Supplemental Data (Molecular Plant Letter to the Editor)

Supplemental Data (Molecular Plant – Letter to the Editor)

Glossy15 plays an important role inthe divergence of the vegetative transition between maize and its progenitor, teosinte

Dingyi Xu1,3, Xufeng Wang1,3,Cheng Huang1, Guanghui Xu1, Yameng Liang1, Qiuyue Chen1, Chenglong Wang1, Dan Li1, Jinge Tian1, Lishuan Wu1, Yaoyao Wu1, Li Guo1, Xuehan Wang1, Weihao Wu1, Weiqiang Zhang2, Xiaohong Yang1 and Feng Tian1*

1National Maize Improvement Center of China, Beijing Key Laboratory of Crop Genetic Improvement, Laboratory of Crop Heterosis and Utilization, China Agricultural University, Beijing 100193, China

2Department of Crop Physiology and Cultivation, China Agricultural University, Beijing 100193, China

3These authors contributed equally to this article

* Correspondence: Feng Tian ()

Contents:

Supplemental Materials and Methods

Supplemental Figures

SupplementalFigure 1. Phenotypic differences in total leaf number and days to anthesis between NILmaize and NILteosinte for qVT9-1.

SupplementalFigure 2. Transcriptional activation assay of Gl15protein in yeast.

SupplementalFigure 3. Subcellular localization of Gl15 in epidermal cells of Nicotiana benthamiana leaves.

Supplemental Tables

Supplemental Table 1. Maize and teosinte materials used for the LLEW investigation

Supplemental Table 2.QTLs for LLEW identified in the maize-teosinte BC2S3 RIL population

Supplemental Table 3.Significant association with the LLEW detected in Gl15 region

Supplemental Table 4.Sequence statistics for the region flanking SNP2154 (F4 region) in maize and teosinte lines

Supplemental Table 5. Molecular markers used for qVT9-1 fine mapping

Supplemental Table 6.Primer sequences used for sequencing Gl15 in the maize association panel

Supplemental Table 7. Teosinte accessions used for F4 region sequencing

Supplemental Table 8.Primer sequences used for quantitative real-time PCR

Supplemental Table 9.Primer sequences used for plasmid construction

Supplemental References

Supplemental Materials and Methods

Plant Materials and Phenotyping

To evaluate the difference in the vegetative transition between maize and teosinte, a panel of 50 diverse maize inbreds and 13 teosintes (Supplemental Table 1) were planted in a winter nursery in Sanya, Hainan province, China in 2013. The 50 maize inbreds were randomly chosen from an association panel that consisted of 517 diverse maize inbred lines (Yang et al., 2011), and the 13 teosintes were obtained from the National Plant Germplasm System (NPGS). The 50 maize inbreds and 13 teosintes were randomly planted. Each line was planted in a single-row plot with 15 plants per row, 25 cm between plants within each row and 50 cm between rows. Following a previous method (Foerster et al., 2015), the last leaf with epicuticular wax (LLEW) was scored as an indicator trait measuring the timing of the juvenile-to-adult transition in maize and teosinte lines. For each line, the middle five plants in the plot were measured for the LLEW. Because the earliest emerging leaves senesce and are no longer visible when scoring the uppermost leaves, the 5th and 10th leaves of the middle five plants were marked for the maize lines. For the teosinte lines that are highly tillering, each emerging leaf from the main stem was marked.

A large population of 866 maize-teosinte BC2S3 recombinant inbred lines (RILs)population was obtained from the Maize Genetics Cooperation Stock Center (Maize COOP) and used to map the QTLs controlling the timing of the juvenile-to-adult vegetative transition. This population was derived from a cross between W22, a typical temperate maize (Zea mays ssp. mays) inbred line, and CIMMYT accession 8759, a typical accession of teosinte (Zea mays ssp. parviglumis). The 866 maize-teosinte BC2S3 RILs were previously genotyped using 19,838 SNPs (Shannon, 2012). Owing to the large sample size and high-density markers, this population has been used as a powerful tool for trait dissection and gene cloning (Huang et al., 2016; Hung et al., 2012; Lang et al., 2014; Li et al., 2016; Lin et al., 2012; Wills et al., 2013). The maize-teosinte BC2S3 RILs were planted at Shangzhuang Experimental Station of China Agricultural University, Beijing (39.9°N, 116.4°E), China, in the summers of 2012 and 2013. The field trials were the same as those described by Li et al. (2016) and Xu et al. (2017), who measured the leaf number, flowering time and tassel traits in the population. Briefly, in each trial, the 866 BC2S3 RILs were grown according to an augmented incomplete randomized block design that has been widely used to phenotype large populations (Buckler et al., 2009; Huang et al., 2016; Kump et al., 2011; Li et al., 2016; Tian et al., 2011). Two maize inbred lines, W22 and Mo17, were randomly inserted in each incomplete block as controls. Each line was grown in a three-row plot in 2012 and a single-row plot in 2013. For accurately counting the number of leaves, the 5th and 10th leaves of the middle five plants of each plot were marked and scored for the LLEW.

Following a previously described approach that minimizes the effects of environmental variation (Buckler et al., 2009; Huang et al., 2016; Kump et al., 2011; Li et al., 2016; Tian et al., 2011; Xu et al., 2017), the phenotypic data collected over two years were ﬁtted with a linear mixed model that included the effects for the year, the genotype, incomplete blocks, the ﬁeld range and the ﬁeld row. For each line, the best linear unbiased predictor (BLUP) was predicted with SAS (v.9.2; SAS Institute Inc., Cary, NC, USA) and used as the phenotype in subsequent QTL mapping.

QTL mapping

The QTL mapping was performed as previously described (Huang et al., 2016; Li et al., 2016; Shannon, 2012; Xu et al., 2017). Brieﬂy, a modiﬁed version of R/QTL software (Broman et al., 2003) that considers the BC2S3 pedigree of the RILs was used for the QTL mapping (Shannon, 2012). A multiple QTL mapping procedure was employed to identify the QTLs affecting the LLEW. A total of 1000 permutation tests were first performed to determine the P<0.05 logarithm of odds (LOD) signiﬁcance threshold level for claiming QTLs. The scanone command was then used to identify an initial QTL list for subsequent multiple QTL ﬁtting (Broman et al., 2003). A drop-one ANOVA was used to test the multiple QTL model; only QTLs with a LOD score greater than the threshold and an ANOVA P-value <0.05 were retained in the model. To further reﬁne the position of each QTL in the model, the reﬁneqtl command that employs the likelihood ratio test was used to measure the improvement of the model. Finally, the addqtl command was used to search for additional QTLs. The ANOVA and reﬁneqtl procedures were repeated to determine whether the newly added QTLs could improve the model. The entire process was repeated until signiﬁcant QTLs could no longer be added. Using the ﬁtqtl function, the total phenotypic variation explained by all QTLs was calculated by fitting all QTLs in the model. The percentage of phenotypic variation explained by each QTL was estimated using the ﬁtqtl function that employs a drop-one ANOVA. A 2-LOD support interval was used as the conﬁdence interval for each QTL.

Fine mapping of qVT9-1

The fine mapping strategy was the same as that employed in previous studies (Huang et al., 2016; Hung et al., 2012; Li et al., 2016; Xu et al., 2017). Briefly, a large F2 population derived from a heterogeneous inbred family (HIF) that was heterozygous only at qVT9-1 was planted in a winter nursery in Sanya, Hainan Province in 2013. 35 recombinants were identified from theF2 populationcontaining 1500 plants, using two markers flanked the target region (M1 and M5). Other additional markers were then developed to resolve the breakpoints of the recombinants. Based on the geneotypes across the qVT9-1 region, these recombinants were grouped into six types (Figure 1D). A within-family comparison strategy was used to delimit the region of qVT9-1. Within each recombinant-derived F3 family, homozygous recombinant (HR) and homozygous nonrecombinant (HNR) plants were identified using markers. The phenotypic difference between the HR and HNR plants within each family was determined. If a significant phenotypic difference was observed between the HR and HNR plants, the parental F2 recombinant was heterozygous for the target QTL; otherwise, the recombinant was homozygous for either parent. A t-test with the Bonferroni correction for multiple testing was used to determine the significance of the phenotypic differences between the HR and HNR plants (P<0.01). By integrating the QTL location information from all recombinants, the causal QTL region was delimited.

Near-isogenic lines (NILs) that were homozygous for the W22 allele or teosinte allele across the qVT9-1 region (designated NILmaize and NILteosinte, respectively) were developed from the HIF-derived F2 population. NILmaize and NILteosinte were used to examine the phenotypic effects of qVT9-1 and profile the Gl15 expression during the early shoot development. To determine the difference in the onset of the adult vegetative phase between NILmaize and NILteosinte, the production of the leaf macrohairs along the leaf stages was investigated. At each leaf stage, the leaf was dissected into three parts (tip, middle, base). For each part, 10 1-cm2 slices were collected to count the number of macrohairs under a Leica EZ4HD (Germany) microscope. Additionally, the days to anthesis (DTA) and total leaf number at maturity were scored for NILmaize and NILteosinte. A t-test was used to determine the significance of the phenotypic difference between NILmaize and NILteosinte at P<0.01.

Sequencing the Gl15 region and association analysis

To identify the functional variants at Gl15 that control the natural variation in the timing of the juvenile-to-adult vegetative transition, we sequenced the Gl15 region and conducted an association analysis in a global germplasm collection that consisted of 517 diverse maize inbreds (Yang et al., 2011). Following a similar field experimental design as the maize-teosinte BC2S3RILs population described above, this association panel was grown in the Wuqiao Experimental Station of China Agricultural University, Hebei province, China, in the summers of 2014 and 2015 according to an augmented incomplete randomized block design. Each line was grown in a single-row plot in both years. The 5th and 10th leaves of the middle five plants of each plot were marked and scored for the LLEW. The best linear unbiased predictor (BLUP) was estimated with SAS (v.9.2; SAS Institute Inc., Cary, NC, USA) for each line and was used as the phenotype in the association analysis.

According to the B73 reference sequence (B73 RefGen_v3) ( six pairs of primers were designed using the software PRIMER 3.0 ( (Supplemental Table 6) to amplify a 6-kb genomic region around Gl15, including a 2.2-kb coding sequence, 1.6-kb upstream sequence and 2.2-kb downstream sequence, in a maize diverse panel containing 517 inbred lines. Sequencing reactions were performed on the PCR products in both directions. Multiple sequence alignments were performed using BIOEDIT (v.7.0.9.0; North Carolina State University, Raleigh, NC, USA) and manually edited if necessary. Polymorphic sites (SNPs and InDels) were extracted, and thelevels of LD between sites were calculated using TASSEL 2.1.0 (Bradbury et al., 2007). The associations between sequence variants with MAF≥0.05 and the LLEW were tested using a model that corrects for familiar population structure (Yu et al., 2006), implemented in TASSEL 2.1.0 (Bradbury et al., 2007). Sequence variants that passed the P<0.01 significance threshold after the Bonferroni multiple test correction were considered to have significant associations.

Molecular population genetic analyses

To examine the selection signals at Gl15, we sequenced a 1-kb region flanking SNP2154 (F4 region, Supplemental Table 6) in diverse teosintes (Supplemental Table 7). PCR products were cloned into the pEASY-T5 vector (TransGen, Beijing, China), and at least three clones were sequenced for each entry. The nucleotide diversity of the teosintes and maize lines in the 1-kb region flanking SNP2154 was compared. The number of segregating sites (S), nucleotide polymorphisms (θ) (Watterson, 1975), nucleotide diversity (п) (Tajima, 1983), and Tajima’s D statistic (Tajima, 1989) were estimated using DnaSP v.5.10.00 (Librado and Rozas, 2009). The retention of the nucleotide diversity, which is the relative ratio of п in maize to п in teosinte, was calculated for the sequenced region. To evaluate whether the observed loss of genetic diversity in maize relative to that in teosinte could be explained by a domestication bottleneck alone, coalescent simulations that incorporated the domestication bottleneck (Eyre-Walker et al., 1998; Tenaillon et al., 2004) were performed using the MS program (Hudson, 2002) following a previously described procedure (Tian et al., 2009; Xu et al., 2017). All parameters in the model were assigned to previously established values (Tian et al., 2009; Wright et al., 2005). The severity of the maize domestication bottleneck (k), which is the ratio of the population size during the bottleneck (Nb) to the duration of the bottleneck (d), was 2.45 (Tian et al., 2009; Wright et al., 2005). The population mutation and population recombination parameters were estimated from the teosinte sequences of the sequenced region. A total of 10,000 coalescent simulations were performed. Significant deviations from the expectations under a neutral domestication bottleneck indicate a selection in the examined region.

RNA sampling and Gl15 expression profiling

NILmaize and NILteosinte for qVT9-1 were planted in a greenhouse under 16 hours of daylight. A previous study has shown that Gl15 expression is found mostly in the shoot apex and is limited to the first 20 days after sowing (DAS) (Lauter et al., 2005). We therefore performed expression profiling on shoot apices collected between 8 and 21 DAS. The basal 1.5-cm part that enclosed the shoot apical meristem was kept and used for RNA isolation. At each sampling stage, at least five plants from each NIL were sampled and pooled as a biological replicate, and four biological replicates were included for each sample combination. The collected samples were instantly frozen in liquid nitrogen and stored at -80°C until the RNA isolation. The total RNA was isolated and purified using RNAprep Pure plant kits (TIANGEN BIOTECH). First-strand cDNA synthesis was synthesized using TransScript® One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGene Biotech). The relative Gl15 transcript abundance was estimated via quantitative real-time PCR using the maize ZmTubulin1 gene as the internal control (Supplemental Table 8). Three technical replicates were performed for each reaction. qRT-PCR was performed on ABI 7500 (Applied Biosystems) using the SYBR Premix Ex Taq II kit (Takara). The comparative CT (2-∆CT) method (Schmittgen et al., 2008) was used to quantify the Gl15 relative expression level.

Transcriptional activation assay in yeast

The yeast strain YRG-2 containing the HIS3 and lacZ reporter genes was used as the assay system (Stratagene, USA). The coding sequences of Gl15 from the maize and teosinte parents of the BC2S3 population were amplified (Supplemental Table 9) and inserted into the pBD-GAL4 plasmid (Stratagene, USA) at the EcoRI and SalI sites to generate vectors pBD-Gl15maize andpBD-Gl15teosinte. The pBD-GAL4 plasmid was used as the negative control and the pGAL4 vector as the positive control. Theseplasmids were independently transformed into the yeastYRG-2 strain.The transformed yeast cells were grown on SDmedium lacking threonine, or lacking threonine and histidine, for3 d at 30°C. Quantitation assay of transcriptional activation activity was performed using o-nitrophenyl β-D-galactopyranoside (ONPG) to test β-galactosidase activity as described in the yeast protocols handbook (PT3024-1; Clontech).

Subcellular localization of Gl15

The coding sequences of Gl15 from the maize and teosinte parent without stop codon were amplified (Supplemental Table 9) and inserted into a modified pCambia1300-GFP plasmid to generate the Gl15maize-GFP and Gl15teosinte-GFP fusion proteins under control of 35S promoter. TheAgrobacterium containing these constructs were resuspendedin infiltration buffer and infiltrated into 5-week-old leaves ofNicotiana benthamiana. After infiltration, plants were placed at 24°C for50 h before GFP observationunder a Zeiss LSM 710confocal microscope.

Protoplast transient expression assay

For transient expression assay in maize protoplasts, the expression vectors were constructed as follows: a microRNA172precursorwas amplified (Supplemental Table 9) and fused into the pGreenII 62-SK at EcoRI and XhoI sites, forming the effector vector p35S::miR172. A commercial dual-luciferase assay vector, pGreenII 0800-LUC, was modified to allow expression of both the mini-promoter controlling the expression of the firefly luciferase (LUC) reporter gene and aninternal control promoter (CaMV 35S) regulating the expression of the Renilla luciferase (REN) reporter gene. We cloned the full length of the 3’-UTR of the two parental alleles based on the B73 reference annotation into XbaI site at the downstream of the LUC gene, forming the reporter vector miniPro::LUC-3′UTRmaize and miniPro::LUC-3′UTRteosinte. All of the constructs were conﬁrmed by sequencing prior to usage in expression assays.

Seeds of B73 inbred line were germinated in pots with nutrient soil and vermiculite mixed 1:1 (v/v) under normal condition (16h light: 8h dark; 28℃) for 3 days, and then moved into the dark for 9 days with same temperature. The maize mesophyll protoplasts wereisolated and transfectedfor transient expression assay using a modified protocol. Briefly, the 2th leafof an etiolated maizeseedling was cut into strips and then incubated in a solution containing1.5% cellulose R10, 0.4% macerozyme R-10, 0.8 M mannitol, 0.2 M MES,1 M CaCl2, 5 mM β-mercaptoethanol and 0.1% BSA, pH 5.7, for 3-4 h in the dark with gentle shaking.Released protoplasts were collected and washed withW5 solution (2 M NaCl, 1 M CaCl2, 2 M KCl and 0.2 M MES, pH 5.7)beforetransfection. The control (pGreenII 62-SK) and effector (p35S::miR172) vectors were co-transfected with each fusion reporter (miniPro::LUC-3’UTRmaize and miniPro::LUC-3’UTRteosinte) in protoplastsat a 1:1 ratio (50 μgof vector in total for 500 μl of protoplasts),using the PEG-mediatedDNAdelivery method (Yoo et al., 2007). After cultivation at 25°Cin the dark for 16-18 h, the transformed maize protoplasts were collected and LUC activities were assayed using a Dual Luciferase Reporter Assay System (Promega). The relative ratio of LUCto RENactivity(LUC/REN) was calculated to normalize each assay.

Supplemental Figures

SupplementalFigure 1.Phenotypic differences in total leaf number and days to anthesis between NILmaize and NILteosinte for qVT9-1. N.S., no significant difference detected.