Molecular modeling and in silico characterization of GmABCC5: a phytate transporter and potential target for low-phytate crops

Abstract

Designing low-phytate crops without affecting the developmental process in plants had led to the identification of ABCC5 gene in soybean. The GmABCC5 gene was identified and a partial gene sequence was cloned from popular Indian soybean genotype Pusa16. Conserved domains and motifs unique to ABC transporters were identified in the 30 homologous sequences retrieved by BLASTP analysis. The homologs were analyzed for their evolutionary relationship and physiochemical properties. Conserved domains, transmembrane architecture and secondary structure of GmABCC5 were predicted with the aid of computational tools. Analysis identified 53 alpha helices and 31 beta strands, predicting 60% residues in alpha conformation. A three-dimensional (3D) model for GmABCC5 was developed based on 5twv.1.B (Homo sapiens) template homology to gain better insight into its molecular mechanism of transport and sequestration. Spatio-temporal real-time PCR analysis identified mid-to-late seed developmental stages as the time window for the maximum GmABCC5 gene expression, a potential target stage for phytate reduction. Results of this study provide valuable insights into the structural and functional characteristics of GmABCC5, which may be further utilized for the development of nutritionally enriched low-phytate soybean with improved mineral bioavailability.

Introduction

Soybean, an agriculturally important crop, is noted for its high contents of protein (40%), oil (20%), dietary fiber and other beneficial phytochemicals like isoflavones and vitamins. Despite the manifestation of these unique nutritional attributes, it has limited consumption, which can be accredited to the presence of an absorption inhibitor, phytic acid. Phytic acid [myo-inositol 1,2,3,4,5,6-hexakisphosphate (InsP₆)] bears a strong negative charge due to the presence of six attached phosphorus groups on the myo-inositol ring, enabling chelation with many essential divalent cations like iron, zinc, calcium and potassium and consequently reduces their bioavailability (Loewus and Murthy 2000). Almost half of the phosphorus fertilizer given to the crops is lost and the remaining half gets assimilated and translocated to the seeds, where a substantial (75% of the total assimilated) amount is bound and stored as phytate in vacuoles and becomes unattainable to the plant. This is a noteworthy bottleneck in the flow of phosphorus through the world’s agrarian ecosystem. Also, the digestive tracts of non-ruminants lack enzymes for the removal of phosphates from the myo-inositol ring of phytic acid, and are therefore unable to utilize the phosphorus found in legumes and cereals. Unutilized phosphorus present as phytic acid is excreted and used as manure for the fertilization of crops. The runoff from these agricultural fields containing high amount of unutilized phosphorus can further result in environmental phosphorus pollution, eutrophication and impaired water quality (Raboy 2003; Lott et al. 2000). The serious consequences have thus sparked off numerous studies and approaches toward reduction of phytate levels in the seeds of cereals and legumes used as food and feed. Considerable efforts have been made for lowering of phytic acid in cereals and oilseeds by targeting the genes involved in its biosynthesis and generating low-phytate mutants (lpa), which accumulate more inorganic phosphate without affecting the total plant phosphorus content (Raboy 2001). The biosynthesis of phytic acid takes place in the cytosol by sequential phosphorylation of the primary substrate, myo-inositol (Greenwood and Bewley 1984; Rasmussen et al. 2010), and subsequent transport and compartmentalization is undertaken by ATP-binding cassette (ABC) transporters. The targeting of ABC transporter gene has been reported to have a far-reaching effect on phytate reduction without adversely affecting plant growth and development (Shi et al. 2007).

The ABC transporters represent one of the largest superfamily of ubiquitous membrane proteins. The plant genome is particularly rich in these transporters, where Arabidopsis and Orzya sativa are known to encode more than 120 ABC transporter proteins. The ABC transporter proteins mediate MgATP-energized transmembrane transport and also regulate other transporters. These transporters in plants have been designated into different subfamilies based on the identity of their nucleotide-binding domain (NBD) and the transmembrane domains (TMD) and further prediction of function based on subfamily membership. The plant members of various subclasses have been implicated in a plethora of processes viz, lipid catabolism, polar auxin transport, xenobiotic detoxification, stomatal function, disease resistance and the transport of mineral ions and lipids. The ABC transporters of subfamily C compartmentalize phytate into protein storage vacuoles coupled with ATP hydrolysis (Raboy 2002; Xu et al. 2009; Regvar et al. 2011). The role of plant ABC transporter family proteins in phytic acid transport has been well documented (Nagy et al. 2009; Dorsch et al. 2003; Panzeri et al. 2011; Xu et al. 2009; Shi et al. 2007), but reports on its expression profile, physiochemical and structural properties in soybean are sparse. Similarly, the elucidation of the complete molecular and functional mechanisms of transmembrane phytate transport requires deep insight into the proteins’ structural features. Therefore, in this study, efforts were made to identify the ABC transporter gene implicated as a potential candidate for developing low-phytate soybean, an agriculturally and nutritionally important crop. Its protein sequence was characterized through bio-computational tools for elucidation of biochemical, physiochemical and structural properties to gain a better insight into the biological mechanism of phytate sequestration and transport.

Materials and methods

Plant material

Field-grown soybean variety Pusa 16 from the Division of Genetics, IARI, New Delhi, was used for the study. For the spatial expression profile, samples from root, stem, leaf and flower were collected from the 30-day-old plants. Developing soybean seeds were collected and sorted based on seed size into developmental stages. Tissue samples and developing seeds were harvested by flash freezing in liquid nitrogen and stored at − 80 °C for RNA isolation.

Amplification and cloning of GmABCC5 partial sequence

Total RNA was isolated from the samples using the TRIZOL reagent (Invitrogen, USA) and treated with RNase-free DNase I (Thermo Scientific, USA) for removal of any residual genomic DNA as per manufacturer’s instructions. RNA yield was quantified by spectrophotometry with Nanodrop 2000 (Thermo Scientific, USA) to achieve uniform concentration for cDNA synthesis, and RNA integrity was checked by separation on 1.2% agarose gel. 1 µg of total RNA was reverse transcribed for cDNA synthesis using the RevertAid^TM H Minus First strand cDNA Synthesis Kit (Thermo Scientific, USA) according to the manufacturer’s protocol. The primer pair designed, GmABCC5F [5′-ATG AAG CCA TAG AGG CCA TGG ATA TCC CTA C-3′] and GmABCC5R [5′-TAG CTT CTG TGC AGC TGC AAG ACC AAA C-3′], were used for the amplification of GmABCC5 partial gene sequence (521 bp) from the previously reported genomic sequence of GmABCC5 deposited in NCBI GenBank (accession number XM_003554257.3), implicated in lowering of phytic acid in soybean (Maroof et al. 2009). The sequence amplified from cDNA was cloned in the pGEM^®-T Easy vector (Promega, USA). The recombinant plasmids were verified by restriction analyses and sequenced using universal primers (SP6 and T7) on an automated sequencer (ABI 3730xl DNA Analyzer, USA). The nucleotide sequence data were submitted to the GenBank (accession number KF183505). To analyze the exon/intron organization of GmABCC5 (XM_003554257.3), the genomic and CDS sequences of the gene were downloaded from the NCBI database and compared in the Gene Structure Display Server 2.0 (Hu et al. 2015) program.

Differential gene expression by quantitative real-time PCR

To study the expression profile, target gene primers were customized from the exonic regions of previously reported complete genomic sequences of GmABCC5 (GenBank: accession number XM_003554257.3) for amplification of approximately 200 bp from the conserved region of the gene of interest (qABCC5F 5′-TCG ACG TTG TTT TCT TCT TCT TCT TC-3′ and qABCC5R 5′-CCG TTC TCC TTA CCG AAT CTA ACT-3′). The data were normalized using the soybean housekeeping gene PEP carboxylase (qPEPCoF 5′-CAT GCA CCA AAG GGT GTT TT-3′ and qPEPCoR 5′-TTT TGC GGC AGC TAT CTC TC-3′). Real-time expression analyses of the target and reference gene was carried out in PikoReal 96 Real Time PCR System (Thermo Scientific, USA). The experimentation was performed according to the standard protocol using DyNAmo Flash SYBER Green qPCR Kit (Thermo Scientific, USA). The three biological samples were run in triplicate under the following conditions: 1 cycle of 3 min at 94 °C followed by 35 cycles of 95 °C for 30 s and 30 s at 60 °C with a final extension of 30 s at 72 °C. The amplified products were subjected to melt-curve analysis and the specificity of the amplification was assessed by dissociation curve analysis. A unique peak on the dissociation curve was confirmed for the gene. Relative expression of target genes was calculated using the efficiency calibrated $2^{[-\mathrm{\Delta }\mathrm{\Delta }C(T)]}$ method (Livak and Schmittgen 2001). For the spatial expression analysis, flower tissue was considered as the calibrator, whereas for temporal analysis, expression in the representative 2–4 mm sized seeds was taken as calibrator and their expression was taken as one and the fold change was calculated. The specificity of the PCR amplification reaction was analyzed on 1.2% agarose gel.

Sequence analysis of GmABCC5 gene

The related GmABCC5 amino acid sequences from other plants were identified via Basic Local Alignment Search Tool (BLASTP) implemented on databases of the National Center for Biotechnology Information (http://blast.ncbi.nlm.nih.gov/blastp/) using the previously deduced GmABCC5 (accession number XM_003554257.3) amino acid sequence as a query. The amino acid compositions of the selected sequences were computed using PEPSTATS (http://emboss.bioinformatics.nl/cgi-bin/emboss/pepstats/) analysis tool (Rice et al. 2000). The ExPASy ProtParam tool (http://web.expasy.org/protparam/) was used to compute the physiochemical characteristics of the selected 31 homologous sequences (Gasteiger et al. 2005). The amino acid sequences were fed into MUSCLE (http://www.ebi.ac.uk/Tools/msa/muscle/) for multiple alignments and the alignment file was imported to BioEdit program (Hall 1999) to identify and shade the conserved amino acid sequences. The phylogenetic analysis was carried out using the neighbor-joining (N-J) method with 2000 bootstrap replicates implemented in MEGA (Molecular Evolutionary Genetic Analysis) version 6 program (Tamura et al. 2013), to evaluate the degree of support for the particular grouping pattern in the constructed cladogram.

Structural analysis, domain prediction and functional characterization of GmABCC5

Secondary structure elements predictions were performed using RaptorX server (http://raptorx.uchicago.edu/) SS3 (3-State Secondary Structure Prediction) (Wang et al. 2016) and PSIPREDv3.2 server (http://bioinf.cs.ucl.ac.uk/psipred/) (Buchan et al. 2013), domain prediction was done using DomPred, protein disorder was predicted using RaptorX DISO and FoldIndex (http://bioportal.weizmann.ac.il/fldbin/findex). Online prediction tools DiANNA (http://clavius.bc.edu/~clotelab/DiANNA) (Ferre and Clote 2005) and CYS-REC (http://linux1.softberry.com/berry.phtml) were utilized to determine the most likely bonding patterns involving the available cysteine residues. Solvent accessibility of the protein was estimated using RaptorX server ACC program. Membrane helix and topology prediction was done using HMMTOP, MEMSAT3 and MEMSAT-SVM programs from the PSIPRED server. The probable subcellular protein location was determined using TargetP server (http://www.cbs.dtu.dk/) (Emanuelsson et al. 2000), MEMSAT SVM and ProtComp v. 9.0 (http://www.softberry.com/berry.phtml?topic=protcomppl&group=programs&subgroup=proloc). For the probable gene term annotations, FFPred 3 program was used for prediction of all Gene Ontology domains (Biological Process, Molecular Function, Cellular Component) from PSIPRED server (http://bioinf.cs.ucl.ac.uk/psipred/).

Homology modeling and quality assessment of the predicted structure

The 3D structure of the target protein was constructed by template-based homology modeling using SWISS-MODEL software. The theoretical model was built by ProMod3 target-template alignment using 5twv.1.B (Homo sapiens) as the reference template. The stereochemical properties of the obtained model were evaluated by Ramachandran plot analysis with RAMPAGE (Lovell et al. 2003). The 3D model of GmABCC5 was fed to Pymol Molecular Graphic System (http://pymol.org/ep) to obtain the final model structure. The PDB file of the modeled GmABCC5 was analyzed on ProSA-web server for obtaining consistent values for the model generated (http://prosa.services.came.sbg.ac.at/prosa.php) (Wiederstein and Sippl 2007) and What IF (http://swift.cmbi.ru.nl/whatif/) for packing quality assessment of the theoretical model.

The GmABCC5 has a well-documented role in phytate transport and storage; hence, identification of probable binding site residues is imperative for detailed mode-of-transport analysis. No probable ligand/substrate was identified by the SWISS-MODEL software, hence RaptorX-binding prediction server (http://raptorx.uchicago.edu/BindingSite/) (Kallberg et al. 2012) was employed for identification of probable substrate and binding site residues.

Results and discussion

Previous studies had reported that a single base mutation in ABC transporter gene resulted in a low-phytate trait in soybean. A single base substitution from A (wild-type) to T (mutant) encoded a stop codon at the amino acid level, generating a loss-of-function mutant ABC transporter protein product (Maroof et al. 2009). Using the sequence-based homology search with the mutated sequence as query helped in identifying the full-length GmABCC5 gene (gene id: XM_003554257.3).

Cloning and sequencing of GmABCC5

Custom-designed primers, synthesized based on sequence data of the GmABCC5 (XM_003554257.3) available in NCBI, were used to amplify the 521 bp open reading frame (ORF) from the seeds of Pusa 16, a popular Indian soybean genotype. This partial gene sequence earmarked for cloning will be further utilized for generation of low-phytate trait soybean. The amplicon was cloned in pGEM^®-T Easy vector, sequenced and submitted in NCBI as KF183505.1 (Fig. 1a, b). The coding region of GmABCC5 was determined to be governed by 11 exons and 10 introns (Fig. 2), where the first exon encodes the maximum number of sequences and the ninth exon encodes the least as predicted by Gene Structure Display Server 2.0. The partial sequence (521 bp) cloned was found to be the part of the fourth exon in the gene.

Fig. 1.

Cloning of GmABCC5 cDNA in pGEMT easy vector system. a PCR amplification of 521 bp fragment isolated from Glycine max cv. Pusa 16 using gene-specific primers and separated on 1% agarose gel along with 100 bp molecular weight marker, b GmABCC5 sequence submitted in NCBI (KF183505.1)

Fig. 2.

Exon/intron organization of GmABCC5. The GmABCC5 genomic sequence and CDS sequences were downloaded and the sequences were compared in Gene Structure Display Server Program to infer the exon/intron organization. The targeted 521 bp sequence lies in the fourth exon of GmABCC5 gene

Expression pattern of GmABCC5

Analysis of the abundance of the GmABCC5 gene transcripts revealed differential expression pattern across the set of tissue samples displayed in Fig. 3. The maximum expression of GmABCC5 gene was found in mature soybean seeds followed by the stem, leaf, root and flower (Fig. 3), respectively. It was observed that the relative expression in seeds was 52-fold higher in comparison to the flowers (taken as the control). In other tissues like the flower, leaf, root and stem, the relative expression was less than fivefold, given that seeds are the major production site and thus necessitating maximum transporter gene activity in seeds for phytic acid transport and compartmentalization (Raboy 1997). Phytic acid is required for several vital functions in plants and is essential for their survival; hence, a basic level of phytate is observed in all tissues, coinciding with the relative abundance of the gene. In the temporal expression profiles, the maximum relative transcript accumulation was observed in the 8–10 mm seed stage. The maximum relative transcript levels observed were in the 8- to 10-mm-sized seeds almost 52-fold higher in comparison to the 2- to 4-mm-sized seeds which were used as the control (Fig. 4). Overall, the expression of GmABBC5 in seeds was found to be higher during the later developmental stages in comparison to the initial stages as observed in the relative abundance profile, which varied from 46-fold in 10–12 mm, followed by 45- and 43-fold in 14–16 and 12–14 mm, respectively, 19-fold in 6–8 mm, and less than 6-fold in 4–6 mm and 0–2 mm seed sizes. The PEPCo gene was used as the internal control of reference for assessing the relative expression pattern of the gene. The pattern of variation in the transcript abundance indicates well toward the probable major role of GmABCC5 during the later stages of seed development. In soybean after anthesis, the initial stages of seed development are devoted to the production of phosphorus compounds needed for growth and development (nucleic acids, membrane phospholipids and signaling molecules), and the synthesis of phosphorus reserve, phytic acid is minimal (Raboy and Dickinson 1987). The gene was highly expressed in the late stages of seed development as compared to the early stages. This coincides evenly with the biosynthesis of phytic acid during seed development. Temporal phytate accumulation has been studied in various crops such as Glycine max (Raboy and Dickinson 1987; Israel et al. 2011); Brassica napus (Dong et al. 2012); Pisum sativum (Shunmugam et al. 2015) and Hordeum vulgare (Hatzack et al. 2001), which revealed that the flux of phosphorus is mainly channeled toward phytic acid biosynthesis from around 21 days after flowering until seed maturity. In spite of early initiation of phytic acid biosynthesis in developing seeds (0.24 g 100 g⁻¹ in 0–2 mm), its major accumulation occurs during middle (1.85–2.87 g 100 g⁻¹ in 4–6 to 8–10 mm) and later stages (3.35 g 100 g⁻¹ in 12–14 mm) of seed development (Pandey et al. 2016).

Fig. 3.

Real-time spatial expression analysis of GmABCC5 gene in soybean tissues (Pusa 16), PEPCo gene was used as the internal control for normalizing the relative expression. The mean value ± SE was calculated from data pooled from three experimental replicates

Fig. 4.

Real-time temporal expression analysis of GmABCC5 gene in soybean (Pusa 16) developing seeds ranging from initial 0–2 to 14–16 mm (full maturity) seed size. PEPCo gene was used as the internal control for normalizing the relative expression. The mean value ± SE was calculated from data pooled from three experimental replicates

Since the ABC transporter protein has a role in the transport and storage of phytic acid, the gene is expressed actively after a threshold level of phytic acid accumulates in the cytoplasm during seed development. The spatial expression profile identifies seed as the major tissue for its expression and the temporal profile suggests that the maximum relative expression occurs in late seed development stages; thus, the relative gene expression observed in seeds coincides with the accumulation of a large amount of phytic acid during the late development stages (Pandey et al. 2016). Thus, the present study illustrates that the initial intermediates of phytic acid biosynthesis, like myo-inositol and other phosphates, are synthesized earlier in seed developmental process, while the accumulation of the final product, phytic acid, occurs more in the later stages. Targeting the GmABBC5 gene expression during late seed development stages may provide a route for the creation of low-phytate soybean with enhanced nutritional value without affecting the intermediates, imperative for normal plant growth and development.

Physicochemical and structural features

GmABCC5 sequence similarity search using BLASTP analysis revealed homology to protein sequences of Glycine soja (KHN02460.1), Vigna radiata var. radiate (XP_014495750.1) and Phaseolus vulgaris (CBX25010.1) with 99, 93 and 92% of identity, respectively. Table 1 shows the selected 30 GmABCC5 homologs retrieved from NCBI BLASTP using GmABCC5 as the query sequence. ABC transporter gene sequences identified with the role in phytate transport were also included from viz., Zea mays (Shi et al. 2007), Arabidopsis thaliana (Nagy et al. 2009) and Triticum aestivum (Bhati et al. 2016) for better relevance.

Table 1.

ABC transporter protein sequences retrieved from the NCBI database (http://www.ncbi.nlm.nih.gov/)

S. no.	Gene Id	Plant species	Sequence length (amino acid residue)
1	XP_003554305.1	Glycine max	1537
2	KHN02460.1	Glycine soja	1690
3	XP_003521316.1	Glycine max	1539
4	XP_014495750.1	Vigna radiate var. radiata	1537
5	XP_007162606.1	Phaseolus vulgaris	1538
6	XP_017411139.1	Vigna angularis	1538
7	XP_019455530.1	Lupinus angustifolius	1533
8	XP_012569380.1	Cicer arietinum	1532
9	XP_003541373.1	Glycine max	1517
10	XP_014514170.1	Vigna radiate var. radiata	1513
11	KHN48949.1	Glycine soja	1495
12	XP_006443721.1	Citrus clementina	1536
13	XP_019427850.1	Lupinus angustifolius	1539
14	EOX94300.1	Theobroma cacao	1539
15	XP_015901176.1	Ziziphus jujuba	1531
16	XP_003625394.2	Medicago trunculata	1514
17	EOX94301.1	Theobroma cacao	1535
18	XP_015948969.1	Arachis duranensis	1531
19	XP_013468420.1	Medicago trunculata	1530
20	XP_007144409.1	Phaseolus vulgaris	1513
21	XP_011024496.1	Populus euphratica	1532
22	XP_017604352.1	Gossypium arboreum	1540
23	OAY34676.1	Manihot esculenta	1531
24	OAY32700.1	Manihot esculenta	1531
25	XP_017649043.1	Gossypium arboreum	1543
26	XP_012484085.1	Gossypium raimondii	1540
27	NP_001106060.1	Zea mays	1510
28	AIK23242.1	Triticum aestivum	1227
29	NP_171908.1	Arabidopsis thaliana	1514
30	XP_002468528.1	Sorghum bicolor	1512

ExPASy ProtParam and PEPSTATS tools were used for the primary structural analysis of the selected GmABCC5 homologs by computing different parameters and are tabulated in Table 2. The analysis revealed Leu as the most abundant amino acid for all the ABCC5 proteins under study, accounting for about 12% in all the protein’s primary structure, followed by Ser and Val; Cys and His were the least abundant amino acid residues encountered. The predicted average molecular weight (MW) determined was 169,358 g mol⁻¹ (Table 2). The calculated average pI of the ABC transport proteins sequences was 7.47 (range 6.06–8.59), indicating that this enzyme is likely to precipitate in basic buffers although eight proteins have pI less than 7, signifying that in those species protein will supposedly precipitate in acidic buffers (Table 2). This information will be of much use in the development of buffer systems for the purification of recombinant ABC transport proteins by isoelectric focussing. The longer proteins are typically composed of more charged amino acids, thus much better to buffer the effect of fluctuation in their composition and can keep near neutral pI (close to 7.4), whereas extreme pI values are particular for smaller proteins (Schwartz et al. 2001). There was more variation in the pI (range 6.06–8.59) of the ABC transport proteins than their size (range 136,696–187,880 g mol⁻¹, partial sequence from Triticum aestivum), concurrent with reports that the size of orthologous proteins found in directly related organisms is much more conserved than their pI (Nandi et al. 2005). The proteins localized to membrane contain more basic residues which are located generally on either side of the membrane-spanning region, essential for the stability of protein in the membrane (Schwartz et al. 2001). Hence, membrane proteins are basic in nature as observed here. The in vivo half-life (T_1/2) of the selected transporters was computed via the instability index (Ii). It has been reported in previous studies that proteins having Ii > 40 have a T_1/2 of less than 5 h, while those having Ii < 40 have a longer T_1/2 of 16 h (Rogers et al. 1986). Our study showed that 13 proteins were stable, and the other ABC proteins including GmABCC5 were thermally unstable. Although the positional or interaction effects of the adjacent residues were not considered for analyzing the stability, the study provides some indication about the physical state of the protein. GRAVY indices of ABC transporter sequences ranged from 0.094 to 0.274 (Table 2), reflecting the hydrophobicity of the amino acids (Kyte and Doolittle 1982) and grand average hydropathicity. These results thus confirm that all ABC transporters tend to be hydrophobic in nature, which endorses its multifaceted role in cellular membrane transport. Aliphatic index (Ai), another parameter that measures the relative volume occupied by the aliphatic side chains of the amino acids such as Ala, Val, Leu and Ile, serves as a measure of the thermostability of proteins (Gupta et al. 2012). In this study, the Ai values of all the transporters under consideration ranged from 106.62 to 111.24. All the ABC transporter protein sequences analyzed had a high Ai (103.15–111.24) value, an indicator of protein stability over a wide temperature range, thus indicating that all the proteins are thermostable. The extinction coefficient (EC), which indicates the light absorbed by the protein at a wavelength of 280 nm, was found to be in a range of 1.19–1.36 M⁻¹ cm⁻¹ estimated with respect to the concentration of Cys, Trp, and Tyr, was less as Cys and Trp were the least abundant amino acids.

Table 2.

Parameters of inositol phosphate kinases (plants) calculated using the ProtParam program: molecular weight (MW) (g/mol); isoelectric point (pI); extinction coefficient (EC) (M⁻¹ cm⁻¹); instability index (Ii); aliphatic index (Ai); grand average hydropathy (GRAVY); number of negative residues (− R); number of positive residues (+R)

S. no.	Plant species	MW	PI	EC	II	AI	GRAVY	R −	R +
1	Glycine max	170,554	7.0	1.28	41.57	111.24	0.274	154	153
2	Glycine soja	187,880	7.61	1.27	42.32	107.62	0.186	172	174
3	Glycine max	170,953	6.64	1.25	40.90	110.60	0.252	158	154
3	Vigna radiate var. radiata	170,429	6.85	1.31	40.62	107.25	0.221	156	154
4	Phaseolus vulgaris	171,203	7.03	1.36	40.85	107.14	0.229	156	155
5	Vigna angularis	170,622	7.43	1.30	40.29	106.62	0.213	155	156
6	Phaseolus vulgaris	171,191	7.03	1.36	40.75	106.89	0.226	155	156
7	Lupinus angustifolius	170,888	7.26	1.27	40.59	107.10	0.188	155	155
8	Cicer arietinum	170,790	6.06	1.30	41.81	108.69	0.236	160	147
9	Glycine max	168,332	6.42	1.26	39.00	106.49	0.216	157	150
10	Vigna radiate var. radiata	168,698	7.47	1.32	39.57	106.50	0.201	151	152
11	Glycine soja	165,946	6.48	1.27	39.09	106.23	0.202	155	149
12	Citrus clementina	170,591	8.44	1.20	37.24	108.62	0.241	150	161
13	Lupinus angustifolius	171,123	7.85	1.31	40.36	108.25	0.176	151	154
14	Theobroma cacao	170,738	8.09	1.20	39.17	109.03	0.246	150	155
15	Ziziphus jujuba	170,310	8.59	1.20	41.38	107.20	0.193	150	163
16	Medicago trunculata	168,549	6.44	1.25	40.28	107.67	0.223	157	151
17	Theobroma cacao	170,234	7.97	1.20	39.32	109.06	0.249	150	154
18	Arachis duranensis	169,911	7.85	1.19	36.75	107.43	0.218	151	154
19	Medicago trunculata	170,057	6.66	1.27	39.74	108.82	0.239	155	151
20	Phaseolus vulgaris	168,932	6.71	1.31	39.26	106.95	0.213	156	152
21	Populus euphratica	170,369	8.13	1.17	41.00	108.53	0.240	149	155
22	Gossypium arboreum	170,660	7.06	1.211	39.49	108.77	0.251	155	154
23	Manihot esculenta	170,411	8.44	1.21	38.87	108.21	0.224	147	157
24	Manihot esculenta	170,418	8.20	1.23	40.33	109.12	0.234	147	153
25	Gossypium arboreum	171,089	7.82	1.20	39.70	107.67	0.223	154	157
26	Gossypium raimondii	170,615	7.47	1.21	39.86	109.21	0.252	153	154
27	Zea mays	166,790	8.44	1.28	46.28	107.19	0.190	144	155
28	Triticum aestivum	136,696	8.18	1.19	43.47	103.15	0.094	127	132
29	Arabidopsis thaliana	168,574	7.64	1.23	40.45	108.37	0.209	159	161
30	Sorghum bicolor	166,545	8.31	1.27	44.67	107.31	0.203	146	154

Domain identification and evolutionary analysis

Multiple alignment of the selected amino acid sequences of the identified GmABCC5 and 30 different homologs was performed using ClustalW server and the sequences were imported to BioEdit software. The significantly conserved amino acids are shown in Fig. 5. Most of the amino acids in the ABC_MRP_domains (accession: cd03250 and cd03244) are well conserved in all the homologs as indicated by low E values (Fig. 6) identified by the CDD program. The ABC_MRP domain is the nucleotide-binding fold/domain (NBD) encompassing highly conserved amino acid motifs which are involved in energy generation and hence power the transport mechanism (Kerr 2002). Two additional ATP-binding motifs were identified within the NBD: the Walker A motif (GXXGXGK(S/T)) and the Walker B motif (hhhhDE, where h is a hydrophobic amino acid) (Fig. 5). The Walker A motif is involved in establishing extensive interactions with the phosphate group of an ATP molecule; the Asp of the Walker B motif coordinates water and Mg²⁺ at the catalytic site while the catalytic glutamate is vital for ATP hydrolysis (Vetter and Wittinghofer 1999; Zaitseva et al. 2005; Jones and George 2002). A conserved Gln identified at the Q-loop, located between the Walker sequences, performs a similar role in ATP molecule coordination. Two additional motifs, conserved D-loop and H (or switch)-loop, were identified downstream of the Walker B motif containing a conserved Asp and His, respectively, involved in the coordination of the γ-phosphate either through interaction with a water molecule (D-loop) or through direct hydrogen bonding (H-loop) (Zaitseva et al. 2006; Jones and George 2012). An ABC signature motif, a helical subdomain, was identified located in between the Walker sequences of each NBD containing the conserved amino acid sequence LSGGQ, also involved in ATP binding (Schmees et al. 1999). The presence of ABC_membrane superfamily domain (accession: cl00549), representing hydrophobic transmembrane helices, was also identified as conserved by the low E values in all the homologs, which is an essential feature of ABC transport proteins, mandatory for transmembrane transport of molecules (Fig. 6) (Schulz and Kolukisaoglum 2006). It has been reported that NBDs tend to have high sequence similarity, but the transmembrane domains from related transporters vary from having a high sequence identity to probably no similarity, thus making phylogenetic analyses desirable to ascertain the similarity and evolutionary link between the ABC transporter proteins.

Fig. 5.

a, b Schematic representation of multiple sequence alignment of GmABCC5 homologs by ClustalW and image generated by BioEdit program for identification of conserved domains and motifs

Fig. 6.

Domain analysis of GmABCC5 and its homologs as analyzed by BLASTP and CDD program

Sequence conservation across different species is an important indicator of functionality and evolution. (Baum 2008). Phylogenetic analysis provides a useful framework to understand the relationship of different homologs and their evolution from a common ancestor. We constructed a neighbor-joining phylogenetic tree of GmABCC5 and its 30 homologs using (Fig. 7) MEGA 6.0 software. The proteins formed four distinct clades, each comprising evolutionarily related members from the selected homologs and the branching patterns were supported by high bootstrap values. Pattern branching of the phylogenetic tree revealed that the ABC transport proteins of plants belonging to Fabaceae formed a monophyletic cluster with the proteins from citrus, coffee, cotton, etc., indicating close evolutionary relationship and suggesting a common ancestor. The Fabaceae family formed a separate clade with two sub-clades comprising members from soybean, chickpea, beans, etc., and the proteins from citrus, coffee, cotton etc. formed another subgroup in the second clade of phylogeny. Out of the four ABC transport proteins having a well-identified role in phytate transport, the protein from Arabidopsis formed a separate third clade, and the three ABC transporters from the Graminae family formed another well-delineated fourth clade. The phylogram thus illustrated a high degree of divergence among the legumes, cereals, and the horticultural crops despite having high sequence similarity.

Fig. 7.

Cladogram of the selected 30 ABC transport proteins based on their amino acid sequences as outlined in Table 1 using Mega 6.0 software with an outgroup. The tree was inferred based on the neighbor-joining algorithm

Secondary structure prediction

Structural analyses are prized sources of information on shapes and domain structure, protein classification, function prediction and interactions with other macromolecules. The structure provides the first framework to understand the interactions of transporter protein with the ligands/substrates at the molecular level. The secondary structure information in addition aids in the prediction of homology-based protein models. The secondary structure of the GmABCC5 protein is dominated by alpha helix (60%), whereas beta sheets and random coils contribute to 7 and 32%, respectively, as revealed by three-state secondary structure prediction (SS3) by RaptorX server. PSIPRED secondary structure prediction determined a total of 53 alpha helices and 31 beta strands in the protein’s structure. The secondary structures are more conserved than the nucleotide sequences, which help in understanding the molecular evolution (Reehana et al. 2013). Protein sequence analysis by RaptorX predicted disorder in 22% of GmABCC5 residues, further confirmed by FoldIndex server (Wang et al. 2016; Prilusky et al. 2005). The 1537 residues of GmABCC5 revealed 0.326 unfoldability, 0.001 charge and 0.530 phobic values. There were 6 predicted disordered regions having a total of 171 disordered residues. Predicted disordered segments were [290]–[297] length: 8 score: − 0.02 ± 0.03, [299]–[307] length: 9 score: − 0.03 ± 0.02, [505]–[533] length: 29 score: –0.13 ± 0.06, [864]–[879] length: 16 score: − 0.06 ± 0.03, [892]–[956] length: 65 score: − 0.15 ± 0.09 (longest disordered region) and [1373]–[1416] length: 44 score: − 0.13 ± 0.06. Protein intrinsic disorder since long has been linked to molecular recognition, binding activities and regulatory processes in the light of both experimental and computational investigations and hence is functionally important (Cozzetto and Jones 2013; Wright and Dyson 2015).

Disulfide predictions using DiANNA server revealed the presence of 29 Cys residues in GmABCC5 and the most probable bonds as predicted by CYS-REC were 82–1313, 158–896, 163–1480 and 654–905. Solvent access was determined by RaptorX (Wang et al. 2016) revealing residues as exposed, intermediate and buried, 30, 42 and 27%, respectively, based on relative solvent accessibility (RSA), where exposed had 40–100%, intermediate 10–40% and buried 0–10% RSA. HMMTOP, MEMSAT SVM and MEMSAT 3 are transmembrane protein topology prediction programs where MEMSAT SVM is a support vector machines (SVM) based predictor, a highly reliable program which can also identify pore-lining residues. These programs predicted 16 transmembrane helices (Table 3) with the total entropy of the model 17.0221. The transmembrane helices illustrated were rich in hydrophobic amino acids, which is well documented by the Kyte and Dolittle mean hydrophobicity profile, and the final annotated topology of this transmembrane protein was visualized by MEMSAT SVM (Fig. 8a, b) (Nugent and Jones 2009). It also identified three pore-lining helices, featuring residues from 323 to 353, 587 to 610 and 1085 to 1108 with a pore stoichiometry of 1. It has been reported that the transmembrane domain region of the transporter proteins, which forms the transport channel, exclusively consists several membrane-spanning alpha-helices, along with kinks and bends, providing considerable structural variability essential for its function (Dahl et al. 2004; Moussatova et al. 2008). Thus, the presence of transmembrane and hydrophobic residues identified its membrane-embedded nature and role in transmembrane transport which can be further validated by structural and functional characterization.

Table 3.

Total number of transmembrane helices and regions involved in the formation of transmembrane helix as predicted by MEMSAT SVM

Helix no.	Residues
1	37–61
2	98–122
3	141–156
4	168–187
5	205–233
6	323–353
7	370–388
8	445–465
9	469–491
10	549–577
11	587–610
12	963–990
13	1012–1041
14	1085–1108
15	1112–1127
16	1188–1218

Fig. 8.

a Schematic diagram for topology analysis prediction of the GmABCC5 query sequence by MEMSAT SVM and MEMSAT3. b Topology analysis cartoon diagram for GmABCC5 by MEMSAT SVM

Structural and functional characterization

Computational protein localization and targeting predictions aid in in silico protein function characterization as well as genome annotation and is therefore a significant study. Signal peptides and transmembrane helices provide useful hints about the protein’s subcellular localization and transmembrane transporter activities. Protein sequence analyses based on TargetP scores (cTP: 0.315, mTP: 0.014, SP: 0.216, other: 0.702) suggested GmABCC5 to be located anywhere in the cell besides mitochondria, with less chance of presence in chloroplast as well. Also, the signal peptide (SP) score reinforced its membranous nature. Results compared to predictions obtained from MEMSAT-SVM server identified a 32-residue long N-terminal signal region positioned at 1–32, indicating that the protein was guided through a secretary pathway to the cellular membrane. ProtComp server identified protein’s location to the plasma membrane and vacuole, and also a probable GPI anchor at the 1501 residue. The GPI anchors are prevalent in membrane proteins and have a confirmed role in providing membrane-attached proteins a stable membrane-anchoring device, and are resistant to most extracellular proteases and lipases (Low 1989). FFPred is an SVM-based program, where a large suite of protein’s physicochemical property predictors cover many features such as signal peptides, membrane helices, secondary structure and disorder terms for prediction of gene ontology (Cozzetto et al. 2016). Gene ontology prediction by FFPred predicted biological function as transmembrane transport and ion transport with 0.940 and 0.907 probabilities, respectively. Molecular function prediction determined anion/substrate-specific transmembrane transporter activity with 0.981 probability and cellular component prediction as integral component of membrane with 1.0 probability. Domain analysis, structural information and GO prediction results, together, strongly recommend GmABCC5 protein’s membrane localization and role in transmembrane transport of substrates, particularly phytate which needs further evidence from 3D structure study.

Homology modeling and model validation

Atomic resolution crystal structures of several soluble proteins have been reported, but similar progress has not been made for the transporter proteins, where the major obstacle encountered is the extreme difficulty in their crystallization owing to conformational flexibility. Hence, the development of homology-based protein model is a relevant alternative (Dahl et al. 2004). Pymol Molecular Graphic System was used to visualize the final 3D structure of GmABCC5 (Fig. 9a), generated through template-based homology model with SWISS-MODEL software. Molecular modeling methods have been routinely used to analyze the structure, dynamics, surface properties and thermodynamics of inorganic, biological and polymeric systems. The 5twv.1.B template [ATP-binding cassette subfamily C member 8, channel conductance-controlling ATPase (cystic fibrosis transmembrane conductance regulator from Homo sapiens)] with a sequence identity score of 35.17% was identified as the top template for homology-based modeling for GmABCC5. In the context of QMEAN global scores, QMEAN-Z score was observed as − 4.96, indicating relatively good model quality (Fig. 9b) and the bio-computed model was further assessed by the ProSA server. The Z score value is a determinant of model quality that predicts the total energy of the structure (Wiederstein and Sippl 2007). As shown in Fig. 9c, the obtained Z score value of GmABCC5 was − 10.47, which is within the range typically found for comparable protein chains of PDB, indicating reliability of the structure. The energy plot (Fig. 9d) shows the local model quality by plotting energies as a function of amino acid sequence position and, in general, positive values correspond to problematic or erroneous parts of a model. The residues with the negative energies further confirmed the consistency of this predicted model (Kulkarni and Devarumath 2014).

Fig. 9.

a Homology model of GmABCC5 protein generated using Swiss-model and rendered using PyMOL. b Plot showing normalized Q mean score of the model. c Z score value of the model generated. d Ramachandran plot validating backbone dihedral angels of the energy-minimized model

The packing environment for residues of the modeled GmABCC5 was compared with the experimental structures by WHAT IF. Poor structural packing is specified by a score of less than or equal to − 5.0 (Singh and Pandey 2015). Analysis reveals that the generated structure has similar packing scores compared to the template X-ray structure with very few residues having poor packing as reflected by their scores being less than − 5.0. Ramachandran plot assessment of GmABCC5 for stereochemical and energetic properties by RAMPAGE server revealed 92% of residues (1188) in the most favored zone, 6.3% of residues (81) in the allowed region and 1.7% of residues (22) in the disallowed/outlier region (Fig. 9d). The statistics in the favored and allowed region and relative low percentage in the outlier suggest that the Ramachandran plot for GmABCC5 is acceptable. The goodness factor (G-factor) from the PROCHECK results showed relevant information between covalent and overall bond angle distances. Analysis of G-factor of the modeled GmABCC5 was − 0.32, which revealed the quality of the predicted model as very good. The overall Ramachandran plot attributes and the G-factor assured the quality of the GmABCC5 structure (Table 4). ERRAT analysis of GmABCC5 showed an overall quality factor of 79.935. The results obtained from various quality evaluation servers certify the quality of the 3D protein model developed in this study. Although GmABCC5 has a well-documented role in ATP-coupled phytate transport, where ATP and phytate are obvious substrates, no substrate/ligand was identified for the model developed and, hence, no binding sites were found. However, another reliable online server RaptorX-binding site prediction determined PO₄ as the probable ligand and identified two binding sites governed by the residues: site 1: W655, V684, G685, S686, G687, K688, S689, Q717 (Fig. 10a); and site 2: F175, F178, W193 (Fig. 10b). The first site identified lies in the ABC_MRP domain, the nucleotide-binding fold specific for the coordination and hydrolysis of the ATP molecule (López-Marqués et al. 2015); thus, the probable ligand (PO₄) can be speculated to be the part of ATP moiety, and the binding site residues might be involved in energy-coupled transport. As for the second site, no probable domain was identified, but PO4 moiety identified as ligand can either be the component of ATP or phytate salt, which requires further validation. Unraveling the complete mechanistic of phytate discrimination and transport across the membrane requires further studies on complete model, preferably developed through atomic resolution. The 3D structure thus developed will open new avenues and experimental approaches for identification of protein–ligand-/substrate-binding sites and interactions for better insights into the transport mechanism for planning of in vivo experiments.

Table 4.

PROCHECK statistics of the predicted 3D model

G factors
Parameter	Score	Average score
Dihedral angles
Phi–psi distribution	− 0.12	− 0.24
Chi1–chi2 distribution	0.22
Chi1 only	0.08
Chi3 and chi4	0.73
Omega	− 1.04
Main-chain covalent forces
Main-chain bond lengths	0.18	− 0.26
Main-chain bond angles	− 0.57
Overall average		− 0.23

Fig. 10.

Interactions of PO4 substrate with residues in the active site of GmABCC5 protein a site 1 and b site 2, generated by RaptorX and visualized using JasMol

Conclusion

ATP-binding cassette (ABC) C family 5 transporters in soybean couple ATP hydrolysis to the sequestration and transport of phytate across the cellular membrane in protein storage vacuoles. Here, we have identified and cloned the partial gene sequence of ABCC5 from Glycine max var. Pusa16. The GmABCC5 gene was predominantly expressed in the later phase of seed development, and hence is an appropriate target window for generation of low-phytate soybean without affecting its growth and development. Analyses by computational tools shed light on its secondary structure, physicochemical features, evolutionary linkages, localization and function which can act as a starting point for experimental studies. The three-dimensional protein model, developed through template-based homology modeling, can provide insight into the functional mechanism and molecular structure and can enable the formulation of hypotheses regarding transporter structure and function, which needs further experimental validation. This study would thus be helpful in developing strategies for the development of low-phytate, nutritionally enhanced crops without affecting other plant parameters.

Abbreviations

ATP

Adenosine triphosphate

ABC

ATP-binding cassette transporter

ABCC5

ABC transporter subfamily C member 5

Gm

Glycine max

PEPCo

PEP carboxylase

PDB

Protein Data Bank

NJ

Neighbor joining

MW

Molecular weight

pI

Isoelectric pH

EC

Extinction coefficient

Ai

Aliphatic index

Ii

Instability index

GRAVY

Grand average hydropathy

Compliance with ethical standards

Conflict of interest

The authors declare that there is no conflict of interest regarding the publication of this article.