Introduction
The goosefoot genus Chenopodium (Chenopodiaceae, x= 9) has a worldwide distribution (Bhargava et al., Reference Bhargava, Shukla and Ohri2006). The foliage constitutes a rich source of carotenoids, minerals and vitamin C (Prakash et al., Reference Prakash, Nath and Pal1993; Bhargava et al., Reference Bhargava, Shukla and Ohri2010). Quinoa (Chenopodium quinoa) seeds have protein in the 12.8–15.7% range, with elevated amounts of the essential amino acids threonine, lysine, methionine and tryptophan (Repo-Carrasco et al., Reference Repo-Carrasco, Espinoza and Jacobsen2003).
Chenopodium berlandieri is a goosefoot that was domesticated at least three times anciently in Eastern North America: (1) as a thin-testa form of C. berlandieri ssp. jonesianum; (2) a thick-testa form of C. berlandieri that may have been utilized as a leafy vegetable and (3) another thin-testa seed crop morphologically similar to a modern Mexican domesticated chenopod, C. berlandieri ssp. nuttalliae cv. ‘huauzontle’ (Smith and Yarnell, Reference Smith and Yarnell2009; Jellen et al., Reference Jellen, Kolano, Sederberg, Bonifacio, Maughan and Kole2011). Genetic data from hybridization studies (Wilson, Reference Wilson1990), karyotype analyses (Bhargava et al., Reference Bhargava, Shukla and Ohri2006; Palomino et al., Reference Palomino, Hernandez and Torres2008; Kolano et al., Reference Kolano, Gardunia, Michalska, Bonifacio, Fairbanks, Maughan, Coleman, Stevens, Jellen and Maluszynska2011) and gene sequencing (Maughan et al., Reference Maughan, Kolano, Maluszynska, Coles, Bonifacio, Rojas, Coleman, Stevens, Fairbanks, Parkinson and Jellen2006; Storchova et al., Reference Storchova, Drabesova, Chab, Kolar and Jellen2014; Walsh et al., Reference Walsh, Adhikary, Maughan, Emshwilller and Jellen2015) indicate that C. berlandieri, its South American weedy ecotype Chenopodium hircinum, and Andean C. quinoa form a New World biological species complex of allotetraploids (2n= 4x= 36), whose two subgenomes, designated A and B, likely originated from diploids in the New World and Old World, respectively. In addition to vegetable huauzontle, subspecies nuttalliae also includes seed and semi-weedy cultigens in Mexico (García-Andrade and De La Cruz, Reference García-Andrade and De La Cruz2011).
Starch, a vital energy component of seeds, consists of molecules of α-d-glucopyranose in two different types of polymers: linear, helical α-1,4 linked chains of insoluble amylose; and mixed α-1,4 and α-1,6 branched, water-soluble amylopectin. Atwell et al. (Reference Atwell, Patrick, Johnson and Gloss1983) and Lindeboom et al. (Reference Lindeboom, Chang, Tyler and Chibbar2005) reported that quinoa starch has amylose contents ranging from 3 to 20%. Amylose is synthesized by GBSS, while amylopectin is synthesized by soluble starch synthases, starch branching enzymes and starch debranching enzymes (Park et al., Reference Park, Nishikawa, Tomooka and Nomoto2012c), in concert with GBSSI (Denyer et al., Reference Denyer, Johnson, Zeeman and Smith2001).
The waxy (wx), amylose-free seed phenotype is due to loss of function of the GBSSI gene. Waxy mutants occur in many cereals including wheat (Huang and Brule-Babel, Reference Huang and Brule-Babel2012), rice (Hirano et al., Reference Hirano, Eiguchi and Sano1998; Crofts et al., Reference Crofts, Abe, Aihara, Itoh, Nakamura, Itoh and Fujita2012) and cassava (Manihot esculenta; Aiemnaka et al., Reference Aiemnaka, Wongkaew, Chanthaworn, Nagashima, Boonma, Authapun, Jenweerawat, Kongsila, Kittipadakul, Nakasathien, Sreewongchai, Wannarat, Vichukit, Lopez-Lavalle, Ceballos, Rojanaridpiched and Phumichai2012). Starches that are waxy or are low in amylose are desirable in rice (Liu et al., Reference Liu, Ma, Liu, Zhu, Jiang, Wang, Shen, Ren, Dong, Chen, Liu, Zhao, Zhai and Wan2009). Upon cooking, waxy starches produce a soft paste with a sticky texture, whereas wild-type starches produce a harder gel that separates easily from the cooking water (Hunt et al., Reference Hunt, Moots, Graybosch, Jones, Parker, Romanova, Jones, Howe and Trafford2013) and can recrystallize (Denyer et al., Reference Denyer, Johnson, Zeeman and Smith2001).
The objective of this study was to survey the distribution of waxy mutants in landrace populations of huauzontle. Preliminary screening of five C. quinoa varieties failed to identify any with a waxy phenotype; however, one out of two huauzontle populations, designated ‘H-02’, was waxy (Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). We hypothesize that huauzontle, being a vegetable crop, was not subjected to as stringent a level of selection pressure for seed quality as quinoa.
Materials and methods
Plant materials and starch phenotyping
In this study, we worked with 11 Mexican populations or distinct cultigens of Chenopodium berlandieri ssp. nuttalliae: (1) H3, Atlacomulco, Mexico State, 19°48′N, 99°52′W, 2570 masl; (2) H5, Tenango del Valle, Mexico State, 18°39′7″N, 99°31′37″W, 2600 masl; (3) H7 opaque, El Capulín, Otzolotepec, Mexico State, 19°25′55″N, 99°33′28″W, 2580 masl; (4) H9, Zolotepec, Xonacatlán, Mexico State, 19°24′N, 99°32′W, 2610 masl; (5) H17 translucent, La Concepción Huchochitlán, Toluca, Mexico State, 19°37′32″N, 99°39′14″W, 2680 masl; (6) H18 translucent, La Concepción Huchochitlán, Toluca, Mexico State, 19°37′32″N, 99°39′14″W, 2689 masl; (7) H35-08 translucent, San Andrés Cuexcontitlán, Mexico State, 19°22′08″N, 99°36′40″W, 2670 masl; (8) PI 433230, Guadalajara, Jalisco, 20°37′46″N, 103°22′24″W, 1500 masl; (9) PI 433231, Atlixco, Puebla, 18°53′45″N, 98°21′41″W, 1880 masl; (10) PI 568155 Cacaloxuchil, Puebla, 18°45′0″N, 98°30′0″W, 1680 masl and (11) PI 568156, Acutzilapan, Mexico State, 19°47′0″N, 99°41′0″W, 2700 masl (Fig. S1, available online). Accessions H17 and H18 were collected together in the same field, but they are distinct cultigens.
The approximate content of both amylose and amylopectin was analysed by staining with Lugol's I2–KI Stain (0.1 g resublimated iodine and 0.2 g KI dissolved in 30 ml distilled water) following the protocol developed by Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014), who had previously demonstrated the utility of this technique in Chenopodium. Lugol's Stain is a common tool for discriminating between non-waxy (purple-blue) and waxy (red-brown) seeds, for example, in the grain amaranths (Park et al., Reference Park, Nemoto, Nishikawa, Matsushima, Minami and Kawase2012a, Reference Park, Nishikawa, Tomooka and Nomotob). Inspection of stained quinoa starch suspensions from powdered seed was performed under 630–1000 × magnification using an Axioskop 2 microscope (Zeiss, Jena, Germany).
Populations one through seven of C. berlandieri ssp. nuttalliae listed above were obtained from the Instituto Nacional de Investigaciones Nucleares (ININ) in Ocoyoacac, Mexico. Most of these accessions contained seeds of a mixture of phenotypes, including seeds that were translucent, opaque and of various colours (mostly brown, orange-red or black). Since previous studies by Park et al. (Reference Park, Nemoto, Nishikawa, Matsushima, Minami and Kawase2012a, Reference Park, Nishikawa, Tomooka and Nomotob) had noted an association between opaque/non-waxy and translucent/waxy perisperm, respectively, in grain amaranths (Amaranthus spp.) – which are in the same family, Amaranthaceae, as Chenopodium – we wanted to see whether this association also held true in huauzontle. Translucent seeds were selected, with the aid of a transmitted light box, for accessions H3, H5, H9, H17, H18 and H35-08. Opaque seeds were selected from accession H7. Population H9 contained black and yellow-translucent seeds and the staining with iodine solution was different between them; for that reason, we sequenced GBSSI from DNA of plants derived from both seed phenotypes in this population, yellow-translucent and black. Hence, after raising single plants from selected seeds of known huauzontle phenotypes, we obtained the whole sequence of GBSSI representing eight different seed phenotypes from seven accessions. Considering that all of the foregoing populations of C. berlandieri ssp. nuttalliae are property of the Mexican government and are, therefore, restricted germplasm sources, we decided to also starch-phenotype and sequence the A-genome homoeoallele in four strains of huauzontle from the United States Department of Agriculture-National Plant Germplasm System (USDA-NPGS) collection to identify potential lines that could be used as parents in developing publicly available breeding lines. This group included: PI 433230, PI 433231, PI 568155 and PI 568156 (Fig. S1, available online). The 512 bp portion of GBSSI from the start codon to the end of exon 2 was sequenced in these accessions.
DNA extraction
Seeds of each population were germinated in the Life Sciences Greenhouse at Brigham Young University, Provo, UT, and young leaves were collected. Fresh leaves of three plants were lyophilized, individually, in a freeze-dryer at 0.7 atm for 2 d. Between 20 and 30 mg of dry leaves were crushed (FastPrep FP120; Bio 101 ThermoFisher Scientific, Waltham, MA, USA), and the DNA was extracted according to Dellaporta and Hicks (Reference Dellaporta and Hicks1983) and modified from Dellaporta (Reference Dellaporta, Freeling and Walbot1993). The total concentration of DNA of each sample was determined using a NanoDrop ND-1000 Spectrophotometer (Thermo Scientific, Wilmington, DE, USA). We used only one DNA extraction of one individual from each population for polymerase chain reaction (PCR) amplification, cloning and sequencing. We selected the sample having the best DNA quality.
GBSSI amplification and cloning
Primers for GBSSI cloning were designed from amaranth and C. quinoa sequences (Geneious v. 6.1.6; Biomatters Ltd., Auckland, New Zealand, available from http://www.geneious.com) as described by Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). Four segments were cloned between nucleotides 176 and 2478: first pair: 180 F, 5′-ACG CGA AAA ATC CTA CTG AGG AGC-3′ and 864 R, 5′- CAC GCT AAA TCG AAG CTG GT-3′; second pair: 646 F, 5′-TTC CAC ACC TAC AAG CGA GG-3′ and 1718 R, 5′-CAG GCA AAT GAA GAC GCG AG-3′; third pair: 1454 F, 5′-GGC ATA GTG CTC TTC TCC CAG CC-3′ and 2223 R, 5′-ACC AAC TTC TGC TTG TAG GGC TTC C-3′; fourth pair: 2020 F, 5′-ATG GAT GTC CTG GAA TGG AA-3′ and 2840 RA, 5′-CCC ATA TGG AAT CCG GTG TA-3′. The genomic DNA (100 ng per reaction) was amplified using each pair of primers in order to obtain the amplification of the GBSSI gene. We used One Taq 2x Master Mix with Standard Buffer (New England Biolabs, Ipswich, MA, USA). The PCR product was purified using a PCR clean-up system (Wizard® SV Gel and PCR Clean-Up System; Promega, Madison, WI, USA).
Subsequently, the PCR product, previously purified and quantified, was introduced into a vector system (pGEM®-T and pGEM®-T Easy Vector Systems; Promega, Madison, WI, USA). Cloning was necessary in order to separate individual sequence variants from heterogeneous PCR amplicons due to the presence of multiple genomes in the allotetraploid. Six or more selected clones were randomly chosen for sequencing per amplified fragment and for each individual (one individual per sample). We used between 50 and 100 ng/ml of DNA per 10 ml reaction volume. The ligation and transformation protocols had been described previously by Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014).
In order to amplify and sequence the ends of GBSSI, we employed a rapid amplification of complementary DNA (cDNA) ends (RACE) strategy (Yeku and Frohman, Reference Yeku and Frohman2011) using the SMARTer RACE cDNA Amplification Kit (Clontech, Mountain View, CA, USA), following the manufacturer's protocol. The mRNA for this procedure was extracted from immature inflorescences (green and pre-anthesis) collected from C. quinoa and C. berlandieri plants as described in Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). The inflorescences were immediately frozen in liquid nitrogen and subsequently stored at − 80°C until use. A Qiagen kit was used for RNA extraction (RNeasy, Plant Mini Kit 50 Cat. No. 74904, CA, USA). At the end, RNA was quantified in a Nano Drop Spectrophotometer (ND-1000, V3. 8.1, 2010, Thermo Scientific, Wilmington, DE, USA).
Transferrin receptor (TFR) primers were designed based on the C. quinoa sequence; we designed different primers for different places at the start and end of the two homoeologous genes. Primers for the 5′-end of the genome A copy were 20F/736R and 20F/2171R. Primers for the 3′-end of the genome A copy were 1820F/3295R, 1783F/3278R and 2224F/3278R. Primers for the 5′-end of the genome B copy were -80F/1539R, 180F/1537R and -80F/624R. Primers for the 3′-end of the genome B copy were 1200F/3415R, 1300F/3276R and 1300F/3334R. The PCR amplicons obtained with these primers were cleaned by the phosphatase/exonuclease protocol (New England Biolabs, Ipswich, MA, USA).
DNA sequence analysis
Sequences generated by Sanger sequencing as described by Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014) were aligned and analysed using the Geneious software (Geneious v. 7.0.6; Biomatters Ltd., Auckland, New Zealand, available from http://www.geneious.com). The first step was identification of A and B subgenomes in order to align sequences by subgenome groups (Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). Previously, both A- and B-genome variants were obtained by sequencing diploid species of Chenopodium. Genome A was sequenced from the North American diploid Chenopodium standleyanum and Genome B (BB) from Eurasian Chenopodium ficifolium. We measured the pairwise % identity of each huauzontle GBSSI sequence and the higher of the two identity values when compared with the two diploids was used to match that sequence with either the A or B genome – usually, between 97.5 and 100% for the A genome and approximately 97.5% for the B genome, as had been previously performed with C. quinoa and C. berlandieri by Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014).
We sequenced between three and six amplicon-containing colonies per population to capture both the A and B subgenomes. After their classification in subgenome A or B, we aligned all the segments, and total GBSSI sequence was obtained for each of the accessions or cultigens. After sequencing flanking ends of the genes with the RACE reaction protocol (Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014), the complete coding sequence could be identified. Sequence annotation was performed (i.e. identification of intron/exon junctions, and start and stop codons) by comparison with sequences of C. quinoa and C. berlandieri previously reported by Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014) and by comparison with Amaranthus sequences deposited in NCBI (http://www.ncbi.nlm.nih.gov).
Results
Starch staining
Potassium iodide staining of each of the eight cultigens showed clear differences in the staining among them (Fig. S2, available online). The presence of stained granules in brown, purple or blue was observed in each of the eight varieties or cultigens, but in different proportions. The percentage of both purple and blue granules was 83% for H3 translucent, 75% for H7 opaque, 87% for H9 black, and 80% for the H18 translucent red seed. Three cultigens showed a high percentage of starch granules stained brown. The proportion of granules stained brown was 98% for H5 translucent, 87% in H9 translucent and 88% in H17 translucent. Only one of the cultigens showed an intermediate staining pattern. Sample H35-08 stained 50% purple-brown and 50% brown. According to these results, we identified the following as non-waxy cultigens: translucent H3, opaque H7, black H9 and translucent H18. We identified cultigens translucent H5, translucent H9 and translucent H17 as being waxy. The H35-08 cultigen was classified as intermediate – perhaps as low-amylose – cultigens (Fig. S2, available online). Subsequently, PI's 433231 and 568155 were observed to be staining brown (waxy), while 433230 and 568156 stained purple or non-waxy (results not shown). Seed starch phenotypes are indicated by accession label colour on the map in Fig. S1 (available online).
GBSSI in Chenopodium berlandieri var. nuttalliae
After sequencing the ends of the gene GBSSI with RACE methodology, we were able to identify both the start and stop codons (Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). The prediction of exon/intron sites was facilitated by the relatedness of the genera Chenopodium and Amaranthus, both of them being confamilial in the family Amaranthaceae. The identity of the GBSSI gene of C. berlandieri (A or B genome) in comparison with the Amaranthus hypochondriacus genome was 80–84%. The identity was essentially the same when C. berlandieri was compared with close relatives of A. hypochondriacus, Amaranthus cruentus or Amaranthus caudatus. The alignment of the coding regions of A. caudatus, A. hypochondriacus and the eight accessions or cultigens of C. berlandieri var. nuttalliae with their entire GBSSIa and GBSSIb genes sequenced revealed a pairwise % identity of 92.5, with 79.2 % of sites identical (481 sites) for GBSSIa. The B genome of C. berlandieri was very different, having a pairwise % identity of 85.3, with 67.9% of sites being identical (412 sites) (Fig. S3, available online). In huauzontle, the A genome of GBSSI had a total of 13 exons – exactly the same as in C. quinoa (Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). These exons were present in all eight of the cultigens, the number of nucleotides being constant in each one (Table 1). The sum of the nucleotides of the 13 exons within the A genome was 1818 (including the stop codon), which resulted in a coding sequence of 605 amino acids (minus Ter). The eight cultigens showed only slight amino acid sequence differences between non-waxy and waxy accessions (Fig. 1). The waxy genotypes H3 and H17 shared three similarities: a Pro (vs. Ser) residue at position 274, an Ile (vs. Val) residue at position 325 and a Val (vs. Leu) residue at position 456. Additionally, within landrace accession H3, there were three polymorphisms, indicative of heterozygosity within the sampled plant: the first was ambiguity for Thr or Ile at position 54, the second was the presence of either Ile or Val at position 325 and the third was a Leu/Val ambiguity at amino acid position 456 (Fig. 3). Since Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014) hypothesized that the Ile → Thr substitution is associated with improper plastid targeting of GBSSIa, we sequenced this portion of the gene in the four USDA-NPGS accessions and noted that PI 433230 was heterozygous T/C (Ile/Thr), PI 433231 was homozygous C/C (Thr), PI 558155 was homozygous C/C (Thr) and PI 558156 was homozygous T/T (Ile) at position 54.
Table 1 Combinations of sequence variants in huauzontle and their associations with seed morphology and starch phenotype
Glu, glutamic acid; Ala, alanine.
Seed morphology: T, translucent; O, opaque; B, black.
Other explanations: Del, deletion present; Wt, wild-type (no deletion).
Fig. 1 Amino acid sequence of granule-bound starch synthase I (GBSSI), subgenome A in huauzontle (Chenopodium berlandieri ssp. nuttalliae). The total length of the wild-type allele is 605 amino acids. Three polymorphisms were detected at different positions along the amino acid chain: position 54, threonine or isoleucine (T/I); position 325, isoleucine or valine (I/V) and position 456, leucine or valine (L/V). The mutation at position 54 lies within a conserved portion of the plastid-targeting transit peptide (Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014).
In contrast, the B genome had a greater range of sequence polymorphisms when compared with the A genome (Figs 1–3). The alignment in Fig. 2 shows 12 sites along the sequence in which there were differences among the accessions. All of the non-waxy plants possessed the normal-length allele with 13 exons. However, the B genome of the waxy accessions had a deletion 79 amino acids in length, resulting in a hypothetical polypeptide consisting of only 11 exons. Differences between the A and B subgenome sequences for GBSSI are presented schematically in Fig. 3. We detected five waxy populations and three non-waxy populations in samples provided from ININ, México. The missing region was towards the 5′-end of the gene and affected the first three exons. Exon 1 was lacking the last 45 amino acids, Exon 2 was completely eliminated and Exon 3 was missing the first 7 amino acids (Fig. 2). The sum of nucleotides in the 13 exons of the B-genome allele in the non-waxy varieties was 1818 (including the last three nucleotides of the stop codon), for a total of 605 amino acids – the same as the A-genome alleles. However, waxy varieties had only 1581 nucleotides (including the last three nucleotides of the stop codon), for a total of 526 amino acids. This represents a reduction of 13% in the length of the coding region. In this way, mutations in the B genome allowed us to clearly identify waxy (null) and non-waxy gene variants. The accessions H7 opaque, H9 black and translucent H18, which we classified as non-waxy, showed the wild-type, 13-exon allele. The deletion mutation was detected in all landraces classified by potassium iodide staining as waxy: translucent H5, translucent H9, translucent H17 and even in the low-amylose translucent H35-08. Interestingly, though we had classified H3 as non-waxy according to potassium iodide staining, it was also homozygous for the gbssIb-del mutation, which suggests that presence of only one functional copy of GBSSIa is sufficient to confer the wild-type phenotype with respect to seed amylose.
Fig. 2 Amino acid sequence of granule-bound starch synthase I (GBSSI), subgenome B in huauzontle (Chenopodium berlandieri ssp. nuttalliae). The total length of the wild-type allele is 605 amino acids. Waxy cultigens have a deletion mutation affecting amino acids 65–143. One polymorphism was also detected at position 417, where either glutamic acid or alanine (E/A) is possible.
Fig. 3 Schematic of granule-bound starch synthase I (GBSSI) gene in base pairs of cultigens of huauzontle (Chenopodium berlandieri ssp. nuttalliae). Waxy genotypes are H3 translucent, H5 translucent, H9 translucent, H17 translucent and H35-08 translucent. Non-waxy genotypes include H7 opaque, H9 black and H18 translucent. Numbers above the boxes represent the number of base pairs in each exon of the coding region.
We separately analysed two seed phenotypes in accession H9: yellow-translucent seeds that had 87% brown-staining starch granules and black seeds having 87% purple-blue starch granules. In keeping with our expectations, yellow-seeded plants having the waxy phenotype bore the gbssIa-tp (Thr at position 54) and gbssIb-del alleles, while black-seeded and non-waxy plants had the gbssIa-tp mutation but wild-type GBSSIb. However, waxy translucent H9 had one ambiguous result: two alternate sets of primers ( − 80 and − 624, 180 and 1537) used to amplify the 5′-region of the gene indicated that the Ile–Thr mutation at position 54 was not present. However, it is interesting that at position 417 in the amino acid sequence of genome B, all non-waxy cultigens, including H9 black seed, encoded the amino acid alanine (Ala). In contrast, all of the waxy genotypes had glutamic acid (Glu). Interestingly, translucent H9 was heterozygous for a polymorphism at this position, having alleles encoding both Glu and Ala.
Discussion
We identified haplogroups associated with the waxy or low-amylose seed starch phenotypes in huauzontle (Table 1 and Fig. S2 (available online)). Individuals having the I54T-I325V-V456L haplotype in the A genome plus the large deletion in the B-genome produced less detectable (H35-08) to no (H5, H9T) seed amylose. Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014) identified the same GBSSI haplogroups in waxy huauzontle H2. Genotype H17 was anomalous, having not only all three A-genome substitutions, but also the substitution S274P. Intriguingly, this mutation, while also present in H3, had been previously reported in diploid Chenopodium neomexicanum accession BYU 843 by Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). This taxon is morphologically very similar to allotetraploid C. berlandieri var. sinuatum, a wild Sonoran ecotype most similar to cultivated ssp. nuttalliae and therefore its candidate progenitor. The additional prior finding of Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014) that the I325V-V456L combination is in wild C. berlandieri ecotypes from such widespread locations as Maine (BYU 803), the Texas Gulf Coast (BYU 937), Utah (BYU 652) and northern Argentina (BYU 1101, C. hircinum) suggest that V325 and L456 may more accurately be considered the wild-type A-genome haplotype, albeit without the I54T substitution.
Several reports have found the existence of a transit peptide in the first 77 amino acids of the gene GBSSI. The GBSSI sequence in Amaranthus cruentus had 606 amino acid residues, including a transit peptide of 77 amino acids (Park et al., Reference Park, Nemoto, Nishikawa, Matsushima, Minami and Kawase2009). In tartary buckwheat (Fagopyrum tataricum), also a pseudocereal, the genomic sequence of FtGBSSI contained 3947 nucleotides and was composed of 14 exons and 13 introns (Wang et al., Reference Wang, Feng, Xu, Sestili, Zhao, Xiang, Lafiandra and Wang2014). Nevertheless, the sequence of deduced FtGBSSI protein contained 605 amino acids, like C. berlandieri. Interestingly, they discovered a cleavage site in the FtGBSSI protein sequence towards the N-terminus with a transit sequence of 78 amino acids (8.4 kDa) and a mature protein of 527 amino acids (58.2 kDa) (Wang et al., Reference Wang, Feng, Xu, Sestili, Zhao, Xiang, Lafiandra and Wang2014). The sweet potato IbGBSSI protein also contained a signal peptide of 77 amino acids (Wang et al., Reference Wang, Yeh and Tsai1999).
In this study, consequently, we hypothesized that the first 77 or so amino acids constitute the transit peptide for CbGBSSIa. Under this scenario, the heterozygous I54T substitution in H3 is interesting because there is a polymorphism at this position, while the B-genome homoeoallele contains a large deletion suggestive of a null allele. These data support the hypothesis of Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014) that the presence of only one functional GBSSIa allele is sufficient to produce seed amylose. Future quantitative tests to measure seed amylose should verify whether or not there is a quantitative or additive reduction in amylose with decreasing doses of functional GBSSI alleles.
Appearance of the perisperm was an accurate indicator of content of amylose or amylopectin in the grain amaranths, with opaque seeds having the waxy mutation and translucent seeds the non-waxy genotype (Park et al., Reference Park, Nemoto, Nishikawa, Matsushima, Minami and Kawase2009). However, this same pattern (Fig. S2, available online) was not verified in either huauzontle or quinoa (Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014).
Although this study included a relatively small number of huauzontle samples, it is evident from Fig. S1 (available online) that the gbssIa-tp I54T and gbssIb-del mutations that result in waxy huauzontle are distributed across a broad geographic area and do not appear to follow a discernable pattern of grouping. Unfortunately, we know very little about the historical, let alone ancient, distribution of this crop (Wilson and Heiser, Reference Wilson and Heiser1979). The relative commonality of the waxy phenotype in huauzontle is intriguing, especially since no waxy phenotypes were identified among 22 previously examined quinoa cultivars (Lindeboom et al., Reference Lindeboom, Chang, Tyler and Chibbar2005; Brown et al., Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014). Lindeboom et al. (Reference Lindeboom, Chang, Tyler and Chibbar2005) had identified geographically diverse South American quinoa genotypes with seed amylose concentrations of 3.5–19.5%. This discrepancy can possibly be explained by one of the two hypotheses. The first possibility is that GBSSI mutants in one or more of the two subgenomes had already been present in a mixed weedy domesticated complex C. berlandieri population from which vegetable huauzontle evolved, whereas these mutations were absent in the ancestral population in South America that gave rise to quinoa. Such mutations would have no effect on phenotype because of gene duplication and the recessive nature of the waxy phenotype. An alternate hypothesis is that stringent selection for seed plumpness, hardiness, amylose-related cooking properties, etc., in South America eliminated waxy grain quinoa genotypes as they periodically emerged. In contrast, huauzontle in the central Mexican highlands was selected as a vegetable, rather than a grain, crop, hence accumulating seed quality mutations would not have been culled out as stringently as in the quinoa growing regions of South America. If waxy mutations reduced the cooking time of more mature huauzontle inflorescences, they might even have been unconsciously selected.
Future research will include efforts to combine the gbssIa-tp mutant allele with mutant B-genome alleles from quinoa to produce waxy quinoa cultivars. Our preliminary screen of four publicly available huauzontle accessions from the USDA-NPGS identified three accessions – PI 433230, PI 433231 and PI 558155 – that are either homogeneous or heterogeneous for waxy mutations and carry the gbssIa-tp allele. Brown et al. (Reference Brown, Cepeda-Cornejo, Maughan and Jellen2014) identified a single putatively null B-genome allele, designated gbssIb-t (W129X), in lowland quinoa genotype ‘G205-95’. A PI 433231 × ‘G205-95’ F2 population, for example, would be expected to harbour waxy:non-waxy plants at a ratio of 1:15 due to duplicate-dominant segregation. Known low-amylose genotypes such as ‘Ames 21 926’ and ‘Baer’ will also be screened for GBSSI mutations (Lindeboom et al., Reference Lindeboom, Chang, Tyler and Chibbar2005). Such mutations will be especially important for breeding waxy quinoa because of the general lack of acceptance of transgenic crops in several Latin American countries.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S1479262115000076
Acknowledgements
The authors gratefully recognize the USDA-ARS, NPGS and David Brenner (Ames, North Central Regional PI Station) for their contribution of germplasm to this study. They also thank the Instituto Nacional de Investigaciones Nucleares (ININ) for Chenopodium seeds provided. Dr Cepeda-Cornejo was supported at Brigham Young University with a Postgraduate Fellowship provided by the National Council of Science and Technology of Mexico (Consejo Nacional de Ciencia y Tecnologia, CONACYT) of the Mexican government. They also acknowledge supplementary internal support from Brigham Young University and a grant from the Doug Holmes Family Foundation. Carmen Gutierrez Cornejo helped with the design of figures. Earl Hansen provided logistic support at the Brigham Young University greenhouse. They also gratefully acknowledge the contributions of Hailey Unice, Ivan Arano and Evan Braithwaite, undergraduates who also worked on this project in the laboratory.
