>4</sup> <sup>+</sup> through rhizosphere alkalization (Foy et al., 1965, 1967; Foy and Fleming, 1978, 1982; Fleming, 1983; Taylor and Foy, 1985a,b,c). The results of subsequent studies, however, indicated that genotypic differences in wheat Al tolerance were not caused by differences in rhizosphere pH induced by the differential uptake of NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> (Taylor, 1988a,b; Miyasaka et al., 1989). Instead, the differences in the uptake of NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> among different wheat genotypes were suggested to be the result of, rather than the cause of, differences in A1 tolerance among genotypes (Taylor, 1988a,b; Miyasaka et al., 1989). Another research demonstrated that the decrease in the growth medium pH under Al stress was greater for an Al-tolerant wheat genotype than an Al-sensitive one (Ikeda and Yamanishi, 1999). Therefore, genotypic differences in the relative Al tolerance of wheat could not be explained by root-induced pH changes due to the uptake of NH<sup>4</sup> + and NO<sup>3</sup> −.

Three reports on rice plants drew different conclusions. In two studies, an Al-tolerant rice genotype had a stronger ability than an Al-sensitive genotype to increase nutrient solution pH through efficient NO<sup>3</sup> <sup>−</sup> uptake and metabolism (Ganesan et al., 1993; Justino et al., 2006). However, another study (van Hai et al., 1989) compared with NO<sup>3</sup>

−.

fpls-09-00807 June 14, 2018 Time: 17:48 # 4

obtained the opposite result, in that an Al-resistant genotype took up more NH<sup>4</sup> <sup>+</sup> and acidified the nutrient solution to a greater degree than did an Al-sensitive one. In barley, Al tolerance of different cultivars was not related to the root-induced pH change by the uptake of inorganic N sources from the growth medium (Wagatsuma and Yamasaku, 1985). Similarly, differences in pH changes in the growth medium were not related to differences in A1 tolerance between two sorghum genotypes (Galvez and Clark, 1991). In fact, the NO<sup>3</sup> <sup>−</sup> uptake rate was found to be higher in an Al-sensitive sorghum genotype than in an Altolerant one (Cambraia et al., 1989). Genotypic differences in the Al tolerance of soybean plants were not associated with the difference in NH<sup>4</sup> <sup>+</sup> uptake vs. NO<sup>3</sup> <sup>−</sup> uptake and root-induced pH changes (Klotz and Horst, 1988b). Changes in the medium pH were also not related to Al tolerance in triticale (Antunes and Antonieta Nunes, 1997). These analyses further demonstrated that genotypic differences in the Al tolerance of diverse plant species cannot be explained only by root-induced pH changes due to NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> uptake.

Since low pH increases the concentrations of soluble Al in soils, the alkalization of the rhizosphere was proposed to be an important mechanism of plant Al tolerance (Matsumoto, 2000; Kochian et al., 2004; Ma, 2007). However, several studies demonstrated that H<sup>+</sup> could alleviate Al toxicity because H<sup>+</sup> competed with Al3<sup>+</sup> for adsorption to the root surface (Kinraide et al., 1992; Godbold et al., 1995; Zhao et al., 2009; Zhao et al., 2014). A supply of H<sup>+</sup> also alleviated Al toxicity in bacteria (Kinraide and Sweeney, 2003) and yeast (Zhao et al., 2017). These results implied that Al toxicity is much lower at low pH than at high pH under a certain acid pH range (pH < 5.0) because of the H<sup>+</sup> alleviation of Al phytotoxicity. The uptake of NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> decreases and increases the pH of the medium, respectively. Many reports have indicated that NH<sup>4</sup> <sup>+</sup> supply can enhance plant Al tolerance, while NO<sup>3</sup> <sup>−</sup> supply aggravates Al toxicity (**Table 2**). In some studies, Al was found to stimulate the growth of some grasses (Rorison, 1985), tropical trees (Watanabe et al., 1998), Lespedeza bicolor (Chen et al., 2010), and rice (Zhao et al., 2013b) when supplied with NH<sup>4</sup> <sup>+</sup>, but not when supplied with NO<sup>3</sup> <sup>−</sup>. The stimulatory effects of Al on plant growth may be related to the effects of Al to alleviate H<sup>+</sup> toxicity (Kinraide et al., 1992). Thus, NH<sup>4</sup> <sup>+</sup> alleviates Al toxicity, and Al enhances NH<sup>4</sup> + utilization.

It is now accepted that the NH<sup>4</sup> <sup>+</sup>-induced rhizosphere acidification is the primary mechanism underlying the NH<sup>4</sup> + enhancement of Al tolerance in plants (Zhao et al., 2009; Wang et al., 2015) (**Figure 1**). Relative to NO<sup>3</sup> <sup>−</sup>, NH<sup>4</sup> <sup>+</sup> uptake by rice roots reduces the pH of the nutrient solution. Lower pH further decreases the number of Al-binding functional groups and enhances the positive electrical potential of the root surface (Wang et al., 2015; Liu et al., 2016). Consequently, NH<sup>4</sup> <sup>+</sup>-fed roots adsorb less Al than do NO<sup>3</sup> <sup>−</sup>-fed roots, thereby alleviating Al toxicity. The ability of NH<sup>4</sup> <sup>+</sup> to alleviate Al toxicity was also observed under constant pH conditions (Rorison, 1985; Klotz and Horst, 1988a,b; Grauer and Horst, 1990), indicating that factors other than pH may be involved. It is possible that intermediate products of N metabolism such as nitric oxide (NO) play a role in the alleviation of Al toxicity by NH<sup>4</sup> <sup>+</sup> (Zhao and Shen, 2013).

TABLE 2 | Summary of NH<sup>4</sup> <sup>+</sup> effects on plant Al tolerance relative to NO<sup>3</sup> <sup>−</sup>: (+) enhancement, (−) decrease, and (0) no change.


<sup>a</sup>Study was conducted using sand culture irrigated with nutrient solutions. Studies not marked by superscript letter were conducted using hydroponic systems. <sup>b</sup>Effect was dependent on plant genotypes.

Several studies found that NH<sup>4</sup> <sup>+</sup> aggravated Al toxicity, relative to NO<sup>3</sup> <sup>−</sup> (**Table 2**), which may reflect differences in plants' sensitivity to NH<sup>4</sup> <sup>+</sup>. Some studies on the aggravating effects of NH<sup>4</sup> <sup>+</sup> on Al toxicity used wheat as the experimental material (Fleming, 1983; Taylor and Foy, 1985a,b,c). Wheat plants prefer NO<sup>3</sup> <sup>−</sup> to NH<sup>4</sup> <sup>+</sup> and are sensitive to both Al and NH<sup>4</sup> + (**Table 1**). If wheat plants are supplied only with NH<sup>4</sup> <sup>+</sup>, then NH<sup>4</sup> <sup>+</sup> toxicity may occur and may be more serious than Al toxicity. Thus, NH<sup>4</sup> <sup>+</sup> may aggravate rather than alleviate Al toxicity in wheat plants. Some sorghum genotypes showed lower Al toxicity and some showed higher Al toxicity with NH<sup>4</sup> + relative to NO<sup>3</sup> <sup>−</sup> N (Tan et al., 1992). Because an Al-sensitive sorghum genotype was more NH<sup>4</sup> <sup>+</sup>-sensitive than an Al-tolerant one, NH<sup>4</sup> <sup>+</sup> toxicity probably masked Al toxicity in sorghum (Keltjens, 1987). Consequently, it is difficult to observe the NH<sup>4</sup> <sup>+</sup> alleviation of Al toxicity in NH<sup>4</sup> <sup>+</sup>-sensitive plant species (Keltjens, 1987). Thus, plants grown in acid soils may suffer from Al toxicity accompanied by NH<sup>4</sup> <sup>+</sup> toxicity due to poor soil nitrification.

Most studies on the effects of NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> on Al tolerance have been conducted using hydroponic experiments (**Table 2**), which might not reflect the real effects of NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> on Al tolerance. In soils, lower root rhizosphere pH will result in greater solubilization of Al ions from the soil into the rhizosphere solution, potentially increasing Al toxicity to plants. However, under nutrient solution culture, lower rhizosphere pH will only affect Al speciation (Keltjens and van Loenen, 1989). Lower pH due to NH<sup>4</sup> <sup>+</sup> uptake by plants increases the solubilization of Al3<sup>+</sup> from bulk soils into the rhizosphere solution (**Figure 1**). Nevertheless, for plant roots, more H<sup>+</sup> in the rhizosphere solution can decrease Al3<sup>+</sup> adsorption by roots through cation competition and increasing the positive electrical potential of the root surface. Thus, whether Al toxicity is exacerbated or alleviated by NH<sup>4</sup> <sup>+</sup> or NO<sup>3</sup> <sup>−</sup> may depend on the relative dominance of the effects of pH on Al desorption from soils into the rhizosphere solution and Al adsorption from the rhizosphere solution into the roots. Further studies on this topic should be conducted on soil-grown plants.

#### EFFECTS OF Al ON N TRANSFORMATIONS IN SOILS

Although the effects of nitrification on soil pH and Al solubility are well known, less is known about the effects of Al on soil N transformations such as nitrification and ammonification. The nitrification rate is lower in acid soils than in neutral to calcareous soils (Che et al., 2015), although the reasons for this are still unclear. It is generally considered that low pH inhibits the activity of nitrifying microbes. Higher levels of soluble Al are often concomitant with lower soil pH. Soil N transformations are controlled by microbes. Most microbes are very sensitive to Al (Piña and Cervantes, 1996), while fungi are relatively more tolerant than bacteria to Al and acids (Zhao et al., 2013a, 2017). Low pH does not always result in high concentrations of active Al in soils, because Al ions can form complexes with various organic and inorganic ligands. Future research should explore the role of Al in regulating soil N transformations and in N cycle as a whole.

In a paper published almost 100 years ago (Denison, 1922), Al salts stimulated ammonifying microbes but adversely affected nitrifying bacteria. However, more recent reports showed that Al did not affect the nitrification potential and abundance of ammonia-oxidizing amoA gene of archaea and bacteria (Kasuga et al., 2010; Lin et al., 2017). Bacterial growth was shown to gradually decrease as the pH decreased from 6.5 to 4.0 (Rousk et al., 2010), while soil exchangeable Al linearly increased as the pH decreased from 5.4 to 3.7 (Aciego Pietri and Brookes, 2008). In addition, the OTU richness and Shannon's diversity index of both ammonia-oxidizing archaea and bacteria showed significantly negative correlation with soil pH ranging from 3.77 to 8.46 (Hu et al., 2013). Therefore, microbial growth was found to be limited at soil pHs lower than 5.4 when Al became soluble, but was limited by low pH rather than Al toxicity at pHs ranging from 6.5 to 5.4. These analyses suggested that the inhibition of soil nitrification that transformed NH<sup>4</sup> <sup>+</sup> to NO<sup>3</sup> <sup>−</sup> was due to acid stress rather than Al toxicity, when soil pH decreased from 6.5 to 5.4. There are several soil N transformation processes such as nitrification, denitrification, and ammonification, and different types of microbes control the different pathways of transformations. To clarify the effects of Al on soil N transformation, further studies should evaluate N transformation-related microbial populations and Al

solubility under controlled conditions with variable soil pH and NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup> supply.

#### EFFECTS OF Al ON NO<sup>3</sup> <sup>−</sup> UPTAKE BY PLANT ROOTS

Approximately 30 published studies have focused on the effects of Al toxicity on NO<sup>3</sup> <sup>−</sup> uptake, and most of them found that Al inhibited NO<sup>3</sup> <sup>−</sup> uptake (**Table 3**). Jerzykiewicz (2001) observed that an extremely high concentration of Al (5 mM) even resulted in NO<sup>3</sup> <sup>−</sup> efflux from cucumber roots. The mechanism by which Al inhibits NO<sup>3</sup> <sup>−</sup> uptake is still unclear, but some possible mechanisms have been proposed. In one study, a high Al concentration resulted in large amounts of Al entering the symplast of soybean roots, leading to symplastic Al concentrations that were high enough to inhibit NO<sup>3</sup> − transport across the membrane (Lazof et al., 1994). Thus, one proposed mechanism by which Al inhibits NO<sup>3</sup> <sup>−</sup> uptake is that intracellular Al may bind to NO<sup>3</sup> <sup>−</sup> transporters, NO<sup>3</sup> − metabolic enzymes, and other components of systems related to NO<sup>3</sup> <sup>−</sup> uptake. Plant NO<sup>3</sup> <sup>−</sup> transport involves at least three systems; the constitutive high-affinity transport system (cHATS), the inducible high-affinity transport system (iHATS), and the constitutive low-affinity transport system (cLATS) (Crawford and Glass, 1998; Miller et al., 2007). The constitutive systems function without NO<sup>3</sup> <sup>−</sup> pretreatment, but the inducible system is stimulated by external NO<sup>3</sup> <sup>−</sup>. The cHATS has low values of both K<sup>m</sup> (6–20 µM) and Vmax (0.3–0.82 µmol g−<sup>1</sup> h −1 ), while the iHATS is characterized by higher K<sup>m</sup> (20–100 µM) and Vmax (3–8 µmol g−<sup>1</sup> h −1 ) values and is induced by exposure to NO<sup>3</sup> <sup>−</sup> for hours to days. The cLATS functions at NO<sup>3</sup> − concentrations above 250 µM and does not become saturated even when NO<sup>3</sup> <sup>−</sup> concentrations are as high as 50 mM. Durieux et al. (1993) reported that Al exerted stronger effects on the inducible system than on the constitutive systems. Their results also suggested that high concentrations of Al inhibited the activity of NO<sup>3</sup> <sup>−</sup> transporters in the inducible system rather than affected the number of NO<sup>3</sup> <sup>−</sup> transporters (Durieux et al., 1993). Pretreatment with Al had little effect on NO<sup>3</sup> <sup>−</sup> uptake by plants (Jarvis and Hatch, 1986; Durieux et al., 1993), and NO<sup>3</sup> <sup>−</sup> transport quickly recovered when Al was removed from the external growth medium (Durieux et al., 1993). These results suggested that Al directly interacts with NO<sup>3</sup> <sup>−</sup> transporters but that this interaction is reversible, leading to the inhibition of NO<sup>3</sup> <sup>−</sup> uptake by Al.

The inhibition of root elongation is the main symptom of Al phytotoxicity. Root elongation was inhibited much more than NO<sup>3</sup> <sup>−</sup> uptake in the presence of high Al concentrations in soybean (Rufty et al., 1995). The Al-inhibition of NO<sup>3</sup> <sup>−</sup> uptake was found to be similar across different Al-tolerant soybean genotypes and different root regions (Lazof et al., 1994). The


<sup>a</sup>Study was conducted using sand culture irrigated with nutrient solutions. <sup>b</sup>Study was conducted using soil culture. Studies not marked by superscript letters were conducted using hydroponic systems.

root apex is the primary target of Al toxicity to plants (Ryan et al., 1993). However, NO<sup>3</sup> <sup>−</sup> uptake rates by corn root tips only accounted for a low percentage of NO<sup>3</sup> <sup>−</sup> taken up by the total root system, and N in root tips was mainly derived from N adsorbed through other root regions (Lazof et al., 1992). The mechanism by which Al inhibits root elongation was suggested to differ from the mechanism of Al inhibition of NO<sup>3</sup> <sup>−</sup> uptake in maize (Durieux et al., 1995). The results of these studies indicated that the mechanism of Al inhibition of NO<sup>3</sup> <sup>−</sup> uptake might differ from the mechanism(s) of plant Al sensitivity and Al-inhibited root elongation, at least in maize and soybean. This should be further tested using more plant species.

The effects of Al on NO<sup>3</sup> <sup>−</sup> uptake may depend on Al concentrations, Al exposure time, plant species, and plant genotype. Aluminum does not always affect NO<sup>3</sup> <sup>−</sup> uptake, for example, in Al-tolerant tea trees (Morita et al., 1998) (**Table 3**). A stimulatory effect of Al on root NO<sup>3</sup> <sup>−</sup> uptake has been observed in studies where Al was supplied at low concentrations (van Hai et al., 1989; Rufty et al., 1995; Jerzykiewicz, 2001), or for a short-term (Nichol et al., 1993; Jerzykiewicz, 2001), and/or in studies on wild plant species that prefer Al (Tomioka et al., 2007) (**Table 3**). Similar to the observed stimulatory effects of Al on NO<sup>3</sup> <sup>−</sup> uptake, N uptake and partitioning were found to be enhanced by lower Al concentrations (20–200 µM Al) but inhibited by high Al concentrations (1000 µM Al) in defoliated grasses (Thornton, 1998). In wheat, N uptake by root tips was inhibited by Al in an Al-sensitive genotype, but stimulated in an Al-tolerant genotype (Ikeda and Yamanishi, 1999). These results suggested that low Al accumulation in plants could stimulate NO<sup>3</sup> <sup>−</sup> uptake.

Several possible mechanisms were suggested to be responsible for the stimulation of NO<sup>3</sup> <sup>−</sup> uptake by low concentrations of Al (Rufty et al., 1995; Jerzykiewicz, 2001) (**Figure 2**). First, the increase in the positive electrical potential of the cell surface by Al3<sup>+</sup> could facilitate the access of negatively charged NO<sup>3</sup> <sup>−</sup> to the root cell surface. Second, Al-induced H<sup>+</sup> extrusion under acid stress could increase NO<sup>3</sup> <sup>−</sup> transport across the membrane via H+/NO<sup>3</sup> <sup>−</sup> co-transport. Finally, NO<sup>3</sup> <sup>−</sup> efflux from cells could be diminished by the binding of extracellular Al to the cell membrane if Al impairs the structural integrity of plasma membranes and alters their permeability (Cakmak and Horst, 1991). However, direct and specific evidence for each of these mechanisms is still lacking.

Rufty et al. (1995) compared experimental conditions including the Al concentration, medium pH, and calcium concentration among several papers reporting different effects of Al on NO<sup>3</sup> <sup>−</sup> uptake. This comparative analysis suggested that pH and calcium levels, rather than Al concentrations, explained the differences in results among studies (Rufty et al., 1995). Under acid stress and low calcium levels, Al ameliorated acid

transporter; NR, nitrate reductase; NiR, nitrite reductase; GS, glutamine synthetase; GOGAT, glutamate synthase; GDH, Glutamate dehydrogenase. When plant roots accumulate low concentrations of Al in the apoplastic space, root NO<sup>3</sup> <sup>−</sup> uptake is stimulated by apoplastic Al because of Al3+-increased positive electrical charge of cell surface (1), enhanced H+-NO<sup>3</sup> <sup>−</sup> cotransport (2), and diminished NO<sup>3</sup> <sup>−</sup> efflux (3). When plant roots accumulate large amounts of Al that enters the symplasm of roots, intracellular Al inhibits NO<sup>3</sup> <sup>−</sup> uptake as Al binds to NO<sup>3</sup> <sup>−</sup> transporter (4) and induces enhanced efflux of NO<sup>3</sup> <sup>−</sup> (5). Al3+-increased positive electrical charge of cell surface results in the Al inhibition of NH<sup>4</sup> <sup>+</sup> uptake (6). Low concentrations of Al stimulates NRA (7) because of Al-stimulated NO<sup>3</sup> <sup>−</sup> uptake by the three ways (1, 2, and 3), while high concentrations of Al inhibits NRA (8) because of Al-inhibited NO<sup>3</sup> <sup>−</sup> uptake by the two ways (4 and 5). Al stimulates GS activity (9) due to the binding of Al with GS while inhibits that (10) due to the inhibition of NH<sup>4</sup> <sup>+</sup> uptake (6). The effects of Al on GOGAT and GDH are still uncertain (?).

stress to roots, thereby enhancing NO<sup>3</sup> <sup>−</sup> influx into cells (Rufty et al., 1995). Further studies using carefully designed experiments should explore how pH and calcium affect the ability of Al to alter NO<sup>3</sup> <sup>−</sup> uptake.

Based on the analyses summarized above, we present a schematic diagram to explain the mechanisms of the effects of Al on NO<sup>3</sup> <sup>−</sup> uptake (**Figure 2**). When plant roots accumulate low concentrations of Al in the apoplastic space of roots, extracellular Al may stimulate NO<sup>3</sup> <sup>−</sup> uptake because of an Al3+-induced increase in the positive electrical charge of the cell surface, enhanced H+-NO<sup>3</sup> <sup>−</sup> cotransport, and diminished NO<sup>3</sup> <sup>−</sup> efflux. When large amounts of Al enter the symplasm of roots, root NO<sup>3</sup> <sup>−</sup> uptake is inhibited by Al because Al binds to the NO<sup>3</sup> − transporter and enhances NO<sup>3</sup> <sup>−</sup> efflux. We emphasize that this schematic diagram is based only on the published reports. There is still no direct evidence for these proposed mechanisms. Just as the molecular basis for N uptake has been discovered in recent years, the molecular basis of both the Al-stimulation and Al-inhibition of NO<sup>3</sup> <sup>−</sup> transport can be explored in molecular studies on plant mutants defective in NO<sup>3</sup> <sup>−</sup> transport.

#### EFFECTS OF Al ON NH<sup>4</sup> <sup>+</sup> UPTAKE BY PLANT ROOTS

Various studies have reported that root NH<sup>4</sup> <sup>+</sup> uptake was either inhibited, stimulated, or unaffected by Al (**Table 4**). However, most studies have reported inhibitory effects of Al on NH<sup>4</sup> + uptake by plants. Nichol et al. (1993) indicated that Al treatment for 5 min suppressed the movement of cations (NH<sup>4</sup> <sup>+</sup>, Ca2+, and K <sup>+</sup>) across the plasma membrane but facilitated the movement of anions (NO<sup>3</sup> <sup>−</sup> and phosphate). Aluminum ions may bind to the cell surface and form a positively charged layer, thereby inhibiting the adsorption of positively charged cations to the cell surface but stimulating the adsorption of negatively charged anions. Thus, similar to the mechanisms responsible for the Al stimulation of NO<sup>3</sup> <sup>−</sup> uptake described above, the Al3+-induced increase in the positive electrical charge of the cell surface is responsible for the inhibition of NH<sup>4</sup> <sup>+</sup> uptake by Al (**Figure 2**).

In general, Al exerts a smaller negative effect on NH<sup>4</sup> + uptake than on NO<sup>3</sup> <sup>−</sup> uptake. In maize roots, Al reduced the uptake of both NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> but increased the uptake ratio NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> <sup>−</sup>, indicating that NH<sup>4</sup> <sup>+</sup> uptake was inhibited much less than NO<sup>3</sup> <sup>−</sup> uptake by Al (Purcino et al., 2003). An Al treatment reduced NO<sup>3</sup> <sup>−</sup> uptake but not NH<sup>4</sup> <sup>+</sup> uptake in maize and triticale (Durieux et al., 1993; Calba and Jaillard, 1997; Domingues, 2010), while Al inhibited NO<sup>3</sup> <sup>−</sup> uptake but stimulated NH<sup>4</sup> <sup>+</sup> uptake in sorghum and triticale (Keltjens and van Ulden, 1987; Antunes and Antonieta Nunes, 1997). Leaf N content was increased by A1 when NH<sup>4</sup> <sup>+</sup> was supplied but reduced by Al when NO<sup>3</sup> <sup>−</sup> was supplied (Van Praag et al., 1985). An Al treatment reduced the NO<sup>3</sup> <sup>−</sup> concentration but increased the free NH<sup>4</sup> <sup>+</sup> concentration in the leaves of corn plants (Souza et al., 2016).

The studies reporting that Al stimulated root NH<sup>4</sup> <sup>+</sup> uptake generally used N sources comprising a mixture of NH<sup>4</sup> <sup>+</sup> and NO<sup>3</sup> <sup>−</sup> (Keltjens, 1987, 1988; Keltjens and van Ulden, 1987; Antunes and Antonieta Nunes, 1997). Since Al inhibited NO<sup>3</sup> − uptake in those studies, we may infer that N deficiency caused by the inhibition of NO<sup>3</sup> <sup>−</sup> uptake might explain the stimulation of NH<sup>4</sup> <sup>+</sup> uptake by Al. When NO<sup>3</sup> <sup>−</sup> cannot meet the N demands of plants under Al stress, plants may take up more NH<sup>4</sup> <sup>+</sup> in place of NO<sup>3</sup> <sup>−</sup> to alleviate N deficiency.

#### EFFECTS OF Al ON NO<sup>3</sup> <sup>−</sup> REDUCTION

Nitrate reductase (NR) represents the first enzymatic and ratelimiting step of NO<sup>3</sup> <sup>−</sup> assimilation in plants. It catalyzes the reduction of nitrate to nitrite and is a substrate-inducible enzyme (Tischner, 2000). A large body of research has indicated that Al inhibits NR activity (NRA) in roots, shoots, or both (**Table 5**). Several studies reported that Al toxicity reduced NRA much more in Al-sensitive plant genotypes than in Al-tolerant ones (Foy

TABLE 4 | Summary of effects of aluminum on NH<sup>4</sup> <sup>+</sup> uptake: (−) inhibition, (+) stimulation, and (0) no change.


All studies used hydroponic systems.

and Fleming, 1982; Keltjens and van Ulden, 1987; Justino et al., 2006). In wheat and sorghum, Al significantly inhibited NRA in shoots rather than roots (Foy and Fleming, 1982; Keltjens and van Ulden, 1987). In contrast, Al inhibited NRA in roots rather than shoots in red spruce (Cumming and Brown, 1994). The inhibitory effect of Al on NRA may result from Al-inhibition of NO<sup>3</sup> <sup>−</sup> uptake, as the decreased level of the substrate, NO<sup>3</sup> −, would lead to decreased NRA (Gomes et al., 1985; Keltjens and van Ulden, 1987; Keltjens, 1988; Justino et al., 2006; Pal'ove-Balang and Mistrík, 2007; Souza et al., 2016). The Al-induced decrease in NO<sup>3</sup> <sup>−</sup> content in plants was proposed to be the main mechanism by which Al inhibits NRA, so the interaction between Al and NR may be indirect. Roots generally accumulate more Al than do shoots. However, Al significantly inhibited NRA in the shoots but not in roots of wheat and sorghum (Foy and Fleming, 1982; Keltjens and van Ulden, 1987), suggesting that a direct interaction between NR and Al is unlikely. The ratio of absorbed <sup>15</sup>NO<sup>3</sup> <sup>−</sup> to reduced ammonia-containing N remained constant with increasing Al, also suggesting an indirect effect of Al on NR (Rufty et al., 1995). However, in another study, Al inhibited the shoot NRA of sorghum, and this could not be reversed by increased NO<sup>3</sup> <sup>−</sup> concentrations (Cambraia et al., 1989). Aluminum decreased NO<sup>3</sup> <sup>−</sup> accumulation in cucumber roots and maize leaves but enhanced their NRA (Lidon et al., 1998; Jerzykiewicz, 2001).

In some studies, Al was found to increase NRA (**Table 5**). At low concentrations, Al stimulated NRA in spruce (<37 µM Al; Peuke and Tischner, 1991) and rice (80 µM Al; Sharma and Dubey, 2005). Aluminum stimulated NRA in the Al-preferring species Quercus serrata (Tomioka et al., 2007, 2012) and tea

TABLE 5 | Summary of effects of aluminum on nitrate reductase activity: (−) inhibition, (+) stimulation, (0) no change and (N) not studied.


a study was conducted using sand culture irrigated with nutrient solutions. <sup>b</sup>Study was conducted using vermiculite culture irrigated with nutrient solutions. Other studies not marked with superscript letters were conducted using hydroponic systems.

(Hajiboland et al., 2014). The production of NO mediated by NR alleviated Al toxicity in red kidney bean, wheat, and soybean by alleviating oxidative stress, where Al significantly enhanced NRA in root tips (Wang et al., 2010, 2017; Sun et al., 2014). In another study, Al more strongly promoted NRA in Al-tolerant wheat than in Al-sensitive wheat (Sun et al., 2014).

The interaction between Al and NR appears to be complex, and can be positive or negative, direct or indirect. Many environmental factors are known to modulate NRA (Tischner, 2000). In various studies, the effects of A1 on NRA depended on the plant genotype (Foy and Fleming, 1982; Keltjens and van Ulden, 1987; Justino et al., 2006; Sun et al., 2014; Mihailovic et al., 2015), plant species (Dinev and Stancheva, 1993), plant part (Foy and Fleming, 1982; Keltjens and van Ulden, 1987), medium pH (Yandow and Klein, 1986), Al levels (Peuke and Tischner, 1991; Sharma and Dubey, 2005), N source and levels (Cumming, 1990; Mihailovic et al., 2015; Gupta et al., 2016), and inoculation treatments (Cumming, 1990). Although the Al–NR interaction is complex, we can conclude that NRA is generally inhibited by high Al concentrations, and stimulated by low Al concentrations (**Figure 2**). This overall trend is similar to the effects of Al on NO<sup>3</sup> <sup>−</sup> uptake, because NO<sup>3</sup> <sup>−</sup> is the primary factor regulating NRA.

Further research with detailed and well-designed experiments using different plant materials is necessary to clarify the details of the interaction between NR and Al. Recently, several genes encoding NR in maize (Zea mays) were found to be differently modulated at the transcriptional level by Al toxicity (Cantú et al., 2016). Molecular biology techniques could be helpful to clarify the detailed mechanisms of the interaction between Al and NR as well as NO<sup>3</sup> <sup>−</sup> uptake.

#### EFFECTS OF Al ON NH<sup>4</sup> <sup>+</sup> ASSIMILATION

In plants, NH<sup>4</sup> <sup>+</sup> is mainly assimilated by the GS/GOGAT (glutamine synthetase/glutamate synthase) cycle, where GS catalyzes the reaction between NH<sup>4</sup> <sup>+</sup> and glutamate to form glutamine. Glutamine subsequently combines with 2 oxoglutarate in a reaction catalyzed by GOGAT to form two molecules of glutamate (Masclaux-Daubresse et al., 2010). Glutamate dehydrogenase (GDH) is considered to be an alternative pathway to incorporate NH<sup>4</sup> <sup>+</sup> into glutamate when plants are exposed to high NH<sup>4</sup> <sup>+</sup> concentrations under stress. However, there is more evidence that GDH functions mainly in glutamate deamination (Masclaux-Daubresse et al., 2010). The presence of Al was shown to decrease the concentrations of NO<sup>3</sup> <sup>−</sup>-N and asparagine but increase the concentrations of amino acid-N and glutamine in the xylem sap of sorghum plants, potentially indicating that Al interferes with the synthesis and/or interconversion of N in plants (Gomes et al., 1985).

Pécsváradi's research group reported the activating effect of the Al(III)-tartrate 1:3 complex and the Al(III)–nitrilotriacetic acid complex on the activity of GS extracted from roots and leaves of wheat (Kertész et al., 2002; Pécsváradi et al., 2009). This activating effect was attributable to the specific binding of Al to the protein chain of GS, similar to the role of Mg in activating GS activity (Pécsváradi et al., 2009). Except for those two reports (Kertész et al., 2002; Pécsváradi et al., 2009), all of the other studies summarized here reported Al inhibition of GS activity in both roots and shoots (**Table 6**). However, Al either activated, suppressed, or did not affect the activities of GOGAT and GDH (**Table 6**). The effects of Al on the activities of N-assimilating enzymes were found to vary between Al-tolerant and Al-sensitive maize varieties and depend on the N form supplied. In maize, NH<sup>4</sup> <sup>+</sup> facilitated the Al stimulation of N assimilation in the roots of an Al-tolerant maize genotype (Mihailovic et al., 2015). Here, we suggest that Al might stimulate GS activity by binding to it, or inhibit it by limiting NH<sup>4</sup> <sup>+</sup> uptake (**Figure 2**). However, it is difficult to draw clear conclusions about the interaction between Al and NH<sup>4</sup> <sup>+</sup> assimilation on the basis of studies published to date. Therefore, more research is required to explore the effects of Al on these enzymes involved in NH<sup>4</sup> <sup>+</sup> assimilation.

## CONCLUDING REMARKS

A complex interaction between Al and N occurs in the soil– plant system. Relative to NO<sup>3</sup> <sup>−</sup>, NH<sup>4</sup> <sup>+</sup> uptake by roots generally alleviates Al phytotoxicity under solution culture conditions,

TABLE 6 | Summary of effects of aluminum on the activities of glutamine synthetase (GS), glutamate synthase (GOGAT), and glutamate dehydrogenase (GDH): (−) inhibition, (+) stimulation, (0) no change, and (N) not studied.


a study was conducted using sand culture irrigated with nutrient solutions. Other studies were conducted using hydroponic systems.

while NH<sup>4</sup> <sup>+</sup> aggravates the solubilization of toxic Al from soils into rhizosphere solutions. Both the alleviation and aggravation effects mainly result from NH<sup>4</sup> <sup>+</sup>-induced H<sup>+</sup> excretion due to NH<sup>4</sup> <sup>+</sup> uptake by plant roots and/or soil nitrification.

Compared with the effects of N on Al, the effects of Al on N are much more complicated because N is involved in multiple physiological processes within plants. Many reports have demonstrated that Al toxicity inhibits NO<sup>3</sup> <sup>−</sup> uptake by plant roots because Al binds to the NO<sup>3</sup> <sup>−</sup> transporter and stimulates NO<sup>3</sup> <sup>−</sup> efflux. In some cases, such as low Al concentrations, shortterm Al exposure, and Al-preferring plants, the Al stimulation of NO<sup>3</sup> <sup>−</sup> uptake is probably because of an increase in the positive electrical charge at the root-surface, enhanced H+-NO<sup>3</sup> − cotransport, and diminished NO<sup>3</sup> <sup>−</sup> efflux. The inhibitory effect of Al is generally smaller for root NH<sup>4</sup> <sup>+</sup> uptake than for NO<sup>3</sup> − uptake. Similar to the Al inhibition of NO<sup>3</sup> <sup>−</sup> uptake, the activity of NR can be inhibited by Al treatment because of decreased internal NO<sup>3</sup> <sup>−</sup> accumulation. Low concentrations of Al can stimulate NR activity as a result of stimulating NO<sup>3</sup> <sup>−</sup> uptake. The effects of Al on the activities of GS, GOGAT, and GDH are still uncertain.

Despite the diverse interactions between Al and N in many studies as described above, it is clear that Al-tolerant plants generally prefer NH<sup>4</sup> <sup>+</sup>, while Al-sensitive plants prefer NO<sup>3</sup> −. This relationship between plant Al tolerance and NH<sup>4</sup> <sup>+</sup>/NO<sup>3</sup> − preference may be the result of ecological evolution and natural selection because acid soils are characterized by a relatively higher ratio of NH<sup>4</sup> <sup>+</sup> to NO<sup>3</sup> <sup>−</sup> and higher concentrations of toxic Al than are neutral to calcareous soils.

Together, the results of numerous studies have suggested that the synergistic interaction between plant Al tolerance and NH<sup>4</sup> +- N nutrition may be an important strategy of plants to thrive in acid soils dominated by both toxic Al and NH<sup>4</sup> <sup>+</sup>. In addition, the Al stimulation of N uptake and assimilation can help to explain why Al stimulates plant growth in some cases.

Many studies have focused on the interactions between Al and N in plants, but the exact mechanisms underlying these interactions are still unclear. The Al–N interactions have been studied mainly at the physiological level rather than the molecular level. Physiological effects are indirectly affected by many factors and are not specific. Many genes that function in N uptake, N assimilation, and Al tolerance/toxicity have been identified (Masclaux-Daubresse et al., 2010; Ryan et al., 2011; Schroeder et al., 2013; Ma et al., 2014). The use of mutants with knocked-out or knocked-down expression of these genes could be helpful to explore the detailed mechanisms of Al–N interactions. In addition, we emphasize the importance of soil experiments for researching Al–N interactions, because the ultimate goal of understanding Al– N interactions is to improve the growth of plants in soils. Unfortunately, most studies on Al–N interactions have been conducted under solution culture conditions. As discussed above, the Al–N interactions in solutions may differ from those in soils.

How can the existing knowledge of Al–N interactions be used to improve the productivity of plants grown in acid soils? Plants need to overcome the dual limitation of Al toxicity and N deficiency in acid soils. Due to poor nitrification, acid soils have a higher NH<sup>4</sup> <sup>+</sup> to NO<sup>3</sup> <sup>−</sup> ratio than do neutral to calcareous soils. Large-area forest decline has been linked to both NH<sup>4</sup> + toxicity and soil acidification, and NH<sup>4</sup> <sup>+</sup> toxicity has become an important issue in global agriculture and ecology (Britto and Kronzucker, 2002). Symptoms of NH<sup>4</sup> <sup>+</sup> toxicity, such as leaf chlorosis, growth suppression, and even death generally appear when the external NH<sup>4</sup> <sup>+</sup> concentrations exceed 0.1 to 0.5 mM, depending on the plant (Britto and Kronzucker, 2002). Thus, any enhancements in plant Al tolerance in acid soils should be accompanied by improvements in plant NH<sup>4</sup> + utilization or reduced plant NH<sup>4</sup> <sup>+</sup> sensitivity. Although NH<sup>4</sup> + supply generally enhances plant Al tolerance, it also increases the concentrations of toxic Al in soils and leads to potentially toxic NH<sup>4</sup> <sup>+</sup> concentrations. How can we solve this contradiction? Which type of N fertilizer should be applied in acid soils, NH<sup>4</sup> + or NO<sup>3</sup> <sup>−</sup>? The NO<sup>3</sup> <sup>−</sup> fertilizers are much more expensive than NH<sup>4</sup> <sup>+</sup> fertilizers. In addition, NO<sup>3</sup> <sup>−</sup> is lost to water more readily than is NH<sup>4</sup> <sup>+</sup> because NO<sup>3</sup> <sup>−</sup> binds weakly to soil particles, which are generally negatively charged. Therefore, applying NO<sup>3</sup> <sup>−</sup> fertilizers to acid soils appears to be impractical at the moment.

Fortunately, plants originating from acid soils are generally both Al-tolerant and NH<sup>4</sup> <sup>+</sup>-preferring. Thus, one way to increase productivity from acid soils is to breed and develop genotypes that are both Al-tolerant and NH<sup>4</sup> <sup>+</sup>-preferring. This strategy may synergistically enhance plant Al tolerance and N-use efficiency, and reduce NH<sup>4</sup> <sup>+</sup> sensitivity and NO<sup>3</sup> <sup>−</sup> loss. The improvement of N-use efficiency could reduce the amounts of N fertilizers applied to soils, thereby alleviating soil acidification and Al toxicity. Recently, an in situ <sup>15</sup>N-labeling experiment showed that soluble soil Al inhibited the relative uptake of NO<sup>3</sup> <sup>−</sup> by six tree species, potentially increasing NO<sup>3</sup> <sup>−</sup> loss from acid soils into the surrounding water environment (Burnham et al., 2017). Thus, knowledge about Al–N interactions is important for agriculture, ecology, and the environment.

## AUTHOR CONTRIBUTIONS

XZ wrote the manuscript. RS checked and revised the manuscript.

## FUNDING

This work was supported financially by the National Natural Science Foundation of China (No. 31672229), the Strategic Priority Research Program of the Chinese Academy of Sciences (Nos. XDB15030202 and XDB15030302), and the National Key Basic Research Program of China (No. 2014CB441000).

#### ACKNOWLEDGMENTS

We are grateful to the National Natural Science Foundation of China, the Chinese Academy of Sciences, and the Ministry of Science and Technology of the People's Republic of China. We thank the Editor and two reviewers for their nice and detailed comments. We also thank Jennifer Smith, Ph.D., from Liwen

#### REFERENCES

fpls-09-00807 June 14, 2018 Time: 17:48 # 12


Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.



carbon in a pilot-scale drinking water treatment plant with different treatment processes. Water Sci. Technol. 61, 3070–3077. doi: 10.2166/wst.2010.204




elongation by regulating the ascorbate-glutathione cycle in soybean roots. Plant Soil 410, 453–465. doi: 10.1007/s11104-016-3045-4


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhao and Shen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Bioinformatic and Functional Analysis of a Key Determinant Underlying the Substrate Selectivity of the Al Transporter, Nrat1

Muxue Lu† , Guangzhe Yang† , Peifang Li† , Zhigang Wang, Shan Fu, Xiang Zhang, Xi Chen, Mingxing Shi, Zhenhua Ming\* and Jixing Xia\*

State Key Laboratory of Conservation and Utilization of Subtropical Agro-Bioresources, College of Life Science and Technology, Guangxi University, Nanning, China

#### Edited by:

Wei Fan, Yunnan Agricultural University, China

#### Reviewed by:

He Qiang Lou, Zhejiang Agriculture and Forestry University, China Dezhi Wu, Zhejiang University, China Miguel A. Pineros, United States Department of Agriculture (USDA), United States

#### \*Correspondence:

Jixing Xia xiajx@gxu.edu.cn Zhenhua Ming zhming@gxu.edu.cn †These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science

Received: 23 December 2017 Accepted: 17 April 2018 Published: 07 May 2018

#### Citation:

Lu M, Yang G, Li P, Wang Z, Fu S, Zhang X, Chen X, Shi M, Ming Z and Xia J (2018) Bioinformatic and Functional Analysis of a Key Determinant Underlying the Substrate Selectivity of the Al Transporter, Nrat1. Front. Plant Sci. 9:606. doi: 10.3389/fpls.2018.00606 Nrat1 is a member of the natural resistance-associated macrophage protein (Nramp) family of metal ion transporters in all organisms. Different from other Nramp members capable of transporting divalent metals, Nrat1 specifically transports trivalent aluminum (Al) ion. However, molecular mechanism underlying the Al transport selectivity of Nrat1 remains unknown. Here, we performed structure-function analyses of Nrat1 and other Nramp members to gain insights into the determinants of ion selectivity. A phylogenetic analysis showed that plant Nramp transporters could be divided into five groups. OsNrat1 was found in one of the individual clades and clustered with SbNrat1 and ZmNrat1 on the evolutionary tree. Structural modeling revealed that Nrat1 transporters adopted a common LeuT fold shared by many Nramp-family transporters that likely employed an identical transport mechanism. Sequence alignment and evolutionary conservation analysis of amino acids identified a metal-permeation pathway of Nrat1 centered at the metal binding site. The metal binding site of Nrat1 was characterized by two conserved sequence motifs, i.e., the Asp-Pro-Ser-Asn motif (motif A) and the Ala-Ile-Ile-Thr motif (motif B). Replacement of the Ala-Met-Val-Met motif B of the OsNramp3 manganese (Mn) transporter to that of Nrat1 resulted in a partial gain of Al transport activity and a total loss of Mn in yeast. Conversely, substitution of the motif B of OsNrat1 with that of OsNramp3 altered the Al transport activity. These observations indicated the metal binding site, particularly the motif B, as a key determinant of Al selectivity of Nrat1.

Keywords: Nrat1, aluminum, Al transporter, selectivity, bioinformatic analysis

## INTRODUCTION

The natural resistance-associated macrophage proteins (Nramps) are widely presented in bacteria, fungi, plants, and mammals (Curie et al., 2000; Nevo and Nelson, 2006). They function as metal ion transporters for a wide range of divalent metal substrates such as Fe2+, Mn2+, Cd2+, Zn2+, Co2+, Ca2+, Cu2+, Ni2+, and Pb2<sup>+</sup> (Gunshin et al., 1997). In higher plants, Nramp proteins play major roles in the transport of mineral elements from soil to different organs and tissues of plants. For example, AtNramp1 is found to be localized at the plasma membrane of root cells and functions as a high-affinity transporter for Mn uptake in Arabidopsis (Cailliatte et al., 2010). AtNramp3 and

**50**

AtNramp4 function redundantly to release Fe and Mn from the vacuole (Thomine et al., 2000; Lanquar et al., 2005, 2010). In rice, OsNramp1 transports Fe and Cd in yeast and is suggested to be involved in Cd uptake (Takahashi et al., 2011). OsNramp3 is localized at the plasma membrane of node cells and is involved in distribution of Mn, but not Fe and Cd (Yamaji et al., 2013). The plasma membrane-localized transporter OsNramp5 is the major contributor for Mn and Cd uptake (Sasaki et al., 2012).

Recently, OsNrat1, an Nramp member, was reported to specifically transport Al3<sup>+</sup> but not divalent metal ions such as Fe2+, Mn2+, and Cd2+, and required for Al tolerance in rice (Xia et al., 2010). In sorghum, SbNrat1, a close homolog of rice OsNrat1, also was shown to selectively transport Al3<sup>+</sup> (Lu et al., 2017). However, the molecular mechanisms underlying the Al transport selectivity of Nrat1 remain unknown.

Several studies have investigated the relationships between the structure and the function in Nramp proteins. For instance, mutational analysis of the first external loop (Loop I) of NRAMP2/DCT1/DMT1 suggested that Loop I is involved in metal ion binding and specificity (Cohen et al., 2003). The mutation (G185R) in NRAMP2/DCT1/DMT1 not only resulted in a decrease in iron transport but increased the permeability to calcium (Xu et al., 2004). In Arabidopsis, three residues (L67, E401, F413) of AtNramp4 have been also shown to play important roles in metal selectivity (Pottier et al., 2015). On the other hand, the crystal structural studies have revealed that Nramp proteins shared a conserved protein fold that was previously found in the amino acid transporter LeuT (Cellier, 2012; Ehrnstorfer et al., 2014). The ScaNramp structure also revealed that a metal binding site consists of conserved aspartate, asparagines, and methionine residues, and a backbone carbonyl from transmembrane segments (TMs) 1 and 6 (Ehrnstorfer et al., 2014). Moreover, the conserved metal-binding site methionine was shown to confer selectivity against the abundant alkaline earth metals calcium and magnesium (Bozzi et al., 2016a). However, the role of the conserved metal-binding site in controlling substrate selectivity is still poorly understood.

In this study, we compared the structure and function of Nrat1 and other initially reported Nramp members in plants by phylogenetic analysis and homology modeling. Furthermore, we performed the site-direct mutagenesis analysis of the conserved metal binding motif in two Nramp proteins, OsNrat1 and OsNramp3, which are known as transporters for Al and Mn (Xia et al., 2010; Yamaji et al., 2013), respectively, and examined their transport activities for Al and Mn. Our results identified a key determinant of Al selectivity of Nrat1, which is essential for Mn selectivity of OsNramp3. It provides novel insights into the molecular basis of Al transport selectivity of Nrat1 and valuable clues to investigate Mn transport selectivity of OsNramp3.

## MATERIALS AND METHODS

#### Sequence and Structure Collection

The amino acid sequences of OsNrat1 homologs from four types of plants, Oryza sativa, Arabidopsis thaliana, Sorghum bicolor, and Zea mays, were obtained by BLAST (Johnson et al., 2008) using the OsNrat1 sequence as a query in the U.S. National Center for Biotechnology Information (NCBI) reference sequence (RefSeq) database. After eliminating the repetitive sequences, we collected a total of 24 sequences. The structures of prokaryotic Nramp transporters were downloaded from the Protein Data Bank (PDB) database. The 25 Nramp transporters and their NCBI accession numbers are as follows: OsNramp1, XP\_015647629; OsNramp2, XP\_015632573; OsNramp3, XP\_015644306; OsNrat1, XP\_015625418; OsNramp5, XP\_015645014; OsNramp6, XP\_015620405; OsNramp7, XP\_015618209; AtNramp1, NP\_178198; AtNramp2, NP\_175157; AtNramp3, NP\_179896; AtNramp4, NP\_201534; AtNramp5, NP\_193614; AtNramp6, NP\_173048; SbNramp1, XP\_002459640; SbNramp2, XP\_002465667; SbNramp3, XP\_002438846; SbNramp4, XP\_021317241; SbNramp5, XP\_002461772; SbNramp6, XP\_002464246; SbNrat1, XP\_002451480; ZmNramp1, XP\_008670084; ZmNramp4, XP\_008670762; ZmNramp5, XP\_008652227; ZmNramp6, XP\_008665146; ZmNrat1, NP\_001334019.

### Sequence Alignment and Phylogenetic Analysis

Multiple-sequence alignment (MSA) was performed by using the T-Coffee server (Di Tommaso et al., 2011). The alignment was produced by combining multiple methods, including mafft\_msa, clustalw\_msa and t\_coffee\_msa. Results were subjected to figure production by ESPript version 3.0 (Robert and Gouet, 2014), evolutionary tree building by MEGA6 (Tamura et al., 2013), or evolutionary conservation analysis by ConSurf (Ashkenazy et al., 2010).

Phylogenetic analysis was conducted in MEGA version 6 by the bootstrap neighbor joining method (Saitou and Nei, 1987). Bootstrap method (Felsenstein, 1985) was used for test of phylogeny and the number of bootstrap replications was set to 1000. The evolutionary distances were calculated using the Poisson correction method (Zuckerkandl and Pauling, 1965) and were in the units of the number of amino acid substitutions per site. The analysis involved all 25 amino acid sequences of the Nramp family transporters in the four types of plants. All positions containing gaps and missing data were eliminated. There was a total of 450 positions in the final dataset.

## Evolutionary Conservation Analysis

MSA of the 25 plant Nramp transporters constructed by T-Coffee and the I-TASSER model for the core domain of OsNrat1 (45–502) was used to calculate the position-specific conservation scores by the empirical Bayesian algorithms (Mayrose et al., 2004) in ConSurf (Ashkenazy et al., 2010). The continuous conservation scores are divided into a discrete scale of nine grades for visualization, from the most variable positions (grade 1) colored turquoise, through intermediately conserved positions (grade 5) colored white, to the most conserved positions (grade 9) colored maroon. Scripts for visualizing the protein colored with ConSurf scores were generated and the colored protein was shown in PyMOL (DeLano, 2002).

Normalized conservation scores were also extracted and used to calculate the average conservation score for each structural element and produce figures in GraphPad Prim version 5.

#### Vector Construction

fpls-09-00606 May 3, 2018 Time: 20:41 # 3

The coding region of OsNrat1 and OsNramp3 was amplified from rice (Oryza sativa, Nipponbare) root cDNA with high-fidelity PCR (KOD Fx polymerase, Toyobo), and the amplified fragments were cloned into the HindIII/EcoRI, BamHI/EcoRI restriction sites of yeast expression vector pYES2 (Invitrogen), respectively. Site-directed mutagenesis of OsNrat1 and OsNramp3 was performed by overlapping PCR (Ho et al., 1989). The wild-type and mutated OsNrat1 or OsNramp3 CDS were verified by sequencing. All the PCR primers used are listed in Supplementary Table S1.

#### Yeast Assays

The yeast strains used in this study were BY4741 (MATa his210 met1510 ura310) and smf1 (MATa his210 met1510 ura310 YOL122c::KanMX4). Al sensitivity test on agar and complementation of the smf1 phenotype were performed as described by Xia et al. (2010). For Al sensitivity evaluation, OsNrat1, OsNramp3, mutated OsNrat1, or OsNramp3, and vector control pYES2 were introduced into yeast strain BY4741 and then spotted on solid media (LPM with 2% galactose for induction of the GAL promoter) containing 0, 200, or 300 µM AlCl<sup>3</sup> buffered with 5 mM succinic acid. For Al uptake in liquid culture, transformants were selected on uracil-deficient medium and grown in synthetic complete (SC-uracil) yeast solution containing 2% glucose. Cells at mid-exponential phase were harvested and transferred to LPM medium containing 2% galactose. Cells were cultured for 2 h. Then AlCl<sup>3</sup> was added to the cell culture at the final concentration of 50 µM AlCl3. After 6 h incubation with shaking, cells were harvested by centrifugation at 12000 × g for 5 min, and washed three times with deionized water (MilliQ; Millipore), dried and then digested with 65% HNO3. Al concentration was measured by inductively coupled plasma optical emission spectrometry.

## RESULTS

#### Plant Nramp Transporters Fall Into Five Groups on the Phylogenetic Tree

The amino acid sequence of OsNrat1 was used to retrieve Nramp homologs in four plant species by BLAST (Johnson et al., 2008). Twenty-four protein sequences were selected for phylogenetic analysis, including six from Oryza sativa (OsNrat1, OsNramp1- 3, and 5-7), six from Arabidopsis thaliana (AtNramp1-6), seven from Sorghum bicolor (SbNrat1 and SbNramp1-6), and five from Zea mays (ZmNrat1, ZmNramp1, and ZmNramp4- 6). The resulting phylogenetic tree includes five main clades corresponding to five distinctive Nramp groups (**Figure 1**). Notably, two known transporters for trivalent Al ion, OsNrat1 (Xia et al., 2010, 2014; Li et al., 2014) and SbNrat1 (Lu et al., 2017), an Nrat1-like transporter of Zea mays (ZmNrat1) as well, are located in the same clade (group III) but separated from other divalent ion transporters on the evolutionary tree.

#### Nrat1 Transporter Adopts a Conserved LeuT Fold

Sequence alignment of the 25 plant Nramp transporters and the Staphylococcus Capitis Divalent Metal Ion Transporter (ScaDMT) reveals that they contain a conserved core domain (amino acid sequence 45–502 in OsNrat1, see Supplementary Figure S1). The available structural information on the Nramp family is limited to the prokaryotic homologues of divalent metal transporters. To understand the mechanism in which trivalent Al is recognized by Nrat1 transporter, we modeled the protein structures of core domain of OsNrat1 and OsNramp3 through an iterative threading algorithm using the I-TASSER server (Roy et al., 2010), as the terminal N- and C- regions with unknown functionality are not as important as the core domain that is highly conserved and constitutes a part of the molecular determinants for ion permeation. To compare the architecture of metal binding sites in the ionic binding state of these two proteins, the Mn-binding structure of ScaDMT (PDBID: 5M95), which showed the highest sequence similarities with OsNrat1 (59%) and OsNramp3 (62%), was used as a template. The estimated TM-score and C-score of OsNrat1 are 0.95 ± 0.05 and 1.68, respectively; while those of OsNramp3 are 0.92 ± 0.06 and 1.50, respectively. Hence, these two models appear to be acceptable.

OsNrat1 adopts a common LeuT fold (**Figure 2**) that is associated with many prokaryotic Nramp-family transporters, including the Staphylococcus divalent metal transporter ScaDMT (59% sequence similarity, PDBID: 5M95), the Deinococcus radiodurans Nramp homolog (DraNramp, 56% similarity, PDBID: 5KTE), and the Eremococcus coleocola Manganese Transporter (EcoDMT, 56% similarity, PDBID: 5M87). OsNrat1 contains a compact globular domain of 12 transmembrane segments (TMs), of which TMs 1–5 and 6–10 form two inverted repeats of the LeuT fold. Like other LeuT-type transporters, the first TM in each of the two inverted repeats (TM1 and TM6) of OsNrat1 contains two α-helices disrupted by a short discontinuous stretch in the middle (Supplementary Figure S1). Overall, a helical bundle comprising TMs 3–5 and 8–10 forms a semicircular (letter C shaped) structure that wraps partway around a second helical bundle formed by TMs 1, 2, 6, and 7. Substrate transport of the LeuT-type transporters is likely to be coupled with a switch from outward-open to inward-open conformation, through a rigid-body rotation (Shi, 2013) of the moving portion (corresponding to the latter α helical bundle) related to the non-moving portion (corresponding to the former helical bundle).

### The Metal Transport Mechanism of Nrat1 Is Conserved

Evolutionary conservation analysis by the ConSurf server (Ashkenazy et al., 2010) reveals an overwhelming conservation of residues that make up the interior of the Nrat1 cylinder structure (**Figure 2A**), which contains a substrate transport path along the central axis that is perpendicular to the lipid

membrane plane. A detailed conservation analysis for each structural element was performed by calculating the normalized evolutionary conservation scores on all amino acid residues. As shown in **Figure 2B**, TMs 1-3, 6, 8-10, along with the L23 loop are highly conserved across all 25 Nramp transporters in plants, suggesting that these conserved elements may play important roles in metal transport. On the contrary, structural elements of NT, CT, TM11 and TM12 in the Nramp family are variable. These observations are consistent with structural and functional analysis of other LeuT-type transporters, indicating a conserved substrate transport mechanism. Similar to other known LeuTtype transporters, the five TMs (TMs 1, 3, 6, 8, and 10) may participate directly in substrate binding and transport. TM2 and TM9, which connect the functional helices TM1/TM3 and TM8/TM10, respectively, may confer transport activity through control of the local conformation. The L23 linker that connects neighboring helices of TM2 and TM3 may parallel the roles of TM2 and TM9.

### The Nrat1 Transporter Contains a Unique Ion Binding Site

The metal recognition site of prokaryotic Nramp transporters is known to be characterized by two structural motifs (**Figure 3** and Supplementary Figure S1), motifs A and B. The highly conserved Asp-Pro-[Gly/Ser]-Asn motif (motif A) occurs in the loop between TM1a and TM1b, as well as the N-terminal portion of TM1b; while the moderate conserved motif (motif B) occurs in the C-terminal portion of TM6a and the loop between TM6a and TM6b. Only four residues, i.e., the first and fourth residues in each motif, are required to coordinate the central metal ion (**Figure 3D**). The Asp and Asn residues in motif A (locating at the first and fourth position, respectively), as well as the fourth residue in motif B, use their side chains to contact the metal ion. By contrast, the first residue in motif B contributes to metal binding by its main-chain carbonyl oxygen. In support of the structural and functional importance of the metal coordination ligands, the Asp and Asn residues in motif A are invariant within the Nramp family and across plant species (**Figure 3A** and Supplementary Figure S1). Notably, the metal ligands in motif B are only moderately conserved (**Figure 3A** and Supplementary Figure S1), suggesting that this motif may contribute to ion-subtype specificity. It is also worth noting that, through refining the proper conformation of terminal residues in each motif, interspace residues of the two motifs may be important for ion binding and selectivity as well.

Sequence alignment results for the two signature motifs show good agreement with the phylogenetic analysis of plant Nramp transporters (**Figure 3A**). Transporters of the Nrat1 group and group II have an Asp-Pro-Ser-Asn sequence pattern of motif A, while other groups consist of a motif A with a uniform Asp-Pro-Gly-Asn sequence. The characteristic sequence patterns of motif B can be clearly divided into five sets, each corresponding to one of the five phylogenetic groups. Among the five sets of motif B, sequences of groups I, III, and IV have the highest conservation. Motif B of group III transporters has an invariant sequence of Ala-Ile-Ile-Thr, while motif B of group I and IV has an identical sequence of Ala-Leu-Val-Met and Ala-Met-Val-Met, respectively. Together, these data indicated that Nrat1 transporters contain a unique pair of sequence motifs which may be critical for mediating metal recognition.

most conserved in this specific protein calculated using a specific multiple sequence alignment (MSA). Error bars represent SEMs. The NT and CT regions of OsNrat1 are not shown in (A).

crystal structure of ScaDMT by the software PyMOL. d1 is the diameter of the binding site circle measured in the figure. d2 is the distance between the main-chain oxygen of Ala223 and the side-chain oxygen of Asn52 of ScaDMT measured in the figure.

We proceeded to compar the putative metal binding sites from the core domain structural models of OsNrat1 and OsNramp3, with that of the crystal structure of ScaDMT. As shown in **Figures 3B–D**, the architecture of the divalent ion recognition site in OsNramp3 is identical to that observed in ScaDMT. By contrast, the trivalent metal binding site in OsNrat1 appears to be slightly larger than those in OsNramp3 and ScaDMT, as calculated in the two models and the ScaDMT crystal structure (**Figures 3B–D**). These differences are likely caused by a replacement of the Met with a Thr, which contains a shorter side-chain compared to that of a Met, at the fourth residue of motif B.

### The Nrat1 Specific Motif B Is a Key Determinant for Al Transport

To determine the functional importance of the signature motifs, we generated several mutations for OsNrat1 and OsNramp3 (OsNrat1I240M,I241V,T242M, OsNrat1T242M, OsNramp3M239I,V240I,M241T, OsNramp3M241T, see Supplementary Table S2) by exchanging corresponding residues of one protein with another and examined their capabilities on Al and Mn transport as well as OsNrat1 or OsNramp3 as a positive control, respectively.

In the absence of Al, all the transformants showed similar growth on the plate (**Figure 4A**). However, in the presence of Al, the growth of yeast cells carrying OsNrat1T242M, OsNramp3M239I,V240I,M241T, or OsNrat1 was significantly inhibited compared with that of the vector control, while that of OsNrat1I240M,I241V,T242M, OsNramp3M241T, or OsNramp3 was not (**Figure 4A**). Al uptake also significantly increased in the yeast carrying OsNrat1T242M, OsNramp3M239I,V240I,M241T , or OsNrat1 and was not affected in the yeast carrying OsNrat1I240M,I241V,T242M, OsNramp3M241T, or OsNramp3 (**Figure 4B**). Furthermore, the Al uptake ability of OsNrat1T242M or OsNramp3M239I,V240I,M241T was lower than that of OsNrat1 (**Figure 4B**). These results suggested that substitution of the intact motif B of OsNrat1 (OsNrat1I240M,I241V,T242M) with that of OsNramp3 completely deprived the Al transport activity of OsNrat1, while a single mutation on the fourth residue Thr242Met (OsNrat1T242M) resulted in a decrease in Al uptake of OsNrat1, and that replacement of the intact motif B of OsNramp3 (OsNramp3M239I,V240I,M241T) with that of OsNrat1, but not a single mutation on the fourth residue Met241Thr (OsNramp3M241T), rendered the Mn specific divalent transporter to gain a function of Al transport.

OsNramp3M239I,V240I,M241T , OsNramp3M241T were spotted on LPM without uracil medium (pH 4.2) buffered with 5 mM succinic acid with or without AlCl<sup>3</sup> at serial dilutions (from left to right: 10 µl cell suspension with OD 0.2, 0.02, 0.002, and 0.0002) and incubated at 30◦C for 3 days. (B) Transport activity of mutated OsNrat1 or OsNramp3 for Al. Yeast cells expressing different mutants were exposed to a solution containing 50 µM AlCl<sup>3</sup> (pH 4.2) for 6 h. Data are mean ± SD of three biological replicates. Different letters above the bars indicate significant differences (P < 0.05, Tukey's test). (C) Complementation of manganese uptake. Transformed smf1 were grown on a medium (pH 6.0) buffered with 50 mM MES in the presence or absence of EGTA. The plates were incubated at 30◦C for 3 days.

Subsequently, we performed a complementation test in the 1smf1 yeast strain to examine whether mutants of OsNrat1 and OsNramp3 could have transport activity for Mn. As expected, OsNramp3 could restore the growth of a yeast mutant (smf1) defective in Mn uptake, while OsNrat1 could not (**Figure 4C**). Surprisingly, as shown in **Figure 4C**, all of the four above mentioned mutations failed to complement the manganese uptake phenotype of the 1smf1 mutant yeast. These observations collectively indicate that the Nrat1 specific motif B is both sufficient and required for Al transport, while the OsNramp3 specific motif B, especially the fourth ionic coordination ligand, is only required but not sufficient for Mn transport.

We also studied the functional importance of residues in close vicinity of the signature motifs by sequence exchange between OsNrat1 and OsNramp3 (OsNrat1A59F, G64A, OsNrat1Y244H, OsNramp3F58A, A63G, OsNramp3H243Y, see Supplementary Table S2). As shown in the Supplementary Figure S2, the expression of OsNrat1A59F,G64A, OsNrat1Y244H, or OsNrat1 increased the sensitivity of yeast to Al toxicity and the Al uptake in yeast compared with that of the vector control, while that of OsNramp3F58A,A63G, OsNramp3H243Y, or OsNramp3 did not (Supplementary Figures S2A,B). Furthermore, the Al uptake ability of OsNrat1A59F,G64A or OsNrat1Y244H was lower than that of OsNrat1 (Supplementary Figure S2B). On the other hand, in contrast to OsNramp3, the expression of OsNrat1A59F,G64A, OsNrat1Y244H, OsNramp3F58A,A63G, or OsNramp3H243Y was not able to complement the growth of the yeast mutant 1smf1 under the Mn-limited condition controlling by EGTA (Supplementary Figure S2C). These results indicated that flanking residues of the characteristic motifs are dispensable for Al selectivity but required, at least in part, for Al transport activity. These data also suggested that residues near the metal binding motifs of OsNramp3, but not OsNrat1, are essential for Mn uptake of the transporter.

#### DISCUSSION

A number of reported variations in the Nrat1 coding region affect transport activity (Li et al., 2014; Xia et al., 2014; Lu et al., 2017) but not selectivity of Nrat1. We reasoned that careful examination of the metal binding site may facilitate to understand the selectivity of the transporter. The Nramp family of transporters utilizes two separate motifs, each from one of the two discontinuous TMs, to coordinate metal ions (Ehrnstorfer et al., 2014, 2017; Bozzi et al., 2016b). Our bioinformatic and functional analyses demonstrate that the metal binding site, particularly the motif B with a sequence of Ala-Ile-Ile-Thr, is a prominent determinant of Al selectivity for Nrat1.

Motif B of OsNramp3 is probably essential for the selectivity of the transporter. However, the interpretation for selectivity of Nrat1 cannot be directly applied to give a simplified explanation for the selectivity of the Mn specific transporters of the Nramp family, as substitution of mere motif B in OsNrat1 by that of OsNramp3 is not sufficient for the former to gain Mn transport activity. This is consistent with the experimental observations for divalent Nramp transporters reported by Bozzi et al. (2016a). The conserved metal-binding methionine (Met230) of motif B is dispensable in the bacterial DraNramp, as the Met-to-Ala mutant can still enable robust transport of the physiological manganese substrate and similar divalent iron and cobalt. In sharp contrast to the DraNramp, the corresponding Met265Ala mutant of human Nramp2 did not transport any of the tested divalent metals, including Co, Mn, Cd, and Ca. These results indicate a dependency of the functional divergence on sequence and structure context (Bozzi et al., 2016a). Supporting this hypothesis, whereas the single mutation (corresponding to Gln76 of OsNrat1) in TM1b of the mammalian transporter DCT1 (Slc11a2) completely blocked Mn transport, a double mutation (corresponding to Asp74 and Gln76 of OsNrat1) in TM1b restored the activity and altered the metal ion specificity in favor of Fe (Cohen et al., 2003). Moreover, random mutagenesis studies revealed that three residues, Leu67 (in the immediate vicinity of motif A) from TM1a and Glu401/Phe413 from TM10, contributed to the selectivity of AtNramp4 for the uptake of another divalent metal Cd (Pottier et al., 2015).

Our work identified that the Nrat1-type motif B is both sufficient and required for Al transport in Nrat1 and OsNramp3, as one of the key determinants for the Al selectivity. Our results also suggested that the OsNramp3-type motif B is necessary, though not sufficient, for the Mn selectivity of OsNramp3. Identification of the important functions of motif B in substrate selectivity of Nrat1 and OsNramp3 may help further elucidate the selectivity of other Nramp transporters.

#### AUTHOR CONTRIBUTIONS

ZM and JX conceived and designed the experiments. ML, GY, PL, ZW, XZ, XC, SF, and MS performed the experiments. ML, GY, ZM, and JX analyzed the data. ZM and JX wrote the paper.

## FUNDING

This work was supported by Guangxi Natural Science Foundation (2015GXNSFCB139001, 2016GXNSFFA380013, 2016GXNSFBA380176), National Natural Science Foundation of China (31670253, 31700052), the Project of High Level Innovation Team and Outstanding Scholar in Guangxi Colleges and Universities (2016), and China Postdoctoral Science Foundation (2017M612863).

## ACKNOWLEDGMENTS

We thank Zheng Zhang (Shandong University) for helpful discussion.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00606/ full#supplementary-material

#### REFERENCES

fpls-09-00606 May 3, 2018 Time: 20:41 # 9


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Lu, Yang, Li, Wang, Fu, Zhang, Chen, Shi, Ming and Xia. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Functional Conservation and Divergence of Soybean GmSTOP1 Members in Proton and Aluminum Tolerance

Weiwei Wu<sup>1</sup> , Yan Lin<sup>1</sup> , Qianqian Chen<sup>1</sup> , Wenting Peng<sup>2</sup> , Junchu Peng<sup>1</sup> , Jiang Tian<sup>1</sup> , Cuiyue Liang<sup>1</sup> \* and Hong Liao<sup>2</sup>

<sup>1</sup> Root Biology Center, State Key Laboratory for Conservation and Utilization of Subtropical Agro-bioresources, South China Agricultural University, Guangzhou, China, <sup>2</sup> Root Biology Center, Fujian Agriculture and Forestry University, Fuzhou, China

Proton (H+) and aluminum (Al) rhizotoxicity are two major factors limiting crop production in acid soils. Orthologs of the zinc-finger transcription factor, Sensitive To Proton Rhizotoxicity1 (STOP1), have been found to play an essential role in the tolerance to both stresses by regulating the transcription of multiple H<sup>+</sup> and Al tolerant genes. In the present study, color three GmSTOP1 homologs were identified in the soybean genome. All three GmSTOP1 exhibited similar properties as reflected by the harboring of four potential zinc finger domains, localizing in the nucleus, and having transactivation activity. Expression profiling showed that H<sup>+</sup> stress slightly modulated transcription of all three GmSTOP1s, while Al significantly up-regulated GmSTOP1-1 and GmSTOP1-3 in root apexes and GmSTOP1-3 in basal root regions. Furthermore, complementation assays in an Arabidopsis Atstop1 mutant line overexpressing these GmSTOP1s demonstrated that all three GmSTOP1s largely reverse the H<sup>+</sup> sensitivity of the Atstop1 mutant and restore the expression of genes involved in H<sup>+</sup> tolerance. In contrast, only GmSTOP1-1 and GmSTOP1-3 could partially recover Al tolerance in the Atstop1 mutant. These results suggest that the function of three GmSTOP1s is evolutionarily conserved in H<sup>+</sup> tolerance, but not in Al tolerance.

Keywords: GmSTOP1, proton rhizotoxicity, Al rhizotoxicity, soybean, transcription factor

## INTRODUCTION

Agricultural production is limited on acid soils, which comprise approximately 50% of the world's potentially arable lands (von Uexküll and Mutert, 1995). There are several constraints limiting plant growth on acid soils, including deficiency of mineral nutrients, such as phosphorus (P), calcium (Ca), and magnesium (Mg), as well as, toxicity of excessive ions, including aluminum (Al3+), hydrogen (H+), and manganese (Mn2+) (Ishitani et al., 2004).

Among these stresses, Al toxicity has been widely acknowledged as a major constraint on crop production (Kochian et al., 2004; Ma, 2007; Bojórquez-Quintal et al., 2017). The Al3<sup>+</sup> ion can cause rapid and severe impairment of root apical development by damaging cell walls (Horst et al., 1999) and cytoskeletons (Chang et al., 1999; Sivaguru et al., 1999), disturbing DNA and plasma membrane processes (Elstner et al., 1988; Yamaguchi et al., 1999; Meriga et al., 2004), blocking

#### Edited by:

Jian Li Yang, Zhejiang University, China

#### Reviewed by:

Meiya Liu, Tea Research Institute (CAAS), China Jiangfeng You, Jilin University, China

> \*Correspondence: Cuiyue Liang liangcy@scau.edu.cn

#### Specialty section:

This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science

Received: 27 December 2017 Accepted: 11 April 2018 Published: 26 April 2018

#### Citation:

Wu W, Lin Y, Chen Q, Peng W, Peng J, Tian J, Liang C and Liao H (2018) Functional Conservation and Divergence of Soybean GmSTOP1 Members in Proton and Aluminum Tolerance. Front. Plant Sci. 9:570. doi: 10.3389/fpls.2018.00570

**59**

production of callose (Sivaguru et al., 2000), and impeding stresssignaling pathways (Ramos-Díaz et al., 2007). Consequently, plant root growth and nutrient acquisition are inhibited, which leads to significant reductions in crop yields (Ryan et al., 2001; Kochian et al., 2004; Ma, 2007; Bojórquez-Quintal et al., 2017).

Often combined with Al toxicity, H<sup>+</sup> rhizotoxicity has also been recognized as a major limiting factor for crop production on acid soils (Kochian et al., 2004). When exposed to strong acid conditions, plant root cells will be structurally and functionally damaged (Foy, 1984). For example, obviously swollen root hairs and cracks between cells in root meristems have been observed in Arabidopsis (Koyama et al., 1995, 2001; Iyer-Pascuzzi et al., 2011) and yorkshire-fog grass (Holcus lanatus) subjected to acid treatments (Kidd and Proctor, 2001). Moreover, a pH drop from 5.5 to 4.0 is associated with significant membrane depolarization, destruction of epidermal and cortical cells, and, ultimately, inhibition of root growth in Lotus corniculatus (Pavlovkin et al., 2009; Palóve-Balang et al., 2012). Similar symptom caused by proton rhizotoxicity have also been observed in many other plant species, such as alfalfa (Medicago sativa) (Yokota and Ojima, 1995), spinach (Spinacia oleracea) (Yang et al., 2005), common bean (Phaseolus vulgaris) (Rangel et al., 2005), and barley (Hordeum vulgare) (Song et al., 2011). Besides direct toxicity, low pH can also increase the solubility of other toxic ions, such as Al3+, in soil, and thus adversely influence plant root growth. In this aspect, Al and H<sup>+</sup> toxicities are physiologically linked to one another.

Over the past few decades, mechanisms of plant tolerance to Al and H<sup>+</sup> rhizotoxicities have been elucidated in many studies. Among them, identification of the C2H2-type zinc finger transcription factor family, STOP1 (Sensitive to Proton Rhizotoxicity1), contributed considerably to understanding of regulatory mechanisms underlying the integration of Al and H<sup>+</sup> tolerance in planta (Iuchi et al., 2007; Yamaji et al., 2009; Ohyama et al., 2013; Sawaki et al., 2014; Fan et al., 2015).

The first STOP1 gene, AtSTOP1, was identified in Arabidopsis (Iuchi et al., 2007). Transcriptome analyses and genetic characterization showed that AtSTOP1 regulates the expression of a set of genes, including three major Al tolerance genes, AtALMT1 (Aluminum activated Malate Transporter1), AtMATE (Multidrug and Toxic Compound Extrusion), and AtALS3 (Aluminum Sensitive3), along with other genes apparently involved in the regulation of cytosolic pH, such as GAD1 (Glutamate Decarboxylase1), ME1/2 (Malic Enzyme1/2), and GDH1/2 (Glutamate Dehydrogenase 1/2) (Liu et al., 2009; Sawaki et al., 2009; Kobayashi et al., 2014). Interestingly, the STOP1 homolog AtSTOP2, which partially accounts for tolerance to Al and H<sup>+</sup> rhizotoxicities, is also regulated by AtSTOP1 (Kobayashi et al., 2014). Moreover, a STOP1 ortholog in rice bean (Vigna umbellata), VuSTOP1, was isolated by suppression subtractive hybridization (Fan et al., 2014). In contrast to the constitutive expression exhibited by AtSTOP1, the expression of VuSTOP1 was inducible by both of Al and H<sup>+</sup> stresses (Fan et al., 2014). However, the assay of planta complementation in Atstop1 mutant showed that VuSTOP1 could fully restore the transcription of several H+-tolerance related genes, but only partially restores the expression of AtMATE and ALS3, indicating that VuSTOP1 might play a major role in H<sup>+</sup> tolerance, but only a minor role in Al tolerance (Fan et al., 2015). Similarly, other STOP1 homologs, including NtSTOP1 in tobacco (Nicotiana tabacum), LjSTOP1 in Lotus japonicas, PnSTOP1 in black poplar (Populus nigra), CsSTOP1 in tea (Camellia sinensis) and EguSTOP1 in Eucalyptus also reportedly possess similar functions in H<sup>+</sup> tolerance, and only partial or even no functionality in Al tolerance (Ohyama et al., 2013; Sawaki et al., 2014). On the other hand, the mutation of ART1 (Al Resistance Transcription Factor1), a STOP1 homolog in rice (Oryza sativa), appears to only affect Al hypersensitivity (Yamaji et al., 2009). In short, previous studies suggest that STOP1 transcription factors are ubiquitous in plants and have conserved functions in plant stress (Al and/or H+) tolerance, though specific responses vary among plant species.

Soybean (Glycine max) is one of the most important leguminous crops globally, comprising approximately 68% of crop legume production in the world and 57% of the global oilseed production (Herridge et al., 2008). Though many studies have elucidated the functions of STOP1 orthologs in other plant species, no information is available on whether GmSTOP family members are also involved in H<sup>+</sup> and Al tolerance in soybean. In the present study, three GmSTOP1 homologs were isolated and characterized from soybean. The function of each GmSTOP1 gene was analyzed in terms of Al and H<sup>+</sup> tolerance in Arabidopsis. The results demonstrate that all three GmSTOP1s play important roles in H<sup>+</sup> tolerance, while only GmSTOP1-1 and GmSTOP1-3 could partially recover Al tolerance in Arabidopsis Atstop1 mutant. Taken together, these results strongly suggest that the three GmSTOP1s in soybean share evolutionary conservation of H<sup>+</sup> tolerance, but not of Al tolerance.

## MATERIALS AND METHODS

## Plant Material and Growth Conditions

The soybean genotype YC03-3 was chosen as the plant material in this study. Soybean seeds were germinated in paper rolls moistened with modified one-half-strength nutrient solution as previously described (Liang et al., 2013). The resultant seedlings were then gown in full strength nutrient solution for 24 h before being used for various treatments. For the low pH treatment, soybean seedlings were subjected to 0.5 mM CaCl<sup>2</sup> (pH 4.2) for 0, 2, 4, 6, and 12 h. After low pH treatment, root tips (0–2 cm) were harvested for gene expression assays. For the tissue specific expression experiment, soybean root tips (0–2 cm), which was further divided into two segments (0–1 cm and 1–2 cm), basal roots (>2 cm) and leaves were harvested after 4 h of Al (0 or 50 µM AlCl<sup>3</sup> in 0.5 mM CaCl2, pH 4.2) treatment. For the Al dose experiment, soybean seedlings were treated with 0, 10, 50, and 100 µM AlCl<sup>3</sup> in 0.5 mM CaCl<sup>2</sup> solution (pH 4.2) for 4 h. For the time-course experiment, soybean seedlings were transplanted to Al (50 µM AlCl<sup>3</sup> in 0.5 mM CaCl2, pH 4.2) treatments for 0, 2, 4, 6, and 12 h. In both of the concentration response experiment and time-course experiment, root tips (0–2 cm) were separately harvested for gene expression assays. All experiments had four biological replicates.

## Phylogenetic Analysis and Characterization of GmSTOP1 Proteins in Soybean

TBLASTN analysis using the AtSTOP1 proteins sequence (accession number: Q9C8N5.1) as the query sequences was conducted at the Phytozome website<sup>1</sup> . Consequently, three STOP1 homologs with high similarity to AtSTOP1 were identified and designated as GmSTOP1-1 (Glyma10g35940), GmSTOP1-2 (Glyma16g27280), and GmSTOP1-3 (Glyma20g31650). Subsequently, multiple sequence alignment and phylogenetic tree construction were conducted using the deduced protein sequences of all three GmSTOP1s together with other STOP1 homologs, including AtSTOP1 from Arabidopsis, NtSTOP1 from tobacco, LjSTOP1 from Lotus japonicas, PnSTOP1 from black poplar (Populus nigra), CsSTOP1 from tea, EguSTOP1 from Eucalyptus, OsART1 from rice, PpSTOP1 from Physcomitrella patens, and TaSTOP1-A, TaSTOP1-B, TaSTOP1-D from wheat (Triticum aestivum). ClustalX2 and MEGA4.1 were used for the multiple sequence alignment and phylogenetic tree construction, respectively. The phylogenetic tree was constructed using the Neighbor-Joining method with 1,000 bootstrap replicates.

## RNA Extraction and Quantitative Real-Time PCR

Total RNA was extracted from plant tissues using RNA-solve reagent (OMEGA Bio-Tek, Norcross, GA, United States). Genomic DNA in the RNA samples was eliminated with RNase-free DNase I (Invitrogen, Carlsbad, CA, United States). The resulting extracts were then used to conduct the reverse transcription via MMLV-reverse transcriptase (Promega, Madison, WI, United States) following the manufacturer's instructions. Subsequently, SYBR Green monitored qRT-PCR (quantitative real-time PCR) analysis was performed using a ABI Step-one Plus real-time PCR system (Thermo Fisher Scientific, Waltham, MA, United States). The primer pairs used for expression analysis are listed in Supplementary Table S1.

#### Subcellular Localization of GmSTOP1s

The full length cDNAs of the three identified GmSTOP1 genes were amplified from first strand cDNA derived from soybean roots using gene specific primer pairs as listed in Supplementary Table S1. The PCR products were then cloned into the pMD18-T vector (Takara, Japan) for sequence confirmation.

Full-length cDNA of the three GmSTOP1s was fused with enhanced green fluorescent protein (GFP) to construct 35S::GmSTOP1s-GFP plasmids. Each construct was introduced into tobacco (Nicotiana tabacum) leaf cells according to previously described methods (Liang et al., 2010; Liu et al., 2016). The 35S::GFP construct was used as the control. Fluorescence signals of GFP were detected at 488 nm by confocal scanning microscope (LSM780; Zeiss, Germany).

## Transcriptional Activation Activity of GmSTOP1s

Gene specific primers with terminal SfiI and SalI restriction sites were used to amplify full-length cDNA of GmSTOP1-1 and GmSTOP1-3, whereas, primers with BamHI and PstI restriction sites were used to amplify full-length cDNA of GmSTOP1-2. Sequence fragments were digested by the corresponding restriction enzymes and inserted into the pGBKT7 vector (Clontech, Japan), producing pGBKT7- GmSTOP1s plasmids. The resultant plasmids and the pGBKT7 empty vector were then transformed into yeast strain AH109. After verification by PCR, transformed AH109 cells were cultured on either SD-Trp or SD-His medium for 3 days. The yeast cells grown on SD-Trp were then printed onto filter paper moistened with X-gal solution. Subsequently, the filter paper was freeze-thawed with liquid nitrogen and moistened again with X-gal solution. The appearance of blue areas on the filter paper was used to determine β-galactosidase activity.

## Complementation of GmSTOP1s in Atstop1 Mutant Plants

The three identified GmSTOP1s were separately introduced into the modified pBEGFP binary vector under the control of a 35S CaMV promoter to produce an over-expression construct that was then transformed into A. tumefaciens strain Gv3101. Subsequently, the constructs were transformed into Arabidopsis Atstop1 mutant plants via the floral dip method (Clough and Bent, 1998). Two independent over-expression lines for each gene were verified by qRT-PCR and used for further analysis as complemented lines.

To investigate the functions of GmSTOP1s in resistance to H<sup>+</sup> and Al toxicity, wild type, Atstop1 mutant and the complemented lines overexpressing GmSTOP1 were germinated on solid Murashige and Skoog (MS) medium for 5 days. Uniform seedlings with ∼1.5 cm root lengths were transferred to modified 1/30 strength Hoagland nutrient solution (without NH4H2PO<sup>4</sup> and plus 1 mM CaCl2) with different treatments as described (Fan et al., 2015). Control plants were grown in media with pH adjusted to 5.8, while treated plants were grown in low pH media (pH 4.7) or media containing 2 µM AlCl<sup>3</sup> (pH 5.0) for 7 days (Fan et al., 2015). Upon harvest, roots of each plant were scanned and analyzed in Image J (National Institutes of Health, United States). All experiments had four biological replicates, each of which contains two plants.

For analysis of H<sup>+</sup> and Al genes expression responses, uniform Arabidopsis seedlings were treated with 1/30 strength Hoagland nutrient solution containing 2 µM AlCl<sup>3</sup> (pH 5.0) and low pH (pH 4.7) for 24 h (Fan et al., 2015). All experiments were conducted in a growth incubator running a 24◦C, 12h/22◦C, 12 h day/night cycle. The whole roots were harvested for gene expression assays. The primer pairs of target genes for qRT-PCR analysis are listed in Supplementary Table S1. Arabidopsis UBQ1 was used as housekeeping gene control to normalize the expression of the corresponding genes. All experiments had four biological replicates.

<sup>1</sup>https://phytozome.jgi.doe.gov/pz/portal.html

### Statistical Analysis

fpls-09-00570 April 24, 2018 Time: 17:17 # 4

All data were analyzed by Student's t-tests using SPSS 13.0 (SPSS Institute, Chicago, IL, United States).

## RESULTS

### Identification of STOP1 Homologs in Soybean

A homolog search resulted in retrieval of three STOP1 homologs in the soybean genome, which were named GmSTOP1- 1 (Glyma.10G215200), GmSTOP1-2 (Glyma.16G156400) and GmSTOP1-3 (Glyma.20G176500) based on genome localization. A phylogenetic tree showed that the STOP1 homologs in dicots were differentiated from those in monocots (**Figure 1A**). Moreover, GmSTOP1-1 and GmSTOP1-3 present as duplicated pair and display high similarity with VuSTOP from rice bean, while GmSTOP1-2 clusters in another sub-clade with LjSTOP1 from Lotus japonicus (**Figure 1A**). Moreover, the deduced amino acid sequences of all three GmSTOP1s contain four putative C2H2 zinc finger domains that are highly conserved in STOP1 orthologs from other plant species (**Figure 1B**).

### Subcellular Localization and Transcription Activation Activity of GmSTOP1s

To determine the subcellular localization of the three identified GmSTOP1s, GmSTOP1-GFP fusion constructs were assayed for transient expression in tobacco leaf cells. The results showed that control GFP fluorescence was detectable in both the nucleus and cytoplasm. In contrast, fluorescence derived from GmSTOP1- GFP constructs was exclusively localized within the nucleus (**Figure 2A**), strongly suggesting that the three GmSTOP1 members are all nucleus localized proteins.

Transcription activity of the three GmSTOP1 members was determined in a one-hybridization expression system in yeast. The results showed that the yeast strain AH109 transformed with either a pGBKT7-GmSTOP1 or the pGBKT7 empty vector could grow well on the SD-Trp medium (**Figure 2B**). However, only the three AH109 strains transformed with pGBKT7- GmSTOP1s grew well on the SD-His medium (**Figure 2B**). Furthermore, all three of the AH109 strains transformed with a pGBKT7-GmSTOP1 showed high β-galactosidase activity as indicated by the blue color on filter paper using X-gal as a substrate (**Figure 2B**). Therefore, the ability of all three GmSTOP1 homologs to activate lacZ expression strongly suggests that each one functions as a transcription factor.

## Expression Patterns of GmSTOP1s in Response to Al and Low pH Stress

Quantitative real-time PCR was used to analyze GmSTOP1 expression patterns in soybean seedlings. Expression levels of the three GmSTOP1s were hardly affected by low pH stress during the 12 h treatment period (Supplementary Figure S2). On the other hand, transcriptional responses varied among the three GmSTOP1 genes in response to Al stress. As shown in **Figure 3A**, transcription of GmSTOP1-2 was not significantly affected after 4 h of Al treatment in any tissues, including root tips (0–2 cm), basal regions of roots (>2 cm) and leaves, (**Figure 3A**). In contrast to the constitutive expression of GmSTOP1-2 in roots, transcript levels of GmSTOP1-1 and GmSTOP1-3 increased by more than 6- and 11-fold, respectively, in root tips after 4 h of Al treatment (**Figure 3A**). However, in root basal regions, only transcription of GmSTOP1-3 increased by more than 1.7-fold in response to Al stress, while no detectable change was observed for GmSTOP1-1.

Dose-responses of GmSTOP1s to Al stress were further analyzed in soybean root tips after 4 h of Al treatment. Transcript accumulations of both GmSTOP1-1 and GmSTOP1-3 were strictly dependent on Al concentration in the medium (**Figure 3B**), with transcript abundances enhanced for both GmSTOP1-1 and GmSTOP1-3 in 50 and 100 µM Al treatments (**Figure 3B**). The expression of GmSTOP1-2 was constitutively expressed at relatively high levels regardless the external Al concentrations (**Figure 3B**).

Results from time-course experiments showed that the expression of both GmSTOP1-1 and GmSTOP1-3 were quickly enhanced in response to Al stress by more than twofold after 2 h of Al treatment, and remained high over 12 h (**Figure 3C**). Meanwhile, the expression of GmSTOP1-2 did not vary during the period of Al treatment (**Figure 3C**).

## Functional Analysis of GmSTOP1s in the Arabidopsis Atstop1 Mutant

In order to examine their functions in plant H<sup>+</sup> and Al tolerance, all three GmSTOP1s were overexpressed in the Arabidopsis Atstop1 mutant. The expression of all three of the GmSTOP1 genes in the Atstop1 mutant was verified by qRT-PCR. Under normal growth conditions, no significant differences were observed among wild type, Atstop1 mutant and complemented lines overexpressing any of the GmSTOP1s (**Figures 4A,B**). However, under low pH condition (pH 4.7), root elongation of wild type and Atstop1 mutant was inhibited by 51% and 80%, respectively (**Figures 4A,C**). In each of two complemented lines of GmSTOP1-1 (#5 and #6), GmSTOP1- 2 (#12 and #15) and GmSTOP1-3 (#47 and #54), root elongation was inhibited much less than that of the Atstop1 mutant (**Figures 4A,C**). These results suggest that all three GmSTOP1s are able to confer H<sup>+</sup> tolerance in Atstop1 mutant plants.

Addition of Al to the low pH culture solution slightly decreased root elongation of wild type plants, but significantly inhibited root elongation of Atstop1 mutants as indicated by a 90% decrease in root elongation compared to root elongation in wild type plants (**Figures 4A,D**). Unlike the role of GmSTOP1s in H<sup>+</sup> tolerance, the functions of GmSTOP1s in Al tolerance varied. Each of the lines complemented with GmSTOP1-1 (#5 and #6) and GmSTOP1-3 (#47 and #54) overexpression recovered elongation to 28% and 32%, and 29% and 19% of that of wild type, respectively (**Figures 4A,D**). In contrast, lines complemented with GmSTOP1-2 overexpression

FIGURE 1 | Phylogenetic tree and amino acid alignments of predicted C2H<sup>2</sup> zinc finger domains in plant STOP1s. (A) Phylogenetic tree was generated based on an amino-acid alignment with STOP1 orthologs from several plant species. (B) Alignment of the amino acid sequences of predicted C2H2 zinc finger domains in STOP1 proteins. Black background indicates identical residues. Asterisks indicate conserved Cys and His residues of C2H2 motifs. The plant STOP1 proteins aligned include representatives from Glycine max (GmSTOP1-1, XP\_006588359.1; GmSTOP1-2, XP\_006598713.1; GmSTOP1-3, XP\_014628358.1), Arabidopsis thaliana (AtSTOP1, NP\_174697.1), Nicotiana tabacum (NtSTOP1, AB811781), Lotus japonicus (LjSTOP1, BAN67817.1), Vigna umbellata (VuSTOP1, KP637172), Camellia sinensis (CsSTOP1, BAN67815.1), Populus nigra (PnSTOP1, BAN67813.1), Eucalyptus (EguSTOP1, BAO56822.1), Triticum aestivum (TaSTOP1-A, AGS15201.1; TaSTOP1-B, AGS15202.1; TaSTOP1-D, AGS15195.1), Oryza sativa (OsART1, AB379846), and Physcomitrella patens (PpSTOP1, BAN67814.1).

did not recover root elongation. These results indicate that GmSTOP1-1 and GmSTOP1-3, but not GmSTOP1-2 can partially reverse the Al hypersensitivity of Atstop1 mutant plants.

### Transcription of STOP1 Down-Stream Genes in GmSTOP1 Complemented Atstop1 Mutants

The differential contributions of GmSTOP1s to H<sup>+</sup> and Al tolerance were further determined by investigating the transcript levels of several related down-stream genes in GmSTOP1 complemented Atstop1 mutants. The results showed that expression of several H<sup>+</sup> tolerance genes were significantly restored in all of the complemented Atstop1 mutant lines (**Figure 5**). Among responsive genes, the expression of GDH2 (At5g07440) was restored the most, with transcription returning to at least 50% of transcript levels in WT plants (**Figure 5**). Although restored to lesser extents, the expression of three other H<sup>+</sup> tolerance genes, GDH1 (At5g18170), GABA-T (At3g22200) and NADP-malate enzyme 2 (NADP-ME2), was restored nonetheless by 37, 25, and 20%, respectively, over expression in the Atstop1 mutant (**Figure 5**).

Unlike expression patterns of H<sup>+</sup> tolerance related genes, expression responses of Al tolerance related genes varied among GmSTOP1 complemented lines. These Al tolerance related genes included pectin methylesterase inhibitor superfamily protein (PMI, At2g45220), AtTDT (At5g47560), NADP-ME2 (At5g11670), and

AtMATE (At1g51340). In both GmSTOP1-1 and GmSTOP1-3 complemented lines, the expression of AtPMI, AtTDT, NADP-ME2, and AtMATE were partially restored. Yet, in GmSTOP1-2 complemented lines, the expression of each of these four genes was not affected relative to expression in Atstop1 mutants (**Figure 5**).

## DISCUSSION

Proton and Al rhizotoxicities are two of the major constraints of plant growth and development on acid soil (Kochian et al., 2004). Plants have adapted to these stresses by developing a variety of coping strategies involving a number of genes (Liu

et al., 2014; Kochian et al., 2015). Recent studies have revealed that H<sup>+</sup> and Al tolerance mechanisms are regulated by STOP1 transcription factors in many plant species (Iuchi et al., 2007; Sawaki et al., 2009, 2014; Yamaji et al., 2009; Garcia-Oliveira et al., 2013; Ohyama et al., 2013; Fan et al., 2015). However, few studies have attempted to systematically dissect the possible roles of all STOP1 members in a single species responding to H<sup>+</sup> stress, Al toxicity, or both.

In the present study, a total of three GmSTOP1 genes were identified in the soybean genome. All of these GmSTOP1

transcript levels were used as the internal standard. Data are expressed as means of four replicates. Asterisks indicate significant differences in comparison to the Atstop1 mutant (∗0.01 < P < 0.05; ∗∗0.001 < P < 0.01; ∗∗∗P < 0.001).

2 µM; pH 5.0) for 24 h. Transcript abundances of GDH1, GDH2, GABA-T and AtNADP-ME2 were quantified from plants grown in the low pH treatment, while expression levels of PMI (At2g45220), AtTDT (At5g47560), AtNADP-ME2 (At5g11670), and AtMATE (At1g51340) were quantified in Al treated samples. UBQ1

homologs are localized to the nucleus, and exhibit transcription activity (**Figure 2**). Sequence analysis revealed that GmSTOP1-2 is phylogenetically distinct from GmSTOP1-1 and GmSTOP1- 3, which appear to be a duplicated pair (**Figure 1**). As gene duplication provides opportunities for functional divergence (Force et al., 1999; Lynch et al., 2001; Flagel and Wendel, 2009; Libault et al., 2010; Schmutz et al., 2010), we hypothesize that the three GmSTOP1 genes might have divergent functions in regard to H<sup>+</sup> and Al tolerance even though they are highly conserved in some features (**Figures 1**, **2**).

To dissect the differential contributions of GmSTOP1s in H<sup>+</sup> tolerance, expression analysis was conducted. The results showed that similar to AtSTOP1 in Arabidopsis (Iuchi et al., 2007), all three GmSTOP1 genes were constitutively expressed and hardly affected by H<sup>+</sup> treatment (Supplementary Figure S2). Further complementation assays showed that all three GmSTOP1 genes are able to confer H<sup>+</sup> tolerance to the Arabidopsis Atstop1 mutant (**Figure 4C**). These results are consistent with the previous studies reporting that STOP1 orthologs in dicots are able to confer H<sup>+</sup> tolerance to the H<sup>+</sup> sensitive Atstop1 mutant (Ohyama et al., 2013; Sawaki et al., 2014; Fan et al., 2015). Furthermore, the H<sup>+</sup> hypersensitivity of the Atstop1 mutant is the result of downregulation of genes in several pH regulation pathways caused by the dysfunction of AtSTOP1 (Sawaki et al., 2009).

The expression of several H<sup>+</sup> tolerance genes, including STOP2, CIPK23, and PGIP1, has been restored in planta in complementation assays of Atstop1 by STOP1 orthologs from rice bean (VuSTOP1), Eucalyptus (EguSTOP1), tobacco (NtSTOP1), black poplar (PnSTOP1), tea (CsSTOP1), Lotus japonicus (LjSTOP1), or Physcomitrella patens (PpSTOP1) (Ohyama et al., 2013; Sawaki et al., 2014; Fan et al., 2015). However, none of these genes were affected by complementation with any of the GmSTOP1 homologs in the Atstop1 mutant (Supplementary Figure S4). Instead, GmSTOP1 complementation restored the transcription of several other H<sup>+</sup> tolerance relative genes, including GDH1, GDH2, GABA-T, and AtNADP-ME2, which are considered to play roles in maintaining pH homeostasis in plants. For example, AtNADP-ME2 has been reported to function in the pH stat pathway through consumption of cytosolic H<sup>+</sup> (Roberts et al., 1992; Sakano, 1998). Meanwhile, GDH1, GDH2 and GABA-T are the major isoforms in the "GABA shunt" pathway, which contributes largely to cytosolic pH homeostasis in plant cells (Crawford et al., 1994; Bown and Shelp, 1997). These results strongly suggest that all three GmSTOP1 homologs participate in conserved functions in H<sup>+</sup> tolerance mainly through regulation of similar pH stat pathways that are distinct from the pH stat pathways regulated by other plant STOP1 orthologs (Ohyama et al., 2013; Sawaki et al., 2014; Fan et al., 2015).

Interestingly, in the presence of Al, expression of both GmSTOP1-1 and GmSTOP1-3 quickly escalated in root tips, while expression of GmSTOP1-2 was not significantly affected (**Figure 3** and Supplementary Figure S3). Similar results have also been reported for bread wheat, in which expression of TaSTOP1-A was found to be responsive to H<sup>+</sup> and Al stresses and divergent from the responses of TaSTOP1-B and TaSTOP1-D (Garcia-Oliveira et al., 2013). It has been suggested that this divergence might be mainly due to the presence of a pyrimidinerich stretch and the absence of a light responsive element in the 5<sup>0</sup> UTR of TaSTOP1-A compared to its homologs TaSTOP1-B and TaSTOP1-D (Garcia-Oliveira et al., 2013). Consistent with this, our investigation revealed that the 5<sup>0</sup> -UTR is more similar between GmSTOP1-1 and GmSTOP1-3 than it is between either of these genes and GmSTOP1-2 (Supplementary Figure S1). Therefore, there is a possibility that the differential expression between GmSTOP1-2 and the other two GmSTOP1s in response to Al stress might be due to divergence in the 5 0 -UTR. Thereby, divergence between GmSTOP1-2 and the other two GmSTOP1s in both transcriptional regulation and protein sequence further suggests that GmSTOP1-2 functions differently than GmSTOP1-1 and GmSTOP1-3 in Al tolerance responses.

Even more evidence in support of divergence among GmSTOP1s in Al tolerance functionality was gathered in complementation assays (**Figure 4**). The results strongly indicate that GmSTOP1-1 and GmSTOP1-3 are involved at least partially in AtSTOP1 related Al tolerance responses, whereas GmSTOP1-2 is not. It has been reported that AtSTOP1 regulates transcription of three major Al tolerance genes in Arabidopsis, namely AtALMT1, ALS3, and AtMATE (Kobayashi et al., 2007; Liu et al., 2009; Sawaki et al., 2009; Tokizawa et al., 2015). Among them, AtALMT1 accounts for more than 70% of the Al tolerance phenotype in Arabidopsis (Iuchi et al., 2007; Liu et al., 2012). In the current study, it was interesting to find that none of the GmSTOP1 homologs restores the expression of AtALMT1 or ALS3 in the Atstop1 mutant, while AtMATE expression was recovered slightly in both GmSTOP1-1 and GmSTOP1-3 complemented lines, but not in GmSTOP1-2 complemented lines (**Figure 5** and Supplementary Figure S4). Similar results were also reported in other plant species, where most STOP1 orthologs are not able to restore the expression of all three Al tolerance genes in the Atstop1 mutant. For example, LjSTOP1, CsSTOP1, and PnSTOP1 can slightly restore the expression of AtALMT1, but not the expression of ALS3 or AtMATE, while VuSTOP1 can partially restore the expression of ALS3 and AtMATE, but not the expression of AtALMT1 (Ohyama et al., 2013; Fan et al., 2015). Placing the current results in the context of previous reports suggests that the regulatory functions of AtSTOP1 in Al tolerance is not entirely conserved among plant STOP1 orthologs.

Potential phenotypic effects of GmSTOP1-1/GmSTOP1-3 in the Atstop1 mutant in response to Al toxicity are revealed by considering functions of PMI, AtTDT, and NADP-ME2, which are down-regulated in Atstop1 mutants subjected to Al stress (Sawaki et al., 2009), and which had expression restored in the complementation experiments herein. Members of the PMI family have been reported to inhibit pectin methylesterase activity, and thereby increasing Al tolerance (Sénéchal et al., 2015; Geng et al., 2017). The other two genes, AtTDT and NADP-ME2, are involved in malate homeostasis and metabolism in the vacuole and cytosol, respectively (Hurth et al., 2005; Badia et al., 2015). Results in the present study showed that all of these genes were partially restored in both

GmSTOP1-1 and GmSTOP1-3 complemented lines, but not in GmSTOP1-2 complemented lines (**Figure 5**). These expression responses are in accord with the variation in Al tolerance observed among GmSTOP1s complemented lines (**Figure 4D**). Therefore, it appears that GmSTOP1-1/GmSTOP1-3 might function in Al tolerance through the regulation of cell wall modifications and malate metabolism.

Overall, the present study identifies three GmSTOP1 homologs in the soybean genome, all of which localize in the nucleus and have the transactivation potential. Complementation assays suggest that all three GmSTOP1 homologs play major roles in H<sup>+</sup> tolerance through transcriptional regulation of H<sup>+</sup> tolerance genes, whereas, only GmSTOP1-1 and GmSTOP1-3 function in Al tolerance. Taken together, the results herein suggest that the functions of the three identified GmSTOP1s are evolutionarily conserved in H<sup>+</sup> tolerance responses, but not in Al tolerance responses.

### AUTHOR CONTRIBUTIONS

WW, YL, QC, WP, JP, and CL performed the experiments and collected the data. CL, JT, and HL designed the research, analyzed the data, and wrote the manuscript.

#### REFERENCES


#### FUNDING

This work was supported by grants from National Natural Science Foundation of China (31672220 and 31422046), the National Key Research and Development Program (2016YFD0100700), Guangdong Natural Science Funds for Distinguished Young Scholars (2015A030306034), Guangdong High-level Personnel of Special Support Program (2015TQ01N078, 2015TX01N042, and YQ201530) and Research Team Project of the Natural Science Foundation of Guangdong Province (2016A030312009).

#### ACKNOWLEDGMENTS

The authors thank Dr. Thomas Walk of Golden Fidelity LLC for critical reading.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00570/ full#supplementary-material



**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Wu, Lin, Chen, Peng, Peng, Tian, Liang and Liao. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Identification of STOP1-Like Proteins Associated With Aluminum Tolerance in Sweet Sorghum (Sorghum bicolor L.)

Sheng Huang† , Jie Gao† , Jiangfeng You, Yanan Liang, Kexing Guan, Siqi Yan, Meiqi Zhan and Zhenming Yang\*

Jilin Province Engineering Laboratory of Plant Genetic Improvement, College of Plant Science, Jilin University, Changchun, China

Aluminum (Al) toxicity in acidic soils affects crop production worldwide. C2H2-type zinc finger transcription factor STOP1/ART1-mediated expression of Al tolerance genes has been shown to be important for Al resistance in Arabidopsis, rice and other crop plants. Here, we identified and characterized four STOP1-like proteins (SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d) in sweet sorghum, a variant of grain sorghum (Sorghum bicolor L.). Al induced the transcription of the four SbSTOP1 genes in both time- and Al concentration-dependent manners. All SbSTOP1 proteins localized to the cell nucleus, and they showed transcriptional activity in a yeast expression system. In the HEK 293 coexpression system, SbSTOP1d showed transcriptional regulation of SbSTAR2 and SbMATE, indicating the possible existence of another SbSTOP1 and SbSTAR2-dependent Al tolerance mechanism in sorghum apart from the reported SbMATE-mediated Al exclusion mechanism. A transgenic complementation assay showed that SbSTOP1d significantly rescued the Al-sensitivity characteristic of the Atstop1 mutant. Additionally, yeast two-hybrid and bimolecular fluorescence complementation (BiFC) assays showed that SbSTOP1d interacted with SbSTOP1b and SbSTOP1d itself, suggesting that SbSTOP1 may function as a homodimer and/or heterodimer. These results indicate that STOP1 plays an important role in Al tolerance in sweet sorghum and extend our understanding of the complex regulatory mechanisms of STOP1-like proteins in response to Al toxicity.

#### Keywords: aluminum toxicity, STOP1, transcriptional regulation, Al tolerance genes, sweet sorghum

#### INTRODUCTION

Acid soils are widespread and limit crop production all over the world. Aluminum (Al) toxicity is a primary limiting factor in acid soils. At pH below 5, Al (the most abundant form, Al3+) inhibits root elongation within a few minutes, which leads to subsequent water and nutrient deficiency (Kochian et al., 2004; Ma, 2007; Bojórquez-Quintal et al., 2017).

To cope with Al stress, plants develop a series of strategies that have been categorized into two main types of Al resistance mechanisms. Al exclusion mechanisms, which are external strategies, aim at preventing toxic Al from entering root cells by exuding organic compounds (e.g., organic acids or phenolics) into the rhizosphere to chelate Al. Al tolerance mechanisms, which are internal strategies, sequester and detoxify Al that enters the plant (Ma, 2000; Ryan et al., 2001;

#### Edited by:

Wei Fan, Yunnan Agricultural University, China

#### Reviewed by:

Jian Li Yang, Zhejiang University, China Chao-Feng Huang, Shanghai Institutes for Biological Sciences (CAS), China Fernando Carlos Gómez-Merino, Inicio COLPOS, Mexico

\*Correspondence:

Zhenming Yang zmyang@jlu.edu.cn †Co-first authors

#### Specialty section:

This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science

Received: 27 November 2017 Accepted: 12 February 2018 Published: 28 February 2018

#### Citation:

Huang S, Gao J, You J, Liang Y, Guan K, Yan S, Zhan M and Yang Z (2018) Identification of STOP1-Like Proteins Associated With Aluminum Tolerance in Sweet Sorghum (Sorghum bicolor L.). Front. Plant Sci. 9:258. doi: 10.3389/fpls.2018.00258

**72**

Kochian et al., 2015). The mechanisms of Al-meditated root exudation of organic acids (citrate, malate or oxalate) are well characterized, with involvement of transporters from the Al-activated malate transporter (ALMT) family and the multidrug and toxic compound extrusion (MATE) family (Sasaki et al., 2004; Furukawa et al., 2007; Magalhaes et al., 2007). SbMATE in sorghum (and MATE in barley, HvAACT1) was the first Al resistance gene to be identified. It encodes a citrate transporter that is primarily responsible for Al resistance in sorghum via citrate release (Furukawa et al., 2007; Magalhaes et al., 2007). Subsequently, homologs of MATE were isolated in other species, including AtMATE1 in Arabidopsis thaliana, VuMATE1 in Vigna umbellata and OsFRD1 in Oryza sativa (Liu et al., 2009; Yang et al., 2011; Yokosho et al., 2011). Once Al traverses the external organic compound barrier, it reaches the root cells. In response, plants develop Al tolerance mechanisms that involve other membrane transporters, including Nramps, ABC transporters and aquaporins (Huang et al., 2009, 2010; Negishi et al., 2012, 2013; Li et al., 2014). The cell wall constitutes the first barrier against Al in cells, while some ABC transporters (such as OsSTAR1/OsSTAR2 protein complex) are thought to mediate the efflux of UDP-glucose into the cell wall, which presumably alters the cell wall composition, limiting Al accumulation and reducing Al toxicity (Huang et al., 2009).

Al induces the coordinated expression of multiple Al tolerance genes in plants. Sensitive to proton rhizotoxicity 1 (STOP1) in Arabidopsis was isolated and further shown to be a key transcription factor that regulates the expression of a range of Al tolerance genes (including AtALMT1, AtMATE, and AtALS3) and some proton tolerance genes (Liu et al., 2009; Sawaki et al., 2009). Al resistance transcription factor 1 (ART1) was also identified in rice. In contrast to AtSTOP1, the rice homolog regulates only Al tolerance genes (such as OsNrat1, OsSTAR1, and OsSTAR2) but not proton tolerance genes (Yamaji et al., 2009). Homologous STOP1-like genes have also been characterized in other plant species. These genes all encode a Cys2His<sup>2</sup> (C2H2) zinc finger protein, but their expression patterns vary. AtSTOP1 in A. thaliana, OsART1 in O. sativa, and NtSTOP1 in Nicotiana tabacum are constitutively expressed in roots, whereas VuSTOP1 in V. umbellata is upregulated by Al toxicity in a dosage-dependent manner (Yamaji et al., 2009; Ohyama et al., 2013; Fan et al., 2015). AtSTOP2, a homolog of AtSTOP1, was identified in Arabidopsis recently. AtSTOP2 activates the expression of some genes for Al- and low pH-tolerance that are regulated by AtSTOP1 (Kobayashi et al., 2014). The distinct roles and/or consociation of AtSTOP1 and AtSTOP2 in Al signaling and regulatory pathways, however, have not yet been clarified.

In this study, four STOP1-like genes (SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d) with diverse expression profiles were identified in sweet sorghum, a variant of grain sorghum (Sorghum bicolor L.). SbSTOP1d, which shares the highest identity with AtSTOP1 and OsART1, regulated the transcription of SbSTAR2, suggesting the existence of a SbSTOP1-mediated Al tolerance mechanism aside from the previously reported SbMATE-dependent Al exclusion mechanism in sorghum. SbSTOP1d interacted with itself and SbSTOP1b in plants, implying that SbSTOP1d might form a homo- and/or heterodimer to function. Taken together, we characterized homologous SbSTOP1s in sweet sorghum and examined the association between diverse SbSTOP1s, which may help to further clarify the complex signal transduction pathways of STOP1-like proteins in response to Al toxicity.

### MATERIALS AND METHODS

## Plant Materials, Culture Conditions, and Al Treatments

The sweet sorghum (S. bicolor L.) cultivar POTCHETSTRM was used in this study (Zhang et al., 2015). Seeds were surface sterilized with 1% (v/v) NaClO for 20 min, rinsed with deionized water five times, spread on wet filter paper in a Petri dish and germinated for 2 days in darkness at 28◦C. The germinated seeds were transplanted into 0.5 mM CaCl<sup>2</sup> solution at pH 4.5 or 5.8 depending on the treatment. The seedlings were grown in an environmentally controlled growth chamber with a 14 h light (400 µmol m−<sup>2</sup> s −1 )/10 h dark photoperiod, 26◦C day/22◦C night temperatures and 80% relative humidity.

For gene expression pattern analysis, seedlings cultured for 3 days in 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) were then exposed to a different treatment. For the time-course assay, seedlings were exposed to 0.5 mM CaCl<sup>2</sup> solution with 15 µM AlCl<sup>3</sup> for 0, 3, 6, 9, or 24 h (pH 4.5), then the root apices (0–1 cm) were excised. For the Al concentration-dependent assay, seedlings were exposed to 0.5 mM CaCl<sup>2</sup> solution with 0, 5, 10, 15, or 30 µM AlCl<sup>3</sup> for 24 h (pH 4.5), then the root apices (0–1 cm) were cut. For the tissue expression pattern assay, seedlings were exposed to 0.5 mM CaCl<sup>2</sup> solution with 0 or 15 µM AlCl<sup>3</sup> for 24 h (pH 4.5) with roots (0–1 cm, 1–2 cm, or 2–3 cm) and shoots excised. For the different metal treatments, seedlings were exposed to 15 µM AlCl3, 10 µM CdCl2, 0.5 µM CuCl<sup>2</sup> or 10 µM LaCl<sup>3</sup> for 24 h (pH 4.5), then the root apices (0–1 cm) were excised. For the low pH treatments, seedlings were cultured in 0.5 mM CaCl<sup>2</sup> solution (pH 5.8) for 3 days, then exposed to the same solution at pH 5.8, 5.0, 4.5, 4.0, or 3.5 for 24 h. Then, the root apices (0–1 cm) were excised for RNA isolation. Each treatment was analyzed using three biological replicates.

#### Sequence Analysis

All sequences were analyzed using BLAST in the sorghum genome database<sup>1</sup> and NCBI. Sequence alignment was performed using Vector NTI and modified in GeneDoc. The phylogenetic tree was constructed according to the neighbor-joining method using MEGA 5.1.

## RNA Isolation and Quantitative Real-Time PCR

Total RNA isolation, cDNA preparation and quantitative realtime PCR (qRT-PCR) were performed as previously described

<sup>1</sup>http://pgsb.helmholtz-muenchen.de/plant/sorghum/

(Zhang et al., 2015). The gene-specific primers were designed using Primer 5.0 software (Supplementary Table 1). The house-keeping gene β-actin (GenBank ID: X79378) was used as an internal control (Zhang et al., 2015). The qRT-PCR was performed using SYBR Premix ExTaq (Takara) in an Mx3005P qPCR system (Stratagene, United States). Thermocycling proceeded as follows: 1 cycle of 30 s at 95◦C, 30 cycles of 5 s at 95◦C and 20 s at 60◦C, and 1 cycle of 60 s at 95◦C, 30 s at 55◦C, and 30 s at 95◦C for the melting curve analysis. The relative expression level of the genes was calculated using the 2 <sup>−</sup>11C<sup>T</sup> method (Livak and Schmittgen, 2001). The experiment was conducted using three biological replicates.

#### Subcellular Localization of SbSTOPs

Arabidopsis protoplasts were isolated from 4-week-old plants. Leaves were cut into strips and transferred quickly into the enzyme solution [1% (w/v) cellulase R10, 0.25% (w/v) macerozyme R10, 0.4 M D-mannitol, 20 mM KCl, 20 mM MES pH 5.7 and 10 mM CaCl2] for 1 h digestion at room temperature in darkness. Protoplasts were filtered through a 100-micron nylon mesh and centrifuged for 2 min at 100 g, rinsed with ice-cold W5 buffer [154 mM NaCl, 125 mM CaCl2, 5 mM KCl and 2 mM MES, pH 5.7], and suspended in MMg buffer [0.4 M mannitol, 15 mM MgCl2, 4 mM MES, pH 5.7]. Afterward, the protoplasts were ready for transformation. The PEG-mediated protoplast transformation method was used in this study. 10 µl of 35S::YFP-SbSTOP1a (or 35S::YFP-SbSTOP1b, 35S::YFP-SbSTOP1c and 35S::YFP-SbSTOP1d) was mixed with 100 µl protoplasts and 110 µl PEG solution [40% (w/v) PEG4000, 0.2 M mannitol, 100 mM CaCl2]. The protoplast/DNA mixture was incubated at room temperature in darkness for 15 min, washed twice with W5 buffer, and incubated in darkness at room temperature for 12–16 h. The fluorescence images were captured using a fluorescence microscope (Axio Observer A1, Zeiss).

## Transcriptional Activity Detection and Yeast Two-Hybrid Assay

To detect the transcriptional activity of SbSTOP1s, the bait vector pBridge expressing SbSTOP1a, SbSTOP1b, SbSTOP1c, SbSTOP1d, SbSTOP1d-NT (1–275 aa) or SbSTOP1d-CT (276–519 aa) fused to the GAL4 DNA-binding domain (BD) was used to transform the yeast strain Y2HGold. Colonies were selected on SD/-Trp-His medium (with or without 3-AT) and cultured for 3 days at 30◦C. For the yeast twohybrid assay, the prey vector pGADT7 expressing SbSTOP1b or SbSTOP1d fused to the GAL4 activation domain (AD) and the bait vector pBridge expressing SbSTOP1d-NT (1–275 aa) fused to the BD were used to co-transform the yeast strain Y2HGold (or the Y190 yeast strain for the β-galactosidase assay). Colonies were selected on SD/-Trp-Leu-His medium and cultured for 3 days at 30◦C. The β-galactosidase assay was performed using chlorophenol redβ-D-galactopyranoside (CPRG) as substrate, and Miller units were calculated according to the Yeast Protocols Handbook (Clontech, PT3024-1). The experiment was conducted using three biological replicates.

## HEK293 Coexpression System and Dual-Luciferase Reporter Assay

To examine the transcriptional regulation of SbMATE or SbSTAR2 by SbSTOP1d, the reporter plasmid (pSbMATE::LUC-SV40::REN or pSbSTAR2::LUC-SV40::REN) and effector plasmid (CMV::SbSTOP1d-Myc) were co-transfected into HEK293 (human embryonic kidney) cells.

HEK293 cells were cultured as previously described (Gao et al., 2015). Cells were maintained in Dulbecco's modified Eagle's medium (DMEM) with FBS (10%) and penicillin/streptomycin (1%) in a cell culture flask, T75 (Eppendorf), which was incubated in a 37◦C incubator with a humidified atmosphere of 5% CO<sup>2</sup> in air. When the cell count reached 2 × 10<sup>7</sup> , cells were subcultured in a 6-well plate the night before and grown to 60–70% confluence by the day of transfection. HEK293 cells were transfected with the constructed plasmids (reporter and effector) using the calcium phosphate transfection method as reported (Gao et al., 2015). After 30–48 h, the transfected cells were ready for the dualluciferase reporter assay.

The dual-luciferase reporter assay was conducted according to the technical manual of the Dual-luciferase Reporter Assay System (Promega, E1910). After removing the growth medium, the transfected cells were gently rinsed with 1 × PBS (pH 7.2, Thermo, 20012050) and lysed in 1 × Passive Lysis Buffer (PLB). The PLB lysate was plated in a 96-well plate with volume ≤20 µl/well. The firefly luciferase activity was measured by adding 100 µl of Luciferase Assay Reagent II (LAR II) to generate a luminescent signal that was measured with a luminometer (Berthold LB960). This reaction was then quenched, and the Renilla luciferase reaction is simultaneously initiated by adding 100 µl of Stop & Glo <sup>R</sup> Reagent to the same well. The Stop & Glo <sup>R</sup> Reagent also produced a luminescent signal from the Renilla luciferase, which served as an internal control. The experiment was conducted using three biological replicates.

### Bimolecular Fluorescence Complementation Assay

Different pairs of plasmids encoding nYFP-SbSTOP1d and cCFP-SbSTOP1b, or encoding nYFP-SbSTOP1d and cCFP-SbSTOP1d were co-transformed into Arabidopsis protoplasts. The protoplast preparation and transformation method are described above. The reconstituted YFP fluorescence images were examined by a fluorescence microscope (Axio Observer A1, Zeiss), and the percentage of cells that exhibited bimolecular fluorescence complementation (BiFC) fluorescence signals were calculated. The experiment was conducted using three biological replicates.

## Overexpression of SbSTOP1d in the Atstop1 Mutant

The open reading frame (ORF) of SbSTOP1d was amplified and cloned into the pEGAD vector (35S::LUC-SbSTOP1d) using the In-Fusion enzyme. The construct was transformed into Agrobacterium tumefaciens strain AGL0, which was further introduced into the Atstop1 mutant using the floral dip method (Clough and Bent, 1998). The transgenic seedlings were first screened with the Basta herbicide, then

confirmed by a three-primer PCR-based genotyping using the following primers: LP, 5<sup>0</sup> - TTCATTGGTGAGAACGACT CC -3<sup>0</sup> , RP, 5<sup>0</sup> - ATCTTCTTGTTGGTCGTGGTG -3<sup>0</sup> , LB, 5<sup>0</sup> - AT TTTGCCGATTTCGGAAC -3<sup>0</sup> . An immunoblot assay was performed to examine the expression of the fusion protein LUC-SbSTOP1d. After seeds were surface sterilized and germinated on solid MS medium vertically for 5 days, uniform seedlings were transferred to solid medium containing 4.3 mM CaCl<sup>2</sup> and 3% sucrose at pH 4.5, with or without 50 µM AlCl<sup>3</sup> for 2 days, and their root growth was measured. At least 20 seedlings were measured for each treatment and independent experiments were performed three times.

## RESULTS

#### Sequence Analysis of SbSTOP1s

Using the amino acid sequences of AtSTOP1 and OsART1 as queries, four sweet sorghum STOP1-like genes, named SbSTOP1a (Sb01g001950.1), SbSTOP1b (Sb04g023670.1), SbSTOP1c (Sb07g023890.1), and SbSTOP1d (Sb03g041170.1), were identified in the sorghum genome database. The SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d coding regions are 795, 1185, 1290, and 1560 bp, respectively, and they encode proteins of 264, 394, 429, and 519 amino acids, respectively. All SbSTOP1s contain four putative Cys2His<sup>2</sup> zinc finger domains that resemble those of AtSTOP1, OsART1 and other homologs in different species (**Figure 1A**). SbSTOP1d shows the highest similarity to AtSTOP1 and OsART1, with identities of 54.9 and 48.5%, respectively, while SbSTOP1a, SbSTOP1b, and SbSTOP1c share relatively lower identities with AtSTOP1 and OsART1 (**Figure 1A**). Phylogenetic analysis revealed that SbSTOP1d clusters closely with AtSTOP1 and OsART1 compared to the other three SbSTOP1s. SbSTOP1b and SbSTOP1c cluster more closely with AtSTOP2 (**Figure 1B**).

## Detection of SbSTOP1 Expression Patterns

The expression patterns of the SbSTOP1s were investigated using quantitative real-time PCR. A time-course experiment indicated that Al induced a gradual increase in SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d expression in root apices (0–1 cm) during the entire 24 h Al treatment, though with different transcript abundances (**Figure 2A**). In addition, the four SbSTOP1s showed increased transcriptional abundances in a dosage-dependent manner when the roots were exposed to increasing external Al concentrations for 24 h (**Figure 2B**). All SbSTOP1s were mainly expressed in roots rather than shoots, and their expression levels in basal roots (1–2 cm) and roots (2–3 cm) were higher than that detected in root apices (0– 1 cm) regardless of Al stress (**Figure 2C**). Al stress induced increasing SbSTOP1s expression in roots (especially in root apices), but there was no detectable effect on the expression of SbSTOP1s in shoots (**Figure 2C**). We also compared the expression of the four SbSTOP1s under Al stress with their expression under other metal and proton stress. The expression of SbSTOP1c was specifically induced by Al stress, while the expression of SbSTOP1a, SbSTOP1b, and SbSTOP1d was induced only by Al and Cd stress but not by other metals (Supplementary Figures 1A–D). In addition, as shown in Supplementary Figure 2, the expression of SbSTOP1a, SbSTOP1b, and SbSTOP1c was increased when the pH value of the treatment solution decreased. A notable exception was the expression of SbSTOP1d, which was relatively unaffected by low pH stress, similar to OsART1 (Yamaji et al., 2009).

## The Subcellular Localization and Transcriptional Ability of SbSTOP1s

The main transcriptional characteristics of the SbSTOP1s were examined, including the subcellular localization, transcriptional activity and DNA-binding property. YFP-SbSTOP1a, YFP-SbSTOP1b, YFP-SbSTOP1c, and YFP-SbSTOP1d fusion genes under the control of the cauliflower mosaic virus 35S promoter were transiently introduced into Arabidopsis protoplasts. As shown in **Figure 3**, the YFP-SbSTOP1a, YFP-SbSTOP1b, YFP-SbSTOP1c, and YFP-SbSTOP1d fusion proteins were strictly localized to the nucleus, while the control YFP protein was distributed throughout the cytosol and nucleus.

The transcriptional activity of SbSTOP1s was assessed in the yeast expression system. SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d were fused to the GAL4 DNA-BD. The resulting plasmids were transformed into the Y2HGold yeast strain with a His auxotrophic marker. As observed in **Figure 4A**, yeast cells carrying BD-SbSTOP1a, BD-SbSTOP1b, BD-SbSTOP1c, and BD-SbSTOP1d grew well in SD medium without His. In contrast, yeast cells containing the GAL4 DNA-BD alone did not. These results indicated that all four SbSTOP1s have transcriptional activity.

We further investigated the DNA-binding property of SbSTOP1d due to its high similarity to AtSTOP1 and OsART1 (**Figure 1**). SbMATE (Sb03g043890), the first and also one of the few reported Al tolerance genes in sorghum (Magalhaes et al., 2007), and SbSTAR2 (Sb09g001990), an ortholog of OsSTAR2 that is transcriptionally regulated by OsART1 (Yamaji et al., 2009; Tsutsui et al., 2011), were both examined as potential downstream genes using the HEK293 coexpression system (Gao et al., 2015) and a dual-luciferase reporter assay. We introduced the SbMATE/SbSTAR2 promoter to drive the firefly luciferase reporter gene with the Renilla luciferase gene as an internal control (**Figure 4B**). As an effector, full-length SbSTOP1d under the control of the cytomegalovirus (CMV) promoter (**Figure 4B**) was co-transformed with the above reporter into HEK293 cells, and luciferase activity was detected. Both the SbMATE and SbSTAR2 promoter-driven reporters showed higher luciferase activity in the presence of the SbSTOP1d effector compared to the vector-only effector, though the SbSTAR2 promoter-driven reporter showed higher absolute value and significant differences at P < 0.01 compared with the SbMATE promoter-driven reporter (**Figures 4C,D**). SbSTOP1a, SbSTOP1b, and SbSTOP1c also showed weak or positive effects on the expression of SbMATE (Supplementary Figure 3). These results demonstrated that SbSTOP1d interacts with the SbSTAR2 and SbMATE promoters to act as a transcriptional activator.

Vigna umbellata (VuSTOP1, KP637172). Horizontal lines indicate zinc finger (ZF) domains, and asterisks show conserved Cys2His<sup>2</sup> or Cys2His2-Cys motifs as predicted (Iuchi et al., 2007). (B) Phylogenic analysis of SbSTOP1s and the above homologous proteins. The phylogenetic tree was constructed according to the neighbor-joining method using MEGA 5.1.

### SbSTOP1d Interacts With Itself or SbSTOP1b in Plants

We performed a yeast two-hybrid assay to screen for proteins that interact with SbSTOP1d, with the N-terminal fragment of SbSTOP1d [SbSTOP1d-NT, 1-275 aa, truncated before the zinc finger (ZF) domain] as bait, since its autoactivation could be readily suppressed by adding 3 mM 3-AT in SD medium (Supplementary Figure 4). Interestingly, the potential interaction proteins of SbSTOP1d included SbSTOP1d itself and SbSTOP1b. Yeast two-hybrid validation indicated that yeast cells co-transformed with SbSTOP1d-NT and SbSTOP1b grew well on SD/-Trp-Leu-His medium (**Figure 5A**). In addition, yeast cells containing SbSTOP1d-NT and SbSTOP1d showed similar result as the above (**Figure 5B**). Moreover, β-galactosidase assays showed that the β-galactosidase activities of yeast cells co-transformed with SbSTOP1d-NT and SbSTOP1b,

and with SbSTOP1d-NT and SbSTOP1d were approximately 80 times and 450 times that of the control (**Figure 5C**). These results demonstrated that SbSTOP1d can interact with SbSTOP1d itself and SbSTOP1b at the N-terminal region (1–275 aa, not include ZF domain). SbSTOP1a and SbSTOP1c, nevertheless, showed no interaction with SbSTOP1d (Supplementary Figure 5). We further tested whether SbSTOP1d could interact with SbSTOP1b or itself in plant cells using the BiFC assay as described previously (Meng et al., 2013). SbSTOP1d was fused to the N-terminal fragment of YFP or to the C-terminal fragment of CFP, and SbSTOP1b was fused to the C-terminal fragment of CFP. Different pairs of constructs were co-transformed into Arabidopsis protoplasts while the protein–protein interaction was observed under a microscope (**Figure 5D**) and analyzed semi-quantitatively by measuring the percentage of cells that showed reconstituted YFP activity (**Figure 5E**). These results demonstrated that SbSTOP1d interacted with SbSTOP1d itself (self-association) as well as with SbSTOP1b in plants (**Figures 5D,E**), suggesting that SbSTOP1d might function as a homo- and/or heterodimer in plants. The homo- and/or heterodimerization of SbSTOP1d might facilitate its specificity and DNA-binding affinity, since this is a strategy used by other transcription factors (Crossley et al., 1995; Jakoby et al., 2002; Xu et al., 2015).

#### SbSTOP1d Overexpression in Arabidopsis Confers Aluminum Tolerance

The primary symptom of Al toxicity is a rapid inhibition of root growth (Foy, 1988; Kochian et al., 2004). The Atstop1 mutant showed a root inhibition phenotype under Al stress, while the overexpression of PpSTOP1, PnSTOP1, or NtSTOP1 in the Atstop1 mutant could recover the Al- sensitive phenotype to varying extents (Ohyama et al., 2013; Fan et al., 2015). To further examine the function of SbSTOP1d, we introduced LUC-SbSTOP1d under the control of the CaMV 35S promoter

in the Atstop1 mutant background (SALK 114108). After a three-primer PCR-based genotyping (Supplementary Figure 6) and an immunoblot analysis for the LUC-SbSTOP1d fusion protein (**Figure 6A**), two independent complemented lines expressing SbSTOP1d were selected for phenotypic analysis. As shown in **Figures 6B,C**, the root growth of the WT, Atstop1, and two complemented lines was similar in the absence of Al. In the presence of Al, the root growth of WT was inhibited, with a relative root elongation (RRE) of 65%. Atstop1, which is sensitive to Al, had only 35% RRE, and in contrast, the two SbSTOP1d complemented lines greatly recovered the Al sensitivity characteristic of the Atstop1 mutant, with 55 and 60% RRE, respectively. These results indicated that heterologous expression of SbSTOP1d improved the Al tolerance of the transgenic plants.

#### DISCUSSION

Aluminum can be a beneficial element for some plant species at low concentrations. At pH values below 5, aluminum concentration (Al3+) rises sharply, inhibiting root growth and function, which leads to significant reductions in crop yields (Foy, 1983; Liu et al., 2014; Bojórquez-Quintal et al., 2017; Moreno-Alvarado et al., 2017). Transcription factors, such as STOP1 and WRKY46 in Arabidopsis, ART1 and ASR5 in rice play important roles in Al signal perception and transduction (Sawaki et al., 2009; Yamaji et al., 2009; Ding et al., 2013; Arenhart et al., 2014; Xu et al., 2017). Moreno-Alvarado et al. (2017) recently reported for the first time the induction of transcription factor NAC gene expression in Al-treated rice plants. Among them, STOP1 like proteins have been shown to be key transcription factors and investigated in many plant species, including Arabidopsis (AtSTOP1, AtSTOP2), rice (OsART1), tobacco (NtSTOP1), eucalyptus (EguSTOP1), and rice bean (VuSTOP1) (Iuchi et al., 2007; Yamaji et al., 2009; Ohyama et al., 2013; Kobayashi et al., 2014; Sawaki et al., 2014; Fan et al., 2015), yet a STOP1-like protein has never been characterized in sorghum. We isolated four sweet sorghum genes, SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d, encoding proteins containing four conserved C2H<sup>2</sup> zinc finger domains, similar to other homologous proteins (**Figure 1**). Compared with the other three SbSTOP1s, SbSTOP1d shares higher similarity with AtSTOP1/OsART1 (**Figure 1**), its expression level was relatively higher under Al stress (data not shown) and it was specifically affected by Al stress but not low pH stress (Supplementary Figure 2D), thus, SbSTOP1d was further investigated and confirmed to be effective for Al tolerance in plants (**Figure 6**).

As previously reported, the expression levels of AtSTOP1 and OsART1 were not significantly affected by Al (Iuchi et al., 2007; Yamaji et al., 2009). However, in this study, the expression of SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d in roots was significantly induced by Al in a time- and Al concentrationdependent manner (**Figure 2**). These results suggested that SbSTOP1s could respond to Al toxicity as early as transcriptional regulation and that different Al response mechanisms may exist between SbSTOP1s and AtSTOP1/OsART1. VuSTOP1 shares similar expression patterns with the SbSTOP1s, since its expression is induced by Al stress, but VuSTOP1 expression is also affected by low pH (Fan et al., 2015). In sweet sorghum, SbSTOP1a, SbSTOP1b, and SbSTOP1c expression were induced by low pH (pH 3.5), but SbSTOP1d expression showed little change under proton stress (Supplementary Figure 2). Thus, it is possible that different STOP1-like proteins from various species or even from the same species differ in function. In addition, Al- and low pH- tolerance seem to be regulated by different gene groups that belong to the STOP1-regulated system (Kobayashi et al., 2014). Therefore, in this study, we focused on SbSTOP1d for further functional analysis due to its high sequence identity to AtSTOP1 and OsART1 and because its expression was specifically affected by Al toxicity but not proton stress. Further research could be conducted to investigate the detailed characterizations of different SbSTOP1s in sweet sorghum.

All SbSTOP1s localized to the nucleus (**Figure 3**), and this agreed with the expectation for transcription factors. The four SbSTOP1s displayed transcriptional activity. SbSTOP1d-CT (276–519 aa, including ZF domains) showed stronger transcriptional activity than that of SbSTOP1d-NT (1–275 aa), which could not be inhibited with 15 mM 3-AT in SD medium (Supplementary Figure 4).

STOP1-like proteins generally regulate the transcription of Al tolerance genes by binding to the cis-acting element of the

promoter. OsART1 regulates multiple genes implicated in Al tolerance, and most of these genes (e.g., OsSTAR1, OsSTAR2) possess a cis-acting element as GGN(T/g/a/C)V(C/A/g)S(C/G) in their promoter (Yamaji et al., 2009; Tsutsui et al., 2011). AtSTOP1 also regulates several genes such as AtALMT1 and AtMATE1 in response to Al toxicity (Liu et al., 2009; Sawaki et al., 2009). Therefore, we tested whether SbSTOP1d regulates the transcription of two typical Al-associated genes, SbMATE (involved in Al exclusion mechanisms) and SbSTAR2 (involved in Al tolerance mechanisms). Both genes contain the above putative cis-acting element in their promoters. SbSTOP1d showed a positive effect on the expression of SbMATE and SbSTAR2 (**Figures 4C,D**), though the SbMATE promoter-driven reporter showed relatively lower luciferase activity than the SbSTAR2 promoter-driven reporter did. Similarly, VuSTOP1 can bind only weakly to the promoter of VuMATE (Fan et al., 2015). It was reported that the promoter of SbMATE harbored a tourist like miniature inverted repeat transposable element (MITE). The copy number (sequence repeats) of this MITE, which varied in different sorghum accessions, was positively correlated with Al tolerance (Magalhaes et al., 2007). Thus, the expression level of SbMATE in the sweet sorghum cultivar we used may also be regulated by this transposable element. It is unlikely, but we cannot exclude the possibility that there are some other cis-acting elements away from the tested promoter (−2039 bp) of SbMATE, since VuSTOP1 can also interact with a DNA sequence lacking

the putative GGN(T/g/a/C)V(C/A/g)S(C/G) cis-acting element (Fan et al., 2015). In addition, even though an increasing expression level of SbSTOP1s was induced by Al (**Figure 2**), some post-translational modifications may restrict the transcriptional activity of SbSTOP1. These modifications, such as protein phosphorylation, are frequently involved in the activation of transcription factors in response to biotic and abiotic stress, e.g., tomato PSEUDOMONAS TOMATO RESISTANCE (PTO) kinase phosphorylates PTI4 to increases the DNA-binding ability of PTI4 (Singh et al., 2002); Phosphorylation of ABAresponsive element binding proteins (AREB) was suggested to be involved in their activation (Uno et al., 2000). Thus, complex SbMATE regulation pathways may exist in sorghum. SbMATE-dependent citrate excretion is an important Al exclusion mechanism in sorghum (Magalhaes et al., 2007), while our study suggested that the SbSTOP1-dependent Al tolerance mechanism may blaze another trail, i.e., SbSTOP1 transcriptionally regulates SbSTAR2 (**Figure 4D**) to fulfill its Al resistance function.

Homo- and/or heterodimerization of transcription factors occurs frequently to facilitate their function at diverse promoters or bring together/stabilize two regulatory elements. Plant basicleucine zipper (bZIP) transcription factors form homodimers or heterodimers to bind DNA and trans-activate downstream gene expression (Schindler et al., 1992; Jakoby et al., 2002). Several types of zinc-finger motifs in transcription factors function as parts of DNA-binding and protein–protein interaction domains,

SbSTOP1b; 1d, SbSTOP1d. Data represent the means ± SD from three independent biological replicates. Columns with different letters are significantly different at P < 0.05.

assays showing the association of SbSTOP1d and SbSTOP1b and the self-association of SbSTOP1d in Arabidopsis protoplasts. YFP, YFP fluorescence; Auto, autofluorescence; Bright, bright field. Scale bar indicates 100 µm. (E) The percentage of protoplasts that exhibit BiFC fluorescence signals was calculated. 1b,

e.g., GATA-1 in erythroid cells self-associates mediated by its zinc finger domain to influence transcription (Crossley et al., 1995). These studies provide a clue that the self-association of SbSTOP1d and association of SbSTOP1d and SbSTOP1b may also be beneficial for the DNA-binding property of SbSTOP1d. Differing from those of other zinc finger TFs, the association of SbSTOP1d itself (or with SbSTOP1b) in the yeast assays occurred at the N-terminal region, which lacks zinc finger domains (**Figures 5A–C**). In addition, SbSTOP1b shows the highest identity with AtSTOP2, except for SbSTOP1c in the sorghum genome database, and closely clusters with AtSTOP2 (**Figure 1B**). AtSTOP2, a homolog of AtSTOP1, was reported to activate transcription of some of the genes regulated by AtSTOP1 (Kobayashi et al., 2014), while the regulatory roles (individual roles or possible partnership) of AtSTOP1 and AtSTOP2 in Aland low pH-tolerance have not yet been clarified. In the present study, SbSTOP1d and SbSTOP1b, as the homologous proteins of AtSTOP1 and AtSTOP2, respectively, displayed protein-protein interaction (**Figures 5A,C**). This result may help to further clarify the complex signal transduction pathways of STOP1-like proteins in response to Al and/or proton stress.

Complementation assays were conducted introducing multiple AtSTOP1 orthologous genes in the Atstop1 mutant background, and the results varied. Overexpression of PpSTOP1 and PnSTOP1 could almost fully or partially rescue the Alsensitive phenotype of Atstop1, while CsSTOP1 was somewhat effective in transgenic lines (Ohyama et al., 2013; Sawaki et al., 2014). Here, SbSTOP1d greatly recovered the Al-sensitive phenotype of the Atstop1 mutant, with nearly 90% of the root elongation of WT, demonstrating the Al tolerance function of SbSTOP1d in plants (**Figure 6**).

In summary, we have identified four STOP1-like genes (SbSTOP1a, SbSTOP1b, SbSTOP1c, and SbSTOP1d) in sweet sorghum that encode C2H<sup>2</sup> zinc finger transcription factors. The expression of all four genes in roots was upregulated by Al stress. Heterologous expression of SbSTOP1d in Atstop1 enhanced the Al tolerance of transgenic plants. SbSTOP1d interacted with itself (self-association) and SbSTOP1b in plants. These results provide a complete characterization of the SbSTOP1s in sweet sorghum and extend the understanding of STOP1-like transcription factors regulating Al tolerance in different plant species.

## AUTHOR CONTRIBUTIONS

ZY designed the research and revised the manuscript. SH, JG, YL, KG, SY, and MZ conducted the experiments. SH and JY analyzed the data. JG wrote the manuscript.

## FUNDING

This work was supported by the National Natural Science Foundation of China (31572192 and 31701982) and the China Postdoctoral Science Foundation (2017M611327).

## SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.00258/ full#supplementary-material

#### REFERENCES

fpls-09-00258 February 26, 2018 Time: 17:56 # 11


soybean GmbZIP60 gene. J. Plant Biochem. Biotechnol. 24, 9–17. doi: 10.1007/ s13562-013-0228-4


sweet sorghum (Sorghum bicolor L.) show reduced callose deposition and increased tolerance to aluminium toxicity. Plant Cell Environ. 38, 1178–1188. doi: 10.1111/pce.12472

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Huang, Gao, You, Liang, Guan, Yan, Zhan and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Cell Cycle Checkpoint Regulator ATR Is Required for Internal Aluminum Toxicity-Mediated Root Growth Inhibition in Arabidopsis

Yang Zhang1,2† , Jinliang Guo1,2† , Mo Chen<sup>1</sup> , Lun Li1,2, Lihua Wang<sup>3</sup> and Chao-Feng Huang1,2 \*

<sup>1</sup> College of Resources and Environmental Sciences, Nanjing Agricultural University, Nanjing, China, <sup>2</sup> Shanghai Center for Plant Stress Biology, National Key Laboratory of Plant Molecular Genetics, CAS Center for Excellence in Molecular Plant Sciences, Chinese Academy of Sciences, Shanghai, China, <sup>3</sup> Flower Research Institute, Yunnan Academy of Agricultural Sciences, Kunming, China

#### Edited by:

Jian Li Yang, Zhejiang University, China

#### Reviewed by:

He Qiang Lou, Zhejiang A & F University, China Wei Fan, Yunnan Agricultural University, China Cuiyue Liang, South China Agricultural University, China

> \*Correspondence: Chao-Feng Huang cfhuang@sibs.ac.cn

†These authors have contributed equally to this work.

#### Specialty section:

This article was submitted to Plant Nutrition, a section of the journal Frontiers in Plant Science

Received: 13 November 2017 Accepted: 22 January 2018 Published: 14 February 2018

#### Citation:

Zhang Y, Guo J, Chen M, Li L, Wang L and Huang C-F (2018) The Cell Cycle Checkpoint Regulator ATR Is Required for Internal Aluminum Toxicity-Mediated Root Growth Inhibition in Arabidopsis. Front. Plant Sci. 9:118. doi: 10.3389/fpls.2018.00118 Aluminum (Al) can target multiple sites of root cells for toxicity, including the cell wall, the plasma membrane and symplastic components. Previous work revealed that the cell cycle checkpoint regulator (ATR) Ataxia Telangiectasia-mutated and Rad3-related is required for Al toxicity-induced root growth inhibition in als3 and that the symplastic component DNA is an important target site of Al for the toxicity. However, whether monitoring DNA integrity through ATR-regulated pathway is required for Al-induced root growth inhibition in other Al-sensitive mutants remains unknown. In this study, we demonstrated that the atr mutation could also rescue the Al hypersensitivity and Alinduced cell cycle arrest in star1, which supports the hypothesis that ALS3 and STAR1 function together to be involved in the detoxification of Al in Arabidopsis. However, mutation of ATR could not rescue the Al-sensitive phenotype of almt1 or stop1, both of which are defective in external detoxification mechanisms of Al. We further showed that the Al hypersensitivity and Al-induced quiescent center (QC) differentiation in als1 could also be rescued by the atr mutation. Therefore, our results suggest that ATRregulated pathway is involved in the modulation of internal Al toxicity-mediated root growth inhibition in Arabidopsis.

Keywords: aluminum toxicity, Arabidopsis thaliana, ATR, cell cycle checkpoint, DNA damage, external, internal

#### INTRODUCTION

Aluminum (Al) comprises about 7% of the earth's crust and is the most abundant metallic element. In neutral or alkaline soils, Al exists as insoluble aluminosilicates or oxides, which are non-toxic to plants. However, in acid soils with a pH of 5.5 or lower, solubilization of Al is enhanced and phytotoxic forms of Al are released into soil to levels that affect root growth. As a consequence, Al toxicity on acid soils becomes one of the most severe global problems since these soils comprise approximately 50% of the world's potentially arable land (von Uexkull and Mutert, 1995; Kochian et al., 2004).

In acidic soils, Al exists as the octahedral hexahydrate Al(H2O)<sup>6</sup> <sup>3</sup>+, which is more commonly referred to as Al3+. The phytotoxic Al3<sup>+</sup> is the hardest Lewis acid, which is characterized by

Zhang et al. ATR for Internal Al Toxicity

a low covalent and a high ionic index. Hard metal ions have strong interactions with organic molecules bearing oxygen groups (Poschenrieder et al., 2008). Therefore, Al3<sup>+</sup> preferentially binds to phosphate, sulfate, and carboxyl groups for toxicity. Considering the components of a plant cell, Al is believed to target multiple sites for toxicity, including the cell wall, the plasma membrane and inside the cells. Cell walls and intercellular spaces are the first sites of the root in contact with Al when the roots are exposed to Al. Many studies have shown that most of the Al is bound to the cell wall. The ratio of cell wall Al to the total Al has been reported to range from 85 to 99.9% (Ma, 2007). Al can also bind to the plasma membrane and alter the membrane fluidity and surface potential (Kinraide, 2001), block ion channel activity (Pineros and Kochian, 2001), and induce the reactive oxygen species (ROS) as well as lipid peroxidation on the plasma membrane (Yamamoto et al., 2001). Furthermore, a small portion of Al can enter the symplasm rapidly and may interact with a number of symplastic targets (Lazof et al., 1996; Silva et al., 2000). For example, Al disrupts the cytoskeleton by interacting with both microtubules and actin filaments (Grabski and Schindler, 1995; Blancaflor et al., 1998), and blocks signal transduction pathways, particularly in Ca2<sup>+</sup> homeostasis and signaling (Jones and Kochian, 1995; Jones et al., 1998; Zhang and Rengel, 1999). Al can also interact with DNA (Karlik et al., 1980; Karlik and Eichhorn, 1989), which is expected to have serious effects on gene expression and chromosome structure.

To cope with Al toxicity, plants have evolved Al-resistance mechanisms, including external and internal detoxification of Al (Ma et al., 2001; Kochian et al., 2004). In Arabidopsis thaliana, external detoxification of Al is primarily achieved through AtALMT1-mediated secretion of malate to form a non-toxic form of Al-malate in the apoplast (Hoekenga et al., 2006), and the citrate transport AtMATE play a minor role in the external detoxification of Al (Liu et al., 2009). STOP1, a C2H2 transcription factor, is involved in the detoxification of Al mainly through the regulation of AtALMT1 expression (Iuchi et al., 2007). For the internal detoxification of Al, the tonoplast-localized ATP-binding cassette (ABC) transporter ALS1 is required, which tolerates Al presumably via the transport of cytosolic Al into vacuoles (Larsen et al., 2007). STAR1 and STAR2/ALS3 encode a nucleotide-binding domain and transmembrane domain of a bacterial-type ABC transporter, respectively, and are suggested to be involved in Al tolerance through modification of cell wall or redistribution of Al from Al-sensitive root tips to other less Al-sensitive tissues (Larsen et al., 2005; Huang et al., 2009, 2010). Recently, Dong et al. (2017) reported that unlike rice STAR1 and STAR2, Arabidopsis ALS3 interacts with AtASTAR1 to be localized to the tonoplast, suggesting that AtSTAR1/ALS3 might be also required for the internal detoxification of Al.

Through the screening of the suppressors of the Al hypersensitivity of als3 mutant, Gabrielson et al. (2006) identified a dozen of suppressor mutants, and two of them had different mutations on the same gene ATR (Rounds and Larsen, 2008). ATR (Ataxia Telangiectasia-mutated and Rad3-related) is a cell cycle checkpoint regulator that functions in detecting DNA damage and then halting cell division (Culligan et al., 2004). atr mutant is hypersensitive to clastogenic and genotoxic stresses, but shows increased tolerance to Al because of failure to halt cell cycle progression. Together with the recovery of the Al hypersensitivity of als3 by the atr mutation, the results suggest that Al acts as a mild genotoxic agent and can target DNA to arrest root growth through ATR-regulated pathway (Rounds and Larsen, 2008).

In this study, to determine whether ATR-dependent pathway is required for the Al hypersensitivity in all Al-sensitive mutants, we created a series of double mutants between Al-sensitive mutants and atr mutant and then evaluated their sensitivity to Al in Arabidopsis. Our results revealed that the atr mutation could rescue the Al-sensitive phenotype of als3, star1 and als1, but not that of almt1 and stop1. These findings suggest that ATRregulated pathway is required for internal Al toxicity-induced root growth inhibition.

#### MATERIALS AND METHODS

#### Plant Materials and Growth Conditions

Arabidopsis thaliana (Columbia ecotype, Col-0) was used for all the control experiments. The T-DNA insertion lines atr (SALK\_032841C), star1 (GABI\_762A06), als3 (SALK\_004094), stop1 (SALK\_114108), almt1 (SALK\_00962) and the mutant als1-1 (CS3847) were all derived from uNASC<sup>1</sup> . Plants were grown in a growth chamber or controlled room at 22–25◦C with 14 h of light and 10 h of darkness.

#### Mutant Genotyping

To select homozygous mutants of atr, als3, star1, stop1, and almt1, primer pairs flanked each T-DNA insertion were used as follows: ATR (5<sup>0</sup> -ACTGCATGCCAT TTACTCCTAC-3<sup>0</sup> and 5 0 -GATCAGCTTGATCATCCAAACT-3<sup>0</sup> ), ALS3 (5<sup>0</sup> - CAA TGTTCTTGCTCGTCCTCCT-3<sup>0</sup> and 5<sup>0</sup> -TGGTTCACGTAGTG GGCCATCG-3<sup>0</sup> ), STAR1 (5<sup>0</sup> -TCGTAGAGTTGGAATGCTTTT TC-3<sup>0</sup> and 5<sup>0</sup> -GTTGAAGAAACCTCTGTGCCATT-3<sup>0</sup> ), ALMT1 (50 -TTGAGAGAGCTGAGTGACCA-3<sup>0</sup> and 5<sup>0</sup> -ACAAC GA TATCAGCGCGAAC-3<sup>0</sup> ), and STOP1 (5<sup>0</sup> -TCTTAAAGCGG CCATTGGTG-3<sup>0</sup> and 5<sup>0</sup> -TTAGAGACTAGTATCTGAAACAG ACTCAC-3<sup>0</sup> ). For als1-1 mutant, a dCAPS (derive Cleaved Amplified Polymorphic sequences) marker was developed by using a primer pair (5<sup>0</sup> -TGTGAAACAGTTTGGTCGCT-3<sup>0</sup> and 5<sup>0</sup> -TGCGTTTAGTCCTCCGAAGA-3<sup>0</sup> ) and a restriction endonuclease TfiI. To generate double or triple mutants, crosses were made between atr and each Al-sensitive mutant or between als3atr and star1 and then the derived F2 plants were genotyped and selected. For genotyping of CyclinB1;1 and QC46 marker lines, a primer pair for the GUS gene was used (5<sup>0</sup> -ATGTTACGTCCTGTAGAAACC-3<sup>0</sup> and 5 0 -TCATTGTTTGCCTCCC TGCTGC-3<sup>0</sup> ).

#### RNA Isolation and Expression Analysis

Seeds were sterilized and stratified at 4◦C for 2 days and then sowed on a 0.3% Gellan gum (G1910; Sigma–Aldrich) nutrient

<sup>1</sup>http://szlapncs01.nottingham.ac.uk/

Scale bar = 1 cm.

medium consisting of 1 mM KNO3, 0.2 mM KH2PO4, 2 mM MgSO4, 0.25 mM (NH4)2SO4, 1 mM Ca(NO3)2, 1 mM CaSO4, 1 mM K2SO4, 1 µM MnSO4, 5 µM H3BO3, 0.05 µM CuSO4, 0.2 µM ZnSO4, 0.02 µM NaMoO4, 0.1 µM CaCl2, 0.001 µM CoCl<sup>2</sup> and 1% sucrose. After 7 days growth, the seedlings were transferred to a 0.5 mM CaCl<sup>2</sup> solution for 6 h pretreatment at pH 4.8 and then exposed to a 0.5 mM CaCl<sup>2</sup> solution (pH 4.8) with or without 20 µM AlCl<sup>3</sup> for 12 h. Total RNA was extracted using TaKaRa MiniBEST plant RNA Extraction Kit (Cat # 9769). Around one microgram total RNA was first digested with DNase I and then subjected for the synthesis of first-strand cDNAs by using HiScript <sup>R</sup> 1st Strand cDNA Synthesis Kit (Vazyme Biotech

Co., Ltd., Nanjing, China). One twentieth of the cDNA products and the SYBR <sup>R</sup> Green Master Mix kit (Vazyme Biotech Co., Ltd., Nanjing, China) were used for RT-PCR and real-time RT-PCR analysis. The primers for RT-PCR analysis of ATR, ALS3, STAR1, ALMT1, and STOP1 were same to those primers for genotyping as shown above. The primers for real-time RT-PCR analysis were as follows: ATR (5<sup>0</sup> -CTGACTGAGGACTGTGGTCTGGT-3<sup>0</sup> and 5 0 -GACGGTCACCAAGCCCAACA-3<sup>0</sup> ), ALS3 (5<sup>0</sup> -CGTATCTC TTCATGGTCTCTGTCG-3<sup>0</sup> and 5<sup>0</sup> -GTAACTCCGGTGACGGT CATG-3<sup>0</sup> ), STAR1 (5<sup>0</sup> -TTCAAGGGACTGTTGCGGATA-3<sup>0</sup> and 5 0 -AAGAGCACTTGTTGGTTCATCG-3<sup>0</sup> ), ALS1 (5<sup>0</sup> -GCCTCA CAGTTGGTTCATCGG-3<sup>0</sup> and 5<sup>0</sup> -GTCGTTTTTCCTCCACCG CT-3<sup>0</sup> ), ALMT1 (5<sup>0</sup> -TGCAAGCTGCGTTGTCGAC-3<sup>0</sup> and 5 0 -CAAAATCTTGAAGGAAGTGGGAG-3<sup>0</sup> ) and STOP1 (5<sup>0</sup> - TCACATAGCTCTGTTCCAGGGA-3<sup>0</sup> and 5<sup>0</sup> -ATCAGTCATTC CAGGCTGTGT-3<sup>0</sup> ). UBQ10 was used as an internal control and the forward and reverse prime sequences of UBQ10 are 5 0 -CGTCTTCGTGGTGGTTTCTAA-3<sup>0</sup> and 5<sup>0</sup> -GGATTATACA AGGCCCCAAAA-3<sup>0</sup> , respectively.

#### Evaluation of Sensitivity to Al

For assessment of Al sensitivity in hydroponic conditions, we referred to a previous method with slight modifications (Huang et al., 2010). Briefly, seeds of each line were stratified at 4◦C for 2 days and then sowed on a plastic mesh floating on a 1/30 strength Hoagland nutrient solution (NH4H2PO<sup>4</sup> omitted) plus 1 mM CaCl<sup>2</sup> and different concentrations of AlCl<sup>3</sup> at pH 5.0 for 7 days. The solution was renewed every 3 days. After the treatment, the seedlings were photographed and root length was measured by ImageJ. Relative root growth expressed as (root length with Al treatment/root length without Al) × 100 was used to evaluate the Al sensitivity. For soaked gel experiments, we adopted the method developed by Larsen et al. (2005). Nutrient agar medium was first prepared, which consisted of 50 ml of 1 mM KNO3, 0.2 mM KH2PO4, 2 mM MgSO4, 0.25 mM (NH4)2SO4, 1 mM Ca(NO3)2, 1 mM CaSO4, 1 mM K2SO4, 1 µM MnSO4, 5 µM H3BO3, 0.05 µM CuSO4, 0.2 µM ZnSO4, 0.02 µM NaMoO4, 0.1 µM CaCl2, 0.001 µM CoCl2, 1% sucrose, and 0.3% Gellan gum (G1910; Sigma–Aldrich). The agar medium was then soaked with 25 ml of the same nutrient medium containing 0, 0.5, 0.75, or 1 mM AlCl3. After 2 days soaking, the solution was removed and seeds were grown on the agar medium plates for 7 days. The seedlings were then pictured and compared and the root length was measured by ImageJ.

#### GUS Activity Assay

To investigate the effect of Al on Cyclin B1;1 accumulation, seeds of CycB1;1:GUS –containing WT, atr, star1, and star1atr were grown on a soaked gel medium containing 0 or 0.5 mM AlCl<sup>3</sup> for 7 days. The seedlings were then stained with a commercialized GUS staining solution (161031; O'Biolab Co., Ltd., Beijing, China) for 2 h at 37◦C. For determination of the status of the quiescent center (QC) after Al treatment, seeds of QC46 (GUS-based QC marker)-containing WT, atr, als1, and als1atr were grown on a soaked gel medium containing 0 or 1.5 mM AlCl3. After growth for 7 days, the seedlings were stained with the GUS staining solution overnight at 37◦C. Stained tissues were observed and photographed with a microscope (Olympus BX53F, Japan).

## RESULTS

## Mutation of ATR Rescued the Al-Sensitive Phenotype of Both als3 and star1 Mutants

To confirm the previous observation that mutation of ATR could rescue the Al-sensitive phenotype of als3 (Rounds and Larsen, 2008), we generated als3atr double mutant through a genetic cross between atr and als3 single mutants. RT-PCR analysis revealed that ATR and ALS3 were knocked out in respective single or double mutants (**Figure 1A**). We evaluated the tolerance of WT, atr, als3, and als3atr mutants to Al in both hydroponic and soaked gel conditions. Consistent with previous results, atr mutant showed more tolerance to Al than WT, and the atr mutation was able to reduce the sensitivity of als3 to Al at all Al concentrations (**Figures 1C,E,F**). Nevertheless, mutation of ATR was not able to fully rescue the Al-sensitive phenotype of als3, especially at high Al concentrations (**Figures 1C,E,F**), suggesting that other Al toxicity mechanisms are also required for Al-induced growth inhibition in als3 mutant. As STAR1 interacts with ALS3 to be involved in the regulation of Al tolerance in Arabidopsis (Huang et al., 2010; Dong et al., 2017), we investigated whether the atr mutation could also rescue the Al-sensitive phenotype of star1. We generated star1atr double mutant through crossing and genotyping and RT-PCR analysis confirmed that both STAR1 and ATR were knocked out in the double mutant (**Figure 1B**). Evaluation of Al tolerance in the

double mutant showed that star1atr was more tolerant to Al than star1 at all Al concentrations (**Figures 1D,G,H**), indicating that ATR is required for Al-induced growth inhibition in star1 mutant. Additionally, similar to that in als3atr mutant, mutation of ATR did not fully rescue the Al-sensitive phenotype of star1 (**Figures 1D,G,H**). We also generated star1als3 and star1als3atr mutants to further investigate whether mutation of ATR could rescue the Al sensitivity in star1als3 double mutant. Results showed that the Al-sensitive phenotype of star1als3 could also be rescued by the introduction of the atr mutation (**Figure 1I**). Together, these results confirm that STAR1 and ALS3 regulate Al tolerance through the same pathway and indicate that ATRdependent pathway is also required for Al-induced growth inhibition in star1 mutant.

Al-induced inhibition of root growth was correlated with the increase in the number of cells trapped in the G2 stage,

which causes the hyperaccumulation of Cyclin B1;1 in root tips (Rounds and Larsen, 2008). To examine the effect of Al on the accumulation of Cyclin B1;1 in star1 mutant background, we introduced CycB1;1:GUS into atr, star1 and star1atr through crossing. In the absence of Al, GUS expression was detected at relatively low levels in all the materials (**Figure 2**). After exposure to a low toxic level of Al, while GUS activity was slightly increased in WT, GUS expression in star1 was dramatically increased in root tips, suggesting that cell cycle progression was halted in star1 (**Figure 2**). In star1atr, GUS activity was detected at similar low

levels to that in WT and atr, which suggested that the arrest of cell cycle progression in star1 was rescued by the atr mutation. The Cyclin B1;1 expression results support the conclusion that knockout of ATR is able to rescue the Al hypersensitivity in star1.

#### The atr Mutation Could Not Rescue the Al Hypersensitivity in Either almt1 or stop1 Mutants

To investigate whether mutation of ATR could rescue the hypersensitivity of almt1 and stop1 to Al, we introduced the atr mutation into stop1 and almt1 mutants by crossing and genotyping, respectively. RT-PCR analysis confirmed that ALMT1 or STOP1 were knocked out in the corresponding mutants (**Figures 3A,B**). Phenotypic analysis of Al tolerance showed that the tolerance of almt1atr to Al did not differ from that of almt1 at all Al concentrations in both hydroponic and soaked gel conditions (**Figures 3C,D,G,H**), indicating that mutation of ATR could not rescue Al-sensitive phenotype of almt1. Similarly, Al tolerance in stop1atr was also not different from that in stop1 under all Al treatment (**Figures 3E,F**), demonstrating that the atr mutation was not able to rescue the Al-sensitive phenotype of stop1 either. These results suggest that ATR is not required for Al-induced growth inhibition in those Al-sensitive mutants that are defective in the external detoxification of Al.

#### The Al-Sensitive Phenotype of als1 Could Also Be Rescued by the atr Mutation

Since ATR is localized in the nucleus and required for Al-induced halting cell division in als3 or star1 (**Figures 1**, **2**), there are two possibilities that ATR might detect general internal Al toxicity signal or star1/als3-specific Al toxicity signal. To distinguish these two, we utilized another Al-sensitive mutant als1, which is deficient in the sequestration of Al into vacuoles (Larsen et al., 2007). Introduction of atr mutation into als1 mutant could also rescue its Al-sensitive phenotype at various Al concentrations (**Figures 4A,B**). These results imply that ATR is required for internal Al toxicity-mediated root growth inhibition.

We also determined the status of the QC after Al treatment by introduction of a GUS-based QC marker, QC46 (Sabatini et al., 2003), into atr, als1 and als1atr. Without Al treatment, GUS expression was well detected in all the materials (**Figure 4C**). However, in the presence of high levels of Al, GUS activity was lost in als1, suggesting that the essential stem cells required for maintenance of root growth was destroyed by Al toxicity in als1 mutant. In contrast, als1atr double mutant displayed normal GUS activity in the QC after Al treatment (**Figure 4C**). These results indicate that the atr mutation could help als1 mutant to maintain the QC integrity for root growth when exposure to highly toxic levels of Al.

#### Expression Pattern of ATR and Al-Resistance Genes

To examine whether ATR expression was altered in Al-sensitive mutants, we compared the expression level of ATR between WT and the Al-sensitive mutants. Results showed that there was no

significant difference in ATR expression between WT and the mutants in the absence of Al (**Figure 5A**). Al treatment slightly decreased the expression of ATR, but no significant difference in ATR expression was found in WT and the mutants. This result suggests that increased Al sensitivity of the mutants was not due to altered ATR expression. The expression of Al-resistance genes in atr mutant was also determined. The expression levels of the Al-resistance genes including ALS3, STAR1, ALS1, ALMT1, and STOP1 in atr mutant were similar to those in WT under both – Al and –Al conditions (**Figure 5B**), suggesting that increased Al tolerance in atr mutant was not caused by elevated expression of Al-resistance genes.

#### DISCUSSION

ATR functions as a cell cycle checkpoint to detect DNA damage and subsequently prevent cell division (Culligan et al., 2004).

Since knockout of ATR is able to rescue the Al hypersensitivity in als3 mutant (Gabrielson et al., 2006; Rounds and Larsen, 2008), two possible mechanisms exist for the increased sensitivity to Al in als3. One possible mechanism is that mutation of ALS3 results in the increased Al accumulation in nucleus and consequently activates ATR-regulated pathway to halt cell division and ultimately inhibit root growth. The other is that Al toxicity-induced specific signal in als3 activates ATR-regulated pathway to cause root growth inhibition. Our results showed that in addition to als3, mutation of ATR can also rescue Al-sensitive phenotype of star1 and als1, indicating that rescue of Al-sensitive phenotype by atr mutation is not specific to als3 mutant. Thus, we prefer the former hypothesis that elevated Al accumulation in the nucleus induces ATR-regulated pathway to inhibit root growth in als3 mutant.

In contrast to its hypersensitivity to clastogenic and genotoxic stresses, atr mutant shows increased tolerance to Al. Al in nucleus might bind to DNA non-covalently and induce a conformational alteration from the B-form to Z-DNA, which affects DNA unwinding during DNA replication (Anitha and Rao, 2002). Nevertheless, unlike other genotoxic stresses, Al is thought to be a mild DNA damage agent and its binding to DNA is likely to be reversible (Rounds and Larsen, 2008; Nezames et al., 2012). This unique interaction of Al with DNA can activate ATR-, ALT2-, and SOG1-regualted transcriptional response to halt cell division and cause the inhibition of root growth (Sjogren et al., 2015). However, it remains unknown about how the interaction of Al with DNA activates the ATR-regulated pathway and what the ATR-regulated downstream transcriptional events that lead to the cease of cell division are.

The inhibition of root growth can be attributed to the disruption of cell division and/or cell elongation. Rapid reduction in root growth suggests an initial impact of Al on cell elongation instead of cell division (Sharp et al., 1988; Kopittke et al., 2015). However, when roots are exposed to Al for a long period of time, inhibition of cell division might also contribute to the reduction of root growth. Al-activated ATR-regulated cease of cell division in als3/star1 or als1 suggests that inhibition of cell division plays a critical role in Al-induced inhibition of root growth in these Al-sensitive mutants. Further work is required to determine whether mutation of atr could rescue the Al-sensitive phenotype of these mutants after a short-term exposure to Al.

Numerous studies have suggested that Al can target multiple sites for toxicity, including apoplastic and symplastic components (Kochian, 1995; Ma, 2007). Nevertheless, it remains debatable about which sites play more important roles in Al-induced inhibition of root growth. We found that the atr mutation could not rescue the Al hypersensitivity in almt1 and stop1, which are defective in the capacity to detoxify Al externally. These results indicate that ATR is not required for Al-induced inhibition of root growth in all Al-sensitive mutants and suggest that both symplastic components such as DNA and apoplastic components including cell wall are important Al target sites that lead to root growth inhibition by Al toxicity. Additionally, our data showed that the atr mutation could not fully rescue the Al hypersensitivity in als3, suggesting that Al also targets other symplastic sites to cause root growth inhibition in als3 mutant.

In rice, OsSTAR1 interacts with OsSTAR2, the rice ortholog of ALS3, to form a functional complex that is suggested to be involved in the modification of cell wall that is required for Al detoxification (Huang et al., 2009). Although Arabidopsis AtSTAR1 can also interact with ALS3 to be involved in the detoxification of Al, AtSTAR1 and ALS3 are localized to tonoplast (Larsen et al., 2005; Huang et al., 2010; Dong et al., 2017), which are different from OsSTAR1 and OsSTAR2 that are localized to vesicle membranes (Huang et al., 2009). We found that in addition to als3, knockout of ATR also rescues the Al-sensitive phenotype of star1. Furthermore, the atr mutation can even rescue Al hypersensitivity in als3star1 double mutant. These results indicate that als3 and star1 share the same mechanism for their hypersensitivity to Al, i.e., ATR-regulated pathway required for Al-induced inhibition of root growth. The results also support the view that STAR1 and STAR2/ALS3 function together to be involved in the same pathway of Al detoxification. We further found that the Al hypersensitivity in als1 was rescued by the atr mutation. als1 has defects in the internal detoxification of Al (Larsen et al., 2007). Together, our results suggest that ATR is required for internal Al toxicity-induced inhibition of root growth and that STAR1 and ALS3 might be involved in the internal detoxification of Al in Arabidopsis. We propose that under Al stress conditions, internal Al detoxification-deficient mutants accumulate high levels of Al in the nucleus, which induces DNA damage and consequently activates ATR-regulated pathway and arrest cell cycle, finally leading to the inhibition of root growth.

#### AUTHOR CONTRIBUTIONS

All authors conceived the project. C-FH drafted the manuscript. YZ, JG, MC, LL, and LW performed the experiments. YZ and JG helped to analyze the data and write the manuscript. All authors read and approved the final manuscript.

## FUNDING

This work was supported by Jiangsu Science Fund for Distinguished Young Scholars (Grant No. BK20150027), National Natural Science Foundation of China (Grant No. 31570253 to C-FH), the Strategic Priority Research Program (Grant No. XDPB0404) of the Chinese Academy of Sciences, and the Shanghai Center for Plant Stress Biology, Chinese Academy of Sciences.

#### ACKNOWLEDGMENTS

We thank Prof. Philip N. Benfey from Duke University for kindly providing the QC46 marker.

### REFERENCES

fpls-09-00118 March 8, 2018 Time: 14:10 # 9


is required for aluminum tolerance in Arabidopsis. Plant J. 41, 353–363. doi: 10.1111/j.1365-313X.2004.02306.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhang, Guo, Chen, Li, Wang and Huang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# BoALMT1, an Al-Induced Malate Transporter in Cabbage, Enhances Aluminum Tolerance in *Arabidopsis thaliana*

Lei Zhang1†, Xin-Xin Wu1†, Jinfang Wang1†, Chuandong Qi <sup>1</sup> , Xiaoyun Wang<sup>1</sup> , Gongle Wang<sup>1</sup> , Mingyue Li <sup>1</sup> , Xingsheng Li <sup>2</sup> and Yang-Dong Guo<sup>1</sup> \*

*<sup>1</sup> Beijing Key Laboratory of Growth and Developmental Regulation for Protected Vegetable Crops, College of Horticulture, China Agricultural University, Beijing, China, <sup>2</sup> Shandong Huasheng Agriculture Co. Ltd, Shandong, China*

#### *Edited by:*

*Wei Fan, Yunnan Agricultural University, China*

#### *Reviewed by:*

*Qi Chen, Kunming University of Science and Technology, China Jiang Tian, South China Agricultural University, China*

*\*Correspondence:*

*Yang-Dong Guo yaguo@cau.edu.cn † These authors have contributed equally to this work.*

#### *Specialty section:*

*This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science*

*Received: 24 July 2017 Accepted: 06 December 2017 Published: 23 January 2018*

#### *Citation:*

*Zhang L, Wu X-X, Wang J, Qi C, Wang X, Wang G, Li M, Li X and Guo Y-D (2018) BoALMT1, an Al-Induced Malate Transporter in Cabbage, Enhances Aluminum Tolerance in Arabidopsis thaliana. Front. Plant Sci. 8:2156. doi: 10.3389/fpls.2017.02156* Aluminum (Al) is present in approximately 50% of the arable land worldwide and is regarded as the main limiting factor of crop yield on acidic soil. Al-induced root malate efflux plays an important role in the Al tolerance of plants. Here, the aluminum induced malate transporter *BoALMT1* (KF322104) was cloned from cabbage (*Brassica oleracea*). *BoALMT1* showed higher expression in roots than in shoots. The expression of *BoALMT1* was specifically induced by Al treatment, but not the trivalent cations lanthanum (La), cadmium (Cd), zinc (Zn), or copper (Cu). Subcellular localization studies were performed in onion epidermal cells and revealed that BoALMT1 was localized at the plasma membrane. Scanning Ion-selective Electrode Technique was used to analyze H <sup>+</sup> flux. *Xenopus* oocytes and *Arabidopsis thaliana* expressing *BoALMT1* excreted more H<sup>+</sup> under Al treatment. Overexpressing *BoALMT1* in transgenic *Arabidopsis* resulted in enhanced Al tolerance and increased malate secretion. The results suggested that *BoALMT1* functions as an Al-resistant gene and encodes a malate transporter. Expressing *BoALMT1* in *Xenopus* oocytes or *A. thaliana* indicated that BoALMT1 could increase malate secretion and H+ efflux to resist Al tolerance.

Keywords: aluminum tolerance, *BoALMT1*, cabbage, malates, SIET

## INTRODUCTION

Al is the most abundant metal and the third most abundant element, making up around 7% of the earth's crust (Tesfaye et al., 2001). When the soil pH value is lower than 5.0, the soluble aluminum in soil solutions is mostly present as the toxic Al3+, which inhibits root growth at micromolar concentrations in many species (Kochian et al., 2005). Micromole levels of Al3<sup>+</sup> can remarkably inhibit root elongation, and impair the absorption, of water and nutrients (Kochian et al., 2005). The well-known mechanism of plant Al tolerance is the Al-induced secretion of organic acids (OA) from the root tips. The OAs chelate Al3<sup>+</sup> and form the non-toxic compound OA-Al (Kochian et al., 2004; Horst et al., 2010; Ryan et al., 2011). The most common OAs involved in the Al detoxification process are malate, citrate, and oxalate, depending on the plant. For example, malate is used in wheat (Delhaize et al., 1993) and Arabidopsis (Hoekenga et al., 2003), citrate is secreted in maize (Pellet et al., 1995), and oxalate is used in buckwheat (Zheng et al., 2005) and tomato (Yang et al., 2008).

Wheat TaALMT1 (ALMT, for Al-activated Malate Transporter) encoding a malate transporter was the first plant gene involved in Al tolerance and the first ALMT family gene. In Al-tolerant wheat genotypes, TaALMT1 is specifically expressed in the root tips (Sasaki et al., 2004; Raman et al., 2005). Overexpression of TaALMT1 in wheat, barley, and tobacco-cell suspension increases the efflux of Al-activated malate and enhances tolerance to Al stress (Delhaize et al., 2004; Sasaki et al., 2004; Pereira et al., 2010). TaALMT1 homologs have now been isolated in Arabidopsis (Hoekenga et al., 2006), oilseed rape (Ligaba et al., 2006), rye (Collins et al., 2008), soybean (Liang et al., 2013), and Medicago sativa (Chen et al., 2013). Multi-antimicrobial extrusion (MATE) proteins are a family of proteins that function as drug/sodium or proton antiporters. MATE proteins can secrete organic anions to contribute to the Al tolerance in plants (Furukawa et al., 2007; Magalhaes et al., 2007; Wang et al., 2007; Liu et al., 2009). In Arabidopsis, the zinc finger transcription factor STOP1 (known as ART1 in rice) plays a critical role in plant Al tolerance by regulating the Al-inducible expression of ALMT and MATE (Liu et al., 2009). In rice, multiple genes implicated in Al tolerance, including MATE transporter family members, are regulated by the transcription factor ART1 (Yamaji et al., 2009).

Cabbage (Brassica oleracea) is one of the most important vegetable crops around the world (Wu et al., 2014). Our previous study has shown that BoMATE encodes a citrate transporter and is induced by Al and enhances aluminum tolerance in Arabidopsis (Wu et al., 2014). Here we report that cabbage BoALMT1 is located in the plasma membrane and induced by Al. A reverse genetic approach was used to characterize the functions of BoALMT1. Overexpressing BoALMT1 in Xenopus oocytes and Arabidopsis facilitated H<sup>+</sup> efflux. Overexpressing BoALMT1 in Arabidopsis resulted in enhanced Al tolerance and increased malate secretion. These results suggested that BoALMT1 has an important role in Al tolerance in cabbage.

## RESULTS

## Sequence Analysis of *BoALMT1* in Cabbage

The ALMT gene was the first Al3<sup>+</sup> tolerance gene identified in plants (Sasaki et al., 2004; Delhaize et al., 2007; Meyer et al., 2010). Membrane protein ALMTs possess 5–7 predicted transmembrane domains and a UPF0005 domain with unknown function (Delhaize et al., 2007). BoALMT1 (KF322104) cloned from cabbage contains an open reading frame of 1,497 bp, encoding a polypeptide of 498 amino acids. BLAST analysis revealed that the sequence of BoALMT1 was a 99% match to BnALMT1 from rape, 73% match to AtALMT1 from Arabidopsis, and 33% match to TaALMT1 from wheat. The HMMTOP transmembrane topology prediction server was used to predict the localization of helical transmembrane segments and the analysis indicated that BoALMT1 contained 5 predicted transmembrane domains (**Figure 1A**). Analysis of BoALMT1 and other reported ALMTs in plants indicated that BoALMT1 was most closely clustered with the BnALMT1 from Brassica napus (**Figure 1B**).

## Expression Pattern of *BoALMT1*

We performed real-time reverse transcription (RT)-PCR analysis to measure the expression of BoALMT1 in the roots and shoots, and found that BoALMT1 expression was primarily localized to the roots (**Figure 2A**). Al treatment enhanced its expression in all tissues (**Figure 2A**). Cabbage plants were exposed to a variety of trivalent cations, and the expression of BoALMT1 was not induced by lanthanum (La), cadmium (Cd), zinc (Zn), or copper (Cu), but was severely induced by aluminum (**Figure 2B**). A dose-response experiment and a time-course experiment indicated that increasing the external Al concentration and treatment time did not further increase the BoALMT1 transcript level (**Figures 2C,D**).

## Subcellular Localization of BoALMT1

The subcellular localization of BoALMT1 was determined via localization of the GFP::BoALMT1 protein transiently expressed in onion epidermal cells (**Figure 3**). The GFP::BoALMT1 green fluorescence was only observed at the outer layer of the cell (**Figures 3a,b**), and the cells expressing GFP showed green fluorescence in the whole cell (**Figures 3e,f**). We induced plasmolysis by the addition of 0.8 M mannitol to distinguish localization in the plasma membrane and observed that the fluorescence of GFP::BoALMT1 was exclusively located in the plasma membrane in the plasmolysis cells (**Figures 3c,d**). These localization results were similar to those of some ALMTs identified in other species [TaALMT1 (Yamaguchi et al., 2005), BnALMT1 (Ligaba et al., 2006), ZmALMT1 (Piñeros et al., 2008), ZmALMT2 (Ligaba et al., 2008), and GmALMT1 (Liang et al., 2013)].

#### Pattern of Malate Secretion

To investigate whether the secretion of malate was induced by Al treatment, we characterized malate exudation from cabbage roots. Cabbage roots secreted a low level of malate under normal conditions. After 3 h treatment with 50µM Al, malate exudation was remarkably induced (**Figure 4**).

## Heterologous Expression of *BoALMT1* Reduced Al-Induced H<sup>+</sup> Efflux in *Xenopus* Oocytes

By treated the Arabidopsis mutant with Al stress, Degenhardt et al. (1998) observed that the pH of the root surface increased, while Bose et al. (2010) further confirmed Al stress correlated with lower H<sup>+</sup> influx. So in our study, we used the non-invasive Scanning Ion-selective Electrode Technique (SIET) system to measure H<sup>+</sup> fluxes crossing the surface of Xenopus oocytes with or without the per-injected malate (**Figure 5A**). We noticed that the H<sup>+</sup> flux had no difference in the control oocytes under the absence or the present of Al. Furthermore, compared with the control cells, BoALMT1-expressing oocytes also secreted similar amount of H<sup>+</sup> without pretreated with malate. However, when the malate was fed, the BoALMT1-expressing oocytes secreted more H<sup>+</sup> compared with the control oocytes under the absence

FIGURE 1 | Amino acid sequence (A) and phylogenetic (B) analysis of *Brassica oleracea BoALMT1*. (A) Multiple sequence alignment of cabbage *BoALMT1*, maize *ZmALMT1*, wheat *TaALMT1*, *Arabidopsis AtALMT1*, rape *BnALMT1*, rye *ScALMT1,* and Barley *HvALMT1*. Identical amino acids and similar amino acids were indicated by dark shading and light shading, respectively. Lines depict the 5 predicted transmembrane domains in BoALMT1 as predicted by HMMTOP. (B) Phylogenetic relationship of BoALMT1 and other known Al-activated malate transporters (ALMT). The amino acid sequences were aligned by ClustalW.

or the present of Al condition (**Figure 5A**). To further elucidate BoALMT1 served as a malate efflux transporter, we fed the control and the BoALMT1-expressing oocytes with <sup>14</sup>C-labeled malate and then measured the efflux of radioactively labeled malate (**Figure 5B**). The BoALMT1-expressing oocytes excreted more labeled malate than the control cells. These results indicated that BoALMT1 was a malate efflux transporter and enhanced the H<sup>+</sup> efflux according to malate secretion in Xenopus oocytes.

## Overexpressing *BoALMT1* in *A. thaliana* Enhanced Al Tolerance

Al-activated membrane transporters, which mediate organic acid release from the root apex, are the primary physiological mechanism of plant Al tolerance (Kochian et al., 2004). Plant ALMTs that have been implicated in malate transport and Al tolerance are TaALMT1 in wheat (Sasaki et al., 2004), AtALMT1 in Arabidopsis (Hoekenga et al., 2006), BnAMLT1 and BnALMT2 in oilseed rape (Ligaba et al., 2006), GmALMT1 in soybean (Liang et al., 2013), and MsALMT1 in M. sativa (Chen et al., 2013).

In this study, to investigate whether the overexpression of BoALMT1 enhances malate exudation and Al tolerance, we induced expression of BoALMT1 driven by the CaMV 35S promoter in Arabidopsis plants. Successful introduction of BoALMT1 in two transgenic lines, but not the control line, was confirmed by RT-PCR (**Figure 6A**). Root malate exudation was then measured in the plants expressing BoALMT1 and demonstrating increased Al tolerance (**Figure 6B**). Plants expressing BoALMT1 showed a remarkable increase in root malate exudation rates in the presence of Al, but no difference was observed in the absence of Al. When grown in in agar plates without Al, the transgenic plants expressing BoALMT1 showed root growth similar to that of wild-type (**Figure 6C**). When grown in agar plates with 400µM AlCl3, root elongation of plants expressing BoALMT1 showed less root growth inhibition than that of the plants without expression (**Figures 6C,D**). To further determine the effect on H<sup>+</sup> flow caused by overexpressing BoALMT1 in Arabidopsis, we performed SIET to detect the H<sup>+</sup> flux at the root DEZ with 0 or 50µM Al (pH = 4.5). Under low pH condition, the pattern of H<sup>+</sup> influx exhibited no statistic difference between WT lines and BoALMT1 transgenic lines. However, treated with 50µM Al, the H<sup>+</sup> influx was inhibited in the WT lines, while the H<sup>+</sup> was secreted from the roots BoALMT1 transgenic lines (**Figure 6E**).

## DISCUSSION

Al-activated malate transporters (ALMT) have been reported to be involved in Al tolerance and have been isolated from Arabidopsis, M. sativa, oil seed rape, rye, wheat, and soybean (Sasaki et al., 2004; Hoekenga et al., 2006; Ligaba et al.,

2006; Collins et al., 2008; Liang et al., 2013). Here we reported that Al-induced cabbage BoALMT1 enhanced malate secretion under Al stress in Arabidopsis. BoALMT1 contains five predicted transmembrane domains (**Figure 1A**) and was most closely clustered with BnALMT1 (**Figure 1B**). The expression of BoALMT1 was rapidly induced by aluminum and was primarily localized to the root (**Figure 2**). Some ALMTs are Al-induced but not Al-activated, such as GmALMT1, but AtALMT1 is both induced and activated by Al (Hoekenga et al., 2006; Liang et al., 2013). If BoALMT1 is activated by Al requires further studies.

BoALMT1 was heterologously expressed in oocytes and Arabidopsis to analyze its function (**Figures 5**, **6**). In oocytes, under the absence of Al condition, cells expressing BoALMT1 secreted more H<sup>+</sup> compared with control cells. After Al treatment, H<sup>+</sup> influx diminished in the wild type cells and slightly reduced in the BoALMT1-expressing cells (**Figure 5A**). In Arabidopsis, the BoALMT1 overexpression lines exhibited longer root elongation and more malate exudation under Al treatment compared with WT lines, but there no difference between WT and transgenic lines (**Figures 6B–D**). These results demonstrate that BoALMT1 increase malate secretion to resist Al tolerance in Arabidopsis. This was similar with the reported homologous ALMTs in Arabidopsis and B. napus (Hoekenga et al., 2006; Ligaba et al., 2006). In **Figure 6E**, compared with the low pH condition, BoALMT1 expressing plants secreted H<sup>+</sup> form root tips while the WT plants only diminished the H<sup>+</sup> influx. As described by Ahn and Matsumoto, the activity of H+-ATPase of Al-tolerance wheat lines was higher than that of Al-sensitive wheat under Al treatment (Ahn and Matsumoto, 2006). In faba bean, the activity of PM H+-ATPase was increased and positively associated with citrate exudation under Al stress (Chen et al., 2015). The similar results were also found in our previous study about BoMATE (Wu et al., 2014). Our results might imply that BoALMT1 mediate malate transport instead of directly mediate H<sup>+</sup> flux, and the H<sup>+</sup> efflux might associate with the secretion of malate (**Figures 5**, **6**). However, the causes of these different H<sup>+</sup> flux patterns between Xenopus oocytes and Arabidopsis are unclear. Expressing ALMTs in yeast and bacteria did not show their functions (Ryan et al., 2011). BoALMT1 may behave differently in these two heterologous expressing systems.

were kept in OCM solution. The radioactivity in the bathing solution was measured at the indicated time points; values are expressed as a percentage of the total radioactivity injected. Data was given as means ± *SD* (*n* = 3). Different letters above the columns indicate significant differences (*P* < 0.05) between treatments.

So combining the previous studies by Wu et al. (2014) and Chen et al. (2015) with our findings, we speculated that the secretions of organic acids such as citrate and malate was associated with the activity of PM H+-ATPase to resist Al stress.

A C2H2-type zinc finger transcription factor STOP1 plays a key role in plant Al tolerance. Multiple Al-induced genes such as ALMTs and MATEs are regulated by STOP1 (Liu et al., 2009; Yamaji et al., 2009). To uncover the Al tolerance mechanism in cabbage and determine if STOP1 or a similar regulator participate in this mechanims, further studies are required.

In addition to the external Al detoxification, ALMTs may also have other uncharacterized functions. Recently, Kobayashi et al. demonstrated that ALMT1 responds to multiple signals such as abscisic acid (ABA), indole-3-acetic acid (IAA), low pH, and hydrogen peroxide, but does not respond to methyl jasmonate and salicylic acid (Kobayashi et al., 2013). A few reports found that aluminum-induced malate efflux is negatively regulated by ethylene by inhibition of the expression of TaALMT1 (Tian et al., 2014), this process can be alleviated by the inhibition of ACS activity (Yu et al., 2016). Furthermore, TAA1 regulates local auxin biosynthesis and influences the aluminum-induced inhibition of root growth (Yang et al., 2014). Further work should examine the complex regulation of BoALMT1 during the resistance of multiple stresses and the mechanism by which plants can sense external Al (Kobayashi et al., 2013).

Our results illustrated that the cabbage BoALMT1 localized to the plasma membrane, and the expression of BoALMT1 was specifically induced by Al treatment. Expression of BoALMT1 in Xenopus oocytes and Arabidopsis could enhance Al tolerance. We identified that BoALMT1 can function as an Al-induced gene, and the BoALMT1 protein is involved in H<sup>+</sup> flux in response to Al stress.

#### MATERIALS AND METHODS

#### Plant Cultivars and Growth Conditions

Cabbage (B. oleracea cv. Zhonggan-11) was seeded at 25◦C on moist filter paper in the dark for 2 days. The seedlings were then moved to a complete nutrient solution (Ligaba et al., 2006). After 5 days of culture, the uniform seedlings were moved to a new plastic pot wetted with 0.5 mM CaCl<sup>2</sup> (pH 4.5) solution and pre-incubated for ∼24 h. To measure the spatial expression patterns of BoALMT1 in root tips (0–1 cm), after 6 h of 50µM Al exposure, the roots and shoots were separately collected and subjected to qRT-PCR analysis. To test the specificity of Al-induced BoALMT1 gene expression, we exposed seedlings in a 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) containing 50µM AlCl3, 25µM CdCl2, 10µM LaCl3, 0.5µM CuCl2, or 2.0µM ZnCl<sup>2</sup> for 6 h. To investigate the dose effects of Al on BoALMT1 expression, the seedlings were exposed to a 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) containing 0, 10, 50, or 100µM AlCl<sup>3</sup> for 6 h. To analyze time-course effects of Al toxicity on BoALMT1 expression, the seedlings were exposed to a 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) containing 50µM AlCl<sup>3</sup> for 0, 2, 4, and 6 h.

#### Gene Cloning and Sequencing

To clone BoALMT1, RNA was isolated from cabbage seedlings roots treated with Al. To identify cabbage BoALMT1, we performed a BLAST search with the known AtALMT1 and BnALMT1 sequences on the NCBI website (http://www.ncbi. nlm.nih.gov/). For further amplification, two expressed sequence tags (ESTs) (DK499842 and DY012377) were selected. The two nucleotide sequences were combined to generate a full-length cDNA. The full-length cDNA of BoALMT1 was amplified with sense primer 5′ -ATGGAGAAAGTGAGAGA GATAGTGAG-3′ and anti-sense primer 5′ -TCAAATCTGA AGTATACGAACACCC-3′ , and then constructed into the pMD18-T vector (Takara, Japan). HMMTOP was used for transmembrane protein prediction analysis. Multiple amino acid alignment was conducted by using ClustalX and MEGA4.1 software.

### Characterization of *BoALMT1* Expression via qPCR

BoALMT1 expression was evaluated using quantitative real-time RT-PCR techniques. Primers for qPCR were designed using Primer 3.0. The first-strand cDNA synthesis was performed by using the Primescript reverse transcriptase (Takara, Japan). We performed real-time PCR with a SYBR Premix Ex TaqTM (perfect real time) kit (Takara, Japan) and using the Applied Biosystems 7500 Real-Time PCR System (ABI) using a relative standard curve method with the following primers: BoALMT1, 5′ -AGAGAAGGAAGGAGGGTAGGAG AA-3′ (forward) and 5′ -GAAGACAACAACGACGGTCA-3′ (reverse); Actin (LOC106327159), 5′ -TAACAGGGAGAAGATG ACTCAGATCA-3′ (forward) and 5′ -AAGATCAAGACGAAGG ATAGCATGAG-3′ (reverse). Quantitative PCR was performed with conditions of 95◦C for 3 min, and then 40 cycles of 95◦C for 10 s, 60◦C for 30 s, and 72◦C for 30 s. Expression data were normalized to the expression level of Actin by the 11Ct method.

## Subcellular Localization of BoALMT1

The subcellular localization of BoALMT1 was determined in onion (Allium cepa) epidermal cells. We constructed a vector as 35S:BoALMT1::GFP. The coding region of BoALMT1 was subcloned into the expression vector pCAMBIA1302 using primers: 5′ -CATGCCATGGTAATGGAGAAACTGAGAGAGA TAGTG-3′ (forward) and 5′ -GGACTAGTAATCTGAAGTAT ACGAACACCC-3′ (reverse). We transferred the chimera by particle bombardment. The gold particles (1µm, 1.5 mg) were coated with 5 µg of plasmid DNA in a solution of 2.5 M CaCl<sup>2</sup> and 0.1 M spermidine (Sigma). We bombarded the epidermal onion peels at a helium pressure of 25–30 Mpa (Bio-rad, U.S.), and then incubated the tissue in MS medium at room temperature in the dark for 24 h. Confocal laser scanning microscopy (Leica DMI 6000B-CS, Germany) with a 488 nm excitation wavelength was used to detect the GFP fluorescence. We induced cell plasmolysis by adding 0.8 M mannitol for 3–5 min.

## *BoALMT1* Expression in *Xenopus laevis* Oocytes

We cloned the coding regions (cDNA) of BoALMT1 into the MCS of a pCS107 vector. According to the manufacturer's (Ambion) recommendations, we synthesized the cRNA from 1 µg of AscI-linearized plasmid DNA template. We harvested stage V–VI Xenopus laevis oocytes as described previously (Golding, 1992; Hoekenga et al., 2006). We injected 50 nl RNasefree water containing 15 ng of cRNA encoding BoALMT1 or 50 nl RNase-free water into oocytes using a micro-injector and then incubated the injected oocytes at 18◦C for 2 d in oocyte culture medium, OCM; 1L OCM contains 600 ml L-15 (Sigma L4386), 400 mg BSA (Sigma A4919), 5 ml Penicillin-Streptomycin (Gibco 15140-122), and 400 ml H2O). Before flux measurements of H+, we preloaded malate in the Xenopus oocytes by injection of 50 nl of 0.1 M sodium malate or water. Two hours after preloading, the H<sup>+</sup> fluxes were measured 30µm away from X. laevis oocytes in a solution of 2 mM KCl, 96 mM NaCl, 1 mM MgCl2, 0.3 mM MES, 1.8 mM CaCl<sup>2</sup> with or without 0.1 mM AlCl<sup>3</sup> and with the pH 4.5. Net H<sup>+</sup> fluxes were measured using SIET (Xuyue Science and Technology Co., Ltd., Beijing, China) under steady conditions for 8–10 min to insure that no fluctuation was present. We used the OCM bath solution (pH = 4.5) to perform the <sup>14</sup>C-labeled malate experiment as our previous study (Wu et al., 2014).

#### Heterologous Expression of *BoALMT1* in *Arabidopsis thaliana*

The coding region (cDNA) of BoALMT1 was amplified with primers (5′ -GCTCTAGAATGGAGAAACTGAGAGAGATAGT G-3′ and 5′ -CGCCCCGGGTCAAATCTGAAGTATACGAACA CCC-3′ ) and was cloned into pBI121. We transformed the construct into Arabidopsis using Agrobacterium tumefaciens via the floral dip method (Clough and Bent, 1998). We used RT-PCR to measure the expression level of BoALMT1 in the transgenic plants. Root malate release and Al tolerance were analyzed in two independent homozygous transgenic T3 lines as follows. Arabidopsis seeds, stratified at 4◦C for 3 days, were surfacesterilized and sown onto solid MS medium for 4 days. After germination, we removed uniform seedlings to 0.5 mM CaCl2 agar plates containing 0 or 400µM AlCl<sup>3</sup> (pH = 4.5). The seedlings were kept on agar plates for 2 days, and then the roots were scanned and the primary root length was measured by the Image J program (Liu et al., 2009). For malate exudation assays,

#### REFERENCES


two transgenic Arabidopsis and wild-type lines were surface sterilized and germinated on solid MS medium for 1 week. Next, we transferred the seedlings to a 25 ml solution with 0.5 mM CaCl<sup>2</sup> (pH 4.5) and without Al for 24 h. After this 24 h preincubation step, we then transfered the plants to 25 ml exudation medium (pH 4.5) with or without Al (50µM AlCl3). We collected the sample for malate assay by capillary electrophoresis, as described by Hoekenga et al. (2006). We measured the fluxes of H<sup>+</sup> by using the non-invasive Scanning Ion-selective Electrode Technique (SIET) (Xuyue Science and Technology Co., Ltd., Beijing, China) as described by Bose et al. (2010). The 4- to 5-dayold wild type and BoALMT1 expressing Arabidopsis seedlings were equilibrated in a solution (0.1 mM CaCl2, 0.1 mM KCl, 0.3 mM MES, pH 4.5) with or without 50 mM Al for 5–10 min. H<sup>+</sup> fluxes were measured 200 mm from the root tip for 6–10 min. The H<sup>+</sup> fluxes were calculated by the JCal V3.1 (a free MS Excel spreadsheet, youngerusa.com or ifluxes.com). The H<sup>+</sup> flux assay was replicated independently 4–6 times and the data were averaged.

#### Statistical Analysis

All the statistical analysis was performed by one-way ANOVA and the t-test to determine the significance at the P < 0.05 level.

#### AUTHOR CONTRIBUTIONS

LZ, X-XW, and Y-DG: designed research; X-XW, LZ, JW, CQ, XW, GW, ML, and XL: performed research; JW, X-XW, LZ, and Y-DG: analyzed the data; X-XW, LZ, JW, and Y-DG: wrote the paper.

#### ACKNOWLEDGMENTS

We thank Prof. S Ren and Dr. S Weeda (Virginia State University) for critical reading of the manuscript and Prof. Q Tao (Tsinghua University, Beijing) for providing Xenopus oocytes test system. This work was supported by the grants to Y-DG (2016YFD0101007, BLVT-03) and to XL (tszy20140808).


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhang, Wu, Wang, Qi, Wang, Wang, Li, Li and Guo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Soybean NADP-Malic Enzyme Functions in Malate and Citrate Metabolism and Contributes to Their Efflux under Al Stress

Ying Zhou, Zhenming Yang, Yuezi Xu, Haoran Sun, Zhitao Sun, Bao Lin, Wenjing Sun and Jiangfeng You\*

Laboratory of Soil and Plant Molecular Genetics, College of Plant Science, Jilin University, Changchun, China

Malate accumulation has been suggested to balance Al-induced citrate synthesis and efflux in soybean roots. To test this hypothesis, characteristics of Al-induced accumulation and efflux of citrate and malate were compared between two soybean genotypes combining a functional analysis of GmME1 putatively encode a cytosolic NADP-malic enzyme. Similar amounts of citrate were released, and root elongation was equally inhibited before 8 h of Al treatment of Jiyu 70 and Jiyu 62 cultivars. Jiyu 70 began to secrete more citrate and exhibited higher Al resistance than did Jiyu 62 at 12 h. A sustained increase in internal malate and citrate concentrations was observed in Jiyu 70 at 24 h of Al treatment. However, Jiyu 62 decreased its malate concentration at 12 h and its citrate concentration at 24 h of Al treatment. GmME1 localized to the cytoplast and clustered closely with cytosolic malic enzymes AtME2 and SgME1 and was constitutively expressed in the roots. Al treatment induced higher NADP-malic enzyme activities and GmME1 expression levels in Jiyu 70 than in Jiyu 62 within 24 h. Compared with wild-type hairy roots, over-expressing GmME1 in hairy roots (GmME1- OE) produced higher expression levels of GmME1 but did not change the expression patterns of either of the putative citrate transporter genes GmAACT1 and GmFRDL or the malate transporter gene GmALMT1, with or without Al treatment. GmME1-OE showed a higher internal concentration and external efflux of both citrate and malate at 4 h of Al stress. Lighter hematoxylin staining and lower Al contents in root apices of GmME1-OE hairy roots indicated greater Al resistance. Comprehensive experimental results suggest that sustaining Al-induced citrate efflux depends on the malate pool in soybean root apices. GmME1 encodes a cytosolic malic enzyme that contributes to increased internal malate and citrate concentrations and their external efflux to confer higher Al resistance.

Keywords: aluminum toxicity, anaplerotic reaction, citrate efflux, malic enzyme, tricarboxylic acid cycle

## INTRODUCTION

Aluminum (Al) toxicity seriously restricts crop yield in acidic soils, which cover almost 40% of the arable land worldwide (Ma et al., 2001). Al can significantly inhibit root growth and disrupt root function rapidly (Delhaize and Ryan, 1995). Fortunately, some plant species have developed Al resistance mechanisms to grow in acidic soils. Al-induced organic acid efflux has

#### Edited by:

Wei Fan, Yunnan Agricultural University, China

#### Reviewed by:

Zhichang Chen, Fujian Agriculture and Forestry University, China Wei Hu, Institute of Tropical Bioscience and Biotechnology (CATAS), China

> \*Correspondence: Jiangfeng You youjf@jlu.edu.cn

#### Specialty section:

This article was submitted to Plant Abiotic Stress, a section of the journal Frontiers in Plant Science

Received: 21 July 2017 Accepted: 21 December 2017 Published: 10 January 2018

#### Citation:

Zhou Y, Yang Z, Xu Y, Sun H, Sun Z, Lin B, Sun W and You J (2018) Soybean NADP-Malic Enzyme Functions in Malate and Citrate Metabolism and Contributes to Their Efflux under Al Stress. Front. Plant Sci. 8:2246. doi: 10.3389/fpls.2017.02246

been well established to detoxify Al internally and externally and thus far is the best-documented Al resistance mechanism in higher plants (Ma et al., 2001; Ryan et al., 2001; Kochian et al., 2004; Kochian et al., 2015). Two patterns have been classified according to the rapidity of organic acid release (Ma et al., 2001). In pattern I, some plant species, such as wheat (Ryan et al., 1995) and buckwheat (Ma and Miyasaka, 1998), can rapidly release malate or oxalate, respectively, after Al stress. In pattern II, some species, such as Cassia tora (Ma et al., 1997) and soybean (Yang et al., 2001), secrete citrate after at least 4 h of Al treatment. In both patterns, organic acid anion transporters are crucial for organic acid efflux under Al stress and for Al resistance (Ryan et al., 2011). Over-expression of TaALMT1 (Aluminum-activated malate transporter) in barley (Liu et al., 2009) and wheat (Collins et al., 2008) increased Al resistance by 8-fold and 20-fold, respectively. The Al resistance of Arabidopsis can be increased by 2.5-fold and 3-fold by over-expression of MATE family citrate transporter genes SbMATE (Magalhaes et al., 2007) and ZmMATE1 (Maron et al., 2013), respectively.

The strategies to over-express enzymes involved in organic acid metabolism have also been proven effective in regulating Al resistance in some plant species. Over-expression of citrate synthase genes in different plant species, including alfalfa, Arabidopsis, canola and tobacco, can increase their citrate efflux and Al3<sup>+</sup> resistance in transgenic plants (Koyama et al., 1999; Anoop et al., 2003; Barone et al., 2008; Deng et al., 2009; Han et al., 2009). Malate dehydrogenase genes of different origins were over-expressed in alfalfa (Tesfaye et al., 2001) and tobacco (Wang et al., 2010) and showed enhanced malate efflux and improved Al3<sup>+</sup> resistance. SgME1 encoding NADP-dependent malic enzyme was found to functionally control malate synthesis and secretion and thus Al detoxification (Sun et al., 2014). Recently, over-expression of VuFDH encoding a mitochondrial formate dehydrogenase and VuAAE3 encoding Acyl activating enzyme 3 in tobacco were found to increase Al tolerance by decreasing formate production and oxalate accumulation, respectively (Lou et al., 2016a,b).

Our previous study showed that Al-induced citrate secretion from soybean required almost 4 h of Al exposure (Yang et al., 2000, 2001), which was clearly classified as pattern II (Ma, 2000). Soybean mitochondrial enzymes, including increased citrate synthase and decreased aconitase, were found to contribute to the citrate efflux from roots under Al stress (Xu et al., 2010). Sustained Al-induced citrate efflux from common bean, the close relative of soybean, was reported to rely on the maintenance of high citrate synthase activity and citrate pool (Rangel et al., 2010). Cytosol phosphoenolpyruvate carboxylase (PEPC) and mitochondrial NAD malic enzyme were suggested to contribute to the accumulation and the secretion of citrate in common bean by fueling the tricarboxylic acid (TCA) cycle (Rangel et al., 2010).

Organic acid metabolism-related enzymes was proposed to contribute to detoxifying Al in some plant species (Rangel et al., 2010; Xu et al., 2010; Sun et al., 2014; Lou et al., 2016a,b). In soybean, during the process of citrate efflux from soybean, malate but not citrate significantly decreased with the increase in Al treatment duration (Yang et al., 2001). Thus, malate was hypothesized to maintain balance between the citrate pool and efflux in the soybean roots exposed to Al. However, there is no direct evidence to support this hypothesis until now. Malate is tightly controlled to affect a series of physiological processes because it is at the branching point of many metabolic pathways (Santelia and Lawson, 2016). The transcript level of NADP-malic enzyme was found by microarray assay to increase in soybean root apices under Al stress (You et al., 2011). In this study, in order to elucidate the role of malate pool in the Al-induced citrate efflux from soybean, GmME1, probably encoding NADP-dependent malic enzyme in soybean, was functionally characterized to evaluate its possible implications in organic acid pool and efflux. Al-induced accumulation and efflux of citrate and malate were also compared between two soybean genotypes in relation to GmME1 enzyme activities and gene expression patterns.

### MATERIALS AND METHODS

### Hydroponic Culture and Al Treatment Conditions

Our previous work has shown that soybean cultivar Jiyu 70 and Jiyu 62 exhibited contrast Al resistance capabilities, thus was used as Al tolerant and Al sensitive cultivars respectively in our lab. Seeds of soybean Jiyu 70 and Jiyu 62 cultivars were germinated in darkness for 3 days. Then, seedlings with roots 4–5 cm long were selected for transplant into 0.5 mM CaCl<sup>2</sup> solution. After 24 h of culture, seedlings were exposed to 0.5 mM CaCl<sup>2</sup> solution containing 0 or 30 µM AlCl<sup>3</sup> (pH 4.5). Root length was measured at 0, 8, 12, and 24 h. The relative root elongation (RRE) was calculated to evaluate Al sensitivity. The formula is root elongation with AlCl<sup>3</sup> treatment/root elongation, without AlCl<sup>3</sup> × 100.

The remaining germinated seedlings were grown in 1-L plastic pots filled with nutrient solution with composition, as described by Horst et al. (1992). The solutions were modified to pH 4.5 by HCl and aerated continually. After 14 days of culture, seedlings were pre-cultured in 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) overnight and then transferred to 0.5 mM CaCl<sup>2</sup> solution containing 0 or 30 µM AlCl<sup>3</sup> (pH 4.5). Treatment solutions were refreshed at 2, 4, 8, 12, and 24 h and collected, respectively, for organic acid analysis. Root exudates were concentrated and purified, as described in Ma et al. (1997). Simultaneously, root apices were excised from the parallel Al-treated soybean seedlings at 0, 2, 4, 8, 12, and 24 h (∼0.5 g for each sample). Citrate and malate were extracted from the excised root apices, as described in Yang et al. (2001). Their concentrations were measured by high-performance liquid chromatography (HPLC) (LC 20AT, Shimadzu, Tokyo, Japan) with a Shodex RSpakKC-811 ion-exclusion column (300 × 8 mm, Shimadzu, Tokyo, Japan). NADP-malic enzyme (EC1.1.1.40) was extracted and quantified by an NADP-ME kit (Comin Biotechnology, Suzhou, China). The rate of increase of NADP was monitored at 340 nm.

The 7-day-old seedlings were transferred into 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) overnight and then exposed to 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) including 0 or 30 µM AlCl3. Then, 0- to 1-cm root apices were excised at a treatment duration of 0, 2, 4, 8, 12,

and 24 h. The collected root apices were immediately placed in liquid nitrogen and stored at −80◦C for RNA isolation.

Soybean seedlings were cultivated in a controlled environment with a 14 h/25◦C day and 10 h/22◦C night cycle. Light intensity was controlled as 300 µmol m−<sup>2</sup> s −1 . Relative humidity was kept at 60%.

Jiyu 70 was sown in the field of the agricultural trial station of Jilin University at the end of April 2014. The soil contained 49.4 ± 4.8 g/kg available nitrogen, 11.8 ± 4.1 g/kg available P, 170 ± 6.2 g/kg K, and 21.8 ± 3.7 g/kg organic carbon at pH 6.5. After 18 days, the roots, shoots, leaves, flowers and pods were sampled in the field-grown soybean. The samples were stored at −80◦C for RNA isolation.

#### Gene Transcriptional Expression

RNA was extracted from root apices by Trizol reagent (Invitrogen, Carlsbad, CA, United States). cDNA was obtained by reverse transcribing with M-MLV reverse transcriptase (TaKaRa Bio, Tokyo, Japan). The gene-specific primers were designed according to the CDS of GmME1 (Glyma.06G087800) by Primer 3.0 online<sup>1</sup> and had the following sequences: forward primer 5<sup>0</sup> -AGCATCTGTGGTATTAGCA-3<sup>0</sup> ; reverse primer 5 0 -GGAATAAGAAGGTATGGTCAAC-3<sup>0</sup> . The housekeeping gene β-Tublin (GenBank ID: 100811275) had the following primer: forward primer 5<sup>0</sup> -GGAAGGCTTTCTTGCATTGGTA-3 0 ; reverse primer 5<sup>0</sup> -AGTGGCATCCTGGTACTGC-3<sup>0</sup> . Quantitative real-time PCR (qRT-PCR) was conducted in an Mx3005P machine (PRIMER Biosoft Company, Palo Alto, CA, United States). The 25 µl reaction system included 2 µl of cDNA template (50 ng), 1 µl of a mixture of forward and reverse primers (10 mM), 12.5 µl of 2× SYBR Taq (TaKaRa, Bio Inc.), and 9.5 µl of milli-Q water. The program was as follows: 95◦C for 30 s; 30 cycles of 95◦C for 5 s, 60◦C for 20 s, 95◦C for 60 s, 55◦C for 30 s, and 95◦C for 30 s. Relative expression was computed according to the 2−11C<sup>t</sup> method (Livak and Schmittgen, 2001).

#### Gene Cloning and Sequence Analysis

PCR was performed with cDNA template prepared by root apices treated with Al for 4 h. The primers were designed according to the CDS of GmME1 (Glyma.06G087800), considering the vector pCAMBIA3301 with BamHI with the following primer sequence: GmME1-NF5<sup>0</sup> - CATTCTGGCGGGATCCGCAGCAGCAGCAGCAATGTCGA GCGCTTCGTTGA-3, BamHI; GmME1-NR5<sup>0</sup> -GAGAAAGC TTGGATCCAACGGTAGCTTCGGTAGCCT-3<sup>0</sup> , BamHI. The PCR products were purified using the TransGen Biotech Kit according to the manufacturer's protocol, confirmed by sequencing, and aligned to vectors (pCAMBIA3301) by in-fusion enzyme. Phylogenetic tree construction and sequence comparison were conducted with MEGA 5.1 and Cluster. Other MATE family gene sequences were blasted at the NCBI website as follows: Arabidopsis thaliana (AtNADP-ME1 GeneID:816509, AtNADP-ME2 GeneID:831039, AtNADP-ME3 GeneID:832657, AtNADP-ME4 GeneID:844314), Flaveria bidentis (FbNADP-ME LOCUS: AAW56450), Lycopersicon esculentum (LeME2 LOCUS: AAB58728), Medicago truncatula (MtNADP-ME GeneID:25490143), Nicotiana sylvestris (NtNADP-ME GeneID:104247285), Oryza sativa (OsNADP-ME GeneID:4338007), Stylosanthes guianensis (SgME1 LOCUS AGH32501), and Vigna Umbellata (VuNADP-ME LOCUS CAA56354).

The subcellular localization of GmME1 was determined as follows: The CDS of GmME1 was cloned into pENSG-N-YFP vector with the cauliflower mosaic virus (CaMV) 35S as a promoter. The resulting constructs were fully sequenced to check the sequence accuracy. Plasmid DNA was transformed into Arabidopsis protoplast cells. The imaging of GFP fluorescence was conducted by microscopy (Zeiss 2012 Observer, Göttingen, Germany).

### Agrobacterium-Mediated Over-expression of GmME1 in Soybean Hairy Roots

With CaMV 35S as the promotor, PCR product was cloned into the modified pCamBIA3301 vector. After verification by sequence, the resulting construct was transformed into the K599 strain by electroporation. Soybean transformation in Jiyu 62 cotyledons and hairy root induction were performed according to Subramanian et al. (2005). Hairy roots with scanning luciferase activity greater than 3000 were considered successfully transformed. The hairy roots induced by only K599 were considered wild type (WT). Both transgenic and WT hairy roots were treated in 0.5 mM CaCl<sup>2</sup> solution (pH 4.5) including 0 or 30 µM AlCl<sup>3</sup> in a 5-ml plastic tube. Root exudates were collected at 4 h for citrate and malate efflux measurement. Root apices (0–1 cm) were cut, and three were stained by hematoxylin. The remaining root apices were stored at −80◦C for RNA isolation, internal organic acid concentration measurement, or Al concentration examination. Internal citrate and malate were extracted according to Yang et al. (2000). Citrate and malate concentrations and efflux were measured by enzymatic method (Delhaize et al., 1993). The Al concentration in hairy root apices was determined by 2 M HCl and assayed by an atomic absorption spectrophotometer equipped with a graphite furnace atomizer (Perkin Elmer AAnalyst 700, United States).

## RESULTS

#### Time Course of Relative Root Elongation and Citrate Efflux in Two Soybean Genotypes under Al Stress

Root elongation was nearly equally inhibited during 8 h of Al exposure for both genotypes, whereas recovery began at 12 h for Jiyu 70 and at 24 h for Jiyu 62 (**Figure 1A**). A significant difference in Al-induced citrate exudation (**Figure 1B**) was found between Jiyu 70 and Jiyu 62 at 12 h of Al treatment. The RRE of Jiyu 70 was approximately 1.7 fold greater than that of Jiyu 62 at 12 h (**Figure 1A**). In addition, an approximately 1.8 fold greater Al-induced citrate efflux was found in Jiyu 70 than in Jiyu 62 (**Figure 1B**). The higher Al resistance of Jiyu 70 depends on the

<sup>1</sup>http://primer3.ut.ee/

exposed to 0 or 30 µM Al in 0.5 mM CaCl<sup>2</sup> solution. Root exudates were collected at 2, 4, 8, 12, and 24 h. Relative root elongation was calculated, and organic acid exudates were analyzed as described in Section "Materials and Methods." The values are the means of three independent experiments ± SDs.

recovery from Al-induced root elongation inhibition, in which maintenance of continuous citrate efflux is necessary. Quimbaya, an Al-resistant common bean genotype, was also found to recover root elongation inhibition by sustaining Al-induced citrate efflux (Rangel et al., 2010).

### Time Course of Internal Citrate and Malate Concentration and NADP-Mali Enzymes Activities in Two Soybean Genotypes under Al Stress

Al treatment increased the internal citrate and malate concentrations beginning at 2 h of Al exposure for both genotypes (**Figures 2A,B**). Higher citrate and malate concentrations were always found in Jiyu 70 during throughout the Al treatment duration. A great decrease in the malate concentration occurred at 12 h of Al treatment, followed by a decrease in the citrate

concentration at 24 h in Jiyu 62 (**Figures 2A,B**). The exhaustion of malate (**Figure 2B**) might have a negative effect on the citrate concentration (**Figure 2A**) and efflux (**Figure 1B**) under Al stress. Compared with Jiyu 70, malate exhaustion at Jiyu 62 at 12 h was consistent with the lower citrate efflux at 24 h (**Figure 1B**). This result is consistent with our previous study in another soybean

genotype, Shuzunari, in which the malate concentration but not the citrate concentration was found to decrease under Al stress (Yang et al., 2001). Malate might contribute to balance citrate synthesis and efflux. Al treatment increased the activities of NADP-malic enzymes beginning at 2 h of Al exposure for Jiyu 70. Activities of NADP-malic enzymes in Jiyu 62 increased at 4 h, peaked at 8 h and decreased in the remaining Al exposure duration (**Figure 2C**). The lower NADP-malic enzyme activities of Jiyu 62 were consistent with its lower malate concentration in

### The Transcriptional Expression of GmME1

calibrator. Data are given as the mean ± SD (n = 3).

GmME1 was constitutively expressed throughout the entire plant of Jiyu 70, especially in the roots (**Figure 3A**). Its transcriptional abundance fluctuated in Jiyu 70 and increased 10 fold and 20 fold at 4 and 24 h, respectively, under Al stress. Jiyu 62 also displayed higher expression at 4 and 24 h, albeit with less magnitude (**Figure 3B**).

#### The Bioinformatic Analysis of GmME1

Full-length GmMe1 was isolated from soybean root apices (GenBank: 100778170). GmMe1 encodes a protein with 619 amino acids. As predicted in http: //prosite.expasy.org/scanprosite, GmMe1 contains a malic enzyme signature (Ps00331) at its 331-347 site. Sequence analysis at http://www.cbs.dtu.dk/services/SignalP/ showed that GmMe1 does not contain any predicted organelle sorting signal (data not shown).

The cloned GmME1 showed high similarity to AtME2 and AtME3, with identities of 77 and 76%, respectively. Arabidopsis NADP-malic enzyme isoforms shared high degrees of identity but have very different roles (Wheeler et al., 2008). With minimal structural differences, AtME2 and AtME3 display the forward (malate oxidative decarboxylation to decompose malate) and reverse (pyruvate reductive carboxylate ion to produce malate) reactions. GmME1 is conserved at the suggested critical regulatory regions of fumarate activation and malate inhibition (**Figure 4A**). Multiple isoforms of GmME1 might function redundantly or display different roles. GmNADP-ME homologues had identities between 35 and 96% (**Figure 4A**). Phylogenetic analysis showed that GmME1 exhibited high similarity to SgME1 (**Figure 4B**), which suggests contribution to more malate synthesis and efflux under Al stress (Sun et al., 2014). GmME1 also closely clustered with VuNADP-ME and OsNADP-ME (Chen et al., 2015).

#### Subcellular Localization of GmME1

Transiently expressed GmME1-YFP in Arabidopsis protoplast cells displayed fluorescence signal throughout the cytosol. The expressed YFP alone exhibited non-specific fluorescence within plasma membrane, cytosol and nucleus (**Figure 5**). Thus, GmME1 was suggested to localize at cytosol, which is similar to the cytosol localization of AtME2 and SgME1.

### Agrobacterium Mediated Over-expression of GmME1 in Soybean Hairy Roots

In comparison with the WT, GmME1-OE hairy roots contained higher internal malate (**Figure 6A**) and citrate (**Figure 6B**) concentrations and secreted more malate (**Figure 6C**) and citrate (**Figure 6D**) under either −Al or +Al stress. Compared with that of WT under −Al treatment, nearly 10-fold higher malate concentrations were found in the GmME1-OE hairy roots (**Figure 6A**). Malate concentrations were further increased by Al treatment of both transgenic and WT hairy roots (**Figure 6A**). A slight but significant increase in citrate concentration was found in GmME1-OE hairy roots (**Figure 6B**). Different from WT hairy roots, the citrate concentration in GmME1-OE hairy roots could not be increased by Al treatment (**Figure 6B**). By sensitive enzymatic assay, malate and citrate were detected in the

root apices (**Figures 2B,C**).

root exudates of both WT and GmME1-OE roots under either −Al or +Al treatment (**Figures 6C,D**). Compared with those of WT, 2.5-fold citrate efflux (**Figure 6D**) and 2.0-fold malate efflux (**Figure 6C**) increases were found in the root exudates of GmME1-OE hairy roots. The amount of malate efflux was approximately one-tenth that of citrate (**Figures 6C,D**).

Transcriptional expression analysis showed that GmME1-OE hairy roots had higher transcriptional abundance of GmME1 (**Figure 7A**). 4 h Al treatment didn't cause significant changes of transcription abundance in GmME1 in WT hairy roots, that was different from roots of Jiyu 62 response to Al stress (**Figure 3B**). The difference might result from distinct culture conditions or the different physiological properties between soybean roots and hairy roots. The over-expression of GmME1 had less effect on the transcriptional patterns of either of the putative citrate transporters GmAACT1 (**Figure 7B**) and GmFRDL (**Figure 7C**) or the malate transporter GmALMT1 (**Figure 7D**).

Compared with the WT roots and in agreement with the higher organic acid concentration and exudation, lighter

hematoxylin staining was found in the GmME1-OE roots after 4 h of Al treatment (**Figure 8A**). The Al content in the WT hairy roots was almost 1.5 fold that in the GmME1-OE hairy roots (**Figure 8B**). Thus, GmME1-OE hairy roots successfully acquired higher Al resistance.

#### DISCUSSION

Al-induced citrate exudation has been well documented as the Al–exclusion mechanism in soybean (Yang et al., 2000, 2001; Silva et al., 2001). In the present study, both genotypes showed sensitivity to Al with similarly lower citrate efflux before 8 h of Al treatment, and significantly more citrate efflux was induced from Jiyu 70 at 12 h, which resulted in its higher RRE (**Figures 1A,B**). Thus, the capacity for maintaining higher citrate exudation is critical for Al resistance in soybean. Malate was hypothesized to maintain the balance between the citrate synthesis and release in soybean root exposed to Al because the internal root concentration of citrate increased, whereas malate dropped (Yang et al., 2001). The sharp decrease in malate preceded citrate under Al stress in the Al-sensitive genotype Jiyu 62 (**Figures 2A,B**). The exhaustion of internal malate in Jiyu 62 was more consistent with its lower malic enzyme activities (**Figure 2C**) than the less Al-induced citrate efflux in Jiyu 62 (**Figures 1B**, **2B**). Thus, malate was suggested to play a crucial role in the sustained Al-induced citrate release from soybean roots. It is necessary to elucidate how malate metabolism affects the Al-induced citrate efflux from soybean roots.

Malate is one of the essential carbon storage molecules in plants (Zell et al., 2010) and has long been thought to be involved in regulating and composing the root exudates or affecting stomatal function as an osmolyte (Fernie and Martinoia, 2009). Malic enzyme reversibly converses between malate and pyruvate, depending on the isoform, cellular conditions and available substrates (Sweetman et al., 2009). Mitochondrial NAD-malic enzyme has been suggested to supply pyruvate for the TCA cycle to increase the citrate pool in common bean under Al stress (Rangel et al., 2010). Cytosolic isoforms of NADPdependent malic enzymes have been found to regulate the

(B) concentration analysis. Root exudates were collected for malate (C) and citrate (D) efflux measurement. Fresh root biomass was weighed immediately. Data are given as the mean ± SD (n = 3).

cytosolic pH or stomatal closure by balancing malate synthesis and degradation (Martinoia and Rentsch, 1994; Laporte et al., 2002). Minimal changes in the primary structure of AtNADP-ME isoforms might result in very different kinetic behaviors of each AtNADP-ME isoform (Wheeler et al., 2008). Cytosol-localized NADP-ME2 and NADP-ME3 share 90% sequence identity but show distinct kinetic properties in their forward (malate oxidative decarboxylation) and reverse (pyruvate reductive carboxylation) reactions to regulation (Wheeler et al., 2009). SgME1 has been verified as a malic enzyme functioning in malate synthesis because of its over-expression in yeast, and A. thaliana and common bean hairy roots can significantly increase their malate concentrations (Sun et al., 2014). GmME1 putatively encoding cytosolic NADP-dependent malic enzyme was revealed to increase its transcriptional abundance (You et al., 2011) and thus was chosen to study its contribution to Al-induced citrate efflux.

Fourteen homology genes putatively encoding cytosolic NADP-dependent malic enzyme existed in the soybean genome (**Figure 4B**). There have been no reports on their functional analysis until now. The transcription expression analysis in the present study revealed that GmME1 was expressed throughout the whole soybean plant, at especially higher levels in the roots (**Figure 3A**). Al increased the transcription abundance of GmME1 in soybean root apices of Jiyu 70 during 24 h (**Figure 3B**), which was consistent with its higher malate concentration (**Figure 2B**) and higher NADP-malic enzyme activities in the root apices (**Figure 2C**). GmME1 was localized to the cytosol (**Figure 5**) and displayed high similarity to AtME2 (77%), AtME3 (76%) and SgME1 (86%) (**Figure 4A**), which were conserved in the regions of suggested fumarate activation and malate inhibition (**Figure 4A**) (Wheeler et al., 2009). AtME3 is restricted to trichomes and pollen (Wheeler et al., 2009). According to sequence comparison (**Figures 4A,B**), subcellular localization (**Figure 5**), and spatial expression pattern (**Figure 3A**), GmME1 functions similarly to AtME2 and SgME1 as a malic enzyme contributing to malate metabolism.

Malate efflux from soybean under Al stress has been considered negligible because of several orders of lower magnitude and small variation between soybean genotypes (Yang et al., 2000; Silva et al., 2001). Soybean root malate exudation and concentrations were also reported to coordinately be influenced by pH changes, phosphorus deficiency, and Al toxicity (Liang et al., 2013). GmALMT1 encoding a malate transporter was successfully cloned from soybean root apices, and GmALMT1-mediated root malate efflux was suggested to underlie soybean Al tolerance in soybean (Liang et al., 2013). The different conclusion in the role of Al-induced malate secretion might result from different experiment conditions and/or genotypes. Different from intact root treatment and HPLC detection in experiments from Yang et al. (2000, 2001), excised root apices and a sensitive enzyme assay were used in

the experiments of Liang et al. (2013). In the present study, malate efflux was not detected by HPLC from Al-induced root exudates in 14-day-old seedlings of Jiyu 62 and Jiyu 70 (data not shown). However, malate exudation could be detected in the hairy root experiment by the enzyme assay (**Figures 6C,D**), although it was approximately one-tenth that of citrate efflux.

Over-expressing GmME1 in soybean hairy roots enhanced its own expression (**Figure 7A**) and resulted in a significant increase in malate and citrate concentrations under either +Al or −Al treatment (**Figures 6A,B**). Al-induced malate efflux was found to increase in the GmME1-OE hairy roots, although the transcription level of GmALMT1 encoding malate transporter remained constant (**Figure 7D**). Thus, GmME1 was verified to be a malic enzyme similar to SgME1, responsible for malate synthesis and efflux under Al stress (Sun et al., 2014).

Citrate efflux was approximately 10-fold higher than malate efflux in the hairy roots of transgenic and wild-type plants (**Figures 6C,D**). With unchanged transcription levels of GmFRDL and GmAACT1 putatively encoding citrate transporter (**Figures 7B,C**), more Al-induced citrate secretion was found from GmME1-OE root exudates than that of WT

FIGURE 9 | A Model proposing the role of GmME1 in the process of Al-induced malate and citrate efflux from soybean roots. Cytosolic malic enzyme, GmME1, was supposed to increase malate synthesis in cytosol and thus supply malate to the TCA cycle as an anaplerotic reaction to promote citrate pool over citrate exudation under Al stress. Imported malate and exported citrate through mitochondrial membrane were supposed to occur by GmDTC. The malate and citrate efflux across plasma membrane might have occurred by GmALMT1 and MATE family proteins (GmFRDL and GmAACT1), respectively.

(**Figure 6D**), which might have resulted from their more internal citrate concentration (**Figure 6B**). GmME1, a malic enzyme, increased the citrate synthesis and citrate efflux, which supports our previous hypothesis that malate contributes to balance citrate synthesis and efflux in soybean. Discussion of how GmME1 affects the citrate pool and then efflux follows.

Mitochondrial TCA cycle-related enzymes, especially citrate synthase, have been proposed to prompt the Al-induced citrate efflux from soybean roots (Xu et al., 2010). Anaplerotic enzyme induction has been revealed to counteract the depletion of TCA intermediates. Root isoforms of PEPC and NAD malic enzyme are known to have various anaplerotic functions involved in carbon skeleton supply during N assimilation, maintenance of cytoplasmic pH or osmolarity regulation (Nisi and Zocchi, 2000; Held, 2005). Both PEPC and mitochondrial NAD-ME were proposed to be involved in anaplerotic functions in common bean under Al stress by fueling the TCA cycle (Rangel et al., 2010). This means that the anaplerotic reaction is necessary for some plant species under Al stress. One report on hypertrophied hearts suggested that cytosolic malic enzyme catalyzes pyruvate carboxylation to supply more malate to the mitochondrial TCA cycle, leading to more citrate synthesis (Pound et al., 2009). "Anaplerotic" influx depends on the direct shuttle of malate between the cytosol and mitochondria (Pound et al., 2009). In the present study, GmME1-OE hairy roots contained more malate, which prompted more synthesis and secretion of citrate (**Figures 6B,C**). This process is depicted in **Figure 9**. Cytosol-localized GmME1 might be involved in an alternate anaplerotic pathway to supply the TCA under Al stress by promoting more malate synthesis (**Figure 9**). Mitochondrial carrier proteins function to export or import metabolite to maintain the pools of TCA cycle intermediates

(Haferkamp and Schmitz-Esser, 2012; Etienne et al., 2013). Dicarboxylate/tricarboxylate carrier (DTC) was suggested to transport dicarboxylates, such as oxaloacetate and malate, and tricarboxylates, including citrate, isocitrate, cis-aconitate, and trans-aconitate across mitochondrial membrane by a counter-exchange mechanism (Deng et al., 2009). CjDTC was suggested to involve in organic acid excretion in Citrus junos because of its higher expression under Al stress (Deng et al., 2009). Consistently, the expression of DTC was increased in soybean root apices under Al stress (You et al., 2011). Citrate carrier inhibitor treatment decreased Al-induced citrate efflux from soybean, indicating its important role in the process of citrate release (Xu et al., 2010). In this paper, the import of malate and export of citrate in mitochondria might depend on mitochondrial carrier proteins such as DTC protein (**Figure 9**). The malate and citrate efflux across the plasma membrane was supposed to depend on GmALMT1 and members of the MATE family (GmAACT and GmFRDL) (**Figure 9**).

Consistent with the increased efflux and concentration of both malate and citrate, the GmME1-OE soybean hairy roots have light hematoxylin staining (**Figure 8A**) and lower Al contents (**Figure 8B**) in root apices, demonstrating better Al exclusion capacity and higher Al resistance.

#### CONCLUSION

GmME1 was revealed to encode a cytosolic malic enzyme, which increased malate and citrate synthesis and Al-induced malate and citrate efflux. Moreover, new evidence was added that GmME1 can function in anaplerotic pathways to supply the TCA cycle

#### REFERENCES


to prompt more citrate synthesis then efflux under Al stress (**Figure 9**).

#### AUTHOR CONTRIBUTIONS

YZ performed most of the experiments. ZY the supervisor of YZ, helped to design the experiments and performed organic acid measurement by HPLC. YX performed the subcellular localization experiments. BL and WS helped in plant culture and hairy root induction experiments. HS and ZS helped in qRT-PCR experiments. JY designed the entire experiment, performed organic acid examination by enzymatic assay, and wrote the manuscript. The final manuscript has been read and approved by all authors.

### FUNDING

This research was financially supported by the National Natural Science Foundation of China (No. 31372124) for JY and Jilin Natural Science Foundation of China (No. 20130101084JC) for JY. This research was also funded by the National Undergraduates Innovating Experimentation Project (No. 2016A82362) for BL and WS.

#### ACKNOWLEDGMENTS

We would like to thank Hongyan Sheng at Washington State University for language revision and proofreading.


through GmALMT1 to improve soybean adaptation to acid soils. Plant Physiol. 161, 1347–1361. doi: 10.1104/pp.112.208934


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Zhou, Yang, Xu, Sun, Sun, Lin, Sun and You. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.