Variation in maize chlorophyll biosynthesis alters plant architecture

Chlorophyll is a tetrapyrrole metabolite essential for photosynthesis in plants. The oil yellow1 (oy1) gene of maize encodes subunit I of Magnesium chelatase, the enzyme catalyzing the first committed step of chlorophyll biosynthesis. A range of chlorophyll contents and net CO2 assimilation rates can be achieved in maize by combining a semi-dominant mutant allele, Oy1-N1989, and cis-regulatory alleles encoded by the Mo17 inbred called very oil yellow1 (vey1). We previously demonstrated that these allelic interactions can delay reproductive maturity. In this study, we demonstrate that multiple gross morphological traits respond to a reduction in chlorophyll. We found that stalk width, number of lateral branches (tillers), and branching of the inflorescence decline with a decrease in chlorophyll level. Chlorophyll variation suppressed tillering in multiple maize mutants including teosinte branched1, grassy tiller1, and Tillering1 as well as the tiller number1 QTL responsible for tillering in many sweet corn varieties. In contrast to these traits, plant height showed a non-linear response to chlorophyll levels. Weak suppression of Oy1-N1989 by vey1B73 resulted in a significant increase in mutant plant height. This was true in multiple mapping populations, isogenic inbreds, and hybrid backgrounds. Enhancement of the Oy1-N1989 mutants by the vey1Mo17 allele reduced chlorophyll contents and plant height in mapping populations and isogenic inbred background. We demonstrate that the effects of reduced chlorophyll content on plant growth and development are complex and that the genetic relationship depends on the trait. We propose that growth control for branching and architecture are downstream of energy balance sensing.


Introduction
Land plants project their aerial parts into three-dimensional space to capture solar radiation and convert it into chemical energy via photosynthesis (Rabinowitch and Govindjee 1965;Govindjee et al. 2017). Energy is stored as carbon-carbon bonds and primarily transported as sugars from photosynthetically active tissues, such as leaves, to rapidly growing parts of the plant. The addition of biomass to young tissue is achieved by the mobilization of this photosynthetically fixed carbon (Yu et al. 2015). Thus, growth of the plant body is achieved by consuming excess sugars from photosynthesis. To sustain maximum growth rates, plants must coordinate growth and metabolism to prevent undershooting or overshooting the available resources (Lemoine et al. 2013).
Unsurprisingly, energy deficits from manipulation of photosynthesis can result in dramatically decreased growth.
One way to manipulate photosynthetic activity and energy resources in plants is by altering chlorophyll biosynthesis (Walker et al. 2017;Khangura et al. 2019a). Magnesium chelatase (MgChl) is a hetero-oligomeric enzyme complex that catalyzes the first committed step of chlorophyll biosynthesis by conversion of protoporphyrin IX (PPIX) into magnesium-PPIX (Walker and Weinstein 1991;Gibson et al. 1995). In maize, one of the subunits of the MgChl complex is encoded by the oil yellow1 gene (oy1; B73 RefGenV3: GRMZM2G419806; B73 RefGenV4: Zm00001d023536). This gene received its name because weak hypomorphic alleles result in a light green to yellow color reminiscent of vegetable oil (Eyster 1933;Sawers et al. 2006). We previously demonstrated that the semi-dominant Oy1-N1989 mutant was strongly modified by a QTL called very oil yellow1 (vey1) that is likely encoded by cis-acting regulatory polymorphisms at oy1 itself (Khangura et al. 2019a). The highly heritable vey1 QTL is a major modifier of chlorophyll contents in heterozygous Oy1-N1989/+ mutants but had no effect on chlorophyll accumulation in wildtype plants from either B73 x Mo17 biparental mapping populations or a maize association panel (Khangura et al. 2019a). The vey1 alleles are detectable as a cis-acting eQTL and generate allele-specific expression at oy1 (Li et al. 2013(Li et al. , 2018Waters et al. 2017;Khangura et al. 2019aKhangura et al. , 2019b. These expression level differences at oy1 are visible at the mRNA level in both mutants and wildtypes, but the impact on chlorophyll content was only visible in the presence of the Oy1-N1989 allele (Khangura et al. 2019a). This allelic interaction indicates the value of using mutants to uncover hidden genetic variation in maize. In the B73 x Mo17 populations, the vey1 allele from B73 (vey1 B73 ) was associated with higher expression of OY1, whereas the vey1 allele from Mo17 (vey1 Mo17 ) was linked with lower expression of OY1.
Using the Oy1-N1989 mutant allele, and the two vey1 modifier alleles from B73 and Mo17, we previously developed populations of maize to investigate the effects of this cryptic variant on chlorophyll content (Khangura et al. 2019a), net CO2 assimilation, and reproductive maturity (Khangura et al. 2019b). These three phenotypes showed high linear correlations with each other and were consistent with expression of the wildtype oy1 allele such that greater expression of the wildtype OY1 transcript resulted in suppression of the Oy1-N1989/+ mutant phenotype (Khangura et al. 2019a(Khangura et al. , 2019b. Our results, thus far, are consistent with an integrative measure of carbon assimilation linking energy status and vegetative to reproductive transition in maize (Minow et al. 2018).
In this report, we show that these allelic combinations at vey1 and oy1 can be used to study the effect of chlorophyll content and photosynthetic rate on plant growth and development. We again used crosses of Oy1-N1989/+ mutants to B73 x Mo17 recombinant inbred lines (RIL) and near-isogenic lines (NIL) carrying alternate alleles at vey1 to vary chlorophyll content in maize.
QTL mapping was used to detect loci that modified stalk width and plant height in both wildtype and mutant F1 hybrid (RIL x tester) populations. The single locus testing of oy1-vey1 interactions for these traits were tested directly using NIL x tester crosses. We found stalk widths of Oy1-N1989 mutants were narrower than wildtype siblings and this was enhanced by the vey1 Mo17 allele, just as we previously found for chlorophyll content, net CO2 assimilation, and time to reproductive maturity (Khangura et al. 2019a(Khangura et al. , 2019b. To our surprise, we found that plant height in the Oy1-N1989 mutants can be either greater or less than their congenic wildtype siblings. This divergence in plant height of heterozygous Oy1-N1989 mutants was contingent upon the vey1 allele. We also tested the effects of Oy1-N1989 and vey1 interaction on tassel phenotypes and other axillary meristem structures such as tiller numbers in isogenic materials. A previous study identified effects of a recessive partial loss-of-function allele at oy1 on tassel branch number (Brewbaker 2015). Our results agree with Brewbaker (2015) as Oy1-N1989 decreased tassel branch number dramatically in the presence of vey1 Mo17 allele and the effect was suppressed by vey1 B73 . Remarkably in hightillering sweet corn, Tillering1 (Tlr1), teosinte branched1 (tb1) and grassy tillers1 (gt1) mutants, Oy1-N1989 reduced the number of tillers at maturity and an epistatic effect of Oy1-N1989 on tb1 and gt1 was further enhanced by the vey1 Mo17 allele. The effect of a chlorophyll biosynthetic mutant on plant architecture and the unexpected opposing effect of the mutant and the vey1 modifier on plant height are discussed.

Location, Experimental design, and planting
All the field experiments were performed at the Purdue Agronomy Center for Research and Education (ACRE) in West Lafayette, Indiana. The test cross populations of IBM F1 and Syn10 F1 hybrid progenies were evaluated in 2013 (IBM) and 2014 (Syn10) in a randomized complete block design (RCBD) with one and two replications, respectively. The F1 progeny of Oy1-N1989/+:B73 crossed with B73 and Mo17 were used as parental checks in each range treated as a block. The F1 progenies of Oy1-N1989/+:B73 with NIL and parents (B73 and Mo17) were planted in a RCBD with five replications in 2016. All F1 families were planted as a single plot of 12-16 plants. Plots were sown in a 3.84 m long row with the inter-row spacing of 0.79 m and an alley space of 0.79 m. No irrigation was applied during the entire crop season as rainfalls were uniformly distributed for satisfactory plant growth. Conventional fertilizer, pest and weed control practices for growing field maize at Purdue University were followed.

Phenotyping and data collection
The plant height and stalk width were measured in both wildtype (WT) and mutant (MT) siblings segregating in each family in the F1 populations of Oy1-N1989/+:B73 with the IBM , Syn10, and NIL populations. As described previously (Khangura et al. 2019a(Khangura et al. , 2019b, genetic backgrounds containing the vey1 B73 allele progressively suppresses the mutant phenotype and were tagged at V5-V7 stage when they were distinguishable from the wildtype siblings so that traits could be recorded at maturity when the genotypes are harder to distinguish. Trait measurements were performed on two to four WT and MT siblings that were picked at random. Plant height and stalk width were measured in cm and caliper units (Chaikam et al., 2011). Briefly, around three weeks after reproductive maturity, plant height was measured as the vertical distance from the soil line to the point of attachment of the flag leaf (abbreviated as FlHT), and ear height was measured as the vertical distance from the soil line to the top/primary ear (abbreviated as EaHT). Both FlHT and EaHT were recorded in cm. Additional height traits were calculated indirectly by subtracting EaHT from FlHT to get the distance between the top ear and flag leaf (abbreviated as Ea2Fl).
Ratios (MT/WT) and differences (WT-MT) of these traits were also derived for each family. The EaHT were not measured in the IBM F1 populations, and consequently, any height trait derived from EaHT is missing in this population. Stalk widths (abbreviated as SW) were measured approximately three weeks after reproductive maturity using a Vernier caliper (Vernier 1631) at the middle of the first internode above the primary ear. The stalk width measurements were Mesocotyl measurements were done on seedlings grown under continuous red light (15-18 µmol m -2 s -1 ), far-red light (6-7 µmol m -2 s -1 ), white light (150 µmol m -2 s -1 ), and dark at 25º C for seven days at previously described peak wavelengths (Puthiyaveetil 2008;Macadlo et al. 2019).

Genotypes and gene expression data
The genotypic and gene expression datasets used in this study are described previously in detail (Khangura et al. 2019a(Khangura et al. , 2019b. Briefly, the public marker data for IBM consists of 2,156 unique markers. The gene expression of the oy1 locus was obtained from MaizeGDB. This public dataset consists of normalized read counts, expressed as reads per kilobase of transcript per million mapped reads (RPKM), of the maize genes from the transcriptome of shoot apices of 14 days old IBM seedlings (Li et al. 2013(Li et al. , 2018. The Syn10 genotypes were obtained from Liu et al., 2015. This dataset consists of SNP genotypes at 6611 positions (B73 RefGen v2). The BM-NIL carrying introgressions of the vey1 region from opposite parents were selected for QTL validation from a set of previously described NIL (Eichten et al. 2011).

Statistical analyses
Preliminary data exploration and descriptive statistics were calculated using JMP 13.0 (SAS Institute Inc. 2016). Trait means for each line were used as input phenotypes for QTL analyses.
The QTL analyses were done using the R/QTL package version 1.40-8 (Broman et al., 2003) using single interval mapping (SIM) for all traits. The statistical thresholds to declare significant loci for each trait were computed by permutation testing using 1000 iterations (Churchill and Doerge 1994).

Oy1-N1989 reduces stalk width and is epistatic to vey1
Heterozygous Oy1-N1989 plants (Oy1-N1989/+) have their eponymous oil-yellow color due to reduced chlorophyll content but are fertile and produce both ears and tassels (Khangura et al. 2019a). In the B73 inbred background, Oy1-N1989/+ plants germinate as yellow-green seedlings but the phenotype is progressively suppressed as they develop and are difficult to distinguish by color from wildtype plants at maturity (Khangura et al. 2019a). We previously described cis-acting expression polymorphisms at oy1 distinguishing the maize inbred lines Mo17 and B73 and proposed that this expression polymorphism is the genetic basis of an Oy1-N1989 modifier QTL very oil yellow 1 (vey1). When the enhancing vey1 allele from Mo17 is introgressed into the B73 background, Oy1-N1989/+ plants with a vey1 Mo17 allele exhibited lower levels of chlorophyll and delayed flowering as compared to suppressed mutants with the vey1 B73 allele (Khangura et al. 2019a(Khangura et al. , 2019b. The further reduction in chlorophyll content in enhanced Oy1-N1989/+ plants with a vey1 Mo17 allele increased the time to reproductive maturity and decreased CO2 assimilation and sugar contents in young leaves (Khangura et al. 2019b).
To determine if the reduced net CO2 assimilation, sugar content, or chlorophyll reduction affected by variation in oy1 had effects on growth and development we measured multiple traits in materials with a range of phenotypic severities. We started with F1 crosses of B73 and Mo17 with and without the Oy1-N1989 allele. As reported previously, chlorophyll measurements at two time-points (CCMI, ~3-4 weeks after sowing; CCMII, ~6-7 weeks after sowing) were lower in mutants in Mo17 x B73 F1 background compared to mutants in B73 background (Table S4; Khangura et al. 2019a). The heterotic Mo17 x B73 F1 wildtype hybrid plants had greater stalk diameters than the B73 plants ( Table S4). The presence of an Oy1-N1989 allele in the Mo17 x B73 F1 background eliminated the hybrid advantage producing mutant hybrids with narrower stalks than both wildtype and mutant B73 inbred plants. We hypothesized that the vey1 allele in Mo17 was responsible for enhancing the effect of Oy1-N1989 on stalk width in the B73 x Mo17 F1 hybrids. To test this, we measured stalk width in F1 populations generated by crossing Oy1-N1989/+ to hundreds of recombinant inbred lines from the IBM and Syn10 populations. As expected, mutant chlorophyll values showed high positive correlations with mutant stalk width in both IBM and Syn10 F1 populations (Tables S5 and S6). Not surprisingly, as previously reported for chlorophyll content, CO2 assimilation, and reproductive maturity, the vey1 locus modified stalk width in the Oy1-N1989/+ mutant F1 siblings but not their wildtype individuals. The vey1 locus was the largest effect QTL for stalk width in mutant siblings in both mapping populations explaining ~43% and ~68% phenotypic variation in IBM and Syn10 F1 populations, respectively ( Figure 1; Tables S7 and S8). Assessment of stalk width to control for congenic wildtype sibling background using ratio only detected vey1 in IBM, whereas in Syn10, an additional QTL was detected on chromosome 9. The regression using linked markers for this additional QTL (chr09.159.5) and vey1 (chr10.94.5) in Syn10 did not detect any interaction between these two QTL for Ratio_SW (data not shown). No genes with previously described roles were present within a 400kb window surrounding the peak maxima on chromosome 9 (data not shown). These results formally extend the epistatic relationship of the Oy1-N1989 allele and vey1 locus to stalk width. The F1 mutant plants that inherited the suppressor vey1 B73 allele were thicker than the mutant plants harboring enhancing vey1 Mo17 allele. This demonstrated that stalk width responded similarly to chlorophyll content, reproductive maturity, and CO2 assimilation. Just as was the case for chlorophyll content, vey1 was the only QTL detected for mutant stalk width in both populations.
The stalk widths of wildtype siblings in both populations were also variable and under genetic control. The localization of small effect QTL are often imprecise, but candidate genes within a window bound by a 2-LOD decrease in the QTL significance were tabulated for each detected QTL (Tables S7 and S8). QTL-candidate genes were assessed for roles in growth regulation and marked for eQTL using data from a previous study of the IBM population (Li et al. 2013(Li et al. , 2018. Three QTL, one in the IBM and two in the Syn10 F1 population, were detected on chromosome 1 for wildtype stalk widths (Tables S7 and S8). There was no overlap of the 2-LOD QTL intervals between IBM and Syn10 for stalk width of the wildtype siblings. Moreover, the Mo17 allele at the QTL detected in the IBM increased wildtype stalk width, whereas both Syn10 QTL showed increase in stalk width with B73 alleles, making an imprecise localization of the same QTL detected in both populations on chromosome 1 unlikely. The QTL in IBM is a moderate effect QTL with an estimated contribution of 13% of the phenotypic variation (Tables S7), but such estimates should be regarded as upper estimates (Beavis 1994). The 2-LOD window at this QTL contains a candidate gene in the ß-expansin family (expb4; B73 v4: Zm00001d033231), whose ortholog in rice have been implicated to increase stem elongation in response to gibberellic acid (Lee and Kende 2001). EXPB4 transcript abundance was influenced by cis and trans eQTL with B73 alleles increasing expression in all cases, making a cis-acting eQTL a possibility for this candidate gene. No other QTL overlapped genes with previously described roles in stem architecture (data not shown).
Confirmation of the impact of the vey1 locus on stalk width was provided by crosses between Oy1-N1989/+ mutants and near-isogenic lines (NIL) with introgressed vey1 alleles. In all NIL F1 families, Oy1-N1989/+ mutants were thinner than their wildtype siblings (Table S3). In F1 crosses of NIL from the B73 and Mo17 isogenic backgrounds to Oy1-N1989/+, the allele at vey1 had no discernable impact on the stem width of the wildtype siblings (Figures 2a and S2). In both B73 and Mo17 x B73 F1 isogenic backgrounds inheritance of the vey1 Mo17 allele enhanced the Oy1-N1989/+ phenotype leading to thinner plants than isogenic siblings with the suppressing vey1 B73 allele (Figures 2a and S2). Thus, Oy1-N1989 controlled stalk width and was epistatic to the vey1 QTL, consistent with the previously reported effects and interactions of these alleles on chlorophyll content, reproductive maturity, and CO2 assimilation (Khangura et al. 2019a(Khangura et al. , 2019b.

Plant height in Oy1-N1989 shows a non-linear effect direction dependent on the vey1 allele
In addition to the previously described delay in reproductive maturity of Oy1-N1989/+ mutants in the B73 background (Khangura et al. 2019b), mutant plants reached or surpassed the total plant height (FlHT) and ear height (EaHT) of their congenic B73 wildtype siblings (Figure 2b and Table S4). In the parental crosses, wildtype Mo17 x B73 F1 hybrid plants had greater plant height and ear height than the B73 inbreds (Table S4). Not only were Oy1-N1989/+ mutants in the Mo17 x B73 F1 hybrid not taller than wildtype hybrids, they showed a significant reduction in ear height, but not plant height, compared to the wildtype siblings. and S8). This effect was clear in both IBM and Syn10 F1 populations. In addition, the average height of mutant plants that inherited the vey1 B73 allele was significantly greater than their much greener wildtype siblings (Figure 4). In both populations, mutant plants that inherited the vey1 Mo17 allele were shorter than their wildtype siblings. It would appear that the relationship between the disruption of chlorophyll biosynthesis and final plant height are complicated and non-linear. As a result, Oy1-N1989/+ mutants can either be taller or shorter than their wildtype siblings with the direction of the deviation conditioned by the vey1 allele.
In the Syn10 F1 crosses, we also measured the height of the primary ear node (EaHT) which allowed us to obtain distance between primary ear node and flag leaf (Ea2Fl). Similar to the results for total plant height, vey1 Mo17 resulted in reduced EaHT and Ea2Fl in Syn10 F1 plants and vey1 B73 resulted in plants with greater EaHT and Ea2Fl than wildtype controls. Thus, plant heights of chlorotic Oy1-N1989/+ mutants can be either increased or decreased by alternate alleles of vey1.
This allele-specific interaction between vey1 and Oy1-N1989, with opposing effects on mutant plant height for vey1 B73 and vey1 Mo17 as compared to the wildtypes, is inconsistent with the simple interpretation that plant height in the Oy1-N9189 mutants is conditioned solely by reduced growth resulting from less carbon assimilation in the Oy1-N9189 mutants.
A number of smaller effect QTL were also identified for plant height in both IBM and Syn10 F1 populations (Tables S7 and S8). Using the same approach described above for stalk width, candidate genes were sought with a 2-LOD window of QTL that were not linked to vey1 and affected height. A QTL for wildtype FlHT was detected on chromosome 3 in the Syn10 F1 population (Table S8) at a similar position to a previously reported plant height and flowering time QTL in both Syn10 and IBM inbred lines . This QTL overlaps the recently described flowering time regulator zmmads69 (B73 v4: Zm00001d042315; Liang et al. 2019) encoding a QTL detected between maize and teosinte (Liang et al. 2019), and in two maize association panels (Hirsch et al. 2014;Lin et al. 2017). Over-expression of zmmads69 hastens flowering time in maize and reduces plant height in all cases (Liang et al. 2019). Consistent with strong cis-eQTL encodes increased accumulation of mRNA from the Mo17 allele of zmmads69 (LOD = 44.17, additive effect = -3.84, R 2 = 0.76; Table S3 from Li et al. 2013; please note that the effect direction was reversed due to the correction in Li et al. 2018). We tested the mutant siblings for an effect of zmmads69 on flowering time and for interaction with vey1 using the data of Khangura et al. 2019b. The marker associated with reduced wildtype FlHT linked to zmmads69 (chr03.1591) was significant in a single marker regression test for an effect on flowering time in the wildtype (LOD = 3.04, additive effect = 0.57, R 2 = 0.05, P-value = 0.0002) and mutant siblings (LOD = 0.86, additive effect = 1.13, R 2 = 0.02, P-value = 0.047) but had no effect on the ratio or difference in flowering time between mutant and wildtype siblings (data not shown). There was no interaction between zmmads69 and vey1 for days to anthesis in mutant siblings (data not shown) indicating that this QTL is not chlorophyll-dependent. Interaction between zmmads69 and vey1 for mutant height traits (FlHT, EaHT, and Ea2Fl) were only significant for mutant Ea2Fl (additive effect = -6.58, R 2 = 0.02, P-value = 0.03) but not in the normalized Ea2Fl (both ratio and difference). To test if two flowering time QTLs, vey1 and other oy1 flowering time locus1 (oof1), from mutant (Khangura et al. 2019b) also result in plant height differences, we performed a regression test using the main and interaction effects of these QTLs on mutant FlHT. There was no effect of oof1 or oof1 x vey1 interaction on mutant FlHT (data not shown).
Multiple FlHT QTL were identified on chromosome 9 in both Syn10 and IBM F1 populations (Tables S7 and S8). In Syn10 mutant F1 plants we detected a FlHT QTL, centered at 3.2 Mbp, at the top of chromosome 9 close to the position of height QTL for wildtype plants from B73 x Mo17 crosses detected in multiple previous studies (Beavis et al. 1991Austin et al. 2001;Liu et al. 2015). Others have noted that this QTL colocalizes with the dwarf plant3 (d3) gene (Beavis et al. 1991;Wang et al. 2006Wang et al. , 2018 encoding Ent-kaurenoic acid oxidase (B73 v4 Zm00001d045563; Winker and Helentjaris 1995). There is a strong cis-eQTL in the IBM transcriptomic data that accounts for 63% of the variation in DWARF PLANT3 mRNA accumulation (Li et al. 2013). Recently, it was revealed that this eQTL study of the IBM population (Li et al. 2013) had reversed the signs of the effects on gene expression (Li et al. 2018). As a result of this correction, it is clear that dwarf plant3 cannot encode this QTL as the gene expression is lower from the Mo17 allele, and the linked marker with Mo17 allele condition an increase in plant height ( Table S8) (Ma et al. 2014). No eQTL was detected for zmgras45 in the previous study of IBM shoot apices (Li et al. 2013(Li et al. , 2018.
Of the remaining QTL, only one had a 2-LOD window that overlapped a gene known to impact plant height. In the Syn10 F1 population, a QTL on chromosome 1 with a peak at 82 Mbp was detected for FlHT in both wildtype and mutant siblings ( Table S8). The 2-LOD window for this locus overlaps a MATE transporter (MATE; B73 v4: Zm00001d029677), which is orthologous to A. thaliana genes known to encode dominant alleles affecting hypocotyl elongation  and facilitating ABA transport (Zhang et al. 2014). No significant eQTL were described for this MATE in the previously published IBM eQTL study (Li et al. 2013(Li et al. , 2018.

Single locus recapitulation and validation of the non-linear relationship between plant height and chlorophyll level
To test the single-locus effect of vey1 on plant height of the Oy1-N1989 mutant, we crossed Oy1-N1989/+ mutants to near-isogenic lines (NIL) in B73 background with vey1 Mo17 allele introgressions. The F1 progenies from these crosses segregated for Oy1-N1989 mutants with alternative vey1 alleles (vey1 B73 and vey1 Mo17 ). Supporting a simple explanation of one-locus determination, in the isogenic B73 background the total plant height of Oy1-N1989/+ mutants with a vey1 B73 allele was greater than both wildtype siblings and Oy1-N1989/+ mutants with a vey1 Mo17 allele (Figures 2 and 5). Similarly, when vey1 Mo17 enhanced the Oy1-N1989/+ phenotype, plants were shorter than all other genotypes. Thus, we were able to recapitulate the effect of vey1 on mutant plant height observed in the RIL and DH population using a single-locus introgression in an isogenic inbred background.
The B73 NIL removed hybridity from our tests, and we considered that some of the variation in the Oy1-N1989 x vey1 interaction effects on plant height could be driven by overall plant vigor and hybridity. For instance, in both IBM and Syn10 crosses, the vey1 Mo17 allele decreased plant height (Figure 4B and 4E), but we did not see such reduction in height in the original B73 x Mo17 F1 materials that also contained vey1 Mo17 allele (Table S4). Each RIL and DH is ~50% Mo17 while the Oy1-N1989 mutant is in a B73 inbred background resulting in F1 progeny that were hybrid at a random collection of 50% of the loci contributing to heterosis in this cross. To perform a test of the effect of vey1 in an isogenic hybrid background, we crossed Oy1-N1989/+:B73 to NIL derived using Mo17 as the recurrent parent with introgressions of alternative alleles at vey1. This test maintains hybridity at most unlinked loci between B73 and Mo17, unlike the RIL crosses which retained hybridity for only approximately half of the genome. The mutant F1 progenies derived from Mo17 NIL containing the vey1 B73 allele were significantly taller than the wildtype siblings (Figures 2 and 5), just as observed in the IBM, Syn10, and isogenic B73 NIL F1 progenies (Table S4; Figures 2, 4, and 5). However, the mutant F1 progeny derived from Mo17 NIL that inherited the vey1 Mo17 allele showed no discernable difference to total plant height as compared to wildtype siblings, but ear height decreased and a compensatory increase in the height between ear and flag leaf was observed ( Figure 5). Thus, vey1 B73 consistently increased plant height of the Oy1-N1989/+ plants over wildtype siblings in isogenic B73 background, across a population of recombinant inbred lines crossed to B73, and in B73 x Mo17 F1 NIL hybrids, whereas, the vey1 Mo17 allele decreased mutant plant heights when compared to wildtype siblings in all materials except the hybrid Mo17 x B73 F1 background.

Expression level variation at oy1 in IBM correlates with stalk width and height in mutants
We previously proposed vey1 encodes a cis-acting expression polymorphism in the regulatory sequence at oy1 (Khangura et al. 2019a(Khangura et al. , 2019b. Our previous reanalysis of transcript data derived from shoot apices from IBM (Li et al. 2013(Li et al. , 2018, a maize association panel (Kremling et al. 2018), and our allele-specific expression assays in NIL demonstrated that the vey1 B73 allele was linked to higher expression of wildtype OY1 transcript (Khangura et al. 2019b). Based on this linkage, we expect stalk width and height variation in mutant plants to also be correlated with OY1 mRNA accumulation. Stalk width and plant height in mutant IBM F1 progenies showed significant linear correlation with OY1 transcript abundance in the shoot apices of the IBM inbred lines (Figure 6). The OY1 transcript variation in IBM did not correlate with stalk width and plant height in the wildtype siblings. These results are consistent with vey1 encoding a cis-regulatory polymorphism controlling OY1 as the genetic basis of stalk width and height traits in Oy1-N1989/+ mutants.
If the effect of vey1 on Oy1-N1989/+ mutants is the result of rescuing chlorophyll biosynthesis, rather than some novel function of the protein encoded by Oy1-N1989, then chlorophyll abundance should correlate better with gross morphology than OY1 transcript abundance. Both CCM measures (CCMI and CCMII) vary within each vey1 genotypic class. CCM increases even within the RIL containing vey1 Mo17 and one containing vey1 B73 allele (Figures S3-S7). If the effects on development are mediated by a loss of net CO2 capture and plant energy status due to reduction in chlorophyll, then the variation in CCM levels in Oy1-N1989 mutants due to phenomena other than vey1 should result in corresponding fluctuation in phenotype.
To explore the relationship between chlorophyll abundance variation affected by Oy1-N1989 and modifiers and development, we explored the phenotypic correlations between the previously measured and described CCM data (Khangura et al. 2019a), the reproductive maturity date (Khangura et al. 2019b), plant height and stalk width measurements detailed in this study. As the effect of vey1 is only visible in the presence of the Oy1-N1989 allele for all traits, we only used difference of congenic wildtype and mutant siblings (MT-WT) to study these variations with respect to chlorophyll. Mutant stalk widths reduced (increase in difference) with a decrease in CCM within both vey1 Mo17 and vey1 B73 genotypes in IBM and Syn10 F1 populations ( Figure S3).
Consistent with the effects on net CO2 capture (Khangura et al. 2019b), the slope of the regression line and therefore the fit was greater for the mutants with vey1 Mo17 allele. Thus, the variation in chlorophyll content among individuals within each vey1 genotype was sufficient to affect stalk width.
Similarly, the reduction in total plant height was also suppressed in mutants with vey1 Mo17 allele that accumulated relatively higher chlorophyll in both IBM (R 2 = 0.18, P-value = <0.0001) and Syn10 (R 2 = 0.09, P-value = 0.008) F1 populations ( Figure S4). There was no discernable impact of a change in chlorophyll content on plant height in the vey1 B73 mutant F1 siblings ( Figure   S4). The effect of chlorophyll variation on ear height was only significant between the later mutant chlorophyll estimate (CCMII) in IBM with vey1 Mo17 allele and ear height, although a similar trend was observed for Syn10 F1 progenies inheriting vey1 Mo17 allele ( Figure S5)

Seedling etiolation in far-red light is not altered in Oy1-N1989 mutants
The non-linearity of the effect of Oy1-N1989 and vey1 on plant height raised the possibility that disrupting CHLI alters light signaling in maize. The shared biosynthetic origins of chlorophyll and bilin chromophores required for phytochrome assembly (Frankenberg and Lagarias 2012) was explored as a possible source of increase in total plant height observed in Oy1-N1989/+ mutants modified by vey1 B73 . Low R/FR ratio is indicative of shade in the plant canopy and elicits a shade avoidance syndrome via phytochrome signaling (Chen et al. 2004;Pierik and De Wit 2014;Roig-Villanova and Martínez-García 2016;Ortiz-Alcaide et al. 2019). If a mutation in oy1 perturbs tetrapyrrole biosynthesis, a reduction in the pool of functional phytochrome should increase mesocotyl length and plant height and reduce the number of days to reproductive maturity as had been reported previously in the elongated mesocotyl1 (elm1) maize mutant which is defective in chromophore synthesis and has reduced phytochrome levels (Sawers et al. 2002). To test whether the range MgChl defects affected by the Oy1-N1989 and vey1 interactions influence the phytochrome pool, we grew plants in a variety of light conditions that stimulate phytochrome mediated elongation.
We grew near-isogenic Oy1-N1989 mutant seedlings with segregating vey1 alleles under four light conditions (continuous dark, white, red and far-red light) and measured mesocotyl length. The elm1 mutant was used as a positive control for these growth conditions (Sawers et al. 2002(Sawers et al. , 2004. As expected, the elm1 mutants showed significantly longer mesocotyls compared to wildtype in all light conditions, and the stimulatory effect of red and shade-mimicking far-red light on mesocotyl length in elm1 was evident (Figure 7). Under all light conditions tested, no statistically significant differences were observed in mesocotyl length between wildtype and Oy1-N1989 mutants with either vey1 Mo17 or vey1 B73 allele. These results indicate no, or very limited, impact Oy1-N1989 on light-mediated growth responses consistent with independent regulation of the tetrapyrrole pathway branches leading to chlorophyll and chromophore biosynthesis.

Tassel branching is reduced in severe Oy1-N1989 mutants
A recessive mutant allele at oy1 was previously shown to reduce the tassel branch number in maize (Brewbaker 2015). The availability of near-isogenic materials with segregating vey1 alleles prompted us to test whether the semi-dominant Oy1-N1989 allele and its interaction with vey1 alters tassel branching and development. Using the F1 populations derived from crosses of both B73-like and Mo17-like NIL with Oy1-N1989/+:B73, we performed a morphometric analysis of the tassels collected from mutant and wildtype siblings of each F1 family. Tassel growth (measured as the length of the total tassel, branching zone, and central spike) and development (primary and secondary branch number) was not affected in Oy1-N1989/+ mutants modified by the suppressing vey1 B73 allele in both B73 and Mo17 NIL backgrounds (Figure 8 and Table S9). The Oy1-N1989/+ mutants carrying the enhancing vey1 Mo17 allele had reduced tassel branch numbers and tassel length. Mutants derived from crosses of Oy1-N1989/+:B73 with B73 and Mo17 NIL carrying vey1 Mo17 introgression displayed a reduction in the primary tassel branch number by ~47% and ~23%, respectively. Similarly, the overall length of the tassels was reduced by ~39% and ~18% in Oy1-N1989 mutants in the F1 generation with B73 and Mo17 NIL with vey1 Mo17 introgression, respectively. These changes in tassel architecture are consistent with the report of Brewbaker (Brewbaker 2015) where ~30% reduction in tassel branching was observed.

Tillering is reduced in Oy1-N1989 mutants and this effect is modulated by vey1
Sugar availability is implicated in axillary bud outgrowth (Barbier et al. 2015). The levels of sucrose and other reducing sugars are lower in Oy1-N1989 and further modified by vey1 (Khangura et al. 2019b). We tested if Oy1-N1989 modulated tiller outgrowth by crossing the mutant to well characterized prolific tillering genetic stocks such as (a) the maize inbred P39 which carries the high tillering allele at tiller number1 (Zhang et al. 2019), (b) the classical dominant tillering mutant Tillering1, and (c) a semi-dominant tb1 in a mixed background. The tillering of the F1 hybrids between B73 and all three high tillering lines (sweet corn inbred P39, Tlr1, and tb1) was suppressed by Oy1-N1989 allele (Figure 9 and Table S10). To eliminate mixed and hybrid genetic background effects in these crosses, we further tested the influence of Oy1-N1989 mutation on tillering in tb1 and gt1 mutants in isogenic B73 backgrounds. We used B73-like NIL carrying both vey1 alleles to generate F2 populations that segregated for Oy1-N1989 along with tb1 or gt1 mutants in an isogenic B73 background. The Oy1-N1989 allele suppressed tillering in both tb1 and gt1 mutants. The effect of Oy1-N1989 on tillering was further modified by vey1 in the tb1;Oy1-N1989/+ and gt1;Oy1-N1989/+ double mutants as compared to tb1 and gt1 mutant siblings (Figure 10). The tb1;Oy1-N1989/+ and gt1;Oy1-N1989/+ plants that inherited vey1 B73 alleles from the NIL had an intermediate number of tillers whereas those with vey1 Mo17 alleles had little to no lateral tiller outgrowth. Though Oy1-N1989/+ suppressed tillering in gt1 homozygotes, these plants were readily distinguishable from gt1 heterozygotes and wildtype siblings due to extended blade tissue from the husk sheath (Figure 10b) that was described previously (Whipple et al. 2011). Thus, the enhancement and suppression of tiller outgrowth in tb1 and gt1 mutants by the Oy1-N1989 and vey1 alleles displayed the additive and linear relationship previously observed for chlorophyll accumulation and net CO2 fixation. This is consistent with sugar availability controlling tiller outgrowth in maize (Mason et al. 2014;Barbier et al. 2015). The possibility that these changes in axillary branch growth in the combinations of tin1;Oy1-N1989/+, tb1;Oy1-N1989/+ and gt1;Oy1-N1989/+ are due to change in photosynthetic capacity, suggest strongly that future investigations seeking identification of relationships between the fundamentals of energy balance sensing and lateral organ growth control are needed.

Discussion
We demonstrate here that modulation of chlorophyll accumulation in Oy1-N1989 mutants by the standing variation in the maize germplasm at vey1 varies metabolism sufficient to alter plant growth and development. Our previous study using these genetic materials showed that cis-acting expression polymorphism at oy1 locus encoded by alleles of vey1 can influence the transcript abundance of wildtype OY1. As a result, relative increase in wildtype OY1 transcripts by suppressing vey1 alleles was consistent with higher chlorophyll accumulation in Oy1-N1989/+ mutants, and vice-versa (Khangura et al. 2019a). We further demonstrated that the lower chlorophyll levels in Oy1-N1989 mutants conditioned by vey1 alleles resulted in lower photosynthetic rate, non-structural carbohydrate synthesis, and delayed reproductive maturity (Khangura et al. 2019b). In this study, we demonstrate that chlorophyll metabolism, and the variation affected by the vey1 locus interactions with the Oy1-N1989 mutant allele, sufficiently disrupts metabolism to alter stem thickness, plant height, tiller production, and tassel branching.
In all traits but plant height, Oy1-N1989 mutants had lower trait expression than the wildtype siblings, consistent with a loss of growth due to decreased photosynthetic output. The performance of the mutant plants was further reduced for these traits by the enhancing vey1 Mo17 allele. So far, all phenotypes except plant height show a linear change in trait value with suppressing (e.g. B73) and enhancing (e.g. Mo17) alleles at vey1 (Figures 2, 4, and 5). This relationship is not a genetic correlation between alleles and phenotypes. As expected for the traits to be determined by the effect of alleles on chlorophyll biosynthesis, the variation in chlorophyll levels between individuals carrying the same vey1 genotype were also correlated with consistent changes in plant phenotypes for all traits other than plant height (Figures S3-S9).
The effect of vey1 alleles on mutant plant height does not follow a linear relationship and cannot be explained by continuous loss of chlorophyll creating a continuum of sugar levels which support differing plant growth. The Oy1-N1989/+ mutants inheriting a vey1 B73 allele had a mild decrease in chlorophyll and lower net CO2 assimilation (Khangura et al. 2019a(Khangura et al. , 2019b) and yet were taller than their wildtype siblings. The Oy1-N1989/+ mutants with the vey1 Mo17 allele that have further reduction in chlorophyll and net CO2 assimilation were shorter than their wildtype siblings (Figures 2, 4, and 5; Table S4). These results contrast to those from Arabidopsis, where a 20% decrease in chlorophyll lead to a small but significant decrease in hypocotyl elongation under the shade and a further decrease lead to greater reduction in hypocotyl lengths (Ortiz-Alcaide et al. 2019). In our experiments neither the ~80% nor 90% reduction in chlorophyll of the Oy1-N1989/+ mutants with the vey1 B73 and vey1 Mo17 alleles (from the CCMI measurements in Khangura et al., 2019b) reduced nor increased mesocotyl elongation under continuous white light, darkness, red, or far-red light. (Figure 7). This may simply relate to the amount of carbohydrates stored in the maize seed available for mesocotyl elongation. The opposing effects of vey1 alleles on the influence of Oy1-N1989 on plant height strongly argues for additional experiments to distinguish whether a biochemical signal related to tetrapyrrole metabolism or an indirect effect of chlorophyll loss, unrelated to CO2 assimilation, lead to the changes in plant height.
In addition to vey1, multiple QTLs were detected for stalk width and plant height measurements in both mutant and wildtype siblings. Putative genes underlying these QTLs discussed above could explain the genetics of these traits. One of the QTL for total plant height variation in the wildtype siblings in the Syn10 F1 population on chromosome 3 contained zmmads69 (Table S8). QTL for plant height and flowering time in the IBM and Syn10 inbred populations were previously mapped to a similar position . Recently, zmmads69 was characterized as a positive regulator of flowering time and was proposed as a component of the ZmRap2.7-ZCN8 regulatory network that allowed latitudinal adaptation of maize (Liang et al. 2019). Overexpression of zmmads69 lead to early flowering time and also reduced plant and ear height (Liang et al. 2019) and a strong cis-eQTL for zmmads69 (LOD=44.17; R 2 =0.76) was detected in the IBM inbred lines, with the Mo17 allele resulting in greater transcript abundance, reduced plant height, and earlier maturity. We demonstrate here that the Mo17 allele at this locus is associated with a QTL that decreased plant height in the wildtype siblings of our Syn10 F1 population (Table S8) and, when considered without a multiple test correction, decreased days to anthesis.
Since we only measured tillering and tassel architecture in NIL crosses, it is a formal possibility that a modifier of plant branching is linked to oy1. For example, a tillering modifier that is common to both the gt1 and tb1 responses may be linked to oy1. However, the changes in the developmental phenotypes were contingent on Oy1-N1989/+ and it retained the epistatic interaction with vey1 that we have found for all traits (Figure 10). This makes it very unlikely that a linked QTL differing between B73 and Mo17 underlie this variation in tillering. In the case of tassel branching, our results are an independent validation of a previous report of reduced tassel branching by a loss-of-function allele at oy1 (Brewbaker 2015). In all cases, lateral branch numbers were reduced by loss of chlorophyll regardless of the modifying allele at vey1 (Figure 8).
The maize gene, tiller number1 (tin1) encodes a C2H2-zinc-finger transcription factor which is a dominant factor responsible for increased tiller growth in the Purdue sweetcorn line P51 (Zhang et al. 2019). TIN1 is upstream of GT1, but acts independently of TB1 (Kebrom and Brutnell 2015;Zhang et al. 2019). The high tillering Purdue Sweetcorn line P39, encodes the same tin1 allele as P51 (Zhang et al. 2019) possibly due to their common derivation from 'Golden Bantam' (www.ars-grin.gov). The Oy1-N1989 allele suppressed tillering conditioned by tin1 variant in P39 (Figure 9). The suppression of tillering in gt1 mutants in Oy1-N1989/+ backgrounds ( Figure 10) is consistent with tin1 acting upstream of gt1 (Zhang et al. 2019) to regulate tiller bud out growth. Oy1-N1989 also suppressed tb1 despite the fact that tb1 acts independently of tin1, suggests that loss of chlorophyll regulates branching upstream of the branch point between tb1 and tin1-gt1 (Figures 9 and 10; Table S10). In isogenic double mutants of Oy1-N1989 allele with gt1 and tb1, vey1 locus that have been previously shown to cause gradual decline in photosynthetic capacity and sugar accumulation of Oy1-N1989 mutants (Khangura et al. 2019b) also influenced tiller number in a vey1-dependent manner (Figure 10). A study analyzing young axillary buds of tb1 and gt1 mutants in maize found increased sugar biosynthesis and metabolism to be critical for breaking bud dormancy and that tb1 bound to the promoters of genes encoding enzymes important for energy sensing, such as enzymes involved in trehalose-6-phosphate metabolism (Dong et al. 2019). Analysis of transcript abundance in axillary buds of tin1 variants found photosynthesis to be the most enriched gene ontology term and chlorophyll biosynthesis to be highly upregulated in non-dormant tiller buds (Zhang et al. 2019). This may be tempting to dismiss, as the initiation of photosynthesis and respiration must be a consequence of bud outgrowth. However, it may be that sugar production at the branches is a critical determinant of branch growth. Based on our data that chlorophyll limitation suppresses tillering by tb1, gt1, and tin1 alleles, we predict that these genes regulate the process of bud outgrowth via stem sucrose concentrations and that these transcription factors act downstream to modify bud growth. We expect genes important for the patterning and placement of axillary meristems to act upstream of the sensor of carbon status.
Flowering time and plant height displayed a confusing relationship that was not resolved by the idea that later flowering plants were taller. Indeed, as expected based on the correlations with CCM and vey1 but in defiance of a simple explanation, as mutant plants accumulated more chlorophyll they flowered earlier and were taller in IBM regardless of the vey1 genotype ( Figure   S8). In the Syn10 F1 population, the plants with vey1 Mo17 allele behaved in this manner, but the vey1 B73 plants went in the opposite direction with increasing height and increasing flowering time co-occurring ( Figure S9). Thus, a secondary relationship between plant height and the flowering time delay does not explain the non-linear change in plant height with respect to chlorophyll levels.
Although the plant height in Oy1-N1989 and vey1 segregating materials is extremely reproducible across range of populations (two RIL populations and NIL) and across multiple years, the mechanism that explains the complicated relationship between the Oy1-N1989 allele, its modifier vey1 and CCM values to plant height and the paradoxical effect of this epistasis on different morphological traits is elusive to us at this time.
There is some separation between the CCM values that are produced by vey1 B73 -Oy1-N1989 and those produced by vey1 Mo17 -Oy1-N1989. It may be that a "goldilocks" value for CCM that maximizes plant height sits between the values we have measured. But if so, Oy1-N1989/+ mutants suppressed by a vey1 B73 allele should get taller as CCM decreases. Instead, the vey1 B73  and (d-f) Mo17-like NILs x Oy1-N1989/+:B73 F 1 hybrid populations. The chlorophyll content measurements (CCMI: early season; CCMII: late season) and time to reproductive maturity (DTA: days to anthesis; and DTS: days to silking) are depicted in panel a and d (data reanalyzed from Khangura et al. 2019aKhangura et al. , 2019b; height in panel b and e; stalk width in panel c and f. The statistical significance was determined using ANOVA, and the connecting letter report indicates mean comparisons using Tukey's HSD at p<0.05.  Figure 7. The seedling phenotypes of elm1, wildtype (WT) and Oy1-N1989/+ mutants (MT) with both vey1 alleles (vey1 B73 and vey1 Mo17 ) after 7 days of growth under continuous (a) white light, (b) no light/dark, (c) red light, and (d) far-red light conditions. The white arrows in panel a-d indicate boundary between mesocotyl and the first internode. (e) The quantification (mean ± S.D.) of various genotypic groups in all four growth conditions at 7 days. The connecting letter report indicates statistical significance determined using the student's t-test at p<0.05. The sample size is 8-19 seedlings per genotype in each growth condition.  Figure 8. The epistatic effect of vey1 Mo17 allele on tassel morphology of Oy1-N1989 mutants in B73 and B73 x Mo17 isogenic backgrounds. The data presented on the schematic of the maize tassel is derived from mutant siblings of B73 and Mo17-like NILs that were crossed to Oy1-N1989/+:B73 tester. Blue colored B73 and Mo17-NILs carry vey1 B73 allele, whereas a golden color with asterisk represents NILs with vey1 Mo17 introgression. The black cross and downward pointing arrow in a give tassel trait represents no significant reduction and a statistically significant reduction in mutants compared to the wildtype siblings, respectively. Full data is provided in Table S9.

No significant difference
Significant reduction