Plant_SNP_TATA_Z-Tester: A Web Service That Unequivocally Estimates the Impact of Proximal Promoter Mutations on Plant Gene Expression

Abstract

Synthetic targeted optimization of plant promoters is becoming a part of progress in mainstream postgenomic agriculture along with hybridization of cultivated plants with wild congeners, as well as marker-assisted breeding. Therefore, here, for the first time, we compiled all the experimental data-on mutational effects in plant proximal promoters on gene expression-that we could find in PubMed. Some of these datasets cast doubt on both the existence and the uniqueness of the sought solution, which could unequivocally estimate effects of proximal promoter mutation on gene expression when plants are grown under various environmental conditions during their development. This means that the inverse problem under study is ill-posed. Furthermore, we found experimental data on in vitro interchangeability of plant and human TATA-binding proteins allowing the application of Tikhonov's regularization, making this problem well-posed. Within these frameworks, we created our Web service Plant_SNP_TATA_Z-tester and then determined the limits of its applicability using those data that cast doubt on both the existence and the uniqueness of the sought solution. We confirmed that the effects (of proximal promoter mutations on gene expression) predicted by Plant_SNP_TATA_Z-tester correlate statistically significantly with all the experimental data under study. Lastly, we exemplified an application of Plant_SNP_TATA_Z-tester to agriculturally valuable mutations in plant promoters.

Keywords: TATA box; TATA-binding protein; Web service; correlation; development; environmental exposure; expression; gene; marker-assisted breeding; mutation; plant; plant hybrid; prediction; promoter; verification.

Figure 1

The effects of mutation within the analyzed artificial Pmec promoter on the β-glucuronidase (GUS) activity of tobacco under the experimental conditions “light” (x-axis) and “dark” (y-axis) in vivo [53] do not correlate with each other, thereby casting doubt on both the existence and the uniqueness of the uniform estimate for the mutational effects of plant proximal promoters on gene expression under various environmental conditions, usually called an ill-posed inverse problem [61]. Legend: circle, the prototype (arrow, →), or a mutant variant of the studied artificial promoter Pmec for plants; dashed and dotted lines are linear regression and limits of its 95% confidence interval, as calculated in the Statistica software (Statsoft^TM, Tulsa, OK, USA); r, R, τ, γ, and p are the linear correlation, Spearman’s rank correlation, Kendall’s rank correlation, Goodman–Kruskal generalized correlation coefficients, and their statistical significance, respectively.

Figure 2

A stepwise flowchart of an empirical solution (found in this work) to the ill-posed inverse problem of how to uniformly estimate the effects of mutations in plant proximal promoters on gene expression. Legend: see the legend of Figure 1. (a) The interchangeability of hsTBP (x-axis) and atTBP (y-axis) for transcription in vitro [45]. (b) Using Human_SNP_TATA_Z-tester [28] (see in-depth description in the Supplementary Materials [18,28,29,30,31,32,33,35,36,37,38,39,41,42,62,63,64,65,66,67]), we estimated the −ln(K_{D; hsTBP}) values of hsTBP affinity for the 90 bp DNA sequence of either the wildtype or the mutant variant of the Agrobacterium tumefaciens T-DNA TC7 promoter (AtTC7) [54] from [45], as shown by arrow 1. (c) Statistically significant linear regression of the predicted K_D;hsTBP values in relation to the F_hsTBP values measured using hsTBP [45], as expressed by Equation (1) written above the dashed line. (d) Statistically significant linear regression of ln(K_D;atTBP) values of atTBP affinity for the promoters in question, which are the F_atTBP values of the reporter gene expression measured in vitro using atTBP [45] and, here, rescaled using Equation (1) via the (calculated above) −ln(K_D;hsTBP) values of hsTBP affinity for the same promoters, as a solution to the ill-posed inverse problem under study, as expressed by Equation (2) written above the dashed line. (e) The Plant_SNP_TATA_Z-tester Web service created in this work implements the abovementioned solution to the ill-posed inverse problem being analyzed. (f) Statistically significant correlations between the −ln(K_D;atTBP) values calculated by Plant_SNP_TATA_Z-tester and measured effects of the mutations in the plant proximal promoter in question on gene expression, using atTBP in vitro [45]. Within Tikhonov’s regularization [61], these correlations characterize the match between the approximate solution found in this work to the ill-posed problem under study and its unknown true solution.

Figure 3

Determining the application limits of Plant_SNP_TATA_Z-tester on experimental data about tobacco development in the dark or under light [53], which indicated that the inverse problem under study is ill-posed. Legend: see the legend of Figure 1. (a) The result of our Web service Plant_SNP_TATA_Z-tester created in this work in the case of the comparison between the prototype Pmec of the artificial promoter for plant genes (the textbox “1st promoter”) and its mutant variant “G13c” (the textbox “2nd promoter”). (b,c) Statistically significant correlations between the in silico predicted −ln(K_D) values of TBP–promoter affinity expressed in ln units, which characterize the complexes formed by tobacco TBP binding to various artificial promoters based on the Pmec prototype [59] (x-axis) and the in vivo efficiency magnitudes of the reporter gene gusA expression (y-axis) on tobacco development in the dark or under light, respectively.

Figure 4

Verification of Plant_SNP_TATA_Z-tester using experimental data on the natural plant promoters (datasets 3–8). (a) The rice gene PAL promoter (phenylalanine ammonia-lyase) and Pol II, in whole-cell extracts of rice cell suspension cultures in vitro [46,47]. (b,c) The bean tRNA-Leu gene prompter, Pol III, in vitro [48] and ex vivo [49], respectively. (d) The thale cress U6-26 snRNA gene promoter (U6 small nuclear RNA; TAIR ID AT3G13855 [57]) and Pol III [50]. (e) The thale cress U2.2 snRNA gene promoter (U2 small nuclear RNA; TAIR ID AT3G57645 [57]). (f) The cauliflower mosaic virus (CaMV) promoter for the viral 35S transcript (GenBank AC MT611510 [58]), (e,f) RNA polymerase II. (d–f) Tobacco protoplasts ex vivo [50]. The natural (WT) and one of the mutant variants of the promoter under study are indicated.

Figure 4

Verification of Plant_SNP_TATA_Z-tester using experimental data on the natural plant promoters (datasets 3–8). (a) The rice gene PAL promoter (phenylalanine ammonia-lyase) and Pol II, in whole-cell extracts of rice cell suspension cultures in vitro [46,47]. (b,c) The bean tRNA-Leu gene prompter, Pol III, in vitro [48] and ex vivo [49], respectively. (d) The thale cress U6-26 snRNA gene promoter (U6 small nuclear RNA; TAIR ID AT3G13855 [57]) and Pol III [50]. (e) The thale cress U2.2 snRNA gene promoter (U2 small nuclear RNA; TAIR ID AT3G57645 [57]). (f) The cauliflower mosaic virus (CaMV) promoter for the viral 35S transcript (GenBank AC MT611510 [58]), (e,f) RNA polymerase II. (d–f) Tobacco protoplasts ex vivo [50]. The natural (WT) and one of the mutant variants of the promoter under study are indicated.

Figure 5

Verification of Plant_SNP_TATA_Z-tester using experimental data on the plant gene synthetic artificial proximal promoters designed on the basis of natural ones (datasets 9 and 10). (a) The synthetic artificial promoter designed on the basis of the thale cress U6-26 snRNA gene promoter, Pol III, and tobacco protoplasts ex vivo [51]. (b) The synthetic artificial promoter designed on the basis of the thale cress U2.2 snRNA gene promoter, Pol II, and tobacco protoplasts ex vivo [52].

Figure 6

Examples of the output of Web service Plant_SNP_TATA_Z-tester regarding assessment of agriculturally valuable mutations in plant proximal promoters for gene expression responsive to environmental factors during wheat development. (a) Deletions of the spacer (dotted box) between a TBP-binding site (solid box) and a TSS of wheat gene VRN1 can downregulate vernalization protein 1 encoded by this gene, which is the widely accepted genome-wide molecular marker of spring wheats in contrast to winter wheats [69]. (b) Statistically significant upregulation of the glutenin high-molecular-weight (HMW) subunit, which determines the gluten level in the grain, in wheatgrass (Thinopyrum) as compared to wheat (Triticum) [70]. This result explains how, in the harsh Siberian climate, wheat–wheatgrass hybrids increase grain baking quality without yield losses in comparison with the mother wheat variety [71].