Identification and functional characterization of ABCC transporters for Cd tolerance and accumulation in Sedum alfredii Hance

Cd is one of the potential toxic elements (PTEs) exerting great threats on the environment and living organisms and arising extensive attentions worldwide. Sedum alfredii Hance, a Cd hyperaccumulator, is of great importance in studying the mechanisms of Cd hyperaccumulation and has potentials for phytoremediation. ATP-binding cassette sub-family C (ABCC) belongs to the ABC transporter family, which is deemed to closely associate with multiple physiological processes including cellular homeostasis, metal detoxification, and transport of metabolites. In the present work, ten ABCC proteins were identified in S. alfredii Hance, exhibiting uniform domain structure and divergently clustering with those from Arabidopsis. Tissue-specific expression analysis indicated that some SaABCC genes had significantly higher expression in roots (Sa23221 and Sa88F144), stems (Sa13F200 and Sa14F98) and leaves (Sa13F200). Co-expression network analysis using these five SaABCC genes as hub genes produced two clades harboring different edge genes. Transcriptional expression profiles responsive to Cd illustrated a dramatic elevation of Sa14F190 and Sa18F186 genes. Heterologous expression in a Cd-sensitive yeast cell line, we confirmed the functions of Sa14F190 gene encoding ABCC in Cd accumulation. Our study performed a comprehensive analysis of ABCCs in S. alfredii Hance, firstly mapped their tissue-specific expression patterns responsive to Cd stress, and characterized the roles of Sa14F190 genes in Cd accumulation.

. The amino-acid sequences of Arabidopsis AtABCCs were chosen as the queries and further applied in a local BLAST using the Blast + software supplied by National Center for Biotechnology Information (NCBI) 65 . The expressed sequence tag (EST) hits with E-value less than 10 −6 were considered as significant ones and their sequences were screened for representative domain signature (Walker A, Walker B, and ABCC-MRP like ATPase domains) as defined in the NCBI conserved domain database (CDD). The location of NBD and TMD was scanned against a large collection of protein families, each represented by multiple sequence alignments and hidden Markov models (PFAM) database version 28.0 using biosequence analysis using profile hidden Markov models (HMMER) v3. 1 66 , and only those with confirmed structure features of ABCCs were applied to further analysis. The chosen EST hits were further filtered by removing redundant sequences and annotated with functions using a simple modular architecture research tool (SMART, http://smart .embl-heide lberg .de) 67 and PFAM (http://pfam.xfam.org/) 68 . Phylogenetic analysis, protein structure and conserved motif analysis of SaABCCs. To verify the evolutionary location of putative SaABCC members, a phylogenetic tree was constructed using MEGA 7.0 69 via the neighbor-joining (NJ) method using ABC proteins of Arabidopsis as reference 45 . Phylogenetic analysis was carried out to characterize the evolutionary relationships among AtABCCs and SaABCCs with MEGA 7.0.
For the analysis of protein structure, all the identified SaABCCs were firstly searched against the PFAM database (version 28.0) 68 and further validated by CDD analysis (http://www.ncbi.nlm.nih.gov/Struc ture/cdd/wrpsb .cgi). The domain composition of each SaABCC protein was subsequently visualized by the software package of Illustrator of Biological Sequences (IBS) 70 . Analysis of conserved motif distributions was performed using the motif analysis online program, Multiple Expectation Maximization for Motif Elicitation (MEME) (http:// meme-suite .org/tools /meme) 71 . Besides default parameters, other parameters were set as follows: any number of repetitions, maximum number of motifs = 20, motif wide between 10 and 30. Finally, the protein structures and the distributions of motifs were combined with the phylogenetic tree using the iTOL tool (http://itol.embl.

Quantifications of SaABCCs expression profiles in tissues and response to Cd stress.
The hyperaccumulator S. alfredii Hance naturally inhabited on an old Pb/Zn mine in Quzhou City (Zhejiang Province, China) were collected and cultured in a greenhouse with a 16 h light/8 h dark cycle at an average temperature of 25-28 °C. To ensure homogeneity, seedlings of S. alfredii Hance were asexual propagated and grew in buckets filled with half-strength Hoagland-Arnon solution (pH = 6.0) in an artificial climate growth chamber for 1 months. The stock solution of CdCl 2 (0.1 M) was prepared by dissolving 22.835 g of CdCl 2 .
2.5H 2 O in 1.0 L of half-strength Hoagland-Arnon solution and the working concentration of CdCl 2 (400 μM) was set based on a previous study verifying the reliable internal genes 73 . Subsequently, 48 vigorous and uniform seedlings were randomly divided according to Cd stress. The exposure duration was 0, 0.5, 2, 4, 8, 16, 32 and 48 h, and six individuals were collected for each treatment and washed thoroughly using RNAase-free water. To avoid the influence of rhythm on gene expression, the sampling time was fixed at 14:00 pm and the exposure starting points for each treatment varied according to the exposure duration. Three tissues (whole roots, the middle part of stems, and young leaves) were collected separately and promptly frozen in liquid nitrogen for RNA extraction.
Total RNA was extracted using Total RNA Purification Kit (Norgan Biotek Corp., Ontario, Canada) from roots, stems and leaves of S. alfredii Hance, and subsequently treated with RNase-free DNase I (NEB BioLabs, Ipswich, MA, USA) to remove genomic DNA. Then, 3 μg of RNA were used to synthesize the first strand of cDNA using the SuperScript III First-Strand Synthesis System (Invitrogen, Carlsbad, CA, USA) with oligo d(T) primers (Invitrogen, Carlsbad, CA, USA). Quantitative real-time polymerase chain reaction (qRT-PCR) was carried out in triplicates on a 7300 Real-Time PCR System (Applied Biosystems, CA, USA) using SYBR Premix Ex Taq™ (TaKaRa, Dalian, China) in a 20 μL reaction system (2 μL of cDNA reaction mixture, 10 μL of SYBR Premix Ex Taq™, 0.4 μL of ROX Reference Dye, and 0.4 μL of each primer). The amplification conditions were set as follows: denaturation at 94℃ for 10 s, 40 cycles of amplifications (94℃ for 5 s, 60℃ for 31 s), and a final gradient heating from 60℃ to 95℃ for the melting dissociation curve. Gene encoding ubiquitin conjugating enzyme 9 (UBC9) was selected as the reference gene 73 and the expression levels of all target genes were adjusted by that of UBC9. The primers are listed in Table S2 (Supporting Information).
To uncover tissue-expression profiles of ten SaABCC genes in roots, stems and leaves of S. alfredii Hance, the relative expression level was calculated as the ratio of expression level of each gene to the one with the lowest expression level (Sa14F190 in roots and stems, Sa88F144 in leaves) as reference. For Cd-responsive expression profiles of SaABCC genes in different tissues of S. alfredii Hance, the relative abundance of each gene was calculated as the ratio of its expression level under Cd pressure to that without Cd exposure (time = 0). Analysis of variance (ANOVA) was performed using R software (v3.0.3) to test the significance of gene expression comparing to control at p = 0.05 level. Significantly up-regulated or down-regulated genes were those with relative abundance > 1.5 or < 0.7 and p < 0.05.
Co-expression network construction. The previously reported transcriptome dataset was constructed from mixed RNA samples of three tissues (roots, stems and leaves) from Cd-stressed (400 μM) S. alfredii plants 64 . Genes responding to Cd stress were annotated and a co-expression network was constructed to identify the modules of highly correlated genes responding to Cd stress using weighted gene co-expression network analysis (WGCNA) package 64 . Among all hub genes referring to co-expressed ones with strong interconnections, SaAB-CCs were screened and analyzed for their correlated edge genes among the annotated differentially expressed genes. The threshold of Pearson correlation coefficient of FPKM (fragments per kilobase of exon per million reads mapped) values for each gene pair was set at 0.40 and all the edges meeting the criterion were categorized by their GO annotations. The correlations of the identified hub SaABCCs and edge genes were visualized using Cytoscape v3.6.1 with the NetworkAnalyzer plugin 74 .

Yeast-expressing vector construction and heterologous expression of Sa14F190 in yeast. As
Sa14F190 gene dramatically responded to Cd stress and behaved as a hub gene in the co-expression network, it was chosen as the target SaABCC genes to characterize the functions in Cd hypertolerance and hyperaccumulation. The open reading frame (ORF) of Sa14F190 gene was amplified using High Fidelity KOD-Plus DNA Polymerase (Toyobo, Japan) from the cDNA of S. alfredii Hance with specific primers listed in Table S2 (Supporting Information). The forward primer Sa14F190-F and reverse primer Sa14F190-R was supplemented with SpeI and SmaI site, respectively. The PCR products were then gel-purified using a DNA gel extraction kit (Axygen, USA) and digested with SpeI and SmaI. The gel-purified digests were ligated into the yeast expression vector pDR196 previously cut with the same restriction enzymes at 4 °C overnight in a 10 μL reaction system (6 μL of digested PCR products, 2 μL of digested pDR196, 1 μL of 10 × T4 DNAase ligase buffer, and 1 μL of T4 DNAase ligase). The ligase was transformed into Top10 competent cells (TIANGEN, China) and cultured overnight in Luria-Bertani (LB) broth containing ampicillin (100 µg/mL). The positive colonies were further cultured in 1 mL of liquid LB broth supplemented with ampicillin (100 µg/mL) and validated by PCR using the primer pair of pDR196-F (ATG TCC TAT CAT TAT CGT CTA) and pDR196-R (CTT TTC GAT CTT TTC GTA ). They were further sequenced and the verified plasmids were designated as pDR196-Sa14F190 which were then transferred into a Cd-sensitive mutant Saccharomyces cerevisiae strain BY4742 Δycf1 (MATα; his3Δ1; leu2Δ0; met15Δ0; ura3Δ0; YDR135c::kanMX4) 54 by the lithium acetate method 75 . Positive yeast clones surviving on half-strength synthetic dextrose agar plates lacking uracil (SD-U) were further validated by PCR using the primer set of pDR196-F and pDR196-R to confirm the successful transformation of pDR196-Sa14F190 and www.nature.com/scientificreports/ designated as Δycf1_Sa14F190. Δycf1 containing empty vector was constructed following the same method and named as Δycf1_EV.

Functions of Sa14F190 gene in Cd tolerance and accumulation. The functions of Sa14F190 gene
in Cd tolerance and accumulation were assessed by yeast spotting assay, growth curve and cellular Cd content. In yeast spotting assay, Δycf1_Sa14F190 and Δycf1_EV were cultivated in liquid SD-U medium until the optical density at 600 nm (OD 600 ) reached 1.0. Cells were then serially diluted to OD 600 of 0.1, 0.01, 0.001, 0.0001 and 0.00001, spotted on SD-U agar plates containing 0 and 30 µM CdCl 2 respectively, and cultured at 28 °C for 3 days before photographs were taken. The inhibition effects of Cd on cell growth was measured in liquid SD-U medium containing 10 µM CdCl 2 , and both Δycf1_Sa14F190 and Δycf1_EV were cultured at 28 °C for 72 h. OD 600 was measured every 12 h.
To compare cellular Cd contents in Δycf1_Sa14F190 and Δycf1_EV, they were cultured in liquid SD-U medium with 10 µM CdCl 2 for 48 h, harvested by centrifugation at 5 000 × g for 10 min and washed three times with distilled water. The cell pellets were then washed by Na 2 EDTA (0.1 mM) for more than three times to remove residual Cd on cell surface. The washed cells were then dried at 65 °C until the weight was constant and finally digested by HNO 3 and perchloric acid (9:1, v:v) to determine Cd content using inductively coupled plasma mass spectrometry (ICP-MS, 7500a, Agilent, Santa Clara, CA, USA). Cd content was expressed in terms of mg/g (dry weight, DW).

Identification and classification of SaABCC genes in S. alfredii
Hance. In total, ten genes belonging to ABCC subfamily were identified from the transcriptome database of S. alfredii Hance after removing those partial or redundant sequences, and the detailed sequence information is deposited in Genbank (accession numbers listed in Table S2, Supporting Information). These SaABCC genes were named after their original IDs, and a phylogenetic tree was constructed using SaABCC genes and all ABC family members of Arabidopsis ( Fig. 1). The results illustrate that members of AtABC genes are segregated into 11 clusters and SaABCC genes cluster with the C-subfamily of AtABCs.
Tissue expression of SaABCCs. The identified SaABCC genes exhibited a diverse tissue-expression profile, as illustrated in Fig. 3. Five genes showing root-specific expression included Sa88F144, Sa23221, Sa48F96, Sa14F98 and Sa12F279. Only one gene (Sa14F190) had a leaf-specific expression pattern, and no gene exhibited specific expression in stems. The other four genes showed no obvious tissue prevalence although they differed in the expression levels. For instance, Sa13F200 and Sa18F186 were highly and lowly expressed in all three tissues.
In roots, the expression levels of SaABCCs had more variation than those in stems and leaves. Comparing the expression level of Sa14F190 gene in roots which was the lowest among all genes, the relative expression level of Sa23221 and Sa88F144 was over 2,000 times higher (p < 0.05), followed by Sa48F96, Sa12F279, Sa13F200 and Sa14F98 genes which exhibited 500 times stronger expression (p < 0.05, Fig. 3A). In stems, Sa13F200 and Sa14F98 genes had the strongest expression, over 200-fold higher than that of Sa14F190 gene (p < 0.05, Fig. 3B). The expression levels of SaABCCs in leaves fell in a relatively narrow range and the highest expressed one was Sa13F200 (Fig. 3C).

Cd-stressed profiles under different treatment time. The expression profiles postexposure to Cd
showed different responses of these SaABCC genes to Cd stress. More precisely, nine SaABCC genes (Sa14F190, Sa18F186, Sa12F279, Sa14F98, Sa118202, Sa13F200, Sa48F96, Sa88F144 and Sa45F39) only exhibited upregulated patterns in the presence of Cd in roots (p < 0.05), but not in stems and leaves (p > 0.05). Particularly, Sa14F190, Sa18F186, Sa12F279, Sa14F98 and Sa118202 genes exhibited a continuous up-regulation throughout the whole stress procedures in roots (Fig. 4A,E), and the peak expression of Sa14F190 and Sa18F186 genes was over 100-fold induced (Fig. 4A,B). Sa12F279, Sa14F98 and Sa25F86 genes displayed moderate up-regulation patterns in roots (p < 0.05, Fig. 4C,E), and the other four genes (Sa13F200, Sa48F96, Sa88F144 and Sa45F39) exhibited slight but significant positive response to Cd stress at certain time points (p < 0.05, Fig. 4F, I) www.nature.com/scientificreports/ the levels of Sa23221 gene did not show significant change (the relative abundance = 1.6, p > 0.05, Fig. 4J). All these genes did not show significant up-regulation in stems or leaves postexposure to Cd (p > 0.05, Fig. 4).

Co-expression network analysis.
Among all the identified SaABCC genes, Sa13F200, Sa45F39, Sa23221, Sa14F190 and Sa12F279 genes are characterized as hub genes showing strong interconnections with those coexpressed genes according to the co-expression network analysis (Table S1, Supporting information). The coexpression network harbors 551 nodes and 1249 connections (Fig. 5), and the edge genes have the functions associated with metabolic process (549 edges), cellular process (420 edges), biological regulation (115 edges), transporter activity (60 edges), response to stimuli (55 edges) and transcription factor (50 edges). Through the circling layout, the five SaABCC hub genes are categorized into two clades. Sa23221 and Sa12F279 genes in clade I share with some identical nodes mainly categorized to metabolic process, transporter activity and response to stimuli (Fig. 5). Nodes of clade II (Sa13F200, Sa45F39 and Sa14F190) cover the functions linked to metabolic process, transporter activity and transcription factor. Sa12F279 gene has 567 edges and is the largest module in the co-expression network, including 225 edges involved in metabolic process, 212 edges participating in cellular process, 39 edges associated with biological regulation, 21 edges executing transporter activity, 17 edges taking parts in response to stimuli and 14 edges related to transcription factor. Sa13F200 gene has more edges with response to stimuli (8/79) and transcription factor (6/79). Sa14F190 gene is assigned 254 edges mostly related to metabolic process (101 edges) and cellular process (75 edges), and harbors 16 edges executing transporter activity, 13 edges responding to stimuli and 11 edges associated with transcription factor. Most edges of Sa23221 and Sa45F39 genes are linked to metabolic process (122/251 and 44/108, respectively) and cellular process (89/251 and 28/108, respectively). www.nature.com/scientificreports/

Heterologous expression and function verification of Sa14F190 gene in Cd-sensitive yeast cells.
As Sa14F190 gene was strongly induced in roots by Cd stress and characterized as a hub gene with edges related to transporter activity and response to stimuli, it was selected for constructing yeast-expressing cell lines to prove its roles in Cd tolerance and accumulation. Results from both spotting assay and growth curves illustrated that Cd exerted a more profound inhibition effect on ∆ycf1_Sa14F190 than ∆ycf1_EV (Fig. 6). Both ∆ycf1_Sa14F190 and ∆ycf1_EV grew vigorously without Cd stress in the spotting assay, whereas the growth of  www.nature.com/scientificreports/ ∆ycf1_Sa14F190 was more significantly retarded under Cd stress than ∆ycf1_EV (Fig. 6A). Similarly, ∆ycf1_ Sa14F190 and ∆ycf1_EV exhibited different growth curves in liquid medium supplemented with CdCl 2 (Fig. 6B) that ∆ycf1_Sa14F190 took much longer time to reach the post-exponential phase (60 h) than that of ∆ycf1_EV (36 h).
To assess whether the growth inhibition was attributing to Cd accumulation, cellular Cd content was compared between ∆ycf1_Sa14F190 and ∆ycf1_EV (Fig. 6C). Cd content in ∆ycf1_Sa14F190 cells was 92.8 μg/g (DW), significantly higher than that in ∆ycf1_EV cells (68.5 μg/g DW, p < 0.01). The results suggested that the expression of Sa14F190 gene may facilitate the import of Cd inside the yeasts.

Discussion
ABC proteins are powerful transporters driving the exchange of compounds across many different biological membranes and the C-subfamily of ABC proteins is an important component 77 . In human, ABCC transporters have been studied extensively on substrate specificity, tissue expression and transport kinetics to provide insights into cellular functions, drug discovery and development 52,78 . In plants, ABCC transporters are originally defined as vacuolar pumps of GS conjugates, and deemed to associate with detoxification, PTEs sequestration, chlorophyll catabolite transport and ion channel regulation 76,79 .
As a plant hyperaccumulating Cd, S. alfredii Hance harbors numerous genes related to metal transport or detoxification and requires intense attentions. Several characterized transporter genes include SaHMA3 34 , SaZIP4 38 and SaNramp6 36 . Recent reports from Arabidopsis confer the roles of ABCC transporters as major detoxifiers sequestering metal-chelators into the plant vacuoles 57,58 , suggesting that ABCC proteins might contribute to PTEs hyperaccumulation and hypertolerance. However, the physiological functions of ABCC proteins are still scarce in metal hyperaccumulators, and it is of great importance to address their roles in S. alfredii Hance. In the present study, the composition and diversification of ABCC subfamily in S. alfredii Hance were identified using bioinformatics tools and their expression profiles were characterized for their possible roles in Cd tolerance and accumulation.
Among the identified ten proteins belonging to the ABCC subfamily, Sa45F39 is largest (1630 aa) and Sa48F96 is the smallest one (1024 aa). Similar size distribution is observed in Arabidopsis 45 , rice 46 and wheat 62 . However, the number of ABCC genes in S. alfredii Hance is fewer than those reported species. For instance, there are   81 . Therefore, we presume that S. alfredii Hance may adopt a more economic strategy of assembling multi-functional genes rather than expanding gene family to cope with the external PTEs stress during the adaptive evolution. This phenomenon is also observed for leucine-rich repeat receptor-like protein kinase (LRR-RLK) gene family which is also smaller in S. alfredii Hance than other plant species 82 . Nevertheless, these results are obtained from the transcriptomics rather than a complete genome sequencing data, which limits more precise evaluation of gene numbers especially for gene families composed by long proteins such as ABCC subfamily. The analysis on phylogeny and protein structure provides hints on the possible functions of SaABCC genes. Categorized into the division of AtABCC clade (Fig. 1), SaABCC genes are members of ABCC subfamily. All SaABCCs have similar distribution of TMD and NBD as that of AtABCCs, possessing similar protein length and spacer region (Fig. 2). However, it is worth noticing that members of SaABCCs and AtABCCs fall in different clades (I, II and III) 76 , and SaABCCs are only in clade I and II, suggesting the possibilities of multi-functionality in S. alfredii Hance. Additionally, SaABCCs have the identical numbers of MEME motifs and possess uniform distribution except for Sa23221 and Sa45F39 (Fig. 2C), demonstrating a higher conservation than AtABCCs. This www.nature.com/scientificreports/ may imply that under severe surroundings with PTEs stress, members of ABCC subfamily have prior functions to detoxify PTEs than other roles. The transcript abundance of ABCC genes across tissues helps in understanding their molecular functions. In the present study, six SaABCC genes exhibited tissue-specific expression patterns in the absence of Cd, five (Sa88F144, Sa23221, Sa48F96, Sa12F279 and Sa14F98) in roots and one (Sa14F190) in leaves (Fig. 3). Among them, Sa14F98 may participate in the plant growth and development as it is clustered with AtABCC6 which shows a similar root-specific expression 62 . Sa14F190 showing the similar expression specificity in leaves as TaABCC14 and TaABCC15 62 may serve as candidate transporters in leaves. Although the clustering of SaAB-CCs and AtABCCs hints their functional similarity, there might be functional divergences as well. For example, though Sa88F144 and Sa13F200 are clustered together with AtABCC4, Sa13F200 was abundantly expressed in all three tissues and Sa88F144 exhibited a root-specific expression pattern, neither consistent with the low expression level of AtABCC4 in all tissues 83 . ABCCs in wheat are reported to show preferential expression in specific tissues, e.g., TaABCC3 in roots, TaABCC1 in stems, TaABCC14 and TaABCC15 in flag leaves 62 . As TaABCC3 clustered with AtABCC6 is a gene highly expressed at the initiation point of secondary roots 84 , it is suggested to play roles in root architecture development 62 . Such different expression patterns imply the possibilities of performing other functions apart from getting involved in the control of stomatal movements as AtABCC4 83 .
Cd-responsive expression of SaABCC genes hints their roles in Cd hyperaccumulation and hypertolerance in S. alfredii Hance. In the presence of Cd, Sa14F190 and Sa18F186 genes were mostly induced in roots (Fig. 4). Sa18F186 is segregated with AtABCC5 which is involved in K + uptake and salt stress tolerance 85 . Although AtABCC5 is regarded as a central regulator of guard cell ion channel 59 , guard cell signaling and phytate storage 86 , rare studies report the participation of AtABCC5 in Cd response. The functional discrepancy between Sa18F186 and AtABCC5 may arise from the different leaf structures that S. alfredii has fleshy leaves and Sa18F186 therefore might enroll to combat Cd stress rather than water loss. Sa14F190 gene falls into the same clade with AtABCC3 and AtABCC7, and AtABCC3 is important vacuolar transporters conferring Cd tolerance in Arabidopsis 57 . In addition, a Cd-responding cluster is found comprising AtABCC3, AtABCC6, AtABCC7 and SAT3 (serine acetyltransferase gene) on chromosome III 84,87 . Although Sa45F39 did not respond significantly to Cd stress, it is clustered with AtABCC1/AtABCC2, which are reported to confer the tolerance to As 56 , Cd and Hg in Arabidopsis 55 . AtABCC2 is particularly involved in vacuolar transport of chlorophyll catabolites 58 and the uptake of cyanidin 3-O-glucoside (C3G) 88 . In rice, OsABCC1 closely related to AtABCC1/AtABCC2 does not confer Cd tolerance; instead, it is involved in As detoxification and reduces the allocation of As in grains 63 . Therefore, Sa45F39 is also speculated with roles in PTEs tolerance.
Han et al. proposed a comparative analysis of S. alfredii Hance transcriptomic datasets and characterized numerous hub genes strongly associated with Cd stress 64 . However, there are lack of details about SaABCC hub genes and their connections with other genes. In the present study, a co-expression network was reconstructed www.nature.com/scientificreports/ and uncovered five SaABCC hub genes (Sa14F190, Sa13F200, Sa12F279, Sa23221 and Sa45F39). The overlapping in the edge genes hints the internal connections among SaABCC hub genes. For Sa14F190 gene, the edge genes are associated with metabolic process, cellular process, biological regulation activity, transporter activity, response to stimuli and transcription factor (Fig. 5).
To be more precise, the edge genes performing transporter activity include multidrug and toxic compound extrusion (MATE) efflux family protein, zinc/iron transporter, ABC transporter B family member and sulfate transporter. Those participating in metabolic process and cellular process consist of respiratory burst oxidase homolog protein D, CBL-interacting serine/threonine-protein kinase (CIPK), plant cysteine oxidase, endoplasmic reticulum oxidoreductin-1, phospholipid-transporting ATPase 1, GDSL esterase/lipase, and heavy metal-associated isoprenylated plant protein (HIPP). Representative transcription factors include bZIP and WRKY transcription factors. Among them, MATE efflux family protein harbors multitasking abilities in encompassing regulation of plant development, secondary metabolite transport, xenobiotic detoxification, Al tolerance, and disease resistance 89 . Zinc/iron transporters differs in substrate range and specificity, involved in the transport of Zn, Fe, Mn and Cd 90,91 . Members belonging to HIPP, CIPK and WRKY transcription factor are reported to confer Cd tolerance in yeasts 92 . Therefore, the functions of Sa14F190 are activated or magnified through either direct or indirect interactions with these edge genes.
Taking Cd-responsive profiles and co-expression network together, we chose Sa14F190 to further assess its roles in Cd tolerance and accumulation by heterologous expression in Cd-sensitive yeasts. Under Cd stress, Cd content in ∆ycf1_Sa14F190 was 35.48% higher than that in ∆ycf1_EV (Fig. 6), indicating the vital roles of Sa14F190 gene in Cd accumulation. However, Sa14F190 increased the sensitivity of transformed yeasts instead of enhancing Cd tolerance, evidenced by the retarded growth which may be due to more Cd accumulation. Similar findings are observed for other transporters in S. alfredii Hance which all enhanced the Cd sensitivity and Cd accumulation of transformed yeasts, and the increase of cellular Cd content was 22.2% by SaNramp6 36 , 25.2% by SaHMA3 34 and 16.4% by SaCAX2h 39 . It has been reported that AtABCC3 could confer Cd tolerance in yeast cells, but not Cd accumulation 93 , consistent with another study on AtABCC3 in Arabidopsis 57 . These findings suggested that members of ABCC subfamily closely located in the phylogenetic tree might have functional divergence, and Cd tolerance and accumulation are mediated by different pathways.
In summary, the ABCC transporter Sa14F190 gene is responsible for Cd hyperaccumulation and other SaABCCs might have diverse roles in the tolerance or accumulation of PTEs. Members of ABCC transporter are a genetic pool of candidates encompassing strong ability to transport, tolerate or accumulate Cd and other PTEs for phytoremediation. This work applies bioinformatic analysis to provide a preliminary study uncovering the interesting functions of ABCCs members in S. alfredii Hance, and these functions need further experimental evidence to enrich our knowledge.

Conclusions
In the present study, a first comprehensive evolutionary analysis of ABCCs genes in S. alfredii Hance was carried out using a transcriptome data set. By characterizing the composition and structure of ten identified ABCC proteins, we found similar domain arrangements and conserved motifs shared by SaABCCs, indicating their similar evolutionary history. Bioinformatics analysis visualized the cluster of the putative SaABCCs and AtAB-CCs and composed with representative distribution of protein domains. These SaABCC genes exhibited tissuespecific expression profiles, particularly in roots and stems. Cd-responsive expression profiles uncovered the up-regulation of five SaABCC genes, among which Sa14F190 and Sa18F186 genes were most strongly induced. Co-expression network also suggested that Sa14F190 gene was one of the five SaABCCs hub genes tensely associating with Cd stress. Heterologous expression of Sa14F190 gene in Cd-sensitive yeast cells proved a stronger accumulation but less tolerance of Cd by Sa14F190-expression cell lines, hinting the possible roles of Sa14F190 genes in Cd transport. Our findings open a new door to understand the evolution and functions of SaABCCs in S. alfredii Hance, unravel their roles in adapting Cd stress, provide new clues on Cd transport and detoxification in S. alfredii Hance, and add perspectives for studying mechanisms of PTEs tolerance and accumulation in other hyperaccumulators.