ARF family identification in Tamarix chinensis reveals the salt responsive expression of TcARF6 targeted by miR167

View article
Plant Biology

Introduction

Tamarisk plants exist as shrubs or small trees that are naturally distributed in saline soils of Eurasia, the Mediterranean basin and northern and southwestern Africa (Marlin et al., 2017; Sanz Elorza et al., 2010). Tamarix chinensis Lour is one of the most highly salt-tolerant tree species that is well suited for investigating salt-stress-response genes (Liu et al., 2014). The high salt tolerance of tamarisk trees was shown by tolerance limit studies and on 340 mM (2% (m/v)) NaCl treatment for 1 week caused no physiological damage to T. chinensis (Wang et al., 2018a). This result indicated that the salt tolerance limit of T. chinensis could be greater than the 2.1% carbonate limit reported by Song et al. (2006), the 300 mM NaCl limit of Tamarix aphylla and the 250 mM NaCl limit of Tamarix ramosissima (Natale et al., 2010). Transcription factors (TFs) coordinate gene expression by activating or inhibiting transcription in response to various abiotic stress signals in plants. In our previous study (Wang et al., 2017b), many TFs of T. chinensis were predicted to be differentially expressed under NaCl stress. The functions of some of these TFs, such as basic leucine zipper (bZIP), NAC-domain-containing (NACs), basic helix-loop-helix (bHLH) and WRKY TFs of Tamarix spp., were verified to increase salt tolerance when heterologously overexpressed (Zheng et al., 2013; Gao et al., 2013; Ji et al., 2013; Yang et al., 2014; Zhou et al., 2014; Wang et al., 2015, 2017a). Other significantly differentially expressed TFs, such as ARFs, which are highly important, may be related to salt tolerance and should be further studied.

ARFs play a key role in regulating the expression of auxin response genes by binding to auxin response elements (AuxREs) within their promoters (Tiwari, 2003). ARFs in Arabidopsis (AtARFs) play important roles in vascular tissue development, phototropism of the hypocotyl (Hardtke & Berleth, 2014; Pekker, Alvarez & Eshed, 2005), root cap formation (Su et al., 2016; Wang et al., 2005) and other plant developmental processes via auxin signaling (Li et al., 2016). The stress response of ARFs was noticed early and Hannah, Heyer & Hincha (2005) proved that the expression of several AtARFs is altered by cold stress. To date, transcriptional responses to various abiotic stresses have been identified in various plant species by genome-wide expression profile analyses of ARFs, including those in rice (Oryza sativa) (drought, cold, salt stress), banana (Musa nana) (cold, salt and osmotic stress), tomato (Lycopersicon esculentum) (abscisic acid (ABA), indoleacetic acid (IAA) and gibberellin (GA3) treatments) and Salvia miltiorrhiza (methyl jasmonate treatment) (Jain & Khurana, 2009; Audran-Delalande et al., 2012; Hu et al., 2015; Xu et al., 2016a). In addition, it seems that auxin regulates stress-induced responses involved in morphogenesis via small RNA (sRNA)-ARF pathways, in which microRNA (miRNA)-ARFs and trans-acting small interfering RNA (tasiRNA)-ARFs are involved. Several studies have suggested that ARF members of group I, such as ARF5, ARF6 and ARF8, targeted by sRNA or miRNA are associated with abiotic stress responses. For example, by fine tuning the expression of auxin-responsive genes, the tasiRNA-ARF pathway has been reported to moderate floral morphogenesis in Arabidopsis in response to drought stress (Matsui et al., 2014). In addition, the miR167a-ARF (ARF6 and -8) pathway mediates plant growth and the response to Pi-starvation stress via homeostasis of reactive oxygen species (ROS) and Pi acquisition in tobacco (Chen et al., 2018). Recently, by increasing ABA and proline contents as well as superoxide dismutase activity, the ARF5 gene from sweet potato (Ipomoea batatas) was shown to increase tolerance to salt and drought in transgenic Arabidopsis (Kang et al., 2018). Arabidopsis double mutants of miR160 and miR165/166 indicated that ARFs and HD-ZIP IIIs may play opposite roles in the regulation of leaf development and drought tolerance (Yang et al., 2019). These reports suggest that the expression of ARFs is induced or repressed by various abiotic stresses and that sRNA-ARF modules play important roles in the crosstalk between auxin and abiotic stress signaling (Sharma et al., 2015).

While the high salt tolerance of T. chinensis is noticeable, its molecular mechanism remains elusive. Studies of ARF profiles under salt stress in a variety of crop species and miRNA-ARF pathway associations with salt tolerance suggested that differentially expressed ARFs might also play important roles in the salt tolerance mechanism of T. chinensis. There have been no studies on ARFs in Tamarix spp.; thus, we aimed to screen the salt-stress-responsive ARF genes in T. chinensis by transcriptome-wide identification of the members of the ARF family, by expression profile construction and by miRNA target prediction. Furthermore, we preliminarily investigated the correlations of miRNAs and ARF target expression to determine the posttranscriptional regulatory mechanisms involved in the response to salt stress. Our work could serve as a basis for studying the roles of the miR164-AtARF pathway in the salt-stress response and would be helpful for further understanding the salt tolerance mechanism of T. chinensis.

Materials and Methods

Plant materials

The acquisition of cuttings and salt treatment of one-month-old T. chinensis ramets followed the methods in our previous study (Wang et al., 2018a). The roots of the ramets were immersed in 340 mM NaCl solution for 0.5 h, 1 h or 4 h and the roots in the control (CK) group were immersed in sterile water for 4 h. The roots, stems and leaves of the salt-treated ramets and CK ramets were harvested (12 tissues with 3 biological replicates for a total of 36 samples). The plant material needed for each sample was collected from 3 to 5 ramets (36–60 plants in total). All samples were quickly frozen in liquid nitrogen and then stored at −80 °C until RNA and DNA extractions were performed.

Prediction and identification of ARFs

To explore the role of the ARF genes in T. chinensis, the members of the TcARF gene family were predicted from the RNA sequencing (RNA-seq) data (Wang et al., 2017b). Using the hidden Markov model (HMM) prediction referring to the Pfam profile of the ARF family (PF06507, http://Pfam.sanger.ac.uk/), we predicted TcARF transcripts from the assembled transcripts of the RNA-seq data (Wang et al., 2017b). The proteins of the predicted TcARFs were verified by the Plant Transcription Factor Database (http://planttfdb.cbi.pku.edu.cn/) via BLASTP. We used an E-value cutoff of 1E−2 for the HMMER search and 1E−4 for the BLASTP identification.

Phylogenetic analysis

The conserved sequences (the B3 domain and the AUX_resp domain) of the TcARFs and AtARFs from Arabidopsis (Table S3) were used to construct a phylogenetic tree on the basis of the maximum likelihood method. The sequences of the AtARFs were downloaded from The Arabidopsis Information Resource database (https://www.arabidopsis.org/). Multiple sequence alignments of the conserved domains (B3 and AUX_resp domains) within the ARF proteins were carried out via ClustalX. A phylogenetic analysis was subsequently performed via the neighbor-joining method by MEGA 7, with 1,000 bootstrap replicates.

Digital gene expression profile analysis

A DGE profile of the tch-miR160 and tch-miR164 TcARFs in the T. chinensis roots was constructed on the basis of the normalized transcripts per kilobase million (TPM). The TPM was calculated according to the RNA-seq data (Wang et al., 2017b) and the small RNA sequencing (sRNA-seq) data (Wang et al., 2018a).

Sequence cloning

Genomic DNA was extracted from newly expanded leaves via a DNeasy Plant Mini Kit (QIAGEN, Hilden, Germany). Total RNA was subsequently extracted via an RNAprep Pure Plant Kit (Tiangen, Beijing, China) (Liu, Sun & Xu, 2018). Full cDNA sequences of TcARF6 were identified according to the manual provided with a SMARTer Rapid Amplification of cDNA Ends (RACE) 5′/3′ Kit (Clontech, Mountain View, CA, USA), with slight modifications (Wang et al., 2018a; 2018b; Cheng et al., 2017). The complete coding DNA sequences (CDSs) of TcARF1-3, 5 and 9–12 were identified via sequence amplification with cDNA and introns within TcARF6 were identified by sequence amplification with DNA. Partial sequences of TcARF12-14 (those lacking a complete CDS) were included in subsequent analyses. All sequences and primers used are listed in Additional Table S1.

Expression analysis

The total RNA of the T. chinensis roots, stems and leaves was reverse transcribed via PrimeScript RT Master Mix (Takara, Dalian, China) to synthesize first-strand cDNAs. Primers (Tm of 59–60 °C) for TcARF6 and the reference gene were designed by Oligo 7 software to amplify the gene-specific PCR products whose length ranged from 70 to 150 bp. Three technical replicates were included; each replicate consisted of a 20 μl mixture that consisted of 10 μl of SYBR Premix Ex Taq (Takara, Dalian, China), three μl of 50× ROX Reference Dye II, 0.2 μl (10 μM) of each primer, two μl of cDNA and 4.6 μl of H2O. Quantitative real-time PCR (qRT-PCR) was performed by a Viia 7 Real-Time PCR System according to the manufacturer’s protocol (Wang et al., 2018b). The thermocycling conditions were the same as those reported by Liu, Sun & Xu (2018) and Liu et al. (2018). For quantification of miRNA, 500 ng of the total RNA from the roots was reverse transcribed via PrimeScript RT Master Mix in conjunction with a manually designed stem-loop primer. The qRT-PCR and analysis methods were the same as those mentioned above. The transcription initiation factor (TIF) gene was used as a reference gene. All primers used are listed in Additional Table S2 and the relative expression level was calculated with the 2−∆∆CT method.

Results

Identification of ARF family members in T. chinensis

Twelve members were identified as belonging to the TcARF family. The TcARF proteins, which ranged from 624 (TcARF9) to 966 (TcARF5) amino acids (AAs) in length, had a predicted molecular mass of 73.3 (TcARF11) to 105.9 (TcARF5) kDa. All 12 TcARFs contained conserved B3 and AUX_resp domains. Four TcARFs containing microRNA response elements (MREs) were determined to be potential targets of tch-miR167 or tch-miR160 (Table 1).

Table 1:
Identification of ARFs in Tamarix chinensis.
Gene Source CDS (bp) MRE MRE location (bp)
TcARF1 PCR 2,031
TcARF2 PCR 2,553
TcARF3 PCR 2,130
TcARF4 Prediction 1,698*
TcARF5 PCR 2,901
TcARF6 RACE 2,631 miR167 2,367–2,387
TcARF7 Prediction 1,584*
TcARF8 Prediction 2,340* miR167 CDS
TcARF9 PCR 1,872
TcARF10 PCR 2,091 miR160 1,344–1,364
TcARF11 PCR 2,007
TcARF12 PCR 2,094 miR160 1,320–1,340
DOI: 10.7717/peerj.8829/table-1

Note:

Means partial CDS.

To identify the subfamilies, a phylogenetic analysis of the TcARF family members was conducted (Fig. 1). The topological structure and corresponding ARF members obtained were identical to those of maize (Zea mays), sorghum (Sorghum bicolor) and rice (Wang et al., 2010a, 2007; Xing et al., 2011). All of the ARF genes clustered into five groups (Fig. 1). Groups I, II, III, IV and V correspond to the ARF5/6-like group (AtARF5-8, 19), ARF3/4 group, ARF16/17 group (AtARF10, 16 and 17), ARF1/2 group and ARF9/11-like group of Arabidopsis, respectively. The deduced protein length and molecular weight of the TcARF proteins in each group varied little (Table S3). The potential TcARFs targeted by miRNAs belong to groups I and III, which was in agreement with the miRNA-ARF modules in Arabidopsis. The ARFs targeted by miR167 belonged to group I and encoded relatively large-molecular-weight proteins. All miR160-targeted ARFs clustered into group III and the MRE sequences were highly conserved in this group.

Phylogenetic analysis of Tamarix chinensis and Arabidopsis ARF proteins.

Figure 1: Phylogenetic analysis of Tamarix chinensis and Arabidopsis ARF proteins.

The numbers on each branch indicate bootstrap values and the yellow background indicates miRNA-targeted ARFs.

Expression profiles and MREs of ARFs targeted by miRNAs

DGE profiles were constructed to analyze the expression of ARFs in groups I and III (Fig. 2A). In group III (miR160-ARF module), the abundance of miR160 resulted in an expression level that was greater for this miRNA (TPM > 200) than for miR167 and the potential targets the former (TcARF10/TcARF12 of) exhibited a lower abundance (TPM < 20). In group I (the miR167-ARF module), the DGE profile suggests that miR167 is upregulated and that the expression of TcARF6 is downregulated (~4-fold). The negative correlation indicated that miR167-ARF6 could respond to salt stress, which was further tested with qRT-PCR.

ARFs targeted by miRNAs in Tamarix chinensis.

Figure 2: ARFs targeted by miRNAs in Tamarix chinensis.

(A) Digital gene expression profiles of ARFs and miR160/167 in T. chinensis roots. A heat map was generated on the basis of the normalized Log2 (TPM), which is represented by the red-blue gradation. The columns represent the 0 h and 0.5 h salt treatments (340 mM NaCl). The rows represent ARF genes targeted by miRNAs except TcARF5 and TcARF7. The numbers in the heat map are TPM values. (B) A schematic diagram of MRE locations. The lines show the relative positions of the domains and the numbers indicate the positions of the AA residues. The B3, Aux_resp and CTD domains; the Gln-rich region; and the MRE are represented by different geometric shapes.

The conserved domains were identified via the NCBI Conserved Domain Search Service (Fig. 2B). For the ARFs in groups I and III, the glutamine-rich (Gln-rich) region and the MREs are specific motifs. All members except TcARF7 in group I contain a Gln-rich region. The presence of sequences similar to those of the Gln-rich regions of AtARFs (e.g., AtARF6, AtARF8) indicated that these ARFs are inhibitory TFs (Wu, Tian & Reed, 2006; Su et al., 2016; Nagpal et al., 2005; Tabata et al., 2010). The miR160 MREs of TcARF10/12 were located within the C-terminal region of the Aux_resp domain (~80 AA apart) and encoded the conserved motif GXQGAR. The miR167 MREs were also located within the C-terminal region. The MRE of TcARF6 overlapped with the AUX/IAA domain, which was absent in TcARF8. The MRE of TcARF6 and 8 encoded the conserved motif GSGWQL.

Cloning of TcARF6 and expression analysis of tch-miR167/TcARF6

Expression profile analysis revealed that TcARF6 was a candidate responsive gene. Gene cloning and expression analysis of tch-miR167/TcARF6 were subsequently performed for validation. The 3,708 bp full cDNA of TcARF6 was obtained by RACE and included an 854 bp 5′-untranslated region (UTR) and a 226 bp 3′-UTR with a polyadenylation (poly-A) signal (50 bp before the poly-A tail). When the CDS was compared with the genome sequence, it was determined that the TcARF6 gene structure had no introns (Fig. 3).

Gene and predicted protein structures of TcARF6.

Figure 3: Gene and predicted protein structures of TcARF6.

In the TcARF6 gene structure with no introns, the green lines represent the 5′-UTR and 3′UTR. “ATAAAA” is the poly-A signal. The yellow line represents the 2,631 bp open reading frame. In the TcARF6 protein structure, rectangles of different colors represent different domains. The rhomboids represent the Gln-rich region and the purple line represents the nuclear localization signal. In the B3 domain, α represents an α-helix, and β represents a β-sheet. The tertiary structure represents the spatial conformation of the B3 conserved motif. “TGTCTC” is the promoter recognition site.

The TcARF6 protein had typical ARF domains (B3, AUX_resp) in the DNA-binding region (DBD) and an AUX/IAA domain in the carboxy-terminal dimerization (CTD) region (Fig. 3). Specifically, the GL-rich motif (462–536 AA, 33% glutamine) in the middle region (MR) was similar to the activation domain (AD) of members in Arabidopsis (AtTcARF5, 6, 7, 8 and 19). The tertiary protein structure of the B3 domain consisting of seven β-sheets and two α-helices was highly conserved in the B3 protein family. These results suggested that TcARF6 may recognize AuxREs by binding to the TGTCTC motif. A predicted nuclear localization signal (NLS) was located at the DBD and MR boundary. The CTD consisted of 148 AA residues that encompassed domains III and IV. Domain IV of TcARF6 is more conserved than domain III; the MRE was located in domain IV (Fig. 3). Only a short deletion from 614 to 684 AA was found from the protein sequence collinearity with AtARF6. A similar protein sequence collinearity indicated that TcARF6 could be an orthologous gene of AtARF6.

The 3217–3235 nt of TcARF6 mRNA (the MRE) was perfectly complementary to the 3–20 nt of tch-miR156, with the exception of one mismatch (the 14th nucleotide of tch-miR156) (Fig. 4A). This indicated that the abundance of TcARF6 RNA would decrease by mRNA cleavage induced by tch-miR167. Analysis of the expression of TcARF6 and miR167 in different tissues under 0.5 h, 1 h and 4 h salt stress treatments (Fig. 4B) revealed that TcARF6 exhibited tissue-specific expression and that its expression was correlated with that of miR167. In the stems and leaves, the expression of TcARF6 remained relatively stable during the 0–4 h salt treatment (a slight change occurred between 0.65 and 1.11). In the roots, the expression of TcARF6 was slightly downregulated during the 0.5–1 h salt treatment and sharply downregulated (~25 times) until 4 h (Fig. 4C). Interestingly, the fold change of the upregulated miR167 expression corresponded to that of the TcARF6 expression. The degree of negative correlation of expression between miR167 and TcARF6 seemed to increase with prolonged salt stress.

Salt-stress responses of TcARF6/tch-miR167 in Tamarix chinensis.

Figure 4: Salt-stress responses of TcARF6/tch-miR167 in Tamarix chinensis.

(A) Sequence complementary of the miRNAs response elements (MRE) and tch-miR167. The numbers indicate the length of tch-miR167 and location of MRE within TcARF6. (B) Relative expression of tch-miR167/TcARF6 in roots of T. chinensis. The Y-axis represents the log2-transformed relative expression. (C) Relative expression of TcARF6 in the aboveground parts of T. chinensis; R, S and L represent the roots, stems and leaves, respectively. W (WR, WS, WL) represents tissues under water; 0.5, 1 and 4 represent salt treatments of 0.5 h, 1 h and 4 h, respectively. **Indicates the significance (0.05) by T-test compared with the control group.

Discussion

ARF family members in T. chinensis

The number of ARFs varies among plant species and can be roughly divided into two categories: plant species with more than 20 ARFs, such as poplar (35), tobacco (42) and soybean (56) and those with less than 15 ARFs, such as Carica papaya (11), peanut (10) and cannabis (13) (Li et al., 2016; Kalluri et al., 2007; Wang et al., 2007; Liu et al., 2015; Xu et al., 2016b; Zhang et al., 2017). In this study, 12 TcARFs were identified that belong to the category with fewer members. TcARFs are grouped into groups I–V, as are most plant ARF families. Eighty-three percent of TcARFs and AtARFs (10/12) are homologous. The high degree linear homology between the TcARFs and AtTcARFs suggested that the ARF family members were linearly correlated in the whole genome after separation from Arabidopsis during Tamarix evolution. In species with more than 20 TcARFs, such as Populus and Salix, the whole genome had duplicated, thus expanding the number to twice that in Arabidopsis (Kalluri et al., 2007).

Introns are the noncoding portions of genes and play a large role in alternative splicing of pre-mRNAs. ARFs compose a family of intron-rich genes (Finet et al., 2010; Kalluri et al., 2007). Apart from Arabidopsis members in group III (the ARF16/17-like group) having relatively few introns (1–3), members in the other four groups have a relatively large number of introns (even more than nine introns). Introns from rice and poplar ARFs have been described and the number and structure of introns in the TcARFs of the same group are similar (Kalluri et al., 2007; Wang et al., 2007). Forty-nine percent of Arabidopsis intron-containing genes are regulated by alternative splicing at the posttranscriptional level in the salt stress response, but most differentially alternatively spliced genes (DAS) are not differentially regulated at the transcriptional level (Ding et al., 2014). Our cloning of the genomic sequences of other TcARFs (several ARFs are still being tested, unpublished data) indicated that most TcARFs should be intron rich and we suspected that these ARFs could be DASs with stable expression at the transcription level.

TcARF6 belongs to group I, whose members are associated with an abundance of introns. The intronless structure of the TcARF6 gene is rare in the TcARF family. Only the smallest SmARF12 (345 AAs, 37.78 kDa) of the 25 ARFs from S. miltiorrhiza has been reported to be intronless. SmARF12 belongs to group III and is predicted to localize to the chloroplast rather than the nucleus (Xu et al., 2016a). This may be due to the loss of introns, including NLSs and flanking introns, within the members of group III. However, TcARF6, which has a long CDS region and a typical NLS, is considered a macromolecule. Though the long gene length of TcARF6 differed from that of most intronless genes encoding small proteins, its characteristic rapid expression in response to salt stress corresponded to the response to stress of the intronless genes (Zhang et al., 2014; Yan et al., 2014; Jeffares, Penkett & Bahler, 2008). Genes that respond rapidly to stress have significantly low intron densities throughout the plant genome, suggesting that introns can delay regulatory responses (Jeffares, Penkett & Bahler, 2008; Zhang et al., 2014; Yan et al., 2014). We speculated that the intronless TcARF6 gene was selected from an intron-containing ancestral gene or paralogs against the requirement for rapid adjustment for survival against environmental challenges during evolution.

Regulation of TcARF6 under salt stress

Many global expression profiles of plants under abiotic stress suggest that the salt- or drought-induced regulatory pattern of ARFs is complex (Die, Gil & Millan, 2018). Different expression patterns can be presented by different members in the same tissue or by the same members in different tissues. Under salt and drought stress, the expression of most of the 23 rice ARFs was downregulated, while that of several ARFs was upregulated (Li et al., 2016; Luo, Zhou & Zhang, 2018). Under severe salt stress, the expression of all 25 ARFs of sorghum was significantly downregulated in the roots, while that of many members was upregulated in the leaves (Wang et al., 2010b). The number of ARFs expressed in response to stress greatly differs among species. The expression of only 4 of 28 ARFs in chickpea (Cicer arietinum) was significantly induced in the roots. The TcARF expression profile in which a low percentage of ARFs is regulated, such as the situation in chickpea, suggested that only TcARF6 responds to salt stress. qRT-PCR further revealed that the expression of TcARF6 was significantly downregulated in the roots but remained stable in the aboveground parts. Though the expression was inhibited by the action of several homologous genes of TcARF6, such as salt-inhibited AtARF6-8, SlARF6/7 of tomato and MnARF6 of banana (Kumar, Tyagi & Sharma, 2011; Audran-Delalande et al., 2012; Hu et al., 2015; Su et al., 2016), root tissue-specific expression patterns were rare. These results indicate that TcARF6 may play a unique role in the root response to salt stress.

Recent studies suggest that miR167-ARFs can link stress signals to auxin signal transduction (Xi et al., 2018). According to the expression profile analysis, though the abundance of tch-miR167 was not high, its upregulation was negatively correlated with miR167-ARF6. Further, qRT-PCR proved that the fold change of upregulated expression increased with salt stress duration. The increasing correlation between the temporal expression profiles of miR167-ARF6 was in agreement with canonical miRNA-ARF interactions, such as those involving the abiotic stress regulatory module MIR167a-ARF6(8) in Arabidopsis and tobacco (Chen et al., 2018). The salt tolerance role of ARF5 (homologous to ARF6) in sweet potato has been functionally verified in transgenic Arabidopsis and is associated with ABA signaling (Kang et al., 2018). According to the results of our comparative analysis and predicted transcriptional inhibition characteristics, we suspected that AtARF6 increases salt tolerance to some extent via posttranscriptional regulation of miR167. In addition, AUX/IAA family members generally present opposite expression patterns under the same abiotic stress conditions. These findings suggest that salt stress may indirectly affect the ARF-AUX/IAA interaction, resulting in more complex regulation of ARFs. Additional studies on the roles of TcARF6 under stress should focus on the verification of molecular interactions between miR167 and downstream AUX/IAA elements.

Conclusions

In the current study, a transcriptome-wide analysis of the TcARF gene family was performed. In total, 12 TcARF genes were identified via analyses of phylogenetic relationships, conserved domains/motifs and expression profiles in plants under salt stress. Only the expression of TcARF6 with a miR167 MRE was significantly downregulated under stress. Further cloning and pattern analysis indicated that TcARF6, whose expression is opposite that of tch-miR167, plays a role in the posttranscriptional regulation of the T. chinensis salt-stress response. This study provides comprehensive information on the TcARF gene family and will aid in investigations of the salt-stress response of TcARFs.

Supplemental Information

All sequences and primers of ARF family in this study.

Primers of ARFs cloning; Primers of qRT-PCR; Protein feature of ARFs in phylogenetic analysis.

DOI: 10.7717/peerj.8829/supp-1
21 Citations   Views   Downloads