Figure 1.
Generic pathway diagram of lignin biosynthesis with species-specific extensions.
The widely accepted generic pathway consists of black arrows. It leads to the biosynthesis of three hydroxycinnamyl alcohol monomers that in turn give rise to p-hydroxyphenyl (H), guaiacyl (G), and syringyl (S) subunits of lignin. Listed next to each reaction arrow are the catalyzing enzymes, which are highlighted in bold if considered major. The lycophyte Selaginella moellendorffi contains two bi-functional enzymes, SmF5H and SmCOMT, which are shown in blue. Co-expression of these two enzymes would permit S. moellendorffi to synthesize coniferyl and sinapyl alcohol directly from p-coumaryl aldehyde and p-coumaryl alcohol. By contrast, Medicago truncatula has two functionally distinct isoforms of CCR, which are shown in red. The green arrow that connects caffeyl aldehyde to coniferyl aldehyde denotes the only non-canonical reaction that is likely to be functional in both S. moellendorffi and M. truncatula. Abbreviations: CAD, cinnamyl alcohol dehydrogenase; CCoAOMT, caffeoyl CoA O-methyltransferase; CCR, cinnamoyl CoA reductase; C3′H, p-coumaroyl shikimate 3′-hydroxylase; C4H, cinnamate 4-hydroxylase; 4CL, 4-coumarate:CoA ligase; COMT, caffeic acid O-methyltransferase; F5H, ferulate 5-hydroxylase; HCT, hydroxycinnamoyl-CoA:shikimate hydroxycinnamoyl transferase; PAL, L-phenylalanine ammonia-lyase.
Figure 2.
Scaffold of topological configurations.
The relevant metabolites and enzymatic reactions (arrows) for the biosynthesis of guaiacyl (G), 5-hydroxyguaiacyl (5HG), and syringyl (S) lignin monomers are shown in black, if they are included in all topological configurations, or gray, if they are included in only some specific configurations. G and S channels are represented as blue and red arrows, respectively. Notice that 5-hydroxyconiferyl alcohol is allowed to be incorporated into lignin polymer as 5HG subunit because this actually occurs when COMT is down-regulated [41]. Enzymes are highlighted in bold and italics.
Figure 3.
Lists of topological configurations and regulatory mechanisms.
Panel A: The topological configurations differ in their numbers of edges. Panel B: The orange arrows represent activation processes, whereas the blocked lines (aqua) represent inhibition processes. Arrows colored in gray are reactions included in only some specific topological configurations. Metabolite names: 1 caffeoyl CoA;
, caffeyl aldehyde;
feruloyl CoA;
coniferyl aldehyde;
coniferyl alcohol;
5-hydroxyconiferyl aldehyde.
Figure 4.
Simulation results for pathway designs without crosstalk.
Each of the 6 panels represents one topological configuration (as indicated with labels (A, C, D, G, K, N) corresponding to model designs in Figure 3) with at least one randomly parameterized S-system model that yields quantitatively correct predictions of S/G ratios for both CCoAOMT and COMT down-regulated alfalfa plants. Each open circle refers to the S/G ratio of CCR1 and CCR2 in a M. truncatula knockout mutant, as predicted by one randomly parameterized S-system model; its color indicates the type of regulation. The gray strips denote regions within 5% of the wild-type level; model predictions within these strips are considered essentially the same as wild-type. Qualitatively correct predictions should fall into the northwest quadrant. It is evident that not a single model instantiation is admissible. The total number of randomly parameterized model instantiations per panel was 105.
Figure 5.
Simulation results for pathway designs using only Mechanism 3.
See Figure 3B for the structure of this mechanism and the legend of Figure 4 for more information on details shown. In contrast to the results in Figure 4, the pathway designs analyzed here permit numerous admissible model instantiations (topologies A, B, E, F, I, and O), which fall into the northwest quadrant. The total number of randomly parameterized model instantiations per panel was 105.
Figure 6.
Simulation results for pathway designs that contain Mechanisms 1 and 3 simultaneously.
See Figure 3B for the structure of these mechanisms and the legend of Figure 4 for more information on details shown. Similar to the results in Figure 5, the pathway designs analyzed here permit numerous admissible model instantiations (topologies A, B, E, F, I, and O), which fall into the northwest quadrant. The total number of randomly parameterized model instantiations per panel was 105.
Figure 7.
Experimental data indicating that the activation of CCoAOMT-mediated methylation of caffeoyl CoA by caffeyl aldehyde, while statistically significant, is too small to be biologically significant.
Error bars, mean ± s.d.; ***p<0.001, *p<0.05 by Student's t-test; n = 3.
Figure 8.
Summary of simulation results from 304 designs.
Each row corresponds to one crosstalk pattern, whereas each column in the table to the right of dashed line corresponds to one topological configuration. A design is represented by a filled circle if at least one of the 105 randomly parameterized model instantiations is valid. Empty circles thus refer to designs that are incongruent with observations. Rows highlighted in red contain at least one topological configuration with valid model instantiations.
Figure 9.
Relative levels of caffeyl aldehyde (compared to wild-type values) in simulations of four down-regulated lines.
Each panel is shaded to highlight the results from four different crosstalk patterns. These patterns, when combined with specific topological configurations, give rise to designs with valid model instantiations (cf. rows with red circles in Figure 8). In contrast to the other three perturbation schemes, where all four crosstalk patterns (and their corresponding designs) exhibit similar responses regarding the caffeyl aldehyde level, knocking out ccr1 is associated with a higher caffeyl aldehyde level only for the two crosstalk patterns including Mechanism 1. The circles, colored according to topological configuration, are the medians, and the error bars represent interquartile ranges. The dashed line in each panel, if present, denotes the wild-type level of caffeyl aldehyde.
Figure 10.
Robust configurations are evolutionarily connected.
Each node represents a specific topological configuration (see Figure 3); two nodes are connected if the corresponding configurations differ only by one edge. The subgraph of all the robust configurations, colored in red, is connected, thereby indicating the potential of direct evolvability.