Rare genetic variation at Zea mays crtRB1 increases β -carotene in maize grain

Articles

table 2 crtRB1 haplotype estimated effects in P1 (2003 phenotypic data)

crtRB1

Haplotypea1,0,11,0,21,0,31,12,12,0,12,12,13,0,3

5′TE1111223

InDel4000120120

3′TE1231113

N28151096155

βC2.150.881.401.9113.346.792.6440%

5.23 × 10 167.6

s.e.m.1.560.711.420.95–4.771.21

βC/βCX2.320.640.954.0919.0610.182.9280%

1.68 × 10 5315.2

s.e.m.1.160.320.612.48–4.001.10

Traitsb

βC/Zb0.410.100.181.243.792.180.4068%

1.57 × 10 3818.3

s.e.m.0.270.070.291.19–1.830.11

βC/ALL0.10.030.060.100.420.390.1150%

1.02 × 10 2110.5

s.e.m.0.060.020.050.04–0.220.04

R2

P (ANOVA)

Avg. change between “2,12,1” and “1,0,2”c

aHaplotype is shown as linear combination of 5′TE allele (1, 397-bp insertion; 2, 206-bp insertion; 3, 0-bp insertion), InDel4 allele (12-bp or 0-bp insertion), 3′TE allele (1, no insertion;

2, 325-bp insertion; 3, 1,250-bp insertion). Favorable alleles are in boldface type, and only observed haplotypes are listed. bTraits are described in table 1. cComparison between best and worst haplotypes as predicted by component allelic effects.

© 2010 Nature America, Inc. All rights reserved.

six haplotype classes resulting from combinations of 5′TE, InDel4 and 3′TE explained 62–95% of the variation for the same derived traits (Supplementary Table 6).

We observed a 7.6-fold change in the average βC values between the most and least favorable haplotype classes in P1 in 2003 (Table 2) and a 10.5-fold change for the same comparison in P3 in 2006 (Supplementary Table 6). Haplotypes containing the favorable 5′TE and 3′TE alleles (n = 6) had an average βC concentration of 7.88 (±5.03) µg g 1, in contrast to all other haplotypes (n = 165) averaging

1 11.53 (±1.41) µg g. This 6.34 µg g difference represents 42% of

the HP target goal and indicates that crtRB1-specific allelic variation yields a marked increase in maize grain βC concentration. The derived traits of βC/βCX, βC/Z and βC/ALL, respectively, showed 15.2-, 18.3- and 10.5-fold change between the most and least favorable haplotype classes in P1 (Table 2), and 43.3-, 94.7- and 14.5-fold changes in P3 (Supplementary Table 6). Derived traits were not calculated for P2, as the frequency of 5′TE and InDel4 were too low to be analyzed.crtRB1 maps to a principal QTL for -carotene

We used five populations consisting of recombinant inbred lines (RILs) or F2:3 progenies (see Online Methods) that segregate for crtRB1 polymorphisms (parental genotypes listed in Supplementary Table 7) for quantitative trait locus (QTL) mapping (Supplementary Table 8). In the B73 × BY804 RIL population, crtRB1 mapped to a genetic interval containing a principal QTL for βC concentration and βC/ALL explaining 16.3% and 32.8% of the trait variation, respec-tively (Supplementary Fig. 2 and Supplementary Table 8). Principal QTLs for βC and other β-branch traits also mapped to an interval containing crtRB1 in DEexp × CI7 and A619 × SC55 F2:3 populations

table 3 crtRB1 genetic effects in three segregating populations

210

(Supplementary Table 8). The effect of crtRB1 on βC concentration was also further confirmed through single-factor analysis in the KI3 × SC55 and KI3 × B77 F2:3 populations (analysis of variance (ANOVA), α = 0.01, P < 0.001; Supplementary Table 8). QTL for variation in βC did not map to the four other BCH maize orthologs.

Three populations (A619 × SC55, KI3 × SC55 and KI3 × B77) seg-regate for at least one significant polymorphism in both crtRB1 and lcyε. To estimate crtRB1 effects with a single-gene model (Table 3), we adjusted for lcyε effects and found crtRB1 to explain significant variation in β-branch traits including βC (R2 range 8–44%), βC/βCX (R2 range 37–69%) and βC/ALL (R2 range 15–60%). Favorable homozygous geno-types lead to β-carotene increases of 2.78 µg g 1 in KI3 × SC55, 1.61 µg g 1 in KI3 × B77 and 0.53 µg g 1 in A619 × SC55 above βC concentrations of the respective unfavorable homozygotes, which averaged 2.79, 0.76 and 4.62 µg g 1 (Table 3). Additive effects for crtRB1 (ANOVA, α = 0.05, P < 0.001) in all three populations were associated with maximal

1per-copy increases in βC of 1.25 µg g, as estimated through linear

regression (Supplementary Table 9). Collectively, these results indicate that favorable crtRB1 alleles can lead to higher βC concentrations across a range of segregating genetic backgrounds.

crtRB1 genotypes associated with higher βC were also associated with a decrease in all other tested carotenoids and total carotenoid concentration (Fig. 2). Comparing favorable homozygous genotypes with unfavorable homozygous genotypes at crtRB1 showed that favo-rable crtRB1 homozygotes resulted in an 8.9–48% reduction in total carotenoids (P < 0.0001). Reduction in total carotenoids is possibly conferred by the favorable 5′TE allele, in that significant variation in total carotenoids was most often observed in populations polymorphic at 5′TE as compared with InDel4 or 3′TE (Supplementary Table 7).

Traitsb

3211450

334235

122628

5.15 5.02 4.627.80.531.11

2.371.490.7643.51.613.11

5.573.39 2.79 25.42.782.00

0.470.410.3817.00.091.23

0.170.070.0359.60.145.67

0.46 0.28 0.1749.00.292.70

8.984.833.6015.05.382.49

0.330.100.0463.10.298.25

9.65

1.500.6246.89.0315.56

12.337.285.436.86.932.28

2.250.630.2658.11.998.65

5.801.630.7769.35.037.53

R2 (%)

Actual difference between homozygotesRelative change between homozygotes

acrtRB1 genotypic class means for βC (µg g 1), βC/ALL, βC/Z and βC/βCX traits. All effects are adjusted by lcyε covariates. Adjusted means are all significantly different at α = 0.05, except

where noted ( ). bTraits are described in table 1. A, population A619 × SC55; B, population KI3 × B77; C, population KI3 × SC55.

Nature GeNetics