Introduction

Waxy maize, also known as sticky maize, has high economic, nutritional, and processing value1,2. The starch in the endosperm of waxy maize is nearly 100% amylopectin, which confers the sticky quality to maize grains. It is mainly consumed as food in Asia, and is also an important ingredient in the textile and paper industries. Waxy maize was first discovered in China in 1908, and was later reported in other locations of Asia2. The results of several studies suggest that the southwest region of China, particularly Yunnan Province and its surrounding areas, is the central origin of Chines waxy maize3,4. In recent decades, many adapted maize lines have been developed for hybrid seed production by different selection methods, but the relationships among these waxy maize lines are unclear.

The glutinous genotypes of waxy maize are null mutations of the waxy gene, which encodes the granule bound starch synthase (GBSSI) that is necessary for amylose synthesis5,6,7. The wild-type waxy gene in maize is 3.93 kb long, located on chromosome 9, and composed of 14 exons. Insertions and deletions in the DNA sequence are the main types of waxy mutations in maize8. Globally, more than 50 mutations in the waxy gene have been characterized at the molecular level8. In Chinese waxy maize, the two deletion mutations wx-D7 and wx-D10 (a 30-bp deletion in the seventh exon and a 15-bp deletion in the tenth exon, respectively) are the main waxy alleles9,10. Transposable element insertions are another type of waxy gene mutation8. Transposons can be divided into DNA transposons and RNA transposons according to their transposition mechanism. Members of the DNA-transposon family can jump from one gene to another in a cut-and-paste fashion to produce unstable mutants11,12. Among the identified mutations in the maize waxy gene, wx-m9, wx-m5, wx-B3, wx-m1, wx-B4, wx-m6, and wx-m7 are insertion mutations of Ac/Ds transposable elements, wx-m8 is an insertion mutation of the dSpm element, and wx-844 is a mutation caused by the En/Spm element8,11. The Ac/Ds, dSpm, and En/Spm elements are all members of the DNA-transposon family12 and their waxy alleles are important resources for transposon research. RNA transposons move within the genome using a copy-and-paste mechanism via reverse transcription of an RNA intermediate3. Among the known mutant alleles of the waxy gene in maize, wx-stonor, wx-B5, wx-G, wx-M, wx-I, wx-K, wx-Cin4 and wx-Rina are RNA transposon insertion mutations3,8. RNA transposons result in stable mutants and can be used as markers to identify waxy mutation loci. However, there are still many waxy maize lines with unknown waxy alleles.

There is wide genetic diversity at the waxy locus among waxy maize accessions. Previous studies on waxy diversity have shown that different waxy mutants have independent origins10. Compared with non-glutinous maize, Chinese waxy maize shows much narrower nucleotide diversity at the waxy locus, suggesting that there has been strong selection in the waxy genomic region during waxy maize breeding13,14. An association analysis between allelic variations of waxy and starch physicochemical properties showed that waxy allelic variation affects the gel consistency, gelatinization temperature, and pasting viscosity properties of rice starch, implying that certain waxy alleles have been favored for grain quality improvement15,16. Therefore, comparison of waxy sequence variation among maize germplasm accessions not only provides insights into selection at the maize waxy locus during domestication, but also can highlight mutations in the waxy gene that will be useful for the germplasm utilization and quality breeding of waxy maize16,17.

In this study, we collected waxy maize breeding accessions from Jilin province, Shanxi province, and Beijing in China, as well as from Korea. First, we used simple sequence repeat (SSR) markers to study the genetic diversity of the waxy maize inbred lines. Then we sequenced the waxy genes from the first to the fourteenth exon to analyze sequence variations and the phylogenetic relationships among waxy maize lines. Additionally, a newly identified allele of the waxy gene in waxy maize was characterized.

Results

Genetic analyses of waxy maize

The genetic diversity of the 200 waxy maize inbred lines (Supplementary Table S2) was evaluated using SSR markers. In total, 458 alleles were found at the 40 SSR loci, with a range of 2 to 25 alleles per marker. The average number of alleles per marker locus across genotypes was 11.45, about double that obtained by Zheng et al.14 using 20 SSR markers and 165 accessions. The polymorphism information content (PIC) values for the 40 SSR loci ranged from 0.17 to 0.89 (average, 0.7). The genetic similarity coefficient was analyzed using the SSR data. The genetic similarity coefficient of 200 waxy maize inbred lines ranged from 0.03 to 0.95, with an average of 0.31, and 86.79% of them were less than 0.45 (Fig. 1a). On the basis of similarities of SSR data, a cluster analysis of the 200 waxy maize inbred lines was performed using the neighbor-joining method18. The cluster analysis grouped the 200 waxy inbred lines into three main groups: group A included 63 accessions, group B included 59 accessions, and group C included 78 accessions (Fig. 1b). These results show that there was wide genetic diversity among the tested waxy maize accessions.

Figure 1
figure 1

Genetic similarity coefficient and cluster analysis of 200 maize accessions using SSR data. (a) Genetic similarity coefficient frequency distribution. The genetic similarity coefficient was calculated by SSRAnalyzer V1.0 (Software copyright registration number: 2018SR003610). (b) Cluster analysis of waxy lines. Cluster analysis based on allele identity was carried out using PowerMarker V3.25 with the neighbor-joining method. Different colors of taxon names represent different mutant alleles in waxy gene. Different colors of subtree markers represent different origin regions of waxy maize. Different colors of branch lines represent different waxy maize groups. wx-hAT: waxy maize with wx-hAT mutant allele; wx-D7: waxy maize with wx-D7 mutant allele; wx-D10: waxy maize with wx-D10 mutant allele; wx-124: waxy maize with wx-124 mutant allele; Other: waxy maize had other mutation in waxy gene, which was different from wx-hAT, wx-D7, wx-D10 and wx-124; Not analyzed: the sequence of these waxy maize lines were not obtained; Beijing, China: waxy maize originated from Beijing city in China; Jilin, China: waxy maize originated from Jilin province of China; Shanxi, China: waxy maize originated from Shanxi province of China; Korea: waxy maize originated from Korea; Group A: waxy maize classified into group A; Group B: waxy maize classified into group B; Group C: waxy maize classified into group C.

Full size image

The genetic distances of waxy maize inbred lines in group A, B and C ranged from 0.11 to 0.97, 0.05 to 0.94, 0.06 to 0.96, with the average values of 0.68, 0.66 and 0.67, respectively. All the three groups contained waxy maize inbred lines originated from Beijing and Jilin province of China, which indicated that waxy maize with different genetic background had been widely used in China, which was consistent with the rapid development of waxy maize breeding and industry in China in recent years. It is worth mentioning that the waxy maize inbred lines originated from Shanxi Province of China and Korea only existed in group C, which indicated that the waxy maize inbred lines from Shanxi Province of China had similar genetic background, and those from Korea had similar genetic background (Fig. 1b). As expected, waxy maize inbred lines with similar pedigree were clustered in the same group. For example, waxy maize inbred lines JYN3, JYN4, JYN5, JYN6, JYN7 and JYN9 all had similar pedigree and were clustered in group A; the waxy maize inbred lines DN1, DN2, DN3 and DN4 had similar pedigree and were clustered in group C (Supplementary Table S2 and Fig. 1b).

Nucleotide variation at waxy locus in waxy maize

Nucleotide sequences from the first to the 14th exon of the waxy genes were determined using four pairs of primers (Supplementary Fig. S1 and Table S1). We examined the DNA nucleotide variations in an approximately 3523-bp region of waxy loci in 167 waxy and 14 flint maize accessions (Supplementary Data File S1). In this region, the variable nucleotide sites of 169 waxy maize and 14 flint maize were 57 and 87, respectively (Table 1). The genetic variation in waxy gene was compared among the three different waxy maize groups and the flint maize. All three waxy maize groups and all waxy maize in three groups showed a minimum level of genetic variation at the waxy locus. The estimate of nucleotide diversity for waxy maize accessions in group A, group B, group C and all waxy maize in three groups was 0.00173, 0.00097, 0.00228 and 0.00150, respectively, indicating that the genetic diversity of the waxy locus differs among waxy maize populations with different genetic backgrounds. Apparent nucleotide diversity was observed at the waxy locus in flint maize (8.4 fold) than in waxy maize. Consistently, the values of S (number of polymorphic sites), and K (average number of pairwise nucleotide differences) were lower in waxy maize accessions than in non-glutinous maize accessions (Table 1). The significant reduction in diversity at the waxy locus in waxy maize suggests that modern waxy maize has experienced a genetic bottleneck during its domestication.

Table 1 Summary of nucleotide diversity for waxy gene within maize taxa.
Full size table

The Tajima’s D and Li & Fu’s D* and F* values were calculated to test the deviation from the neutral equilibrium model. All three tests identified negative selection at the waxy locus in waxy maize populations, but not in flint maize, suggesting that there has been strong selection acting on waxy maize accessions (Table 1).

The haplotypes of waxy gene in waxy maize and flint maize were calculated using DnaSP19 software and the results were shown in Table 2. Thirty haplotypes were detected in 167 waxy maize accessions, among which 117 lines shared haplotype Hap_19. Eight haplotypes were detected in 14 flint maize accessions.

Table 2 Haplotype of waxy gene in studied 167 waxy maize and 14 flint maize.
Full size table

Phylogenetic analysis based on sequence polymorphisms

We obtained the sequences of the 3523-bp region at the waxy loci in 167 waxy and 14 flint maize. Waxy sequence data for eight wild relatives of maize and seven landraces from Southwestern China were downloaded from the GenBank database. Based on these sequences, a phylogenetic tree including waxy maize inbred lines, flint maize, waxy maize landraces and their relatives was constructed by the neighbor-joining method20. According to the tree, five wild relatives of maize formed a branch that was basal to five separate branches. All five branches contained waxy maize and flint maize, indicating that a number of glutinous maize accessions may have been developed from domesticated non-glutinous maize. Qbviously, maize with the same mutation allele clustered together, and waxy maize in different branches carried different mutation alleles. Five waxy maize inbred lines and seven waxy maize landraces formed a branch, and these maize accessions carried the wx-D10 allele in waxy gene. In another branch, eight waxy maize with the wx-hAT allele formed a subgroup, and then clustered with one waxy maize harboring the wx-124 mutant allele. The other two branches contained two and four waxy maize inbred lines, respectively. No insertion or deletion mutations were detected in the amplification region of waxy gene in these maize lines, suggesting that they had other mutation alleles in other regions of the waxy gene. Furthermore, the remaining 147 waxy inbred lines carrying the wx-D7 mutant allele clustered together and formed an independent branch, which was significantly separated from waxy maize carrying wx-D10, wx-124 and wx-hAT mutant alleles (Fig. 2).

Figure 2
figure 2

Phylogenetic analysis of maize accessions based on waxy gene. The neighbor-joining phylogenetic tree based on the Kimura 2-parameter model was constructed with MEGAX64 software using waxy gene sequence data with 1000 bootstrap replicates to assess tree reliability. Different colors of branch lines represent different groups. Different colors of taxon names represent waxy maize inbred lines carrying different waxy gene mutation alleles, and flint maize, wild relatives of maize, as well as landraces from Southwestern China. Subtree markers pointed out landraces from Southwestern China and waxy maize inbred lines from Shanxi province of China. wx-hAT: waxy maize with wx-hAT mutant allele; wx-D7: waxy maize with wx-D7 mutant allele; wx-D10: waxy maize with wx-D10 mutant allele; wx-124: waxy maize with wx-124 mutant allele; Other mutation: waxy maize had other mutation in waxy gene, which was different from wx-hAT, wx-D7, wx-D10 and wx-124; Shanxi, China: waxy maize inbred lines originated from Shanxi province of China; Southwestern China: landraces from Southwestern China.

Full size image

Identification of accessions with novel waxy genotype

After sequencing, four types of waxy mutation were identified among the waxy maize accessions: 147 accessions carried the wx-D7 allele with a 30-bp deletion in the seventh exon–intron region; five maize accessions carried the wx-D10 allele with a 15-bp deletion in the 10th exon region; one maize accession carried the wx-124 allele with a 125-bp insertion in the seventh exon; and eight maize accessions carried an allele with a new insertion mutation. This mutation was identified as a 2.2 kb insertion in the middle of the third exon of waxy gene, which would lead to abnormal gene coding (Fig. 3). The conservative domain of waxy gene was analyzed in Pfam database (https://pfam.xfam.org/search/sequence), and found that the 2.2 kb insertion mutation was located on the conserved starch synthase catalytic domain of waxy gene. The amylopectin content analysis showed that all eight of these accessions had a high amylopectin content (> 94.5%) in grain starch, like that in wx-D7 mutant maize lines (Supplementary Table S3). Consistently, the GBSS activity in seeds of these accessions (19.9–28.1 nmol/min/g) was remarkably lower than that in wild type flint maize (66.7–79.3 nmol/min/g) (Supplementary Table S4).

Figure 3
figure 3

Mutation types of waxy gene. (a) Wild-type waxy gene is about 4.5 kb long and contains 14 exons numbered e1 to e14. Important gene cassettes in promoter region as well as the start and stop codons are indicated. Arrows above schematic of waxy gene mark sites of insertion mutations; lines below mark sites of deletion mutations. wx-hAT is a new mutation identified in this study. (b) Detection of wx-hAT mutation by 1% agarose electrophoresis of PCR products. M, marker; lane 1–9, homozygous-type wx-hAT; lane 10–12, wild-type waxy gene. (c) Structural features of wx-hAT waxy gene mutation. wx-hAT is 2286 bp long, and contains a partial-length of the hAT element (grey box), 8-bp TSD (red box), 3-bp (yellow box) and 9-bp (green box) TIRs.

Full size image

As shown in Fig. 1b, waxy maize in the same group had different origins and carried different waxy mutation alleles based on cluster analysis using SSR data. Waxy maize from the same origin possessed different waxy mutation alleles. Therefore, based on cluster analysis, there was no correlation among the groups, origins and waxy mutation alleles in waxy maize in this study.

Molecular characteristics of wx-hAT

The newly identified insertion mutation was 2286 bp in length and was inserted after the 43rd nucleotide in the third exon of the waxy gene (Fig. 3a,b and Supplementary Data File S1). Transposon analysis using the CENSOR program21 indicated that the 2.2-kb insertion contained part of the hAT transposon with sequences ranging in length from 813 to 1558 bp (Fig. 3c). We named this waxy allele wx-hAT. Further analyses revealed that wx-hAT had a 5′-CAG-3′ terminal inverted repeat (TIR) and a 5′-GGCGGATCT-3′ near-terminal inverted repeat, generated a 5′-AGTACAAG-3′ target site repeat (TSD). The 3-bp TIR was flanked by the 8-bp TSD and the 3-bp and the 9-bp TIRs were separated by one nucleotide (Fig. 3c). BLAST searches in the MaizeGDB database revealed that the sequences from 9 to 1559 bp and from 1561 to 2286 bp of wx-hAT showed 99.03% (e-value = 0) and 99.59% (e-value = 0) identities to the regions of 27,081,986–27,083,536 bp and 27,084,569–27,085,294 bp on chromosome 4 (Supplementary Fig. S2), respectively. This result suggests that this transposon might have jumped from chromosome 4 to the waxy gene on chromosome 9. Those regions of chromosome 4 were not annotated.

Intragenic selection marker for recessive waxy gene

As shown in Fig. 3b, using the primers of E1-4F/E1-4R two amplicons 3.1 kb and 884 bp in length were amplified from waxy lines and flint lines, respectively. We anticipated that the mutation of the waxy gene in these waxy lines was caused by the insertion of the 2.2 kb fragment in its allelic dominant gene. However, the heterozygous type and the wild type of the wx-hAT allele could not be differentiated by analyzing fragments amplified using this set of PCR primers, because only the 884 bp fragment amplified from the wild-type waxy gene was amplified in the heterozygous type maize, which may due to the effect of long fragment amplification being affected by the priority amplification of short fragment. A forward primer waxyF2 was designed on the basis of the 2.2 kb insertion sequence (Supplementary Table S1). Using the three primers E1-4F, E1-4R and waxyF2, a 608-bp and an 884-bp fragment were amplified for the homozygous mutant waxy gene and the wild type waxy gene, respectively, while both the 608-bp and the 884-bp fragments were amplified for the heterozygous-type waxy gene (Fig. 4). Therefore, this set of three primers functioned as a molecular marker for selection of the 2.2-kb insertion mutant of the waxy locus. This molecular marker was able to distinguish among maize lines with wild-type, homozygous mutant-type, and heterozygous-type waxy genes.

Figure 4
figure 4

Molecular marker detection of wx-hAT mutation by 1.5% agarose electrophoresis of PCR products. M, marker; lane 1–6, SKN5, JN1, HN2, YN1M, 16–585, 80,482, and these waxy maize lines carried the homozygous-type wx-hAT; lane 7–12, SKN5 × B73, JN1 × B73, HN2 × B73, YN1M × B73, 16–585 × B73, 80,482 × B73, and these lines carried the heterozygous-type waxy gene; lane 13–15, Jing2416, MC01, Jing 464, and these lines carried the wild-type waxy gene.

Full size image

Discussion

Because of the popularity of glutinous maize in China, waxy maize lines are frequently selected by maize breeders. An abundant of waxy maize germplasm has been obtained through decades of waxy maize breeding. The aim of our study was to reveal the genetic diversity and relationships among modern waxy maize inbred lines, as well as to identify new mutant alleles of the waxy gene in maize.

In previous studies, SSR markers have been widely used for genetic analyses of date palm22, barley23, angelica gigas24 and maize14,25. In this study, 40 core SSR markers with proven performance for maize genotype identification26 were used to genetically analyze 200 waxy maize accessions. The high average PIC value (0.7), high average number of alleles (11.45) and low average genetic similarity coefficient (0.31) are indicative of a high degree of genetic diversity among the waxy maize germplasm.

Compared with rice, maize shows a much higher level of sequence variation at the waxy locus8. To explore the genetic differentiation of the waxy gene in maize, the waxy gene sequences (3523 bp) from 167 waxy maize and 14 flint maize were compared and analyzed. The results show that the sequence diversity at the waxy locus in waxy maize is only 11.8% of that in flint maize (Table 1). This result is consistent with the findings of previous studies, i.e., that the DNA polymorphism of the waxy gene is much reduced in waxy maize10,13,14. Consistent the lower level of genetic diversity at the waxy locus in waxy maize than in non-glutinous maize, the neutral test revealed negative selection for the waxy gene in waxy maize, but not in non-glutinous maize (Table 1). Similar domestication selection in the waxy genomic region has also been detected in rice27,28,29, suggesting that the waxy gene in waxy crops has experienced a genetic bottleneck and strong artificial selection has acted on this locus during the improvement of waxy crops.

We conducted a phylogenetic analysis using the nucleotide sequences of the waxy gene in 167 waxy maize accessions, including five waxy maize accessions (Jing2, HN17, 1029, 80,453, 80,452) with the wx-D10 allele, one waxy maize accession with the wx-124 allele (SXBN4), eight waxy maize with the wx-hAT allele (SKN6, 80,482, SKN5, JN1, JN2, 16–585, YN1M and HN2), six waxy maize accessions with other waxy allele (SXBN1, SXBN3, BN9, 6013, 6003 and Zhonghang3M), and 147 waxy maize accessions with the wx-D7 allele. A previous study showed that the waxy maize with the wx-D10 genotype originated from the Yunnan-Guangxi region while that with the wx-D7 genotype originated from the Yangzi River region10. Consistent with these findings, waxy maize harboring wx-D10 formed a branch while waxy maize with the wx-D7 mutation was on a distinct independent branch in the tree (Fig. 2). Some waxy maize accessions had wx-124 and wx-hAT mutations and were located on a separate branch, suggesting that there may be additional origins of waxy maize.

The phylogenetic tree showed a clear genetic relationship among waxy maize, flint maize, waxy maize landraces and their wild relatives. Five wild maize relatives (parviglumis P1331783, PI1331786 and PI384061, and mexicana P1566683 and PI56685) formed a branch that was basal to all flint maize and waxy maize accessions, indicating that wild maize relatives (Mexicana and parviglumis) might be the ancestor of maize. Seven waxy maize landraces (CWM057, CWM056, CWM069, CWM052, CWM050) and five inbred lines (Jing2, HN17, 1029, 80,453, 80,452) clustered together first and then with one wild relative of maize (parviglumis M106) and two flint maize accessions (B73 and D9H) to form an independent branch. All waxy maize in this branch carried the wx-D10 allele. Next to this branch, eight maize inbred lines (SKN6, 80,482, SKN5, JN1, JN2, 16-585, YN1M and HN2) carrying the wx-124 allele clustered together first and then with one waxy maize inbred line (SXBN4) harboring the wx-124 allele, as well as six flint maize (Jing464, MC01, Jing724, HZS, C92 and Chang7-2) to form a separate branch. At the same time, far away from these two branches, two waxy maize (Zhonghang3M and 6003) were intermixed with three flint maize (Dan340, P178 and Qi319), and then clustered with two maize wild relatives (parviglumis P1331785 and mexicana P1566691) to form an independent branch. Four waxy maize (SXBN1, SXBN3, BN9 and 6013) and two flint maize (Ye478 and Zheng58) formed another branch. The remaining 147 waxy maize carrying the wx-D7 allele formed a separate branch with one flint maize (Jing2416) (Fig. 2). Waxy maize and flint maize were intermixed in each branch, which indicated that some waxy maize might be domesticated from flint maize. Moreover, waxy maize carrying the wx-D7 mutation was most abundant in the collected waxy maize inbred lines, suggesting that wx-D7 might be the main mutation type used in modern waxy maize breeding.

More than 50 waxy maize mutations had been described previously8. Previous studies identified two deletion mutations including wx-D7 and wx-D1010,13, together with three insertion mutations including wx-Cin411, wx-12411 and wx-Reina3 in Chinese waxy maize accessions. The wx-D7, wx-D10, and wx-124 mutations were also identified in our study, and wx-D7 was the main waxy mutant allele type among the accessions we studied. Furthermore, we found a new insertion mutation allele, wx-hAT, in eight of 200 waxy maize accessions. A truncated hAT element was found in the insertion sequence, so we named this new insertion mutation wx-hAT. The hAT element is 3182-bp DNA transposon containing a 14-bp TIR flanked by an 8-bp TSD30. Only 746 nucleotides of the hAT element were retained in wx-hAT allele. The whole wx-hAT sequence was 2286 bp long with a 3-bp TIR flanked by an 8-bp TSD as well as another 9-bp TIR near the terminal (Fig. 3c). The identification of a new allele of the waxy gene in this study enriches the collection of maize waxy alleles and will be useful for breeding and germplasm preservation.

Interestingly, by comparing sequences with the B73 reference genome, we detected 9–12 bp deletions in the gene sequence of 5′-CAGCACCAGCAGCAG-3′ in the second exon of the waxy gene in nine flint maize and all waxy maize accessions (Supplementary Data File S1). Flint maize accessions have this deletion mutation at the waxy locus, suggesting that this 15-bp nucleotide sequence is not required for the function of the wild-type waxy gene.

To detect wx-hAT effectively, we developed PCR molecular markers for this allele. Our results show that the wx-hAT mutation can be readily amplified using the primers E1-4Fb, E1-4Rb, and waxyF2. This set of primers amplify an 884-bp product for the wild-type waxy gene, a 608-bp product for the mutant type and two fragments (884 bp and 608 bp) for the heterozygous-type. Therefore, these allele-specific primers can be used as molecular markers to discriminate among mutant type, heterozygous-type, and wild type allelic genotypes.

Materials and methods

Germplasm accessions, amylopectin content and GBSS activity analysis

The research materials, 200 waxy and 14 flint maize inbred lines, were provided by the Maize Research Center, Beijing Academy of Agriculture and Forestry Sciences (BAAFS). The starch content in seeds was measured according to the National Standards of People’s Republic of China, GB 5009.9-2016. The amylopectin content of seeds was determined using a commercial amylose/amylopectin assay kit (Megazyme, Wicklow, Ireland). The activity of GBSS in harvested maize seeds was determined using a commercial GBSS assay kit (Ziker, Shenzhen, China).

SSR analyses

Genomic DNA was extracted from leaves using the CTAB procedure31. The core 40 SSR markers were developed by BAAFS26, and have been approved by the state of China for maize DNA fingerprinting. The PCR protocols and reaction conditions were those specified in the Sector Standard of Agriculture (NY/T 1432-2014). The core 40 SSR primers covered the entire maize genome26. The SSR primers were labeled with fluorescent dyes during amplification and the SSR DNA fragments were separated by capillary electrophoresis. The polymorphism information content (PIC) and allele number for each marker was determined using PowerMarker V3.25 software18. The genetic similarity coefficient was calculated by SSRAnalyzer V1.0 (Software copyright registration number: 2018SR003610). Cluster analysis based on allele identity was carried out using PowerMarker V3.25 with the neighbor-joining method18.

PCR and DNA sequencing

The waxy genes were sequenced from the first to the 14th exon. Primers specific for the waxy gene were designed using the Primer3 software (https://primer3.ut.ee/). Exons 1–4, exons 4–8, exons 8–12, and exons 12–14 were amplified using the primer pairs 1-4F/R (5′-AGAAGTGTACTGCTCCGTCC-3′ and 5′-AGAACCTGACCGTCTCGTAC-3′), 4-8F/R (5′-TACGAGACGGTCAGGTTC-3′and 5′- GGTAGGAGATGTTGTGGAT-3′)11, 8-12F/R (5′-GATTTCATCGACGGGTCTGT-3′and 5′-TCTGTCCCTCTCGTCAGGAT-3′)14 and 12-14F/R (5′-ATCCTGACGAGAGGGACAGA-3′ and 5′- CACCGAACAGCAGGGATTAT-3′)14, respectively (Supplementary Table S1). Phanta Max Super-Fidelity DNA Polymerase (Vazyme Biotech Nanjing, China) was used for PCR amplification. For PCR with the 1-4F/R primers, a PCR enhancer (Vazyme Biotech, Nanjing, China) was used with the following program: 5 min at 94 °C, 34 cycles of 60 s at 94 °C, 60 s at 59 °C, 2 min at 72 °C, and final extension for 10 min at 72 °C. The PCR enhancer was not added to the PCR reaction mixtures using the other primers. For PCRs with the 4-8F/R primers, the PCR amplification program was as follows: 5 min at 94 °C, 34 cycles of 60 s at 94 °C, 60 s at 55 °C, 2 min at 72 °C, and final extension for 10 min at 72 °C. For PCRs with the of 8-12F/R and 12-14F/R primers, the following program was used: 5 min at 94 °C, 34 cycles of 60 s at 94 °C, 60 s at 60 °C, 2 min at 72 °C, and final extension for 10 min at 72 °C. The PCR products were purified and sequenced by the TsingKe Biological Technology Co. Ltd. (Beijing, China). The gene sequences obtained in this study were shown in Supplementary Data File S1 and were also submitted to the NCBI GenBank database (MT863356–MT863536).

Sequence alignment analysis

The waxy gene sequences of the wild relatives (Zea mays ssp. Mexicana and Zea mays ssp. Parviglumis) of maize and waxy maize landraces had been studied in previous studies14, and they were downloaded from the GenBank database (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Muscle3.8.31_i86win32 software was used for multiple sequence alignment with manual refinement32.

Nucleotide variation analysis

Nucleotide variation analysis was performed with the DnaSP_v61203_x64 program19. The number of sites, number of polymorphic sites (S), haplotypes (h), number of haplotypes, nucleotide diversity (Pi), average number of pairwise nucleotide differences (K), minimum number of recombination events (Rm), and neutrality tests (Tajuma’s D*, Fu and Li's D* and F* test) were calculated using DnaSP software.

Phylogenetic tree reconstruction

A neighbor-joining phylogenetic tree based on the Kimura 2-parameter model was constructed with MEGAX64 software using waxy gene sequence data with 1000 bootstrap replicates to assess tree reliability20.

PCR marker development

The sequence of a newly identified insertion mutant of the waxy gene was used to develop a PCR molecular marker. The molecular marker was designed with Primer3 software. The primers were 1-4F (5′-AGAAGTGTACTGCTCCGTCC-3′), 1-4R (5′-AGAACCTGACCGTCTCGTAC-3′), and WaxyF2 (5′-AGTATTGCTTCTACCTGTGGCA-3′) (Supplementary Table S1). The PCR reaction conditions were as follows: 5 min at 94 °C, 34 cycles of 60 s at 94 °C, 60 s at 60 °C, 2 min at 72 °C, and final extension for 10 min at 72 °C. The PCR products were detected by 1.5% agarose gel electrophoresis.

Sequence analysis

Transposon prediction of the inserted sequence was performed using CENSOR (https://www.girinst.org/censor/index.php)21. The similarity between the inserted sequence and the B73 reference sequence was compared by conducting BLAST searches against the MaizeGDB database (https://www.maizegdb.org). Gene structure was drawn by Gene Structure Display Server (https://gsds.cbi.pku.edu.cn/)33.