Fig 1.
Workflow of model construction (left and central white forms), experimental model simulations to study the effects of ethanol stress and lncRNAs on the yeast cell cycle (yellow forms), and simulation results (green forms). ’Test model reliability’ includes simulations of the model with several random cell cycle perturbations and the effect of checkpoint nodes on cell cycle arrest. ’EtOH’ indicates ethanol, ’HT’ and ’LT’ are higher and lower ethanol-tolerant phenotypes, respectively. Details concerning the methods and the description of experimental model simulations (ranging from 1st to 4th in the figure) are described in the Materials and Methods section.
Fig 2.
Black circles are phenomenological nodes, white circles are genomic regulatory elements, green circles are checkpoint nodes, red circles are lncRNA nodes, white rectangles are protein or protein complex nodes, and blue rectangles are the S_proteins and G2_proteins nodes responsible for activating nodes from S and G2 phases, respectively. The lncRNA-protein interactions were predicted independently for the SEY6210 and BY4742 strains. The computational GINsim model is available at https://figshare.com/articles/software/Yeast_Cell_Cycle_Logic_Model/14503035.
Table 1.
Node types, range of values, and their biological meaning.
’a’ indicates that this node type also has some nodes that reach only three values. Thus, ’1’ and ’2’ indicate normal and high levels (or activity), respectively. ’b’, the checkpoint node Mating is a Boolean node. Then, ’0’ or ’1’ indicate inactivation or activation of the Mating node, respectively. ’c’ refers to the lncRNAs or selected genes for the experimental model simulations (details provided in the ‘Simulating the effects of ethanol stress-responsive lncRNAs and ethanol on the cell cycle’ in the Materials and Methods section).
Fig 3.
Model evaluation (A and B), growth curve analysis (C and D), and DNA damage-related gene expression (E). A: Specificity and sensitivity of simulating cell cycle mutants (mutants that had a disruption in the cell cycle) calculated by comparing the simulation outcomes and the reported description of each mutant. ’Inv.’ and ’Via’ are ’Inviable’ and ’Viable’ phenotypes, respectively; B: the percentage of functional cell cycles (details in the ’Model cycling rationale’ in the Materials and Methods section) from random cell cycle perturbation simulations (100, 1,000, and 10,000 simulations) and in the random perturbations from G1 phase. C: Population growth curve analysis of wild-type (WT) SEY6210 and BY4742 strains in the population rebound experiment after exposure to the highest ethanol stress level supported for each strain (’X’ axis). D: Population growth curve analysis in the population rebound experiment after ethanol (EtOH) stress relief (’X’ axis). The SEY6210 WT and SEY6210 lnc9136Δ1 mutant (partial deletion of lnc9136) strains were analyzed. E: Log2 fold-change (’X’ axis) in the expression of DNA damage-related genes between the treatment and control groups under the highest ethanol stress level for each strain [24]. The logic equations used to simulate these expressions as systems constraints are described in S4 Table. A higher populational growth rate (Log K in ’Y’ axis in ’C’ and ’D’) indicates the population with a higher ability to recover the growth rate after stress relief, and the number above bar indicates p values.
Fig 4.
Cell cycle predictions of the LT (A) and HT (B) phenotypes under ethanol stress (the first experimental model simulation in Fig 1). The box color indicates the node values in each simulation state. Cell cycle arrest is defined when the simulation usually displays single-state attractors without the MASS node returning to ’0’, as observed here; otherwise, the simulation of the model returned a functional cell cycle. The M phase arrest in the LT (A) is evidenced here by MASS > ’0’ and MITOSIS_EXIT = ’1’. G1 arrest in the HT (B) is evidenced here by the MASS > ’0’, and DNA_Replication = ’0’. Details concerning arrests are reported in the ’Model cycling rationale’ and Eq 1 in the Materials and Methods section. Since the model simulations in A and B represented cell cycle arrests, the attractors are the last states depicted on the X-axis.
Fig 5.
Simulating the SEY6210 cell cycle under ethanol stress with in silico overexpression of lnc9136 (the third experimental model simulation in Fig 1).
The box color indicates the node levels in each simulation state. The simulation presented a cyclic attractor related to a functional cell cycle, which includes all states depicted on the X-axis. Thus, the functional cell cycle is evidenced when the simulation outcomes cyclic attractors presenting the activation of all phenomenological nodes, further inhibited when the MITOSIS_EXIT node reaches ’2’, and restarting the cell cycle (MASS returning to ’0’), as observed here (see details in the ’Model cycling rationale’ in the Materials and Methods). The upper picture in the box on the left is the LT cell cycle arrest mechanism reported in Fig 4, while the bottom picture reports our suggested M arrest skip mechanism that may occur in cells under ethanol stress. The red ’X’ depicts suppression of a given regulation. The red ’X’ on the edge head ended at the Swe1 node, indicates that this node was blocked neither by Gin4 nor Hsl1 when lnc9136 was overexpressed under ethanol (EtOH) stress.
Fig 6.
Simulating the BY4742 cell cycle with active spindle checkpoint arrest (Misaligned_Spindle and Unattached_Kinetochores nodes fixed at ’2’) to assess the role of lnc10883 overexpression (’Over’) in silico under these conditions (the fourth experimental model simulation in Fig 1).
The box color indicates the node levels in each simulation state. The simulation presented a cyclic attractor related to a functional cell cycle, which includes all states depicted on the X-axis. Thus, the functional cell cycle is evidenced when the simulation outcomes cyclic attractors presenting the activation of all phenomenological nodes, further inhibited when the MITOSIS_EXIT node reaches ’2’, and restarting the cell cycle (MASS returning to ’0’), as observed here (see details in the ’Model cycling rationale’ in the Materials and Methods). The upper picture in the box on the left is the arrest mechanism mediated by spindle checkpoints, while the bottom picture reports our suggested arrest skip mechanism. The red ’X’ on the edges depicts the suppression of given regulations.
Table 2.
Oligonucleotides used in this study.
Bold nucleotides indicate the genomic target sequence. “Δ1”: oligonucleotides used to obtain the SEY6210 lnc9136Δ1 mutant; “Δ2”: oligonucleotides used to obtain the SEY6210 lnc9136Δ2; *: phosphorylated primer end. Details of target-homology repairs and Cas9 cleavage loci are shown in S4 Fig.