Redundancy, Feedback, and Robustness in the Arabidopsis thaliana BZR/BEH Gene Family

Abstract

Organismal development is remarkably robust, tolerating stochastic errors to produce consistent, so-called canalized adult phenotypes. The mechanistic underpinnings of developmental robustness are poorly understood, but recent studies implicate certain features of genetic networks such as functional redundancy, connectivity, and feedback. Here, we examine the BZR/BEH gene family, whose function contributes to embryonic stem development in the plant Arabidopsis thaliana, to test current assumptions on functional redundancy and trait robustness. Our analyses of BZR/BEH gene mutants and mutant combinations revealed that functional redundancy among these gene family members is not necessary for trait robustness. Connectivity is another commonly cited determinant of robustness; however, we found no correlation between connectivity among gene family members or their connectivity with other transcription factors and effects on developmental robustness. Instead, our data suggest that BEH4, the earliest diverged family member, modulates developmental robustness. We present evidence indicating that regulatory cross-talk among gene family members is integrated by BEH4 to promote wild-type levels of developmental robustness. Further, the chaperone HSP90, a known determinant of developmental robustness, appears to act via BEH4 in maintaining robustness of embryonic stem length. In summary, we demonstrate that even among closely related transcription factors, trait robustness can arise through the activity of a single gene family member, challenging common assumptions about the molecular underpinnings of robustness.

Keywords: BES1; BZR1; canalization; developmental robustness; hypocotyl; plant; stochasticity; variance.

FIGURE 1

The BZR/BEH family encodes genes with similar effects on hypocotyl length. (A) Seedlings were grown for seven days in the dark, and hypocotyls were measured. beh3-1, beh4-1, bes1-2, bzr1-2 hypocotyls were significantly shorter than those of wild-type (^∗p < 0.0001, linear mixed effects model with genotype as a fixed effect and replicate as a random effect). (B) The phenotype of the bes1-2;bzr1-2 double mutant suggests that BZR1 is epistatic to BES1 because there was no significant difference in hypocotyl length between bes1-2 and bes1-2;bzr1-2. Significant differences (p < 0.05) are displayed by the horizontal bars as determined by linear mixed effect modeling. (C) No significant differences in hypocotyl length were observed for beh1-1 and beh2-1 single mutants, or for the double mutant beh1-1;beh2-1. For (A–C) one representative replicate experiment with standard error of the mean for n > 20 is shown.

FIGURE 2

beh4-1 decreases robustness of dark grown hypocotyls | (A) The beh4-1 mutant exhibits significantly greater variation in hypocotyl length compared to wild-type (^∗∗∗p < 0.0001, Levene’s test, n = 210). None of the other single mutants increase hypocotyl length variance significantly. (B) The double mutant bes1-2;beh4-1 showed an intermediate effect on hypocotyl length robustness compared to either single mutant (^∗∗∗p < 0.0001, Levene’s test, n = 210). CV was estimated in three biological replicates. Standard error of the mean for n = 3 is shown for both (A) and (B). (C) The double mutant bes1-2;beh4-1 also showed an intermediate effect on hypocotyl mean values compared to either single mutant (^∗∗∗p < 0.0001, ^∗∗p < 0.001, linear mixed effects model with genotype as a fixed effect and replicate as a random effect).

FIGURE 3

BZR/BEH family members engage in extensive regulatory cross-talk. Level of gene expression (mRNA) in mutant backgrounds was determined using qPCR for (A) BES1, (B) BEH1, (C) BEH3, (D) BZR1, (E) BEH2, and (F) BEH4. (G) Direct and indirect regulatory relationships among BZR/BEH family members were determined from results in (A–F). A regulatory relationship was called for a gene if a greater than a 2-fold expression difference between wild-type and mutant backgrounds was measured. Both positive (arrow) and negative (bar) regulatory relationships are indicated.

FIGURE 4

Robustness provided by HSP90 likely arises from the chaperone’s interaction with BEH4 | (A) Seedlings were grown with or without HSP90 inhibition, and hypocotyl length was measured in three replicate experiments. CV was calculated for each replicate and the standard errors of the mean for n = 3 are shown. BES1 is a known HSP90 client in this gene family. (B) Hypocotyl length mean data for the same conditions are shown. One representative replicate experiment with standard error of the mean for n > 20 is shown. ^∗Significant differences in mean trait response to HSP90 inhibition are shown (p < 0.03, linear mixed model with genotype, treatment, and interaction effects as fixed effects and replicate as a random effect).